The Cookbook - Symfony
The Cookbook
Version: master
generated on December 3, 2014
The Cookbook (master)
This work is licensed under the “Attribution-Share Alike 3.0 Unported” license (http://creativecommons.org/
licenses/by-sa/3.0/).
You are free to share (to copy, distribute and transmit the work), and to remix (to adapt the work) under the
following conditions:
• Attribution: You must attribute the work in the manner specified by the author or licensor (but
not in any way that suggests that they endorse you or your use of the work).
• Share Alike: If you alter, transform, or build upon this work, you may distribute the resulting work
only under the same, similar or a compatible license. For any reuse or distribution, you must make
clear to others the license terms of this work.
The information in this book is distributed on an “as is” basis, without warranty. Although every precaution
has been taken in the preparation of this work, neither the author(s) nor SensioLabs shall have any liability to
any person or entity with respect to any loss or damage caused or alleged to be caused directly or indirectly by
the information contained in this work.
If you find typos or errors, feel free to report them by creating a ticket on the Symfony ticketing system
(http://github.com/symfony/symfony-docs/issues). Based on tickets and users feedback, this book is
continuously updated.
Contents at a Glance
How to Use Assetic for Asset Management ..........................................................................................6
How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) ...............................................................12
How to Minify JavaScripts and Stylesheets with YUI Compressor.......................................................16
How to Use Assetic for Image Optimization with Twig Functions ......................................................19
How to Apply an Assetic Filter to a specific File Extension .................................................................22
How to Install 3rd Party Bundles .......................................................................................................24
Best Practices for Reusable Bundles ...................................................................................................27
How to Use Bundle Inheritance to Override Parts of a Bundle ............................................................34
How to Override any Part of a Bundle ...............................................................................................36
How to Remove the AcmeDemoBundle .............................................................................................39
How to Load Service Configuration inside a Bundle ...........................................................................42
How to Create Friendly Configuration for a Bundle ...........................................................................45
How to Simplify Configuration of multiple Bundles ...........................................................................51
How to Use Varnish to Speed up my Website ....................................................................................54
How to Master and Create new Environments ...................................................................................59
How to Override Symfony's default Directory Structure .....................................................................64
Using Parameters within a Dependency Injection Class ......................................................................67
Understanding how the Front Controller, Kernel and Environments Work together............................70
How to Set external Parameters in the Service Container ....................................................................73
How to Use PdoSessionHandler to Store Sessions in the Database ......................................................76
How to Use the Apache Router .........................................................................................................79
Configuring a Web Server .................................................................................................................82
How to Organize Configuration Files ................................................................................................87
How to Create a Console Command .................................................................................................92
How to Use the Console....................................................................................................................97
How to Generate URLs and Send Emails from the Console ................................................................99
How to Enable Logging in Console Commands ............................................................................... 101
How to Define Commands as Services ............................................................................................. 105
How to Customize Error Pages ........................................................................................................ 107
How to Define Controllers as Services ............................................................................................. 112
How to Optimize your Development Environment for Debugging.................................................... 116
How to Deploy a Symfony Application ............................................................................................ 118
Deploying to Microsoft Azure Website Cloud .................................................................................. 122
Deploying to Heroku Cloud ............................................................................................................ 135
How to Handle File Uploads with Doctrine ..................................................................................... 139
How to use Doctrine Extensions: Timestampable, Sluggable, Translatable, etc. ................................ 148
PDF brought to you by
generated on December 3, 2014
Contents at a Glance | iii
How to Register Event Listeners and Subscribers ............................................................................. 149
How to Use Doctrine DBAL ............................................................................................................ 152
How to Generate Entities from an Existing Database........................................................................ 154
How to Work with multiple Entity Managers and Connections ........................................................ 158
How to Register custom DQL Functions.......................................................................................... 161
How to Define Relationships with Abstract Classes and Interfaces.................................................... 162
How to Provide Model Classes for several Doctrine Implementations ............................................... 165
How to Implement a simple Registration Form ................................................................................ 169
Console Commands........................................................................................................................ 175
How to Send an Email..................................................................................................................... 176
How to Use Gmail to Send Emails ................................................................................................... 179
How to Use the Cloud to Send Emails ............................................................................................. 180
How to Work with Emails during Development............................................................................... 182
How to Spool Emails....................................................................................................................... 184
How to Test that an Email is Sent in a functional Test...................................................................... 186
How to Setup before and after Filters............................................................................................... 188
How to Extend a Class without Using Inheritance............................................................................ 192
How to Customize a Method Behavior without Using Inheritance .................................................... 195
How to use Expressions in Security, Routing, Services, and Validation ............................................. 197
How to Customize Form Rendering ................................................................................................ 198
How to Use Data Transformers ....................................................................................................... 211
How to Dynamically Modify Forms Using Form Events ................................................................... 218
How to Embed a Collection of Forms .............................................................................................. 230
How to Create a Custom Form Field Type....................................................................................... 243
How to Create a Form Type Extension ............................................................................................ 248
How to Reduce Code Duplication with "inherit_data" ..................................................................... 253
How to Unit Test your Forms.......................................................................................................... 257
How to Configure empty Data for a Form Class............................................................................... 262
How to Use the submit() Function to Handle Form Submissions...................................................... 264
How to Use the virtual Form Field Option....................................................................................... 267
How to Use Monolog to Write Logs ................................................................................................ 268
How to Configure Monolog to Email Errors .................................................................................... 272
How to Configure Monolog to Display Console Messages................................................................ 274
How to Configure Monolog to Exclude 404 Errors from the Log ...................................................... 276
How to Log Messages to different Files ............................................................................................ 277
How to Create a custom Data Collector........................................................................................... 279
How to Use Matchers to Enable the Profiler Conditionally ............................................................... 282
Switching the Profiler Storage .......................................................................................................... 284
How to Configure Symfony to Work behind a Load Balancer or a Reverse Proxy .............................. 285
How to Register a new Request Format and Mime Type................................................................... 287
How to Force Routes to always Use HTTPS or HTTP ...................................................................... 289
How to Allow a "/" Character in a Route Parameter ......................................................................... 290
How to Configure a Redirect without a custom Controller ............................................................... 291
How to Use HTTP Methods beyond GET and POST in Routes ........................................................ 293
How to Use Service Container Parameters in your Routes ................................................................ 295
How to Create a custom Route Loader ............................................................................................ 297
Redirect URLs with a Trailing Slash................................................................................................. 301
iv | Contents at a Glance
Contents at a Glance | 4
How to Pass Extra Information from a Route to a Controller............................................................ 303
How to Load Security Users from the Database (the Entity Provider)................................................ 304
How to Add "Remember Me" Login Functionality ........................................................................... 317
How to Impersonate a User ............................................................................................................. 320
How to Implement your own Voter to Blacklist IP Addresses ........................................................... 322
How to Use Voters to Check User Permissions................................................................................. 325
How to Use Access Control Lists (ACLs) ......................................................................................... 329
How to Use advanced ACL Concepts .............................................................................................. 333
How to Force HTTPS or HTTP for different URLs........................................................................... 337
How to Restrict Firewalls to a Specific Request ................................................................................ 338
How to Customize your Form Login................................................................................................ 340
How to Secure any Service or Method in your Application ............................................................... 343
How to Create a custom User Provider ............................................................................................ 347
How to Create a Custom Form Password Authenticator................................................................... 352
How to Authenticate Users with API Keys ....................................................................................... 356
How to Create a custom Authentication Provider............................................................................. 365
Using pre Authenticated Security Firewalls ...................................................................................... 374
How to Change the default Target Path Behavior ............................................................................. 376
Using CSRF Protection in the Login Form........................................................................................ 378
How to Choose the Password Encoder Algorithm Dynamically ........................................................ 380
How to Use the Serializer ................................................................................................................ 382
How to Create an Event Listener ..................................................................................................... 384
How to Work with Scopes .............................................................................................................. 387
How to Work with Compiler Passes in Bundles ............................................................................... 392
Session Proxy Examples .................................................................................................................. 393
Making the Locale "Sticky" during a User's Session .......................................................................... 395
Configuring the Directory where Session Files are Saved .................................................................. 397
Bridge a legacy Application with Symfony Sessions .......................................................................... 399
Limit Session Metadata Writes ........................................................................................................ 401
How Symfony2 Differs from Symfony1 ............................................................................................ 402
How to Inject Variables into all Templates (i.e. global Variables) ...................................................... 407
How to Use and Register Namespaced Twig Paths ........................................................................... 409
How to Use PHP instead of Twig for Templates............................................................................... 411
How to Write a custom Twig Extension .......................................................................................... 417
How to Render a Template without a custom Controller.................................................................. 420
How to Simulate HTTP Authentication in a Functional Test ............................................................ 422
How to Simulate Authentication with a Token in a Functional Test.................................................. 423
How to Test the Interaction of several Clients .................................................................................. 425
How to Use the Profiler in a Functional Test.................................................................................... 427
How to Test Code that Interacts with the Database.......................................................................... 429
How to Test Doctrine Repositories .................................................................................................. 432
How to Customize the Bootstrap Process before Running Tests........................................................ 434
How to Create a custom Validation Constraint ................................................................................ 436
How to Use PHP's built-in Web Server ............................................................................................ 440
How to Create a SOAP Web Service in a Symfony Controller ........................................................... 443
How to Create and Store a Symfony Project in Git ........................................................................... 447
How to Create and Store a Symfony Project in Subversion................................................................ 451
PDF brought to you by
generated on December 3, 2014
Contents at a Glance | v
Chapter 1
How to Use Assetic for Asset Management
Assetic combines two major ideas: assets and filters. The assets are files such as CSS, JavaScript and image
files. The filters are things that can be applied to these files before they are served to the browser. This
allows a separation between the asset files stored in the application and the files actually presented to the
user.
Without Assetic, you just serve the files that are stored in the application directly:
Listing 1-1
1 <script src="{{ asset('js/script.js') }}" type="text/javascript"></script>
But with Assetic, you can manipulate these assets however you want (or load them from anywhere) before
serving them. This means you can:
• Minify and combine all of your CSS and JS files
• Run all (or just some) of your CSS or JS files through some sort of compiler, such as LESS,
SASS or CoffeeScript
• Run image optimizations on your images
Assets
Using Assetic provides many advantages over directly serving the files. The files do not need to be stored
where they are served from and can be drawn from various sources such as from within a bundle.
You can use Assetic to process CSS stylesheets, JavaScript files and images. The philosophy behind adding
either is basically the same, but with a slightly different syntax.
Including JavaScript Files
To include JavaScript files, use the javascripts tag in any template:
Listing 1-2
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' %}
2
<script type="text/javascript" src="{{ asset_url }}"></script>
3 {% endjavascripts %}
PDF brought to you by
generated on December 3, 2014
Chapter 1: How to Use Assetic for Asset Management | 6
If you're using the default block names from the Symfony Standard Edition, the javascripts tag
will most commonly live in the javascripts block:
Listing 1-3
1
2
3
4
5
6
7
{# ... #}
{% block javascripts %}
{% javascripts '@AcmeFooBundle/Resources/public/js/*' %}
<script type="text/javascript" src="{{ asset_url }}"></script>
{% endjavascripts %}
{% endblock %}
{# ... #}
You can also include CSS Stylesheets: see Including CSS Stylesheets.
In this example, all of the files in the Resources/public/js/ directory of the AcmeFooBundle will be
loaded and served from a different location. The actual rendered tag might simply look like:
Listing 1-4
1 <script src="/app_dev.php/js/abcd123.js"></script>
This is a key point: once you let Assetic handle your assets, the files are served from a different location.
This will cause problems with CSS files that reference images by their relative path. See Fixing CSS Paths
with the cssrewrite Filter.
Including CSS Stylesheets
To bring in CSS stylesheets, you can use the same methodologies seen above, except with the
stylesheets tag:
Listing 1-5
1 {% stylesheets 'bundles/acme_foo/css/*' filter='cssrewrite' %}
2
<link rel="stylesheet" href="{{ asset_url }}" />
3 {% endstylesheets %}
If you're using the default block names from the Symfony Standard Edition, the stylesheets tag
will most commonly live in the stylesheets block:
Listing 1-6
1
2
3
4
5
6
7
{# ... #}
{% block stylesheets %}
{% stylesheets 'bundles/acme_foo/css/*' filter='cssrewrite' %}
<link rel="stylesheet" href="{{ asset_url }}" />
{% endstylesheets %}
{% endblock %}
{# ... #}
But because Assetic changes the paths to your assets, this will break any background images (or other
paths) that uses relative paths, unless you use the cssrewrite filter.
PDF brought to you by
generated on December 3, 2014
Chapter 1: How to Use Assetic for Asset Management | 7
Notice that in the original example that included JavaScript files, you referred to the files using
a path like @AcmeFooBundle/Resources/public/file.js, but that in this example, you referred
to the CSS files using their actual, publicly-accessible path: bundles/acme_foo/css. You can use
either, except that there is a known issue that causes the cssrewrite filter to fail when using the
@AcmeFooBundle syntax for CSS Stylesheets.
Including Images
To include an image you can use the image tag.
Listing 1-7
1 {% image '@AcmeFooBundle/Resources/public/images/example.jpg' %}
2
<img src="{{ asset_url }}" alt="Example" />
3 {% endimage %}
You can also use Assetic for image optimization. More information in How to Use Assetic for Image
Optimization with Twig Functions.
Fixing CSS Paths with the cssrewrite Filter
Since Assetic generates new URLs for your assets, any relative paths inside your CSS files will break. To
fix this, make sure to use the cssrewrite filter with your stylesheets tag. This parses your CSS files
and corrects the paths internally to reflect the new location.
You can see an example in the previous section.
When using the cssrewrite filter, don't refer to your CSS files using the @AcmeFooBundle syntax.
See the note in the above section for details.
Combining Assets
One feature of Assetic is that it will combine many files into one. This helps to reduce the number of
HTTP requests, which is great for front end performance. It also allows you to maintain the files more
easily by splitting them into manageable parts. This can help with re-usability as you can easily split
project-specific files from those which can be used in other applications, but still serve them as a single
file:
Listing 1-8
1 {% javascripts
2
'@AcmeFooBundle/Resources/public/js/*'
3
'@AcmeBarBundle/Resources/public/js/form.js'
4
'@AcmeBarBundle/Resources/public/js/calendar.js' %}
5
<script src="{{ asset_url }}"></script>
6 {% endjavascripts %}
In the dev environment, each file is still served individually, so that you can debug problems more easily.
However, in the prod environment (or more specifically, when the debug flag is false), this will be
rendered as a single script tag, which contains the contents of all of the JavaScript files.
If you're new to Assetic and try to use your application in the prod environment (by using the
app.php controller), you'll likely see that all of your CSS and JS breaks. Don't worry! This is on
purpose. For details on using Assetic in the prod environment, see Dumping Asset Files.
PDF brought to you by
generated on December 3, 2014
Chapter 1: How to Use Assetic for Asset Management | 8
And combining files doesn't only apply to your files. You can also use Assetic to combine third party
assets, such as jQuery, with your own into a single file:
Listing 1-9
1 {% javascripts
2
'@AcmeFooBundle/Resources/public/js/thirdparty/jquery.js'
3
'@AcmeFooBundle/Resources/public/js/*' %}
4
<script src="{{ asset_url }}"></script>
5 {% endjavascripts %}
Using Named Assets
AsseticBundle configuration directives allow you to define named asset sets. You can do so by defining
the input files, filters and output files in your configuration under the assetic section. Read more in the
assetic config reference.
Listing 1-10
1 # app/config/config.yml
2 assetic:
3
assets:
4
jquery_and_ui:
5
inputs:
6
- '@AcmeFooBundle/Resources/public/js/thirdparty/jquery.js'
7
- '@AcmeFooBundle/Resources/public/js/thirdparty/jquery.ui.js'
After you have defined the named assets, you can reference them in your templates with the
@named_asset notation:
Listing 1-11
1 {% javascripts
2
'@jquery_and_ui'
3
'@AcmeFooBundle/Resources/public/js/*' %}
4
<script src="{{ asset_url }}"></script>
5 {% endjavascripts %}
Filters
Once they're managed by Assetic, you can apply filters to your assets before they are served. This includes
filters that compress the output of your assets for smaller file sizes (and better front-end optimization).
Other filters can compile JavaScript file from CoffeeScript files and process SASS into CSS. In fact, Assetic
has a long list of available filters.
Many of the filters do not do the work directly, but use existing third-party libraries to do the heavylifting. This means that you'll often need to install a third-party library to use a filter. The great advantage
of using Assetic to invoke these libraries (as opposed to using them directly) is that instead of having to
run them manually after you work on the files, Assetic will take care of this for you and remove this step
altogether from your development and deployment processes.
To use a filter, you first need to specify it in the Assetic configuration. Adding a filter here doesn't mean
it's being used - it just means that it's available to use (you'll use the filter below).
For example to use the UglifyJS JavaScript minifier the following config should be added:
Listing 1-12
1 # app/config/config.yml
2 assetic:
3
filters:
PDF brought to you by
generated on December 3, 2014
Chapter 1: How to Use Assetic for Asset Management | 9
4
5
uglifyjs2:
bin: /usr/local/bin/uglifyjs
Now, to actually use the filter on a group of JavaScript files, add it into your template:
Listing 1-13
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' filter='uglifyjs2' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
A more detailed guide about configuring and using Assetic filters as well as details of Assetic's debug
mode can be found in How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS).
Controlling the URL Used
If you wish to, you can control the URLs that Assetic produces. This is done from the template and is
relative to the public document root:
Listing 1-14
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' output='js/compiled/main.js' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
Symfony also contains a method for cache busting, where the final URL generated by Assetic
contains a query parameter that can be incremented via configuration on each deployment. For
more information, see the assets_version configuration option.
Dumping Asset Files
In the dev environment, Assetic generates paths to CSS and JavaScript files that don't physically exist on
your computer. But they render nonetheless because an internal Symfony controller opens the files and
serves back the content (after running any filters).
This kind of dynamic serving of processed assets is great because it means that you can immediately see
the new state of any asset files you change. It's also bad, because it can be quite slow. If you're using a lot
of filters, it might be downright frustrating.
Fortunately, Assetic provides a way to dump your assets to real files, instead of being generated
dynamically.
Dumping Asset Files in the prod Environment
In the prod environment, your JS and CSS files are represented by a single tag each. In other words,
instead of seeing each JavaScript file you're including in your source, you'll likely just see something like
this:
Listing 1-15
1 <script src="/js/abcd123.js"></script>
Moreover, that file does not actually exist, nor is it dynamically rendered by Symfony (as the asset files
are in the dev environment). This is on purpose - letting Symfony generate these files dynamically in a
production environment is just too slow.
PDF brought to you by
generated on December 3, 2014
Chapter 1: How to Use Assetic for Asset Management | 10
Instead, each time you use your app in the prod environment (and therefore, each time you deploy), you
should run the following task:
Listing 1-16
1 $ php app/console assetic:dump --env=prod --no-debug
This will physically generate and write each file that you need (e.g. /js/abcd123.js). If you update any
of your assets, you'll need to run this again to regenerate the file.
Dumping Asset Files in the dev Environment
By default, each asset path generated in the dev environment is handled dynamically by Symfony. This
has no disadvantage (you can see your changes immediately), except that assets can load noticeably slow.
If you feel like your assets are loading too slowly, follow this guide.
First, tell Symfony to stop trying to process these files dynamically. Make the following change in your
config_dev.yml file:
Listing 1-17
1 # app/config/config_dev.yml
2 assetic:
3
use_controller: false
Next, since Symfony is no longer generating these assets for you, you'll need to dump them manually. To
do so, run the following:
Listing 1-18
1 $ php app/console assetic:dump
This physically writes all of the asset files you need for your dev environment. The big disadvantage is
that you need to run this each time you update an asset. Fortunately, by passing the --watch option, the
command will automatically regenerate assets as they change:
Listing 1-19
1 $ php app/console assetic:dump --watch
Since running this command in the dev environment may generate a bunch of files, it's usually a good
idea to point your generated assets files to some isolated directory (e.g. /js/compiled), to keep things
organized:
Listing 1-20
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' output='js/compiled/main.js' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
PDF brought to you by
generated on December 3, 2014
Chapter 1: How to Use Assetic for Asset Management | 11
Chapter 2
How to Minify CSS/JS Files (Using UglifyJS and
UglifyCSS)
UglifyJS1 is a JavaScript parser/compressor/beautifier toolkit. It can be used to combine and minify
JavaScript assets so that they require less HTTP requests and make your site load faster. UglifyCSS2 is a
CSS compressor/beautifier that is very similar to UglifyJS.
In this cookbook, the installation, configuration and usage of UglifyJS is shown in detail. UglifyCSS
works pretty much the same way and is only talked about briefly.
Install UglifyJS
UglifyJS is available as an Node.js3 npm module and can be installed using npm. First, you need to install
Node.js4. Afterwards you can install UglifyJS using npm:
Listing 2-1
1 $ npm install -g uglify-js
This command will install UglifyJS globally and you may need to run it as a root user.
1. https://github.com/mishoo/UglifyJS
2. https://github.com/fmarcia/UglifyCSS
3. http://nodejs.org/
4. http://nodejs.org/
PDF brought to you by
generated on December 3, 2014
Chapter 2: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 12
It's also possible to install UglifyJS inside your project only. To do this, install it without the -g
option and specify the path where to put the module:
Listing 2-2
1 $ cd /path/to/symfony
2 $ mkdir app/Resources/node_modules
3 $ npm install uglify-js --prefix app/Resources
It is recommended that you install UglifyJS in your app/Resources folder and add the
node_modules folder to version control. Alternatively, you can create an npm package.json5 file and
specify your dependencies there.
Depending on your installation method, you should either be able to execute the uglifyjs executable
globally, or execute the physical file that lives in the node_modules directory:
Listing 2-3
1 $ uglifyjs --help
2
3 $ ./app/Resources/node_modules/.bin/uglifyjs --help
Configure the uglifyjs2 Filter
Now we need to configure Symfony to use the uglifyjs2 filter when processing your JavaScripts:
Listing 2-4
1 # app/config/config.yml
2 assetic:
3
filters:
4
uglifyjs2:
5
# the path to the uglifyjs executable
6
bin: /usr/local/bin/uglifyjs
The path where UglifyJS is installed may vary depending on your system. To find out where npm
stores the bin folder, you can use the following command:
Listing 2-5
1 $ npm bin -g
It should output a folder on your system, inside which you should find the UglifyJS executable.
If you installed UglifyJS locally, you can find the bin folder inside the node_modules folder. It's
called .bin in this case.
You now have access to the uglifyjs2 filter in your application.
Configure the node Binary
Assetic tries to find the node binary automatically. If it cannot be found, you can configure its location
using the node key:
Listing 2-6
5. http://package.json.nodejitsu.com/
PDF brought to you by
generated on December 3, 2014
Chapter 2: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 13
1 # app/config/config.yml
2 assetic:
3
# the path to the node executable
4
node: /usr/bin/nodejs
5
filters:
6
uglifyjs2:
7
# the path to the uglifyjs executable
8
bin: /usr/local/bin/uglifyjs
Minify your Assets
In order to use UglifyJS on your assets, you need to apply it to them. Since your assets are a part of the
view layer, this work is done in your templates:
Listing 2-7
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' filter='uglifyjs2' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
The above example assumes that you have a bundle called AcmeFooBundle and your JavaScript
files are in the Resources/public/js directory under your bundle. This isn't important however you can include your JavaScript files no matter where they are.
With the addition of the uglifyjs2 filter to the asset tags above, you should now see minified JavaScripts
coming over the wire much faster.
Disable Minification in Debug Mode
Minified JavaScripts are very difficult to read, let alone debug. Because of this, Assetic lets you disable a
certain filter when your application is in debug (e.g. app_dev.php) mode. You can do this by prefixing
the filter name in your template with a question mark: ?. This tells Assetic to only apply this filter when
debug mode is off (e.g. app.php):
Listing 2-8
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' filter='?uglifyjs2' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
To try this out, switch to your prod environment (app.php). But before you do, don't forget to clear your
cache and dump your assetic assets.
Instead of adding the filter to the asset tags, you can also globally enable it by adding the apply_to
attribute to the filter configuration, for example in the uglifyjs2 filter apply_to: "\.js$". To
only have the filter applied in production, add this to the config_prod file rather than the common
config file. For details on applying filters by file extension, see Filtering Based on a File Extension.
Install, Configure and Use UglifyCSS
The usage of UglifyCSS works the same way as UglifyJS. First, make sure the node package is installed:
PDF brought to you by
generated on December 3, 2014
Chapter 2: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 14
Listing 2-9
1 $ npm install -g uglifycss
Next, add the configuration for this filter:
Listing 2-10
1 # app/config/config.yml
2 assetic:
3
filters:
4
uglifycss:
5
bin: /usr/local/bin/uglifycss
To use the filter for your CSS files, add the filter to the Assetic stylesheets helper:
Listing 2-11
1 {% stylesheets 'bundles/AcmeFoo/css/*' filter='uglifycss' filter='cssrewrite' %}
2
<link rel="stylesheet" href="{{ asset_url }}" />
3 {% endstylesheets %}
Just like with the uglifyjs2 filter, if you prefix the filter name with ? (i.e. ?uglifycss), the minification
will only happen when you're not in debug mode.
PDF brought to you by
generated on December 3, 2014
Chapter 2: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 15
Chapter 3
How to Minify JavaScripts and Stylesheets with
YUI Compressor
Yahoo! provides an excellent utility for minifying JavaScripts and stylesheets so they travel over the wire
faster, the YUI Compressor1. Thanks to Assetic, you can take advantage of this tool very easily.
The YUI Compressor is no longer maintained by Yahoo2 but by an independent volunteer.
Moreover, Yahoo has decided to stop all new development on YUI3 and to move to other modern
alternatives such as Node.js.
That's why you are strongly advised to avoid using YUI utilities unless strictly necessary. Read
How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) for a modern and up-to-date alternative.
Download the YUI Compressor JAR
The YUI Compressor is written in Java and distributed as a JAR. Download the JAR4 from the Yahoo! site
and save it to app/Resources/java/yuicompressor.jar.
Configure the YUI Filters
Now you need to configure two Assetic filters in your application, one for minifying JavaScripts with the
YUI Compressor and one for minifying stylesheets:
Listing 3-1
1 # app/config/config.yml
2 assetic:
1. http://developer.yahoo.com/yui/compressor/
2. http://www.yuiblog.com/blog/2013/01/24/yui-compressor-has-a-new-owner/
3. http://yahooeng.tumblr.com/post/96098168666/important-announcement-regarding-yui
4. https://github.com/yui/yuicompressor/releases
PDF brought to you by
generated on December 3, 2014
Chapter 3: How to Minify JavaScripts and Stylesheets with YUI Compressor | 16
3
4
5
6
7
8
# java: "/usr/bin/java"
filters:
yui_css:
jar: "%kernel.root_dir%/Resources/java/yuicompressor.jar"
yui_js:
jar: "%kernel.root_dir%/Resources/java/yuicompressor.jar"
Windows users need to remember to update config to proper Java location. In Windows7 x64 bit
by default it's C:\Program Files (x86)\Java\jre6\bin\java.exe.
You now have access to two new Assetic filters in your application: yui_css and yui_js. These will use
the YUI Compressor to minify stylesheets and JavaScripts, respectively.
Minify your Assets
You have YUI Compressor configured now, but nothing is going to happen until you apply one of these
filters to an asset. Since your assets are a part of the view layer, this work is done in your templates:
Listing 3-2
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' filter='yui_js' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
The above example assumes that you have a bundle called AcmeFooBundle and your JavaScript
files are in the Resources/public/js directory under your bundle. This isn't important however you can include your JavaScript files no matter where they are.
With the addition of the yui_js filter to the asset tags above, you should now see minified JavaScripts
coming over the wire much faster. The same process can be repeated to minify your stylesheets.
Listing 3-3
1 {% stylesheets '@AcmeFooBundle/Resources/public/css/*' filter='yui_css' %}
2
<link rel="stylesheet" type="text/css" media="screen" href="{{ asset_url }}" />
3 {% endstylesheets %}
Disable Minification in Debug Mode
Minified JavaScripts and Stylesheets are very difficult to read, let alone debug. Because of this, Assetic lets
you disable a certain filter when your application is in debug mode. You can do this by prefixing the filter
name in your template with a question mark: ?. This tells Assetic to only apply this filter when debug
mode is off.
Listing 3-4
1 {% javascripts '@AcmeFooBundle/Resources/public/js/*' filter='?yui_js' %}
2
<script src="{{ asset_url }}"></script>
3 {% endjavascripts %}
PDF brought to you by
generated on December 3, 2014
Chapter 3: How to Minify JavaScripts and Stylesheets with YUI Compressor | 17
Instead of adding the filter to the asset tags, you can also globally enable it by adding the apply_to
attribute to the filter configuration, for example in the yui_js filter apply_to: "\.js$". To only
have the filter applied in production, add this to the config_prod file rather than the common
config file. For details on applying filters by file extension, see Filtering Based on a File Extension.
PDF brought to you by
generated on December 3, 2014
Chapter 3: How to Minify JavaScripts and Stylesheets with YUI Compressor | 18
Chapter 4
How to Use Assetic for Image Optimization
with Twig Functions
Amongst its many filters, Assetic has four filters which can be used for on-the-fly image optimization.
This allows you to get the benefits of smaller file sizes without having to use an image editor to process
each image. The results are cached and can be dumped for production so there is no performance hit for
your end users.
Using Jpegoptim
Jpegoptim1 is a utility for optimizing JPEG files. To use it with Assetic, add the following to the Assetic
config:
Listing 4-1
1 # app/config/config.yml
2 assetic:
3
filters:
4
jpegoptim:
5
bin: path/to/jpegoptim
Notice that to use jpegoptim, you must have it already installed on your system. The bin option
points to the location of the compiled binary.
It can now be used from a template:
Listing 4-2
1 {% image '@AcmeFooBundle/Resources/public/images/example.jpg'
2
filter='jpegoptim' output='/images/example.jpg' %}
3
<img src="{{ asset_url }}" alt="Example"/>
4 {% endimage %}
1. http://www.kokkonen.net/tjko/projects.html
PDF brought to you by
generated on December 3, 2014
Chapter 4: How to Use Assetic for Image Optimization with Twig Functions | 19
Removing all EXIF Data
By default, running this filter only removes some of the meta information stored in the file. Any EXIF
data and comments are not removed, but you can remove these by using the strip_all option:
Listing 4-3
1 # app/config/config.yml
2 assetic:
3
filters:
4
jpegoptim:
5
bin: path/to/jpegoptim
6
strip_all: true
Lowering maximum Quality
The quality level of the JPEG is not affected by default. You can gain further file size reductions by setting
the max quality setting lower than the current level of the images. This will of course be at the expense of
image quality:
Listing 4-4
1 # app/config/config.yml
2 assetic:
3
filters:
4
jpegoptim:
5
bin: path/to/jpegoptim
6
max: 70
Shorter Syntax: Twig Function
If you're using Twig, it's possible to achieve all of this with a shorter syntax by enabling and using a
special Twig function. Start by adding the following config:
Listing 4-5
1 # app/config/config.yml
2 assetic:
3
filters:
4
jpegoptim:
5
bin: path/to/jpegoptim
6
twig:
7
functions:
8
jpegoptim: ~
The Twig template can now be changed to the following:
Listing 4-6
1 <img src="{{ jpegoptim('@AcmeFooBundle/Resources/public/images/example.jpg') }}"
alt="Example"/>
You can specify the output directory in the config in the following way:
Listing 4-7
1 # app/config/config.yml
2 assetic:
3
filters:
4
jpegoptim:
5
bin: path/to/jpegoptim
6
twig:
PDF brought to you by
generated on December 3, 2014
Chapter 4: How to Use Assetic for Image Optimization with Twig Functions | 20
7
8
functions:
jpegoptim: { output: images/*.jpg }
PDF brought to you by
generated on December 3, 2014
Chapter 4: How to Use Assetic for Image Optimization with Twig Functions | 21
Chapter 5
How to Apply an Assetic Filter to a specific File
Extension
Assetic filters can be applied to individual files, groups of files or even, as you'll see here, files that have
a specific extension. To show you how to handle each option, suppose that you want to use Assetic's
CoffeeScript filter, which compiles CoffeeScript files into JavaScript.
The main configuration is just the paths to coffee, node and node_modules. An example configuration
might look like this:
Listing 5-1
1 # app/config/config.yml
2 assetic:
3
filters:
4
coffee:
5
bin:
/usr/bin/coffee
6
node:
/usr/bin/node
7
node_paths: [/usr/lib/node_modules/]
Filter a single File
You can now serve up a single CoffeeScript file as JavaScript from within your templates:
Listing 5-2
1 {% javascripts '@AcmeFooBundle/Resources/public/js/example.coffee' filter='coffee' %}
2
<script src="{{ asset_url }}" type="text/javascript"></script>
3 {% endjavascripts %}
This is all that's needed to compile this CoffeeScript file and serve it as the compiled JavaScript.
Filter multiple Files
You can also combine multiple CoffeeScript files into a single output file:
PDF brought to you by
generated on December 3, 2014
Chapter 5: How to Apply an Assetic Filter to a specific File Extension | 22
Listing 5-3
1 {% javascripts '@AcmeFooBundle/Resources/public/js/example.coffee'
2
'@AcmeFooBundle/Resources/public/js/another.coffee'
3
filter='coffee' %}
4
<script src="{{ asset_url }}" type="text/javascript"></script>
5 {% endjavascripts %}
Both the files will now be served up as a single file compiled into regular JavaScript.
Filtering Based on a File Extension
One of the great advantages of using Assetic is reducing the number of asset files to lower HTTP requests.
In order to make full use of this, it would be good to combine all your JavaScript and CoffeeScript files
together since they will ultimately all be served as JavaScript. Unfortunately just adding the JavaScript
files to the files to be combined as above will not work as the regular JavaScript files will not survive the
CoffeeScript compilation.
This problem can be avoided by using the apply_to option in the config, which allows you to specify
that a filter should always be applied to particular file extensions. In this case you can specify that the
coffee filter is applied to all .coffee files:
Listing 5-4
# app/config/config.yml
assetic:
filters:
coffee:
bin:
node:
node_paths:
apply_to:
/usr/bin/coffee
/usr/bin/node
[/usr/lib/node_modules/]
"\.coffee$"
With this, you no longer need to specify the coffee filter in the template. You can also list regular
JavaScript files, all of which will be combined and rendered as a single JavaScript file (with only the
.coffee files being run through the CoffeeScript filter):
Listing 5-5
1 {% javascripts '@AcmeFooBundle/Resources/public/js/example.coffee'
2
'@AcmeFooBundle/Resources/public/js/another.coffee'
3
'@AcmeFooBundle/Resources/public/js/regular.js' %}
4
<script src="{{ asset_url }}" type="text/javascript"></script>
5 {% endjavascripts %}
PDF brought to you by
generated on December 3, 2014
Chapter 5: How to Apply an Assetic Filter to a specific File Extension | 23
Chapter 6
How to Install 3rd Party Bundles
Most bundles provide their own installation instructions. However, the basic steps for installing a bundle
are the same:
• A) Add Composer Dependencies
• B) Enable the Bundle
• C) Configure the Bundle
A) Add Composer Dependencies
Dependencies are managed with Composer, so if Composer is new to you, learn some basics in their
documentation1. This has 2 steps:
1) Find out the Name of the Bundle on Packagist
The README for a bundle (e.g. FOSUserBundle2) usually tells you its name (e.g. friendsofsymfony/
user-bundle). If it doesn't, you can search for the library on the Packagist.org3 site.
Looking for bundles? Try searching at KnpBundles.com4: the unofficial archive of Symfony
Bundles.
2) Install the Bundle via Composer
Now that you know the package name, you can install it via Composer:
1. http://getcomposer.org/doc/00-intro.md
2. https://github.com/FriendsOfSymfony/FOSUserBundle
3. https://packagist.org
4. http://knpbundles.com/
PDF brought to you by
generated on December 3, 2014
Chapter 6: How to Install 3rd Party Bundles | 24
1 $ php composer.phar require friendsofsymfony/user-bundle
Listing 6-1
This will choose the best version for your project, add it to composer.json and download the library into
the vendor/ directory. If you need a specific version, add a : and the version right after the library name
(see composer require5).
B) Enable the Bundle
At this point, the bundle is installed in your Symfony project (in vendor/friendsofsymfony/) and the
autoloader recognizes its classes. The only thing you need to do now is register the bundle in AppKernel:
Listing 6-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// app/AppKernel.php
// ...
class AppKernel extends Kernel
{
// ...
public function registerBundles()
{
$bundles = array(
// ...,
new FOS\UserBundle\FOSUserBundle(),
);
// ...
}
}
C) Configure the Bundle
It's pretty common for a bundle to need some additional setup or configuration in app/config/
config.yml. The bundle's documentation will tell you about the configuration, but you can also get a
reference of the bundle's config via the config:dump-reference command.
For instance, in order to look the reference of the assetic config you can use this:
Listing 6-3
1 $ app/console config:dump-reference AsseticBundle
or this:
Listing 6-4
1 $ app/console config:dump-reference assetic
The output will look like this:
Listing 6-5
1 assetic:
2
debug:
%kernel.debug%
5. https://getcomposer.org/doc/03-cli.md#require
PDF brought to you by
generated on December 3, 2014
Chapter 6: How to Install 3rd Party Bundles | 25
3
4
5
6
7
8
9
10
11
use_controller:
enabled:
profiler:
read_from:
write_to:
java:
node:
node_paths:
# ...
%kernel.debug%
false
%kernel.root_dir%/../web
%assetic.read_from%
/usr/bin/java
/usr/local/bin/node
[]
Other Setup
At this point, check the README file of your brand new bundle to see what to do next. Have fun!
PDF brought to you by
generated on December 3, 2014
Chapter 6: How to Install 3rd Party Bundles | 26
Chapter 7
Best Practices for Reusable Bundles
There are 2 types of bundles:
• Application-specific bundles: only used to build your application;
• Reusable bundles: meant to be shared across many projects.
This article is all about how to structure your reusable bundles so that they're easy to configure and
extend. Many of these recommendations do not apply to application bundles because you'll want to keep
those as simple as possible. For application bundles, just follow the practices shown throughout the book
and cookbook.
The best practices for application-specific bundles are discussed in The Symfony Framework Best Practices.
Bundle Name
A bundle is also a PHP namespace. The namespace must follow the technical interoperability standards1
for PHP 5.3 namespaces and class names: it starts with a vendor segment, followed by zero or more
category segments, and it ends with the namespace short name, which must end with a Bundle suffix.
A namespace becomes a bundle as soon as you add a bundle class to it. The bundle class name must
follow these simple rules:
•
•
•
•
Use only alphanumeric characters and underscores;
Use a CamelCased name;
Use a descriptive and short name (no more than 2 words);
Prefix the name with the concatenation of the vendor (and optionally the category
namespaces);
• Suffix the name with Bundle.
Here are some valid bundle namespaces and class names:
1. http://www.php-fig.org/psr/psr-0/
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 27
Namespace
Bundle Class Name
Acme\Bundle\BlogBundle
AcmeBlogBundle
Acme\Bundle\Social\BlogBundle AcmeSocialBlogBundle
Acme\BlogBundle
AcmeBlogBundle
By convention, the getName() method of the bundle class should return the class name.
If you share your bundle publicly, you must use the bundle class name as the name of the
repository (AcmeBlogBundle and not BlogBundle for instance).
Symfony2 core Bundles do not prefix the Bundle class with Symfony and always add a Bundle subnamespace; for example: FrameworkBundle2.
Each bundle has an alias, which is the lower-cased short version of the bundle name using underscores
(acme_hello for AcmeHelloBundle, or acme_social_blog for Acme\Social\BlogBundle for instance).
This alias is used to enforce uniqueness within a bundle (see below for some usage examples).
Directory Structure
The basic directory structure of a HelloBundle bundle must read as follows:
Listing 7-1
1 XXX/...
2
HelloBundle/
3
HelloBundle.php
4
Controller/
5
Resources/
6
meta/
7
LICENSE
8
config/
9
doc/
10
index.rst
11
translations/
12
views/
13
public/
14
Tests/
The XXX directory(ies) reflects the namespace structure of the bundle.
The following files are mandatory:
• HelloBundle.php;
• Resources/meta/LICENSE: The full license for the code;
• Resources/doc/index.rst: The root file for the Bundle documentation.
These conventions ensure that automated tools can rely on this default structure to work.
2. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/FrameworkBundle.html
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 28
The depth of sub-directories should be kept to the minimal for most used classes and files (2 levels at a
maximum). More levels can be defined for non-strategic, less-used files.
The bundle directory is read-only. If you need to write temporary files, store them under the cache/ or
log/ directory of the host application. Tools can generate files in the bundle directory structure, but only
if the generated files are going to be part of the repository.
The following classes and files have specific emplacements:
Type
Directory
Commands
Command/
Controllers
Controller/
Service Container Extensions DependencyInjection/
Event Listeners
EventListener/
Configuration
Resources/config/
Web Resources
Resources/public/
Translation files
Resources/translations/
Templates
Resources/views/
Unit and Functional Tests
Tests/
When building a reusable bundle, model classes should be placed in the Model namespace. See
How to Provide Model Classes for several Doctrine Implementations for how to handle the mapping
with a compiler pass.
Classes
The bundle directory structure is used as the namespace hierarchy. For instance, a HelloController
controller is stored in Bundle/HelloBundle/Controller/HelloController.php and the fully qualified
class name is Bundle\HelloBundle\Controller\HelloController.
All classes and files must follow the Symfony2 coding standards.
Some classes should be seen as facades and should be as short as possible, like Commands, Helpers,
Listeners, and Controllers.
Classes that connect to the event dispatcher should be suffixed with Listener.
Exceptions classes should be stored in an Exception sub-namespace.
Vendors
A bundle must not embed third-party PHP libraries. It should rely on the standard Symfony2 autoloading
instead.
A bundle should not embed third-party libraries written in JavaScript, CSS, or any other language.
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 29
Tests
A bundle should come with a test suite written with PHPUnit and stored under the Tests/ directory.
Tests should follow the following principles:
• The test suite must be executable with a simple phpunit command run from a sample
application;
• The functional tests should only be used to test the response output and some profiling
information if you have some;
• The tests should cover at least 95% of the code base.
A test suite must not contain AllTests.php scripts, but must rely on the existence of a
phpunit.xml.dist file.
Documentation
All classes and functions must come with full PHPDoc.
Extensive documentation should also be provided in the reStructuredText format, under the Resources/
doc/ directory; the Resources/doc/index.rst file is the only mandatory file and must be the entry point
for the documentation.
Installation Instructions
In order to ease the installation of third-party bundles, consider using the following standardized
instructions in your README.md file.
Listing 7-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Installation
============
Step 1: Download the Bundle
--------------------------Open a command console, enter your project directory and execute the
following command to download the latest stable version of this bundle:
```bash
$ composer require <package-name> "~1"
```
This command requires you to have Composer installed globally, as explained
in the [installation chapter](https://getcomposer.org/doc/00-intro.md)
of the Composer documentation.
Step 2: Enable the Bundle
------------------------Then, enable the bundle by adding the following line in the `app/AppKernel.php`
file of your project:
```php
<?php
// app/AppKernel.php
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 30
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
// ...
class AppKernel extends Kernel
{
public function registerBundles()
{
$bundles = array(
// ...
new <vendor>\<bundle-name>\<bundle-long-name>(),
);
// ...
}
// ...
}
```
This template assumes that your bundle is in its 1.x version. If not, change the "~1" installation version
accordingly ("~2", "~3", etc.)
Optionally, you can add more installation steps (Step 3, Step 4, etc.) to explain other required installation
tasks, such as registering routes or dumping assets.
Routing
If the bundle provides routes, they must be prefixed with the bundle alias. For an AcmeBlogBundle for
instance, all routes must be prefixed with acme_blog_.
Templates
If a bundle provides templates, they must use Twig. A bundle must not provide a main layout, except if
it provides a full working application.
Translation Files
If a bundle provides message translations, they must be defined in the XLIFF format; the domain should
be named after the bundle name (bundle.hello).
A bundle must not override existing messages from another bundle.
Configuration
To provide more flexibility, a bundle can provide configurable settings by using the Symfony2 built-in
mechanisms.
For simple configuration settings, rely on the default parameters entry of the Symfony2 configuration.
Symfony2 parameters are simple key/value pairs; a value being any valid PHP value. Each parameter
name should start with the bundle alias, though this is just a best-practice suggestion. The rest of the
parameter name will use a period (.) to separate different parts (e.g. acme_hello.email.from).
The end user can provide values in any configuration file:
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 31
Listing 7-3
1 # app/config/config.yml
2 parameters:
3
acme_hello.email.from: [email protected]
Retrieve the configuration parameters in your code from the container:
Listing 7-4
1 $container->getParameter('acme_hello.email.from');
Even if this mechanism is simple enough, you are highly encouraged to use the semantic configuration
described in the cookbook.
If you are defining services, they should also be prefixed with the bundle alias.
Custom Validation Constraints
Starting with Symfony 2.5, a new Validation API was introduced. In fact, there are 3 modes, which the
user can configure in their project:
• 2.4: the original 2.4 and earlier validation API;
• 2.5: the new 2.5 and later validation API;
• 2.5-BC: the new 2.5 API with a backwards-compatible layer so that the 2.4 API still works.
This is only available in PHP 5.3.9+.
As a bundle author, you'll want to support both API's, since some users may still be using the 2.4 API.
Specifically, if your bundle adds a violation directly to the ExecutionContext3 (e.g. like in a custom
validation constraint), you'll need to check for which API is being used. The following code, would work
for all users:
Listing 7-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
use Symfony\Component\Validator\ConstraintValidator;
use Symfony\Component\Validator\Constraint;
use Symfony\Component\Validator\Context\ExecutionContextInterface;
// ...
class ContainsAlphanumericValidator extends ConstraintValidator
{
public function validate($value, Constraint $constraint)
{
if ($this->context instanceof ExecutionContextInterface) {
// the 2.5 API
$this->context->buildViolation($constraint->message)
->setParameter('%string%', $value)
->addViolation();
);
} else {
// the 2.4 API
$this->context->addViolation(
$constraint->message,
array('%string%' => $value)
);
3. http://api.symfony.com/master/Symfony/Component/Validator/Context/ExecutionContext.html
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 32
22
23
24 }
}
}
Learn more from the Cookbook
• How to Load Service Configuration inside a Bundle
PDF brought to you by
generated on December 3, 2014
Chapter 7: Best Practices for Reusable Bundles | 33
Chapter 8
How to Use Bundle Inheritance to Override
Parts of a Bundle
When working with third-party bundles, you'll probably come across a situation where you want to
override a file in that third-party bundle with a file in one of your own bundles. Symfony gives you a very
convenient way to override things like controllers, templates, and other files in a bundle's Resources/
directory.
For example, suppose that you're installing the FOSUserBundle1, but you want to override its base
layout.html.twig template, as well as one of its controllers. Suppose also that you have your own
AcmeUserBundle where you want the overridden files to live. Start by registering the FOSUserBundle as
the "parent" of your bundle:
Listing 8-1
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/UserBundle/AcmeUserBundle.php
namespace Acme\UserBundle;
use Symfony\Component\HttpKernel\Bundle\Bundle;
class AcmeUserBundle extends Bundle
{
public function getParent()
{
return 'FOSUserBundle';
}
}
By making this simple change, you can now override several parts of the FOSUserBundle simply by
creating a file with the same name.
Despite the method name, there is no parent/child relationship between the bundles, it is just a
way to extend and override an existing bundle.
1. https://github.com/friendsofsymfony/fosuserbundle
PDF brought to you by
generated on December 3, 2014
Chapter 8: How to Use Bundle Inheritance to Override Parts of a Bundle | 34
Overriding Controllers
Suppose you want to add some functionality to the registerAction of a RegistrationController that
lives inside FOSUserBundle. To do so, just create your own RegistrationController.php file, override
the bundle's original method, and change its functionality:
Listing 8-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/Acme/UserBundle/Controller/RegistrationController.php
namespace Acme\UserBundle\Controller;
use FOS\UserBundle\Controller\RegistrationController as BaseController;
class RegistrationController extends BaseController
{
public function registerAction()
{
$response = parent::registerAction();
// ... do custom stuff
return $response;
}
}
Depending on how severely you need to change the behavior, you might call
parent::registerAction() or completely replace its logic with your own.
Overriding controllers in this way only works if the bundle refers to the controller using the
standard FOSUserBundle:Registration:register syntax in routes and templates. This is the
best practice.
Overriding Resources: Templates, Routing, etc
Most resources can also be overridden, simply by creating a file in the same location as your parent
bundle.
For example, it's very common to need to override the FOSUserBundle's layout.html.twig template so
that it uses your application's base layout. Since the file lives at Resources/views/layout.html.twig in
the FOSUserBundle, you can create your own file in the same location of AcmeUserBundle. Symfony will
ignore the file that lives inside the FOSUserBundle entirely, and use your file instead.
The same goes for routing files and some other resources.
The overriding of resources only works when you refer to resources with the @FosUserBundle/
Resources/config/routing/security.xml method. If you refer to resources without using the
@BundleName shortcut, they can't be overridden in this way.
Translation and validation files do not work in the same way as described above. Read
"Translations" if you want to learn how to override translations and see "Validation Metadata" for
tricks to override the validation.
PDF brought to you by
generated on December 3, 2014
Chapter 8: How to Use Bundle Inheritance to Override Parts of a Bundle | 35
Chapter 9
How to Override any Part of a Bundle
This document is a quick reference for how to override different parts of third-party bundles.
Templates
For information on overriding templates, see
• Overriding Bundle Templates.
• How to Use Bundle Inheritance to Override Parts of a Bundle
Routing
Routing is never automatically imported in Symfony2. If you want to include the routes from any
bundle, then they must be manually imported from somewhere in your application (e.g. app/config/
routing.yml).
The easiest way to "override" a bundle's routing is to never import it at all. Instead of importing a thirdparty bundle's routing, simply copying that routing file into your application, modify it, and import it
instead.
Controllers
Assuming the third-party bundle involved uses non-service controllers (which is almost always the case),
you can easily override controllers via bundle inheritance. For more information, see How to Use Bundle
Inheritance to Override Parts of a Bundle. If the controller is a service, see the next section on how to
override it.
PDF brought to you by
generated on December 3, 2014
Chapter 9: How to Override any Part of a Bundle | 36
Services & Configuration
In order to override/extend a service, there are two options. First, you can set the parameter holding the
service's class name to your own class by setting it in app/config/config.yml. This of course is only
possible if the class name is defined as a parameter in the service config of the bundle containing the
service. For example, to override the class used for Symfony's translator service, you would override the
translator.class parameter. Knowing exactly which parameter to override may take some research.
For the translator, the parameter is defined and used in the Resources/config/translation.xml file in
the core FrameworkBundle:
Listing 9-1
1 # app/config/config.yml
2 parameters:
3
translator.class: Acme\HelloBundle\Translation\Translator
Secondly, if the class is not available as a parameter, you want to make sure the class is always overridden
when your bundle is used, or you need to modify something beyond just the class name, you should use
a compiler pass:
Listing 9-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/DemoBundle/DependencyInjection/Compiler/OverrideServiceCompilerPass.php
namespace Acme\DemoBundle\DependencyInjection\Compiler;
use Symfony\Component\DependencyInjection\Compiler\CompilerPassInterface;
use Symfony\Component\DependencyInjection\ContainerBuilder;
class OverrideServiceCompilerPass implements CompilerPassInterface
{
public function process(ContainerBuilder $container)
{
$definition = $container->getDefinition('original-service-id');
$definition->setClass('Acme\DemoBundle\YourService');
}
}
In this example you fetch the service definition of the original service, and set its class name to your own
class.
See How to Work with Compiler Passes in Bundles for information on how to use compiler passes. If you
want to do something beyond just overriding the class - like adding a method call - you can only use the
compiler pass method.
Entities & Entity Mapping
Due to the way Doctrine works, it is not possible to override entity mapping of a bundle. However,
if a bundle provides a mapped superclass (such as the User entity in the FOSUserBundle) one can
override attributes and associations. Learn more about this feature and its limitations in the Doctrine
documentation1.
Forms
In order to override a form type, it has to be registered as a service (meaning it is tagged as "form.type").
You can then override it as you would override any service as explained in Services & Configuration.
This, of course, will only work if the type is referred to by its alias rather than being instantiated, e.g.:
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/inheritance-mapping.html#overrides
PDF brought to you by
generated on December 3, 2014
Chapter 9: How to Override any Part of a Bundle | 37
Listing 9-3
1 $builder->add('name', 'custom_type');
rather than:
Listing 9-4
1 $builder->add('name', new CustomType());
Validation Metadata
Symfony loads all validation configuration files from every bundle and combines them into one validation
metadata tree. This means you are able to add new constraints to a property, but you cannot override
them.
To override this, the 3rd party bundle needs to have configuration for validation groups. For instance,
the FOSUserBundle has this configuration. To create your own validation, add the constraints to a new
validation group:
Listing 9-5
1 # src/Acme/UserBundle/Resources/config/validation.yml
2 Fos\UserBundle\Model\User:
3
properties:
4
plainPassword:
5
- NotBlank:
6
groups: [AcmeValidation]
7
- Length:
8
min: 6
9
minMessage: fos_user.password.short
10
groups: [AcmeValidation]
Now, update the FOSUserBundle configuration, so it uses your validation groups instead of the original
ones.
Translations
Translations are not related to bundles, but to domains. That means that you can override the
translations from any translation file, as long as it is in the correct domain.
The last translation file always wins. That means that you need to make sure that the bundle
containing your translations is loaded after any bundle whose translations you're overriding. This
is done in AppKernel.
The file that always wins is the one that is placed in app/Resources/translations, as those files
are always loaded last.
PDF brought to you by
generated on December 3, 2014
Chapter 9: How to Override any Part of a Bundle | 38
Chapter 10
How to Remove the AcmeDemoBundle
The Symfony2 Standard Edition comes with a complete demo that lives inside a bundle called
AcmeDemoBundle. It is a great boilerplate to refer to while starting a project, but you'll probably want to
eventually remove it.
This article uses the AcmeDemoBundle as an example, but you can use these steps to remove any
bundle.
1. Unregister the Bundle in the AppKernel
To disconnect the bundle from the framework, you should remove the bundle from the
AppKernel::registerBundles() method. The bundle is normally found in the $bundles array but the
AcmeDemoBundle is only registered in the development environment and you can find it inside the if
statement below:
Listing 10-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// app/AppKernel.php
// ...
class AppKernel extends Kernel
{
public function registerBundles()
{
$bundles = array(...);
if (in_array($this->getEnvironment(), array('dev', 'test'))) {
// comment or remove this line:
// $bundles[] = new Acme\DemoBundle\AcmeDemoBundle();
// ...
}
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 10: How to Remove the AcmeDemoBundle | 39
2. Remove Bundle Configuration
Now that Symfony doesn't know about the bundle, you need to remove any configuration and routing
configuration inside the app/config directory that refers to the bundle.
2.1 Remove Bundle Routing
The routing for the AcmeDemoBundle can be found in app/config/routing_dev.yml. Remove the
_acme_demo entry at the bottom of this file.
2.2 Remove Bundle Configuration
Some bundles contain configuration in one of the app/config/config*.yml files. Be sure to remove
the related configuration from these files. You can quickly spot bundle configuration by looking for a
acme_demo (or whatever the name of the bundle is, e.g. fos_user for the FOSUserBundle) string in the
configuration files.
The AcmeDemoBundle doesn't have configuration. However, the bundle is used in the configuration for
the app/config/security.yml file. You can use it as a boilerplate for your own security, but you can
also remove everything: it doesn't matter to Symfony if you remove it or not.
3. Remove the Bundle from the Filesystem
Now you have removed every reference to the bundle in your application, you should remove the bundle
from the filesystem. The bundle is located in the src/Acme/DemoBundle directory. You should remove
this directory and you can remove the Acme directory as well.
If you don't know the location of a bundle, you can use the getPath()1 method to get the path of
the bundle:
Listing 10-2
1 echo $this->container->get('kernel')->getBundle('AcmeDemoBundle')->getPath();
3.1 Remove Bundle Assets
Remove the assets of the bundle in the web/ directory (e.g. web/bundles/acmedemo for the
AcmeDemoBundle).
4. Remove Integration in other Bundles
This doesn't apply to the AcmeDemoBundle - no other bundles depend on it, so you can skip this
step.
Some bundles rely on other bundles, if you remove one of the two, the other will probably not work. Be
sure that no other bundles, third party or self-made, rely on the bundle you are about to remove.
1. http://api.symfony.com/master/Symfony/Component/HttpKernel/Bundle/BundleInterface.html#getPath()
PDF brought to you by
generated on December 3, 2014
Chapter 10: How to Remove the AcmeDemoBundle | 40
If one bundle relies on another, in most cases it means that it uses some services from the
bundle. Searching for the bundle alias string may help you spot them (e.g. acme_demo for bundles
depending on AcmeDemoBundle).
If a third party bundle relies on another bundle, you can find that bundle mentioned in the
composer.json file included in the bundle directory.
PDF brought to you by
generated on December 3, 2014
Chapter 10: How to Remove the AcmeDemoBundle | 41
Chapter 11
How to Load Service Configuration inside a
Bundle
In Symfony, you'll find yourself using many services. These services can be registered in the app/config
directory of your application. But when you want to decouple the bundle for use in other projects, you
want to include the service configuration in the bundle itself. This article will teach you how to do that.
Creating an Extension Class
In order to load service configuration, you have to create a Dependency Injection Extension for your
bundle. This class has some conventions in order to be detected automatically. But you'll later see how
you can change it to your own preferences. By default, the Extension has to comply with the following
conventions:
• It has to live in the DependencyInjection namespace of the bundle;
• The name is equal to the bundle name with the Bundle suffix replaced by Extension (e.g. the
Extension class of AcmeHelloBundle would be called AcmeHelloExtension).
The Extension class should implement the ExtensionInterface1, but usually you would simply extend
the Extension2 class:
Listing 11-1
1
2
3
4
5
6
7
8
9
// src/Acme/HelloBundle/DependencyInjection/AcmeHelloExtension.php
namespace Acme\HelloBundle\DependencyInjection;
use Symfony\Component\HttpKernel\DependencyInjection\Extension;
use Symfony\Component\DependencyInjection\ContainerBuilder;
class AcmeHelloExtension extends Extension
{
public function load(array $configs, ContainerBuilder $container)
1. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/ExtensionInterface.html
2. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/Extension.html
PDF brought to you by
generated on December 3, 2014
Chapter 11: How to Load Service Configuration inside a Bundle | 42
10
11
12
13 }
{
// ... you'll load the files here later
}
Manually Registering an Extension Class
When not following the conventions, you will have to manually register your extension. To do this,
you should override the Bundle::getContainerExtension()3 method to return the instance of the
extension:
Listing 11-2
1
2
3
4
5
6
7
8
9
10
// ...
use Acme\HelloBundle\DependencyInjection\UnconventionalExtensionClass;
class AcmeHelloBundle extends Bundle
{
public function getContainerExtension()
{
return new UnconventionalExtensionClass();
}
}
Since the new Extension class name doesn't follow the naming conventions, you should also override
Extension::getAlias()4 to return the correct DI alias. The DI alias is the name used to refer to the
bundle in the container (e.g. in the app/config/config.yml file). By default, this is done by removing
the Extension prefix and converting the class name to underscores (e.g. AcmeHelloExtension's DI alias
is acme_hello).
Using the load() Method
In the load() method, all services and parameters related to this extension will be loaded. This method
doesn't get the actual container instance, but a copy. This container only has the parameters from the
actual container. After loading the services and parameters, the copy will be merged into the actual
container, to ensure all services and parameters are also added to the actual container.
In the load() method, you can use PHP code to register service definitions, but it is more common if you
put these definitions in a configuration file (using the Yaml, XML or PHP format). Luckily, you can use
the file loaders in the extension!
For instance, assume you have a file called services.xml in the Resources/config directory of your
bundle, your load method looks like:
Listing 11-3
1
2
3
4
5
6
7
8
9
use Symfony\Component\DependencyInjection\Loader\XmlFileLoader;
use Symfony\Component\Config\FileLocator;
// ...
public function load(array $configs, ContainerBuilder $container)
{
$loader = new XmlFileLoader(
$container,
new FileLocator(__DIR__.'/../Resources/config')
3. http://api.symfony.com/master/Symfony/Component/HttpKernel/Bundle/Bundle.html#build()
4. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/Extension.html#getAlias()
PDF brought to you by
generated on December 3, 2014
Chapter 11: How to Load Service Configuration inside a Bundle | 43
10
11
12 }
);
$loader->load('services.xml');
Other available loaders are the YamlFileLoader, PhpFileLoader and IniFileLoader.
The IniFileLoader can only be used to load parameters and it can only load them as strings.
Using Configuration to Change the Services
The Extension is also the class that handles the configuration for that particular bundle (e.g. the
configuration in app/config/config.yml). To read more about it, see the "How to Create Friendly
Configuration for a Bundle" article.
PDF brought to you by
generated on December 3, 2014
Chapter 11: How to Load Service Configuration inside a Bundle | 44
Chapter 12
How to Create Friendly Configuration for a
Bundle
If you open your application configuration file (usually app/config/config.yml), you'll see a number of
different configuration "namespaces", such as framework, twig and doctrine. Each of these configures
a specific bundle, allowing you to configure things at a high level and then let the bundle make all the
low-level, complex changes based on your settings.
For example, the following tells the FrameworkBundle to enable the form integration, which involves the
definition of quite a few services as well as integration of other related components:
Listing 12-1
1 framework:
2
form: true
Using Parameters to Configure your Bundle
If you don't have plans to share your bundle between projects, it doesn't make sense to use this
more advanced way of configuration. Since you use the bundle only in one project, you can just
change the service configuration each time.
If you do want to be able to configure something from within config.yml, you can always create a
parameter there and use that parameter somewhere else.
Using the Bundle Extension
The basic idea is that instead of having the user override individual parameters, you let the user
configure just a few, specifically created, options. As the bundle developer, you then parse through that
configuration and load correct services and parameters inside an "Extension" class.
As an example, imagine you are creating a social bundle, which provides integration with Twitter and
such. To be able to reuse your bundle, you have to make the client_id and client_secret variables
configurable. Your bundle configuration would look like:
PDF brought to you by
generated on December 3, 2014
Chapter 12: How to Create Friendly Configuration for a Bundle | 45
Listing 12-2
1 # app/config/config.yml
2 acme_social:
3
twitter:
4
client_id: 123
5
client_secret: $ecret
Read more about the extension in How to Load Service Configuration inside a Bundle.
If a bundle provides an Extension class, then you should not generally override any service
container parameters from that bundle. The idea is that if an Extension class is present, every
setting that should be configurable should be present in the configuration made available by that
class. In other words, the extension class defines all the public configuration settings for which
backward compatibility will be maintained.
For parameter handling within a Dependency Injection class see Using Parameters within a Dependency
Injection Class.
Processing the $configs Array
First things first, you have to create an extension class as explained in How to Load Service Configuration
inside a Bundle.
Whenever a user includes the acme_social key (which is the DI alias) in a configuration file, the
configuration under it is added to an array of configurations and passed to the load() method of your
extension (Symfony2 automatically converts XML and YAML to an array).
For the configuration example in the previous section, the array passed to your load() method will look
like this:
Listing 12-3
1 array(
2
array(
3
'twitter' => array(
4
'client_id' => 123,
5
'client_secret' => '$ecret',
6
),
7
),
8 )
Notice that this is an array of arrays, not just a single flat array of the configuration values. This is
intentional, as it allows Symfony to parse several configuration resources. For example, if acme_social
appears in another configuration file - say config_dev.yml - with different values beneath it, the
incoming array might look like this:
Listing 12-4
1 array(
2
// values from config.yml
3
array(
4
'twitter' => array(
5
'client_id' => 123,
6
'client_secret' => '$secret',
7
),
8
),
9
// values from config_dev.yml
PDF brought to you by
generated on December 3, 2014
Chapter 12: How to Create Friendly Configuration for a Bundle | 46
10
11
12
13
14
15 )
array(
'twitter' => array(
'client_id' => 456,
),
),
The order of the two arrays depends on which one is set first.
But don't worry! Symfony's Config component will help you merge these values, provide defaults and
give the user validation errors on bad configuration. Here's how it works. Create a Configuration
class in the DependencyInjection directory and build a tree that defines the structure of your bundle's
configuration.
The Configuration class to handle the sample configuration looks like:
Listing 12-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/Acme/SocialBundle/DependencyInjection/Configuration.php
namespace Acme\SocialBundle\DependencyInjection;
use Symfony\Component\Config\Definition\Builder\TreeBuilder;
use Symfony\Component\Config\Definition\ConfigurationInterface;
class Configuration implements ConfigurationInterface
{
public function getConfigTreeBuilder()
{
$treeBuilder = new TreeBuilder();
$rootNode = $treeBuilder->root('acme_social');
$rootNode
->children()
->arrayNode('twitter')
->children()
->integerNode('client_id')->end()
->scalarNode('client_secret')->end()
->end()
->end() // twitter
->end()
;
return $treeBuilder;
}
}
The Configuration class can be much more complicated than shown here, supporting "prototype" nodes,
advanced validation, XML-specific normalization and advanced merging. You can read more about this in the
Config component documentation. You can also see it in action by checking out some of the core Configuration
classes, such as the one from the FrameworkBundle Configuration1 or the TwigBundle Configuration2.
This class can now be used in your load() method to merge configurations and force validation (e.g. if
an additional option was passed, an exception will be thrown):
Listing 12-6
1 public function load(array $configs, ContainerBuilder $container)
2 {
1. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/DependencyInjection/Configuration.php
2. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/TwigBundle/DependencyInjection/Configuration.php
PDF brought to you by
generated on December 3, 2014
Chapter 12: How to Create Friendly Configuration for a Bundle | 47
3
4
5
6
7 }
$configuration = new Configuration();
$config = $this->processConfiguration($configuration, $configs);
// ...
The processConfiguration() method uses the configuration tree you've defined in the Configuration
class to validate, normalize and merge all of the configuration arrays together.
Instead of calling processConfiguration() in your extension each time you provide some
configuration options, you might want to use the ConfigurableExtension3 to do this
automatically for you:
Listing 12-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/HelloBundle/DependencyInjection/AcmeHelloExtension.php
namespace Acme\HelloBundle\DependencyInjection;
use Symfony\Component\DependencyInjection\ContainerBuilder;
use Symfony\Component\HttpKernel\DependencyInjection\ConfigurableExtension;
class AcmeHelloExtension extends ConfigurableExtension
{
// note that this method is called loadInternal and not load
protected function loadInternal(array $mergedConfig, ContainerBuilder $container)
{
// ...
}
}
This class uses the getConfiguration() method to get the Configuration instance, you should
override it if your Configuration class is not called Configuration or if it is not placed in the same
namespace as the extension.
Processing the Configuration yourself
Using the Config component is fully optional. The load() method gets an array of configuration
values. You can simply parse these arrays yourself (e.g. by overriding configurations and using
isset4 to check for the existence of a value). Be aware that it'll be very hard to support XML.
Listing 12-8
1 public function load(array $configs, ContainerBuilder $container)
2 {
3
$config = array();
4
// let resources override the previous set value
5
foreach ($configs as $subConfig) {
6
$config = array_merge($config, $subConfig);
7
}
8
9
// ... now use the flat $config array
10 }
3. http://api.symfony.com/master/Symfony/Component/HttpKernel/DependencyInjection/ConfigurableExtension.html
4. http://php.net/manual/en/function.isset.php
PDF brought to you by
generated on December 3, 2014
Chapter 12: How to Create Friendly Configuration for a Bundle | 48
Modifying the Configuration of Another Bundle
If you have multiple bundles that depend on each other, it may be useful to allow one Extension class to
modify the configuration passed to another bundle's Extension class, as if the end-developer has actually
placed that configuration in their app/config/config.yml file. This can be achieved using a prepend
extension. For more details, see How to Simplify Configuration of multiple Bundles.
Dump the Configuration
The config:dump-reference command dumps the default configuration of a bundle in the console
using the Yaml format.
As long as your bundle's configuration is located in the standard location
(YourBundle\DependencyInjection\Configuration) and does not require arguments to be passed to
the constructor it will work automatically. If you have something different, your Extension class must
override the Extension::getConfiguration()5 method and return an instance of your Configuration.
Supporting XML
Symfony allows people to provide the configuration in three different formats: Yaml, XML and PHP.
Both Yaml and PHP use the same syntax and are supported by default when using the Config component.
Supporting XML requires you to do some more things. But when sharing your bundle with others, it is
recommended that you follow these steps.
Make your Config Tree ready for XML
The Config component provides some methods by default to allow it to correctly process XML
configuration. See "Normalization" of the component documentation. However, you can do some
optional things as well, this will improve the experience of using XML configuration:
Choosing an XML Namespace
In XML, the XML namespace6 is used to determine which elements belong to the configuration of
a specific bundle. The namespace is returned from the Extension::getNamespace()7 method. By
convention, the namespace is a URL (it doesn't have to be a valid URL nor does it need to exists). By
default, the namespace for a bundle is http://example.org/dic/schema/DI_ALIAS, where DI_ALIAS is
the DI alias of the extension. You might want to change this to a more professional URL:
Listing 12-9
1
2
3
4
5
6
7
8
9
10
// src/Acme/HelloBundle/DependencyInjection/AcmeHelloExtension.php
// ...
class AcmeHelloExtension extends Extension
{
// ...
public function getNamespace()
{
return 'http://acme_company.com/schema/dic/hello';
5. http://api.symfony.com/master/Symfony/Component/HttpKernel/DependencyInjection/Extension.html#getConfiguration()
6. http://en.wikipedia.org/wiki/XML_namespace
7. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/Extension.html#getNamespace()
PDF brought to you by
generated on December 3, 2014
Chapter 12: How to Create Friendly Configuration for a Bundle | 49
11
12 }
}
Providing an XML Schema
XML has a very useful feature called XML schema8. This allows you to describe all possible elements and
attributes and their values in an XML Schema Definition (an xsd file). This XSD file is used by IDEs for
auto completion and it is used by the Config component to validate the elements.
In order to use the schema, the XML configuration file must provide an xsi:schemaLocation attribute
pointing to the XSD file for a certain XML namespace. This location always starts with the XML
namespace. This XML namespace is then replaced with the XSD validation base path returned from
Extension::getXsdValidationBasePath()9 method. This namespace is then followed by the rest of the
path from the base path to the file itself.
By convention, the XSD file lives in the Resources/config/schema, but you can place it anywhere you
like. You should return this path as the base path:
Listing 12-10
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/HelloBundle/DependencyInjection/AcmeHelloExtension.php
// ...
class AcmeHelloExtension extends Extension
{
// ...
public function getXsdValidationBasePath()
{
return __DIR__.'/../Resources/config/schema';
}
}
Assume the XSD file is called hello-1.0.xsd, the schema location will be http://acme_company.com/
schema/dic/hello/hello-1.0.xsd:
Listing 12-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
<!-- app/config/config.xml -->
<?xml version="1.0" ?>
<container xmlns="http://symfony.com/schema/dic/services"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:acme-hello="http://acme_company.com/schema/dic/hello"
xsi:schemaLocation="http://acme_company.com/schema/dic/hello http://acme_company.com/
schema/dic/hello/hello-1.0.xsd">
<acme-hello:config>
<!-- ... -->
</acme-hello:config>
<!-- ... -->
</container>
8. http://en.wikipedia.org/wiki/XML_schema
9. http://api.symfony.com/master/Symfony/Component/DependencyInjection/ExtensionInterface.html#getXsdValidationBasePath()
PDF brought to you by
generated on December 3, 2014
Chapter 12: How to Create Friendly Configuration for a Bundle | 50
Chapter 13
How to Simplify Configuration of multiple
Bundles
When building reusable and extensible applications, developers are often faced with a choice: either
create a single large bundle or multiple smaller bundles. Creating a single bundle has the drawback that
it's impossible for users to choose to remove functionality they are not using. Creating multiple bundles
has the drawback that configuration becomes more tedious and settings often need to be repeated for
various bundles.
Using the below approach, it is possible to remove the disadvantage of the multiple bundle approach by
enabling a single Extension to prepend the settings for any bundle. It can use the settings defined in the
app/config/config.yml to prepend settings just as if they would have been written explicitly by the user
in the application configuration.
For example, this could be used to configure the entity manager name to use in multiple bundles. Or it
can be used to enable an optional feature that depends on another bundle being loaded as well.
To give an Extension the power to do this, it needs to implement PrependExtensionInterface1:
Listing 13-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// src/Acme/HelloBundle/DependencyInjection/AcmeHelloExtension.php
namespace Acme\HelloBundle\DependencyInjection;
use Symfony\Component\HttpKernel\DependencyInjection\Extension;
use Symfony\Component\DependencyInjection\Extension\PrependExtensionInterface;
use Symfony\Component\DependencyInjection\ContainerBuilder;
class AcmeHelloExtension extends Extension implements PrependExtensionInterface
{
// ...
public function prepend(ContainerBuilder $container)
{
// ...
}
}
1. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/PrependExtensionInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 13: How to Simplify Configuration of multiple Bundles | 51
Inside the prepend()2 method, developers have full access to the ContainerBuilder3 instance just
before the load()4 method is called on each of the registered bundle Extensions. In order to prepend
settings to a bundle extension developers can use the prependExtensionConfig()5 method on the
ContainerBuilder6 instance. As this method only prepends settings, any other settings done explicitly
inside the app/config/config.yml would override these prepended settings.
The following example illustrates how to prepend a configuration setting in multiple bundles as well as
disable a flag in multiple bundles in case a specific other bundle is not registered:
Listing 13-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
public function prepend(ContainerBuilder $container)
{
// get all bundles
$bundles = $container->getParameter('kernel.bundles');
// determine if AcmeGoodbyeBundle is registered
if (!isset($bundles['AcmeGoodbyeBundle'])) {
// disable AcmeGoodbyeBundle in bundles
$config = array('use_acme_goodbye' => false);
foreach ($container->getExtensions() as $name => $extension) {
switch ($name) {
case 'acme_something':
case 'acme_other':
// set use_acme_goodbye to false in the config of acme_something and
acme_other
// note that if the user manually configured use_acme_goodbye to true
in the
// app/config/config.yml then the setting would in the end be true and
not false
$container->prependExtensionConfig($name, $config);
break;
}
}
}
// process the configuration of AcmeHelloExtension
$configs = $container->getExtensionConfig($this->getAlias());
// use the Configuration class to generate a config array with the settings
"acme_hello"
$config = $this->processConfiguration(new Configuration(), $configs);
// check if entity_manager_name is set in the "acme_hello" configuration
if (isset($config['entity_manager_name'])) {
// prepend the acme_something settings with the entity_manager_name
$config = array('entity_manager_name' => $config['entity_manager_name']);
$container->prependExtensionConfig('acme_something', $config);
}
}
The above would be the equivalent of writing the following into the app/config/config.yml in case
AcmeGoodbyeBundle is not registered and the entity_manager_name setting for acme_hello is set to
non_default:
Listing 13-3
1 # app/config/config.yml
2 acme_something:
2. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/PrependExtensionInterface.html#prepend()
3. http://api.symfony.com/master/Symfony/Component/DependencyInjection/ContainerBuilder.html
4. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Extension/ExtensionInterface.html#load()
5. http://api.symfony.com/master/Symfony/Component/DependencyInjection/ContainerBuilder.html#prependExtensionConfig()
6. http://api.symfony.com/master/Symfony/Component/DependencyInjection/ContainerBuilder.html
PDF brought to you by
generated on December 3, 2014
Chapter 13: How to Simplify Configuration of multiple Bundles | 52
3
# ...
4
use_acme_goodbye: false
5
entity_manager_name: non_default
6
7 acme_other:
8
# ...
9
use_acme_goodbye: false
PDF brought to you by
generated on December 3, 2014
Chapter 13: How to Simplify Configuration of multiple Bundles | 53
Chapter 14
How to Use Varnish to Speed up my Website
Because Symfony's cache uses the standard HTTP cache headers, the Symfony Reverse Proxy can easily
be replaced with any other reverse proxy. Varnish1 is a powerful, open-source, HTTP accelerator capable
of serving cached content quickly and including support for Edge Side Includes.
Trusting Reverse Proxies
For ESI to work correctly and for the X-FORWARDED headers to be used, you need to configure Varnish
as a trusted proxy.
Configuration
As seen previously, Symfony is smart enough to detect whether it talks to a reverse proxy that
understands ESI or not. It works out of the box when you use the Symfony reverse proxy, but you need a
special configuration to make it work with Varnish. Thankfully, Symfony relies on yet another standard
written by Akamai (Edge Architecture2), so the configuration tips in this chapter can be useful even if you
don't use Symfony.
Varnish only supports the src attribute for ESI tags (onerror and alt attributes are ignored).
First, configure Varnish so that it advertises its ESI support by adding a Surrogate-Capability header
to requests forwarded to the backend application:
Listing 14-1
1 sub vcl_recv {
2
// Add a Surrogate-Capability header to announce ESI support.
1. https://www.varnish-cache.org
2. http://www.w3.org/TR/edge-arch
PDF brought to you by
generated on December 3, 2014
Chapter 14: How to Use Varnish to Speed up my Website | 54
3
4 }
set req.http.Surrogate-Capability = "abc=ESI/1.0";
The abc part of the header isn't important unless you have multiple "surrogates" that need to
advertise their capabilities. See Surrogate-Capability Header3 for details.
Then, optimize Varnish so that it only parses the Response contents when there is at least one ESI tag by
checking the Surrogate-Control header that Symfony adds automatically:
Listing 14-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
/* (https://www.varnish-cache.org/docs/4.0/whats-new/
upgrading.html#req-not-available-in-vcl-backend-response) */
sub vcl_backend_response {
// Check for ESI acknowledgement and remove Surrogate-Control header
if (beresp.http.Surrogate-Control ~ "ESI/1.0") {
unset beresp.http.Surrogate-Control;
set beresp.do_esi = true;
}
/* By default Varnish ignores Pragma: nocache
(https://www.varnish-cache.org/docs/4.0/users-guide/
increasing-your-hitrate.html#cache-control)
so in order avoid caching it has to be done explicitly */
if (beresp.http.Pragma ~ "no-cache") {
// https://www.varnish-cache.org/docs/4.0/whats-new/
upgrading.html#hit-for-pass-objects-are-created-using-beresp-uncacheable
set beresp.uncacheable = true;
set beresp.ttl = 120s;
return (deliver);
}
}
Compression with ESI was not supported in Varnish until version 3.0 (read GZIP and Varnish4). If
you're not using Varnish 3.0, put a web server in front of Varnish to perform the compression.
Cache Invalidation
If you want to cache content that changes frequently and still serve the most recent version to users, you
need to invalidate that content. While cache invalidation5 allows you to purge content from your proxy
before it has expired, it adds complexity to your caching setup.
The open source FOSHttpCacheBundle6 takes the pain out of cache invalidation by helping you to
organize your caching and invalidation setup.
3. http://www.w3.org/TR/edge-arch
4. https://www.varnish-cache.org/docs/3.0/phk/gzip.html
5. http://tools.ietf.org/html/rfc2616#section-13.10
6. http://foshttpcachebundle.readthedocs.org/
PDF brought to you by
generated on December 3, 2014
Chapter 14: How to Use Varnish to Speed up my Website | 55
Varnish can be configured to accept a special HTTP PURGE method that will invalidate the cache for a
given resource:
Listing 14-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
/*
Connect to the backend server
on the local machine on port 8080
*/
backend default {
.host = "127.0.0.1";
.port = "8080";
}
sub vcl_recv {
/*
Varnish default behavior doesn't support PURGE.
Match the PURGE request and immediately do a cache lookup,
otherwise Varnish will directly pipe the request to the backend
and bypass the cache
*/
if (req.request == "PURGE") {
return(lookup);
}
}
sub vcl_hit {
// Match PURGE request
if (req.request == "PURGE") {
// Force object expiration for Varnish < 3.0
set obj.ttl = 0s;
// Do an actual purge for Varnish >= 3.0
// purge;
error 200 "Purged";
}
}
sub vcl_miss {
/*
Match the PURGE request and
indicate the request wasn't stored in cache.
*/
if (req.request == "PURGE") {
error 404 "Not purged";
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 14: How to Use Varnish to Speed up my Website | 56
You must protect the PURGE HTTP method somehow to avoid random people purging your cached
data. You can do this by setting up an access list:
Listing 14-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
/*
Connect to the backend server
on the local machine on port 8080
*/
backend default {
.host = "127.0.0.1";
.port = "8080";
}
// ACL's can contain IP's, subnets and hostnames
acl purge {
"localhost";
"192.168.55.0"/24;
}
sub vcl_recv {
// Match PURGE request to avoid cache bypassing
if (req.request == "PURGE") {
// Match client IP to the ACL
if (!client.ip ~ purge) {
// Deny access
error 405 "Not allowed.";
}
// Perform a cache lookup
return(lookup);
}
}
sub vcl_hit {
// Match PURGE request
if (req.request == "PURGE") {
// Force object expiration for Varnish < 3.0
set obj.ttl = 0s;
// Do an actual purge for Varnish >= 3.0
// purge;
error 200 "Purged";
}
}
sub vcl_miss {
// Match PURGE request
if (req.request == "PURGE") {
// Indicate that the object isn't stored in cache
error 404 "Not purged";
}
}
Routing and X-FORWARDED Headers
To ensure that the Symfony Router generates URLs correctly with Varnish, proper `X-Forwarded`
headers must be added so that Symfony is aware of the original port number of the request. Exactly how
this is done depends on your setup. As a simple example, Varnish and your web server are on the same
machine and that Varnish is listening on one port (e.g. 80) and Apache on another (e.g. 8080). In this
situation, Varnish should add the X-Forwarded-Port header so that the Symfony application knows that
the original port number is 80 and not 8080.
PDF brought to you by
generated on December 3, 2014
Chapter 14: How to Use Varnish to Speed up my Website | 57
If this header weren't set properly, Symfony may append 8080 when generating absolute URLs:
Listing 14-5
1 sub vcl_recv {
2
if (req.http.X-Forwarded-Proto == "https" ) {
3
set req.http.X-Forwarded-Port = "443";
4
} else {
5
set req.http.X-Forwarded-Port = "80";
6
}
7 }
Remember to configure framework.trusted_proxies in the Symfony configuration so that Varnish is
seen as a trusted proxy and the X-Forwarded- headers are used.
PDF brought to you by
generated on December 3, 2014
Chapter 14: How to Use Varnish to Speed up my Website | 58
Chapter 15
How to Master and Create new Environments
Every application is the combination of code and a set of configuration that dictates how that code should
function. The configuration may define the database being used, whether or not something should be
cached, or how verbose logging should be. In Symfony, the idea of "environments" is the idea that the
same codebase can be run using multiple different configurations. For example, the dev environment
should use configuration that makes development easy and friendly, while the prod environment should
use a set of configuration optimized for speed.
Different Environments, different Configuration Files
A typical Symfony application begins with three environments: dev, prod, and test. As discussed, each
"environment" simply represents a way to execute the same codebase with different configuration. It
should be no surprise then that each environment loads its own individual configuration file. If you're
using the YAML configuration format, the following files are used:
• for the dev environment: app/config/config_dev.yml
• for the prod environment: app/config/config_prod.yml
• for the test environment: app/config/config_test.yml
This works via a simple standard that's used by default inside the AppKernel class:
Listing 15-1
1
2
3
4
5
6
7
8
9
10
11
12
13
// app/AppKernel.php
// ...
class AppKernel extends Kernel
{
// ...
public function registerContainerConfiguration(LoaderInterface $loader)
{
$loader->load(__DIR__.'/config/config_'.$this->getEnvironment().'.yml');
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 15: How to Master and Create new Environments | 59
As you can see, when Symfony is loaded, it uses the given environment to determine which configuration
file to load. This accomplishes the goal of multiple environments in an elegant, powerful and transparent
way.
Of course, in reality, each environment differs only somewhat from others. Generally, all environments
will share a large base of common configuration. Opening the "dev" configuration file, you can see how
this is accomplished easily and transparently:
Listing 15-2
1 imports:
2
- { resource: config.yml }
3
4 # ...
To share common configuration, each environment's configuration file simply first imports from a central
configuration file (config.yml). The remainder of the file can then deviate from the default configuration
by overriding individual parameters. For example, by default, the web_profiler toolbar is disabled.
However, in the dev environment, the toolbar is activated by modifying the default value in the dev
configuration file:
Listing 15-3
1 # app/config/config_dev.yml
2 imports:
3
- { resource: config.yml }
4
5 web_profiler:
6
toolbar: true
7
# ...
Executing an Application in different Environments
To execute the application in each environment, load up the application using either the app.php (for the
prod environment) or the app_dev.php (for the dev environment) front controller:
Listing 15-4
1 http://localhost/app.php
2 http://localhost/app_dev.php
-> *prod* environment
-> *dev* environment
The given URLs assume that your web server is configured to use the web/ directory of the
application as its root. Read more in Installing Symfony.
If you open up one of these files, you'll quickly see that the environment used by each is explicitly set:
Listing 15-5
1
2
3
4
5
6
// web/app.php
// ...
$kernel = new AppKernel('prod', false);
// ...
As you can see, the prod key specifies that this application will run in the prod environment. A Symfony
application can be executed in any environment by using this code and changing the environment string.
PDF brought to you by
generated on December 3, 2014
Chapter 15: How to Master and Create new Environments | 60
The test environment is used when writing functional tests and is not accessible in the browser
directly via a front controller. In other words, unlike the other environments, there is no
app_test.php front controller file.
Debug Mode
Important, but unrelated to the topic of environments is the false argument as the second
argument to the AppKernel constructor. This specifies whether or not the application should run
in "debug mode". Regardless of the environment, a Symfony application can be run with debug
mode set to true or false. This affects many things in the application, such as whether or not
errors should be displayed or if cache files are dynamically rebuilt on each request. Though not a
requirement, debug mode is generally set to true for the dev and test environments and false
for the prod environment.
Internally, the value of the debug mode becomes the kernel.debug parameter used inside the
service container. If you look inside the application configuration file, you'll see the parameter used,
for example, to turn logging on or off when using the Doctrine DBAL:
Listing 15-6
1 doctrine:
2
dbal:
3
logging: "%kernel.debug%"
4
# ...
As of Symfony 2.3, showing errors or not no longer depends on the debug mode. You'll need to
enable that in your front controller by calling enable()1.
Creating a new Environment
By default, a Symfony application has three environments that handle most cases. Of course, since an
environment is nothing more than a string that corresponds to a set of configuration, creating a new
environment is quite easy.
Suppose, for example, that before deployment, you need to benchmark your application. One way to
benchmark the application is to use near-production settings, but with Symfony's web_profiler enabled.
This allows Symfony to record information about your application while benchmarking.
The best way to accomplish this is via a new environment called, for example, benchmark. Start by
creating a new configuration file:
Listing 15-7
1 # app/config/config_benchmark.yml
2 imports:
3
- { resource: config_prod.yml }
4
5 framework:
6
profiler: { only_exceptions: false }
1. http://api.symfony.com/master/Symfony/Component/Debug/Debug.html#enable()
PDF brought to you by
generated on December 3, 2014
Chapter 15: How to Master and Create new Environments | 61
Due to the way in which parameters are resolved, you cannot use them to build paths in imports
dynamically. This means that something like the following doesn't work:
Listing 15-8
1 # app/config/config.yml
2 imports:
3
- { resource: "%kernel.root_dir%/parameters.yml" }
And with this simple addition, the application now supports a new environment called benchmark.
This new configuration file imports the configuration from the prod environment and modifies it. This
guarantees that the new environment is identical to the prod environment, except for any changes
explicitly made here.
Because you'll want this environment to be accessible via a browser, you should also create a front
controller for it. Copy the web/app.php file to web/app_benchmark.php and edit the environment to be
benchmark:
Listing 15-9
1
2
3
4
5
6
7
// web/app_benchmark.php
// change just this line
$kernel = new AppKernel('benchmark', false);
// ...
The new environment is now accessible via:
Listing 15-10
1 http://localhost/app_benchmark.php
Some environments, like the dev environment, are never meant to be accessed on any deployed
server by the general public. This is because certain environments, for debugging purposes, may
give too much information about the application or underlying infrastructure. To be sure these
environments aren't accessible, the front controller is usually protected from external IP addresses
via the following code at the top of the controller:
Listing 15-11
1 if (!in_array(@$_SERVER['REMOTE_ADDR'], array('127.0.0.1', '::1'))) {
2
die('You are not allowed to access this file. Check '.basename(__FILE__).' for
3 more information.');
}
Environments and the Cache Directory
Symfony takes advantage of caching in many ways: the application configuration, routing configuration,
Twig templates and more are cached to PHP objects stored in files on the filesystem.
By default, these cached files are largely stored in the app/cache directory. However, each environment
caches its own set of files:
Listing 15-12
1 <your-project>/
2 в”њв”Ђ app/
3 в”‚ в”њв”Ђ cache/
PDF brought to you by
generated on December 3, 2014
Chapter 15: How to Master and Create new Environments | 62
4 в”‚ в”‚ в”њв”Ђ dev/
5 в”‚ в”‚ в””в”Ђ prod/
6 в”‚ в”њв”Ђ ...
# cache directory for the *dev* environment
# cache directory for the *prod* environment
Sometimes, when debugging, it may be helpful to inspect a cached file to understand how something
is working. When doing so, remember to look in the directory of the environment you're using (most
commonly dev while developing and debugging). While it can vary, the app/cache/dev directory
includes the following:
• appDevDebugProjectContainer.php - the cached "service container" that represents the
cached application configuration;
• appDevUrlGenerator.php - the PHP class generated from the routing configuration and used
when generating URLs;
• appDevUrlMatcher.php - the PHP class used for route matching - look here to see the
compiled regular expression logic used to match incoming URLs to different routes;
• twig/ - this directory contains all the cached Twig templates.
You can easily change the directory location and name. For more information read the article How
to Override Symfony's default Directory Structure.
Going further
Read the article on How to Set external Parameters in the Service Container.
PDF brought to you by
generated on December 3, 2014
Chapter 15: How to Master and Create new Environments | 63
Chapter 16
How to Override Symfony's default Directory
Structure
Symfony automatically ships with a default directory structure. You can easily override this directory
structure to create your own. The default directory structure is:
Listing 16-1
1
2
3
4
5
6
7
8
9
10
11
12
13
your-project/
в”њв”Ђ app/
в”‚ в”њв”Ђ cache/
в”‚ в”њв”Ђ config/
в”‚ в”њв”Ђ logs/
в”‚ в””в”Ђ ...
в”њв”Ђ src/
в”‚ в””в”Ђ ...
в”њв”Ђ vendor/
в”‚ в””в”Ђ ...
в””в”Ђ web/
в”њв”Ђ app.php
в””в”Ђ ...
Override the cache Directory
You can override the cache directory by overriding the getCacheDir method in the AppKernel class of
you application:
Listing 16-2
1
2
3
4
5
6
7
// app/AppKernel.php
// ...
class AppKernel extends Kernel
{
// ...
PDF brought to you by
generated on December 3, 2014
Chapter 16: How to Override Symfony's default Directory Structure | 64
8
9
10
11
12 }
public function getCacheDir()
{
return $this->rootDir.'/'.$this->environment.'/cache';
}
$this->rootDir is the absolute path to the app directory and $this->environment is the current
environment (i.e. dev). In this case you have changed the location of the cache directory to app/
{environment}/cache.
You should keep the cache directory different for each environment, otherwise some unexpected
behavior may happen. Each environment generates its own cached config files, and so each needs
its own directory to store those cache files.
Override the logs Directory
Overriding the logs directory is the same as overriding the cache directory, the only difference is that
you need to override the getLogDir method:
Listing 16-3
1
2
3
4
5
6
7
8
9
10
11
12
// app/AppKernel.php
// ...
class AppKernel extends Kernel
{
// ...
public function getLogDir()
{
return $this->rootDir.'/'.$this->environment.'/logs';
}
}
Here you have changed the location of the directory to app/{environment}/logs.
Override the web Directory
If you need to rename or move your web directory, the only thing you need to guarantee is that the path
to the app directory is still correct in your app.php and app_dev.php front controllers. If you simply
renamed the directory, you're fine. But if you moved it in some way, you may need to modify the paths
inside these files:
Listing 16-4
1 require_once __DIR__.'/../Symfony/app/bootstrap.php.cache';
2 require_once __DIR__.'/../Symfony/app/AppKernel.php';
You also need to change the extra.symfony-web-dir option in the composer.json file:
Listing 16-5
1 {
2
3
4
...
"extra": {
...
PDF brought to you by
generated on December 3, 2014
Chapter 16: How to Override Symfony's default Directory Structure | 65
5
6
7 }
"symfony-web-dir": "my_new_web_dir"
}
Some shared hosts have a public_html web directory root. Renaming your web directory from
web to public_html is one way to make your Symfony project work on your shared host. Another
way is to deploy your application to a directory outside of your web root, delete your public_html
directory, and then replace it with a symbolic link to the web in your project.
If you use the AsseticBundle you need to configure this, so it can use the correct web directory:
Listing 16-6
1 # app/config/config.yml
2
3 # ...
4 assetic:
5
# ...
6
read_from: "%kernel.root_dir%/../../public_html"
Now you just need to clear the cache and dump the assets again and your application should work:
Listing 16-7
1 $ php app/console cache:clear --env=prod
2 $ php app/console assetic:dump --env=prod --no-debug
PDF brought to you by
generated on December 3, 2014
Chapter 16: How to Override Symfony's default Directory Structure | 66
Chapter 17
Using Parameters within a Dependency
Injection Class
You have seen how to use configuration parameters within Symfony service containers. There are special
cases such as when you want, for instance, to use the %kernel.debug% parameter to make the services
in your bundle enter debug mode. For this case there is more work to do in order to make the system
understand the parameter value. By default your parameter %kernel.debug% will be treated as a simple
string. Consider this example with the AcmeDemoBundle:
Listing 17-1
1
2
3
4
5
6
7
8
9
10
11
// Inside Configuration class
$rootNode
->children()
->booleanNode('logging')->defaultValue('%kernel.debug%')->end()
// ...
->end()
;
// Inside the Extension class
$config = $this->processConfiguration($configuration, $configs);
var_dump($config['logging']);
Now, examine the results to see this closely:
Listing 17-2
1
2
3
4
5
6
7
8
9
10
11
12
my_bundle:
logging: true
# true, as expected
my_bundle:
logging: "%kernel.debug%"
# true/false (depends on 2nd parameter of AppKernel),
# as expected, because %kernel.debug% inside configuration
# gets evaluated before being passed to the extension
my_bundle: ~
# passes the string "%kernel.debug%".
PDF brought to you by
generated on December 3, 2014
Chapter 17: Using Parameters within a Dependency Injection Class | 67
13 # Which is always considered as true.
14 # The Configurator does not know anything about
15 # "%kernel.debug%" being a parameter.
In order to support this use case, the Configuration class has to be injected with this parameter via the
extension as follows:
Listing 17-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
namespace Acme\DemoBundle\DependencyInjection;
use Symfony\Component\Config\Definition\Builder\ArrayNodeDefinition;
use Symfony\Component\Config\Definition\Builder\TreeBuilder;
use Symfony\Component\Config\Definition\ConfigurationInterface;
class Configuration implements ConfigurationInterface
{
private $debug;
public function __construct($debug)
{
$this->debug = (Boolean) $debug;
}
public function getConfigTreeBuilder()
{
$treeBuilder = new TreeBuilder();
$rootNode = $treeBuilder->root('acme_demo');
$rootNode
->children()
// ...
->booleanNode('logging')->defaultValue($this->debug)->end()
// ...
->end()
;
return $treeBuilder;
}
}
And set it in the constructor of Configuration via the Extension class:
Listing 17-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
namespace Acme\DemoBundle\DependencyInjection;
use
use
use
use
Symfony\Component\DependencyInjection\ContainerBuilder;
Symfony\Component\DependencyInjection\Loader\XmlFileLoader;
Symfony\Component\HttpKernel\DependencyInjection\Extension;
Symfony\Component\Config\FileLocator;
class AcmeDemoExtension extends Extension
{
// ...
public function getConfiguration(array $config, ContainerBuilder $container)
{
return new Configuration($container->getParameter('kernel.debug'));
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 17: Using Parameters within a Dependency Injection Class | 68
Setting the Default in the Extension
There are some instances of %kernel.debug% usage within a Configurator class in TwigBundle
and AsseticBundle, however this is because the default parameter value is set by the Extension
class. For example in AsseticBundle, you can find:
Listing 17-5
1 $container->setParameter('assetic.debug', $config['debug']);
The string %kernel.debug% passed here as an argument handles the interpreting job to the
container which in turn does the evaluation. Both ways accomplish similar goals. AsseticBundle
will not use %kernel.debug% but rather the new %assetic.debug% parameter.
PDF brought to you by
generated on December 3, 2014
Chapter 17: Using Parameters within a Dependency Injection Class | 69
Chapter 18
Understanding how the Front Controller,
Kernel and Environments Work together
The section How to Master and Create new Environments explained the basics on how Symfony uses
environments to run your application with different configuration settings. This section will explain a bit
more in-depth what happens when your application is bootstrapped. To hook into this process, you need
to understand three parts that work together:
• The Front Controller
• The Kernel Class
• The Environments
Usually, you will not need to define your own front controller or AppKernel class as the Symfony
Standard Edition1 provides sensible default implementations.
This documentation section is provided to explain what is going on behind the scenes.
The Front Controller
The front controller2 is a well-known design pattern; it is a section of code that all requests served by an
application run through.
In the Symfony Standard Edition3, this role is taken by the app.php4 and app_dev.php5 files in the web/
directory. These are the very first PHP scripts executed when a request is processed.
The main purpose of the front controller is to create an instance of the AppKernel (more on that in a
second), make it handle the request and return the resulting response to the browser.
1. https://github.com/symfony/symfony-standard
2. http://en.wikipedia.org/wiki/Front_Controller_pattern
3. https://github.com/symfony/symfony-standard
4. https://github.com/symfony/symfony-standard/blob/master/web/app.php
5. https://github.com/symfony/symfony-standard/blob/master/web/app_dev.php
PDF brought to you by
generated on December 3, 2014
Chapter 18: Understanding how the Front Controller, Kernel and Environments Work together | 70
Because every request is routed through it, the front controller can be used to perform global
initializations prior to setting up the kernel or to decorate6 the kernel with additional features. Examples
include:
•
•
•
•
Configuring the autoloader or adding additional autoloading mechanisms;
Adding HTTP level caching by wrapping the kernel with an instance of AppCache;
Enabling (or skipping) the ClassCache;
Enabling the Debug Component.
The front controller can be chosen by requesting URLs like:
Listing 18-1
1 http://localhost/app_dev.php/some/path/...
As you can see, this URL contains the PHP script to be used as the front controller. You can use
that to easily switch the front controller or use a custom one by placing it in the web/ directory (e.g.
app_cache.php).
When using Apache and the RewriteRule shipped with the Standard Edition7, you can omit the filename
from the URL and the RewriteRule will use app.php as the default one.
Pretty much every other web server should be able to achieve a behavior similar to that of the
RewriteRule described above. Check your server documentation for details or see Configuring a
Web Server.
Make sure you appropriately secure your front controllers against unauthorized access. For
example, you don't want to make a debugging environment available to arbitrary users in your
production environment.
Technically, the app/console8 script used when running Symfony on the command line is also a front
controller, only that is not used for web, but for command line requests.
The Kernel Class
The Kernel9 is the core of Symfony. It is responsible for setting up all the bundles that make up your
application and providing them with the application's configuration. It then creates the service container
before serving requests in its handle()10 method.
There are two methods declared in the KernelInterface11 that are left unimplemented in Kernel12 and
thus serve as template methods13:
• registerBundles()14, which must return an array of all bundles needed to run the
application;
• registerContainerConfiguration()15, which loads the application configuration.
6. http://en.wikipedia.org/wiki/Decorator_pattern
7. https://github.com/symfony/symfony-standard/blob/master/web/.htaccess
8. https://github.com/symfony/symfony-standard/blob/master/app/console
9. http://api.symfony.com/master/Symfony/Component/HttpKernel/Kernel.html
10. http://api.symfony.com/master/Symfony/Component/HttpKernel/HttpKernelInterface.html#handle()
11. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html
12. http://api.symfony.com/master/Symfony/Component/HttpKernel/Kernel.html
13. http://en.wikipedia.org/wiki/Template_method_pattern
14. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html#registerBundles()
15. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html#registerContainerConfiguration()
PDF brought to you by
generated on December 3, 2014
Chapter 18: Understanding how the Front Controller, Kernel and Environments Work together | 71
To fill these (small) blanks, your application needs to subclass the Kernel and implement these methods.
The resulting class is conventionally called the AppKernel.
Again, the Symfony Standard Edition provides an AppKernel16 in the app/ directory. This class uses the
name of the environment - which is passed to the Kernel's constructor17 method and is available via
getEnvironment()18 - to decide which bundles to create. The logic for that is in registerBundles(), a
method meant to be extended by you when you start adding bundles to your application.
You are, of course, free to create your own, alternative or additional AppKernel variants. All you need is
to adapt your (or add a new) front controller to make use of the new kernel.
The name and location of the AppKernel is not fixed. When putting multiple Kernels into a
single application, it might therefore make sense to add additional sub-directories, for example
app/admin/AdminKernel.php and app/api/ApiKernel.php. All that matters is that your front
controller is able to create an instance of the appropriate kernel.
Having different AppKernels might be useful to enable different front controllers (on potentially different
servers) to run parts of your application independently (for example, the admin UI, the frontend UI and
database migrations).
There's a lot more the AppKernel can be used for, for example overriding the default directory
structure. But odds are high that you don't need to change things like this on the fly by having
several AppKernel implementations.
The Environments
As
just
mentioned,
the
AppKernel
has
to
implement
another
method
registerContainerConfiguration()19. This method is responsible for loading the application's
configuration from the right environment.
Environments have been covered extensively in the previous chapter, and you probably remember that
the Standard Edition comes with three of them - dev, prod and test.
More technically, these names are nothing more than strings passed from the front controller to the
AppKernel's constructor. This name can then be used in the registerContainerConfiguration()20
method to decide which configuration files to load.
The Standard Edition's AppKernel21 class implements this method by simply loading the app/config/
config_*environment*.yml file. You are, of course, free to implement this method differently if you
need a more sophisticated way of loading your configuration.
16. https://github.com/symfony/symfony-standard/blob/master/app/AppKernel.php
17. http://api.symfony.com/master/Symfony/Component/HttpKernel/Kernel.html#__construct()
18. http://api.symfony.com/master/Symfony/Component/HttpKernel/Kernel.html#getEnvironment()
19. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html#registerContainerConfiguration()
20. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html#registerContainerConfiguration()
21. https://github.com/symfony/symfony-standard/blob/master/app/AppKernel.php
PDF brought to you by
generated on December 3, 2014
Chapter 18: Understanding how the Front Controller, Kernel and Environments Work together | 72
Chapter 19
How to Set external Parameters in the Service
Container
In the chapter How to Master and Create new Environments, you learned how to manage your application
configuration. At times, it may benefit your application to store certain credentials outside of your project
code. Database configuration is one such example. The flexibility of the Symfony service container allows
you to easily do this.
Environment Variables
Symfony will grab any environment variable prefixed with SYMFONY__ and set it as a parameter in the
service container. Some transformations are applied to the resulting parameter name:
• SYMFONY__ prefix is removed;
• Parameter name is lowercased;
• Double underscores are replaced with a period, as a period is not a valid character in an
environment variable name.
For example, if you're using Apache, environment variables can be set using the following VirtualHost
configuration:
Listing 19-1
1 <VirtualHost *:80>
2
ServerName
Symfony
3
DocumentRoot
"/path/to/symfony_2_app/web"
4
DirectoryIndex index.php index.html
5
SetEnv
SYMFONY__DATABASE__USER user
6
SetEnv
SYMFONY__DATABASE__PASSWORD secret
7
8
<Directory "/path/to/symfony_2_app/web">
9
AllowOverride All
10
Allow from All
11
</Directory>
12 </VirtualHost>
PDF brought to you by
generated on December 3, 2014
Chapter 19: How to Set external Parameters in the Service Container | 73
The example above is for an Apache configuration, using the SetEnv1 directive. However, this will
work for any web server which supports the setting of environment variables.
Also, in order for your console to work (which does not use Apache), you must export these as
shell variables. On a Unix system, you can run the following:
Listing 19-2
1 $ export SYMFONY__DATABASE__USER=user
2 $ export SYMFONY__DATABASE__PASSWORD=secret
Now that you have declared an environment variable, it will be present in the PHP $_SERVER global
variable. Symfony then automatically sets all $_SERVER variables prefixed with SYMFONY__ as parameters
in the service container.
You can now reference these parameters wherever you need them.
Listing 19-3
1 doctrine:
2
dbal:
3
driver
4
dbname:
5
user:
6
password:
pdo_mysql
symfony_project
"%database.user%"
"%database.password%"
Even in debug mode, setting or changing an environment variable requires your cache to be
cleared to make the parameter available. In debug mode, this is required since only a change to a
configuration file that is loaded by Symfony triggers your configuration to be re-evaluated.
Constants
The container also has support for setting PHP constants as parameters. See Constants as Parameters for
more details.
Miscellaneous Configuration
The imports directive can be used to pull in parameters stored elsewhere. Importing a PHP file gives
you the flexibility to add whatever is needed in the container. The following imports a file named
parameters.php.
Listing 19-4
1 # app/config/config.yml
2 imports:
3
- { resource: parameters.php }
A resource file can be one of many types. PHP, XML, YAML, INI, and closure resources are all
supported by the imports directive.
1. http://httpd.apache.org/docs/current/env.html
PDF brought to you by
generated on December 3, 2014
Chapter 19: How to Set external Parameters in the Service Container | 74
In parameters.php, tell the service container the parameters that you wish to set. This is useful when
important configuration is in a non-standard format. The example below includes a Drupal database
configuration in the Symfony service container.
Listing 19-5
1 // app/config/parameters.php
2 include_once('/path/to/drupal/sites/default/settings.php');
3 $container->setParameter('drupal.database.url', $db_url);
PDF brought to you by
generated on December 3, 2014
Chapter 19: How to Set external Parameters in the Service Container | 75
Chapter 20
How to Use PdoSessionHandler to Store
Sessions in the Database
The default Symfony session storage writes the session information to file(s). Most medium to large
websites use a database to store the session values instead of files, because databases are easier to use and
scale in a multi-webserver environment.
Symfony has a built-in solution for database session storage called PdoSessionHandler1. To use it, you
just need to change some parameters in config.yml (or the configuration format of your choice):
Listing 20-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
# app/config/config.yml
framework:
session:
# ...
handler_id: session.handler.pdo
parameters:
pdo.db_options:
db_table:
db_id_col:
db_data_col:
db_time_col:
session
session_id
session_value
session_time
services:
pdo:
class: PDO
arguments:
dsn:
"mysql:dbname=mydatabase"
user:
myuser
password: mypassword
calls:
- [setAttribute, [3, 2]] # \PDO::ATTR_ERRMODE, \PDO::ERRMODE_EXCEPTION
session.handler.pdo:
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/Session/Storage/Handler/PdoSessionHandler.html
PDF brought to you by
generated on December 3, 2014
Chapter 20: How to Use PdoSessionHandler to Store Sessions in the Database | 76
25
class:
26 Symfony\Component\HttpFoundation\Session\Storage\Handler\PdoSessionHandler
arguments: ["@pdo", "%pdo.db_options%"]
•
•
•
•
db_table: The name of the session table in your database
db_id_col: The name of the id column in your session table (VARCHAR(255) or larger)
db_data_col: The name of the value column in your session table (TEXT or CLOB)
db_time_col: The name of the time column in your session table (INTEGER)
Sharing your Database Connection Information
With the given configuration, the database connection settings are defined for the session storage
connection only. This is OK when you use a separate database for the session data.
But if you'd like to store the session data in the same database as the rest of your project's data, you can
use the connection settings from the parameters.yml file by referencing the database-related parameters
defined there:
Listing 20-2
1 pdo:
2
class: PDO
3
arguments:
4
- "mysql:host=%database_host%;port=%database_port%;dbname=%database_name%"
5
- "%database_user%"
6
- "%database_password%"
Example SQL Statements
MySQL
The SQL statement for creating the needed database table might look like the following (MySQL):
Listing 20-3
1 CREATE TABLE `session` (
2
`session_id` varchar(255) NOT NULL,
3
`session_value` text NOT NULL,
4
`session_time` int(11) NOT NULL,
5
PRIMARY KEY (`session_id`)
6 ) ENGINE=InnoDB DEFAULT CHARSET=utf8;
PostgreSQL
For PostgreSQL, the statement should look like this:
Listing 20-4
1 CREATE TABLE session (
2
session_id character varying(255) NOT NULL,
3
session_value text NOT NULL,
4
session_time integer NOT NULL,
5
CONSTRAINT session_pkey PRIMARY KEY (session_id)
6 );
PDF brought to you by
generated on December 3, 2014
Chapter 20: How to Use PdoSessionHandler to Store Sessions in the Database | 77
Microsoft SQL Server
For MSSQL, the statement might look like the following:
Listing 20-5
1 CREATE TABLE [dbo].[session](
2
[session_id] [nvarchar](255) NOT NULL,
3
[session_value] [ntext] NOT NULL,
4
[session_time] [int] NOT NULL,
5
PRIMARY KEY CLUSTERED(
6
[session_id] ASC
7
) WITH (
8
PAD_INDEX = OFF,
9
STATISTICS_NORECOMPUTE = OFF,
10
IGNORE_DUP_KEY = OFF,
11
ALLOW_ROW_LOCKS = ON,
12
ALLOW_PAGE_LOCKS = ON
13
) ON [PRIMARY]
14 ) ON [PRIMARY] TEXTIMAGE_ON [PRIMARY]
PDF brought to you by
generated on December 3, 2014
Chapter 20: How to Use PdoSessionHandler to Store Sessions in the Database | 78
Chapter 21
How to Use the Apache Router
Symfony, while fast out of the box, also provides various ways to increase that speed with a little bit of
tweaking. One of these ways is by letting Apache handle routes directly, rather than using Symfony for
this task.
Apache router was deprecated in Symfony 2.5 and will be removed in Symfony 3.0. Since the PHP
implementation of the Router was improved, performance gains were no longer significant (while
it's very hard to replicate the same behavior).
Change Router Configuration Parameters
To dump Apache routes you must first tweak some configuration parameters to tell Symfony to use the
ApacheUrlMatcher instead of the default one:
Listing 21-1
1 # app/config/config_prod.yml
2 parameters:
3
router.options.matcher.cache_class: ~ # disable router cache
4
router.options.matcher_class: Symfony\Component\Routing\Matcher\ApacheUrlMatcher
Note that ApacheUrlMatcher1 extends UrlMatcher2 so even if you don't regenerate the
mod_rewrite rules, everything will work (because at the end of ApacheUrlMatcher::match() a call
to parent::match() is done).
Generating mod_rewrite Rules
To test that it's working, create a very basic route for the AcmeDemoBundle:
1. http://api.symfony.com/master/Symfony/Component/Routing/Matcher/ApacheUrlMatcher.html
2. http://api.symfony.com/master/Symfony/Component/Routing/Matcher/UrlMatcher.html
PDF brought to you by
generated on December 3, 2014
Chapter 21: How to Use the Apache Router | 79
Listing 21-2
1 # app/config/routing.yml
2 hello:
3
path: /hello/{name}
4
defaults: { _controller: AcmeDemoBundle:Demo:hello }
Now generate the mod_rewrite rules:
Listing 21-3
1 $ php app/console router:dump-apache -e=prod --no-debug
Which should roughly output the following:
Listing 21-4
# skip "real" requests
RewriteCond %{REQUEST_FILENAME} -f
RewriteRule .* - [QSA,L]
1
2
3
4
5
6
7
# hello
RewriteCond %{REQUEST_URI} ^/hello/([^/]+?)$
RewriteRule .* app.php
[QSA,L,E=_ROUTING__route:hello,E=_ROUTING_name:%1,E=_ROUTING__controller:AcmeDemoBundle\:Demo\:hello]
You can now rewrite web/.htaccess to use the new rules, so with this example it should look like this:
Listing 21-5
1 <IfModule mod_rewrite.c>
2
RewriteEngine On
3
4
# skip "real" requests
5
RewriteCond %{REQUEST_FILENAME} -f
6
RewriteRule .* - [QSA,L]
7
8
# hello
9
RewriteCond %{REQUEST_URI} ^/hello/([^/]+?)$
10
RewriteRule .* app.php
11 [QSA,L,E=_ROUTING__route:hello,E=_ROUTING_name:%1,E=_ROUTING__controller:AcmeDemoBundle\:Demo\:hello]
</IfModule>
The procedure above should be done each time you add/change a route if you want to take full
advantage of this setup.
That's it! You're now all set to use Apache routes.
Additional Tweaks
To save a little bit of processing time, change occurrences of Request to ApacheRequest in web/app.php:
Listing 21-6
1
2
3
4
5
6
7
// web/app.php
require_once __DIR__.'/../app/bootstrap.php.cache';
require_once __DIR__.'/../app/AppKernel.php';
// require_once __DIR__.'/../app/AppCache.php';
use Symfony\Component\HttpFoundation\ApacheRequest;
PDF brought to you by
generated on December 3, 2014
Chapter 21: How to Use the Apache Router | 80
8
9
10
11
12
$kernel = new AppKernel('prod', false);
$kernel->loadClassCache();
// $kernel = new AppCache($kernel);
$kernel->handle(ApacheRequest::createFromGlobals())->send();
PDF brought to you by
generated on December 3, 2014
Chapter 21: How to Use the Apache Router | 81
Chapter 22
Configuring a Web Server
The preferred way to develop your Symfony application is to use PHP's internal web server. However,
when using an older PHP version or when running the application in the production environment, you'll
need to use a fully-featured web server. This article describes several ways to use Symfony with Apache2
or Nginx.
When using Apache2, you can configure PHP as an Apache module or with FastCGI using PHP FPM.
FastCGI also is the preferred way to use PHP with Nginx.
The Web Directory
The web directory is the home of all of your application's public and static files, including images,
stylesheets and JavaScript files. It is also where the front controllers live. For more details, see the
The Web Directory.
The web directory services as the document root when configuring your web server. In the
examples below, the web/ directory will be the document root. This directory is /var/www/
project/web/.
Apache2 with mod_php/PHP-CGI
For advanced Apache configuration options, see the official Apache1 documentation. The minimum
basics to get your application running under Apache2 are:
Listing 22-1
1 <VirtualHost *:80>
2
ServerName domain.tld
3
ServerAlias www.domain.tld
4
5
DocumentRoot /var/www/project/web
6
<Directory /var/www/project/web>
7
# enable the .htaccess rewrites
8
AllowOverride All
1. http://httpd.apache.org/docs/current/mod/core.html#documentroot
PDF brought to you by
generated on December 3, 2014
Chapter 22: Configuring a Web Server | 82
9
Order allow,deny
10
Allow from All
11
</Directory>
12
13
ErrorLog /var/log/apache2/project_error.log
14
CustomLog /var/log/apache2/project_access.log combined
15 </VirtualHost>
If your system supports the APACHE_LOG_DIR variable, you may want to use ${APACHE_LOG_DIR}/
instead of /var/log/apache2/.
For performance reasons, you will probably want to set AllowOverride None and implement the
rewrite rules in the web/.htaccess into the VirtualHost config.
If you are using php-cgi, Apache does not pass HTTP basic username and password to PHP by default.
To work around this limitation, you should use the following configuration snippet:
Listing 22-2
1 RewriteRule .* - [E=HTTP_AUTHORIZATION:%{HTTP:Authorization}]
In Apache 2.4, Order allow,deny has been replaced by Require all granted, and hence you
need to modify your Directory permission settings as follows:
Listing 22-3
1 <Directory /var/www/project/web>
2
# enable the .htaccess rewrites
3
AllowOverride All
4
Require all granted
5 </Directory>
Apache2 with PHP-FPM
To make use of PHP5-FPM with Apache, you first have to ensure that you have the FastCGI process
manager php-fpm binary and Apache's FastCGI module installed (for example, on a Debian based system
you have to install the libapache2-mod-fastcgi and php5-fpm packages).
PHP-FPM uses so-called pools to handle incoming FastCGI requests. You can configure an arbitrary
number of pools in the FPM configuration. In a pool you configure either a TCP socket (IP and port) or
a unix domain socket to listen on. Each pool can also be run under a different UID and GID:
Listing 22-4
1
2
3
4
5
6
7
8
; a pool called www
[www]
user = www-data
group = www-data
; use a unix domain socket
listen = /var/run/php5-fpm.sock
PDF brought to you by
generated on December 3, 2014
Chapter 22: Configuring a Web Server | 83
9 ; or listen on a TCP socket
10 listen = 127.0.0.1:9000
Using mod_proxy_fcgi with Apache 2.4
If you are running Apache 2.4, you can easily use mod_proxy_fcgi to pass incoming requests to PHPFPM. Configure PHP-FPM to listen on a TCP socket (mod_proxy currently does not support unix sockets2),
enable mod_proxy and mod_proxy_fcgi in your Apache configuration and use the ProxyPassMatch
directive to pass requests for PHP files to PHP FPM:
Listing 22-5
1 <VirtualHost *:80>
2
ServerName domain.tld
3
ServerAlias www.domain.tld
4
5
ProxyPassMatch ^/(.*\.php(/.*)?)$ fcgi://127.0.0.1:9000/var/www/project/web/$1
6
7
DocumentRoot /var/www/project/web
8
<Directory /var/www/project/web>
9
# enable the .htaccess rewrites
10
AllowOverride All
11
Require all granted
12
</Directory>
13
14
ErrorLog /var/log/apache2/project_error.log
15
CustomLog /var/log/apache2/project_access.log combined
16 </VirtualHost>
When you run your Symfony application on a subpath of your document root, the regular
expression used in ProxyPassMatch directive must be changed accordingly:
Listing 22-6
1 ProxyPassMatch ^/path-to-app/(.*\.php(/.*)?)$ fcgi://127.0.0.1:9000/var/www/project/
web/$1
PHP-FPM with Apache 2.2
On Apache 2.2 or lower, you cannot use mod_proxy_fcgi. You have to use the FastCgiExternalServer3
directive instead. Therefore, your Apache configuration should look something like this:
Listing 22-7
1 <VirtualHost *:80>
2
ServerName domain.tld
3
ServerAlias www.domain.tld
4
5
AddHandler php5-fcgi .php
6
Action php5-fcgi /php5-fcgi
7
Alias /php5-fcgi /usr/lib/cgi-bin/php5-fcgi
8
FastCgiExternalServer /usr/lib/cgi-bin/php5-fcgi -host 127.0.0.1:9000 -pass-header
9 Authorization
10
11
DocumentRoot /var/www/project/web
2. https://issues.apache.org/bugzilla/show_bug.cgi?id=54101
3. http://www.fastcgi.com/mod_fastcgi/docs/mod_fastcgi.html#FastCgiExternalServer
PDF brought to you by
generated on December 3, 2014
Chapter 22: Configuring a Web Server | 84
12
13
14
15
16
17
18
19
20
<Directory /var/www/project/web>
# enable the .htaccess rewrites
AllowOverride All
Order allow,deny
Allow from all
</Directory>
ErrorLog /var/log/apache2/project_error.log
CustomLog /var/log/apache2/project_access.log combined
</VirtualHost>
If you prefer to use a unix socket, you have to use the -socket option instead:
Listing 22-8
1 FastCgiExternalServer /usr/lib/cgi-bin/php5-fcgi -socket /var/run/php5-fpm.sock
-pass-header Authorization
Nginx
For advanced Nginx configuration options, see the official Nginx4 documentation. The minimum basics
to get your application running under Nginx are:
Listing 22-9
1 server {
2
server_name domain.tld www.domain.tld;
3
root /var/www/project/web;
4
5
location / {
6
# try to serve file directly, fallback to app.php
7
try_files $uri /app.php$is_args$args;
8
}
9
# DEV
10
# This rule should only be placed on your development environment
11
# In production, don't include this and don't deploy app_dev.php or config.php
12
location ~ ^/(app_dev|config)\.php(/|$) {
13
fastcgi_pass unix:/var/run/php5-fpm.sock;
14
fastcgi_split_path_info ^(.+\.php)(/.*)$;
15
include fastcgi_params;
16
fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
17
fastcgi_param HTTPS off;
18
}
19
# PROD
20
location ~ ^/app\.php(/|$) {
21
fastcgi_pass unix:/var/run/php5-fpm.sock;
22
fastcgi_split_path_info ^(.+\.php)(/.*)$;
23
include fastcgi_params;
24
fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
25
fastcgi_param HTTPS off;
26
# Prevents URIs that include the front controller. This will 404:
27
# http://domain.tld/app.php/some-path
28
# Remove the internal directive to allow URIs like this
29
internal;
30
}
31
32
error_log /var/log/nginx/project_error.log;
4. http://wiki.nginx.org/Symfony
PDF brought to you by
generated on December 3, 2014
Chapter 22: Configuring a Web Server | 85
33
34 }
access_log /var/log/nginx/project_access.log;
Depending on your PHP-FPM config, the fastcgi_pass can also be fastcgi_pass
127.0.0.1:9000.
This executes only app.php, app_dev.php and config.php in the web directory. All other files will
be served as text. You must also make sure that if you do deploy app_dev.php or config.php that
these files are secured and not available to any outside user (the IP checking code at the top of each
file does this by default).
If you have other PHP files in your web directory that need to be executed, be sure to include them
in the location block above.
PDF brought to you by
generated on December 3, 2014
Chapter 22: Configuring a Web Server | 86
Chapter 23
How to Organize Configuration Files
The default Symfony Standard Edition defines three execution environments called dev, prod and test.
An environment simply represents a way to execute the same codebase with different configurations.
In order to select the configuration file to load for each environment, Symfony executes the
registerContainerConfiguration() method of the AppKernel class:
Listing 23-1
1
2
3
4
5
6
7
8
9
10
11
12
13
// app/AppKernel.php
use Symfony\Component\HttpKernel\Kernel;
use Symfony\Component\Config\Loader\LoaderInterface;
class AppKernel extends Kernel
{
// ...
public function registerContainerConfiguration(LoaderInterface $loader)
{
$loader->load(__DIR__.'/config/config_'.$this->getEnvironment().'.yml');
}
}
This method loads the app/config/config_dev.yml file for the dev environment and so on. In turn,
this file loads the common configuration file located at app/config/config.yml. Therefore, the
configuration files of the default Symfony Standard Edition follow this structure:
Listing 23-2
1
2
3
4
5
6
7
8
9
10
11
<your-project>/
в”њв”Ђ app/
в”‚ в””в”Ђ config/
в”‚
в”њв”Ђ config.yml
в”‚
в”њв”Ђ config_dev.yml
в”‚
в”њв”Ђ config_prod.yml
в”‚
в”њв”Ђ config_test.yml
в”‚
в”њв”Ђ parameters.yml
в”‚
в”њв”Ђ parameters.yml.dist
в”‚
в”њв”Ђ routing.yml
в”‚
в”њв”Ђ routing_dev.yml
PDF brought to you by
generated on December 3, 2014
Chapter 23: How to Organize Configuration Files | 87
12
13
14
15
в”‚
в””в”Ђ security.yml
в”њв”Ђ src/
в”њв”Ђ vendor/
в””в”Ђ web/
This default structure was chosen for its simplicity — one file per environment. But as any other Symfony
feature, you can customize it to better suit your needs. The following sections explain different ways to
organize your configuration files. In order to simplify the examples, only the dev and prod environments
are taken into account.
Different Directories per Environment
Instead of suffixing the files with _dev and _prod, this technique groups all the related configuration files
under a directory with the same name as the environment:
Listing 23-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
<your-project>/
в”њв”Ђ app/
в”‚ в””в”Ђ config/
в”‚
в”њв”Ђ common/
в”‚
в”‚ в”њв”Ђ config.yml
в”‚
в”‚ в”њв”Ђ parameters.yml
в”‚
в”‚ в”њв”Ђ routing.yml
в”‚
в”‚ в””в”Ђ security.yml
в”‚
в”њв”Ђ dev/
в”‚
в”‚ в”њв”Ђ config.yml
в”‚
в”‚ в”њв”Ђ parameters.yml
в”‚
в”‚ в”њв”Ђ routing.yml
в”‚
в”‚ в””в”Ђ security.yml
в”‚
в””в”Ђ prod/
в”‚
в”њв”Ђ config.yml
в”‚
в”њв”Ђ parameters.yml
в”‚
в”њв”Ђ routing.yml
в”‚
в””в”Ђ security.yml
в”њв”Ђ src/
в”њв”Ђ vendor/
в””в”Ђ web/
To make this work, change the code of the registerContainerConfiguration()1 method:
Listing 23-4
1
2
3
4
5
6
7
8
9
10
11
12
13
// app/AppKernel.php
use Symfony\Component\HttpKernel\Kernel;
use Symfony\Component\Config\Loader\LoaderInterface;
class AppKernel extends Kernel
{
// ...
public function registerContainerConfiguration(LoaderInterface $loader)
{
$loader->load(__DIR__.'/config/'.$this->getEnvironment().'/config.yml');
}
}
1. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html#registerContainerConfiguration()
PDF brought to you by
generated on December 3, 2014
Chapter 23: How to Organize Configuration Files | 88
Then, make sure that each config.yml file loads the rest of the configuration files, including the common
files. For instance, this would be the imports needed for the app/config/dev/config.yml file:
Listing 23-5
1 # app/config/dev/config.yml
2 imports:
3
- { resource: '../common/config.yml' }
4
- { resource: 'parameters.yml' }
5
- { resource: 'security.yml' }
6
7 # ...
Due to the way in which parameters are resolved, you cannot use them to build paths in imports
dynamically. This means that something like the following doesn't work:
Listing 23-6
1 # app/config/config.yml
2 imports:
3
- { resource: "%kernel.root_dir%/parameters.yml" }
Semantic Configuration Files
A different organization strategy may be needed for complex applications with large configuration files.
For instance, you could create one file per bundle and several files to define all application services:
Listing 23-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
<your-project>/
в”њв”Ђ app/
в”‚ в””в”Ђ config/
в”‚
в”њв”Ђ bundles/
в”‚
в”‚ в”њв”Ђ bundle1.yml
в”‚
в”‚ в”њв”Ђ bundle2.yml
в”‚
в”‚ в”њв”Ђ ...
в”‚
в”‚ в””в”Ђ bundleN.yml
в”‚
в”њв”Ђ environments/
в”‚
в”‚ в”њв”Ђ common.yml
в”‚
в”‚ в”њв”Ђ dev.yml
в”‚
в”‚ в””в”Ђ prod.yml
в”‚
в”њв”Ђ routing/
в”‚
в”‚ в”њв”Ђ common.yml
в”‚
в”‚ в”њв”Ђ dev.yml
в”‚
в”‚ в””в”Ђ prod.yml
в”‚
в””в”Ђ services/
в”‚
в”њв”Ђ frontend.yml
в”‚
в”њв”Ђ backend.yml
в”‚
в”њв”Ђ ...
в”‚
в””в”Ђ security.yml
в”њв”Ђ src/
в”њв”Ђ vendor/
в””в”Ђ web/
Again, change the code of the registerContainerConfiguration() method to make Symfony aware of
the new file organization:
Listing 23-8
PDF brought to you by
generated on December 3, 2014
Chapter 23: How to Organize Configuration Files | 89
1
2
3
4
5
6
7
8
9
10
11
12
13
// app/AppKernel.php
use Symfony\Component\HttpKernel\Kernel;
use Symfony\Component\Config\Loader\LoaderInterface;
class AppKernel extends Kernel
{
// ...
public function registerContainerConfiguration(LoaderInterface $loader)
{
$loader->load(__DIR__.'/config/environments/'.$this->getEnvironment().'.yml');
}
}
Following the same technique explained in the previous section, make sure to import the appropriate
configuration files from each main file (common.yml, dev.yml and prod.yml).
Advanced Techniques
Symfony loads configuration files using the Config component, which provides some advanced features.
Mix and Match Configuration Formats
Configuration files can import files defined with any other built-in configuration format (.yml, .xml,
.php, .ini):
Listing 23-9
1 # app/config/config.yml
2 imports:
3
- { resource: 'parameters.yml' }
4
- { resource: 'services.xml' }
5
- { resource: 'security.yml' }
6
- { resource: 'legacy.php' }
7
8 # ...
The IniFileLoader parses the file contents using the parse_ini_file2 function. Therefore, you
can only set parameters to string values. Use one of the other loaders if you want to use other data
types (e.g. boolean, integer, etc.).
If you use any other configuration format, you have to define your own loader class extending it from
FileLoader3. When the configuration values are dynamic, you can use the PHP configuration file to
execute your own logic. In addition, you can define your own services to load configurations from
databases or web services.
Global Configuration Files
Some system administrators may prefer to store sensitive parameters in files outside the project directory.
Imagine that the database credentials for your website are stored in the /etc/sites/mysite.com/
parameters.yml file. Loading this file is as simple as indicating the full file path when importing it from
any other configuration file:
2. http://php.net/manual/en/function.parse-ini-file.php
3. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Loader/FileLoader.html
PDF brought to you by
generated on December 3, 2014
Chapter 23: How to Organize Configuration Files | 90
Listing 23-10
1 # app/config/config.yml
2 imports:
3
- { resource: 'parameters.yml' }
4
- { resource: '/etc/sites/mysite.com/parameters.yml' }
5
6 # ...
Most of the time, local developers won't have the same files that exist on the production servers. For that
reason, the Config component provides the ignore_errors option to silently discard errors when the
loaded file doesn't exist:
Listing 23-11
1 # app/config/config.yml
2 imports:
3
- { resource: 'parameters.yml' }
4
- { resource: '/etc/sites/mysite.com/parameters.yml', ignore_errors: true }
5
6 # ...
As you've seen, there are lots of ways to organize your configuration files. You can choose one of these or
even create your own custom way of organizing the files. Don't feel limited by the Standard Edition that
comes with Symfony. For even more customization, see "How to Override Symfony's default Directory
Structure".
PDF brought to you by
generated on December 3, 2014
Chapter 23: How to Organize Configuration Files | 91
Chapter 24
How to Create a Console Command
The Console page of the Components section (The Console Component) covers how to create a console
command. This cookbook article covers the differences when creating console commands within the
Symfony framework.
Automatically Registering Commands
To make the console commands available automatically with Symfony, create a Command directory inside
your bundle and create a PHP file suffixed with Command.php for each command that you want to
provide. For example, if you want to extend the AcmeDemoBundle to greet you from the command line,
create GreetCommand.php and add the following to it:
Listing 24-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/DemoBundle/Command/GreetCommand.php
namespace Acme\DemoBundle\Command;
use
use
use
use
use
Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;
Symfony\Component\Console\Input\InputArgument;
Symfony\Component\Console\Input\InputInterface;
Symfony\Component\Console\Input\InputOption;
Symfony\Component\Console\Output\OutputInterface;
class GreetCommand extends ContainerAwareCommand
{
protected function configure()
{
$this
->setName('demo:greet')
->setDescription('Greet someone')
->addArgument('name', InputArgument::OPTIONAL, 'Who do you want to greet?')
->addOption('yell', null, InputOption::VALUE_NONE, 'If set, the task will yell
in uppercase letters')
;
}
protected function execute(InputInterface $input, OutputInterface $output)
PDF brought to you by
generated on December 3, 2014
Chapter 24: How to Create a Console Command | 92
24
25
26
27
28
29
30
31
32
33
34
35
36
37
{
$name = $input->getArgument('name');
if ($name) {
$text = 'Hello '.$name;
} else {
$text = 'Hello';
}
if ($input->getOption('yell')) {
$text = strtoupper($text);
}
$output->writeln($text);
}
}
This command will now automatically be available to run:
Listing 24-2
1 $ php app/console demo:greet Fabien
Register Commands in the Service Container
Just like controllers, commands can be declared as services. See the dedicated cookbook entry for details.
Getting Services from the Service Container
By using ContainerAwareCommand1 as the base class for the command (instead of the more basic
Command2), you have access to the service container. In other words, you have access to any configured
service:
Listing 24-3
1 protected function execute(InputInterface $input, OutputInterface $output)
2 {
3
$name = $input->getArgument('name');
4
$logger = $this->getContainer()->get('logger');
5
6
$logger->info('Executing command for '.$name);
7
// ...
8 }
However, due to the container scopes this code doesn't work for some services. For instance, if you try to
get the request service or any other service related to it, you'll get the following error:
Listing 24-4
1 You cannot create a service ("request") of an inactive scope ("request").
Consider the following example that uses the translator service to translate some contents using a
console command:
Listing 24-5
1. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Command/ContainerAwareCommand.html
2. http://api.symfony.com/master/Symfony/Component/Console/Command/Command.html
PDF brought to you by
generated on December 3, 2014
Chapter 24: How to Create a Console Command | 93
1 protected function execute(InputInterface $input, OutputInterface $output)
2 {
3
$name = $input->getArgument('name');
4
$translator = $this->getContainer()->get('translator');
5
if ($name) {
6
$output->writeln($translator->trans('Hello %name%!', array('%name%' => $name)));
7
} else {
8
$output->writeln($translator->trans('Hello!'));
9
}
10 }
If you dig into the Translator component classes, you'll see that the request service is required to get the
locale into which the contents are translated:
Listing 24-6
1
2
3
4
5
6
7
8
9
10
// vendor/symfony/symfony/src/Symfony/Bundle/FrameworkBundle/Translation/Translator.php
public function getLocale()
{
if (null === $this->locale && $this->container->isScopeActive('request')
&& $this->container->has('request')) {
$this->locale = $this->container->get('request')->getLocale();
}
return $this->locale;
}
Therefore, when using the translator service inside a command, you'll get the previous "You cannot
create a service of an inactive scope" error message. The solution in this case is as easy as setting the locale
value explicitly before translating contents:
Listing 24-7
1 protected function execute(InputInterface $input, OutputInterface $output)
2 {
3
$name = $input->getArgument('name');
4
$locale = $input->getArgument('locale');
5
6
$translator = $this->getContainer()->get('translator');
7
$translator->setLocale($locale);
8
9
if ($name) {
10
$output->writeln($translator->trans('Hello %name%!', array('%name%' => $name)));
11
} else {
12
$output->writeln($translator->trans('Hello!'));
13
}
14 }
However for other services the solution might be more complex. For more details, see How to Work with
Scopes.
Testing Commands
When
testing
commands
used
as
part
Symfony\Bundle\FrameworkBundle\Console\Application3
Symfony\Component\Console\Application4:
of
should
the
be
full
used
framework
instead
of
3. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Console/Application.html
4. http://api.symfony.com/master/Symfony/Component/Console/Application.html
PDF brought to you by
generated on December 3, 2014
Chapter 24: How to Create a Console Command | 94
Listing 24-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
use Symfony\Component\Console\Tester\CommandTester;
use Symfony\Bundle\FrameworkBundle\Console\Application;
use Acme\DemoBundle\Command\GreetCommand;
class ListCommandTest extends \PHPUnit_Framework_TestCase
{
public function testExecute()
{
// mock the Kernel or create one depending on your needs
$application = new Application($kernel);
$application->add(new GreetCommand());
$command = $application->find('demo:greet');
$commandTester = new CommandTester($command);
$commandTester->execute(
array(
'name'
=> 'Fabien',
'--yell' => true,
)
);
$this->assertRegExp('/.../', $commandTester->getDisplay());
// ...
}
}
New in version 2.4: Since Symfony 2.4, the CommandTester automatically detects the name of the
command to execute. Prior to Symfony 2.4, you need to pass it via the command key.
In the specific case above, the name parameter and the --yell option are not mandatory for
the command to work, but are shown so you can see how to customize them when calling the
command.
To be able to use the fully set up service container for your console tests you can extend your test from
KernelTestCase5:
Listing 24-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
use
use
use
use
Symfony\Component\Console\Tester\CommandTester;
Symfony\Bundle\FrameworkBundle\Console\Application;
Symfony\Bundle\FrameworkBundle\Test\KernelTestCase;
Acme\DemoBundle\Command\GreetCommand;
class ListCommandTest extends KernelTestCase
{
public function testExecute()
{
$kernel = $this->createKernel();
$kernel->boot();
$application = new Application($kernel);
$application->add(new GreetCommand());
5. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Test/KernelTestCase.html
PDF brought to you by
generated on December 3, 2014
Chapter 24: How to Create a Console Command | 95
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29 }
$command = $application->find('demo:greet');
$commandTester = new CommandTester($command);
$commandTester->execute(
array(
'name'
=> 'Fabien',
'--yell' => true,
)
);
$this->assertRegExp('/.../', $commandTester->getDisplay());
// ...
}
New in version 2.5: KernelTestCase6 was extracted from WebTestCase7 in Symfony 2.5.
WebTestCase inherits from KernelTestCase. The WebTestCase creates an instance of Client8
via createClient(), while KernelTestCase creates an instance of KernelInterface9 via
createKernel().
6.
7.
8.
9.
http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Test/KernelTestCase.html
http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Test/WebTestCase.html
http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Client.html
http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 24: How to Create a Console Command | 96
Chapter 25
How to Use the Console
The Using Console Commands, Shortcuts and Built-in Commands page of the components documentation
looks at the global console options. When you use the console as part of the full stack framework, some
additional global options are available as well.
By default, console commands run in the dev environment and you may want to change this for
some commands. For example, you may want to run some commands in the prod environment for
performance reasons. Also, the result of some commands will be different depending on the environment.
For example, the cache:clear command will clear and warm the cache for the specified environment
only. To clear and warm the prod cache you need to run:
Listing 25-1
1 $ php app/console cache:clear --env=prod
or the equivalent:
Listing 25-2
1 $ php app/console cache:clear -e prod
In addition to changing the environment, you can also choose to disable debug mode. This can be useful
where you want to run commands in the dev environment but avoid the performance hit of collecting
debug data:
Listing 25-3
1 $ php app/console list --no-debug
There is an interactive shell which allows you to enter commands without having to specify php app/
console each time, which is useful if you need to run several commands. To enter the shell run:
Listing 25-4
1 $ php app/console --shell
2 $ php app/console -s
You can now just run commands with the command name:
Listing 25-5
1 Symfony > list
When using the shell you can choose to run each command in a separate process:
PDF brought to you by
generated on December 3, 2014
Chapter 25: How to Use the Console | 97
Listing 25-6
1 $ php app/console --shell --process-isolation
2 $ php app/console -s --process-isolation
When you do this, the output will not be colorized and interactivity is not supported so you will need to
pass all command params explicitly.
Unless you are using isolated processes, clearing the cache in the shell will not have an effect on
subsequent commands you run. This is because the original cached files are still being used.
PDF brought to you by
generated on December 3, 2014
Chapter 25: How to Use the Console | 98
Chapter 26
How to Generate URLs and Send Emails from
the Console
Unfortunately, the command line context does not know about your VirtualHost or domain name. This
means that if you generate absolute URLs within a Console Command you'll probably end up with
something like http://localhost/foo/bar which is not very useful.
To fix this, you need to configure the "request context", which is a fancy way of saying that you need to
configure your environment so that it knows what URL it should use when generating URLs.
There are two ways of configuring the request context: at the application level and per Command.
Configuring the Request Context Globally
To configure the Request Context - which is used by the URL Generator - you can redefine the
parameters it uses as default values to change the default host (localhost) and scheme (http). You can
also configure the base path if Symfony is not running in the root directory.
Note that this does not impact URLs generated via normal web requests, since those will override the
defaults.
Listing 26-1
1 # app/config/parameters.yml
2 parameters:
3
router.request_context.host: example.org
4
router.request_context.scheme: https
5
router.request_context.base_url: my/path
Configuring the Request Context per Command
To change it only in one command you can simply fetch the Request Context from the router service
and override its settings:
Listing 26-2
PDF brought to you by
generated on December 3, 2014
Chapter 26: How to Generate URLs and Send Emails from the Console | 99
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/Acme/DemoBundle/Command/DemoCommand.php
// ...
class DemoCommand extends ContainerAwareCommand
{
protected function execute(InputInterface $input, OutputInterface $output)
{
$context = $this->getContainer()->get('router')->getContext();
$context->setHost('example.com');
$context->setScheme('https');
$context->setBaseUrl('my/path');
// ... your code here
}
}
Using Memory Spooling
New in version 2.3: When using Symfony 2.3+ and SwiftmailerBundle 2.3.5+, the memory spool is
now handled automatically in the CLI and the code below is not necessary anymore.
Sending emails in a console command works the same way as described in the How to Send an Email
cookbook except if memory spooling is used.
When using memory spooling (see the How to Spool Emails cookbook for more information), you must
be aware that because of how Symfony handles console commands, emails are not sent automatically.
You must take care of flushing the queue yourself. Use the following code to send emails inside your
console command:
Listing 26-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
$message = new \Swift_Message();
// ... prepare the message
$container = $this->getContainer();
$mailer = $container->get('mailer');
$mailer->send($message);
// now manually flush the queue
$spool = $mailer->getTransport()->getSpool();
$transport = $container->get('swiftmailer.transport.real');
$spool->flushQueue($transport);
Another option is to create an environment which is only used by console commands and uses a different
spooling method.
Taking care of the spooling is only needed when memory spooling is used. If you are using file
spooling (or no spooling at all), there is no need to flush the queue manually within the command.
PDF brought to you by
generated on December 3, 2014
Chapter 26: How to Generate URLs and Send Emails from the Console | 100
Chapter 27
How to Enable Logging in Console Commands
The Console component doesn't provide any logging capabilities out of the box. Normally, you run
console commands manually and observe the output, which is why logging is not provided. However,
there are cases when you might need logging. For example, if you are running console commands
unattended, such as from cron jobs or deployment scripts, it may be easier to use Symfony's logging
capabilities instead of configuring other tools to gather console output and process it. This can be
especially handful if you already have some existing setup for aggregating and analyzing Symfony logs.
There are basically two logging cases you would need:
• Manually logging some information from your command;
• Logging uncaught Exceptions.
Manually Logging from a Console Command
This one is really simple. When you create a console command within the full framework as described
in "How to Create a Console Command", your command extends ContainerAwareCommand1. This means
that you can simply access the standard logger service through the container and use it to do the logging:
Listing 27-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/DemoBundle/Command/GreetCommand.php
namespace Acme\DemoBundle\Command;
use
use
use
use
use
use
Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;
Symfony\Component\Console\Input\InputArgument;
Symfony\Component\Console\Input\InputInterface;
Symfony\Component\Console\Input\InputOption;
Symfony\Component\Console\Output\OutputInterface;
Psr\Log\LoggerInterface;
class GreetCommand extends ContainerAwareCommand
{
// ...
1. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Command/ContainerAwareCommand.html
PDF brought to you by
generated on December 3, 2014
Chapter 27: How to Enable Logging in Console Commands | 101
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36 }
protected function execute(InputInterface $input, OutputInterface $output)
{
/** @var $logger LoggerInterface */
$logger = $this->getContainer()->get('logger');
$name = $input->getArgument('name');
if ($name) {
$text = 'Hello '.$name;
} else {
$text = 'Hello';
}
if ($input->getOption('yell')) {
$text = strtoupper($text);
$logger->warning('Yelled: '.$text);
} else {
$logger->info('Greeted: '.$text);
}
$output->writeln($text);
}
Depending on the environment in which you run your command (and your logging setup), you should
see the logged entries in app/logs/dev.log or app/logs/prod.log.
Enabling automatic Exceptions Logging
To get your console application to automatically log uncaught exceptions for all of your commands, you
can use console events.
New in version 2.3: Console events were introduced in Symfony 2.3.
First configure a listener for console exception events in the service container:
Listing 27-2
1 # app/config/services.yml
2 services:
3
kernel.listener.command_dispatch:
4
class: Acme\DemoBundle\EventListener\ConsoleExceptionListener
5
arguments:
6
logger: "@logger"
7
tags:
8
- { name: kernel.event_listener, event: console.exception }
Then implement the actual listener:
Listing 27-3
1
2
3
4
5
6
// src/Acme/DemoBundle/EventListener/ConsoleExceptionListener.php
namespace Acme\DemoBundle\EventListener;
use Symfony\Component\Console\Event\ConsoleExceptionEvent;
use Psr\Log\LoggerInterface;
PDF brought to you by
generated on December 3, 2014
Chapter 27: How to Enable Logging in Console Commands | 102
7 class ConsoleExceptionListener
8 {
9
private $logger;
10
11
public function __construct(LoggerInterface $logger)
12
{
13
$this->logger = $logger;
14
}
15
16
public function onConsoleException(ConsoleExceptionEvent $event)
17
{
18
$command = $event->getCommand();
19
$exception = $event->getException();
20
21
$message = sprintf(
22
'%s: %s (uncaught exception) at %s line %s while running console command `%s`',
23
get_class($exception),
24
$exception->getMessage(),
25
$exception->getFile(),
26
$exception->getLine(),
27
$command->getName()
28
);
29
30
$this->logger->error($message);
31
}
32 }
In the code above, when any command throws an exception, the listener will receive an event. You
can simply log it by passing the logger service via the service configuration. Your method receives
a ConsoleExceptionEvent2 object, which has methods to get information about the event and the
exception.
Logging non-0 Exit Statuses
The logging capabilities of the console can be further extended by logging non-0 exit statuses. This way
you will know if a command had any errors, even if no exceptions were thrown.
First configure a listener for console terminate events in the service container:
Listing 27-4
1 # app/config/services.yml
2 services:
3
kernel.listener.command_dispatch:
4
class: Acme\DemoBundle\EventListener\ConsoleTerminateListener
5
arguments:
6
logger: "@logger"
7
tags:
8
- { name: kernel.event_listener, event: console.terminate }
Then implement the actual listener:
Listing 27-5
1 // src/Acme/DemoBundle/EventListener/ConsoleExceptionListener.php
2 namespace Acme\DemoBundle\EventListener;
3
4 use Symfony\Component\Console\Event\ConsoleTerminateEvent;
2. http://api.symfony.com/master/Symfony/Component/Console/Event/ConsoleExceptionEvent.html
PDF brought to you by
generated on December 3, 2014
Chapter 27: How to Enable Logging in Console Commands | 103
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
use Psr\Log\LoggerInterface;
class ConsoleExceptionListener
{
private $logger;
public function __construct(LoggerInterface $logger)
{
$this->logger = $logger;
}
public function onConsoleTerminate(ConsoleTerminateEvent $event)
{
$statusCode = $event->getExitCode();
$command = $event->getCommand();
if ($statusCode === 0) {
return;
}
if ($statusCode > 255) {
$statusCode = 255;
$event->setExitCode($statusCode);
}
$this->logger->warning(sprintf(
'Command `%s` exited with status code %d',
$command->getName(),
$statusCode
));
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 27: How to Enable Logging in Console Commands | 104
Chapter 28
How to Define Commands as Services
New in version 2.4: Support for registering commands in the service container was introduced in
Symfony 2.4.
By default, Symfony will take a look in the Command directory of each bundle and automatically register
your commands. If a command extends the ContainerAwareCommand1, Symfony will even inject the
container. While making life easier, this has some limitations:
• Your command must live in the Command directory;
• There's no way to conditionally register your service based on the environment or availability
of some dependencies;
• You can't access the container in the configure() method (because setContainer hasn't been
called yet);
• You can't use the same class to create many commands (i.e. each with different configuration).
To solve these problems, you can register your command as a service and tag it with console.command:
Listing 28-1
1 # app/config/config.yml
2 services:
3
acme_hello.command.my_command:
4
class: Acme\HelloBundle\Command\MyCommand
5
tags:
6
- { name: console.command }
Using Dependencies and Parameters to Set Default Values for Options
Imagine you want to provide a default value for the name option. You could pass one of the following as
the 5th argument of addOption():
• a hardcoded string;
1. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Command/ContainerAwareCommand.html
PDF brought to you by
generated on December 3, 2014
Chapter 28: How to Define Commands as Services | 105
• a container parameter (e.g. something from parameters.yml);
• a value computed by a service (e.g. a repository).
By extending ContainerAwareCommand, only the first is possible, because you can't access the container
inside the configure() method. Instead, inject any parameter or service you need into the constructor.
For example, suppose you have some NameRepository service that you'll use to get your default value:
Listing 28-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
// src/Acme/DemoBundle/Command/GreetCommand.php
namespace Acme\DemoBundle\Command;
use
use
use
use
use
Acme\DemoBundle\Entity\NameRepository;
Symfony\Component\Console\Command\Command;
Symfony\Component\Console\Input\InputInterface;
Symfony\Component\Console\Input\InputOption;
Symfony\Component\Console\Output\OutputInterface;
class GreetCommand extends Command
{
protected $nameRepository;
public function __construct(NameRepository $nameRepository)
{
$this->nameRepository = $nameRepository;
parent::__construct();
}
protected function configure()
{
$defaultName = $this->nameRepository->findLastOne();
$this
->setName('demo:greet')
->setDescription('Greet someone')
->addOption('name', '-n', InputOption::VALUE_REQUIRED, 'Who do you want to
greet?', $defaultName)
;
}
protected function execute(InputInterface $input, OutputInterface $output)
{
$name = $input->getOption('name');
$output->writeln($name);
}
}
Now, just update the arguments of your service configuration like normal to inject the NameRepository.
Great, you now have a dynamic default value!
Be careful not to actually do any work in configure (e.g. make database queries), as your code will
be run, even if you're using the console to execute a different command.
PDF brought to you by
generated on December 3, 2014
Chapter 28: How to Define Commands as Services | 106
Chapter 29
How to Customize Error Pages
When an exception is thrown, the core HttpKernel class catches it and dispatches a kernel.exception
event. This gives you the power to convert the exception into a Response in a few different ways.
The core TwigBundle sets up a listener for this event which will run a configurable (but otherwise
arbitrary) controller to generate the response. The default controller used has a sensible way of picking
one out of the available set of error templates.
Thus, error pages can be customized in different ways, depending on how much control you need:
1. Use the default ExceptionController and create a few templates that allow you to customize how
your different error pages look (easy);
2. Replace the default exception controller with your own (intermediate).
3. Use the kernel.exception event to come up with your own handling (advanced).
Using the Default ExceptionController
By default, the showAction() method of the ExceptionController1 will be called when an exception
occurs.
This controller will either display an exception or error page, depending on the setting of the
kernel.debug flag. While exception pages give you a lot of helpful information during development,
error pages are meant to be shown to the user in production.
You can also preview your error pages in kernel.debug mode.
How the Template for the Error and Exception Pages Is Selected
The TwigBundle contains some default templates for error and exception pages in its Resources/views/
Exception directory.
1. http://api.symfony.com/master/Symfony/Bundle/TwigBundle/Controller/ExceptionController.html
PDF brought to you by
generated on December 3, 2014
Chapter 29: How to Customize Error Pages | 107
In a standard Symfony installation, the TwigBundle can be found at vendor/symfony/symfony/
src/Symfony/Bundle/TwigBundle. In addition to the standard HTML error page, it also provides
a default error page for many of the most common response formats, including JSON
(error.json.twig), XML (error.xml.twig) and even JavaScript (error.js.twig), to name a
few.
Here is how the ExceptionController will pick one of the available templates based on the HTTP status
code and request format:
• For error pages, it first looks for a template for the given format and status code (like
error404.json.twig);
• If that does not exist or apply, it looks for a general template for the given format (like
error.json.twig or exception.json.twig);
• Finally, it ignores the format and falls back to the HTML template (like error.html.twig or
exception.html.twig).
If the exception being handled implements the HttpExceptionInterface2, the getStatusCode()
method will be called to obtain the HTTP status code to use. Otherwise, the status code will be
"500".
Overriding or Adding Templates
To override these templates, simply rely on the standard method for overriding templates that live inside
a bundle. For more information, see Overriding Bundle Templates.
For example, to override the default error template, create a new template located at app/Resources/
TwigBundle/views/Exception/error.html.twig:
Listing 29-1
1
2
3
4
5
6
7
8
9
10
11
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>An Error Occurred: {{ status_text }}</title>
</head>
<body>
<h1>Oops! An Error Occurred</h1>
<h2>The server returned a "{{ status_code }} {{ status_text }}".</h2>
</body>
</html>
You must not use is_granted in your error pages (or layout used by your error pages), because
the router runs before the firewall. If the router throws an exception (for instance, when the route
does not match), then using is_granted will throw a further exception. You can use is_granted
safely by saying {% if app.user and is_granted('...') %}.
If you're not familiar with Twig, don't worry. Twig is a simple, powerful and optional templating
engine that integrates with Symfony. For more information about Twig see Creating and Using
Templates.
2. http://api.symfony.com/master/Symfony/Component/HttpKernel/Exception/HttpExceptionInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 29: How to Customize Error Pages | 108
This works not only to replace the default templates, but also to add new ones.
For instance, create an app/Resources/TwigBundle/views/Exception/error404.html.twig template
to display a special page for 404 (page not found) errors. Refer to the previous section for the order in
which the ExceptionController tries different template names.
Often, the easiest way to customize an error page is to copy it from the TwigBundle into app/
Resources/TwigBundle/views/Exception and then modify it.
The debug-friendly exception pages shown to the developer can even be customized in the same
way by creating templates such as exception.html.twig for the standard HTML exception page
or exception.json.twig for the JSON exception page.
Testing Error Pages during Development
The default ExceptionController also allows you to preview your error pages during development.
New in version 2.6: This feature was introduced in Symfony 2.6. Before, the third-party
WebfactoryExceptionsBundle3 could be used for the same purpose.
To use this feature, you need to have a definition in your routing_dev.yml file like so:
Listing 29-2
1 # app/config/routing_dev.yml
2 _errors:
3
resource: "@TwigBundle/Resources/config/routing/errors.xml"
4
prefix:
/_error
If you're coming from an older version of Symfony, you might need to add this to your routing_dev.yml
file. If you're starting from scratch, the Symfony Standard Edition4 already contains it for you.
With this route added, you can use URLs like
Listing 29-3
1 http://localhost/app_dev.php/_error/{statusCode}
2 http://localhost/app_dev.php/_error/{statusCode}.{format}
to preview the error page for a given status code as HTML or for a given status code and format.
Replacing the Default ExceptionController
If you need a little more flexibility beyond just overriding the template, then you can change the
controller that renders the error page. For example, you might need to pass some additional variables
into your template.
Make sure you don't lose the exception pages that render the helpful error messages during
development.
3. https://github.com/webfactory/exceptions-bundle
4. https://github.com/symfony/symfony-standard/
PDF brought to you by
generated on December 3, 2014
Chapter 29: How to Customize Error Pages | 109
To do this, simply create a new controller and set the twig.exception_controller option to point to it.
Listing 29-4
1 # app/config/config.yml
2 twig:
3
exception_controller:
AcmeFooBundle:Exception:showException
You can also set up your controller as a service.
The default value of twig.controller.exception:showAction refers to the showAction method
of the ExceptionController described previously, which is registered in the DIC as the
twig.controller.exception service.
Your controller will be passed two parameters: exception, which is a FlattenException5 instance
created from the exception being handled, and logger, an instance of DebugLoggerInterface6 (which
may be null).
The Request that will be dispatched to your controller is created in the ExceptionListener7. This
event listener is set up by the TwigBundle.
You can, of course, also extend the previously described ExceptionController8. In that case, you might
want to override one or both of the showAction and findTemplate methods. The latter one locates the
template to be used.
As of writing, the ExceptionController is not part of the Symfony API, so be aware that it might
change in following releases.
The error page preview also works for your own controllers set up this way.
Working with the kernel.exception Event
As mentioned in the beginning, the kernel.exception event is dispatched whenever the Symfony Kernel
needs to handle an exception. For more information on that, see kernel.exception Event.
Working with this event is actually much more powerful than what has been explained before but also
requires a thorough understanding of Symfony internals.
To give one example, assume your application throws specialized exceptions with a particular meaning
to your domain.
In that case, all the default ExceptionListener and ExceptionController could do for you was trying
to figure out the right HTTP status code and display your nice-looking error page.
5. http://api.symfony.com/master//Symfony/Component/Debug/Exception/FlattenException.html
6. http://api.symfony.com/master//Symfony/Component/HttpKernel/Log/DebugLoggerInterface.html
7. http://api.symfony.com/master/Symfony/Component/HttpKernel/EventListener/ExceptionListener.html
8. http://api.symfony.com/master/Symfony/Bundle/TwigBundle/Controller/ExceptionController.html
PDF brought to you by
generated on December 3, 2014
Chapter 29: How to Customize Error Pages | 110
Writing your own event listener for the kernel.exception event allows you to have a closer look at the
exception and take different actions depending on it. Those actions might include logging the exception,
redirecting the user to another page or rendering specialized error pages.
If your listener calls setResponse() on the GetResponseForExceptionEvent9, event propagation
will be stopped and the response will be sent to the client.
This approach allows you to create centralized and layered error handling: Instead of catching (and
handling) the same exceptions in various controllers again and again, you can have just one (or several)
listeners deal with them.
To see an example, have a look at the ExceptionListener10 in the Security Component.
It handles various security-related exceptions that are thrown in your application (like
AccessDeniedException11) and takes measures like redirecting the user to the login page, logging
them out and other things.
Good luck!
9. http://api.symfony.com/master/Symfony/Component/HttpKernel/Event/GetResponseForExceptionEvent.html
10. https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Security/Http/Firewall/ExceptionListener.php
11. http://api.symfony.com/master/Symfony/Component/Security/Core/Exception/AccessDeniedException.html
PDF brought to you by
generated on December 3, 2014
Chapter 29: How to Customize Error Pages | 111
Chapter 30
How to Define Controllers as Services
In the book, you've learned how easily a controller can be used when it extends the base Controller1
class. While this works fine, controllers can also be specified as services.
Specifying a controller as a service takes a little bit more work. The primary advantage is that the
entire controller or any services passed to the controller can be modified via the service container
configuration. This is especially useful when developing an open-source bundle or any bundle that
will be used in many different projects.
A second advantage is that your controllers are more "sandboxed". By looking at the constructor
arguments, it's easy to see what types of things this controller may or may not do. And because
each dependency needs to be injected manually, it's more obvious (i.e. if you have many
constructor arguments) when your controller has become too big, and may need to be split into
multiple controllers.
So, even if you don't specify your controllers as services, you'll likely see this done in some opensource Symfony bundles. It's also important to understand the pros and cons of both approaches.
Defining the Controller as a Service
A controller can be defined as a service in the same way as any other class. For example, if you have the
following simple controller:
Listing 30-1
1
2
3
4
5
6
7
8
9
// src/Acme/HelloBundle/Controller/HelloController.php
namespace Acme\HelloBundle\Controller;
use Symfony\Component\HttpFoundation\Response;
class HelloController
{
public function indexAction($name)
{
1. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Controller/Controller.html
PDF brought to you by
generated on December 3, 2014
Chapter 30: How to Define Controllers as Services | 112
10
11
12 }
return new Response('<html><body>Hello '.$name.'!</body></html>');
}
Then you can define it as a service as follows:
Listing 30-2
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 parameters:
3
# ...
4
acme.controller.hello.class: Acme\HelloBundle\Controller\HelloController
5
6 services:
7
acme.hello.controller:
8
class: "%acme.controller.hello.class%"
Referring to the Service
To refer to a controller that's defined as a service, use the single colon (:) notation. For example, to
forward to the indexAction() method of the service defined above with the id acme.hello.controller:
Listing 30-3
1 $this->forward('acme.hello.controller:indexAction', array('name' => $name));
You cannot drop the Action part of the method name when using this syntax.
You can also route to the service by using the same notation when defining the route _controller value:
Listing 30-4
1 # app/config/routing.yml
2 hello:
3
path:
/hello
4
defaults: { _controller: acme.hello.controller:indexAction }
You can also use annotations to configure routing using a controller defined as a service. See the
FrameworkExtraBundle documentation2 for details.
New in version 2.6: If your controller service implements the __invoke method, you can simply
refer to the service id (acme.hello.controller).
Alternatives to base Controller Methods
When using a controller defined as a service, it will most likely not extend the base Controller class.
Instead of relying on its shortcut methods, you'll interact directly with the services that you need.
2. http://symfony.com/doc/current/bundles/SensioFrameworkExtraBundle/annotations/routing.html
PDF brought to you by
generated on December 3, 2014
Chapter 30: How to Define Controllers as Services | 113
Fortunately, this is usually pretty easy and the base Controller class source code3 is a great source on how
to perform many common tasks.
For example, if you want to render a template instead of creating the Response object directly, then your
code would look like this if you were extending Symfony's base controller:
Listing 30-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/Acme/HelloBundle/Controller/HelloController.php
namespace Acme\HelloBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
class HelloController extends Controller
{
public function indexAction($name)
{
return $this->render(
'AcmeHelloBundle:Hello:index.html.twig',
array('name' => $name)
);
}
}
If you look at the source code for the render function in Symfony's base Controller class4, you'll see that
this method actually uses the templating service:
Listing 30-6
1 public function render($view, array $parameters = array(), Response $response = null)
2 {
3
return $this->container->get('templating')->renderResponse($view, $parameters,
4 $response);
}
In a controller that's defined as a service, you can instead inject the templating service and use it directly:
Listing 30-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/HelloBundle/Controller/HelloController.php
namespace Acme\HelloBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Templating\EngineInterface;
use Symfony\Component\HttpFoundation\Response;
class HelloController
{
private $templating;
public function __construct(EngineInterface $templating)
{
$this->templating = $templating;
}
public function indexAction($name)
{
return $this->templating->renderResponse(
'AcmeHelloBundle:Hello:index.html.twig',
array('name' => $name)
);
}
}
3. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/Controller/Controller.php
4. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/Controller/Controller.php
PDF brought to you by
generated on December 3, 2014
Chapter 30: How to Define Controllers as Services | 114
The service definition also needs modifying to specify the constructor argument:
Listing 30-8
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 parameters:
3
# ...
4
acme.controller.hello.class: Acme\HelloBundle\Controller\HelloController
5
6 services:
7
acme.hello.controller:
8
class:
"%acme.controller.hello.class%"
9
arguments: ["@templating"]
Rather than fetching the templating service from the container, you can inject only the exact service(s)
that you need directly into the controller.
This does not mean that you cannot extend these controllers from your own base controller. The
move away from the standard base controller is because its helper methods rely on having the
container available which is not the case for controllers that are defined as services. It may be a
good idea to extract common code into a service that's injected rather than place that code into a
base controller that you extend. Both approaches are valid, exactly how you want to organize your
reusable code is up to you.
PDF brought to you by
generated on December 3, 2014
Chapter 30: How to Define Controllers as Services | 115
Chapter 31
How to Optimize your Development
Environment for Debugging
When you work on a Symfony project on your local machine, you should use the dev environment
(app_dev.php front controller). This environment configuration is optimized for two main purposes:
• Give the developer accurate feedback whenever something goes wrong (web debug toolbar,
nice exception pages, profiler, ...);
• Be as similar as possible as the production environment to avoid problems when deploying the
project.
Disabling the Bootstrap File and Class Caching
And to make the production environment as fast as possible, Symfony creates big PHP files in your
cache containing the aggregation of PHP classes your project needs for every request. However, this
behavior can confuse your IDE or your debugger. This recipe shows you how you can tweak this caching
mechanism to make it friendlier when you need to debug code that involves Symfony classes.
The app_dev.php front controller reads as follows by default:
Listing 31-1
1
2
3
4
5
6
7
8
// ...
$loader = require_once __DIR__.'/../app/bootstrap.php.cache';
require_once __DIR__.'/../app/AppKernel.php';
$kernel = new AppKernel('dev', true);
$kernel->loadClassCache();
$request = Request::createFromGlobals();
To make your debugger happier, disable all PHP class caches by removing the call to loadClassCache()
and by replacing the require statements like below:
Listing 31-2
PDF brought to you by
generated on December 3, 2014
Chapter 31: How to Optimize your Development Environment for Debugging | 116
1
2
3
4
5
6
7
8
9
// ...
// $loader = require_once __DIR__.'/../app/bootstrap.php.cache';
$loader = require_once __DIR__.'/../app/autoload.php';
require_once __DIR__.'/../app/AppKernel.php';
$kernel = new AppKernel('dev', true);
// $kernel->loadClassCache();
$request = Request::createFromGlobals();
If you disable the PHP caches, don't forget to revert after your debugging session.
Some IDEs do not like the fact that some classes are stored in different locations. To avoid problems, you
can either tell your IDE to ignore the PHP cache files, or you can change the extension used by Symfony
for these files:
Listing 31-3
1 $kernel->loadClassCache('classes', '.php.cache');
PDF brought to you by
generated on December 3, 2014
Chapter 31: How to Optimize your Development Environment for Debugging | 117
Chapter 32
How to Deploy a Symfony Application
Deploying can be a complex and varied task depending on your setup and needs. This entry
doesn't try to explain everything, but rather offers the most common requirements and ideas for
deployment.
Symfony Deployment Basics
The typical steps taken while deploying a Symfony application include:
1. Upload your modified code to the live server;
2. Update your vendor dependencies (typically done via Composer, and may be done before
uploading);
3. Running database migrations or similar tasks to update any changed data structures;
4. Clearing (and perhaps more importantly, warming up) your cache.
A deployment may also include other things, such as:
•
•
•
•
•
Tagging a particular version of your code as a release in your source control repository;
Creating a temporary staging area to build your updated setup "offline";
Running any tests available to ensure code and/or server stability;
Removal of any unnecessary files from web to keep your production environment clean;
Clearing of external cache systems (like Memcached1 or Redis2).
How to Deploy a Symfony Application
There are several ways you can deploy a Symfony application.
Start with a few basic deployment strategies and build up from there.
1. http://memcached.org/
2. http://redis.io/
PDF brought to you by
generated on December 3, 2014
Chapter 32: How to Deploy a Symfony Application | 118
Basic File Transfer
The most basic way of deploying an application is copying the files manually via ftp/scp (or similar
method). This has its disadvantages as you lack control over the system as the upgrade progresses. This
method also requires you to take some manual steps after transferring the files (see Common PostDeployment Tasks)
Using Source Control
If you're using source control (e.g. Git or SVN), you can simplify by having your live installation also be a
copy of your repository. When you're ready to upgrade it is as simple as fetching the latest updates from
your source control system.
This makes updating your files easier, but you still need to worry about manually taking other steps (see
Common Post-Deployment Tasks).
Using Build Scripts and other Tools
There are also high-quality tools to help ease the pain of deployment. There are even a few tools which
have been specifically tailored to the requirements of Symfony, and which take special care to ensure that
everything before, during, and after a deployment has gone correctly.
See The Tools for a list of tools that can help with deployment.
Common Post-Deployment Tasks
After deploying your actual source code, there are a number of common things you'll need to do:
A) Check Requirements
Check if your server meets the requirements by running:
Listing 32-1
1 $ php app/check.php
B) Configure your app/config/parameters.yml File
This file should be customized on each system. The method you use to deploy your source code should
not deploy this file. Instead, you should set it up manually (or via some build process) on your server(s).
C) Update your Vendors
Your vendors can be updated before transferring your source code (i.e. update the vendor/ directory,
then transfer that with your source code) or afterwards on the server. Either way, just update your
vendors as you normally do:
Listing 32-2
1 $ php composer.phar install --no-dev --optimize-autoloader
The --optimize-autoloader flag makes Composer's autoloader more performant by building
a "class map". The --no-dev flag ensures that development packages are not installed in the
production environment.
PDF brought to you by
generated on December 3, 2014
Chapter 32: How to Deploy a Symfony Application | 119
If you get a "class not found" error during this step, you may need to run export
SYMFONY_ENV=prod before running this command so that the post-install-cmd scripts run in the
prod environment.
D) Clear your Symfony Cache
Make sure you clear (and warm-up) your Symfony cache:
Listing 32-3
1 $ php app/console cache:clear --env=prod --no-debug
E) Dump your Assetic Assets
If you're using Assetic, you'll also want to dump your assets:
Listing 32-4
1 $ php app/console assetic:dump --env=prod --no-debug
F) Other Things!
There may be lots of other things that you need to do, depending on your setup:
•
•
•
•
•
•
Running any database migrations
Clearing your APC cache
Running assets:install (taken care of already in composer.phar install)
Add/edit CRON jobs
Pushing assets to a CDN
...
Application Lifecycle: Continuous Integration, QA, etc
While this entry covers the technical details of deploying, the full lifecycle of taking code from
development up to production may have a lot more steps (think deploying to staging, QA, running tests,
etc).
The use of staging, testing, QA, continuous integration, database migrations and the capability to roll
back in case of failure are all strongly advised. There are simple and more complex tools and one can
make the deployment as easy (or sophisticated) as your environment requires.
Don't forget that deploying your application also involves updating any dependency (typically via
Composer), migrating your database, clearing your cache and other potential things like pushing assets
to a CDN (see Common Post-Deployment Tasks).
The Tools
Capifony3:
This tool provides a specialized set of tools on top of Capistrano, tailored specifically to symfony
and Symfony projects.
3. http://capifony.org/
PDF brought to you by
generated on December 3, 2014
Chapter 32: How to Deploy a Symfony Application | 120
sf2debpkg4:
This tool helps you build a native Debian package for your Symfony project.
Magallanes5:
This Capistrano-like deployment tool is built in PHP, and may be easier for PHP developers to
extend for their needs.
Bundles:
There are many bundles that add deployment features6 directly into your Symfony console.
Basic scripting:
You can of course use shell, Ant7, or any other build tool to script the deploying of your project.
Platform as a Service Providers:
PaaS is a relatively new way to deploy your application. Typically a PaaS will use a single
configuration file in your project's root directory to determine how to build an environment on the
fly that supports your software. One provider with confirmed Symfony support is PagodaBox8.
Looking for more? Talk to the community on the Symfony IRC channel9 #symfony (on freenode)
for more information.
4. https://github.com/liip/sf2debpkg
5. https://github.com/andres-montanez/Magallanes
6. http://knpbundles.com/search?q=deploy
7. http://blog.sznapka.pl/deploying-symfony2-applications-with-ant
8. https://github.com/jmather/pagoda-symfony-sonata-distribution/blob/master/Boxfile
9. http://webchat.freenode.net/?channels=symfony
PDF brought to you by
generated on December 3, 2014
Chapter 32: How to Deploy a Symfony Application | 121
Chapter 33
Deploying to Microsoft Azure Website Cloud
This step by step cookbook describes how to deploy a small Symfony web application to the Microsoft
Azure Website cloud platform. It will explain how to setup a new Azure website including configuring the
right PHP version and global environment variables. The document also shows how to you can leverage
Git and Composer to deploy your Symfony application to the cloud.
Setting up the Azure Website
To setup a new Microsoft Azure Website, first signup with Azure1 or sign in with your credentials. Once
you're connected to your Azure Portal2 interface, scroll down to the bottom and select the New panel.
On this panel, click Web Site and choose Custom Create:
1. https://signup.live.com/signup.aspx
2. https://manage.windowsazure.com
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 122
Step 1: Create Web Site
Here, you will be prompted to fill in some basic information.
For the URL, enter the URL that you would like to use for your Symfony application, then pick Create
new web hosting plan in the region you want. By default, a free 20 MB SQL database is selected in the
database dropdown list. In this tutorial, the Symfony app will connect to a MySQL database. Pick the
Create a new MySQL database option in the dropdown list. You can keep the DefaultConnection
string name. Finally, check the box Publish from source control to enable a Git repository and go to
the next step.
Step 2: New MySQL Database
On this step, you will be prompted to setup your MySQL database storage with a database name and a
region. The MySQL database storage is provided by Microsoft in partnership with ClearDB. Choose the
same region you selected for the hosting plan configuration in the previous step.
Agree to the terms and conditions and click on the right arrow to continue.
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 123
Step 3: Where Is your Source Code
Now, on the third step, select a Local Git repository item and click on the right arrow to configure your
Azure Website credentials.
Step 4: New Username and Password
Great! You're now on the final step. Create a username and a secure password: these will become
essential identifiers to connect to the FTP server and also to push your application code to the Git
repository.
Congratulations! Your Azure Website is now up and running. You can check it by browsing to the
Website url you configured in the first step. You should see the following display in your web browser:
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 124
The Microsoft Azure portal also provides a complete control panel for the Azure Website.
Your Azure Website is ready! But to run a Symfony site, you need to configure just a few additional
things.
Configuring the Azure Website for Symfony
This section of the tutorial details how to configure the correct version of PHP to run Symfony. It also
shows you how to enable some mandatory PHP extensions and how to properly configure PHP for a
production environment.
Configuring the latest PHP Runtime
Even though Symfony only requires PHP 5.3.3 to run, it's always recommended to use the most recent
PHP version whenever possible. PHP 5.3 is no longer supported by the PHP core team, but you can
update it easily in Azure.
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 125
To update your PHP version on Azure, go to the Configure tab of the control panel and select the version
you want.
Click the Save button in the bottom bar to save your changes and restart the web server.
Choosing a more recent PHP version can greatly improve runtime performance. PHP 5.5 ships with
a new built-in PHP accelerator called OPCache that replaces APC. On an Azure Website, OPCache
is already enabled and there is no need to install and setup APC.
The following screenshot shows the output of a phpinfo3 script run from an Azure Website to
verify that PHP 5.5 is running with OPCache enabled.
Tweaking php.ini Configuration Settings
Microsoft Azure allows you to override the php.ini global configuration settings by creating a custom
.user.ini file under the project root directory (site/wwwroot).
Listing 33-1
3. http://php.net/manual/en/function.phpinfo.php
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 126
1
2
3
4
; .user.ini
expose_php = Off
memory_limit = 256M
upload_max_filesize = 10M
None of these settings needs to be overridden. The default PHP configuration is already pretty good, so
this is just an example to show how you can easily tweak PHP internal settings by uploading your custom
.ini file.
You can either manually create this file on your Azure Website FTP server under the site/wwwroot
directory or deploy it with Git. You can get your FTP server credentials from the Azure Website Control
panel under the Dashboard tab on the right sidebar. If you want to use Git, simply put your .user.ini
file at the root of your local repository and push your commits to your Azure Website repository.
This cookbook has a section dedicated to explaining how to configure your Azure Website Git
repository and how to push the commits to be deployed. See Deploying from Git. You can also
learn more about configuring PHP internal settings on the official PHP MSDN documentation4
page.
Enabling the PHP intl Extension
This is the tricky part of the guide! At the time of writing this cookbook, Microsoft Azure Website
provided the intl extension, but it's not enabled by default. To enable the intl extension, there is no
need to upload any DLL files as the php_intl.dll file already exists on Azure. In fact, this file just needs
to be moved into the custom website extension directory.
The Microsoft Azure team is currently working on enabling the intl PHP extension by default. In
the near future, the following steps will no longer be necessary.
To get the php_intl.dll file under your site/wwwroot directory, simply access the online Kudu tool by
browsing to the following url:
Listing 33-2
1 https://[your-website-name].scm.azurewebsites.net
Kudu is a set of tools to manage your application. It comes with a file explorer, a command line prompt,
a log stream and a configuration settings summary page. Of course, this section can only be accessed if
you're logged in to your main Azure Website account.
4. http://blogs.msdn.com/b/silverlining/archive/2012/07/10/configuring-php-in-windows-azure-websites-with-user-ini-files.aspx
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 127
From the Kudu front page, click on the Debug Console navigation item in the main menu and choose
CMD. This should open the Debug Console page that shows a file explorer and a console prompt
below.
In the console prompt, type the following three commands to copy the original php_intl.dll extension
file into a custom website ext/ directory. This new directory must be created under the main directory
site/wwwroot.
Listing 33-3
1 $ cd site\wwwroot
2 $ mkdir ext
3 $ copy "D:\Program Files (x86)\PHP\v5.5\ext\php_intl.dll" ext
The whole process and output should look like this:
To complete the activation of the php_intl.dll extension, you must tell Azure Website to load it from
the newly created ext directory. This can be done by registering a global PHP_EXTENSIONS environment
variable from the Configure tab of the main Azure Website Control panel.
In the app settings section, register the PHP_EXTENSIONS environment variable with the value
ext\php_intl.dll as shown in the screenshot below:
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 128
Hit "save" to confirm your changes and restart the web server. The PHP Intl extension should now be
available in your web server environment. The following screenshot of a phpinfo5 page verifies the intl
extension is properly enabled:
Great! The PHP environment setup is now complete. Next, you'll learn how to configure the Git
repository and push code to production. You'll also learn how to install and configure the Symfony app
after it's deployed.
Deploying from Git
First, make sure Git is correctly installed on your local machine using the following command in your
terminal:
Listing 33-4
1 $ git --version
5. http://php.net/manual/en/function.phpinfo.php
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 129
Get your Git from the git-scm.com6 website and follow the instructions to install and configure it
on your local machine.
In the Azure Website Control panel, browse the Deployment tab to get the Git repository URL where
you should push your code:
Now, you'll want to connect your local Symfony application with this remote Git repository on Azure
Website. If your Symfony application is not yet stored with Git, you must first create a Git repository in
your Symfony application directory with the git init command and commit to it with the git commit
command.
Also, make sure your Symfony repository has a .gitignore file at its root directory with at least the
following contents:
Listing 33-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
/app/bootstrap.php.cache
/app/cache/*
/app/config/parameters.yml
/app/logs/*
!app/cache/.gitkeep
!app/logs/.gitkeep
/app/SymfonyRequirements.php
/build/
/vendor/
/bin/
/composer.phar
/web/app_dev.php
/web/bundles/
/web/config.php
The .gitignore file asks Git not to track any of the files and directories that match these patterns. This
means these files won't be deployed to the Azure Website.
Now, from the command line on your local machine, type the following at the root of your Symfony
project:
Listing 33-6
6. http://git-scm.com/download
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 130
1 $ git remote add azure https://<username>@<your-website-name>.scm.azurewebsites.net:443/
2 <your-website-name>.git
$ git push azure master
Don't forget to replace the values enclosed by < and > with your custom settings displayed in the
Deployment tab of your Azure Website panel. The git remote command connects the Azure Website
remote Git repository and assigns an alias to it with the name azure. The second git push command
pushes all your commits to the remote master branch of your remote azure Git repository.
The deployment with Git should produce an output similar to the screenshot below:
The code of the Symfony application has now been deployed to the Azure Website which you can browse
from the file explorer of the Kudu application. You should see the app/, src/ and web/ directories under
your site/wwwroot directory on the Azure Website filesystem.
Configure the Symfony Application
PHP has been configured and your code has been pushed with Git. The last step is to configure the
application and install the third party dependencies it requires that aren't tracked by Git. Switch back to
the online Console of the Kudu application and execute the following commands in it:
Listing 33-7
1 $ cd site\wwwroot
2 $ curl -sS https://getcomposer.org/installer | php
3 $ php -d extension=php_intl.dll composer.phar install
The curl command retrieves and downloads the Composer command line tool and installs it at the root
of the site/wwwroot directory. Then, running the Composer install command downloads and installs
all necessary third-party libraries.
This may take a while depending on the number of third-party dependencies you've configured in your
composer.json file.
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 131
The -d switch allows you to quickly override/add any php.ini settings. In this command, we are
forcing PHP to use the intl extension, because it is not enabled by default in Azure Website at
the moment. Soon, this -d option will no longer be needed since Microsoft will enable the intl
extension by default.
At the end of the composer install command, you will be prompted to fill in the values of some
Symfony settings like database credentials, locale, mailer credentials, CSRF token protection, etc. These
parameters come from the app/config/parameters.yml.dist file.
The most important thing in this cookbook is to correctly setup your database settings. You can get your
MySQL database settings on the right sidebar of the Azure Website Dashboard panel. Simply click on
the View Connection Strings link to make them appear in a pop-in.
The displayed MySQL database settings should be something similar to the code below. Of course, each
value depends on what you've already configured.
Listing 33-8
1 Database=mysymfonyMySQL;Data Source=eu-cdbr-azure-north-c.cloudapp.net;User
Id=bff2481a5b6074;Password=bdf50b42
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 132
Switch back to the console and answer the prompted questions and provide the following answers. Don't
forget to adapt the values below with your real values from the MySQL connection string.
Listing 33-9
1
2
3
4
5
6
7
database_driver: pdo_mysql
database_host: u-cdbr-azure-north-c.cloudapp.net
database_port: null
database_name: mysymfonyMySQL
database_user: bff2481a5b6074
database_password: bdf50b42
// ...
Don't forget to answer all the questions. It's important to set a unique random string for the secret
variable. For the mailer configuration, Azure Website doesn't provide a built-in mailer service. You
should consider configuring the host-name and credentials of some other third-party mailing service if
your application needs to send emails.
Your Symfony application is now configured and should be almost operational. The final step is to build
the database schema. This can easily be done with the command line interface if you're using Doctrine.
In the online Console tool of the Kudu application, run the following command to mount the tables into
your MySQL database.
Listing 33-10
1 $ php app/console doctrine:schema:update --force
This command builds the tables and indexes for your MySQL database. If your Symfony application is
more complex than a basic Symfony Standard Edition, you may have additional commands to execute
for setup (see How to Deploy a Symfony Application).
Make sure that your application is running by browsing the app.php front controller with your web
browser and the following url:
Listing 33-11
1 http://<your-website-name>.azurewebsites.net/web/app.php
If Symfony is correctly installed, you should see the front page of your Symfony application showing.
Configure the Web Server
At this point, the Symfony application has been deployed and works perfectly on the Azure Website.
However, the web folder is still part of the url, which you definitely don't want. But don't worry! You can
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 133
easily configure the web server to point to the web folder and remove the web in the URL (and guarantee
that nobody can access files outside of the web directory.)
To do this, create and deploy (see previous section about Git) the following web.config file. This file
must be located at the root of your project next to the composer.json file. This file is the Microsoft IIS
Server equivalent to the well-known .htaccess file from Apache. For a Symfony application, configure
it with the following content:
Listing 33-12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
<!-- web.config -->
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
<system.webServer>
<rewrite>
<rules>
<clear />
<rule name="BlockAccessToPublic" patternSyntax="Wildcard" stopProcessing="true">
<match url="*" />
<conditions logicalGrouping="MatchAll" trackAllCaptures="false">
<add input="{URL}" pattern="/web/*" />
</conditions>
<action type="CustomResponse" statusCode="403" statusReason="Forbidden: Access
is denied." statusDescription="You do not have permission to view this directory or page
using the credentials that you supplied." />
</rule>
<rule name="RewriteAssetsToPublic" stopProcessing="true">
<match url="^(.*)(\.css|\.js|\.jpg|\.png|\.gif)$" />
<conditions logicalGrouping="MatchAll" trackAllCaptures="false">
</conditions>
<action type="Rewrite" url="web/{R:0}" />
</rule>
<rule name="RewriteRequestsToPublic" stopProcessing="true">
<match url="^(.*)$" />
<conditions logicalGrouping="MatchAll" trackAllCaptures="false">
</conditions>
<action type="Rewrite" url="web/app.php/{R:0}" />
</rule>
</rules>
</rewrite>
</system.webServer>
</configuration>
As you can see, the latest rule RewriteRequestsToPublic is responsible for rewriting any urls to the
web/app.php front controller which allows you to skip the web/ folder in the URL. The first rule called
BlockAccessToPublic matches all url patterns that contain the web/ folder and serves a 403 Forbidden
HTTP response instead. This example is based on Benjamin Eberlei's sample you can find on GitHub in
the SymfonyAzureEdition7 bundle.
Deploy this file under the site/wwwroot directory of the Azure Website and browse to your application
without the web/app.php segment in the URL.
Conclusion
Nice work! You've now deployed your Symfony application to the Microsoft Azure Website Cloud
platform. You also saw that Symfony can be easily configured and executed on a Microsoft IIS web server.
The process is simple and easy to implement. And as a bonus, Microsoft is continuing to reduce the
number of steps needed so that deployment becomes even easier.
7. https://github.com/beberlei/symfony-azure-edition/
PDF brought to you by
generated on December 3, 2014
Chapter 33: Deploying to Microsoft Azure Website Cloud | 134
Chapter 34
Deploying to Heroku Cloud
This step by step cookbook describes how to deploy a Symfony web application to the Heroku cloud
platform. Its contents are based on the original article1 published by Heroku.
Setting up
To setup a new Heroku website, first signup with Heroku2 or sign in with your credentials. Then
download and install the Heroku Toolbelt3 on your local computer.
You can also check out the getting Started with PHP on Heroku4 guide to gain more familiarity with the
specifics of working with PHP applications on Heroku.
Preparing your Application
Deploying a Symfony application to Heroku doesn't require any change in its code, but it requires some
minor tweaks to its configuration.
By default, the Symfony app will log into your application's app/log/ directory. This is not ideal as
Heroku uses an ephemeral file system5. On Heroku, the best way to handle logging is using Logplex6. And
the best way to send log data to Logplex is by writing to STDERR or STDOUT. Luckily, Symfony uses the
excellent Monolog library for logging. So, a new log destination is just a change to a config file away.
Open the app/config/config_prod.yml file, locate the monolog/handlers/nested section (or create
it if it doesn't exist yet) and change the value of path from "%kernel.logs_dir%/
%kernel.environment%.log" to "php://stderr":
Listing 34-1
1 # app/config/config_prod.yml
2 monolog:
1. https://devcenter.heroku.com/articles/getting-started-with-symfony2
2. https://signup.heroku.com/signup/dc
3. https://devcenter.heroku.com/articles/getting-started-with-php#local-workstation-setup
4. https://devcenter.heroku.com/articles/getting-started-with-php
5. https://devcenter.heroku.com/articles/dynos#ephemeral-filesystem
6. https://devcenter.heroku.com/articles/logplex
PDF brought to you by
generated on December 3, 2014
Chapter 34: Deploying to Heroku Cloud | 135
3
4
5
6
7
8
# ...
handlers:
# ...
nested:
# ...
path: "php://stderr"
Once the application is deployed, run heroku logs --tail to keep the stream of logs from Heroku open
in your terminal.
Creating a new Application on Heroku
To create a new Heroku application that you can push to, use the CLI create command:
Listing 34-2
1
2
3
4
5
$ heroku create
Creating mighty-hamlet-1981 in organization heroku... done, stack is cedar
http://mighty-hamlet-1981.herokuapp.com/ | [email protected]:mighty-hamlet-1981.git
Git remote heroku added
You are now ready to deploy the application as explained in the next section.
Deploying your Application on Heroku
To deploy your application to Heroku, you must first create a Procfile, which tells Heroku what
command to use to launch the web server with the correct settings. After you've done that, you can simply
git push and you're done!
Creating a Procfile
By default, Heroku will launch an Apache web server together with PHP to serve applications. However,
two special circumstances apply to Symfony applications:
1. The document root is in the web/ directory and not in the root directory of the application;
2. The Composer bin-dir, where vendor binaries (and thus Heroku's own boot scripts) are
placed, is bin/ , and not the default vendor/bin.
Vendor binaries are usually installed to vendor/bin by Composer, but sometimes (e.g. when
running a Symfony Standard Edition project!), the location will be different. If in doubt, you can
always run composer config bin-dir to figure out the right location.
Create a new file called Procfile (without any extension) at the root directory of the application and add
just the following content:
Listing 34-3
1 web: bin/heroku-php-apache2 web/
If you prefer working on the command console, execute the following commands to create the Procfile
file and to add it to the repository:
Listing 34-4
PDF brought to you by
generated on December 3, 2014
Chapter 34: Deploying to Heroku Cloud | 136
1
2
3
4
5
$ echo "web: bin/heroku-php-apache2 web/" > Procfile
$ git add .
$ git commit -m "Procfile for Apache and PHP"
[master 35075db] Procfile for Apache and PHP
1 file changed, 1 insertion(+)
Pushing to Heroku
Next up, it's finally time to deploy your application to Heroku. If you are doing this for the very first time,
you may see a message such as the following:
Listing 34-5
The authenticity of host 'heroku.com (50.19.85.132)' can't be established.
RSA key fingerprint is 8b:48:5e:67:0e:c9:16:47:32:f2:87:0c:1f:c8:60:ad.
Are you sure you want to continue connecting (yes/no)?
In this case, you need to confirm by typing yes and hitting <Enter> key - ideally after you've verified that
the RSA key fingerprint is correct7.
Then, deploy your application executing this command:
Listing 34-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
$ git push heroku master
Initializing repository, done.
Counting objects: 130, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (107/107), done.
Writing objects: 100% (130/130), 70.88 KiB | 0 bytes/s, done.
Total 130 (delta 17), reused 0 (delta 0)
-----> PHP app detected
-----> Setting up runtime environment...
- PHP 5.5.12
- Apache 2.4.9
- Nginx 1.4.6
-----> Installing PHP extensions:
- opcache (automatic; bundled, using 'ext-opcache.ini')
-----> Installing dependencies...
Composer version 64ac32fca9e64eb38e50abfadc6eb6f2d0470039 2014-05-24 20:57:50
Loading composer repositories with package information
Installing dependencies from lock file
- ...
Generating optimized autoload files
Creating the "app/config/parameters.yml" file
Clearing the cache for the dev environment with debug true
Installing assets using the hard copy option
Installing assets for Symfony\Bundle\FrameworkBundle into web/bundles/framework
Installing assets for Acme\DemoBundle into web/bundles/acmedemo
Installing assets for Sensio\Bundle\DistributionBundle into web/bundles/
sensiodistribution
-----> Building runtime environment...
7. https://devcenter.heroku.com/articles/git-repository-ssh-fingerprints
PDF brought to you by
generated on December 3, 2014
Chapter 34: Deploying to Heroku Cloud | 137
36
37
38
39
40
41
42
43
44
45
-----> Discovering process types
Procfile declares types -> web
-----> Compressing... done, 61.5MB
-----> Launching... done, v3
http://mighty-hamlet-1981.herokuapp.com/ deployed to Heroku
To [email protected]:mighty-hamlet-1981.git
* [new branch]
master -> master
And that's it! If you now open your browser, either by manually pointing it to the URL heroku create
gave you, or by using the Heroku Toolbelt, the application will respond:
Listing 34-7
1 $ heroku open
2 Opening mighty-hamlet-1981... done
You should be seeing your Symfony application in your browser.
PDF brought to you by
generated on December 3, 2014
Chapter 34: Deploying to Heroku Cloud | 138
Chapter 35
How to Handle File Uploads with Doctrine
Handling file uploads with Doctrine entities is no different than handling any other file upload. In other
words, you're free to move the file in your controller after handling a form submission. For examples of
how to do this, see the file type reference page.
If you choose to, you can also integrate the file upload into your entity lifecycle (i.e. creation, update and
removal). In this case, as your entity is created, updated, and removed from Doctrine, the file uploading
and removal processing will take place automatically (without needing to do anything in your controller).
To make this work, you'll need to take care of a number of details, which will be covered in this cookbook
entry.
Basic Setup
First, create a simple Doctrine entity class to work with:
Listing 35-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/Acme/DemoBundle/Entity/Document.php
namespace Acme\DemoBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
use Symfony\Component\Validator\Constraints as Assert;
/**
* @ORM\Entity
*/
class Document
{
/**
* @ORM\Id
* @ORM\Column(type="integer")
* @ORM\GeneratedValue(strategy="AUTO")
*/
public $id;
/**
* @ORM\Column(type="string", length=255)
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 139
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57 }
* @Assert\NotBlank
*/
public $name;
/**
* @ORM\Column(type="string", length=255, nullable=true)
*/
public $path;
public function getAbsolutePath()
{
return null === $this->path
? null
: $this->getUploadRootDir().'/'.$this->path;
}
public function getWebPath()
{
return null === $this->path
? null
: $this->getUploadDir().'/'.$this->path;
}
protected function getUploadRootDir()
{
// the absolute directory path where uploaded
// documents should be saved
return __DIR__.'/../../../../web/'.$this->getUploadDir();
}
protected function getUploadDir()
{
// get rid of the __DIR__ so it doesn't screw up
// when displaying uploaded doc/image in the view.
return 'uploads/documents';
}
The Document entity has a name and it is associated with a file. The path property stores the relative path
to the file and is persisted to the database. The getAbsolutePath() is a convenience method that returns
the absolute path to the file while the getWebPath() is a convenience method that returns the web path,
which can be used in a template to link to the uploaded file.
If you have not done so already, you should probably read the file type documentation first to
understand how the basic upload process works.
If you're using annotations to specify your validation rules (as shown in this example), be sure that
you've enabled validation by annotation (see validation configuration).
To handle the actual file upload in the form, use a "virtual" file field. For example, if you're building
your form directly in a controller, it might look like this:
Listing 35-2
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 140
1 public function uploadAction()
2 {
3
// ...
4
5
$form = $this->createFormBuilder($document)
6
->add('name')
7
->add('file')
8
->getForm();
9
10
// ...
11 }
Next, create this property on your Document class and add some validation rules:
Listing 35-3
Listing 35-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
use Symfony\Component\HttpFoundation\File\UploadedFile;
// ...
class Document
{
/**
* @Assert\File(maxSize="6000000")
*/
private $file;
/**
* Sets file.
*
* @param UploadedFile $file
*/
public function setFile(UploadedFile $file = null)
{
$this->file = $file;
}
/**
* Get file.
*
* @return UploadedFile
*/
public function getFile()
{
return $this->file;
}
}
1 # src/Acme/DemoBundle/Resources/config/validation.yml
2 Acme\DemoBundle\Entity\Document:
3
properties:
4
file:
5
- File:
6
maxSize: 6000000
As you are using the File constraint, Symfony will automatically guess that the form field is a file
upload input. That's why you did not have to set it explicitly when creating the form above (>add('file')).
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 141
The following controller shows you how to handle the entire process:
Listing 35-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// ...
use Acme\DemoBundle\Entity\Document;
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Template;
use Symfony\Component\HttpFoundation\Request;
// ...
/**
* @Template()
*/
public function uploadAction(Request $request)
{
$document = new Document();
$form = $this->createFormBuilder($document)
->add('name')
->add('file')
->getForm();
$form->handleRequest($request);
if ($form->isValid()) {
$em = $this->getDoctrine()->getManager();
$em->persist($document);
$em->flush();
return $this->redirect($this->generateUrl(...));
}
return array('form' => $form->createView());
}
The previous controller will automatically persist the Document entity with the submitted name, but it
will do nothing about the file and the path property will be blank.
An easy way to handle the file upload is to move it just before the entity is persisted and then set the path
property accordingly. Start by calling a new upload() method on the Document class, which you'll create
in a moment to handle the file upload:
Listing 35-6
1 if ($form->isValid()) {
2
$em = $this->getDoctrine()->getManager();
3
4
$document->upload();
5
6
$em->persist($document);
7
$em->flush();
8
9
return $this->redirect(...);
10 }
The upload() method will take advantage of the UploadedFile1 object, which is what's returned after a
file field is submitted:
Listing 35-7
1 public function upload()
2 {
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/File/UploadedFile.html
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 142
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23 }
// the file property can be empty if the field is not required
if (null === $this->getFile()) {
return;
}
// use the original file name here but you should
// sanitize it at least to avoid any security issues
// move takes the target directory and then the
// target filename to move to
$this->getFile()->move(
$this->getUploadRootDir(),
$this->getFile()->getClientOriginalName()
);
// set the path property to the filename where you've saved the file
$this->path = $this->getFile()->getClientOriginalName();
// clean up the file property as you won't need it anymore
$this->file = null;
Using Lifecycle Callbacks
Using lifecycle callbacks is a limited technique that has some drawbacks. If you want to remove the
hardcoded __DIR__ reference inside the Document::getUploadRootDir() method, the best way is
to start using explicit doctrine listeners. There you will be able to inject kernel parameters such as
kernel.root_dir to be able to build absolute paths.
Even if this implementation works, it suffers from a major flaw: What if there is a problem when the
entity is persisted? The file would have already moved to its final location even though the entity's path
property didn't persist correctly.
To avoid these issues, you should change the implementation so that the database operation and the
moving of the file become atomic: if there is a problem persisting the entity or if the file cannot be moved,
then nothing should happen.
To do this, you need to move the file right as Doctrine persists the entity to the database. This can be
accomplished by hooking into an entity lifecycle callback:
Listing 35-8
1
2
3
4
5
6
7
/**
* @ORM\Entity
* @ORM\HasLifecycleCallbacks
*/
class Document
{
}
Next, refactor the Document class to take advantage of these callbacks:
Listing 35-9
1 use Symfony\Component\HttpFoundation\File\UploadedFile;
2
3 /**
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 143
4 * @ORM\Entity
5 * @ORM\HasLifecycleCallbacks
6 */
7 class Document
8 {
9
private $temp;
10
11
/**
12
* Sets file.
13
*
14
* @param UploadedFile $file
15
*/
16
public function setFile(UploadedFile $file = null)
17
{
18
$this->file = $file;
19
// check if we have an old image path
20
if (isset($this->path)) {
21
// store the old name to delete after the update
22
$this->temp = $this->path;
23
$this->path = null;
24
} else {
25
$this->path = 'initial';
26
}
27
}
28
29
/**
30
* @ORM\PrePersist()
31
* @ORM\PreUpdate()
32
*/
33
public function preUpload()
34
{
35
if (null !== $this->getFile()) {
36
// do whatever you want to generate a unique name
37
$filename = sha1(uniqid(mt_rand(), true));
38
$this->path = $filename.'.'.$this->getFile()->guessExtension();
39
}
40
}
41
42
/**
43
* @ORM\PostPersist()
44
* @ORM\PostUpdate()
45
*/
46
public function upload()
47
{
48
if (null === $this->getFile()) {
49
return;
50
}
51
52
// if there is an error when moving the file, an exception will
53
// be automatically thrown by move(). This will properly prevent
54
// the entity from being persisted to the database on error
55
$this->getFile()->move($this->getUploadRootDir(), $this->path);
56
57
// check if we have an old image
58
if (isset($this->temp)) {
59
// delete the old image
60
unlink($this->getUploadRootDir().'/'.$this->temp);
61
// clear the temp image path
62
$this->temp = null;
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 144
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77 }
}
$this->file = null;
}
/**
* @ORM\PostRemove()
*/
public function removeUpload()
{
$file = $this->getAbsolutePath();
if ($file) {
unlink($file);
}
}
If changes to your entity are handled by a Doctrine event listener or event subscriber, the
preUpdate() callback must notify Doctrine about the changes being done. For full reference on
preUpdate event restrictions, see preUpdate2 in the Doctrine Events documentation.
The class now does everything you need: it generates a unique filename before persisting, moves the file
after persisting, and removes the file if the entity is ever deleted.
Now that the moving of the file is handled atomically by the entity, the call to $document->upload()
should be removed from the controller:
Listing 35-10
1 if ($form->isValid()) {
2
$em = $this->getDoctrine()->getManager();
3
4
$em->persist($document);
5
$em->flush();
6
7
return $this->redirect(...);
8 }
The @ORM\PrePersist() and @ORM\PostPersist() event callbacks are triggered before and after
the entity is persisted to the database. On the other hand, the @ORM\PreUpdate() and
@ORM\PostUpdate() event callbacks are called when the entity is updated.
The PreUpdate and PostUpdate callbacks are only triggered if there is a change in one of the
entity's fields that are persisted. This means that, by default, if you modify only the $file property,
these events will not be triggered, as the property itself is not directly persisted via Doctrine. One
solution would be to use an updated field that's persisted to Doctrine, and to modify it manually
when changing the file.
2. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html#preupdate
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 145
Using the id as the Filename
If you want to use the id as the name of the file, the implementation is slightly different as you need to
save the extension under the path property, instead of the actual filename:
Listing 35-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
use Symfony\Component\HttpFoundation\File\UploadedFile;
/**
* @ORM\Entity
* @ORM\HasLifecycleCallbacks
*/
class Document
{
private $temp;
/**
* Sets file.
*
* @param UploadedFile $file
*/
public function setFile(UploadedFile $file = null)
{
$this->file = $file;
// check if we have an old image path
if (is_file($this->getAbsolutePath())) {
// store the old name to delete after the update
$this->temp = $this->getAbsolutePath();
} else {
$this->path = 'initial';
}
}
/**
* @ORM\PrePersist()
* @ORM\PreUpdate()
*/
public function preUpload()
{
if (null !== $this->getFile()) {
$this->path = $this->getFile()->guessExtension();
}
}
/**
* @ORM\PostPersist()
* @ORM\PostUpdate()
*/
public function upload()
{
if (null === $this->getFile()) {
return;
}
// check if we have an old image
if (isset($this->temp)) {
// delete the old image
unlink($this->temp);
// clear the temp image path
$this->temp = null;
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 146
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92 }
}
// you must throw an exception here if the file cannot be moved
// so that the entity is not persisted to the database
// which the UploadedFile move() method does
$this->getFile()->move(
$this->getUploadRootDir(),
$this->id.'.'.$this->getFile()->guessExtension()
);
$this->setFile(null);
}
/**
* @ORM\PreRemove()
*/
public function storeFilenameForRemove()
{
$this->temp = $this->getAbsolutePath();
}
/**
* @ORM\PostRemove()
*/
public function removeUpload()
{
if (isset($this->temp)) {
unlink($this->temp);
}
}
public function getAbsolutePath()
{
return null === $this->path
? null
: $this->getUploadRootDir().'/'.$this->id.'.'.$this->path;
}
You'll notice in this case that you need to do a little bit more work in order to remove the file. Before it's
removed, you must store the file path (since it depends on the id). Then, once the object has been fully
removed from the database, you can safely delete the file (in PostRemove).
PDF brought to you by
generated on December 3, 2014
Chapter 35: How to Handle File Uploads with Doctrine | 147
Chapter 36
How to use Doctrine Extensions:
Timestampable, Sluggable, Translatable, etc.
Doctrine2 is very flexible, and the community has already created a series of useful Doctrine extensions
to help you with common entity-related tasks.
One library in particular - the DoctrineExtensions1 library - provides integration functionality for
Sluggable2, Translatable3, Timestampable4, Loggable5, Tree6 and Sortable7 behaviors.
The usage for each of these extensions is explained in that repository.
However, to install/activate each extension you must register and activate an Event Listener. To do this,
you have two options:
1. Use the StofDoctrineExtensionsBundle8, which integrates the above library.
2. Implement this services directly by following the documentation for integration with Symfony:
Install Gedmo Doctrine2 extensions in Symfony29
1. https://github.com/Atlantic18/DoctrineExtensions
2. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/sluggable.md
3. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/translatable.md
4. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/timestampable.md
5. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/loggable.md
6. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/tree.md
7. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/sortable.md
8. https://github.com/stof/StofDoctrineExtensionsBundle
9. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/symfony2.md
PDF brought to you by
generated on December 3, 2014
Chapter 36: How to use Doctrine Extensions: Timestampable, Sluggable, Translatable, etc. | 148
Chapter 37
How to Register Event Listeners and
Subscribers
Doctrine packages a rich event system that fires events when almost anything happens inside the system.
For you, this means that you can create arbitrary services and tell Doctrine to notify those objects
whenever a certain action (e.g. prePersist) happens within Doctrine. This could be useful, for example,
to create an independent search index whenever an object in your database is saved.
Doctrine defines two types of objects that can listen to Doctrine events: listeners and subscribers. Both
are very similar, but listeners are a bit more straightforward. For more, see The Event System1 on
Doctrine's website.
The Doctrine website also explains all existing events that can be listened to.
Configuring the Listener/Subscriber
To register a service to act as an event listener or subscriber you just have to tag it with the appropriate
name. Depending on your use-case, you can hook a listener into every DBAL connection and ORM entity
manager or just into one specific DBAL connection and all the entity managers that use this connection.
Listing 37-1
1 doctrine:
2
dbal:
3
default_connection: default
4
connections:
5
default:
6
driver: pdo_sqlite
7
memory: true
8
9 services:
10
my.listener:
11
class: Acme\SearchBundle\EventListener\SearchIndexer
12
tags:
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html
PDF brought to you by
generated on December 3, 2014
Chapter 37: How to Register Event Listeners and Subscribers | 149
13
14
15
16
17
18
19
20
21
- { name: doctrine.event_listener, event: postPersist }
my.listener2:
class: Acme\SearchBundle\EventListener\SearchIndexer2
tags:
- { name: doctrine.event_listener, event: postPersist, connection: default }
my.subscriber:
class: Acme\SearchBundle\EventListener\SearchIndexerSubscriber
tags:
- { name: doctrine.event_subscriber, connection: default }
Creating the Listener Class
In the previous example, a service my.listener was configured as a Doctrine listener on the event
postPersist. The class behind that service must have a postPersist method, which will be called when
the event is dispatched:
Listing 37-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/Acme/SearchBundle/EventListener/SearchIndexer.php
namespace Acme\SearchBundle\EventListener;
use Doctrine\ORM\Event\LifecycleEventArgs;
use Acme\StoreBundle\Entity\Product;
class SearchIndexer
{
public function postPersist(LifecycleEventArgs $args)
{
$entity = $args->getEntity();
$entityManager = $args->getEntityManager();
// perhaps you only want to act on some "Product" entity
if ($entity instanceof Product) {
// ... do something with the Product
}
}
}
In each event, you have access to a LifecycleEventArgs object, which gives you access to both the entity
object of the event and the entity manager itself.
One important thing to notice is that a listener will be listening for all entities in your application. So,
if you're interested in only handling a specific type of entity (e.g. a Product entity but not a BlogPost
entity), you should check for the entity's class type in your method (as shown above).
In Doctrine 2.4, a feature called Entity Listeners was introduced. It is a lifecycle listener class used
for an entity. You can read about it in the Doctrine Documentation2.
Creating the Subscriber Class
A Doctrine event subscriber must implement the Doctrine\Common\EventSubscriber interface and
have an event method for each event it subscribes to:
2. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html#entity-listeners
PDF brought to you by
generated on December 3, 2014
Chapter 37: How to Register Event Listeners and Subscribers | 150
Listing 37-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
// src/Acme/SearchBundle/EventListener/SearchIndexerSubscriber.php
namespace Acme\SearchBundle\EventListener;
use Doctrine\Common\EventSubscriber;
use Doctrine\ORM\Event\LifecycleEventArgs;
// for Doctrine 2.4: Doctrine\Common\Persistence\Event\LifecycleEventArgs;
use Acme\StoreBundle\Entity\Product;
class SearchIndexerSubscriber implements EventSubscriber
{
public function getSubscribedEvents()
{
return array(
'postPersist',
'postUpdate',
);
}
public function postUpdate(LifecycleEventArgs $args)
{
$this->index($args);
}
public function postPersist(LifecycleEventArgs $args)
{
$this->index($args);
}
public function index(LifecycleEventArgs $args)
{
$entity = $args->getEntity();
$entityManager = $args->getEntityManager();
// perhaps you only want to act on some "Product" entity
if ($entity instanceof Product) {
// ... do something with the Product
}
}
}
Doctrine event subscribers can not return a flexible array of methods to call for the events like the
Symfony event subscriber can. Doctrine event subscribers must return a simple array of the event
names they subscribe to. Doctrine will then expect methods on the subscriber with the same name
as each subscribed event, just as when using an event listener.
For a full reference, see chapter The Event System3 in the Doctrine documentation.
3. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html
PDF brought to you by
generated on December 3, 2014
Chapter 37: How to Register Event Listeners and Subscribers | 151
Chapter 38
How to Use Doctrine DBAL
This article is about the Doctrine DBAL. Typically, you'll work with the higher level Doctrine ORM
layer, which simply uses the DBAL behind the scenes to actually communicate with the database.
To read more about the Doctrine ORM, see "Databases and Doctrine".
The Doctrine1 Database Abstraction Layer (DBAL) is an abstraction layer that sits on top of PDO2 and
offers an intuitive and flexible API for communicating with the most popular relational databases. In
other words, the DBAL library makes it easy to execute queries and perform other database actions.
Read the official Doctrine DBAL Documentation3 to learn all the details and capabilities of
Doctrine's DBAL library.
To get started, configure the database connection parameters:
Listing 38-1
1 # app/config/config.yml
2 doctrine:
3
dbal:
4
driver:
pdo_mysql
5
dbname:
Symfony
6
user:
root
7
password: null
8
charset: UTF8
For full DBAL configuration options, or to learn how to configure multiple connections, see Doctrine
DBAL Configuration.
You can then access the Doctrine DBAL connection by accessing the database_connection service:
Listing 38-2
1. http://www.doctrine-project.org
2. http://www.php.net/pdo
3. http://docs.doctrine-project.org/projects/doctrine-dbal/en/latest/index.html
PDF brought to you by
generated on December 3, 2014
Chapter 38: How to Use Doctrine DBAL | 152
1 class UserController extends Controller
2 {
3
public function indexAction()
4
{
5
$conn = $this->get('database_connection');
6
$users = $conn->fetchAll('SELECT * FROM users');
7
8
// ...
9
}
10 }
Registering custom Mapping Types
You can register custom mapping types through Symfony's configuration. They will be added to all
configured connections. For more information on custom mapping types, read Doctrine's Custom
Mapping Types4 section of their documentation.
Listing 38-3
1 # app/config/config.yml
2 doctrine:
3
dbal:
4
types:
5
custom_first: Acme\HelloBundle\Type\CustomFirst
6
custom_second: Acme\HelloBundle\Type\CustomSecond
Registering custom Mapping Types in the SchemaTool
The SchemaTool is used to inspect the database to compare the schema. To achieve this task, it needs to
know which mapping type needs to be used for each database types. Registering new ones can be done
through the configuration.
Now, map the ENUM type (not supported by DBAL by default) to the string mapping type:
Listing 38-4
1 # app/config/config.yml
2 doctrine:
3
dbal:
4
mapping_types:
5
enum: string
4. http://docs.doctrine-project.org/projects/doctrine-dbal/en/latest/reference/types.html#custom-mapping-types
PDF brought to you by
generated on December 3, 2014
Chapter 38: How to Use Doctrine DBAL | 153
Chapter 39
How to Generate Entities from an Existing
Database
When starting work on a brand new project that uses a database, two different situations comes
naturally. In most cases, the database model is designed and built from scratch. Sometimes, however,
you'll start with an existing and probably unchangeable database model. Fortunately, Doctrine comes
with a bunch of tools to help generate model classes from your existing database.
As the Doctrine tools documentation1 says, reverse engineering is a one-time process to get started
on a project. Doctrine is able to convert approximately 70-80% of the necessary mapping
information based on fields, indexes and foreign key constraints. Doctrine can't discover inverse
associations, inheritance types, entities with foreign keys as primary keys or semantical operations
on associations such as cascade or lifecycle events. Some additional work on the generated entities
will be necessary afterwards to design each to fit your domain model specificities.
This tutorial assumes you're using a simple blog application with the following two tables: blog_post
and blog_comment. A comment record is linked to a post record thanks to a foreign key constraint.
Listing 39-1
1 CREATE TABLE `blog_post` (
2
`id` bigint(20) NOT NULL AUTO_INCREMENT,
3
`title` varchar(100) COLLATE utf8_unicode_ci NOT NULL,
4
`content` longtext COLLATE utf8_unicode_ci NOT NULL,
5
`created_at` datetime NOT NULL,
6
PRIMARY KEY (`id`)
7 ) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;
8
9 CREATE TABLE `blog_comment` (
10
`id` bigint(20) NOT NULL AUTO_INCREMENT,
11
`post_id` bigint(20) NOT NULL,
12
`author` varchar(20) COLLATE utf8_unicode_ci NOT NULL,
13
`content` longtext COLLATE utf8_unicode_ci NOT NULL,
14
`created_at` datetime NOT NULL,
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/tools.html#reverse-engineering
PDF brought to you by
generated on December 3, 2014
Chapter 39: How to Generate Entities from an Existing Database | 154
15
PRIMARY KEY (`id`),
16
KEY `blog_comment_post_id_idx` (`post_id`),
17
CONSTRAINT `blog_post_id` FOREIGN KEY (`post_id`) REFERENCES `blog_post` (`id`) ON
18 DELETE CASCADE
) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;
Before diving into the recipe, be sure your database connection parameters are correctly setup in the
app/config/parameters.yml file (or wherever your database configuration is kept) and that you have
initialized a bundle that will host your future entity class. In this tutorial it's assumed that an
AcmeBlogBundle exists and is located under the src/Acme/BlogBundle folder.
The first step towards building entity classes from an existing database is to ask Doctrine to introspect
the database and generate the corresponding metadata files. Metadata files describe the entity class to
generate based on table fields.
Listing 39-2
1 $ php app/console doctrine:mapping:import --force AcmeBlogBundle xml
This command line tool asks Doctrine to introspect the database and generate the XML metadata files
under the src/Acme/BlogBundle/Resources/config/doctrine folder of your bundle. This generates
two files: BlogPost.orm.xml and BlogComment.orm.xml.
It's also possible to generate the metadata files in YAML format by changing the last argument to
yml.
The generated BlogPost.orm.xml metadata file looks as follows:
Listing 39-3
1
2
3
4
5
6
7
8
9
10
11
<?xml version="1.0" encoding="utf-8"?>
<doctrine-mapping xmlns="http://doctrine-project.org/schemas/orm/doctrine-mapping"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://doctrine-project.org/schemas/orm/doctrine-mapping
http://doctrine-project.org/schemas/orm/doctrine-mapping.xsd">
<entity name="Acme\BlogBundle\Entity\BlogPost" table="blog_post">
<id name="id" type="bigint" column="id">
<generator strategy="IDENTITY"/>
</id>
<field name="title" type="string" column="title" length="100" nullable="false"/>
<field name="content" type="text" column="content" nullable="false"/>
<field name="createdAt" type="datetime" column="created_at" nullable="false"/>
</entity>
</doctrine-mapping>
Once the metadata files are generated, you can ask Doctrine to build related entity classes by executing
the following two commands.
Listing 39-4
1 $ php app/console doctrine:mapping:convert annotation ./src
2 $ php app/console doctrine:generate:entities AcmeBlogBundle
The first command generates entity classes with annotation mappings. But if you want to use YAML or
XML mapping instead of annotations, you should execute the second command only.
If you want to use annotations, you can safely delete the XML (or YAML) files after running these
two commands.
PDF brought to you by
generated on December 3, 2014
Chapter 39: How to Generate Entities from an Existing Database | 155
For example, the newly created BlogComment entity class looks as follow:
Listing 39-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
// src/Acme/BlogBundle/Entity/BlogComment.php
namespace Acme\BlogBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
/**
* Acme\BlogBundle\Entity\BlogComment
*
* @ORM\Table(name="blog_comment")
* @ORM\Entity
*/
class BlogComment
{
/**
* @var integer $id
*
* @ORM\Column(name="id", type="bigint")
* @ORM\Id
* @ORM\GeneratedValue(strategy="IDENTITY")
*/
private $id;
/**
* @var string $author
*
* @ORM\Column(name="author", type="string", length=100, nullable=false)
*/
private $author;
/**
* @var text $content
*
* @ORM\Column(name="content", type="text", nullable=false)
*/
private $content;
/**
* @var datetime $createdAt
*
* @ORM\Column(name="created_at", type="datetime", nullable=false)
*/
private $createdAt;
/**
* @var BlogPost
*
* @ORM\ManyToOne(targetEntity="BlogPost")
* @ORM\JoinColumn(name="post_id", referencedColumnName="id")
*/
private $post;
}
As you can see, Doctrine converts all table fields to pure private and annotated class properties. The most
impressive thing is that it also discovered the relationship with the BlogPost entity class based on the
foreign key constraint. Consequently, you can find a private $post property mapped with a BlogPost
entity in the BlogComment entity class.
PDF brought to you by
generated on December 3, 2014
Chapter 39: How to Generate Entities from an Existing Database | 156
If you want to have a one-to-many relationship, you will need to add it manually into the entity
or to the generated XML or YAML files. Add a section on the specific entities for one-to-many
defining the inversedBy and the mappedBy pieces.
The generated entities are now ready to be used. Have fun!
PDF brought to you by
generated on December 3, 2014
Chapter 39: How to Generate Entities from an Existing Database | 157
Chapter 40
How to Work with multiple Entity Managers
and Connections
You can use multiple Doctrine entity managers or connections in a Symfony application. This is
necessary if you are using different databases or even vendors with entirely different sets of entities. In
other words, one entity manager that connects to one database will handle some entities while another
entity manager that connects to another database might handle the rest.
Using multiple entity managers is pretty easy, but more advanced and not usually required. Be sure
you actually need multiple entity managers before adding in this layer of complexity.
The following configuration code shows how you can configure two entity managers:
Listing 40-1
1 doctrine:
2
dbal:
3
default_connection: default
4
connections:
5
default:
6
driver:
"%database_driver%"
7
host:
"%database_host%"
8
port:
"%database_port%"
9
dbname:
"%database_name%"
10
user:
"%database_user%"
11
password: "%database_password%"
12
charset: UTF8
13
customer:
14
driver:
"%database_driver2%"
15
host:
"%database_host2%"
16
port:
"%database_port2%"
17
dbname:
"%database_name2%"
18
user:
"%database_user2%"
19
password: "%database_password2%"
20
charset: UTF8
PDF brought to you by
generated on December 3, 2014
Chapter 40: How to Work with multiple Entity Managers and Connections | 158
21
22
23
24
25
26
27
28
29
30
31
32
33
orm:
default_entity_manager: default
entity_managers:
default:
connection: default
mappings:
AcmeDemoBundle: ~
AcmeStoreBundle: ~
customer:
connection: customer
mappings:
AcmeCustomerBundle: ~
In this case, you've defined two entity managers and called them default and customer. The default
entity manager manages entities in the AcmeDemoBundle and AcmeStoreBundle, while the customer
entity manager manages entities in the AcmeCustomerBundle. You've also defined two connections, one
for each entity manager.
When working with multiple connections and entity managers, you should be explicit about which
configuration you want. If you do omit the name of the connection or entity manager, the default
(i.e. default) is used.
When working with multiple connections to create your databases:
Listing 40-2
1
2
3
4
5
# Play only with "default" connection
$ php app/console doctrine:database:create
# Play only with "customer" connection
$ php app/console doctrine:database:create --connection=customer
When working with multiple entity managers to update your schema:
Listing 40-3
1
2
3
4
5
# Play only with "default" mappings
$ php app/console doctrine:schema:update --force
# Play only with "customer" mappings
$ php app/console doctrine:schema:update --force --em=customer
If you do omit the entity manager's name when asking for it, the default entity manager (i.e. default) is
returned:
Listing 40-4
1 class UserController extends Controller
2 {
3
public function indexAction()
4
{
5
// All three return the "default" entity manager
6
$em = $this->get('doctrine')->getManager();
7
$em = $this->get('doctrine')->getManager('default');
8
$em = $this->get('doctrine.orm.default_entity_manager');
9
10
// Both of these return the "customer" entity manager
11
$customerEm = $this->get('doctrine')->getManager('customer');
12
$customerEm = $this->get('doctrine.orm.customer_entity_manager');
PDF brought to you by
generated on December 3, 2014
Chapter 40: How to Work with multiple Entity Managers and Connections | 159
13
14 }
}
You can now use Doctrine just as you did before - using the default entity manager to persist and fetch
entities that it manages and the customer entity manager to persist and fetch its entities.
The same applies to repository calls:
Listing 40-5
1 class UserController extends Controller
2 {
3
public function indexAction()
4
{
5
// Retrieves a repository managed by the "default" em
6
$products = $this->get('doctrine')
7
->getRepository('AcmeStoreBundle:Product')
8
->findAll()
9
;
10
11
// Explicit way to deal with the "default" em
12
$products = $this->get('doctrine')
13
->getRepository('AcmeStoreBundle:Product', 'default')
14
->findAll()
15
;
16
17
// Retrieves a repository managed by the "customer" em
18
$customers = $this->get('doctrine')
19
->getRepository('AcmeCustomerBundle:Customer', 'customer')
20
->findAll()
21
;
22
}
23 }
PDF brought to you by
generated on December 3, 2014
Chapter 40: How to Work with multiple Entity Managers and Connections | 160
Chapter 41
How to Register custom DQL Functions
Doctrine allows you to specify custom DQL functions. For more information on this topic, read
Doctrine's cookbook article "DQL User Defined Functions1".
In Symfony, you can register your custom DQL functions as follows:
Listing 41-1
1 # app/config/config.yml
2 doctrine:
3
orm:
4
# ...
5
dql:
6
string_functions:
7
test_string: Acme\HelloBundle\DQL\StringFunction
8
second_string: Acme\HelloBundle\DQL\SecondStringFunction
9
numeric_functions:
10
test_numeric: Acme\HelloBundle\DQL\NumericFunction
11
datetime_functions:
12
test_datetime: Acme\HelloBundle\DQL\DatetimeFunction
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/cookbook/dql-user-defined-functions.html
PDF brought to you by
generated on December 3, 2014
Chapter 41: How to Register custom DQL Functions | 161
Chapter 42
How to Define Relationships with Abstract
Classes and Interfaces
One of the goals of bundles is to create discreet bundles of functionality that do not have many (if any)
dependencies, allowing you to use that functionality in other applications without including unnecessary
items.
Doctrine 2.2 includes a new utility called the ResolveTargetEntityListener, that functions by
intercepting certain calls inside Doctrine and rewriting targetEntity parameters in your metadata
mapping at runtime. It means that in your bundle you are able to use an interface or abstract class in your
mappings and expect correct mapping to a concrete entity at runtime.
This functionality allows you to define relationships between different entities without making them hard
dependencies.
Background
Suppose you have an InvoiceBundle which provides invoicing functionality and a CustomerBundle that
contains customer management tools. You want to keep these separated, because they can be used in
other systems without each other, but for your application you want to use them together.
In this case, you have an Invoice entity with a relationship to a non-existent object, an
InvoiceSubjectInterface. The goal is to get the ResolveTargetEntityListener to replace any
mention of the interface with a real object that implements that interface.
Set up
This article uses the following two basic entities (which are incomplete for brevity) to explain how to set
up and use the ResolveTargetEntityListener.
A Customer entity:
Listing 42-1
PDF brought to you by
generated on December 3, 2014
Chapter 42: How to Define Relationships with Abstract Classes and Interfaces | 162
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/AppBundle/Entity/Customer.php
namespace Acme\AppBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
use Acme\CustomerBundle\Entity\Customer as BaseCustomer;
use Acme\InvoiceBundle\Model\InvoiceSubjectInterface;
/**
* @ORM\Entity
* @ORM\Table(name="customer")
*/
class Customer extends BaseCustomer implements InvoiceSubjectInterface
{
// In this example, any methods defined in the InvoiceSubjectInterface
// are already implemented in the BaseCustomer
}
An Invoice entity:
Listing 42-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/InvoiceBundle/Entity/Invoice.php
namespace Acme\InvoiceBundle\Entity;
use Doctrine\ORM\Mapping AS ORM;
use Acme\InvoiceBundle\Model\InvoiceSubjectInterface;
/**
* Represents an Invoice.
*
* @ORM\Entity
* @ORM\Table(name="invoice")
*/
class Invoice
{
/**
* @ORM\ManyToOne(targetEntity="Acme\InvoiceBundle\Model\InvoiceSubjectInterface")
* @var InvoiceSubjectInterface
*/
protected $subject;
}
An InvoiceSubjectInterface:
Listing 42-3
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/Acme/InvoiceBundle/Model/InvoiceSubjectInterface.php
namespace Acme\InvoiceBundle\Model;
/**
* An interface that the invoice Subject object should implement.
* In most circumstances, only a single object should implement
* this interface as the ResolveTargetEntityListener can only
* change the target to a single object.
*/
interface InvoiceSubjectInterface
{
// List any additional methods that your InvoiceBundle
PDF brought to you by
generated on December 3, 2014
Chapter 42: How to Define Relationships with Abstract Classes and Interfaces | 163
14
15
16
17
18
19
20
21 }
// will need to access on the subject so that you can
// be sure that you have access to those methods.
/**
* @return string
*/
public function getName();
Next, you need to configure the listener, which tells the DoctrineBundle about the replacement:
Listing 42-4
1 # app/config/config.yml
2 doctrine:
3
# ...
4
orm:
5
# ...
6
resolve_target_entities:
7
Acme\InvoiceBundle\Model\InvoiceSubjectInterface: Acme\AppBundle\Entity\Customer
Final Thoughts
With the ResolveTargetEntityListener, you are able to decouple your bundles, keeping them usable
by themselves, but still being able to define relationships between different objects. By using this method,
your bundles will end up being easier to maintain independently.
PDF brought to you by
generated on December 3, 2014
Chapter 42: How to Define Relationships with Abstract Classes and Interfaces | 164
Chapter 43
How to Provide Model Classes for several
Doctrine Implementations
When building a bundle that could be used not only with Doctrine ORM but also the CouchDB ODM,
MongoDB ODM or PHPCR ODM, you should still only write one model class. The Doctrine bundles
provide a compiler pass to register the mappings for your model classes.
For non-reusable bundles, the easiest option is to put your model classes in the default locations:
Entity for the Doctrine ORM or Document for one of the ODMs. For reusable bundles, rather than
duplicate model classes just to get the auto-mapping, use the compiler pass.
New in version 2.3: The base mapping compiler pass was introduced in Symfony 2.3. The Doctrine
bundles support it from DoctrineBundle >= 1.2.1, MongoDBBundle >= 3.0.0, PHPCRBundle
>= 1.0.0 and the (unversioned) CouchDBBundle supports the compiler pass since the CouchDB
Mapping Compiler Pass pull request1 was merged.
New in version 2.6: Support for defining namespace aliases was introduced in Symfony 2.6. It is
safe to define the aliases with older versions of Symfony as the aliases are the last argument to
createXmlMappingDriver and are ignored by PHP if that argument doesn't exist.
In your bundle class, write the following code to register the compiler pass. This one is written for the
CmfRoutingBundle, so parts of it will need to be adapted for your case:
Listing 43-1
1
2
3
4
5
6
use
use
use
use
Doctrine\Bundle\DoctrineBundle\DependencyInjection\Compiler\DoctrineOrmMappingsPass;
Doctrine\Bundle\MongoDBBundle\DependencyInjection\Compiler\DoctrineMongoDBMappingsPass;
Doctrine\Bundle\CouchDBBundle\DependencyInjection\Compiler\DoctrineCouchDBMappingsPass;
Doctrine\Bundle\PHPCRBundle\DependencyInjection\Compiler\DoctrinePhpcrMappingsPass;
class CmfRoutingBundle extends Bundle
1. https://github.com/doctrine/DoctrineCouchDBBundle/pull/27
PDF brought to you by
generated on December 3, 2014
Chapter 43: How to Provide Model Classes for several Doctrine Implementations | 165
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
{
public function build(ContainerBuilder $container)
{
parent::build($container);
// ...
$modelDir = realpath(__DIR__.'/Resources/config/doctrine/model');
$mappings = array(
$modelDir => 'Symfony\Cmf\RoutingBundle\Model',
);
$ormCompilerClass =
'Doctrine\Bundle\DoctrineBundle\DependencyInjection\Compiler\DoctrineOrmMappingsPass';
if (class_exists($ormCompilerClass)) {
$container->addCompilerPass(
DoctrineOrmMappingsPass::createXmlMappingDriver(
$mappings,
array('cmf_routing.model_manager_name'),
'cmf_routing.backend_type_orm',
array('CmfRoutingBundle' => 'Symfony\Cmf\RoutingBundle\Model')
));
}
$mongoCompilerClass =
'Doctrine\Bundle\MongoDBBundle\DependencyInjection\Compiler\DoctrineMongoDBMappingsPass';
if (class_exists($mongoCompilerClass)) {
$container->addCompilerPass(
DoctrineMongoDBMappingsPass::createXmlMappingDriver(
$mappings,
array('cmf_routing.model_manager_name'),
'cmf_routing.backend_type_mongodb',
array('CmfRoutingBundle' => 'Symfony\Cmf\RoutingBundle\Model')
));
}
$couchCompilerClass =
'Doctrine\Bundle\CouchDBBundle\DependencyInjection\Compiler\DoctrineCouchDBMappingsPass';
if (class_exists($couchCompilerClass)) {
$container->addCompilerPass(
DoctrineCouchDBMappingsPass::createXmlMappingDriver(
$mappings,
array('cmf_routing.model_manager_name'),
'cmf_routing.backend_type_couchdb',
array('CmfRoutingBundle' => 'Symfony\Cmf\RoutingBundle\Model')
));
}
$phpcrCompilerClass =
'Doctrine\Bundle\PHPCRBundle\DependencyInjection\Compiler\DoctrinePhpcrMappingsPass';
if (class_exists($phpcrCompilerClass)) {
$container->addCompilerPass(
DoctrinePhpcrMappingsPass::createXmlMappingDriver(
$mappings,
array('cmf_routing.model_manager_name'),
'cmf_routing.backend_type_phpcr',
array('CmfRoutingBundle' => 'Symfony\Cmf\RoutingBundle\Model')
));
}
PDF brought to you by
generated on December 3, 2014
Chapter 43: How to Provide Model Classes for several Doctrine Implementations | 166
}
}
Note the class_exists2 check. This is crucial, as you do not want your bundle to have a hard
dependency on all Doctrine bundles but let the user decide which to use.
The compiler pass provides factory methods for all drivers provided by Doctrine: Annotations, XML,
Yaml, PHP and StaticPHP. The arguments are:
• A map/hash of absolute directory path to namespace;
• An array of container parameters that your bundle uses to specify the name of the Doctrine
manager that it is using. In the example above, the CmfRoutingBundle stores the manager
name that's being used under the cmf_routing.model_manager_name parameter. The
compiler pass will append the parameter Doctrine is using to specify the name of the default
manager. The first parameter found is used and the mappings are registered with that manager;
• An optional container parameter name that will be used by the compiler pass to determine if
this Doctrine type is used at all. This is relevant if your user has more than one type of Doctrine
bundle installed, but your bundle is only used with one type of Doctrine;
• A map/hash of aliases to namespace. This should be the same convention used by Doctrine
auto-mapping. In the example above, this allows the user to call $om>getRepository('CmfRoutingBundle:Route').
2. http://php.net/manual/en/function.class-exists.php
PDF brought to you by
generated on December 3, 2014
Chapter 43: How to Provide Model Classes for several Doctrine Implementations | 167
The factory method is using the SymfonyFileLocator of Doctrine, meaning it will only see XML
and YML mapping files if they do not contain the full namespace as the filename. This is by design:
the SymfonyFileLocator simplifies things by assuming the files are just the "short" version of the
class as their filename (e.g. BlogPost.orm.xml)
If you also need to map a base class, you can register a compiler pass with the
DefaultFileLocator like this. This code is taken from the DoctrineOrmMappingsPass and
adapted to use the DefaultFileLocator instead of the SymfonyFileLocator:
Listing 43-2
1
2
3
4
5
6
7
8
9
10
11
12
13
private function buildMappingCompilerPass()
{
$arguments = array(array(realpath(__DIR__ . '/Resources/config/doctrine-base')),
'.orm.xml');
$locator = new
Definition('Doctrine\Common\Persistence\Mapping\Driver\DefaultFileLocator',
$arguments);
$driver = new Definition('Doctrine\ORM\Mapping\Driver\XmlDriver',
array($locator));
return new DoctrineOrmMappingsPass(
$driver,
array('Full\Namespace'),
array('your_bundle.manager_name'),
'your_bundle.orm_enabled'
);
}
Note that you do not need to provide a namespace alias unless your users are expected to ask
Doctrine for the base classes.
Now place your mapping file into /Resources/config/doctrine-base with the fully qualified
class name, separated by . instead of \, for example Other.Namespace.Model.Name.orm.xml. You
may not mix the two as otherwise the SymfonyFileLocator will get confused.
Adjust accordingly for the other Doctrine implementations.
PDF brought to you by
generated on December 3, 2014
Chapter 43: How to Provide Model Classes for several Doctrine Implementations | 168
Chapter 44
How to Implement a simple Registration Form
Some forms have extra fields whose values don't need to be stored in the database. For example, you
may want to create a registration form with some extra fields (like a "terms accepted" checkbox field) and
embed the form that actually stores the account information.
The simple User Model
You have a simple User entity mapped to the database:
Listing 44-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/Acme/AccountBundle/Entity/User.php
namespace Acme\AccountBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
use Symfony\Component\Validator\Constraints as Assert;
use Symfony\Bridge\Doctrine\Validator\Constraints\UniqueEntity;
/**
* @ORM\Entity
* @UniqueEntity(fields="email", message="Email already taken")
*/
class User
{
/**
* @ORM\Id
* @ORM\Column(type="integer")
* @ORM\GeneratedValue(strategy="AUTO")
*/
protected $id;
/**
* @ORM\Column(type="string", length=255)
* @Assert\NotBlank()
* @Assert\Email()
*/
protected $email;
PDF brought to you by
generated on December 3, 2014
Chapter 44: How to Implement a simple Registration Form | 169
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59 }
/**
* @ORM\Column(type="string", length=255)
* @Assert\NotBlank()
* @Assert\Length(max = 4096)
*/
protected $plainPassword;
public function getId()
{
return $this->id;
}
public function getEmail()
{
return $this->email;
}
public function setEmail($email)
{
$this->email = $email;
}
public function getPlainPassword()
{
return $this->plainPassword;
}
public function setPlainPassword($password)
{
$this->plainPassword = $password;
}
This User entity contains three fields and two of them (email and plainPassword) should display on the
form. The email property must be unique in the database, this is enforced by adding this validation at the
top of the class.
If you want to integrate this User within the security system, you need to implement the
UserInterface of the Security component.
Why the 4096 Password Limit?
Notice that the plainPassword field has a max length of 4096 characters. For security purposes
(CVE-2013-57501), Symfony limits the plain password length to 4096 characters when encoding
it. Adding this constraint makes sure that your form will give a validation error if anyone tries a
super-long password.
You'll need to add this constraint anywhere in your application where your user submits a plaintext
password (e.g. change password form). The only place where you don't need to worry about this
is your login form, since Symfony's Security component handles this for you.
1. http://symfony.com/blog/cve-2013-5750-security-issue-in-fosuserbundle-login-form
PDF brought to you by
generated on December 3, 2014
Chapter 44: How to Implement a simple Registration Form | 170
Create a Form for the Model
Next, create the form for the User model:
Listing 44-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
// src/Acme/AccountBundle/Form/Type/UserType.php
namespace Acme\AccountBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
class UserType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('email', 'email');
$builder->add('plainPassword', 'repeated', array(
'first_name' => 'password',
'second_name' => 'confirm',
'type'
=> 'password',
));
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'data_class' => 'Acme\AccountBundle\Entity\User'
));
}
public function getName()
{
return 'user';
}
}
There are just two fields: email and plainPassword (repeated to confirm the entered password). The
data_class option tells the form the name of the underlying data class (i.e. your User entity).
To explore more things about the Form component, read Forms.
Embedding the User Form into a Registration Form
The form that you'll use for the registration page is not the same as the form used to simply modify the
User (i.e. UserType). The registration form will contain further fields like "accept the terms", whose value
won't be stored in the database.
Start by creating a simple class which represents the "registration":
Listing 44-3
1 // src/Acme/AccountBundle/Form/Model/Registration.php
2 namespace Acme\AccountBundle\Form\Model;
3
4 use Symfony\Component\Validator\Constraints as Assert;
PDF brought to you by
generated on December 3, 2014
Chapter 44: How to Implement a simple Registration Form | 171
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
use Acme\AccountBundle\Entity\User;
class Registration
{
/**
* @Assert\Type(type="Acme\AccountBundle\Entity\User")
* @Assert\Valid()
*/
protected $user;
/**
* @Assert\NotBlank()
* @Assert\True()
*/
protected $termsAccepted;
public function setUser(User $user)
{
$this->user = $user;
}
public function getUser()
{
return $this->user;
}
public function getTermsAccepted()
{
return $this->termsAccepted;
}
public function setTermsAccepted($termsAccepted)
{
$this->termsAccepted = (Boolean) $termsAccepted;
}
}
Next, create the form for this Registration model:
Listing 44-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/Acme/AccountBundle/Form/Type/RegistrationType.php
namespace Acme\AccountBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
class RegistrationType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('user', new UserType());
$builder->add(
'terms',
'checkbox',
array('property_path' => 'termsAccepted')
);
$builder->add('Register', 'submit');
}
PDF brought to you by
generated on December 3, 2014
Chapter 44: How to Implement a simple Registration Form | 172
19
20
21
22
23
24 }
public function getName()
{
return 'registration';
}
You don't need to use a special method for embedding the UserType form. A form is a field, too - so you
can add this like any other field, with the expectation that the Registration.user property will hold an
instance of the User class.
Handling the Form Submission
Next, you need a controller to handle the form. Start by creating a simple controller for displaying the
registration form:
Listing 44-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/AccountBundle/Controller/AccountController.php
namespace Acme\AccountBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
use Acme\AccountBundle\Form\Type\RegistrationType;
use Acme\AccountBundle\Form\Model\Registration;
class AccountController extends Controller
{
public function registerAction()
{
$registration = new Registration();
$form = $this->createForm(new RegistrationType(), $registration, array(
'action' => $this->generateUrl('account_create'),
));
return $this->render(
'AcmeAccountBundle:Account:register.html.twig',
array('form' => $form->createView())
);
}
}
And its template:
Listing 44-6
1 {# src/Acme/AccountBundle/Resources/views/Account/register.html.twig #}
2 {{ form(form) }}
Next, create the controller which handles the form submission. This performs the validation and saves
the data into the database:
Listing 44-7
1
2
3
4
5
6
use Symfony\Component\HttpFoundation\Request;
// ...
public function createAction(Request $request)
{
$em = $this->getDoctrine()->getManager();
PDF brought to you by
generated on December 3, 2014
Chapter 44: How to Implement a simple Registration Form | 173
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25 }
$form = $this->createForm(new RegistrationType(), new Registration());
$form->handleRequest($request);
if ($form->isValid()) {
$registration = $form->getData();
$em->persist($registration->getUser());
$em->flush();
return $this->redirect(...);
}
return $this->render(
'AcmeAccountBundle:Account:register.html.twig',
array('form' => $form->createView())
);
Add new Routes
Next, update your routes. If you're placing your routes inside your bundle (as shown here), don't forget
to make sure that the routing file is being imported.
Listing 44-8
1 # src/Acme/AccountBundle/Resources/config/routing.yml
2 account_register:
3
path:
/register
4
defaults: { _controller: AcmeAccountBundle:Account:register }
5
6 account_create:
7
path:
/register/create
8
defaults: { _controller: AcmeAccountBundle:Account:create }
Update your Database Schema
Of course, since you've added a User entity during this tutorial, make sure that your database schema has
been updated properly:
Listing 44-9
1 $ php app/console doctrine:schema:update --force
That's it! Your form now validates, and allows you to save the User object to the database. The extra
terms checkbox on the Registration model class is used during validation, but not actually used
afterwards when saving the User to the database.
PDF brought to you by
generated on December 3, 2014
Chapter 44: How to Implement a simple Registration Form | 174
Chapter 45
Console Commands
The Doctrine2 ORM integration offers several console commands under the doctrine namespace. To
view the command list you can use the list command:
Listing 45-1
1 $ php app/console list doctrine
A list of available commands will print out. You can find out more information about any of these
commands (or any Symfony command) by running the help command. For example, to get details about
the doctrine:database:create task, run:
Listing 45-2
1 $ php app/console help doctrine:database:create
Some notable or interesting tasks include:
• doctrine:ensure-production-settings - checks to see if the current environment is
configured efficiently for production. This should always be run in the prod environment:
Listing 45-3
1 $ php app/console doctrine:ensure-production-settings --env=prod
• doctrine:mapping:import - allows Doctrine to introspect an existing database and create
mapping information. For more information, see How to Generate Entities from an Existing
Database.
• doctrine:mapping:info - tells you all of the entities that Doctrine is aware of and whether or
not there are any basic errors with the mapping.
• doctrine:query:dql and doctrine:query:sql - allow you to execute DQL or SQL queries
directly from the command line.
PDF brought to you by
generated on December 3, 2014
Chapter 45: Console Commands | 175
Chapter 46
How to Send an Email
Sending emails is a classic task for any web application and one that has special complications and
potential pitfalls. Instead of recreating the wheel, one solution to send emails is to use the
SwiftmailerBundle, which leverages the power of the Swift Mailer1 library.
Don't forget to enable the bundle in your kernel before using it:
Listing 46-1
1 public function registerBundles()
2 {
3
$bundles = array(
4
// ...
5
6
new Symfony\Bundle\SwiftmailerBundle\SwiftmailerBundle(),
7
);
8
9
// ...
10 }
Configuration
Before using Swift Mailer, be sure to include its configuration. The only mandatory configuration
parameter is transport:
Listing 46-2
1 # app/config/config.yml
2 swiftmailer:
3
transport: smtp
4
encryption: ssl
5
auth_mode: login
6
host:
smtp.gmail.com
7
username:
your_username
8
password:
your_password
1. http://swiftmailer.org/
PDF brought to you by
generated on December 3, 2014
Chapter 46: How to Send an Email | 176
The majority of the Swift Mailer configuration deals with how the messages themselves should be
delivered.
The following configuration attributes are available:
•
•
•
•
•
•
•
•
transport (smtp, mail, sendmail, or gmail)
username
password
host
port
encryption (tls, or ssl)
auth_mode (plain, login, or cram-md5)
spool
• type (how to queue the messages, file or memory is supported, see How to Spool
Emails)
• path (where to store the messages)
• delivery_address (an email address where to send ALL emails)
• disable_delivery (set to true to disable delivery completely)
Sending Emails
The Swift Mailer library works by creating, configuring and then sending Swift_Message objects. The
"mailer" is responsible for the actual delivery of the message and is accessible via the mailer service.
Overall, sending an email is pretty straightforward:
Listing 46-3
1 public function indexAction($name)
2 {
3
$message = \Swift_Message::newInstance()
4
->setSubject('Hello Email')
5
->setFrom('[email protected]')
6
->setTo('[email protected]')
7
->setBody(
8
$this->renderView(
9
'HelloBundle:Hello:email.txt.twig',
10
array('name' => $name)
11
)
12
)
13
;
14
$this->get('mailer')->send($message);
15
16
return $this->render(...);
17 }
To keep things decoupled, the email body has been stored in a template and rendered with the
renderView() method.
The $message object supports many more options, such as including attachments, adding HTML
content, and much more. Fortunately, Swift Mailer covers the topic of Creating Messages2 in great detail
in its documentation.
2. http://swiftmailer.org/docs/messages.html
PDF brought to you by
generated on December 3, 2014
Chapter 46: How to Send an Email | 177
Several other cookbook articles are available related to sending emails in Symfony:
• How to Use Gmail to Send Emails
• How to Work with Emails during Development
• How to Spool Emails
PDF brought to you by
generated on December 3, 2014
Chapter 46: How to Send an Email | 178
Chapter 47
How to Use Gmail to Send Emails
During development, instead of using a regular SMTP server to send emails, you might find using Gmail
easier and more practical. The SwiftmailerBundle makes it really easy.
Instead of using your regular Gmail account, it's of course recommended that you create a special
account.
In the development configuration file, change the transport setting to gmail and set the username and
password to the Google credentials:
Listing 47-1
1 # app/config/config_dev.yml
2 swiftmailer:
3
transport: gmail
4
username: your_gmail_username
5
password: your_gmail_password
You're done!
If you are using the Symfony Standard Edition, configure the parameters in parameters.yml:
Listing 47-2
1 # app/config/parameters.yml
2 parameters:
3
# ...
4
mailer_transport: gmail
5
mailer_host:
~
6
mailer_user:
your_gmail_username
7
mailer_password: your_gmail_password
The gmail transport is simply a shortcut that uses the smtp transport and sets encryption,
auth_mode and host to work with Gmail.
PDF brought to you by
generated on December 3, 2014
Chapter 47: How to Use Gmail to Send Emails | 179
Chapter 48
How to Use the Cloud to Send Emails
Requirements for sending emails from a production system differ from your development setup as you
don't want to be limited in the number of emails, the sending rate or the sender address. Thus, using
Gmail or similar services is not an option. If setting up and maintaining your own reliable mail server
causes you a headache there's a simple solution: Leverage the cloud to send your emails.
This cookbook shows how easy it is to integrate Amazon's Simple Email Service (SES)1 into Symfony.
You can use the same technique for other mail services, as most of the time there is nothing more
to it than configuring an SMTP endpoint for Swift Mailer.
In the Symfony configuration, change the Swift Mailer settings transport, host, port and encryption
according to the information provided in the SES console2. Create your individual SMTP credentials in
the SES console and complete the configuration with the provided username and password:
Listing 48-1
1 # app/config/config.yml
2 swiftmailer:
3
transport: smtp
4
host:
email-smtp.us-east-1.amazonaws.com
5
port:
465 # different ports are available, see SES console
6
encryption: tls # TLS encryption is required
7
username:
AWS_ACCESS_KEY # to be created in the SES console
8
password:
AWS_SECRET_KEY # to be created in the SES console
The port and encryption keys are not present in the Symfony Standard Edition configuration by default,
but you can simply add them as needed.
And that's it, you're ready to start sending emails through the cloud!
1. http://aws.amazon.com/ses
2. https://console.aws.amazon.com/ses
PDF brought to you by
generated on December 3, 2014
Chapter 48: How to Use the Cloud to Send Emails | 180
If you are using the Symfony Standard Edition, configure the parameters in parameters.yml and
use them in your configuration files. This allows for different Swift Mailer configurations for each
installation of your application. For instance, use Gmail during development and the cloud in
production.
Listing 48-2
1 # app/config/parameters.yml
2 parameters:
3
# ...
4
mailer_transport: smtp
5
mailer_host:
email-smtp.us-east-1.amazonaws.com
6
mailer_port:
465 # different ports are available, see SES console
7
mailer_encryption: tls # TLS encryption is required
8
mailer_user:
AWS_ACCESS_KEY # to be created in the SES console
9
mailer_password:
AWS_SECRET_KEY # to be created in the SES console
If you intend to use Amazon SES, please note the following:
• You have to sign up to Amazon Web Services (AWS)3;
• Every sender address used in the From or Return-Path (bounce address) header needs
to be confirmed by the owner. You can also confirm an entire domain;
• Initially you are in a restricted sandbox mode. You need to request production access
before being allowed to send to arbitrary recipients;
• SES may be subject to a charge.
3. http://aws.amazon.com
PDF brought to you by
generated on December 3, 2014
Chapter 48: How to Use the Cloud to Send Emails | 181
Chapter 49
How to Work with Emails during Development
When developing an application which sends email, you will often not want to actually send the email
to the specified recipient during development. If you are using the SwiftmailerBundle with Symfony,
you can easily achieve this through configuration settings without having to make any changes to
your application's code at all. There are two main choices when it comes to handling email during
development: (a) disabling the sending of email altogether or (b) sending all email to a specific address.
Disabling Sending
You can disable sending email by setting the disable_delivery option to true. This is the default in the
test environment in the Standard distribution. If you do this in the test specific config then email will
not be sent when you run tests, but will continue to be sent in the prod and dev environments:
Listing 49-1
1 # app/config/config_test.yml
2 swiftmailer:
3
disable_delivery: true
If you'd also like to disable deliver in the dev environment, simply add this same configuration to the
config_dev.yml file.
Sending to a Specified Address
You can also choose to have all email sent to a specific address, instead of the address actually specified
when sending the message. This can be done via the delivery_address option:
Listing 49-2
1 # app/config/config_dev.yml
2 swiftmailer:
3
delivery_address: [email protected]
Now, suppose you're sending an email to [email protected]
Listing 49-3
PDF brought to you by
generated on December 3, 2014
Chapter 49: How to Work with Emails during Development | 182
1 public function indexAction($name)
2 {
3
$message = \Swift_Message::newInstance()
4
->setSubject('Hello Email')
5
->setFrom('[email protected]')
6
->setTo('[email protected]')
7
->setBody(
8
$this->renderView(
9
'HelloBundle:Hello:email.txt.twig',
10
array('name' => $name)
11
)
12
)
13
;
14
$this->get('mailer')->send($message);
15
16
return $this->render(...);
17 }
In the dev environment, the email will instead be sent to [email protected] Swift Mailer will add an extra
header to the email, X-Swift-To, containing the replaced address, so you can still see who it would have
been sent to.
In addition to the to addresses, this will also stop the email being sent to any CC and BCC addresses
set for it. Swift Mailer will add additional headers to the email with the overridden addresses in
them. These are X-Swift-Cc and X-Swift-Bcc for the CC and BCC addresses respectively.
Viewing from the Web Debug Toolbar
You can view any email sent during a single response when you are in the dev environment using the
Web Debug Toolbar. The email icon in the toolbar will show how many emails were sent. If you click it,
a report will open showing the details of the sent emails.
If you're sending an email and then immediately redirecting to another page, the web debug toolbar will
not display an email icon or a report on the next page.
Instead, you can set the intercept_redirects option to true in the config_dev.yml file, which will
cause the redirect to stop and allow you to open the report with details of the sent emails.
Listing 49-4
1 # app/config/config_dev.yml
2 web_profiler:
3
intercept_redirects: true
Alternatively, you can open the profiler after the redirect and search by the submit URL used on
the previous request (e.g. /contact/handle). The profiler's search feature allows you to load the
profiler information for any past requests.
PDF brought to you by
generated on December 3, 2014
Chapter 49: How to Work with Emails during Development | 183
Chapter 50
How to Spool Emails
When you are using the SwiftmailerBundle to send an email from a Symfony application, it will default
to sending the email immediately. You may, however, want to avoid the performance hit of the
communication between Swift Mailer and the email transport, which could cause the user to wait for
the next page to load while the email is sending. This can be avoided by choosing to "spool" the emails
instead of sending them directly. This means that Swift Mailer does not attempt to send the email but
instead saves the message to somewhere such as a file. Another process can then read from the spool and
take care of sending the emails in the spool. Currently only spooling to file or memory is supported by
Swift Mailer.
Spool Using Memory
When you use spooling to store the emails to memory, they will get sent right before the kernel
terminates. This means the email only gets sent if the whole request got executed without any unhandled
Exception or any errors. To configure swiftmailer with the memory option, use the following
configuration:
Listing 50-1
1 # app/config/config.yml
2 swiftmailer:
3
# ...
4
spool: { type: memory }
Spool Using a File
In order to use the spool with a file, use the following configuration:
Listing 50-2
1 # app/config/config.yml
2 swiftmailer:
3
# ...
4
spool:
PDF brought to you by
generated on December 3, 2014
Chapter 50: How to Spool Emails | 184
5
6
type: file
path: /path/to/spool
If you want to store the spool somewhere with your project directory, remember that you can use
the %kernel.root_dir% parameter to reference the project's root:
Listing 50-3
1 path: "%kernel.root_dir%/spool"
Now, when your app sends an email, it will not actually be sent but instead added to the spool. Sending
the messages from the spool is done separately. There is a console command to send the messages in the
spool:
Listing 50-4
1 $ php app/console swiftmailer:spool:send --env=prod
It has an option to limit the number of messages to be sent:
Listing 50-5
1 $ php app/console swiftmailer:spool:send --message-limit=10 --env=prod
You can also set the time limit in seconds:
Listing 50-6
1 $ php app/console swiftmailer:spool:send --time-limit=10 --env=prod
Of course you will not want to run this manually in reality. Instead, the console command should be
triggered by a cron job or scheduled task and run at a regular interval.
PDF brought to you by
generated on December 3, 2014
Chapter 50: How to Spool Emails | 185
Chapter 51
How to Test that an Email is Sent in a
functional Test
Sending e-mails with Symfony is pretty straightforward thanks to the SwiftmailerBundle, which leverages
the power of the Swift Mailer1 library.
To functionally test that an email was sent, and even assert the email subject, content or any other
headers, you can use the Symfony Profiler.
Start with an easy controller action that sends an e-mail:
Listing 51-1
1 public function sendEmailAction($name)
2 {
3
$message = \Swift_Message::newInstance()
4
->setSubject('Hello Email')
5
->setFrom('[email protected]')
6
->setTo('[email protected]')
7
->setBody('You should see me from the profiler!')
8
;
9
10
$this->get('mailer')->send($message);
11
12
return $this->render(...);
13 }
Don't forget to enable the profiler as explained in How to Use the Profiler in a Functional Test.
In your functional test, use the swiftmailer collector on the profiler to get information about the
messages send on the previous request:
Listing 51-2
1. http://swiftmailer.org/
PDF brought to you by
generated on December 3, 2014
Chapter 51: How to Test that an Email is Sent in a functional Test | 186
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
// src/Acme/DemoBundle/Tests/Controller/MailControllerTest.php
use Symfony\Bundle\FrameworkBundle\Test\WebTestCase;
class MailControllerTest extends WebTestCase
{
public function testMailIsSentAndContentIsOk()
{
$client = static::createClient();
// Enable the profiler for the next request (it does nothing if the profiler is
not available)
$client->enableProfiler();
$crawler = $client->request('POST', '/path/to/above/action');
$mailCollector = $client->getProfile()->getCollector('swiftmailer');
// Check that an e-mail was sent
$this->assertEquals(1, $mailCollector->getMessageCount());
$collectedMessages = $mailCollector->getMessages();
$message = $collectedMessages[0];
// Asserting e-mail data
$this->assertInstanceOf('Swift_Message', $message);
$this->assertEquals('Hello Email', $message->getSubject());
$this->assertEquals('[email protected]', key($message->getFrom()));
$this->assertEquals('[email protected]', key($message->getTo()));
$this->assertEquals(
'You should see me from the profiler!',
$message->getBody()
);
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 51: How to Test that an Email is Sent in a functional Test | 187
Chapter 52
How to Setup before and after Filters
It is quite common in web application development to need some logic to be executed just before or just
after your controller actions acting as filters or hooks.
In symfony1, this was achieved with the preExecute and postExecute methods. Most major frameworks
have similar methods but there is no such thing in Symfony2. The good news is that there is a much
better way to interfere with the Request -> Response process using the EventDispatcher component.
Token Validation Example
Imagine that you need to develop an API where some controllers are public but some others are restricted
to one or some clients. For these private features, you might provide a token to your clients to identify
themselves.
So, before executing your controller action, you need to check if the action is restricted or not. If it is
restricted, you need to validate the provided token.
Please note that for simplicity in this recipe, tokens will be defined in config and neither database
setup nor authentication via the Security component will be used.
Before Filters with the kernel.controller Event
First, store some basic token configuration using config.yml and the parameters key:
Listing 52-1
1 # app/config/config.yml
2 parameters:
3
tokens:
4
client1: pass1
5
client2: pass2
PDF brought to you by
generated on December 3, 2014
Chapter 52: How to Setup before and after Filters | 188
Tag Controllers to Be Checked
A kernel.controller listener gets notified on every request, right before the controller is executed. So,
first, you need some way to identify if the controller that matches the request needs token validation.
A clean and easy way is to create an empty interface and make the controllers implement it:
Listing 52-2
1
2
3
4
5
6
namespace Acme\DemoBundle\Controller;
interface TokenAuthenticatedController
{
// ...
}
A controller that implements this interface simply looks like this:
Listing 52-3
1
2
3
4
5
6
7
8
9
10
11
12
13
namespace Acme\DemoBundle\Controller;
use Acme\DemoBundle\Controller\TokenAuthenticatedController;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
class FooController extends Controller implements TokenAuthenticatedController
{
// An action that needs authentication
public function barAction()
{
// ...
}
}
Creating an Event Listener
Next, you'll need to create an event listener, which will hold the logic that you want executed before your
controllers. If you're not familiar with event listeners, you can learn more about them at How to Create
an Event Listener:
Listing 52-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/DemoBundle/EventListener/TokenListener.php
namespace Acme\DemoBundle\EventListener;
use Acme\DemoBundle\Controller\TokenAuthenticatedController;
use Symfony\Component\HttpKernel\Exception\AccessDeniedHttpException;
use Symfony\Component\HttpKernel\Event\FilterControllerEvent;
class TokenListener
{
private $tokens;
public function __construct($tokens)
{
$this->tokens = $tokens;
}
public function onKernelController(FilterControllerEvent $event)
{
$controller = $event->getController();
/*
PDF brought to you by
generated on December 3, 2014
Chapter 52: How to Setup before and after Filters | 189
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37 }
* $controller passed can be either a class or a Closure.
* This is not usual in Symfony but it may happen.
* If it is a class, it comes in array format
*/
if (!is_array($controller)) {
return;
}
if ($controller[0] instanceof TokenAuthenticatedController) {
$token = $event->getRequest()->query->get('token');
if (!in_array($token, $this->tokens)) {
throw new AccessDeniedHttpException('This action needs a valid token!');
}
}
}
Registering the Listener
Finally, register your listener as a service and tag it as an event listener. By listening on
kernel.controller, you're telling Symfony that you want your listener to be called just before any
controller is executed.
Listing 52-5
1 # app/config/config.yml (or inside your services.yml)
2 services:
3
demo.tokens.action_listener:
4
class: Acme\DemoBundle\EventListener\TokenListener
5
arguments: ["%tokens%"]
6
tags:
7
- { name: kernel.event_listener, event: kernel.controller, method:
onKernelController }
With this configuration, your TokenListener onKernelController method will be executed on each
request. If the controller that is about to be executed implements TokenAuthenticatedController,
token authentication is applied. This lets you have a "before" filter on any controller that you want.
After Filters with the kernel.response Event
In addition to having a "hook" that's executed before your controller, you can also add a hook that's
executed after your controller. For this example, imagine that you want to add a sha1 hash (with a salt
using that token) to all responses that have passed this token authentication.
Another core Symfony event - called kernel.response - is notified on every request, but after the
controller returns a Response object. Creating an "after" listener is as easy as creating a listener class and
registering it as a service on this event.
For example, take the TokenListener from the previous example and first record the authentication
token inside the request attributes. This will serve as a basic flag that this request underwent token
authentication:
Listing 52-6
1 public function onKernelController(FilterControllerEvent $event)
2 {
3
// ...
4
PDF brought to you by
generated on December 3, 2014
Chapter 52: How to Setup before and after Filters | 190
5
6
7
8
9
10
11
12
13
14 }
if ($controller[0] instanceof TokenAuthenticatedController) {
$token = $event->getRequest()->query->get('token');
if (!in_array($token, $this->tokens)) {
throw new AccessDeniedHttpException('This action needs a valid token!');
}
// mark the request as having passed token authentication
$event->getRequest()->attributes->set('auth_token', $token);
}
Now, add another method to this class - onKernelResponse - that looks for this flag on the request object
and sets a custom header on the response if it's found:
Listing 52-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// add the new use statement at the top of your file
use Symfony\Component\HttpKernel\Event\FilterResponseEvent;
public function onKernelResponse(FilterResponseEvent $event)
{
// check to see if onKernelController marked this as a token "auth'ed" request
if (!$token = $event->getRequest()->attributes->get('auth_token')) {
return;
}
$response = $event->getResponse();
// create a hash and set it as a response header
$hash = sha1($response->getContent().$token);
$response->headers->set('X-CONTENT-HASH', $hash);
}
Finally, a second "tag" is needed on the service definition to notify Symfony that the onKernelResponse
event should be notified for the kernel.response event:
Listing 52-8
1 # app/config/config.yml (or inside your services.yml)
2 services:
3
demo.tokens.action_listener:
4
class: Acme\DemoBundle\EventListener\TokenListener
5
arguments: ["%tokens%"]
6
tags:
7
- { name: kernel.event_listener, event: kernel.controller, method:
8 onKernelController }
- { name: kernel.event_listener, event: kernel.response, method:
onKernelResponse }
That's it! The TokenListener is now notified before every controller is executed (onKernelController)
and after every controller returns a response (onKernelResponse). By making specific controllers
implement the TokenAuthenticatedController interface, your listener knows which controllers it
should take action on. And by storing a value in the request's "attributes" bag, the onKernelResponse
method knows to add the extra header. Have fun!
PDF brought to you by
generated on December 3, 2014
Chapter 52: How to Setup before and after Filters | 191
Chapter 53
How to Extend a Class without Using
Inheritance
To allow multiple classes to add methods to another one, you can define the magic __call() method in
the class you want to be extended like this:
Listing 53-1
1 class Foo
2 {
3
// ...
4
5
public function __call($method, $arguments)
6
{
7
// create an event named 'foo.method_is_not_found'
8
$event = new HandleUndefinedMethodEvent($this, $method, $arguments);
9
$this->dispatcher->dispatch('foo.method_is_not_found', $event);
10
11
// no listener was able to process the event? The method does not exist
12
if (!$event->isProcessed()) {
13
throw new \Exception(sprintf('Call to undefined method %s::%s.',
14 get_class($this), $method));
15
}
16
17
// return the listener returned value
18
return $event->getReturnValue();
19
}
}
This uses a special HandleUndefinedMethodEvent that should also be created. This is a generic class that
could be reused each time you need to use this pattern of class extension:
Listing 53-2
1 use Symfony\Component\EventDispatcher\Event;
2
3 class HandleUndefinedMethodEvent extends Event
4 {
5
protected $subject;
PDF brought to you by
generated on December 3, 2014
Chapter 53: How to Extend a Class without Using Inheritance | 192
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52 }
protected
protected
protected
protected
$method;
$arguments;
$returnValue;
$isProcessed = false;
public function __construct($subject, $method, $arguments)
{
$this->subject = $subject;
$this->method = $method;
$this->arguments = $arguments;
}
public function getSubject()
{
return $this->subject;
}
public function getMethod()
{
return $this->method;
}
public function getArguments()
{
return $this->arguments;
}
/**
* Sets the value to return and stops other listeners from being notified
*/
public function setReturnValue($val)
{
$this->returnValue = $val;
$this->isProcessed = true;
$this->stopPropagation();
}
public function getReturnValue()
{
return $this->returnValue;
}
public function isProcessed()
{
return $this->isProcessed;
}
Next, create a class that will listen to the foo.method_is_not_found event and add the method bar():
Listing 53-3
1 class Bar
2 {
3
public function onFooMethodIsNotFound(HandleUndefinedMethodEvent $event)
4
{
5
// only respond to the calls to the 'bar' method
6
if ('bar' != $event->getMethod()) {
7
// allow another listener to take care of this unknown method
8
return;
PDF brought to you by
generated on December 3, 2014
Chapter 53: How to Extend a Class without Using Inheritance | 193
9
10
11
12
13
14
15
16
17
18
19
20
21
22 }
}
// the subject object (the foo instance)
$foo = $event->getSubject();
// the bar method arguments
$arguments = $event->getArguments();
// ... do something
// set the return value
$event->setReturnValue($someValue);
}
Finally, add the new bar method to the Foo class by registering an instance of Bar with the
foo.method_is_not_found event:
Listing 53-4
1 $bar = new Bar();
2 $dispatcher->addListener('foo.method_is_not_found', array($bar, 'onFooMethodIsNotFound'));
PDF brought to you by
generated on December 3, 2014
Chapter 53: How to Extend a Class without Using Inheritance | 194
Chapter 54
How to Customize a Method Behavior without
Using Inheritance
Doing something before or after a Method Call
If you want to do something just before, or just after a method is called, you can dispatch an event
respectively at the beginning or at the end of the method:
Listing 54-1
1 class Foo
2 {
3
// ...
4
5
public function send($foo, $bar)
6
{
7
// do something before the method
8
$event = new FilterBeforeSendEvent($foo, $bar);
9
$this->dispatcher->dispatch('foo.pre_send', $event);
10
11
// get $foo and $bar from the event, they may have been modified
12
$foo = $event->getFoo();
13
$bar = $event->getBar();
14
15
// the real method implementation is here
16
$ret = ...;
17
18
// do something after the method
19
$event = new FilterSendReturnValue($ret);
20
$this->dispatcher->dispatch('foo.post_send', $event);
21
22
return $event->getReturnValue();
23
}
24 }
PDF brought to you by
generated on December 3, 2014
Chapter 54: How to Customize a Method Behavior without Using Inheritance | 195
In this example, two events are thrown: foo.pre_send, before the method is executed, and
foo.post_send after the method is executed. Each uses a custom Event class to communicate
information to the listeners of the two events. These event classes would need to be created by you and
should allow, in this example, the variables $foo, $bar and $ret to be retrieved and set by the listeners.
For example, assuming the FilterSendReturnValue has a setReturnValue method, one listener might
look like this:
Listing 54-2
1 public function onFooPostSend(FilterSendReturnValue $event)
2 {
3
$ret = $event->getReturnValue();
4
// modify the original ``$ret`` value
5
6
$event->setReturnValue($ret);
7 }
PDF brought to you by
generated on December 3, 2014
Chapter 54: How to Customize a Method Behavior without Using Inheritance | 196
Chapter 55
How to use Expressions in Security, Routing,
Services, and Validation
New in version 2.4: The expression functionality was introduced in Symfony 2.4.
In Symfony 2.4, a powerful ExpressionLanguage component was added to Symfony. This allows us to
add highly customized logic inside configuration.
The Symfony Framework leverages expressions out of the box in the following ways:
•
•
•
•
Configuring services;
Route matching conditions;
Checking security and access controls with allow_if;
Validation.
For more information about how to create and work with expressions, see The Expression Syntax.
PDF brought to you by
generated on December 3, 2014
Chapter 55: How to use Expressions in Security, Routing, Services, and Validation | 197
Chapter 56
How to Customize Form Rendering
Symfony gives you a wide variety of ways to customize how a form is rendered. In this guide, you'll learn
how to customize every possible part of your form with as little effort as possible whether you use Twig
or PHP as your templating engine.
Form Rendering Basics
Recall that the label, error and HTML widget of a form field can easily be rendered by using the form_row
Twig function or the row PHP helper method:
Listing 56-1
1 {{ form_row(form.age) }}
You can also render each of the three parts of the field individually:
Listing 56-2
1 <div>
2
{{ form_label(form.age) }}
3
{{ form_errors(form.age) }}
4
{{ form_widget(form.age) }}
5 </div>
In both cases, the form label, errors and HTML widget are rendered by using a set of markup that ships
standard with Symfony. For example, both of the above templates would render:
Listing 56-3
1 <div>
2
<label for="form_age">Age</label>
3
<ul>
4
<li>This field is required</li>
5
</ul>
6
<input type="number" id="form_age" name="form[age]" />
7 </div>
To quickly prototype and test a form, you can render the entire form with just one line:
Listing 56-4
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 198
1
2
3
4
5
{# renders all fields #}
{{ form_widget(form) }}
{# renders all fields *and* the form start and end tags #}
{{ form(form) }}
The remainder of this recipe will explain how every part of the form's markup can be modified at
several different levels. For more information about form rendering in general, see Rendering a Form in a
Template.
What are Form Themes?
Symfony uses form fragments - a small piece of a template that renders just one part of a form - to render
each part of a form - field labels, errors, input text fields, select tags, etc.
The fragments are defined as blocks in Twig and as template files in PHP.
A theme is nothing more than a set of fragments that you want to use when rendering a form. In other
words, if you want to customize one portion of how a form is rendered, you'll import a theme which
contains a customization of the appropriate form fragments.
Symfony comes with four built-in form themes that define each and every fragment needed to render
every part of a form:
• form_div_layout.html.twig1, wraps each form field inside a <div> element.
• form_table_layout.html.twig2, wraps the entire form inside a <table> element and each form
field inside a <tr> element.
• bootstrap_3_layout.html.twig3, wraps each form field inside a <div> element with the
appropriate CSS classes to apply the default Bootstrap 3 CSS framework4 styles.
• bootstrap_3_horizontal_layout.html.twig5, it's similar to the previous theme, but the CSS
classes applied are the ones used to display the forms horizontally (i.e. the label and the widget
in the same row).
In the next section you will learn how to customize a theme by overriding some or all of its fragments.
For example, when the widget of an integer type field is rendered, an input number field is generated
Listing 56-5
1 {{ form_widget(form.age) }}
renders:
Listing 56-6
1 <input type="number" id="form_age" name="form[age]" required="required" value="33" />
Internally, Symfony uses the integer_widget fragment to render the field. This is because the field type
is integer and you're rendering its widget (as opposed to its label or errors).
In Twig that would default to the block integer_widget from the form_div_layout.html.twig6 template.
In PHP it would rather be the integer_widget.html.php file located in the FrameworkBundle/
Resources/views/Form folder.
1. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_div_layout.html.twig
2. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_table_layout.html.twig
3. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/bootstrap_3_layout.html.twig
4. http://getbootstrap.com/
5. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/bootstrap_3_horizontal_layout.html.twig
6. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_div_layout.html.twig
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 199
The default implementation of the integer_widget fragment looks like this:
Listing 56-7
1 {# form_div_layout.html.twig #}
2 {% block integer_widget %}
3
{% set type = type|default('number') %}
4
{{ block('form_widget_simple') }}
5 {% endblock integer_widget %}
As you can see, this fragment itself renders another fragment - form_widget_simple:
Listing 56-8
1 {# form_div_layout.html.twig #}
2 {% block form_widget_simple %}
3
{% set type = type|default('text') %}
4
<input type="{{ type }}" {{ block('widget_attributes') }} {% if value is not empty
5 %}value="{{ value }}" {% endif %}/>
{% endblock form_widget_simple %}
The point is, the fragments dictate the HTML output of each part of a form. To customize the form
output, you just need to identify and override the correct fragment. A set of these form fragment
customizations is known as a form "theme". When rendering a form, you can choose which form
theme(s) you want to apply.
In Twig a theme is a single template file and the fragments are the blocks defined in this file.
In PHP a theme is a folder and the fragments are individual template files in this folder.
Knowing which Block to Customize
In this example, the customized fragment name is integer_widget because you want to override
the HTML widget for all integer field types. If you need to customize textarea fields, you would
customize textarea_widget.
As you can see, the fragment name is a combination of the field type and which part of the field is
being rendered (e.g. widget, label, errors, row). As such, to customize how errors are rendered
for just input text fields, you should customize the text_errors fragment.
More commonly, however, you'll want to customize how errors are displayed across all fields.
You can do this by customizing the form_errors fragment. This takes advantage of field type
inheritance. Specifically, since the text type extends from the form type, the Form component
will first look for the type-specific fragment (e.g. text_errors) before falling back to its parent
fragment name if it doesn't exist (e.g. form_errors).
For more information on this topic, see Form Fragment Naming.
Form Theming
To see the power of form theming, suppose you want to wrap every input number field with a div tag.
The key to doing this is to customize the integer_widget fragment.
Form Theming in Twig
When customizing the form field block in Twig, you have two options on where the customized form
block can live:
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 200
Method
Pros
Cons
Inside the same template as the
form
Quick and easy
Can't be reused in other templates
Inside a separate template
Can be reused by many
templates
Requires an extra template to be
created
Both methods have the same effect but are better in different situations.
Method 1: Inside the same Template as the Form
The easiest way to customize the integer_widget block is to customize it directly in the template that's
actually rendering the form.
Listing 56-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
{% extends '::base.html.twig' %}
{% form_theme form _self %}
{% block integer_widget %}
<div class="integer_widget">
{% set type = type|default('number') %}
{{ block('form_widget_simple') }}
</div>
{% endblock %}
{% block content %}
{# ... render the form #}
{{ form_row(form.age) }}
{% endblock %}
By using the special {% form_theme form _self %} tag, Twig looks inside the same template for
any overridden form blocks. Assuming the form.age field is an integer type field, when its widget is
rendered, the customized integer_widget block will be used.
The disadvantage of this method is that the customized form block can't be reused when rendering other
forms in other templates. In other words, this method is most useful when making form customizations
that are specific to a single form in your application. If you want to reuse a form customization across
several (or all) forms in your application, read on to the next section.
Method 2: Inside a separate Template
You can also choose to put the customized integer_widget form block in a separate template entirely.
The code and end-result are the same, but you can now re-use the form customization across many
templates:
Listing 56-10
1 {# src/Acme/DemoBundle/Resources/views/Form/fields.html.twig #}
2 {% block integer_widget %}
3
<div class="integer_widget">
4
{% set type = type|default('number') %}
5
{{ block('form_widget_simple') }}
6
</div>
7 {% endblock %}
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 201
Now that you've created the customized form block, you need to tell Symfony to use it. Inside the
template where you're actually rendering your form, tell Symfony to use the template via the form_theme
tag:
Listing 56-11
1 {% form_theme form 'AcmeDemoBundle:Form:fields.html.twig' %}
2
3 {{ form_widget(form.age) }}
When the form.age widget is rendered, Symfony will use the integer_widget block from the new
template and the input tag will be wrapped in the div element specified in the customized block.
Multiple Templates
A form can also be customized by applying several templates. To do this, pass the name of all the
templates as an array using the with keyword:
Listing 56-12
1 {% form_theme form with ['::common.html.twig', ':Form:fields.html.twig',
2
'AcmeDemoBundle:Form:fields.html.twig'] %}
3
4 {# ... #}
The templates can be located at different bundles and they can even be stored at the global app/
Resources/views/ directory.
Child Forms
You can also apply a form theme to a specific child of your form:
Listing 56-13
1 {% form_theme form.child 'AcmeDemoBundle:Form:fields.html.twig' %}
This is useful when you want to have a custom theme for a nested form that's different than the one of
your main form. Just specify both your themes:
Listing 56-14
1 {% form_theme form 'AcmeDemoBundle:Form:fields.html.twig' %}
2
3 {% form_theme form.child 'AcmeDemoBundle:Form:fields_child.html.twig' %}
Form Theming in PHP
When using PHP as a templating engine, the only method to customize a fragment is to create a new
template file - this is similar to the second method used by Twig.
The template file must be named after the fragment. You must create a integer_widget.html.php file
in order to customize the integer_widget fragment.
Listing 56-15
1 <!-- src/Acme/DemoBundle/Resources/views/Form/integer_widget.html.php -->
2 <div class="integer_widget">
3
<?php echo $view['form']->block($form, 'form_widget_simple', array('type' =>
4 isset($type) ? $type : "number")) ?>
</div>
Now that you've created the customized form template, you need to tell Symfony to use it. Inside the
template where you're actually rendering your form, tell Symfony to use the theme via the setTheme
helper method:
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 202
Listing 56-16
1 <?php $view['form']->setTheme($form, array('AcmeDemoBundle:Form')); ?>
2
3 <?php $view['form']->widget($form['age']) ?>
When the form.age widget is rendered, Symfony will use the customized integer_widget.html.php
template and the input tag will be wrapped in the div element.
If you want to apply a theme to a specific child form, pass it to the setTheme method:
Listing 56-17
1 <?php $view['form']->setTheme($form['child'], 'AcmeDemoBundle:Form/Child'); ?>
Referencing base Form Blocks (Twig specific)
So far, to override a particular form block, the best method is to copy the default block from
form_div_layout.html.twig7, paste it into a different template, and then customize it. In many cases, you
can avoid doing this by referencing the base block when customizing it.
This is easy to do, but varies slightly depending on if your form block customizations are in the same
template as the form or a separate template.
Referencing Blocks from inside the same Template as the Form
Import the blocks by adding a use tag in the template where you're rendering the form:
Listing 56-18
1 {% use 'form_div_layout.html.twig' with integer_widget as base_integer_widget %}
Now, when the blocks from form_div_layout.html.twig8 are imported, the integer_widget block is
called base_integer_widget. This means that when you redefine the integer_widget block, you can
reference the default markup via base_integer_widget:
Listing 56-19
1 {% block integer_widget %}
2
<div class="integer_widget">
3
{{ block('base_integer_widget') }}
4
</div>
5 {% endblock %}
Referencing base Blocks from an external Template
If your form customizations live inside an external template, you can reference the base block by using
the parent() Twig function:
Listing 56-20
1
2
3
4
5
6
7
8
{# src/Acme/DemoBundle/Resources/views/Form/fields.html.twig #}
{% extends 'form_div_layout.html.twig' %}
{% block integer_widget %}
<div class="integer_widget">
{{ parent() }}
</div>
{% endblock %}
7. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_div_layout.html.twig
8. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_div_layout.html.twig
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 203
It is not possible to reference the base block when using PHP as the templating engine. You have
to manually copy the content from the base block to your new template file.
Making Application-wide Customizations
If you'd like a certain form customization to be global to your application, you can accomplish this by
making the form customizations in an external template and then importing it inside your application
configuration.
Twig
By using the following configuration, any customized form blocks inside the
AcmeDemoBundle:Form:fields.html.twig template will be used globally when a form is rendered.
Listing 56-21
1 # app/config/config.yml
2 twig:
3
form_themes:
4
- 'AcmeDemoBundle:Form:fields.html.twig'
5
# ...
By default, Twig uses a div layout when rendering forms. Some people, however, may prefer to render
forms in a table layout. Use the form_table_layout.html.twig resource to use such a layout:
Listing 56-22
1 # app/config/config.yml
2 twig:
3
form_themes: ['form_table_layout.html.twig']
4
# ...
If you only want to make the change in one template, add the following line to your template file rather
than adding the template as a resource:
Listing 56-23
1 {% form_theme form 'form_table_layout.html.twig' %}
Note that the form variable in the above code is the form view variable that you passed to your template.
PHP
By using the following configuration, any customized form fragments inside the src/Acme/DemoBundle/
Resources/views/Form folder will be used globally when a form is rendered.
Listing 56-24
1 # app/config/config.yml
2 framework:
3
templating:
4
form:
5
resources:
6
- 'AcmeDemoBundle:Form'
7
# ...
By default, the PHP engine uses a div layout when rendering forms. Some people, however, may prefer to
render forms in a table layout. Use the FrameworkBundle:FormTable resource to use such a layout:
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 204
Listing 56-25
1 # app/config/config.yml
2 framework:
3
templating:
4
form:
5
resources:
6
- 'FrameworkBundle:FormTable'
If you only want to make the change in one template, add the following line to your template file rather
than adding the template as a resource:
Listing 56-26
1 <?php $view['form']->setTheme($form, array('FrameworkBundle:FormTable')); ?>
Note that the $form variable in the above code is the form view variable that you passed to your template.
How to Customize an individual Field
So far, you've seen the different ways you can customize the widget output of all text field types. You
can also customize individual fields. For example, suppose you have two text fields in a product form
- name and description - but you only want to customize one of the fields. This can be accomplished
by customizing a fragment whose name is a combination of the field's id attribute and which part of the
field is being customized. For example, to customize the name field only:
Listing 56-27
1
2
3
4
5
6
7
8
9
{% form_theme form _self %}
{% block _product_name_widget %}
<div class="text_widget">
{{ block('form_widget_simple') }}
</div>
{% endblock %}
{{ form_widget(form.name) }}
Here, the _product_name_widget fragment defines the template to use for the field whose id is
product_name (and name is product[name]).
The product portion of the field is the form name, which may be set manually or generated
automatically based on your form type name (e.g. ProductType equates to product). If you're not
sure what your form name is, just view the source of your generated form.
If you want to change the product or name portion of the block name _product_name_widget you
can set the block_name option in your form type:
Listing 56-28
1
2
3
4
5
6
7
8
9
10
use Symfony\Component\Form\FormBuilderInterface;
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
$builder->add('name', 'text', array(
'block_name' => 'custom_name',
));
}
Then the block name will be _product_custom_name_widget.
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 205
You can also override the markup for an entire field row using the same method:
Listing 56-29
1
2
3
4
5
6
7
8
9
10
11
{% form_theme form _self %}
{% block _product_name_row %}
<div class="name_row">
{{ form_label(form) }}
{{ form_errors(form) }}
{{ form_widget(form) }}
</div>
{% endblock %}
{{ form_row(form.name) }}
Other common Customizations
So far, this recipe has shown you several different ways to customize a single piece of how a form is
rendered. The key is to customize a specific fragment that corresponds to the portion of the form you
want to control (see naming form blocks).
In the next sections, you'll see how you can make several common form customizations. To apply these
customizations, use one of the methods described in the Form Theming section.
Customizing Error Output
The Form component only handles how the validation errors are rendered, and not the actual
validation error messages. The error messages themselves are determined by the validation
constraints you apply to your objects. For more information, see the chapter on validation.
There are many different ways to customize how errors are rendered when a form is submitted with
errors. The error messages for a field are rendered when you use the form_errors helper:
Listing 56-30
1 {{ form_errors(form.age) }}
By default, the errors are rendered inside an unordered list:
Listing 56-31
1 <ul>
2
<li>This field is required</li>
3 </ul>
To override how errors are rendered for all fields, simply copy, paste and customize the form_errors
fragment.
Listing 56-32
1 {# form_errors.html.twig #}
2 {% block form_errors %}
3
{% spaceless %}
4
{% if errors|length > 0 %}
5
<ul>
6
{% for error in errors %}
7
<li>{{ error.message }}</li>
8
{% endfor %}
9
</ul>
10
{% endif %}
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 206
11
{% endspaceless %}
12 {% endblock form_errors %}
See Form Theming for how to apply this customization.
You can also customize the error output for just one specific field type. To customize only the markup
used for these errors, follow the same directions as above but put the contents in a relative _errors block
(or file in case of PHP templates). For example: text_errors (or text_errors.html.php).
See Form Fragment Naming to find out which specific block or file you have to customize.
Certain errors that are more global to your form (i.e. not specific to just one field) are rendered separately,
usually at the top of your form:
Listing 56-33
1 {{ form_errors(form) }}
To customize only the markup used for these errors, follow the same directions as above, but now check
if the compound variable is set to true. If it is true, it means that what's being currently rendered is a
collection of fields (e.g. a whole form), and not just an individual field.
Listing 56-34
1 {# form_errors.html.twig #}
2 {% block form_errors %}
3
{% spaceless %}
4
{% if errors|length > 0 %}
5
{% if compound %}
6
<ul>
7
{% for error in errors %}
8
<li>{{ error.message }}</li>
9
{% endfor %}
10
</ul>
11
{% else %}
12
{# ... display the errors for a single field #}
13
{% endif %}
14
{% endif %}
15
{% endspaceless %}
16 {% endblock form_errors %}
Customizing the "Form Row"
When you can manage it, the easiest way to render a form field is via the form_row function, which
renders the label, errors and HTML widget of a field. To customize the markup used for rendering all
form field rows, override the form_row fragment. For example, suppose you want to add a class to the
div element around each row:
Listing 56-35
1 {# form_row.html.twig #}
2 {% block form_row %}
3
<div class="form_row">
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 207
4
{{ form_label(form) }}
5
{{ form_errors(form) }}
6
{{ form_widget(form) }}
7
</div>
8 {% endblock form_row %}
See Form Theming for how to apply this customization.
Adding a "Required" Asterisk to Field Labels
If you want to denote all of your required fields with a required asterisk (*), you can do this by
customizing the form_label fragment.
In Twig, if you're making the form customization inside the same template as your form, modify the use
tag and add the following:
Listing 56-36
1 {% use 'form_div_layout.html.twig' with form_label as base_form_label %}
2
3 {% block form_label %}
4
{{ block('base_form_label') }}
5
6
{% if required %}
7
<span class="required" title="This field is required">*</span>
8
{% endif %}
9 {% endblock %}
In Twig, if you're making the form customization inside a separate template, use the following:
Listing 56-37
1 {% extends 'form_div_layout.html.twig' %}
2
3 {% block form_label %}
4
{{ parent() }}
5
6
{% if required %}
7
<span class="required" title="This field is required">*</span>
8
{% endif %}
9 {% endblock %}
When using PHP as a templating engine you have to copy the content from the original template:
Listing 56-38
1
2
3
4
5
6
7
8
9
10
11
12
<!-- form_label.html.php -->
<!-- original content -->
<?php if ($required) { $label_attr['class'] = trim((isset($label_attr['class']) ?
$label_attr['class'] : '').' required'); } ?>
<?php if (!$compound) { $label_attr['for'] = $id; } ?>
<?php if (!$label) { $label = $view['form']->humanize($name); } ?>
<label <?php foreach ($label_attr as $k => $v) { printf('%s="%s" ', $view->escape($k),
$view->escape($v)); } ?>><?php echo $view->escape($view['translator']->trans($label,
array(), $translation_domain)) ?></label>
<!-- customization -->
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 208
<?php if ($required) : ?>
<span class="required" title="This field is required">*</span>
<?php endif ?>
See Form Theming for how to apply this customization.
Using CSS only
By default, label tags of required fields are rendered with a required CSS class. Thus, you can
also add an asterisk using CSS only:
Listing 56-39
1 label.required:before {
2
content: "* ";
3 }
Adding "help" Messages
You can also customize your form widgets to have an optional "help" message.
In Twig, if you're making the form customization inside the same template as your form, modify the use
tag and add the following:
Listing 56-40
1 {% use 'form_div_layout.html.twig' with form_widget_simple as base_form_widget_simple %}
2
3 {% block form_widget_simple %}
4
{{ block('base_form_widget_simple') }}
5
6
{% if help is defined %}
7
<span class="help">{{ help }}</span>
8
{% endif %}
9 {% endblock %}
In Twig, if you're making the form customization inside a separate template, use the following:
Listing 56-41
1 {% extends 'form_div_layout.html.twig' %}
2
3 {% block form_widget_simple %}
4
{{ parent() }}
5
6
{% if help is defined %}
7
<span class="help">{{ help }}</span>
8
{% endif %}
9 {% endblock %}
When using PHP as a templating engine you have to copy the content from the original template:
Listing 56-42
1 <!-- form_widget_simple.html.php -->
2
3 <!-- Original content -->
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 209
4
5
6
7
8
9
10
11
12
13
<input
type="<?php echo isset($type) ? $view->escape($type) : 'text' ?>"
<?php if (!empty($value)): ?>value="<?php echo $view->escape($value) ?>"<?php endif ?>
<?php echo $view['form']->block($form, 'widget_attributes') ?>
/>
<!-- Customization -->
<?php if (isset($help)) : ?>
<span class="help"><?php echo $view->escape($help) ?></span>
<?php endif ?>
To render a help message below a field, pass in a help variable:
Listing 56-43
1 {{ form_widget(form.title, {'help': 'foobar'}) }}
See Form Theming for how to apply this customization.
Using Form Variables
Most of the functions available for rendering different parts of a form (e.g. the form widget, form label,
form errors, etc.) also allow you to make certain customizations directly. Look at the following example:
Listing 56-44
1 {# render a widget, but add a "foo" class to it #}
2 {{ form_widget(form.name, { 'attr': {'class': 'foo'} }) }}
The array passed as the second argument contains form "variables". For more details about this concept
in Twig, see More about Form Variables.
PDF brought to you by
generated on December 3, 2014
Chapter 56: How to Customize Form Rendering | 210
Chapter 57
How to Use Data Transformers
You'll often find the need to transform the data the user entered in a form into something else for use
in your program. You could easily do this manually in your controller, but what if you want to use this
specific form in different places?
Say you have a one-to-one relation of Task to Issue, e.g. a Task optionally has an issue linked to it. Adding
a listbox with all possible issues can eventually lead to a really long listbox in which it is impossible to find
something. You might want to add a textbox instead, where the user can simply enter the issue number.
You could try to do this in your controller, but it's not the best solution. It would be better if this issue
were automatically converted to an Issue object. This is where Data Transformers come into play.
Creating the Transformer
First, create an IssueToNumberTransformer class - this class will be responsible for converting to and
from the issue number and the Issue object:
Listing 57-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/Acme/TaskBundle/Form/DataTransformer/IssueToNumberTransformer.php
namespace Acme\TaskBundle\Form\DataTransformer;
use
use
use
use
Symfony\Component\Form\DataTransformerInterface;
Symfony\Component\Form\Exception\TransformationFailedException;
Doctrine\Common\Persistence\ObjectManager;
Acme\TaskBundle\Entity\Issue;
class IssueToNumberTransformer implements DataTransformerInterface
{
/**
* @var ObjectManager
*/
private $om;
/**
* @param ObjectManager $om
*/
public function __construct(ObjectManager $om)
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 211
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68 }
{
$this->om = $om;
}
/**
* Transforms an object (issue) to a string (number).
*
* @param Issue|null $issue
* @return string
*/
public function transform($issue)
{
if (null === $issue) {
return "";
}
return $issue->getNumber();
}
/**
* Transforms a string (number) to an object (issue).
*
* @param string $number
*
* @return Issue|null
*
* @throws TransformationFailedException if object (issue) is not found.
*/
public function reverseTransform($number)
{
if (!$number) {
return null;
}
$issue = $this->om
->getRepository('AcmeTaskBundle:Issue')
->findOneBy(array('number' => $number))
;
if (null === $issue) {
throw new TransformationFailedException(sprintf(
'An issue with number "%s" does not exist!',
$number
));
}
return $issue;
}
If you want a new issue to be created when an unknown number is entered, you can instantiate it
rather than throwing the TransformationFailedException.
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 212
When null is passed to the transform() method, your transformer should return an equivalent
value of the type it is transforming to (e.g. an empty string, 0 for integers or 0.0 for floats).
Using the Transformer
Now that you have the transformer built, you just need to add it to your issue field in some form.
You can also use transformers without creating a new custom form type by calling addModelTransformer
(or addViewTransformer - see Model and View Transformers) on any field builder:
Listing 57-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
use Symfony\Component\Form\FormBuilderInterface;
use Acme\TaskBundle\Form\DataTransformer\IssueToNumberTransformer;
class TaskType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
// this assumes that the entity manager was passed in as an option
$entityManager = $options['em'];
$transformer = new IssueToNumberTransformer($entityManager);
// add a normal text field, but add your transformer to it
$builder->add(
$builder->create('issue', 'text')
->addModelTransformer($transformer)
);
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver
->setDefaults(array(
'data_class' => 'Acme\TaskBundle\Entity\Task',
))
->setRequired(array(
'em',
))
->setAllowedTypes(array(
'em' => 'Doctrine\Common\Persistence\ObjectManager',
));
// ...
}
// ...
}
This example requires that you pass in the entity manager as an option when creating your form. Later,
you'll learn how you could create a custom issue field type to avoid needing to do this in your controller:
Listing 57-3
1 $taskForm = $this->createForm(new TaskType(), $task, array(
2
'em' => $this->getDoctrine()->getManager(),
3 ));
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 213
Cool, you're done! Your user will be able to enter an issue number into the text field and it will
be transformed back into an Issue object. This means that, after a successful submission, the Form
framework will pass a real Issue object to Task::setIssue() instead of the issue number.
If the issue isn't found, a form error will be created for that field and its error message can be controlled
with the invalid_message field option.
Notice that adding a transformer requires using a slightly more complicated syntax when adding
the field. The following is wrong, as the transformer would be applied to the entire form, instead
of just this field:
Listing 57-4
1 // THIS IS WRONG - TRANSFORMER WILL BE APPLIED TO THE ENTIRE FORM
2 // see above example for correct code
3 $builder->add('issue', 'text')
4
->addModelTransformer($transformer);
Model and View Transformers
In the above example, the transformer was used as a "model" transformer. In fact, there are two different
types of transformers and three different types of underlying data.
In any form, the three different types of data are:
1. Model data - This is the data in the format used in your application (e.g. an Issue object). If
you call Form::getData or Form::setData, you're dealing with the "model" data.
2. Norm Data - This is a normalized version of your data, and is commonly the same as your
"model" data (though not in our example). It's not commonly used directly.
3. View Data - This is the format that's used to fill in the form fields themselves. It's also the
format in which the user will submit the data. When you call Form::submit($data), the $data
is in the "view" data format.
The two different types of transformers help convert to and from each of these types of data:
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 214
Model transformers:
• transform: "model data" => "norm data"
• reverseTransform: "norm data" => "model data"
View transformers:
• transform: "norm data" => "view data"
• reverseTransform: "view data" => "norm data"
Which transformer you need depends on your situation.
To use the view transformer, call addViewTransformer.
So why Use the Model Transformer?
In this example, the field is a text field, and a text field is always expected to be a simple, scalar format
in the "norm" and "view" formats. For this reason, the most appropriate transformer was the "model"
transformer (which converts to/from the norm format - string issue number - to the model format - Issue
object).
The difference between the transformers is subtle and you should always think about what the "norm"
data for a field should really be. For example, the "norm" data for a text field is a string, but is a DateTime
object for a date field.
Using Transformers in a custom Field Type
In the above example, you applied the transformer to a normal text field. This was easy, but has two
downsides:
1) You need to always remember to apply the transformer whenever you're adding a field for issue
numbers.
2) You need to worry about passing in the em option whenever you're creating a form that uses the
transformer.
Because of these, you may choose to create a custom field type. First, create the custom field type class:
Listing 57-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/Acme/TaskBundle/Form/Type/IssueSelectorType.php
namespace Acme\TaskBundle\Form\Type;
use
use
use
use
use
Symfony\Component\Form\AbstractType;
Symfony\Component\Form\FormBuilderInterface;
Acme\TaskBundle\Form\DataTransformer\IssueToNumberTransformer;
Doctrine\Common\Persistence\ObjectManager;
Symfony\Component\OptionsResolver\OptionsResolverInterface;
class IssueSelectorType extends AbstractType
{
/**
* @var ObjectManager
*/
private $om;
/**
* @param ObjectManager $om
*/
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 215
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47 }
public function __construct(ObjectManager $om)
{
$this->om = $om;
}
public function buildForm(FormBuilderInterface $builder, array $options)
{
$transformer = new IssueToNumberTransformer($this->om);
$builder->addModelTransformer($transformer);
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'invalid_message' => 'The selected issue does not exist',
));
}
public function getParent()
{
return 'text';
}
public function getName()
{
return 'issue_selector';
}
Next, register your type as a service and tag it with form.type so that it's recognized as a custom field
type:
Listing 57-6
1 services:
2
acme_demo.type.issue_selector:
3
class: Acme\TaskBundle\Form\Type\IssueSelectorType
4
arguments: ["@doctrine.orm.entity_manager"]
5
tags:
6
- { name: form.type, alias: issue_selector }
Now, whenever you need to use your special issue_selector field type, it's quite easy:
Listing 57-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// src/Acme/TaskBundle/Form/Type/TaskType.php
namespace Acme\TaskBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
class TaskType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('task')
->add('dueDate', null, array('widget' => 'single_text'))
->add('issue', 'issue_selector');
}
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 216
17
18
19
20
21 }
public function getName()
{
return 'task';
}
PDF brought to you by
generated on December 3, 2014
Chapter 57: How to Use Data Transformers | 217
Chapter 58
How to Dynamically Modify Forms Using Form
Events
Often times, a form can't be created statically. In this entry, you'll learn how to customize your form
based on three common use-cases:
1. Customizing your Form Based on the Underlying Data
Example: you have a "Product" form and need to modify/add/remove a field
based on the data on the underlying Product being edited.
2. How to dynamically Generate Forms Based on user Data
Example: you create a "Friend Message" form and need to build a drop-down that contains only
users that are friends with the current authenticated user.
3. Dynamic Generation for Submitted Forms
Example: on a registration form, you have a "country" field and a "state" field which should
populate dynamically based on the value in the "country" field.
If you wish to learn more about the basics behind form events, you can take a look at the Form Events
documentation.
Customizing your Form Based on the Underlying Data
Before jumping right into dynamic form generation, hold on and recall what a bare form class looks like:
Listing 58-1
1
2
3
4
5
6
7
8
// src/Acme/DemoBundle/Form/Type/ProductType.php
namespace Acme\DemoBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
class ProductType extends AbstractType
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 218
9 {
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27 }
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('name');
$builder->add('price');
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'data_class' => 'Acme\DemoBundle\Entity\Product'
));
}
public function getName()
{
return 'product';
}
If this particular section of code isn't already familiar to you, you probably need to take a step back
and first review the Forms chapter before proceeding.
Assume for a moment that this form utilizes an imaginary "Product" class that has only two properties
("name" and "price"). The form generated from this class will look the exact same regardless if a new
Product is being created or if an existing product is being edited (e.g. a product fetched from the
database).
Suppose now, that you don't want the user to be able to change the name value once the object has been
created. To do this, you can rely on Symfony's EventDispatcher system to analyze the data on the object
and modify the form based on the Product object's data. In this entry, you'll learn how to add this level
of flexibility to your forms.
Adding an Event Listener to a Form Class
So, instead of directly adding that name widget, the responsibility of creating that particular field is
delegated to an event listener:
Listing 58-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// src/Acme/DemoBundle/Form/Type/ProductType.php
namespace Acme\DemoBundle\Form\Type;
// ...
use Symfony\Component\Form\FormEvent;
use Symfony\Component\Form\FormEvents;
class ProductType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('price');
$builder->addEventListener(FormEvents::PRE_SET_DATA, function (FormEvent $event) {
// ... adding the name field if needed
});
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 219
17
18
19
20 }
}
// ...
The goal is to create a name field only if the underlying Product object is new (e.g. hasn't been persisted
to the database). Based on that, the event listener might look like the following:
Listing 58-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// ...
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
$builder->addEventListener(FormEvents::PRE_SET_DATA, function (FormEvent $event) {
$product = $event->getData();
$form = $event->getForm();
//
//
//
if
check if the Product object is "new"
If no data is passed to the form, the data is "null".
This should be considered a new "Product"
(!$product || null === $product->getId()) {
$form->add('name', 'text');
}
});
}
The FormEvents::PRE_SET_DATA line actually resolves to the string form.pre_set_data.
FormEvents1 serves an organizational purpose. It is a centralized location in which you can find all
of the various form events available. You can view the full list of form events via the FormEvents2
class.
Adding an Event Subscriber to a Form Class
For better reusability or if there is some heavy logic in your event listener, you can also move the logic for
creating the name field to an event subscriber:
Listing 58-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/Acme/DemoBundle/Form/Type/ProductType.php
namespace Acme\DemoBundle\Form\Type;
// ...
use Acme\DemoBundle\Form\EventListener\AddNameFieldSubscriber;
class ProductType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('price');
$builder->addEventSubscriber(new AddNameFieldSubscriber());
}
1. http://api.symfony.com/master/Symfony/Component/Form/FormEvents.html
2. http://api.symfony.com/master/Symfony/Component/Form/FormEvents.html
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 220
16
17 }
// ...
Now the logic for creating the name field resides in it own subscriber class:
Listing 58-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/Acme/DemoBundle/Form/EventListener/AddNameFieldSubscriber.php
namespace Acme\DemoBundle\Form\EventListener;
use Symfony\Component\Form\FormEvent;
use Symfony\Component\Form\FormEvents;
use Symfony\Component\EventDispatcher\EventSubscriberInterface;
class AddNameFieldSubscriber implements EventSubscriberInterface
{
public static function getSubscribedEvents()
{
// Tells the dispatcher that you want to listen on the form.pre_set_data
// event and that the preSetData method should be called.
return array(FormEvents::PRE_SET_DATA => 'preSetData');
}
public function preSetData(FormEvent $event)
{
$product = $event->getData();
$form = $event->getForm();
if (!$product || null === $product->getId()) {
$form->add('name', 'text');
}
}
}
How to dynamically Generate Forms Based on user Data
Sometimes you want a form to be generated dynamically based not only on data from the form but also
on something else - like some data from the current user. Suppose you have a social website where a
user can only message people marked as friends on the website. In this case, a "choice list" of whom to
message should only contain users that are the current user's friends.
Creating the Form Type
Using an event listener, your form might look like this:
Listing 58-6
1
2
3
4
5
6
7
8
9
10
11
// src/Acme/DemoBundle/Form/Type/FriendMessageFormType.php
namespace Acme\DemoBundle\Form\Type;
use
use
use
use
use
use
Symfony\Component\Form\AbstractType;
Symfony\Component\Form\FormBuilderInterface;
Symfony\Component\Form\FormEvents;
Symfony\Component\Form\FormEvent;
Symfony\Component\Security\Core\SecurityContext;
Symfony\Component\OptionsResolver\OptionsResolverInterface;
class FriendMessageFormType extends AbstractType
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 221
12 {
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32 }
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('subject', 'text')
->add('body', 'textarea')
;
$builder->addEventListener(FormEvents::PRE_SET_DATA, function (FormEvent $event) {
// ... add a choice list of friends of the current application user
});
}
public function getName()
{
return 'acme_friend_message';
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
}
The problem is now to get the current user and create a choice field that contains only this user's friends.
Luckily it is pretty easy to inject a service inside of the form. This can be done in the constructor:
Listing 58-7
1
2
3
4
5
6
private $securityContext;
public function __construct(SecurityContext $securityContext)
{
$this->securityContext = $securityContext;
}
You might wonder, now that you have access to the User (through the security context), why
not just use it directly in buildForm and omit the event listener? This is because doing so in the
buildForm method would result in the whole form type being modified and not just this one form
instance. This may not usually be a problem, but technically a single form type could be used on a
single request to create many forms or fields.
Customizing the Form Type
Now that you have all the basics in place you can take advantage of the SecurityContext and fill in the
listener logic:
Listing 58-8
1
2
3
4
5
6
7
8
9
10
11
// src/Acme/DemoBundle/FormType/FriendMessageFormType.php
use Symfony\Component\Security\Core\SecurityContext;
use Doctrine\ORM\EntityRepository;
// ...
class FriendMessageFormType extends AbstractType
{
private $securityContext;
public function __construct(SecurityContext $securityContext)
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 222
12
{
13
$this->securityContext = $securityContext;
14
}
15
16
public function buildForm(FormBuilderInterface $builder, array $options)
17
{
18
$builder
19
->add('subject', 'text')
20
->add('body', 'textarea')
21
;
22
23
// grab the user, do a quick sanity check that one exists
24
$user = $this->securityContext->getToken()->getUser();
25
if (!$user) {
26
throw new \LogicException(
27
'The FriendMessageFormType cannot be used without an authenticated user!'
28
);
29
}
30
31
$builder->addEventListener(
32
FormEvents::PRE_SET_DATA,
33
function (FormEvent $event) use ($user) {
34
$form = $event->getForm();
35
36
$formOptions = array(
37
'class' => 'Acme\DemoBundle\Entity\User',
38
'property' => 'fullName',
39
'query_builder' => function (EntityRepository $er) use ($user) {
40
// build a custom query
41
// return $er->createQueryBuilder('u')->addOrderBy('fullName',
42 'DESC');
43
44
// or call a method on your repository that returns the query
45 builder
46
// the $er is an instance of your UserRepository
47
// return $er->createOrderByFullNameQueryBuilder();
48
},
49
);
50
51
// create the field, this is similar the $builder->add()
52
// field name, field type, data, options
53
$form->add('friend', 'entity', $formOptions);
54
}
55
);
56
}
57
// ...
}
The multiple and expanded form options will default to false because the type of the friend field
is entity.
Using the Form
Our form is now ready to use and there are two possible ways to use it inside of a controller:
1. create it manually and remember to pass the security context to it;
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 223
or
2. define it as a service.
a) Creating the Form manually
This is very simple, and is probably the better approach unless you're using your new form type in many
places or embedding it into other forms:
Listing 58-9
1 class FriendMessageController extends Controller
2 {
3
public function newAction(Request $request)
4
{
5
$securityContext = $this->container->get('security.context');
6
$form = $this->createForm(
7
new FriendMessageFormType($securityContext)
8
);
9
10
// ...
11
}
12 }
b) Defining the Form as a Service
To define your form as a service, just create a normal service and then tag it with form.type.
Listing 58-10
1 # app/config/config.yml
2 services:
3
acme.form.friend_message:
4
class: Acme\DemoBundle\Form\Type\FriendMessageFormType
5
arguments: ["@security.context"]
6
tags:
7
- { name: form.type, alias: acme_friend_message }
If you wish to create it from within a controller or any other service that has access to the form factory,
you then use:
Listing 58-11
1
2
3
4
5
6
7
8
9
10
11
use Symfony\Component\DependencyInjection\ContainerAware;
class FriendMessageController extends ContainerAware
{
public function newAction(Request $request)
{
$form = $this->get('form.factory')->create('acme_friend_message');
// ...
}
}
If you extend the Symfony\Bundle\FrameworkBundle\Controller\Controller class, you can simply
call:
Listing 58-12
1 $form = $this->createForm('acme_friend_message');
You can also easily embed the form type into another form:
Listing 58-13
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 224
1
2
3
4
5
// inside some other "form type" class
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('message', 'acme_friend_message');
}
Dynamic Generation for Submitted Forms
Another case that can appear is that you want to customize the form specific to the data that was
submitted by the user. For example, imagine you have a registration form for sports gatherings. Some
events will allow you to specify your preferred position on the field. This would be a choice field for
example. However the possible choices will depend on each sport. Football will have attack, defense,
goalkeeper etc... Baseball will have a pitcher but will not have a goalkeeper. You will need the correct
options in order for validation to pass.
The meetup is passed as an entity field to the form. So we can access each sport like this:
Listing 58-14
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
// src/Acme/DemoBundle/Form/Type/SportMeetupType.php
namespace Acme\DemoBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\Form\FormEvent;
use Symfony\Component\Form\FormEvents;
// ...
class SportMeetupType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('sport', 'entity', array(
'class'
=> 'AcmeDemoBundle:Sport',
'placeholder' => '',
))
;
$builder->addEventListener(
FormEvents::PRE_SET_DATA,
function (FormEvent $event) {
$form = $event->getForm();
// this would be your entity, i.e. SportMeetup
$data = $event->getData();
$sport = $data->getSport();
$positions = null === $sport ? array() : $sport->getAvailablePositions();
$form->add('position', 'entity', array(
'class'
=> 'AcmeDemoBundle:Position',
'placeholder' => '',
'choices'
=> $positions,
));
}
);
}
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 225
40
41
42 }
// ...
New in version 2.6: The placeholder option was introduced in Symfony 2.6 in favor of
empty_value, which is available prior to 2.6.
When you're building this form to display to the user for the first time, then this example works perfectly.
However, things get more difficult when you handle the form submission. This is because the
PRE_SET_DATA event tells us the data that you're starting with (e.g. an empty SportMeetup object), not
the submitted data.
On a form, we can usually listen to the following events:
•
•
•
•
•
PRE_SET_DATA
POST_SET_DATA
PRE_SUBMIT
SUBMIT
POST_SUBMIT
New in version 2.3: The events PRE_SUBMIT, SUBMIT and POST_SUBMIT were introduced in Symfony
2.3. Before, they were named PRE_BIND, BIND and POST_BIND.
The key is to add a POST_SUBMIT listener to the field that your new field depends on. If you add a
POST_SUBMIT listener to a form child (e.g. sport), and add new children to the parent form, the Form
component will detect the new field automatically and map it to the submitted client data.
The type would now look like:
Listing 58-15
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/DemoBundle/Form/Type/SportMeetupType.php
namespace Acme\DemoBundle\Form\Type;
// ...
use Symfony\Component\Form\FormInterface;
use Acme\DemoBundle\Entity\Sport;
class SportMeetupType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('sport', 'entity', array(
'class'
=> 'AcmeDemoBundle:Sport',
'placeholder' => '',
));
;
$formModifier = function (FormInterface $form, Sport $sport = null) {
$positions = null === $sport ? array() : $sport->getAvailablePositions();
$form->add('position', 'entity', array(
'class'
=> 'AcmeDemoBundle:Position',
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 226
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54 }
'placeholder' => '',
'choices'
=> $positions,
));
};
$builder->addEventListener(
FormEvents::PRE_SET_DATA,
function (FormEvent $event) use ($formModifier) {
// this would be your entity, i.e. SportMeetup
$data = $event->getData();
$formModifier($event->getForm(), $data->getSport());
}
);
$builder->get('sport')->addEventListener(
FormEvents::POST_SUBMIT,
function (FormEvent $event) use ($formModifier) {
// It's important here to fetch $event->getForm()->getData(), as
// $event->getData() will get you the client data (that is, the ID)
$sport = $event->getForm()->getData();
// since we've added the listener to the child, we'll have to pass on
// the parent to the callback functions!
$formModifier($event->getForm()->getParent(), $sport);
}
);
}
// ...
You can see that you need to listen on these two events and have different callbacks only because in two
different scenarios, the data that you can use is available in different events. Other than that, the listeners
always perform exactly the same things on a given form.
One piece that is still missing is the client-side updating of your form after the sport is selected. This
should be handled by making an AJAX call back to your application. Assume that you have a sport
meetup creation controller:
Listing 58-16
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/Acme/DemoBundle/Controller/MeetupController.php
namespace Acme\DemoBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
use Symfony\Component\HttpFoundation\Request;
use Acme\DemoBundle\Entity\SportMeetup;
use Acme\DemoBundle\Form\Type\SportMeetupType;
// ...
class MeetupController extends Controller
{
public function createAction(Request $request)
{
$meetup = new SportMeetup();
$form = $this->createForm(new SportMeetupType(), $meetup);
$form->handleRequest($request);
if ($form->isValid()) {
// ... save the meetup, redirect etc.
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 227
19
20
21
22
23
24
25
26
27
28 }
}
return $this->render(
'AcmeDemoBundle:Meetup:create.html.twig',
array('form' => $form->createView())
);
}
// ...
The associated template uses some JavaScript to update the position form field according to the current
selection in the sport field:
Listing 58-17
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
{# src/Acme/DemoBundle/Resources/views/Meetup/create.html.twig #}
{{ form_start(form) }}
{{ form_row(form.sport) }}
{# <select id="meetup_sport" ... #}
{{ form_row(form.position) }} {# <select id="meetup_position" ... #}
{# ... #}
{{ form_end(form) }}
<script>
var $sport = $('#meetup_sport');
// When sport gets selected ...
$sport.change(function() {
// ... retrieve the corresponding form.
var $form = $(this).closest('form');
// Simulate form data, but only include the selected sport value.
var data = {};
data[$sport.attr('name')] = $sport.val();
// Submit data via AJAX to the form's action path.
$.ajax({
url : $form.attr('action'),
type: $form.attr('method'),
data : data,
success: function(html) {
// Replace current position field ...
$('#meetup_position').replaceWith(
// ... with the returned one from the AJAX response.
$(html).find('#meetup_position')
);
// Position field now displays the appropriate positions.
}
});
});
</script>
The major benefit of submitting the whole form to just extract the updated position field is that no
additional server-side code is needed; all the code from above to generate the submitted form can be
reused.
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 228
Suppressing Form Validation
To suppress form validation you can use the POST_SUBMIT event and prevent the ValidationListener3
from being called.
The reason for needing to do this is that even if you set group_validation to false there are still some
integrity checks executed. For example an uploaded file will still be checked to see if it is too large and
the form will still check to see if non-existing fields were submitted. To disable all of this, use a listener:
Listing 58-18
1
2
3
4
5
6
7
8
9
10
11
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\Form\FormEvents;
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->addEventListener(FormEvents::POST_SUBMIT, function ($event) {
$event->stopPropagation();
}, 900); // Always set a higher priority than ValidationListener
// ...
}
By doing this, you may accidentally disable something more than just form validation, since the
POST_SUBMIT event may have other listeners.
3. http://api.symfony.com/master/Symfony/Component/Form/Extension/Validator/EventListener/ValidationListener.html
PDF brought to you by
generated on December 3, 2014
Chapter 58: How to Dynamically Modify Forms Using Form Events | 229
Chapter 59
How to Embed a Collection of Forms
In this entry, you'll learn how to create a form that embeds a collection of many other forms. This could
be useful, for example, if you had a Task class and you wanted to edit/create/remove many Tag objects
related to that Task, right inside the same form.
In this entry, it's loosely assumed that you're using Doctrine as your database store. But if you're
not using Doctrine (e.g. Propel or just a database connection), it's all very similar. There are only a
few parts of this tutorial that really care about "persistence".
If you are using Doctrine, you'll need to add the Doctrine metadata, including the ManyToMany
association mapping definition on the Task's tags property.
First, suppose that each Task belongs to multiple Tag objects. Start by creating a simple Task class:
Listing 59-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// src/Acme/TaskBundle/Entity/Task.php
namespace Acme\TaskBundle\Entity;
use Doctrine\Common\Collections\ArrayCollection;
class Task
{
protected $description;
protected $tags;
public function __construct()
{
$this->tags = new ArrayCollection();
}
public function getDescription()
{
return $this->description;
}
public function setDescription($description)
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 230
23
24
25
26
27
28
29
30
31 }
{
$this->description = $description;
}
public function getTags()
{
return $this->tags;
}
The ArrayCollection is specific to Doctrine and is basically the same as using an array (but it
must be an ArrayCollection if you're using Doctrine).
Now, create a Tag class. As you saw above, a Task can have many Tag objects:
Listing 59-2
1
2
3
4
5
6
7
// src/Acme/TaskBundle/Entity/Tag.php
namespace Acme\TaskBundle\Entity;
class Tag
{
public $name;
}
The name property is public here, but it can just as easily be protected or private (but then it would
need getName and setName methods).
Then, create a form class so that a Tag object can be modified by the user:
Listing 59-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
// src/Acme/TaskBundle/Form/Type/TagType.php
namespace Acme\TaskBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
class TagType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('name');
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'data_class' => 'Acme\TaskBundle\Entity\Tag',
));
}
public function getName()
{
return 'tag';
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 231
25
26 }
}
With this, you have enough to render a tag form by itself. But since the end goal is to allow the tags of a
Task to be modified right inside the task form itself, create a form for the Task class.
Notice that you embed a collection of TagType forms using the collection field type:
Listing 59-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
// src/Acme/TaskBundle/Form/Type/TaskType.php
namespace Acme\TaskBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
class TaskType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('description');
$builder->add('tags', 'collection', array('type' => new TagType()));
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'data_class' => 'Acme\TaskBundle\Entity\Task',
));
}
public function getName()
{
return 'task';
}
}
In your controller, you'll now initialize a new instance of TaskType:
Listing 59-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/Acme/TaskBundle/Controller/TaskController.php
namespace Acme\TaskBundle\Controller;
use
use
use
use
use
Acme\TaskBundle\Entity\Task;
Acme\TaskBundle\Entity\Tag;
Acme\TaskBundle\Form\Type\TaskType;
Symfony\Component\HttpFoundation\Request;
Symfony\Bundle\FrameworkBundle\Controller\Controller;
class TaskController extends Controller
{
public function newAction(Request $request)
{
$task = new Task();
// dummy code - this is here just so that the Task has some tags
// otherwise, this isn't an interesting example
$tag1 = new Tag();
$tag1->name = 'tag1';
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 232
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38 }
$task->getTags()->add($tag1);
$tag2 = new Tag();
$tag2->name = 'tag2';
$task->getTags()->add($tag2);
// end dummy code
$form = $this->createForm(new TaskType(), $task);
$form->handleRequest($request);
if ($form->isValid()) {
// ... maybe do some form processing, like saving the Task and Tag objects
}
return $this->render('AcmeTaskBundle:Task:new.html.twig', array(
'form' => $form->createView(),
));
}
The corresponding template is now able to render both the description field for the task form as well
as all the TagType forms for any tags that are already related to this Task. In the above controller, I added
some dummy code so that you can see this in action (since a Task has zero tags when first created).
Listing 59-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
{# src/Acme/TaskBundle/Resources/views/Task/new.html.twig #}
{# ... #}
{{ form_start(form) }}
{# render the task's only field: description #}
{{ form_row(form.description) }}
<h3>Tags</h3>
<ul class="tags">
{# iterate over each existing tag and render its only field: name #}
{% for tag in form.tags %}
<li>{{ form_row(tag.name) }}</li>
{% endfor %}
</ul>
{{ form_end(form) }}
{# ... #}
When the user submits the form, the submitted data for the tags field are used to construct an
ArrayCollection of Tag objects, which is then set on the tag field of the Task instance.
The tags collection is accessible naturally via $task->getTags() and can be persisted to the database or
used however you need.
So far, this works great, but this doesn't allow you to dynamically add new tags or delete existing tags.
So, while editing existing tags will work great, your user can't actually add any new tags yet.
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 233
In this entry, you embed only one collection, but you are not limited to this. You can also embed
nested collection as many level down as you like. But if you use Xdebug in your development setup,
you may receive a Maximum function nesting level of '100' reached, aborting! error.
This is due to the xdebug.max_nesting_level PHP setting, which defaults to 100.
This directive limits recursion to 100 calls which may not be enough for rendering the form in
the template if you render the whole form at once (e.g form_widget(form)). To fix this you can
set this directive to a higher value (either via a php.ini file or via ini_set1, for example in app/
autoload.php) or render each form field by hand using form_row.
Allowing "new" Tags with the "Prototype"
Allowing the user to dynamically add new tags means that you'll need to use some JavaScript. Previously
you added two tags to your form in the controller. Now let the user add as many tag forms as they need
directly in the browser. This will be done through a bit of JavaScript.
The first thing you need to do is to let the form collection know that it will receive an unknown number
of tags. So far you've added two tags and the form type expects to receive exactly two, otherwise an
error will be thrown: This form should not contain extra fields. To make this flexible, add the
allow_add option to your collection field:
Listing 59-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/TaskBundle/Form/Type/TaskType.php
// ...
use Symfony\Component\Form\FormBuilderInterface;
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('description');
$builder->add('tags', 'collection', array(
'type'
=> new TagType(),
'allow_add'
=> true,
));
}
In addition to telling the field to accept any number of submitted objects, the allow_add also makes a
"prototype" variable available to you. This "prototype" is a little "template" that contains all the HTML
to be able to render any new "tag" forms. To render it, make the following change to your template:
Listing 59-8
1 <ul class="tags" data-prototype="{{ form_widget(form.tags.vars.prototype)|e }}">
2
...
3 </ul>
If you render your whole "tags" sub-form at once (e.g. form_row(form.tags)), then the prototype
is automatically available on the outer div as the data-prototype attribute, similar to what you
see above.
1. http://php.net/manual/en/function.ini-set.php
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 234
The form.tags.vars.prototype is a form element that looks and feels just like the individual
form_widget(tag) elements inside your for loop. This means that you can call form_widget,
form_row or form_label on it. You could even choose to render only one of its fields (e.g. the name
field):
Listing 59-9
1 {{ form_widget(form.tags.vars.prototype.name)|e }}
On the rendered page, the result will look something like this:
Listing 59-10
1 <ul class="tags" data-prototype="&lt;div&gt;&lt;label class=&quot;
required&quot;&gt;__name__&lt;/label&gt;&lt;div
id=&quot;task_tags___name__&quot;&gt;&lt;div&gt;&lt;label
for=&quot;task_tags___name___name&quot; class=&quot; required&quot;&gt;Name&lt;/
label&gt;&lt;input type=&quot;text&quot; id=&quot;task_tags___name___name&quot;
name=&quot;task[tags][__name__][name]&quot; required=&quot;required&quot;
maxlength=&quot;255&quot; /&gt;&lt;/div&gt;&lt;/div&gt;&lt;/div&gt;">
The goal of this section will be to use JavaScript to read this attribute and dynamically add new tag forms
when the user clicks a "Add a tag" link. To make things simple, this example uses jQuery and assumes
you have it included somewhere on your page.
Add a script tag somewhere on your page so you can start writing some JavaScript.
First, add a link to the bottom of the "tags" list via JavaScript. Second, bind to the "click" event of that
link so you can add a new tag form (addTagForm will be show next):
Listing 59-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
var $collectionHolder;
// setup an "add a tag" link
var $addTagLink = $('<a href="#" class="add_tag_link">Add a tag</a>');
var $newLinkLi = $('<li></li>').append($addTagLink);
jQuery(document).ready(function() {
// Get the ul that holds the collection of tags
$collectionHolder = $('ul.tags');
// add the "add a tag" anchor and li to the tags ul
$collectionHolder.append($newLinkLi);
// count the current form inputs we have (e.g. 2), use that as the new
// index when inserting a new item (e.g. 2)
$collectionHolder.data('index', $collectionHolder.find(':input').length);
$addTagLink.on('click', function(e) {
// prevent the link from creating a "#" on the URL
e.preventDefault();
// add a new tag form (see next code block)
addTagForm($collectionHolder, $newLinkLi);
});
});
The addTagForm function's job will be to use the data-prototype attribute to dynamically add a new
form when this link is clicked. The data-prototype HTML contains the tag text input element with a
name of task[tags][__name__][name] and id of task_tags___name___name. The __name__ is a little
"placeholder", which you'll replace with a unique, incrementing number (e.g. task[tags][3][name]).
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 235
The actual code needed to make this all work can vary quite a bit, but here's one example:
Listing 59-12
1 function addTagForm($collectionHolder, $newLinkLi) {
2
// Get the data-prototype explained earlier
3
var prototype = $collectionHolder.data('prototype');
4
5
// get the new index
6
var index = $collectionHolder.data('index');
7
8
// Replace '__name__' in the prototype's HTML to
9
// instead be a number based on how many items we have
10
var newForm = prototype.replace(/__name__/g, index);
11
12
// increase the index with one for the next item
13
$collectionHolder.data('index', index + 1);
14
15
// Display the form in the page in an li, before the "Add a tag" link li
16
var $newFormLi = $('<li></li>').append(newForm);
17
$newLinkLi.before($newFormLi);
18 }
It is better to separate your JavaScript in real JavaScript files than to write it inside the HTML as is
done here.
Now, each time a user clicks the Add a tag link, a new sub form will appear on the page. When the form
is submitted, any new tag forms will be converted into new Tag objects and added to the tags property
of the Task object.
You can find a working example in this JSFiddle2.
To make handling these new tags easier, add an "adder" and a "remover" method for the tags in the Task
class:
Listing 59-13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/Acme/TaskBundle/Entity/Task.php
namespace Acme\TaskBundle\Entity;
// ...
class Task
{
// ...
public function addTag(Tag $tag)
{
$this->tags->add($tag);
}
public function removeTag(Tag $tag)
{
// ...
}
}
Next, add a by_reference option to the tags field and set it to false:
2. http://jsfiddle.net/847Kf/4/
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 236
Listing 59-14
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/TaskBundle/Form/Type/TaskType.php
// ...
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
$builder->add('tags', 'collection', array(
// ...
'by_reference' => false,
));
}
With these two changes, when the form is submitted, each new Tag object is added to the Task class by
calling the addTag method. Before this change, they were added internally by the form by calling $task>getTags()->add($tag). That was just fine, but forcing the use of the "adder" method makes handling
these new Tag objects easier (especially if you're using Doctrine, which we talk about next!).
You have to create both addTag and removeTag methods, otherwise the form will still use setTag
even if by_reference is false. You'll learn more about the removeTag method later in this article.
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 237
Doctrine: Cascading Relations and saving the "Inverse" side
To save the new tags with Doctrine, you need to consider a couple more things. First, unless you
iterate over all of the new Tag objects and call $em->persist($tag) on each, you'll receive an error
from Doctrine:
A new entity was found through the relationship Acme\TaskBundle\Entity\Task#tags
that was not configured to cascade persist operations for entity...
To fix this, you may choose to "cascade" the persist operation automatically from the Task object
to any related tags. To do this, add the cascade option to your ManyToMany metadata:
Listing 59-15
1
2
3
4
5
6
7
8
// src/Acme/TaskBundle/Entity/Task.php
// ...
/**
* @ORM\ManyToMany(targetEntity="Tag", cascade={"persist"})
*/
protected $tags;
A second potential issue deals with the Owning Side and Inverse Side3 of Doctrine relationships. In
this example, if the "owning" side of the relationship is "Task", then persistence will work fine as
the tags are properly added to the Task. However, if the owning side is on "Tag", then you'll need
to do a little bit more work to ensure that the correct side of the relationship is modified.
The trick is to make sure that the single "Task" is set on each "Tag". One easy way to do this is to
add some extra logic to addTag(), which is called by the form type since by_reference is set to
false:
Listing 59-16
1
2
3
4
5
6
7
8
9
// src/Acme/TaskBundle/Entity/Task.php
// ...
public function addTag(Tag $tag)
{
$tag->addTask($this);
$this->tags->add($tag);
}
Inside Tag, just make sure you have an addTask method:
Listing 59-17
1
2
3
4
5
6
7
8
9
// src/Acme/TaskBundle/Entity/Tag.php
// ...
public function addTask(Task $task)
{
if (!$this->tasks->contains($task)) {
$this->tasks->add($task);
}
}
If you have a one-to-many relationship, then the workaround is similar, except that you can simply
call setTask from inside addTag.
3. http://docs.doctrine-project.org/en/latest/reference/unitofwork-associations.html
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 238
Allowing Tags to be Removed
The next step is to allow the deletion of a particular item in the collection. The solution is similar to
allowing tags to be added.
Start by adding the allow_delete option in the form Type:
Listing 59-18
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/TaskBundle/Form/Type/TaskType.php
// ...
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
$builder->add('tags', 'collection', array(
// ...
'allow_delete' => true,
));
}
Now, you need to put some code into the removeTag method of Task:
Listing 59-19
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/TaskBundle/Entity/Task.php
// ...
class Task
{
// ...
public function removeTag(Tag $tag)
{
$this->tags->removeElement($tag);
}
}
Template Modifications
The allow_delete option has one consequence: if an item of a collection isn't sent on submission, the
related data is removed from the collection on the server. The solution is thus to remove the form element
from the DOM.
First, add a "delete this tag" link to each tag form:
Listing 59-20
1 jQuery(document).ready(function() {
2
// Get the ul that holds the collection of tags
3
$collectionHolder = $('ul.tags');
4
5
// add a delete link to all of the existing tag form li elements
6
$collectionHolder.find('li').each(function() {
7
addTagFormDeleteLink($(this));
8
});
9
10
// ... the rest of the block from above
11 });
12
13 function addTagForm() {
14
// ...
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 239
15
16
17
18 }
// add a delete link to the new form
addTagFormDeleteLink($newFormLi);
The addTagFormDeleteLink function will look something like this:
Listing 59-21
1 function addTagFormDeleteLink($tagFormLi) {
2
var $removeFormA = $('<a href="#">delete this tag</a>');
3
$tagFormLi.append($removeFormA);
4
5
$removeFormA.on('click', function(e) {
6
// prevent the link from creating a "#" on the URL
7
e.preventDefault();
8
9
// remove the li for the tag form
10
$tagFormLi.remove();
11
});
12 }
When a tag form is removed from the DOM and submitted, the removed Tag object will not be included
in the collection passed to setTags. Depending on your persistence layer, this may or may not be enough
to actually remove the relationship between the removed Tag and Task object.
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 240
Doctrine: Ensuring the database persistence
When removing objects in this way, you may need to do a little bit more work to ensure that the
relationship between the Task and the removed Tag is properly removed.
In Doctrine, you have two sides of the relationship: the owning side and the inverse side. Normally
in this case you'll have a many-to-many relationship and the deleted tags will disappear and persist
correctly (adding new tags also works effortlessly).
But if you have a one-to-many relationship or a many-to-many relationship with a mappedBy on the
Task entity (meaning Task is the "inverse" side), you'll need to do more work for the removed tags
to persist correctly.
In this case, you can modify the controller to remove the relationship on the removed tag. This
assumes that you have some editAction which is handling the "update" of your Task:
Listing 59-22
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
// src/Acme/TaskBundle/Controller/TaskController.php
use Doctrine\Common\Collections\ArrayCollection;
// ...
public function editAction($id, Request $request)
{
$em = $this->getDoctrine()->getManager();
$task = $em->getRepository('AcmeTaskBundle:Task')->find($id);
if (!$task) {
throw $this->createNotFoundException('No task found for is '.$id);
}
$originalTags = new ArrayCollection();
// Create an ArrayCollection of the current Tag objects in the database
foreach ($task->getTags() as $tag) {
$originalTags->add($tag);
}
$editForm = $this->createForm(new TaskType(), $task);
$editForm->handleRequest($request);
if ($editForm->isValid()) {
// remove the relationship between the tag and the Task
foreach ($originalTags as $tag) {
if (false === $task->getTags()->contains($tag)) {
// remove the Task from the Tag
$tag->getTasks()->removeElement($task);
// if it was a many-to-one relationship, remove the relationship
like this
PDF brought to you by
generated on December 3, 2014
// $tag->setTask(null);
$em->persist($tag);
// if you wanted to delete the Tag entirely, you can also do that
// $em->remove($tag);
}
}
$em->persist($task);
Chapter 59: How to Embed a Collection of Forms | 241
46
47
48
49
50
51
52
$em->flush();
// redirect back to some edit page
return $this->redirect($this->generateUrl('task_edit', array('id' => $id)));
}
// render some form template
}
As you can see, adding and removing the elements correctly can be tricky. Unless you have a manyto-many relationship where Task is the "owning" side, you'll need to do extra work to make sure
that the relationship is properly updated (whether you're adding new tags or removing existing
tags) on each Tag object itself.
PDF brought to you by
generated on December 3, 2014
Chapter 59: How to Embed a Collection of Forms | 242
Chapter 60
How to Create a Custom Form Field Type
Symfony comes with a bunch of core field types available for building forms. However there are situations
where you may want to create a custom form field type for a specific purpose. This recipe assumes
you need a field definition that holds a person's gender, based on the existing choice field. This section
explains how the field is defined, how you can customize its layout and finally, how you can register it
for use in your application.
Defining the Field Type
In order to create the custom field type, first you have to create the class representing the field. In this
situation the class holding the field type will be called GenderType and the file will be stored in the default
location for form fields, which is <BundleName>\Form\Type. Make sure the field extends AbstractType1:
Listing 60-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/Acme/DemoBundle/Form/Type/GenderType.php
namespace Acme\DemoBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
class GenderType extends AbstractType
{
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'choices' => array(
'm' => 'Male',
'f' => 'Female',
)
));
}
public function getParent()
{
1. http://api.symfony.com/master/Symfony/Component/Form/AbstractType.html
PDF brought to you by
generated on December 3, 2014
Chapter 60: How to Create a Custom Form Field Type | 243
21
22
23
24
25
26
27
28 }
return 'choice';
}
public function getName()
{
return 'gender';
}
The location of this file is not important - the Form\Type directory is just a convention.
Here, the return value of the getParent function indicates that you're extending the choice field type.
This means that, by default, you inherit all of the logic and rendering of that field type. To see some of
the logic, check out the ChoiceType2 class. There are three methods that are particularly important:
• buildForm() - Each field type has a buildForm method, which is where you configure and
build any field(s). Notice that this is the same method you use to setup your forms, and it
works the same here.
• buildView() - This method is used to set any extra variables you'll need when rendering your
field in a template. For example, in ChoiceType3, a multiple variable is set and used in the
template to set (or not set) the multiple attribute on the select field. See Creating a Template
for the Field for more details.
• setDefaultOptions() - This defines options for your form type that can be used in
buildForm() and buildView(). There are a lot of options common to all fields (see form Field
Type), but you can create any others that you need here.
If you're creating a field that consists of many fields, then be sure to set your "parent" type as form
or something that extends form. Also, if you need to modify the "view" of any of your child types
from your parent type, use the finishView() method.
The getName() method returns an identifier which should be unique in your application. This is used in
various places, such as when customizing how your form type will be rendered.
The goal of this field was to extend the choice type to enable selection of a gender. This is achieved by
fixing the choices to a list of possible genders.
Creating a Template for the Field
Each field type is rendered by a template fragment, which is determined in part by the value of your
getName() method. For more information, see What are Form Themes?.
In this case, since the parent field is choice, you don't need to do any work as the custom field type will
automatically be rendered like a choice type. But for the sake of this example, suppose that when your
field is "expanded" (i.e. radio buttons or checkboxes, instead of a select field), you want to always render
it in a ul element. In your form theme template (see above link for details), create a gender_widget block
to handle this:
Listing 60-2
2. https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Form/Extension/Core/Type/ChoiceType.php
3. https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Form/Extension/Core/Type/ChoiceType.php
PDF brought to you by
generated on December 3, 2014
Chapter 60: How to Create a Custom Form Field Type | 244
1 {# src/Acme/DemoBundle/Resources/views/Form/fields.html.twig #}
2 {% block gender_widget %}
3
{% spaceless %}
4
{% if expanded %}
5
<ul {{ block('widget_container_attributes') }}>
6
{% for child in form %}
7
<li>
8
{{ form_widget(child) }}
9
{{ form_label(child) }}
10
</li>
11
{% endfor %}
12
</ul>
13
{% else %}
14
{# just let the choice widget render the select tag #}
15
{{ block('choice_widget') }}
16
{% endif %}
17
{% endspaceless %}
18 {% endblock %}
Make sure the correct widget prefix is used. In this example the name should be gender_widget,
according to the value returned by getName. Further, the main config file should point to the
custom form template so that it's used when rendering all forms.
When using Twig this is:
Listing 60-3
1 # app/config/config.yml
2 twig:
3
form_themes:
4
- 'AcmeDemoBundle:Form:fields.html.twig'
For the PHP templating engine, your configuration should look like this:
Listing 60-4
1 # app/config/config.yml
2 framework:
3
templating:
4
form:
5
resources:
6
- 'AcmeDemoBundle:Form'
Using the Field Type
You can now use your custom field type immediately, simply by creating a new instance of the type in
one of your forms:
Listing 60-5
1
2
3
4
5
6
7
8
9
// src/Acme/DemoBundle/Form/Type/AuthorType.php
namespace Acme\DemoBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
class AuthorType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
PDF brought to you by
generated on December 3, 2014
Chapter 60: How to Create a Custom Form Field Type | 245
10
11
12
13
14
15 }
{
$builder->add('gender_code', new GenderType(), array(
'placeholder' => 'Choose a gender',
));
}
But this only works because the GenderType() is very simple. What if the gender codes were stored
in configuration or in a database? The next section explains how more complex field types solve this
problem.
New in version 2.6: The placeholder option was introduced in Symfony 2.6 in favor of
empty_value, which is available prior to 2.6.
Creating your Field Type as a Service
So far, this entry has assumed that you have a very simple custom field type. But if you need access to
configuration, a database connection, or some other service, then you'll want to register your custom type
as a service. For example, suppose that you're storing the gender parameters in configuration:
Listing 60-6
1 # app/config/config.yml
2 parameters:
3
genders:
4
m: Male
5
f: Female
To use the parameter, define your custom field type as a service, injecting the genders parameter value
as the first argument to its to-be-created __construct function:
Listing 60-7
1 # src/Acme/DemoBundle/Resources/config/services.yml
2 services:
3
acme_demo.form.type.gender:
4
class: Acme\DemoBundle\Form\Type\GenderType
5
arguments:
6
- "%genders%"
7
tags:
8
- { name: form.type, alias: gender }
Make sure the services file is being imported. See Importing Configuration with imports for details.
Be sure that the alias attribute of the tag corresponds with the value returned by the getName method
defined earlier. You'll see the importance of this in a moment when you use the custom field type. But
first, add a __construct method to GenderType, which receives the gender configuration:
Listing 60-8
1 // src/Acme/DemoBundle/Form/Type/GenderType.php
2 namespace Acme\DemoBundle\Form\Type;
3
PDF brought to you by
generated on December 3, 2014
Chapter 60: How to Create a Custom Form Field Type | 246
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
// ...
// ...
class GenderType extends AbstractType
{
private $genderChoices;
public function __construct(array $genderChoices)
{
$this->genderChoices = $genderChoices;
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'choices' => $this->genderChoices,
));
}
// ...
}
Great! The GenderType is now fueled by the configuration parameters and registered as a service.
Additionally, because you used the form.type alias in its configuration, using the field is now much
easier:
Listing 60-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// src/Acme/DemoBundle/Form/Type/AuthorType.php
namespace Acme\DemoBundle\Form\Type;
use Symfony\Component\Form\FormBuilderInterface;
// ...
class AuthorType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder->add('gender_code', 'gender', array(
'placeholder' => 'Choose a gender',
));
}
}
Notice that instead of instantiating a new instance, you can just refer to it by the alias used in your service
configuration, gender. Have fun!
PDF brought to you by
generated on December 3, 2014
Chapter 60: How to Create a Custom Form Field Type | 247
Chapter 61
How to Create a Form Type Extension
Custom form field types are great when you need field types with a specific purpose, such as a gender
selector, or a VAT number input.
But sometimes, you don't really need to add new field types - you want to add features on top of existing
types. This is where form type extensions come in.
Form type extensions have 2 main use-cases:
1. You want to add a generic feature to several types (such as adding a "help" text to every field
type);
2. You want to add a specific feature to a single type (such as adding a "download" feature to
the "file" field type).
In both those cases, it might be possible to achieve your goal with custom form rendering, or custom
form field types. But using form type extensions can be cleaner (by limiting the amount of business logic
in templates) and more flexible (you can add several type extensions to a single form type).
Form type extensions can achieve most of what custom field types can do, but instead of being field types
of their own, they plug into existing types.
Imagine that you manage a Media entity, and that each media is associated to a file. Your Media form
uses a file type, but when editing the entity, you would like to see its image automatically rendered next
to the file input.
You could of course do this by customizing how this field is rendered in a template. But field type
extensions allow you to do this in a nice DRY fashion.
Defining the Form Type Extension
Your first task will be to create the form type extension class (called ImageTypeExtension in this article).
By standard, form extensions usually live in the Form\Extension directory of one of your bundles.
When creating a form type extension, you can either implement the FormTypeExtensionInterface1
interface or extend the AbstractTypeExtension2 class. In most cases, it's easier to extend the abstract
class:
1. http://api.symfony.com/master/Symfony/Component/Form/FormTypeExtensionInterface.html
2. http://api.symfony.com/master/Symfony/Component/Form/AbstractTypeExtension.html
PDF brought to you by
generated on December 3, 2014
Chapter 61: How to Create a Form Type Extension | 248
Listing 61-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/DemoBundle/Form/Extension/ImageTypeExtension.php
namespace Acme\DemoBundle\Form\Extension;
use Symfony\Component\Form\AbstractTypeExtension;
class ImageTypeExtension extends AbstractTypeExtension
{
/**
* Returns the name of the type being extended.
*
* @return string The name of the type being extended
*/
public function getExtendedType()
{
return 'file';
}
}
The only method you must implement is the getExtendedType function. It is used to indicate the name
of the form type that will be extended by your extension.
The value you return in the getExtendedType method corresponds to the value returned by the
getName method in the form type class you wish to extend.
In addition to the getExtendedType function, you will probably want to override one of the following
methods:
•
•
•
•
buildForm()
buildView()
setDefaultOptions()
finishView()
For more information on what those methods do, you can refer to the Creating Custom Field Types
cookbook article.
Registering your Form Type Extension as a Service
The next step is to make Symfony aware of your extension. All you need to do is to declare it as a service
by using the form.type_extension tag:
Listing 61-2
1 services:
2
acme_demo_bundle.image_type_extension:
3
class: Acme\DemoBundle\Form\Extension\ImageTypeExtension
4
tags:
5
- { name: form.type_extension, alias: file }
The alias key of the tag is the type of field that this extension should be applied to. In your case, as you
want to extend the file field type, you will use file as an alias.
PDF brought to you by
generated on December 3, 2014
Chapter 61: How to Create a Form Type Extension | 249
Adding the extension Business Logic
The goal of your extension is to display nice images next to file inputs (when the underlying model
contains images). For that purpose, suppose that you use an approach similar to the one described in
How to handle File Uploads with Doctrine: you have a Media model with a file property (corresponding
to the file field in the form) and a path property (corresponding to the image path in the database):
Listing 61-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
// src/Acme/DemoBundle/Entity/Media.php
namespace Acme\DemoBundle\Entity;
use Symfony\Component\Validator\Constraints as Assert;
class Media
{
// ...
/**
* @var string The path - typically stored in the database
*/
private $path;
/**
* @var \Symfony\Component\HttpFoundation\File\UploadedFile
* @Assert\File(maxSize="2M")
*/
public $file;
// ...
/**
* Get the image URL
*
* @return null|string
*/
public function getWebPath()
{
// ... $webPath being the full image URL, to be used in templates
return $webPath;
}
}
Your form type extension class will need to do two things in order to extend the file form type:
1. Override the setDefaultOptions method in order to add an image_path option;
2. Override the buildForm and buildView methods in order to pass the image URL to the view.
The logic is the following: when adding a form field of type file, you will be able to specify a new option:
image_path. This option will tell the file field how to get the actual image URL in order to display it in
the view:
Listing 61-4
1
2
3
4
5
6
7
8
// src/Acme/DemoBundle/Form/Extension/ImageTypeExtension.php
namespace Acme\DemoBundle\Form\Extension;
use
use
use
use
use
Symfony\Component\Form\AbstractTypeExtension;
Symfony\Component\Form\FormView;
Symfony\Component\Form\FormInterface;
Symfony\Component\PropertyAccess\PropertyAccess;
Symfony\Component\OptionsResolver\OptionsResolverInterface;
PDF brought to you by
generated on December 3, 2014
Chapter 61: How to Create a Form Type Extension | 250
9
10 class ImageTypeExtension extends AbstractTypeExtension
11 {
12
/**
13
* Returns the name of the type being extended.
14
*
15
* @return string The name of the type being extended
16
*/
17
public function getExtendedType()
18
{
19
return 'file';
20
}
21
22
/**
23
* Add the image_path option
24
*
25
* @param OptionsResolverInterface $resolver
26
*/
27
public function setDefaultOptions(OptionsResolverInterface $resolver)
28
{
29
$resolver->setOptional(array('image_path'));
30
}
31
32
/**
33
* Pass the image URL to the view
34
*
35
* @param FormView $view
36
* @param FormInterface $form
37
* @param array $options
38
*/
39
public function buildView(FormView $view, FormInterface $form, array $options)
40
{
41
if (array_key_exists('image_path', $options)) {
42
$parentData = $form->getParent()->getData();
43
44
if (null !== $parentData) {
45
$accessor = PropertyAccess::createPropertyAccessor();
46
$imageUrl = $accessor->getValue($parentData, $options['image_path']);
47
} else {
48
$imageUrl = null;
49
}
50
51
// set an "image_url" variable that will be available when rendering this field
52
$view->vars['image_url'] = $imageUrl;
53
}
54
}
55
56 }
Override the File Widget Template Fragment
Each field type is rendered by a template fragment. Those template fragments can be overridden in order
to customize form rendering. For more information, you can refer to the What are Form Themes? article.
In your extension class, you have added a new variable (image_url), but you still need to take advantage
of this new variable in your templates. Specifically, you need to override the file_widget block:
PDF brought to you by
generated on December 3, 2014
Chapter 61: How to Create a Form Type Extension | 251
Listing 61-5
1
2
3
4
5
6
7
8
9
10
11
12
13
{# src/Acme/DemoBundle/Resources/views/Form/fields.html.twig #}
{% extends 'form_div_layout.html.twig' %}
{% block file_widget %}
{% spaceless %}
{{ block('form_widget') }}
{% if image_url is not null %}
<img src="{{ asset(image_url) }}"/>
{% endif %}
{% endspaceless %}
{% endblock %}
You will need to change your config file or explicitly specify how you want your form to be
themed in order for Symfony to use your overridden block. See What are Form Themes? for more
information.
Using the Form Type Extension
From now on, when adding a field of type file in your form, you can specify an image_path option that
will be used to display an image next to the file field. For example:
Listing 61-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/Acme/DemoBundle/Form/Type/MediaType.php
namespace Acme\DemoBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
class MediaType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('name', 'text')
->add('file', 'file', array('image_path' => 'webPath'));
}
public function getName()
{
return 'media';
}
}
When displaying the form, if the underlying model has already been associated with an image, you will
see it displayed next to the file input.
PDF brought to you by
generated on December 3, 2014
Chapter 61: How to Create a Form Type Extension | 252
Chapter 62
How to Reduce Code Duplication with
"inherit_data"
New in version 2.3: This inherit_data option was introduced in Symfony 2.3. Before, it was
known as virtual.
The inherit_data form field option can be very useful when you have some duplicated fields in different
entities. For example, imagine you have two entities, a Company and a Customer:
Listing 62-1
Listing 62-2
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/Acme/HelloBundle/Entity/Company.php
namespace Acme\HelloBundle\Entity;
1
2
3
4
5
6
7
8
// src/Acme/HelloBundle/Entity/Customer.php
namespace Acme\HelloBundle\Entity;
class Company
{
private $name;
private $website;
private
private
private
private
$address;
$zipcode;
$city;
$country;
}
class Customer
{
private $firstName;
private $lastName;
PDF brought to you by
generated on December 3, 2014
Chapter 62: How to Reduce Code Duplication with "inherit_data" | 253
9
10
11
12
13 }
private
private
private
private
$address;
$zipcode;
$city;
$country;
As you can see, each entity shares a few of the same fields: address, zipcode, city, country.
Start with building two forms for these entities, CompanyType and CustomerType:
Listing 62-3
Listing 62-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/Acme/HelloBundle/Form/Type/CompanyType.php
namespace Acme\HelloBundle\Form\Type;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/Acme/HelloBundle/Form/Type/CustomerType.php
namespace Acme\HelloBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
class CompanyType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('name', 'text')
->add('website', 'text');
}
}
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\Form\AbstractType;
class CustomerType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
->add('firstName', 'text')
->add('lastName', 'text');
}
}
Instead of including the duplicated fields address, zipcode, city and country in both of these forms,
create a third form called LocationType for that:
Listing 62-5
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/HelloBundle/Form/Type/LocationType.php
namespace Acme\HelloBundle\Form\Type;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
class LocationType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
PDF brought to you by
generated on December 3, 2014
Chapter 62: How to Reduce Code Duplication with "inherit_data" | 254
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30 }
->add('address', 'textarea')
->add('zipcode', 'text')
->add('city', 'text')
->add('country', 'text');
}
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'inherit_data' => true
));
}
public function getName()
{
return 'location';
}
The location form has an interesting option set, namely inherit_data. This option lets the form inherit
its data from its parent form. If embedded in the company form, the fields of the location form will
access the properties of the Company instance. If embedded in the customer form, the fields will access
the properties of the Customer instance instead. Easy, eh?
Instead of setting the inherit_data option inside LocationType, you can also (just like with any
option) pass it in the third argument of $builder->add().
Finally, make this work by adding the location form to your two original forms:
Listing 62-6
Listing 62-7
1
2
3
4
5
6
7
8
9
// src/Acme/HelloBundle/Form/Type/CompanyType.php
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
1
2
3
4
5
6
7
8
9
// src/Acme/HelloBundle/Form/Type/CustomerType.php
public function buildForm(FormBuilderInterface $builder, array $options)
{
// ...
$builder->add('foo', new LocationType(), array(
'data_class' => 'Acme\HelloBundle\Entity\Company'
));
}
$builder->add('bar', new LocationType(), array(
'data_class' => 'Acme\HelloBundle\Entity\Customer'
));
}
That's it! You have extracted duplicated field definitions to a separate location form that you can reuse
wherever you need it.
PDF brought to you by
generated on December 3, 2014
Chapter 62: How to Reduce Code Duplication with "inherit_data" | 255
Forms with the inherit_data option set cannot have *_SET_DATA event listeners.
PDF brought to you by
generated on December 3, 2014
Chapter 62: How to Reduce Code Duplication with "inherit_data" | 256
Chapter 63
How to Unit Test your Forms
The Form component consists of 3 core objects: a form type (implementing FormTypeInterface1), the
Form2 and the FormView3.
The only class that is usually manipulated by programmers is the form type class which serves as a form
blueprint. It is used to generate the Form and the FormView. You could test it directly by mocking its
interactions with the factory but it would be complex. It is better to pass it to FormFactory like it is done
in a real application. It is simple to bootstrap and you can trust the Symfony components enough to use
them as a testing base.
There is already a class that you can benefit from for simple FormTypes testing: TypeTestCase4. It is used
to test the core types and you can use it to test your types too.
New in version 2.3: The TypeTestCase has moved to the Symfony\Component\Form\Test
namespace
in
2.3.
Previously,
the
class
was
located
in
Symfony\Component\Form\Tests\Extension\Core\Type.
Depending on the way you installed your Symfony or Symfony Form component the tests may not
be downloaded. Use the --prefer-source option with Composer if this is the case.
The Basics
The simplest TypeTestCase implementation looks like the following:
Listing 63-1
1 // src/Acme/TestBundle/Tests/Form/Type/TestedTypeTest.php
2 namespace Acme\TestBundle\Tests\Form\Type;
1. http://api.symfony.com/master/Symfony/Component/Form/FormTypeInterface.html
2. http://api.symfony.com/master/Symfony/Component/Form/Form.html
3. http://api.symfony.com/master/Symfony/Component/Form/FormView.html
4. http://api.symfony.com/master/Symfony/Component/Form/Test/TypeTestCase.html
PDF brought to you by
generated on December 3, 2014
Chapter 63: How to Unit Test your Forms | 257
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
use Acme\TestBundle\Form\Type\TestedType;
use Acme\TestBundle\Model\TestObject;
use Symfony\Component\Form\Test\TypeTestCase;
class TestedTypeTest extends TypeTestCase
{
public function testSubmitValidData()
{
$formData = array(
'test' => 'test',
'test2' => 'test2',
);
$type = new TestedType();
$form = $this->factory->create($type);
$object = new TestObject();
$object->fromArray($formData);
// submit the data to the form directly
$form->submit($formData);
$this->assertTrue($form->isSynchronized());
$this->assertEquals($object, $form->getData());
$view = $form->createView();
$children = $view->children;
foreach (array_keys($formData) as $key) {
$this->assertArrayHasKey($key, $children);
}
}
}
So, what does it test? Here comes a detailed explanation.
First you verify if the FormType compiles. This includes basic class inheritance, the buildForm function
and options resolution. This should be the first test you write:
Listing 63-2
1 $type = new TestedType();
2 $form = $this->factory->create($type);
This test checks that none of your data transformers used by the form failed. The isSynchronized()5
method is only set to false if a data transformer throws an exception:
Listing 63-3
1 $form->submit($formData);
2 $this->assertTrue($form->isSynchronized());
Don't test the validation: it is applied by a listener that is not active in the test case and it relies on
validation configuration. Instead, unit test your custom constraints directly.
5. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#isSynchronized()
PDF brought to you by
generated on December 3, 2014
Chapter 63: How to Unit Test your Forms | 258
Next, verify the submission and mapping of the form. The test below checks if all the fields are correctly
specified:
Listing 63-4
1 $this->assertEquals($object, $form->getData());
Finally, check the creation of the FormView. You should check if all widgets you want to display are
available in the children property:
Listing 63-5
1
2
3
4
5
6
$view = $form->createView();
$children = $view->children;
foreach (array_keys($formData) as $key) {
$this->assertArrayHasKey($key, $children);
}
Adding a Type your Form Depends on
Your form may depend on other types that are defined as services. It might look like this:
Listing 63-6
1 // src/Acme/TestBundle/Form/Type/TestedType.php
2
3 // ... the buildForm method
4 $builder->add('acme_test_child_type');
To create your form correctly, you need to make the type available to the form factory in your test. The
easiest way is to register it manually before creating the parent form using the PreloadedExtension class:
Listing 63-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/Acme/TestBundle/Tests/Form/Type/TestedTypeTests.php
namespace Acme\TestBundle\Tests\Form\Type;
use
use
use
use
Acme\TestBundle\Form\Type\TestedType;
Acme\TestBundle\Model\TestObject;
Symfony\Component\Form\Test\TypeTestCase;
Symfony\Component\Form\PreloadedExtension;
class TestedTypeTest extends TypeTestCase
{
protected function getExtensions()
{
$childType = new TestChildType();
return array(new PreloadedExtension(array(
$childType->getName() => $childType,
), array()));
}
public function testSubmitValidData()
{
$type = new TestedType();
$form = $this->factory->create($type);
// ... your test
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 63: How to Unit Test your Forms | 259
Make sure the child type you add is well tested. Otherwise you may be getting errors that are not
related to the form you are currently testing but to its children.
Adding custom Extensions
It often happens that you use some options that are added by form extensions. One of the cases may be
the ValidatorExtension with its invalid_message option. The TypeTestCase loads only the core form
extension so an "Invalid option" exception will be raised if you try to use it for testing a class that depends
on other extensions. You need add those extensions to the factory object:
Listing 63-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
// src/Acme/TestBundle/Tests/Form/Type/TestedTypeTests.php
namespace Acme\TestBundle\Tests\Form\Type;
use
use
use
use
use
use
use
Acme\TestBundle\Form\Type\TestedType;
Acme\TestBundle\Model\TestObject;
Symfony\Component\Form\Test\TypeTestCase;
Symfony\Component\Form\Forms;
Symfony\Component\Form\FormBuilder;
Symfony\Component\Form\Extension\Validator\Type\FormTypeValidatorExtension;
Symfony\Component\Validator\ConstraintViolationList;
class TestedTypeTest extends TypeTestCase
{
protected function setUp()
{
parent::setUp();
$validator =
$this->getMock('\Symfony\Component\Validator\Validator\ValidatorInterface');
$validator->method('validate')->will($this->returnValue(new
ConstraintViolationList()));
$this->factory = Forms::createFormFactoryBuilder()
->addExtensions($this->getExtensions())
->addTypeExtension(
new FormTypeValidatorExtension(
$validator
)
)
->addTypeGuesser(
$this->getMockBuilder(
'Symfony\Component\Form\Extension\Validator\ValidatorTypeGuesser'
)
->disableOriginalConstructor()
->getMock()
)
->getFormFactory();
$this->dispatcher =
$this->getMock('Symfony\Component\EventDispatcher\EventDispatcherInterface');
$this->builder = new FormBuilder(null, null, $this->dispatcher, $this->factory);
}
// ... your tests
}
PDF brought to you by
generated on December 3, 2014
Chapter 63: How to Unit Test your Forms | 260
Testing against different Sets of Data
If you are not familiar yet with PHPUnit's data providers6, this might be a good opportunity to use them:
Listing 63-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
// src/Acme/TestBundle/Tests/Form/Type/TestedTypeTests.php
namespace Acme\TestBundle\Tests\Form\Type;
use Acme\TestBundle\Form\Type\TestedType;
use Acme\TestBundle\Model\TestObject;
use Symfony\Component\Form\Test\TypeTestCase;
class TestedTypeTest extends TypeTestCase
{
/**
* @dataProvider getValidTestData
*/
public function testForm($data)
{
// ... your test
}
public function getValidTestData()
{
return array(
array(
'data' => array(
'test' => 'test',
'test2' => 'test2',
),
),
array(
'data' => array(),
),
array(
'data' => array(
'test' => null,
'test2' => null,
),
),
);
}
}
The code above will run your test three times with 3 different sets of data. This allows for decoupling the
test fixtures from the tests and easily testing against multiple sets of data.
You can also pass another argument, such as a boolean if the form has to be synchronized with the given
set of data or not etc.
6. http://www.phpunit.de/manual/current/en/writing-tests-for-phpunit.html#writing-tests-for-phpunit.data-providers
PDF brought to you by
generated on December 3, 2014
Chapter 63: How to Unit Test your Forms | 261
Chapter 64
How to Configure empty Data for a Form Class
The empty_data option allows you to specify an empty data set for your form class. This empty data set
would be used if you submit your form, but haven't called setData() on your form or passed in data
when you created your form. For example:
Listing 64-1
1 public function indexAction()
2 {
3
$blog = ...;
4
5
// $blog is passed in as the data, so the empty_data option is not needed
6
$form = $this->createForm(new BlogType(), $blog);
7
8
// no data is passed in, so empty_data is used to get the "starting data"
9
$form = $this->createForm(new BlogType());
10 }
By default, empty_data is set to null. Or, if you have specified a data_class option for your form class,
it will default to a new instance of that class. That instance will be created by calling the constructor with
no arguments.
If you want to override this default behavior, there are two ways to do this.
Option 1: Instantiate a new Class
One reason you might use this option is if you want to use a constructor that takes arguments.
Remember, the default data_class option calls that constructor with no arguments:
Listing 64-2
1
2
3
4
5
6
7
// src/Acme/DemoBundle/Form/Type/BlogType.php
// ...
use Symfony\Component\Form\AbstractType;
use Acme\DemoBundle\Entity\Blog;
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
PDF brought to you by
generated on December 3, 2014
Chapter 64: How to Configure empty Data for a Form Class | 262
8 class BlogType extends AbstractType
9 {
10
private $someDependency;
11
12
public function __construct($someDependency)
13
{
14
$this->someDependency = $someDependency;
15
}
16
// ...
17
18
public function setDefaultOptions(OptionsResolverInterface $resolver)
19
{
20
$resolver->setDefaults(array(
21
'empty_data' => new Blog($this->someDependency),
22
));
23
}
24 }
You can instantiate your class however you want. In this example, we pass some dependency into the
BlogType when we instantiate it, then use that to instantiate the Blog class. The point is, you can set
empty_data to the exact "new" object that you want to use.
Option 2: Provide a Closure
Using a closure is the preferred method, since it will only create the object if it is needed.
The closure must accept a FormInterface instance as the first argument:
Listing 64-3
1
2
3
4
5
6
7
8
9
10
11
12
use Symfony\Component\OptionsResolver\OptionsResolverInterface;
use Symfony\Component\Form\FormInterface;
// ...
public function setDefaultOptions(OptionsResolverInterface $resolver)
{
$resolver->setDefaults(array(
'empty_data' => function (FormInterface $form) {
return new Blog($form->get('title')->getData());
},
));
}
PDF brought to you by
generated on December 3, 2014
Chapter 64: How to Configure empty Data for a Form Class | 263
Chapter 65
How to Use the submit() Function to Handle
Form Submissions
New in version 2.3: The handleRequest()1 method was introduced in Symfony 2.3.
With the handleRequest() method, it is really easy to handle form submissions:
Listing 65-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
use Symfony\Component\HttpFoundation\Request;
// ...
public function newAction(Request $request)
{
$form = $this->createFormBuilder()
// ...
->getForm();
$form->handleRequest($request);
if ($form->isValid()) {
// perform some action...
return $this->redirect($this->generateUrl('task_success'));
}
return $this->render('AcmeTaskBundle:Default:new.html.twig', array(
'form' => $form->createView(),
));
}
1. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#handleRequest()
PDF brought to you by
generated on December 3, 2014
Chapter 65: How to Use the submit() Function to Handle Form Submissions | 264
To see more about this method, read Handling Form Submissions.
Calling Form::submit() manually
New in version 2.3: Before Symfony 2.3, the submit() method was known as bind().
In some cases, you want better control over when exactly your form is submitted and what data is passed
to it. Instead of using the handleRequest()2 method, pass the submitted data directly to submit()3:
Listing 65-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
use Symfony\Component\HttpFoundation\Request;
// ...
public function newAction(Request $request)
{
$form = $this->createFormBuilder()
// ...
->getForm();
if ($request->isMethod('POST')) {
$form->submit($request->request->get($form->getName()));
if ($form->isValid()) {
// perform some action...
return $this->redirect($this->generateUrl('task_success'));
}
}
return $this->render('AcmeTaskBundle:Default:new.html.twig', array(
'form' => $form->createView(),
));
}
Forms consisting of nested fields expect an array in submit()4. You can also submit individual
fields by calling submit()5 directly on the field:
Listing 65-3
1 $form->get('firstName')->submit('Fabien');
2. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#handleRequest()
3. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#submit()
4. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#submit()
5. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#submit()
PDF brought to you by
generated on December 3, 2014
Chapter 65: How to Use the submit() Function to Handle Form Submissions | 265
Passing a Request to Form::submit() (Deprecated)
New in version 2.3: Before Symfony 2.3, the submit method was known as bind.
Before Symfony 2.3, the submit()6 method accepted a Request7 object as a convenient shortcut to the
previous example:
Listing 65-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
use Symfony\Component\HttpFoundation\Request;
// ...
public function newAction(Request $request)
{
$form = $this->createFormBuilder()
// ...
->getForm();
if ($request->isMethod('POST')) {
$form->submit($request);
if ($form->isValid()) {
// perform some action...
return $this->redirect($this->generateUrl('task_success'));
}
}
return $this->render('AcmeTaskBundle:Default:new.html.twig', array(
'form' => $form->createView(),
));
}
Passing the Request8 directly to submit()9 still works, but is deprecated and will be removed in Symfony
3.0. You should use the method handleRequest()10 instead.
6. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#submit()
7. http://api.symfony.com/master/Symfony/Component/HttpFoundation/Request.html
8. http://api.symfony.com/master/Symfony/Component/HttpFoundation/Request.html
9. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#submit()
10. http://api.symfony.com/master/Symfony/Component/Form/FormInterface.html#handleRequest()
PDF brought to you by
generated on December 3, 2014
Chapter 65: How to Use the submit() Function to Handle Form Submissions | 266
Chapter 66
How to Use the virtual Form Field Option
As of Symfony 2.3, the virtual option is renamed to inherit_data. You can read everything about the
new option in "How to Reduce Code Duplication with "inherit_data"".
PDF brought to you by
generated on December 3, 2014
Chapter 66: How to Use the virtual Form Field Option | 267
Chapter 67
How to Use Monolog to Write Logs
Monolog1 is a logging library for PHP 5.3 used by Symfony. It is inspired by the Python LogBook library.
Usage
To log a message simply get the logger service from the container in your controller:
Listing 67-1
1 public function indexAction()
2 {
3
$logger = $this->get('logger');
4
$logger->info('I just got the logger');
5
$logger->error('An error occurred');
6
7
// ...
8 }
The logger service has different methods for different logging levels. See LoggerInterface2 for details on
which methods are available.
Handlers and Channels: Writing Logs to different Locations
In Monolog each logger defines a logging channel, which organizes your log messages into different
"categories". Then, each channel has a stack of handlers to write the logs (the handlers can be shared).
When injecting the logger in a service you can use a custom channel control which "channel" the
logger will log to.
1. https://github.com/Seldaek/monolog
2. https://github.com/php-fig/log/blob/master/Psr/Log/LoggerInterface.php
PDF brought to you by
generated on December 3, 2014
Chapter 67: How to Use Monolog to Write Logs | 268
The basic handler is the StreamHandler which writes logs in a stream (by default in the app/logs/
prod.log in the prod environment and app/logs/dev.log in the dev environment).
Monolog comes also with a powerful built-in handler for the logging in prod environment:
FingersCrossedHandler. It allows you to store the messages in a buffer and to log them only if a message
reaches the action level (error in the configuration provided in the Standard Edition) by forwarding the
messages to another handler.
Using several Handlers
The logger uses a stack of handlers which are called successively. This allows you to log the messages in
several ways easily.
Listing 67-2
1 # app/config/config.yml
2 monolog:
3
handlers:
4
applog:
5
type: stream
6
path: /var/log/symfony.log
7
level: error
8
main:
9
type: fingers_crossed
10
action_level: warning
11
handler: file
12
file:
13
type: stream
14
level: debug
15
syslog:
16
type: syslog
17
level: error
The above configuration defines a stack of handlers which will be called in the order where they are
defined.
The handler named "file" will not be included in the stack itself as it is used as a nested handler of
the fingers_crossed handler.
If you want to change the config of MonologBundle in another config file you need to redefine the
whole stack. It cannot be merged because the order matters and a merge does not allow to control
the order.
Changing the Formatter
The handler uses a Formatter to format the record before logging it. All Monolog handlers use an
instance of Monolog\Formatter\LineFormatter by default but you can replace it easily. Your formatter
must implement Monolog\Formatter\FormatterInterface.
Listing 67-3
1 # app/config/config.yml
2 services:
3
my_formatter:
4
class: Monolog\Formatter\JsonFormatter
5 monolog:
6
handlers:
PDF brought to you by
generated on December 3, 2014
Chapter 67: How to Use Monolog to Write Logs | 269
7
8
9
10
file:
type: stream
level: debug
formatter: my_formatter
Adding some extra Data in the Log Messages
Monolog allows to process the record before logging it to add some extra data. A processor can be applied
for the whole handler stack or only for a specific handler.
A processor is simply a callable receiving the record as its first argument.
Processors are configured using the monolog.processor DIC tag. See the reference about it.
Adding a Session/Request Token
Sometimes it is hard to tell which entries in the log belong to which session and/or request. The following
example will add a unique token for each request using a processor.
Listing 67-4
Listing 67-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
namespace Acme\MyBundle;
use Symfony\Component\HttpFoundation\Session\Session;
class SessionRequestProcessor
{
private $session;
private $token;
public function __construct(Session $session)
{
$this->session = $session;
}
public function processRecord(array $record)
{
if (null === $this->token) {
try {
$this->token = substr($this->session->getId(), 0, 8);
} catch (\RuntimeException $e) {
$this->token = '????????';
}
$this->token .= '-' . substr(uniqid(), -8);
}
$record['extra']['token'] = $this->token;
return $record;
}
}
1 # app/config/config.yml
2 services:
3
monolog.formatter.session_request:
4
class: Monolog\Formatter\LineFormatter
5
arguments:
6
- "[%%datetime%%] [%%extra.token%%] %%channel%%.%%level_name%%: %%message%%\n"
PDF brought to you by
generated on December 3, 2014
Chapter 67: How to Use Monolog to Write Logs | 270
7
8
monolog.processor.session_request:
9
class: Acme\MyBundle\SessionRequestProcessor
10
arguments: ["@session"]
11
tags:
12
- { name: monolog.processor, method: processRecord }
13
14 monolog:
15
handlers:
16
main:
17
type: stream
18
path: "%kernel.logs_dir%/%kernel.environment%.log"
19
level: debug
20
formatter: monolog.formatter.session_request
If you use several handlers, you can also register a processor at the handler level or at the channel
level instead of registering it globally (see the following sections).
Registering Processors per Handler
You can register a processor per handler using the handler option of the monolog.processor tag:
Listing 67-6
1 # app/config/config.yml
2 services:
3
monolog.processor.session_request:
4
class: Acme\MyBundle\SessionRequestProcessor
5
arguments: ["@session"]
6
tags:
7
- { name: monolog.processor, method: processRecord, handler: main }
Registering Processors per Channel
You can register a processor per channel using the channel option of the monolog.processor tag:
Listing 67-7
1 # app/config/config.yml
2 services:
3
monolog.processor.session_request:
4
class: Acme\MyBundle\SessionRequestProcessor
5
arguments: ["@session"]
6
tags:
7
- { name: monolog.processor, method: processRecord, channel: main }
PDF brought to you by
generated on December 3, 2014
Chapter 67: How to Use Monolog to Write Logs | 271
Chapter 68
How to Configure Monolog to Email Errors
Monolog1 can be configured to send an email when an error occurs with an application. The configuration
for this requires a few nested handlers in order to avoid receiving too many emails. This configuration
looks complicated at first but each handler is fairly straight forward when it is broken down.
Listing 68-1
1 # app/config/config_prod.yml
2 monolog:
3
handlers:
4
mail:
5
type:
fingers_crossed
6
# 500 errors are logged at the critical level
7
action_level: critical
8
# to also log 400 level errors (but not 404's):
9
# action_level: error
10
# excluded_404:
11
#
- ^/
12
handler:
buffered
13
buffered:
14
type:
buffer
15
handler: swift
16
swift:
17
type:
swift_mailer
18
from_email: [email protected]
19
to_email:
[email protected]
20
# or list of recipients
21
# to_email: [[email protected], [email protected], ...]
22
subject:
An Error Occurred!
23
level:
debug
The mail handler is a fingers_crossed handler which means that it is only triggered when the action
level, in this case critical is reached. It then logs everything including messages below the action level.
The critical level is only triggered for 5xx HTTP code errors. The handler setting means that the
output is then passed onto the buffered handler.
1. https://github.com/Seldaek/monolog
PDF brought to you by
generated on December 3, 2014
Chapter 68: How to Configure Monolog to Email Errors | 272
If you want both 400 level and 500 level errors to trigger an email, set the action_level to error
instead of critical. See the code above for an example.
The buffered handler simply keeps all the messages for a request and then passes them onto the nested
handler in one go. If you do not use this handler then each message will be emailed separately. This is
then passed to the swift handler. This is the handler that actually deals with emailing you the error. The
settings for this are straightforward, the to and from addresses and the subject.
You can combine these handlers with other handlers so that the errors still get logged on the server as
well as the emails being sent:
Listing 68-2
1 # app/config/config_prod.yml
2 monolog:
3
handlers:
4
main:
5
type:
fingers_crossed
6
action_level: critical
7
handler:
grouped
8
grouped:
9
type:
group
10
members: [streamed, buffered]
11
streamed:
12
type: stream
13
path: "%kernel.logs_dir%/%kernel.environment%.log"
14
level: debug
15
buffered:
16
type:
buffer
17
handler: swift
18
swift:
19
type:
swift_mailer
20
from_email: [email protected]
21
to_email:
[email protected]
22
subject:
An Error Occurred!
23
level:
debug
This uses the group handler to send the messages to the two group members, the buffered and the
stream handlers. The messages will now be both written to the log file and emailed.
PDF brought to you by
generated on December 3, 2014
Chapter 68: How to Configure Monolog to Email Errors | 273
Chapter 69
How to Configure Monolog to Display Console
Messages
New in version 2.4: This feature was introduced to the MonologBridge in Symfony 2.4.
It is possible to use the console to print messages for certain verbosity levels using the OutputInterface1
instance that is passed when a command gets executed.
Alternatively, you can use the standalone PSR-3 logger provided with the console component.
When a lot of logging has to happen, it's cumbersome to print information depending on the verbosity
settings (-v, -vv, -vvv) because the calls need to be wrapped in conditions. The code quickly gets verbose
or dirty. For example:
Listing 69-1
1
2
3
4
5
6
7
8
9
10
11
12
13
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Output\OutputInterface;
protected function execute(InputInterface $input, OutputInterface $output)
{
if ($output->getVerbosity() >= OutputInterface::VERBOSITY_DEBUG) {
$output->writeln('Some info');
}
if ($output->getVerbosity() >= OutputInterface::VERBOSITY_VERBOSE) {
$output->writeln('Some more info');
}
}
1. http://api.symfony.com/master/Symfony/Component/Console/Output/OutputInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 69: How to Configure Monolog to Display Console Messages | 274
Instead of using these semantic methods to test for each of the verbosity levels, the MonologBridge2
provides a ConsoleHandler3 that listens to console events and writes log messages to the console output
depending on the current log level and the console verbosity.
The example above could then be rewritten as:
Listing 69-2
1
2
3
4
5
6
7
8
9
10
11
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Output\OutputInterface;
protected function execute(InputInterface $input, OutputInterface $output)
{
// assuming the Command extends ContainerAwareCommand...
$logger = $this->getContainer()->get('logger');
$logger->debug('Some info');
$logger->notice('Some more info');
}
Depending on the verbosity level that the command is run in and the user's configuration (see below),
these messages may or may not be displayed to the console. If they are displayed, they are timestamped
and colored appropriately. Additionally, error logs are written to the error output (php://stderr). There
is no need to conditionally handle the verbosity settings anymore.
The Monolog console handler is enabled in the Monolog configuration. This is the default in Symfony
Standard Edition 2.4 too.
Listing 69-3
1 # app/config/config.yml
2 monolog:
3
handlers:
4
console:
5
type: console
With the verbosity_levels option you can adapt the mapping between verbosity and log level. In
the given example it will also show notices in normal verbosity mode (instead of warnings only).
Additionally, it will only use messages logged with the custom my_channel channel and it changes the
display style via a custom formatter (see the MonologBundle reference for more information):
Listing 69-4
1 # app/config/config.yml
2 monolog:
3
handlers:
4
console:
5
type:
console
6
verbosity_levels:
7
VERBOSITY_NORMAL: NOTICE
8
channels: my_channel
9
formatter: my_formatter
Listing 69-5
1 # app/config/services.yml
2 services:
3
my_formatter:
4
class: Symfony\Bridge\Monolog\Formatter\ConsoleFormatter
5
arguments:
6
- "[%%datetime%%] %%start_tag%%%%message%%%%end_tag%% (%%level_name%%)
%%context%% %%extra%%\n"
2. https://github.com/symfony/MonologBridge
3. https://github.com/symfony/MonologBridge/blob/master/Handler/ConsoleHandler.php
PDF brought to you by
generated on December 3, 2014
Chapter 69: How to Configure Monolog to Display Console Messages | 275
Chapter 70
How to Configure Monolog to Exclude 404
Errors from the Log
New in version 2.4: This feature was introduced to the MonologBundle 2.4, which was first
packaged with Symfony 2.4.
Sometimes your logs become flooded with unwanted 404 HTTP errors, for example, when an attacker
scans your app for some well-known application paths (e.g. /phpmyadmin). When using a
fingers_crossed handler, you can exclude logging these 404 errors based on a regular expression in the
MonologBundle configuration:
Listing 70-1
1 # app/config/config.yml
2 monolog:
3
handlers:
4
main:
5
# ...
6
type: fingers_crossed
7
handler: ...
8
excluded_404s:
9
- ^/phpmyadmin
PDF brought to you by
generated on December 3, 2014
Chapter 70: How to Configure Monolog to Exclude 404 Errors from the Log | 276
Chapter 71
How to Log Messages to different Files
The Symfony Standard Edition contains a bunch of channels for logging: doctrine, event, security
and request. Each channel corresponds to a logger service (monolog.logger.XXX) in the container and
is injected to the concerned service. The purpose of channels is to be able to organize different types of
log messages.
By default, Symfony logs every message into a single file (regardless of the channel).
Switching a Channel to a different Handler
Now, suppose you want to log the doctrine channel to a different file.
To do so, just create a new handler and configure it like this:
Listing 71-1
1 # app/config/config.yml
2 monolog:
3
handlers:
4
main:
5
type:
stream
6
path:
/var/log/symfony.log
7
channels: ["!doctrine"]
8
doctrine:
9
type:
stream
10
path:
/var/log/doctrine.log
11
channels: [doctrine]
YAML Specification
You can specify the configuration by many forms:
Listing 71-2
1 channels: ~
2
3 channels: foo
# Include all the channels
# Include only channel "foo"
PDF brought to you by
generated on December 3, 2014
Chapter 71: How to Log Messages to different Files | 277
4 channels: "!foo" # Include all channels, except "foo"
5
6 channels: [foo, bar]
# Include only channels "foo" and "bar"
7 channels: ["!foo", "!bar"] # Include all channels, except "foo" and "bar"
Creating your own Channel
You can change the channel monolog logs to one service at a time. This is done either via the
configuration below or by tagging your service with monolog.logger and specifying which channel the
service should log to. With the tag, the logger that is injected into that service is preconfigured to use the
channel you've specified.
Configure Additional Channels without Tagged Services
New in version 2.4: This feature was introduced to the MonologBundle 2.4, which was first
packaged with Symfony 2.4.
With MonologBundle 2.4 you can configure additional channels without the need to tag your services:
Listing 71-3
1 # app/config/config.yml
2 monolog:
3
channels: ["foo", "bar"]
With this, you can now send log messages to the foo channel by using the automatically registered logger
service monolog.logger.foo.
Learn more from the Cookbook
• How to Use Monolog to Write Logs
PDF brought to you by
generated on December 3, 2014
Chapter 71: How to Log Messages to different Files | 278
Chapter 72
How to Create a custom Data Collector
The Symfony Profiler delegates data collecting to data collectors. Symfony comes bundled with a few of
them, but you can easily create your own.
Creating a custom Data Collector
Creating a custom data collector is as simple as implementing the DataCollectorInterface1:
Listing 72-1
1 interface DataCollectorInterface
2 {
3
/**
4
* Collects data for the given Request and Response.
5
*
6
* @param Request
$request A Request instance
7
* @param Response $response A Response instance
8
* @param \Exception $exception An Exception instance
9
*/
10
function collect(Request $request, Response $response, \Exception $exception = null);
11
12
/**
13
* Returns the name of the collector.
14
*
15
* @return string The collector name
16
*/
17
function getName();
18 }
The getName() method must return a unique name. This is used to access the information later on (see
How to Use the Profiler in a Functional Test for instance).
The collect() method is responsible for storing the data it wants to give access to in local properties.
1. http://api.symfony.com/master/Symfony/Component/HttpKernel/DataCollector/DataCollectorInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 72: How to Create a custom Data Collector | 279
As the profiler serializes data collector instances, you should not store objects that cannot be
serialized (like PDO objects), or you need to provide your own serialize() method.
Most of the time, it is convenient to extend DataCollector2 and populate the $this->data property (it
takes care of serializing the $this->data property):
Listing 72-2
1 class MemoryDataCollector extends DataCollector
2 {
3
public function collect(Request $request, Response $response, \Exception $exception =
4 null)
5
{
6
$this->data = array(
7
'memory' => memory_get_peak_usage(true),
8
);
9
}
10
11
public function getMemory()
12
{
13
return $this->data['memory'];
14
}
15
16
public function getName()
17
{
18
return 'memory';
19
}
}
Enabling custom Data Collectors
To enable a data collector, add it as a regular service in one of your configuration, and tag it with
data_collector:
Listing 72-3
1 services:
2
data_collector.your_collector_name:
3
class: Fully\Qualified\Collector\Class\Name
4
tags:
5
- { name: data_collector }
Adding Web Profiler Templates
When you want to display the data collected by your data collector in the web debug toolbar or the web
profiler, create a Twig template following this skeleton:
Listing 72-4
1 {% extends 'WebProfilerBundle:Profiler:layout.html.twig' %}
2
3 {% block toolbar %}
4
{# the web debug toolbar content #}
5 {% endblock %}
2. http://api.symfony.com/master/Symfony/Component/HttpKernel/DataCollector/DataCollector.html
PDF brought to you by
generated on December 3, 2014
Chapter 72: How to Create a custom Data Collector | 280
6
7
8
9
10
11
12
13
14
15
16
17
{% block head %}
{# if the web profiler panel needs some specific JS or CSS files #}
{% endblock %}
{% block menu %}
{# the menu content #}
{% endblock %}
{% block panel %}
{# the panel content #}
{% endblock %}
Each block is optional. The toolbar block is used for the web debug toolbar and menu and panel are
used to add a panel to the web profiler.
All blocks have access to the collector object.
Built-in templates use a base64 encoded image for the toolbar:
Listing 72-5
1 <img src="data:image/png;base64,..." />
You can easily calculate the base64 value for an image with this little script:
Listing 72-6
1 #!/usr/bin/env php
2 <?php
3 echo base64_encode(file_get_contents($_SERVER['argv'][1]));
To enable the template, add a template attribute to the data_collector tag in your configuration. For
example, assuming your template is in some AcmeDebugBundle:
Listing 72-7
1 services:
2
data_collector.your_collector_name:
3
class: Acme\DebugBundle\Collector\Class\Name
4
tags:
5
- { name: data_collector, template: "AcmeDebugBundle:Collector:templatename",
id: "your_collector_name" }
PDF brought to you by
generated on December 3, 2014
Chapter 72: How to Create a custom Data Collector | 281
Chapter 73
How to Use Matchers to Enable the Profiler
Conditionally
By default, the profiler is only activated in the development environment. But it's imaginable that a
developer may want to see the profiler even in production. Another situation may be that you want to
show the profiler only when an admin has logged in. You can enable the profiler in these situations by
using matchers.
Using the built-in Matcher
Symfony provides a built-in matcher1 which can match paths and IPs. For example, if you want to only
show the profiler when accessing the page with the 168.0.0.1 IP, then you can use this configuration:
Listing 73-1
1 # app/config/config.yml
2 framework:
3
# ...
4
profiler:
5
matcher:
6
ip: 168.0.0.1
You can also set a path option to define the path on which the profiler should be enabled. For instance,
setting it to ^/admin/ will enable the profiler only for the /admin/ URLs.
Creating a custom Matcher
You can also create a custom matcher. This is a service that checks whether the profiler should be
enabled or not. To create that service, create a class which implements RequestMatcherInterface2. This
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/RequestMatcher.html
2. http://api.symfony.com/master/Symfony/Component/HttpFoundation/RequestMatcherInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 73: How to Use Matchers to Enable the Profiler Conditionally | 282
interface requires one method: matches()3. This method returns false to disable the profiler and true to
enable the profiler.
To enable the profiler when a ROLE_SUPER_ADMIN is logged in, you can use something like:
Listing 73-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/DemoBundle/Profiler/SuperAdminMatcher.php
namespace Acme\DemoBundle\Profiler;
use Symfony\Component\Security\Core\SecurityContext;
use Symfony\Component\HttpFoundation\Request;
use Symfony\Component\HttpFoundation\RequestMatcherInterface;
class SuperAdminMatcher implements RequestMatcherInterface
{
protected $securityContext;
public function __construct(SecurityContext $securityContext)
{
$this->securityContext = $securityContext;
}
public function matches(Request $request)
{
return $this->securityContext->isGranted('ROLE_SUPER_ADMIN');
}
}
Then, you need to configure the service:
Listing 73-3
1 parameters:
2
acme_demo.profiler.matcher.super_admin.class: Acme\DemoBundle\Profiler\SuperAdminMatcher
3
4 services:
5
acme_demo.profiler.matcher.super_admin:
6
class: "%acme_demo.profiler.matcher.super_admin.class%"
7
arguments: ["@security.context"]
Now the service is registered, the only thing left to do is configure the profiler to use this service as the
matcher:
Listing 73-4
1 # app/config/config.yml
2 framework:
3
# ...
4
profiler:
5
matcher:
6
service: acme_demo.profiler.matcher.super_admin
3. http://api.symfony.com/master/Symfony/Component/HttpFoundation/RequestMatcherInterface.html#matches()
PDF brought to you by
generated on December 3, 2014
Chapter 73: How to Use Matchers to Enable the Profiler Conditionally | 283
Chapter 74
Switching the Profiler Storage
By default the profile stores the collected data in files in the cache directory. You can control the storage
being used through the dsn, username, password and lifetime options. For example, the following
configuration uses MySQL as the storage for the profiler with a lifetime of one hour:
Listing 74-1
1 # app/config/config.yml
2 framework:
3 profiler:
4
dsn:
"mysql:host=localhost;dbname=%database_name%"
5
username: "%database_user%"
6
password: "%database_password%"
7
lifetime: 3600
The HttpKernel component currently supports the following profiler storage implementations:
•
•
•
•
•
•
•
FileProfilerStorage1
MemcachedProfilerStorage2
MemcacheProfilerStorage3
MongoDbProfilerStorage4
MysqlProfilerStorage5
RedisProfilerStorage6
SqliteProfilerStorage7
1. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/FileProfilerStorage.html
2. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/MemcachedProfilerStorage.html
3. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/MemcacheProfilerStorage.html
4. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/MongoDbProfilerStorage.html
5. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/MysqlProfilerStorage.html
6. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/RedisProfilerStorage.html
7. http://api.symfony.com/master/Symfony/Component/HttpKernel/Profiler/SqliteProfilerStorage.html
PDF brought to you by
generated on December 3, 2014
Chapter 74: Switching the Profiler Storage | 284
Chapter 75
How to Configure Symfony to Work behind a
Load Balancer or a Reverse Proxy
When you deploy your application, you may be behind a load balancer (e.g. an AWS Elastic Load
Balancer) or a reverse proxy (e.g. Varnish for caching).
For the most part, this doesn't cause any problems with Symfony. But, when a request passes through a
proxy, certain request information is sent using special X-Forwarded-* headers. For example, instead of
reading the REMOTE_ADDR header (which will now be the IP address of your reverse proxy), the user's true
IP will be stored in an X-Forwarded-For header.
If you don't configure Symfony to look for these headers, you'll get incorrect information about the
client's IP address, whether or not the client is connecting via HTTPS, the client's port and the hostname
being requested.
Solution: trusted_proxies
This is no problem, but you do need to tell Symfony that this is happening and which reverse proxy IP
addresses will be doing this type of thing:
Listing 75-1
1 # app/config/config.yml
2 # ...
3 framework:
4
trusted_proxies: [192.0.0.1, 10.0.0.0/8]
In this example, you're saying that your reverse proxy (or proxies) has the IP address 192.0.0.1 or
matches the range of IP addresses that use the CIDR notation 10.0.0.0/8. For more details, see the
framework.trusted_proxies option.
That's it! Symfony will now look for the correct X-Forwarded-* headers to get information like the
client's IP address, host, port and whether or not the request is using HTTPS.
PDF brought to you by
generated on December 3, 2014
Chapter 75: How to Configure Symfony to Work behind a Load Balancer or a Reverse Proxy | 285
But what if the IP of my Reverse Proxy Changes Constantly!
Some reverse proxies (like Amazon's Elastic Load Balancers) don't have a static IP address or even a range
that you can target with the CIDR notation. In this case, you'll need to - very carefully - trust all proxies.
1. Configure your web server(s) to not respond to traffic from any clients other than your load
balancers. For AWS, this can be done with security groups1.
2. Once you've guaranteed that traffic will only come from your trusted reverse proxies, configure
Symfony to always trust incoming request. This is done inside of your front controller:
Listing 75-2
1
2
3
4
5
6
7
// web/app.php
// ...
Request::setTrustedProxies(array($request->server->get('REMOTE_ADDR')));
$response = $kernel->handle($request);
// ...
That's it! It's critical that you prevent traffic from all non-trusted sources. If you allow outside traffic, they
could "spoof" their true IP address and other information.
My Reverse Proxy Uses Non-Standard (not X-Forwarded) Headers
Most reverse proxies store information on specific X-Forwarded-* headers. But if your reverse proxy uses
non-standard header names, you can configure these (see "Trusting Proxies"). The code for doing this will
need to live in your front controller (e.g. web/app.php).
1. http://docs.aws.amazon.com/ElasticLoadBalancing/latest/DeveloperGuide/using-elb-security-groups.html
PDF brought to you by
generated on December 3, 2014
Chapter 75: How to Configure Symfony to Work behind a Load Balancer or a Reverse Proxy | 286
Chapter 76
How to Register a new Request Format and
Mime Type
Every Request has a "format" (e.g. html, json), which is used to determine what type of content to
return in the Response. In fact, the request format, accessible via getRequestFormat()1, is used to
set the MIME type of the Content-Type header on the Response object. Internally, Symfony contains
a map of the most common formats (e.g. html, json) and their associated MIME types (e.g. text/
html, application/json). Of course, additional format-MIME type entries can easily be added. This
document will show how you can add the jsonp format and corresponding MIME type.
New in version 2.5: The possibility to configure request formats was introduced in Symfony 2.5.
Configure your New Format
The FrameworkBundle registers a subscriber that will add formats to incoming requests.
All you have to do is to configure the jsonp format:
Listing 76-1
1 # app/config/config.yml
2 framework:
3
request:
4
formats:
5
jsonp: 'application/javascript'
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/Request.html#getRequestFormat()
PDF brought to you by
generated on December 3, 2014
Chapter 76: How to Register a new Request Format and Mime Type | 287
You can also associate multiple mime types to a format, but please note that the preferred one must
be the first as it will be used as the content type:
Listing 76-2
1 # app/config/config.yml
2 framework:
3
request:
4
formats:
5
csv: ['text/csv', 'text/plain']
PDF brought to you by
generated on December 3, 2014
Chapter 76: How to Register a new Request Format and Mime Type | 288
Chapter 77
How to Force Routes to always Use HTTPS or
HTTP
Sometimes, you want to secure some routes and be sure that they are always accessed via the HTTPS
protocol. The Routing component allows you to enforce the URI scheme via schemes:
Listing 77-1
1 secure:
2
path:
/secure
3
defaults: { _controller: AcmeDemoBundle:Main:secure }
4
schemes: [https]
The above configuration forces the secure route to always use HTTPS.
When generating the secure URL, and if the current scheme is HTTP, Symfony will automatically
generate an absolute URL with HTTPS as the scheme:
Listing 77-2
1
2
3
4
5
6
7
{# If the current scheme is HTTPS #}
{{ path('secure') }}
{# generates /secure #}
{# If the current scheme is HTTP #}
{{ path('secure') }}
{# generates https://example.com/secure #}
The requirement is also enforced for incoming requests. If you try to access the /secure path with HTTP,
you will automatically be redirected to the same URL, but with the HTTPS scheme.
The above example uses https for the scheme, but you can also force a URL to always use http.
The Security component provides another way to enforce HTTP or HTTPS via the
requires_channel setting. This alternative method is better suited to secure an "area" of your
website (all URLs under /admin) or when you want to secure URLs defined in a third party bundle
(see How to Force HTTPS or HTTP for different URLs for more details).
PDF brought to you by
generated on December 3, 2014
Chapter 77: How to Force Routes to always Use HTTPS or HTTP | 289
Chapter 78
How to Allow a "/" Character in a Route
Parameter
Sometimes, you need to compose URLs with parameters that can contain a slash /. For example, take
the classic /hello/{username} route. By default, /hello/Fabien will match this route but not /hello/
Fabien/Kris. This is because Symfony uses this character as separator between route parts.
This guide covers how you can modify a route so that /hello/Fabien/Kris matches the /hello/
{username} route, where {username} equals Fabien/Kris.
Configure the Route
By default, the Symfony Routing component requires that the parameters match the following regex
path: [^/]+. This means that all characters are allowed except /.
You must explicitly allow / to be part of your parameter by specifying a more permissive regex path.
Listing 78-1
1 _hello:
2
path:
/hello/{username}
3
defaults: { _controller: AcmeDemoBundle:Demo:hello }
4
requirements:
5
username: .+
That's it! Now, the {username} parameter can contain the / character.
PDF brought to you by
generated on December 3, 2014
Chapter 78: How to Allow a "/" Character in a Route Parameter | 290
Chapter 79
How to Configure a Redirect without a custom
Controller
Sometimes, a URL needs to redirect to another URL. You can do that by creating a new controller action
whose only task is to redirect, but using the RedirectController1 of the FrameworkBundle is even
easier.
You can redirect to a specific path (e.g. /about) or to a specific route using its name (e.g. homepage).
Redirecting Using a Path
Assume there is no default controller for the / path of your application and you want to redirect these
requests to /app. You will need to use the urlRedirect()2 action to redirect to this new url:
Listing 79-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
# app/config/routing.yml
# load some routes - one should ultimately have the path "/app"
AppBundle:
resource: "@AcmeAppBundle/Controller/"
type:
annotation
prefix:
/app
# redirecting the root
root:
path: /
defaults:
_controller: FrameworkBundle:Redirect:urlRedirect
path: /app
permanent: true
1. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Controller/RedirectController.html
2. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Controller/RedirectController.html#urlRedirect()
PDF brought to you by
generated on December 3, 2014
Chapter 79: How to Configure a Redirect without a custom Controller | 291
In this example, you configured a route for the / path and let the RedirectController redirect it to
/app. The permanent switch tells the action to issue a 301 HTTP status code instead of the default 302
HTTP status code.
Redirecting Using a Route
Assume you are migrating your website from WordPress to Symfony, you want to redirect /wp-admin
to the route sonata_admin_dashboard. You don't know the path, only the route name. This can be
achieved using the redirect()3 action:
Listing 79-2
1
2
3
4
5
6
7
8
9
10
11
# app/config/routing.yml
# ...
# redirecting the admin home
root:
path: /wp-admin
defaults:
_controller: FrameworkBundle:Redirect:redirect
route: sonata_admin_dashboard
permanent: true
Because you are redirecting to a route instead of a path, the required option is called route in the
redirect action, instead of path in the urlRedirect action.
3. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Controller/RedirectController.html#redirect()
PDF brought to you by
generated on December 3, 2014
Chapter 79: How to Configure a Redirect without a custom Controller | 292
Chapter 80
How to Use HTTP Methods beyond GET and
POST in Routes
The HTTP method of a request is one of the requirements that can be checked when seeing if it matches
a route. This is introduced in the routing chapter of the book "Routing" with examples using GET and
POST. You can also use other HTTP verbs in this way. For example, if you have a blog post entry then
you could use the same URL path to show it, make changes to it and delete it by matching on GET, PUT
and DELETE.
Listing 80-1
1 blog_show:
2
path:
3
defaults:
4
methods:
5
6 blog_update:
7
path:
8
defaults:
9
methods:
10
11 blog_delete:
12
path:
13
defaults:
14
methods:
/blog/{slug}
{ _controller: AcmeDemoBundle:Blog:show }
[GET]
/blog/{slug}
{ _controller: AcmeDemoBundle:Blog:update }
[PUT]
/blog/{slug}
{ _controller: AcmeDemoBundle:Blog:delete }
[DELETE]
Faking the Method with _method
The _method functionality shown here is disabled by default in Symfony 2.2 and enabled by
default in Symfony 2.3. To control it in Symfony 2.2, you must call
Request::enableHttpMethodParameterOverride1 before you handle the request (e.g. in your
front controller). In Symfony 2.3, use the http_method_override option.
PDF brought to you by
generated on December 3, 2014
Chapter 80: How to Use HTTP Methods beyond GET and POST in Routes | 293
Unfortunately, life isn't quite this simple, since most browsers do not support sending PUT and DELETE
requests. Fortunately, Symfony provides you with a simple way of working around this limitation. By
including a _method parameter in the query string or parameters of an HTTP request, Symfony will use
this as the method when matching routes. Forms automatically include a hidden field for this parameter
if their submission method is not GET or POST. See the related chapter in the forms documentation for
more information.
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/Request.html#enableHttpMethodParameterOverride()
PDF brought to you by
generated on December 3, 2014
Chapter 80: How to Use HTTP Methods beyond GET and POST in Routes | 294
Chapter 81
How to Use Service Container Parameters in
your Routes
Sometimes you may find it useful to make some parts of your routes globally configurable. For instance,
if you build an internationalized site, you'll probably start with one or two locales. Surely you'll add a
requirement to your routes to prevent a user from matching a locale other than the locales your support.
You could hardcode your _locale requirement in all your routes. But a better solution is to use a
configurable service container parameter right inside your routing configuration:
Listing 81-1
1 # app/config/routing.yml
2 contact:
3
path:
/{_locale}/contact
4
defaults: { _controller: AcmeDemoBundle:Main:contact }
5
requirements:
6
_locale: "%acme_demo.locales%"
You can now control and set the acme_demo.locales parameter somewhere in your container:
Listing 81-2
1 # app/config/config.yml
2 parameters:
3
acme_demo.locales: en|es
You can also use a parameter to define your route path (or part of your path):
Listing 81-3
1 # app/config/routing.yml
2 some_route:
3
path:
/%acme_demo.route_prefix%/contact
4
defaults: { _controller: AcmeDemoBundle:Main:contact }
PDF brought to you by
generated on December 3, 2014
Chapter 81: How to Use Service Container Parameters in your Routes | 295
Just like in normal service container configuration files, if you actually need a % in your route, you
can escape the percent sign by doubling it, e.g. /score-50%%, which would resolve to /score-50%.
However, as the % characters included in any URL are automatically encoded, the resulting URL of
this example would be /score-50%25 (%25 is the result of encoding the % character).
For parameter handling within a Dependency Injection class see Using Parameters within a Dependency
Injection Class.
PDF brought to you by
generated on December 3, 2014
Chapter 81: How to Use Service Container Parameters in your Routes | 296
Chapter 82
How to Create a custom Route Loader
A custom route loader allows you to add routes to an application without including them, for example,
in a YAML file. This comes in handy when you have a bundle but don't want to manually add the routes
for the bundle to app/config/routing.yml. This may be especially important when you want to make
the bundle reusable, or when you have open-sourced it as this would slow down the installation process
and make it error-prone.
Alternatively, you could also use a custom route loader when you want your routes to be automatically
generated or located based on some convention or pattern. One example is the FOSRestBundle1 where
routing is generated based off the names of the action methods in a controller.
There are many bundles out there that use their own route loaders to accomplish cases like
those described above, for instance FOSRestBundle2, JMSI18nRoutingBundle3, KnpRadBundle4 and
SonataAdminBundle5.
Loading Routes
The routes in a Symfony application are loaded by the DelegatingLoader6. This loader uses several other
loaders (delegates) to load resources of different types, for instance YAML files or @Route and @Method
annotations in controller files. The specialized loaders implement LoaderInterface7 and therefore have
two important methods: supports()8 and load()9.
Take these lines from the routing.yml in the AcmeDemoBundle of the Standard Edition:
Listing 82-1
1.
2.
3.
4.
5.
6.
https://github.com/FriendsOfSymfony/FOSRestBundle
https://github.com/FriendsOfSymfony/FOSRestBundle
https://github.com/schmittjoh/JMSI18nRoutingBundle
https://github.com/KnpLabs/KnpRadBundle
https://github.com/sonata-project/SonataAdminBundle
http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Routing/DelegatingLoader.html
7. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html
8. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html#supports()
9. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html#load()
PDF brought to you by
generated on December 3, 2014
Chapter 82: How to Create a custom Route Loader | 297
1 # src/Acme/DemoBundle/Resources/config/routing.yml
2 _demo:
3
resource: "@AcmeDemoBundle/Controller/DemoController.php"
4
type:
annotation
5
prefix:
/demo
When the main loader parses this, it tries all the delegate loaders and calls their supports()10 method
with the given resource (@AcmeDemoBundle/Controller/DemoController.php) and type (annotation)
as arguments. When one of the loader returns true, its load()11 method will be called, which should
return a RouteCollection12 containing Route13 objects.
Creating a custom Loader
To load routes from some custom source (i.e. from something other than annotations, YAML or XML
files), you need to create a custom route loader. This loader should implement LoaderInterface14.
The sample loader below supports loading routing resources with a type of extra. The type extra isn't
important - you can just invent any resource type you want. The resource name itself is not actually used
in the example:
Listing 82-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
namespace Acme\DemoBundle\Routing;
use
use
use
use
Symfony\Component\Config\Loader\LoaderInterface;
Symfony\Component\Config\Loader\LoaderResolverInterface;
Symfony\Component\Routing\Route;
Symfony\Component\Routing\RouteCollection;
class ExtraLoader implements LoaderInterface
{
private $loaded = false;
public function load($resource, $type = null)
{
if (true === $this->loaded) {
throw new \RuntimeException('Do not add the "extra" loader twice');
}
$routes = new RouteCollection();
// prepare a new route
$pattern = '/extra/{parameter}';
$defaults = array(
'_controller' => 'AcmeDemoBundle:Demo:extra',
);
$requirements = array(
'parameter' => '\d+',
);
$route = new Route($pattern, $defaults, $requirements);
// add the new route to the route collection:
10. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html#supports()
11. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html#load()
12. http://api.symfony.com/master/Symfony/Component/Routing/RouteCollection.html
13. http://api.symfony.com/master/Symfony/Component/Routing/Route.html
14. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 82: How to Create a custom Route Loader | 298
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54 }
$routeName = 'extraRoute';
$routes->add($routeName, $route);
$this->loaded = true;
return $routes;
}
public function supports($resource, $type = null)
{
return 'extra' === $type;
}
public function getResolver()
{
// needed, but can be blank, unless you want to load other resources
// and if you do, using the Loader base class is easier (see below)
}
public function setResolver(LoaderResolverInterface $resolver)
{
// same as above
}
Make sure the controller you specify really exists.
Now define a service for the ExtraLoader:
Listing 82-3
1 services:
2
acme_demo.routing_loader:
3
class: Acme\DemoBundle\Routing\ExtraLoader
4
tags:
5
- { name: routing.loader }
Notice the tag routing.loader. All services with this tag will be marked as potential route loaders and
added as specialized routers to the DelegatingLoader15.
Using the custom Loader
If you did nothing else, your custom routing loader would not be called. Instead, you only need to add a
few extra lines to the routing configuration:
Listing 82-4
1 # app/config/routing.yml
2 AcmeDemoBundle_Extra:
3
resource: .
4
type: extra
The important part here is the type key. Its value should be "extra". This is the type which the
ExtraLoader supports and this will make sure its load() method gets called. The resource key is
insignificant for the ExtraLoader, so it is set to ".".
15. http://api.symfony.com/master/Symfony/Bundle/FrameworkBundle/Routing/DelegatingLoader.html
PDF brought to you by
generated on December 3, 2014
Chapter 82: How to Create a custom Route Loader | 299
The routes defined using custom route loaders will be automatically cached by the framework. So
whenever you change something in the loader class itself, don't forget to clear the cache.
More advanced Loaders
In most cases it's better not to implement LoaderInterface16 yourself, but extend from Loader17. This
class knows how to use a LoaderResolver18 to load secondary routing resources.
Of course you still need to implement supports()19 and load()20. Whenever you want to load another
resource - for instance a YAML routing configuration file - you can call the import()21 method:
Listing 82-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
namespace Acme\DemoBundle\Routing;
use Symfony\Component\Config\Loader\Loader;
use Symfony\Component\Routing\RouteCollection;
class AdvancedLoader extends Loader
{
public function load($resource, $type = null)
{
$collection = new RouteCollection();
$resource = '@AcmeDemoBundle/Resources/config/import_routing.yml';
$type = 'yaml';
$importedRoutes = $this->import($resource, $type);
$collection->addCollection($importedRoutes);
return $collection;
}
public function supports($resource, $type = null)
{
return $type === 'advanced_extra';
}
}
The resource name and type of the imported routing configuration can be anything that would
normally be supported by the routing configuration loader (YAML, XML, PHP, annotation, etc.).
16. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html
17. http://api.symfony.com/master/Symfony/Component/Config/Loader/Loader.html
18. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderResolver.html
19. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html#supports()
20. http://api.symfony.com/master/Symfony/Component/Config/Loader/LoaderInterface.html#load()
21. http://api.symfony.com/master/Symfony/Component/Config/Loader/Loader.html#import()
PDF brought to you by
generated on December 3, 2014
Chapter 82: How to Create a custom Route Loader | 300
Chapter 83
Redirect URLs with a Trailing Slash
The goal of this cookbook is to demonstrate how to redirect URLs with a trailing slash to the same URL
without a trailing slash (for example /en/blog/ to /en/blog).
Create a controller that will match any URL with a trailing slash, remove the trailing slash (keeping query
parameters if any) and redirect to the new URL with a 301 response status code:
Listing 83-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/Acme/DemoBundle/Controller/RedirectingController.php
namespace Acme\DemoBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
use Symfony\Component\HttpFoundation\Request;
class RedirectingController extends Controller
{
public function removeTrailingSlashAction(Request $request)
{
$pathInfo = $request->getPathInfo();
$requestUri = $request->getRequestUri();
$url = str_replace($pathInfo, rtrim($pathInfo, ' /'), $requestUri);
return $this->redirect($url, 301);
}
}
After that, create a route to this controller that's matched whenever a URL with a trailing slash is
requested. Be sure to put this route last in your system, as explained below:
Listing 83-2
1 remove_trailing_slash:
2
path: /{url}
3
defaults: { _controller: AcmeDemoBundle:Redirecting:removeTrailingSlash }
4
requirements:
5
url: .*/$
6
methods: [GET]
PDF brought to you by
generated on December 3, 2014
Chapter 83: Redirect URLs with a Trailing Slash | 301
Redirecting a POST request does not work well in old browsers. A 302 on a POST request would
send a GET request after the redirection for legacy reasons. For that reason, the route here only
matches GET requests.
Make sure to include this route in your routing configuration at the very end of your route listing.
Otherwise, you risk redirecting real routes (including Symfony core routes) that actually do have a
trailing slash in their path.
PDF brought to you by
generated on December 3, 2014
Chapter 83: Redirect URLs with a Trailing Slash | 302
Chapter 84
How to Pass Extra Information from a Route to
a Controller
Parameters inside the defaults collection don't necessarily have to match a placeholder in the route
path. In fact, you can use the defaults array to specify extra parameters that will then be accessible as
arguments to your controller:
Listing 84-1
1 # app/config/routing.yml
2 blog:
3
path:
/blog/{page}
4
defaults:
5
_controller: AcmeBlogBundle:Blog:index
6
page:
1
7
title:
"Hello world!"
Now, you can access this extra parameter in your controller:
Listing 84-2
1 public function indexAction($page, $title)
2 {
3
// ...
4 }
As you can see, the $title variable was never defined inside the route path, but you can still access its
value from inside your controller.
PDF brought to you by
generated on December 3, 2014
Chapter 84: How to Pass Extra Information from a Route to a Controller | 303
Chapter 85
How to Load Security Users from the Database
(the Entity Provider)
The security layer is one of the smartest tools of Symfony. It handles two things: the authentication
and the authorization processes. Although it may seem difficult to understand how it works internally,
the security system is very flexible and allows you to integrate your application with any authentication
backend, like Active Directory, an OAuth server or a database.
Introduction
This article focuses on how to authenticate users against a database table managed by a Doctrine entity
class. The content of this cookbook entry is split in three parts. The first part is about designing a
Doctrine User entity class and making it usable in the security layer of Symfony. The second part
describes how to easily authenticate a user with the Doctrine EntityUserProvider1 object bundled with
the framework and some configuration. Finally, the tutorial will demonstrate how to create a custom
EntityUserProvider2 object to retrieve users from a database with custom conditions.
Code along with the Example
If you want to follow along with the example in this chapter, create an AcmeUserBundle via:
Listing 85-1
1 $ php app/console generate:bundle --namespace=Acme/UserBundle
1. http://api.symfony.com/master/Symfony/Bridge/Doctrine/Security/User/EntityUserProvider.html
2. http://api.symfony.com/master/Symfony/Bridge/Doctrine/Security/User/EntityUserProvider.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 304
The Data Model
For the purpose of this cookbook, the AcmeUserBundle bundle contains a User entity class with the
following fields: id, username, password, email and isActive. The isActive field tells whether or not
the user account is active.
To make it shorter, the getter and setter methods for each have been removed to focus on the most
important methods that come from the UserInterface3.
You can generate the missing getter and setters by running:
Listing 85-2
Listing 85-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
1 $ php app/console doctrine:generate:entities Acme/UserBundle/Entity/User
// src/Acme/UserBundle/Entity/User.php
namespace Acme\UserBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
use Symfony\Component\Security\Core\User\UserInterface;
/**
* Acme\UserBundle\Entity\User
*
* @ORM\Table(name="acme_users")
* @ORM\Entity(repositoryClass="Acme\UserBundle\Entity\UserRepository")
*/
class User implements UserInterface, \Serializable
{
/**
* @ORM\Column(type="integer")
* @ORM\Id
* @ORM\GeneratedValue(strategy="AUTO")
*/
private $id;
/**
* @ORM\Column(type="string", length=25, unique=true)
*/
private $username;
/**
* @ORM\Column(type="string", length=64)
*/
private $password;
/**
* @ORM\Column(type="string", length=60, unique=true)
*/
private $email;
/**
* @ORM\Column(name="is_active", type="boolean")
*/
private $isActive;
3. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 305
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
public function __construct()
{
$this->isActive = true;
// may not be needed, see section on salt below
// $this->salt = md5(uniqid(null, true));
}
/**
* @inheritDoc
*/
public function getUsername()
{
return $this->username;
}
/**
* @inheritDoc
*/
public function getSalt()
{
// you *may* need a real salt depending on your encoder
// see section on salt below
return null;
}
/**
* @inheritDoc
*/
public function getPassword()
{
return $this->password;
}
/**
* @inheritDoc
*/
public function getRoles()
{
return array('ROLE_USER');
}
/**
* @inheritDoc
*/
public function eraseCredentials()
{
}
/**
* @see \Serializable::serialize()
*/
public function serialize()
{
return serialize(array(
$this->id,
$this->username,
$this->password,
// see section on salt below
// $this->salt,
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 306
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117 }
));
}
/**
* @see \Serializable::unserialize()
*/
public function unserialize($serialized)
{
list (
$this->id,
$this->username,
$this->password,
// see section on salt below
// $this->salt
) = unserialize($serialized);
}
If you choose to implement EquatableInterface4, you determine yourself which properties need
to be compared to distinguish your user objects.
Generate the database table for your User entity by running:
Listing 85-4
1 $ php app/console doctrine:schema:update --force
In order to use an instance of the AcmeUserBundle:User class in the Symfony security layer, the entity
class must implement the UserInterface5. This interface forces the class to implement the five following
methods:
•
•
•
•
•
getRoles()6
getPassword()7
getSalt()8
getUsername()9
eraseCredentials()10
For more details on each of these, see UserInterface11.
4. http://api.symfony.com/master/Symfony/Component/Security/Core/User/EquatableInterface.html
5. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html
6. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getRoles()
7. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getPassword()
8. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getSalt()
9. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getUsername()
10. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#eraseCredentials()
11. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 307
What is the importance of serialize and unserialize?
The Serializable12 interface and its serialize and unserialize methods have been added to
allow the User class to be serialized to the session. This may or may not be needed depending
on your setup, but it's probably a good idea. The id is the most important value that needs to be
serialized because the refreshUser()13 method reloads the user on each request by using the id.
In practice, this means that the User object is reloaded from the database on each request using the
id from the serialized object. This makes sure all of the User's data is fresh.
Symfony also uses the username, salt, and password to verify that the User has not changed
between requests. Failing to serialize these may cause you to be logged out on each request. If your
User implements EquatableInterface14, then instead of these properties being checked, your
isEqualTo method is simply called, and you can check whatever properties you want. Unless you
understand this, you probably won't need to implement this interface or worry about it.
Below is an export of the User table from MySQL with user admin and password admin (which has been
encoded). For details on how to create user records and encode their password, see Encoding the User's
Password.
Listing 85-5
1
2
3
4
5
6
$ mysql> SELECT * FROM acme_users;
+----+----------+------------------------------------------+--------------------+-----------+
| id | username | password
| email
| is_active |
+----+----------+------------------------------------------+--------------------+-----------+
| 1 | admin
| d033e22ae348aeb5660fc2140aec35850c4da997 | [email protected] |
1 |
+----+----------+------------------------------------------+--------------------+-----------+
The next part will focus on how to authenticate one of these users thanks to the Doctrine entity user
provider and a couple of lines of configuration.
Do you need to use a Salt?
Yes. Hashing a password with a salt is a necessary step so that encoded passwords can't be
decoded. However, some encoders - like Bcrypt - have a built-in salt mechanism. If you configure
bcrypt as your encoder in security.yml (see the next section), then getSalt() should return
null, so that Bcrypt generates the salt itself.
However, if you use an encoder that does not have a built-in salting ability (e.g. sha512), you must
(from a security perspective) generate your own, random salt, store it on a salt property that is
saved to the database, and return it from getSalt(). Some of the code needed is commented out
in the above example.
Authenticating Someone against a Database
Authenticating a Doctrine user against the database with the Symfony security layer is a piece of cake.
Everything resides in the configuration of the SecurityBundle stored in the app/config/security.yml
file.
Below is an example of configuration where the user will enter their username and password via HTTP
basic authentication. That information will then be checked against your User entity records in the
database:
12. http://php.net/manual/en/class.serializable.php
13. http://api.symfony.com/master/Symfony/Bridge/Doctrine/Security/User/EntityUserProvider.html#refreshUser()
14. http://api.symfony.com/master/Symfony/Component/Security/Core/User/EquatableInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 308
Listing 85-6
1 # app/config/security.yml
2 security:
3
encoders:
4
Acme\UserBundle\Entity\User:
5
algorithm: bcrypt
6
7
role_hierarchy:
8
ROLE_ADMIN:
ROLE_USER
9
ROLE_SUPER_ADMIN: [ ROLE_ADMIN, ROLE_ALLOWED_TO_SWITCH ]
10
11
providers:
12
administrators:
13
entity: { class: AcmeUserBundle:User, property: username }
14
15
firewalls:
16
admin_area:
17
pattern:
^/admin
18
http_basic: ~
19
20
access_control:
21
- { path: ^/admin, roles: ROLE_ADMIN }
The encoders section associates the bcrypt password encoder to the entity class. This means that
Symfony will expect the password that's stored in the database to be encoded using this encoder. For
details on how to create a new User object with a properly encoded password, see the Encoding the User's
Password section of the security chapter.
If you're using PHP 5.4 or lower, you'll need to install the ircmaxell/password-compat library via
Composer in order to be able to use the bcrypt encoder:
Listing 85-7
{
"require": {
...
"ircmaxell/password-compat": "~1.0.3"
}
}
The providers section defines an administrators user provider. A user provider is a "source" of where
users are loaded during authentication. In this case, the entity keyword means that Symfony will use
the Doctrine entity user provider to load User entity objects from the database by using the username
unique field. In other words, this tells Symfony how to fetch the user from the database before checking
the password validity.
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 309
By default, the entity provider uses the default entity manager to fetch user information from the
database. If you use multiple entity managers, you can specify which manager to use with the
manager_name option:
Listing 85-8
1 # app/config/config.yml
2 security:
3
# ...
4
5
providers:
6
administrators:
7
entity:
8
class: AcmeUserBundle:User
9
property: username
10
manager_name: customer
11
12
# ...
Forbid inactive Users
If a User's isActive property is set to false (i.e. is_active is 0 in the database), the user will still be
able to login access the site normally. To prevent "inactive" users from logging in, you'll need to do a little
more work.
The easiest way to exclude inactive users is to implement the AdvancedUserInterface15 interface
that takes care of checking the user's account status. The AdvancedUserInterface16 extends the
UserInterface17 interface, so you just need to switch to the new interface in the AcmeUserBundle:User
entity class to benefit from simple and advanced authentication behaviors.
The AdvancedUserInterface18 interface adds four extra methods to validate the account status:
•
•
•
•
isAccountNonExpired()19 checks whether the user's account has expired;
isAccountNonLocked()20 checks whether the user is locked;
isCredentialsNonExpired()21 checks whether the user's credentials (password) has expired;
isEnabled()22 checks whether the user is enabled.
For this example, the first three methods will return true whereas the isEnabled() method will return
the boolean value in the isActive field.
Listing 85-9
1
2
3
4
5
6
7
8
9
// src/Acme/UserBundle/Entity/User.php
namespace Acme\UserBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
use Symfony\Component\Security\Core\User\AdvancedUserInterface;
class User implements AdvancedUserInterface, \Serializable
{
// ...
15. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html
16. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html
17. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html
18. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html
19. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#isAccountNonExpired()
20. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#isAccountNonLocked()
21. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#isCredentialsNonExpired()
22. http://api.symfony.com/master/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#isEnabled()
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 310
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30 }
public function isAccountNonExpired()
{
return true;
}
public function isAccountNonLocked()
{
return true;
}
public function isCredentialsNonExpired()
{
return true;
}
public function isEnabled()
{
return $this->isActive;
}
Now, if you try to authenticate as a user who's is_active database field is set to 0, you won't be allowed.
When using the AdvancedUserInterface, you should also add any of the properties used by these
methods (like isActive()) to the serialize() method. If you don't do this, your user may not be
deserialized correctly from the session on each request.
The next session will focus on how to write a custom entity provider to authenticate a user with their
username or email address.
Authenticating Someone with a Custom Entity Provider
The next step is to allow a user to authenticate with their username or email address as they are both
unique in the database. Unfortunately, the native entity provider is only able to handle a single property
to fetch the user from the database.
To accomplish this, create a custom entity provider that looks for a user whose username or email field
matches the submitted login username. The good news is that a Doctrine repository object can act as
an entity user provider if it implements the UserProviderInterface23. This interface comes with three
methods to implement: loadUserByUsername($username), refreshUser(UserInterface $user), and
supportsClass($class). For more details, see UserProviderInterface24.
The code below shows the implementation of the UserProviderInterface25 in the UserRepository
class:
Listing 85-10
1
2
3
4
5
// src/Acme/UserBundle/Entity/UserRepository.php
namespace Acme\UserBundle\Entity;
use Symfony\Component\Security\Core\User\UserInterface;
use Symfony\Component\Security\Core\User\UserProviderInterface;
23. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserProviderInterface.html
24. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserProviderInterface.html
25. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserProviderInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 311
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
use
use
use
use
Symfony\Component\Security\Core\Exception\UsernameNotFoundException;
Symfony\Component\Security\Core\Exception\UnsupportedUserException;
Doctrine\ORM\EntityRepository;
Doctrine\ORM\NoResultException;
class UserRepository extends EntityRepository implements UserProviderInterface
{
public function loadUserByUsername($username)
{
$q = $this
->createQueryBuilder('u')
->where('u.username = :username OR u.email = :email')
->setParameter('username', $username)
->setParameter('email', $username)
->getQuery();
try {
// The Query::getSingleResult() method throws an exception
// if there is no record matching the criteria.
$user = $q->getSingleResult();
} catch (NoResultException $e) {
$message = sprintf(
'Unable to find an active admin AcmeUserBundle:User object identified by
"%s".',
$username
);
throw new UsernameNotFoundException($message, 0, $e);
}
return $user;
}
public function refreshUser(UserInterface $user)
{
$class = get_class($user);
if (!$this->supportsClass($class)) {
throw new UnsupportedUserException(
sprintf(
'Instances of "%s" are not supported.',
$class
)
);
}
return $this->find($user->getId());
}
public function supportsClass($class)
{
return $this->getEntityName() === $class
|| is_subclass_of($class, $this->getEntityName());
}
}
To finish the implementation, the configuration of the security layer must be changed to tell Symfony to
use the new custom entity provider instead of the generic Doctrine entity provider. It's trivial to achieve
by removing the property field in the security.providers.administrators.entity section of the
security.yml file.
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 312
Listing 85-11
1 # app/config/security.yml
2 security:
3
# ...
4
providers:
5
administrators:
6
entity: { class: AcmeUserBundle:User }
7
# ...
By doing this, the security layer will use an instance of UserRepository and call its
loadUserByUsername() method to fetch a user from the database whether they filled in their username
or email address.
Managing Roles in the Database
The end of this tutorial focuses on how to store and retrieve a list of roles from the database. As
mentioned previously, when your user is loaded, its getRoles() method returns the array of security
roles that should be assigned to the user. You can load this data from anywhere - a hardcoded list
used for all users (e.g. array('ROLE_USER')), a Doctrine array property called roles, or via a Doctrine
relationship, as you'll learn about in this section.
In a typical setup, you should always return at least 1 role from the getRoles() method. By
convention, a role called ROLE_USER is usually returned. If you fail to return any roles, it may appear
as if your user isn't authenticated at all.
In order to work with the security configuration examples on this page all roles must be prefixed
with ROLE_ (see the section about roles in the book). For example, your roles will be ROLE_ADMIN or
ROLE_USER instead of ADMIN or USER.
In this example, the AcmeUserBundle:User entity class defines a many-to-many relationship with a
AcmeUserBundle:Role entity class. A user can be related to several roles and a role can be composed of
one or more users. The previous getRoles() method now returns the list of related roles. Notice that
__construct() and getRoles() methods have changed:
Listing 85-12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/Acme/UserBundle/Entity/User.php
namespace Acme\UserBundle\Entity;
use Doctrine\Common\Collections\ArrayCollection;
// ...
class User implements AdvancedUserInterface, \Serializable
{
// ...
/**
* @ORM\ManyToMany(targetEntity="Role", inversedBy="users")
*
*/
private $roles;
public function __construct()
{
$this->roles = new ArrayCollection();
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 313
20
21
22
23
24
25
26
27
28
29 }
}
public function getRoles()
{
return $this->roles->toArray();
}
// ...
The AcmeUserBundle:Role entity class defines three fields (id, name and role). The unique role field
contains the role name (e.g. ROLE_ADMIN) used by the Symfony security layer to secure parts of the
application:
Listing 85-13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
// src/Acme/Bundle/UserBundle/Entity/Role.php
namespace Acme\UserBundle\Entity;
use Symfony\Component\Security\Core\Role\RoleInterface;
use Doctrine\Common\Collections\ArrayCollection;
use Doctrine\ORM\Mapping as ORM;
/**
* @ORM\Table(name="acme_role")
* @ORM\Entity()
*/
class Role implements RoleInterface
{
/**
* @ORM\Column(name="id", type="integer")
* @ORM\Id()
* @ORM\GeneratedValue(strategy="AUTO")
*/
private $id;
/**
* @ORM\Column(name="name", type="string", length=30)
*/
private $name;
/**
* @ORM\Column(name="role", type="string", length=20, unique=true)
*/
private $role;
/**
* @ORM\ManyToMany(targetEntity="User", mappedBy="roles")
*/
private $users;
public function __construct()
{
$this->users = new ArrayCollection();
}
/**
* @see RoleInterface
*/
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 314
44
45
46
47
48
49
50 }
public function getRole()
{
return $this->role;
}
// ... getters and setters for each property
For brevity, the getter and setter methods are hidden, but you can generate them:
Listing 85-14
1 $ php app/console doctrine:generate:entities Acme/UserBundle/Entity/User
Don't forget also to update your database schema:
Listing 85-15
1 $ php app/console doctrine:schema:update --force
This will create the acme_role table and a user_role that stores the many-to-many relationship between
acme_user and acme_role. If you had one user linked to one role, your database might look something
like this:
Listing 85-16
1
2
3
4
5
6
7
8
9
10
11
12
13
$ mysql> SELECT * FROM acme_role;
+----+-------+------------+
| id | name | role
|
+----+-------+------------+
| 1 | admin | ROLE_ADMIN |
+----+-------+------------+
$ mysql> SELECT * FROM user_role;
+---------+---------+
| user_id | role_id |
+---------+---------+
|
1 |
1 |
+---------+---------+
And that's it! When the user logs in, Symfony security system will call the User::getRoles method. This
will return an array of Role objects that Symfony will use to determine if the user should have access to
certain parts of the system.
What's the purpose of the RoleInterface?
Notice that the Role class implements RoleInterface26. This is because Symfony's security system
requires that the User::getRoles method returns an array of either role strings or objects that
implement this interface. If Role didn't implement this interface, then User::getRoles would
need to iterate over all the Role objects, call getRole on each, and create an array of strings to
return. Both approaches are valid and equivalent.
Improving Performance with a Join
To improve performance and avoid lazy loading of roles when retrieving a user from the custom
entity provider, you can use a Doctrine join to the roles relationship in the
UserRepository::loadUserByUsername() method. This will fetch the user and their associated roles
with a single query:
26. http://api.symfony.com/master/Symfony/Component/Security/Core/Role/RoleInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 315
Listing 85-17
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/UserBundle/Entity/UserRepository.php
namespace Acme\UserBundle\Entity;
// ...
class UserRepository extends EntityRepository implements UserProviderInterface
{
public function loadUserByUsername($username)
{
$q = $this
->createQueryBuilder('u')
->select('u, r')
->leftJoin('u.roles', 'r')
->where('u.username = :username OR u.email = :email')
->setParameter('username', $username)
->setParameter('email', $username)
->getQuery();
// ...
}
// ...
}
The QueryBuilder::leftJoin() method joins and fetches related roles from the AcmeUserBundle:User
model class when a user is retrieved by their email address or username.
Understanding serialize and how a User is Saved in the Session
If you're curious about the importance of the serialize() method inside the User class or how the User
object is serialized or deserialized, then this section is for you. If not, feel free to skip this.
Once the user is logged in, the entire User object is serialized into the session. On the next request, the
User object is deserialized. Then, value of the id property is used to re-query for a fresh User object from
the database. Finally, the fresh User object is compared in some way to the deserialized User object to
make sure that they represent the same user. For example, if the username on the 2 User objects doesn't
match for some reason, then the user will be logged out for security reasons.
Even though this all happens automatically, there are a few important side-effects.
First, the Serializable27 interface and its serialize and unserialize methods have been added to
allow the User class to be serialized to the session. This may or may not be needed depending on
your setup, but it's probably a good idea. In theory, only the id needs to be serialized, because the
refreshUser()28 method refreshes the user on each request by using the id (as explained above).
However in practice, this means that the User object is reloaded from the database on each request using
the id from the serialized object. This makes sure all of the User's data is fresh.
Symfony also uses the username, salt, and password to verify that the User has not changed between
requests. Failing to serialize these may cause you to be logged out on each request. If your User
implements the EquatableInterface29, then instead of these properties being checked, your isEqualTo
method is simply called, and you can check whatever properties you want. Unless you understand this,
you probably won't need to implement this interface or worry about it.
27. http://php.net/manual/en/class.serializable.php
28. http://api.symfony.com/master/Symfony/Bridge/Doctrine/Security/User/EntityUserProvider.html#refreshUser()
29. http://api.symfony.com/master/Symfony/Component/Security/Core/User/EquatableInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 85: How to Load Security Users from the Database (the Entity Provider) | 316
Chapter 86
How to Add "Remember Me" Login
Functionality
Once a user is authenticated, their credentials are typically stored in the session. This means that when
the session ends they will be logged out and have to provide their login details again next time they wish
to access the application. You can allow users to choose to stay logged in for longer than the session lasts
using a cookie with the remember_me firewall option. The firewall needs to have a secret key configured,
which is used to encrypt the cookie's content. It also has several options with default values which are
shown here:
Listing 86-1
1 # app/config/security.yml
2 firewalls:
3
main:
4
remember_me:
5
key:
"%secret%"
6
lifetime: 31536000 # 365 days in seconds
7
path:
/
8
domain:
~ # Defaults to the current domain from $_SERVER
It's a good idea to provide the user with the option to use or not use the remember me functionality, as
it will not always be appropriate. The usual way of doing this is to add a checkbox to the login form. By
giving the checkbox the name _remember_me, the cookie will automatically be set when the checkbox is
checked and the user successfully logs in. So, your specific login form might ultimately looks like this:
Listing 86-2
1
2
3
4
5
6
7
8
9
10
{# src/Acme/SecurityBundle/Resources/views/Security/login.html.twig #}
{% if error %}
<div>{{ error.message }}</div>
{% endif %}
<form action="{{ path('login_check') }}" method="post">
<label for="username">Username:</label>
<input type="text" id="username" name="_username" value="{{ last_username }}" />
<label for="password">Password:</label>
PDF brought to you by
generated on December 3, 2014
Chapter 86: How to Add "Remember Me" Login Functionality | 317
11
<input
12
13
<input
14
<label
15
16
<input
17 </form>
type="password" id="password" name="_password" />
type="checkbox" id="remember_me" name="_remember_me" checked />
for="remember_me">Keep me logged in</label>
type="submit" name="login" />
The user will then automatically be logged in on subsequent visits while the cookie remains valid.
Forcing the User to Re-authenticate before Accessing certain Resources
When the user returns to your site, they are authenticated automatically based on the information stored
in the remember me cookie. This allows the user to access protected resources as if the user had actually
authenticated upon visiting the site.
In some cases, however, you may want to force the user to actually re-authenticate before accessing
certain resources. For example, you might allow "remember me" users to see basic account information,
but then require them to actually re-authenticate before modifying that information.
The Security component provides an easy way to do this. In addition to roles explicitly assigned to them,
users are automatically given one of the following roles depending on how they are authenticated:
• IS_AUTHENTICATED_ANONYMOUSLY - automatically assigned to a user who is in a firewall
protected part of the site but who has not actually logged in. This is only possible if anonymous
access has been allowed.
• IS_AUTHENTICATED_REMEMBERED - automatically assigned to a user who was authenticated via
a remember me cookie.
• IS_AUTHENTICATED_FULLY - automatically assigned to a user that has provided their login
details during the current session.
You can use these to control access beyond the explicitly assigned roles.
If you have the IS_AUTHENTICATED_REMEMBERED role, then you also have the
IS_AUTHENTICATED_ANONYMOUSLY role. If you have the IS_AUTHENTICATED_FULLY role, then you
also have the other two roles. In other words, these roles represent three levels of increasing
"strength" of authentication.
You can use these additional roles for finer grained control over access to parts of a site. For example,
you may want your user to be able to view their account at /account when authenticated by cookie but
to have to provide their login details to be able to edit the account details. You can do this by securing
specific controller actions using these roles. The edit action in the controller could be secured using the
service context.
In the following example, the action is only allowed if the user has the IS_AUTHENTICATED_FULLY role.
Listing 86-3
1
2
3
4
5
6
7
8
9
// ...
use Symfony\Component\Security\Core\Exception\AccessDeniedException
public function editAction()
{
if (false === $this->get('security.context')->isGranted(
'IS_AUTHENTICATED_FULLY'
)) {
throw new AccessDeniedException();
PDF brought to you by
generated on December 3, 2014
Chapter 86: How to Add "Remember Me" Login Functionality | 318
10
11
12
13 }
}
// ...
You can also choose to install and use the optional JMSSecurityExtraBundle1, which can secure your
controller using annotations:
Listing 86-4
1
2
3
4
5
6
7
8
9
use JMS\SecurityExtraBundle\Annotation\Secure;
/**
* @Secure(roles="IS_AUTHENTICATED_FULLY")
*/
public function editAction($name)
{
// ...
}
If you also had an access control in your security configuration that required the user to have a
ROLE_USER role in order to access any of the account area, then you'd have the following situation:
• If a non-authenticated (or anonymously authenticated user) tries to access the account
area, the user will be asked to authenticate.
• Once the user has entered their username and password, assuming the user receives
the ROLE_USER role per your configuration, the user will have the
IS_AUTHENTICATED_FULLY role and be able to access any page in the account section,
including the editAction controller.
• If the user's session ends, when the user returns to the site, they will be able to access
every account page - except for the edit page - without being forced to re-authenticate.
However, when they try to access the editAction controller, they will be forced to reauthenticate, since they are not, yet, fully authenticated.
For more information on securing services or methods in this way, see How to Secure any Service or
Method in your Application.
1. https://github.com/schmittjoh/JMSSecurityExtraBundle
PDF brought to you by
generated on December 3, 2014
Chapter 86: How to Add "Remember Me" Login Functionality | 319
Chapter 87
How to Impersonate a User
Sometimes, it's useful to be able to switch from one user to another without having to log out and log
in again (for instance when you are debugging or trying to understand a bug a user sees that you can't
reproduce). This can be easily done by activating the switch_user firewall listener:
Listing 87-1
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
# ...
6
switch_user: true
To switch to another user, just add a query string with the _switch_user parameter and the username as
the value to the current URL:
Listing 87-2
1 http://example.com/somewhere?_switch_user=thomas
To switch back to the original user, use the special _exit username:
Listing 87-3
1 http://example.com/somewhere?_switch_user=_exit
During impersonation, the user is provided with a special role called ROLE_PREVIOUS_ADMIN. In a
template, for instance, this role can be used to show a link to exit impersonation:
Listing 87-4
1 {% if is_granted('ROLE_PREVIOUS_ADMIN') %}
2
<a href="{{ path('homepage', {'_switch_user': '_exit'}) }}">Exit impersonation</a>
3 {% endif %}
Of course, this feature needs to be made available to a small group of users. By default, access is restricted
to users having the ROLE_ALLOWED_TO_SWITCH role. The name of this role can be modified via the role
setting. For extra security, you can also change the query parameter name via the parameter setting:
Listing 87-5
PDF brought to you by
generated on December 3, 2014
Chapter 87: How to Impersonate a User | 320
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
# ...
6
switch_user: { role: ROLE_ADMIN, parameter: _want_to_be_this_user }
PDF brought to you by
generated on December 3, 2014
Chapter 87: How to Impersonate a User | 321
Chapter 88
How to Implement your own Voter to Blacklist
IP Addresses
The Symfony Security component provides several layers to authorize users. One of the layers is called
a "voter". A voter is a dedicated class that checks if the user has the rights to connect to the application
or access a specific resource/URL. For instance, Symfony provides a layer that checks if the user is fully
authorized or if it has some expected roles.
It is sometimes useful to create a custom voter to handle a specific case not handled by the framework.
In this section, you'll learn how to create a voter that will allow you to blacklist users by their IP.
The Voter Interface
A custom voter must implement VoterInterface1, which requires the following three methods:
Listing 88-1
1 interface VoterInterface
2 {
3
public function supportsAttribute($attribute);
4
public function supportsClass($class);
5
public function vote(TokenInterface $token, $object, array $attributes);
6 }
The supportsAttribute()2 method is used to check if the voter supports the given user attribute (i.e: a
role like ROLE_USER, an ACL EDIT, etc.).
The supportsClass()3 method is used to check if the voter supports the class of the object whose access
is being checked.
The vote()4 method must implement the business logic that verifies whether or not the user has access.
This method must return one of the following values:
1. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html
2. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#supportsAttribute()
3. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#supportsClass()
4. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#vote()
PDF brought to you by
generated on December 3, 2014
Chapter 88: How to Implement your own Voter to Blacklist IP Addresses | 322
• VoterInterface::ACCESS_GRANTED: The authorization will be granted by this voter;
• VoterInterface::ACCESS_ABSTAIN: The voter cannot decide if authorization should be
granted;
• VoterInterface::ACCESS_DENIED: The authorization will be denied by this voter.
In this example, you'll check if the user's IP address matches against a list of blacklisted addresses
and "something" will be the application. If the user's IP is blacklisted, you'll return
VoterInterface::ACCESS_DENIED, otherwise you'll return VoterInterface::ACCESS_ABSTAIN as this
voter's purpose is only to deny access, not to grant access.
Creating a custom Voter
To blacklist a user based on its IP, you can use the request_stack service and compare the IP address
against a set of blacklisted IP addresses:
Listing 88-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
// src/Acme/DemoBundle/Security/Authorization/Voter/ClientIpVoter.php
namespace Acme\DemoBundle\Security\Authorization\Voter;
use Symfony\Component\HttpFoundation\RequestStack;
use Symfony\Component\Security\Core\Authorization\Voter\VoterInterface;
use Symfony\Component\Security\Core\Authentication\Token\TokenInterface;
class ClientIpVoter implements VoterInterface
{
protected $requestStack;
private $blacklistedIp;
public function __construct(RequestStack $requestStack, array $blacklistedIp = array())
{
$this->requestStack = $requestStack;
$this->blacklistedIp = $blacklistedIp;
}
public function supportsAttribute($attribute)
{
// you won't check against a user attribute, so return true
return true;
}
public function supportsClass($class)
{
// your voter supports all type of token classes, so return true
return true;
}
public function vote(TokenInterface $token, $object, array $attributes)
{
$request = $this->requestStack->getCurrentRequest();
if (in_array($request->getClientIp(), $this->blacklistedIp)) {
return VoterInterface::ACCESS_DENIED;
}
return VoterInterface::ACCESS_ABSTAIN;
}
}
That's it! The voter is done. The next step is to inject the voter into the security layer. This can be done
easily through the service container.
PDF brought to you by
generated on December 3, 2014
Chapter 88: How to Implement your own Voter to Blacklist IP Addresses | 323
Your implementation of the methods supportsAttribute()5 and supportsClass()6 are not
being called internally by the framework. Once you have registered your voter the vote() method
will always be called, regardless of whether or not these two methods return true. Therefore
you need to call those methods in your implementation of the vote() method and return
ACCESS_ABSTAIN if your voter does not support the class or attribute.
Declaring the Voter as a Service
To inject the voter into the security layer, you must declare it as a service, and tag it as a security.voter:
Listing 88-3
1 # src/Acme/AcmeBundle/Resources/config/services.yml
2 services:
3
security.access.blacklist_voter:
4
class:
Acme\DemoBundle\Security\Authorization\Voter\ClientIpVoter
5
arguments: ["@request_stack", [123.123.123.123, 171.171.171.171]]
6
public:
false
7
tags:
8
- { name: security.voter }
Be sure to import this configuration file from your main application configuration file (e.g. app/
config/config.yml). For more information see Importing Configuration with imports. To read
more about defining services in general, see the Service Container chapter.
Changing the Access Decision Strategy
In order for the new voter to take effect, you need to change the default access decision strategy, which,
by default, grants access if any voter grants access.
In this case, choose the unanimous strategy. Unlike the affirmative strategy (the default), with the
unanimous strategy, if only one voter denies access (e.g. the ClientIpVoter), access is not granted to the
end user.
To do that, override the default access_decision_manager section of your application configuration file
with the following code.
Listing 88-4
1 # app/config/security.yml
2 security:
3
access_decision_manager:
4
# strategy can be: affirmative, unanimous or consensus
5
strategy: unanimous
That's it! Now, when deciding whether or not a user should have access, the new voter will deny access
to any user in the list of blacklisted IPs.
For a more advanced usage see Access Decision Manager.
5. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#supportsAttribute()
6. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#supportsClass()
PDF brought to you by
generated on December 3, 2014
Chapter 88: How to Implement your own Voter to Blacklist IP Addresses | 324
Chapter 89
How to Use Voters to Check User Permissions
In Symfony, you can check the permission to access data by using the ACL module, which is a bit
overwhelming for many applications. A much easier solution is to work with custom voters, which are
like simple conditional statements.
Voters can also be used in other ways, like, for example, blacklisting IP addresses from the entire application:
How to Implement your own Voter to Blacklist IP Addresses.
Take a look at the authorization chapter for an even deeper understanding on voters.
How Symfony Uses Voters
In order to use voters, you have to understand how Symfony works with them. All voters are called
each time you use the isGranted() method on Symfony's security context (i.e. the security.context
service). Each one decides if the current user should have access to some resource.
Ultimately, Symfony uses one of three different approaches on what to do with the feedback from all
voters: affirmative, consensus and unanimous.
For more information take a look at the section about access decision managers.
The Voter Interface
A custom voter must implement VoterInterface1, which has this structure:
Listing 89-1
1 interface VoterInterface
2 {
1. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 89: How to Use Voters to Check User Permissions | 325
3
4
5
6 }
public function supportsAttribute($attribute);
public function supportsClass($class);
public function vote(TokenInterface $token, $object, array $attributes);
The supportsAttribute()2 method is used to check if the voter supports the given user attribute (i.e: a
role like ROLE_USER, an ACL EDIT, etc.).
The supportsClass()3 method is used to check if the voter supports the class of the object whose access
is being checked.
The vote()4 method must implement the business logic that verifies whether or not the user has access.
This method must return one of the following values:
• VoterInterface::ACCESS_GRANTED: The authorization will be granted by this voter;
• VoterInterface::ACCESS_ABSTAIN: The voter cannot decide if authorization should be
granted;
• VoterInterface::ACCESS_DENIED: The authorization will be denied by this voter.
In this example, the voter will check if the user has access to a specific object according to your
custom conditions (e.g. they must be the owner of the object). If the condition fails, you'll return
VoterInterface::ACCESS_DENIED, otherwise you'll return VoterInterface::ACCESS_GRANTED. In case
the responsibility for this decision does not belong to this voter, it will return
VoterInterface::ACCESS_ABSTAIN.
Creating the custom Voter
The goal is to create a voter that checks if a user has access to view or edit a particular object. Here's an
example implementation:
Listing 89-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// src/Acme/DemoBundle/Security/Authorization/Voter/PostVoter.php
namespace Acme\DemoBundle\Security\Authorization\Voter;
use Symfony\Component\Security\Core\Authorization\Voter\VoterInterface;
use Symfony\Component\Security\Core\Authentication\Token\TokenInterface;
use Symfony\Component\Security\Core\User\UserInterface;
class PostVoter implements VoterInterface
{
const VIEW = 'view';
const EDIT = 'edit';
public function supportsAttribute($attribute)
{
return in_array($attribute, array(
self::VIEW,
self::EDIT,
));
}
public function supportsClass($class)
{
2. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#supportsAttribute()
3. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#supportsClass()
4. http://api.symfony.com/master/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html#vote()
PDF brought to you by
generated on December 3, 2014
Chapter 89: How to Use Voters to Check User Permissions | 326
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
$supportedClass = 'Acme\DemoBundle\Entity\Post';
return $supportedClass === $class || is_subclass_of($class, $supportedClass);
}
/**
* @var \Acme\DemoBundle\Entity\Post $post
*/
public function vote(TokenInterface $token, $post, array $attributes)
{
// check if class of this object is supported by this voter
if (!$this->supportsClass(get_class($post))) {
return VoterInterface::ACCESS_ABSTAIN;
}
//
//
//
if
check if the voter is used correct, only allow one attribute
this isn't a requirement, it's just one easy way for you to
design your voter
(1 !== count($attributes)) {
throw new \InvalidArgumentException(
'Only one attribute is allowed for VIEW or EDIT'
);
}
// set the attribute to check against
$attribute = $attributes[0];
// check if the given attribute is covered by this voter
if (!$this->supportsAttribute($attribute)) {
return VoterInterface::ACCESS_ABSTAIN;
}
// get current logged in user
$user = $token->getUser();
// make sure there is a user object (i.e. that the user is logged in)
if (!$user instanceof UserInterface) {
return VoterInterface::ACCESS_DENIED;
}
switch($attribute) {
case self::VIEW:
// the data object could have for example a method isPrivate()
// which checks the Boolean attribute $private
if (!$post->isPrivate()) {
return VoterInterface::ACCESS_GRANTED;
}
break;
case self::EDIT:
// we assume that our data object has a method getOwner() to
// get the current owner user entity for this data object
if ($user->getId() === $post->getOwner()->getId()) {
return VoterInterface::ACCESS_GRANTED;
}
break;
}
return VoterInterface::ACCESS_DENIED;
PDF brought to you by
generated on December 3, 2014
Chapter 89: How to Use Voters to Check User Permissions | 327
82
83 }
}
That's it! The voter is done. The next step is to inject the voter into the security layer.
Declaring the Voter as a Service
To inject the voter into the security layer, you must declare it as a service and tag it with security.voter:
Listing 89-3
1 # src/Acme/DemoBundle/Resources/config/services.yml
2 services:
3
security.access.post_voter:
4
class:
Acme\DemoBundle\Security\Authorization\Voter\PostVoter
5
public:
false
6
tags:
7
- { name: security.voter }
How to Use the Voter in a Controller
The registered voter will then always be asked as soon as the method isGranted() from the security
context is called.
Listing 89-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// src/Acme/DemoBundle/Controller/PostController.php
namespace Acme\DemoBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
use Symfony\Component\HttpFoundation\Response;
use Symfony\Component\Security\Core\Exception\AccessDeniedException;
class PostController extends Controller
{
public function showAction($id)
{
// get a Post instance
$post = ...;
// keep in mind, this will call all registered security voters
if (false === $this->get('security.context')->isGranted('view', $post)) {
throw new AccessDeniedException('Unauthorised access!');
}
return new Response('<h1>'.$post->getName().'</h1>');
}
}
It's that easy!
PDF brought to you by
generated on December 3, 2014
Chapter 89: How to Use Voters to Check User Permissions | 328
Chapter 90
How to Use Access Control Lists (ACLs)
In complex applications, you will often face the problem that access decisions cannot only be based
on the person (Token) who is requesting access, but also involve a domain object that access is being
requested for. This is where the ACL system comes in.
Alternatives to ACLs
Using ACL's isn't trivial, and for simpler use cases, it may be overkill. If your permission logic
could be described by just writing some code (e.g. to check if a Blog is owned by the current User),
then consider using voters. A voter is passed the object being voted on, which you can use to
make complex decisions and effectively implement your own ACL. Enforcing authorization (e.g.
the isGranted part) will look similar to what you see in this entry, but your voter class will handle
the logic behind the scenes, instead of the ACL system.
Imagine you are designing a blog system where your users can comment on your posts. Now, you want a
user to be able to edit their own comments, but not those of other users; besides, you yourself want to be
able to edit all comments. In this scenario, Comment would be the domain object that you want to restrict
access to. You could take several approaches to accomplish this using Symfony, two basic approaches are
(non-exhaustive):
• Enforce security in your business methods: Basically, that means keeping a reference inside each
Comment to all users who have access, and then compare these users to the provided Token.
• Enforce security with roles: In this approach, you would add a role for each Comment object, i.e.
ROLE_COMMENT_1, ROLE_COMMENT_2, etc.
Both approaches are perfectly valid. However, they couple your authorization logic to your business code
which makes it less reusable elsewhere, and also increases the difficulty of unit testing. Besides, you could
run into performance issues if many users would have access to a single domain object.
Fortunately, there is a better way, which you will find out about now.
PDF brought to you by
generated on December 3, 2014
Chapter 90: How to Use Access Control Lists (ACLs) | 329
Bootstrapping
Now, before you can finally get into action, you need to do some bootstrapping. First, you need to
configure the connection the ACL system is supposed to use:
Listing 90-1
1 # app/config/security.yml
2 security:
3
acl:
4
connection: default
The ACL system requires a connection from either Doctrine DBAL (usable by default) or Doctrine
MongoDB (usable with MongoDBAclBundle1). However, that does not mean that you have to use
Doctrine ORM or ODM for mapping your domain objects. You can use whatever mapper you like
for your objects, be it Doctrine ORM, MongoDB ODM, Propel, raw SQL, etc. The choice is yours.
After the connection is configured, you have to import the database structure. Fortunately, there is a task
for this. Simply run the following command:
Listing 90-2
1 $ php app/console init:acl
Getting Started
Coming back to the small example from the beginning, you can now implement ACL for it.
Once the ACL is created, you can grant access to objects by creating an Access Control Entity (ACE) to
solidify the relationship between the entity and your user.
Creating an ACL and Adding an ACE
Listing 90-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/DemoBundle/Controller/BlogController.php
namespace Acme\DemoBundle\Controller;
use
use
use
use
use
Symfony\Bundle\FrameworkBundle\Controller\Controller;
Symfony\Component\Security\Core\Exception\AccessDeniedException;
Symfony\Component\Security\Acl\Domain\ObjectIdentity;
Symfony\Component\Security\Acl\Domain\UserSecurityIdentity;
Symfony\Component\Security\Acl\Permission\MaskBuilder;
class BlogController extends Controller
{
// ...
public function addCommentAction(Post $post)
{
$comment = new Comment();
// ... setup $form, and submit data
if ($form->isValid()) {
$entityManager = $this->getDoctrine()->getManager();
1. https://github.com/IamPersistent/MongoDBAclBundle
PDF brought to you by
generated on December 3, 2014
Chapter 90: How to Use Access Control Lists (ACLs) | 330
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40 }
$entityManager->persist($comment);
$entityManager->flush();
// creating the ACL
$aclProvider = $this->get('security.acl.provider');
$objectIdentity = ObjectIdentity::fromDomainObject($comment);
$acl = $aclProvider->createAcl($objectIdentity);
// retrieving the security identity of the currently logged-in user
$securityContext = $this->get('security.context');
$user = $securityContext->getToken()->getUser();
$securityIdentity = UserSecurityIdentity::fromAccount($user);
// grant owner access
$acl->insertObjectAce($securityIdentity, MaskBuilder::MASK_OWNER);
$aclProvider->updateAcl($acl);
}
}
There are a couple of important implementation decisions in this code snippet. For now, I only want to
highlight two:
First, you may have noticed that ->createAcl() does not accept domain objects directly, but only
implementations of the ObjectIdentityInterface. This additional step of indirection allows you to
work with ACLs even when you have no actual domain object instance at hand. This will be extremely
helpful if you want to check permissions for a large number of objects without actually hydrating these
objects.
The other interesting part is the ->insertObjectAce() call. In the example, you are granting the user
who is currently logged in owner access to the Comment. The MaskBuilder::MASK_OWNER is a predefined integer bitmask; don't worry the mask builder will abstract away most of the technical details,
but using this technique you can store many different permissions in one database row which gives a
considerable boost in performance.
The order in which ACEs are checked is significant. As a general rule, you should place more
specific entries at the beginning.
Checking Access
Listing 90-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/DemoBundle/Controller/BlogController.php
// ...
class BlogController
{
// ...
public function editCommentAction(Comment $comment)
{
$securityContext = $this->get('security.context');
// check for edit access
if (false === $securityContext->isGranted('EDIT', $comment)) {
PDF brought to you by
generated on December 3, 2014
Chapter 90: How to Use Access Control Lists (ACLs) | 331
15
16
17
18
19
20 }
throw new AccessDeniedException();
}
// ... retrieve actual comment object, and do your editing here
}
In this example, you check whether the user has the EDIT permission. Internally, Symfony maps the
permission to several integer bitmasks, and checks whether the user has any of them.
You can define up to 32 base permissions (depending on your OS PHP might vary between 30 to
32). In addition, you can also define cumulative permissions.
Cumulative Permissions
In the first example above, you only granted the user the OWNER base permission. While this effectively
also allows the user to perform any operation such as view, edit, etc. on the domain object, there are
cases where you may want to grant these permissions explicitly.
The MaskBuilder can be used for creating bit masks easily by combining several base permissions:
Listing 90-5
1
2
3
4
5
6
7
8
$builder = new MaskBuilder();
$builder
->add('view')
->add('edit')
->add('delete')
->add('undelete')
;
$mask = $builder->get(); // int(29)
This integer bitmask can then be used to grant a user the base permissions you added above:
Listing 90-6
1 $identity = new UserSecurityIdentity('johannes', 'Acme\UserBundle\Entity\User');
2 $acl->insertObjectAce($identity, $mask);
The user is now allowed to view, edit, delete, and un-delete objects.
PDF brought to you by
generated on December 3, 2014
Chapter 90: How to Use Access Control Lists (ACLs) | 332
Chapter 91
How to Use advanced ACL Concepts
The aim of this chapter is to give a more in-depth view of the ACL system, and also explain some of the
design decisions behind it.
Design Concepts
Symfony's object instance security capabilities are based on the concept of an Access Control List. Every
domain object instance has its own ACL. The ACL instance holds a detailed list of Access Control
Entries (ACEs) which are used to make access decisions. Symfony's ACL system focuses on two main
objectives:
• providing a way to efficiently retrieve a large amount of ACLs/ACEs for your domain objects,
and to modify them;
• providing a way to easily make decisions of whether a person is allowed to perform an action
on a domain object or not.
As indicated by the first point, one of the main capabilities of Symfony's ACL system is a highperformance way of retrieving ACLs/ACEs. This is extremely important since each ACL might have
several ACEs, and inherit from another ACL in a tree-like fashion. Therefore, no ORM is leveraged,
instead the default implementation interacts with your connection directly using Doctrine's DBAL.
Object Identities
The ACL system is completely decoupled from your domain objects. They don't even have to be stored
in the same database, or on the same server. In order to achieve this decoupling, in the ACL system your
objects are represented through object identity objects. Every time you want to retrieve the ACL for a
domain object, the ACL system will first create an object identity from your domain object, and then pass
this object identity to the ACL provider for further processing.
Security Identities
This is analog to the object identity, but represents a user, or a role in your application. Each role, or user
has its own security identity.
PDF brought to you by
generated on December 3, 2014
Chapter 91: How to Use advanced ACL Concepts | 333
New in version 2.5: For users, the security identity is based on the username. This means that, if
for any reason, a user's username was to change, you must ensure its security identity is updated
too. The MutableAclProvider::updateUserSecurityIdentity()1 method is there to handle the
update, it was introduced in Symfony 2.5.
Database Table Structure
The default implementation uses five database tables as listed below. The tables are ordered from least
rows to most rows in a typical application:
• acl_security_identities: This table records all security identities (SID) which hold ACEs. The
default implementation ships with two security identities: RoleSecurityIdentity2 and
UserSecurityIdentity3.
• acl_classes: This table maps class names to a unique ID which can be referenced from other
tables.
• acl_object_identities: Each row in this table represents a single domain object instance.
• acl_object_identity_ancestors: This table allows all the ancestors of an ACL to be determined
in a very efficient way.
• acl_entries: This table contains all ACEs. This is typically the table with the most rows. It can
contain tens of millions without significantly impacting performance.
Scope of Access Control Entries
Access control entries can have different scopes in which they apply. In Symfony, there are basically two
different scopes:
• Class-Scope: These entries apply to all objects with the same class.
• Object-Scope: This was the scope solely used in the previous chapter, and it only applies to
one specific object.
Sometimes, you will find the need to apply an ACE only to a specific field of the object. Suppose you
want the ID only to be viewable by an administrator, but not by your customer service. To solve this
common problem, two more sub-scopes have been added:
• Class-Field-Scope: These entries apply to all objects with the same class, but only to a specific
field of the objects.
• Object-Field-Scope: These entries apply to a specific object, and only to a specific field of that
object.
Pre-Authorization Decisions
For pre-authorization decisions, that is decisions made before any secure method (or secure action) is
invoked, the proven AccessDecisionManager service is used. The AccessDecisionManager is also used
for reaching authorization decisions based on roles. Just like roles, the ACL system adds several new
attributes which may be used to check for different permissions.
1. http://api.symfony.com/master/Symfony/Component/Security/Acl/Dbal/MutableAclProvider.html#updateUserSecurityIdentity()
2. http://api.symfony.com/master/Symfony/Component/Security/Acl/Domain/RoleSecurityIdentity.html
3. http://api.symfony.com/master/Symfony/Component/Security/Acl/Domain/UserSecurityIdentity.html
PDF brought to you by
generated on December 3, 2014
Chapter 91: How to Use advanced ACL Concepts | 334
Built-in Permission Map
Attribute
Intended Meaning
Integer Bitmasks
VIEW
Whether someone is allowed to
view the domain object.
VIEW, EDIT, OPERATOR,
MASTER, or OWNER
EDIT
Whether someone is allowed to
EDIT, OPERATOR, MASTER, or
make changes to the domain object. OWNER
CREATE
Whether someone is allowed to
create the domain object.
CREATE, OPERATOR, MASTER, or
OWNER
DELETE
Whether someone is allowed to
delete the domain object.
DELETE, OPERATOR, MASTER, or
OWNER
UNDELETE
Whether someone is allowed to
restore a previously deleted domain
object.
UNDELETE, OPERATOR,
MASTER, or OWNER
OPERATOR
Whether someone is allowed to
perform all of the above actions.
OPERATOR, MASTER, or OWNER
MASTER
Whether someone is allowed to
perform all of the above actions,
and in addition is allowed to grant
any of the above permissions to
others.
MASTER, or OWNER
OWNER
Whether someone owns the
domain object. An owner can
perform any of the above actions
and grant master and owner
permissions.
OWNER
Permission Attributes vs. Permission Bitmasks
Attributes are used by the AccessDecisionManager, just like roles. Often, these attributes represent in
fact an aggregate of integer bitmasks. Integer bitmasks on the other hand, are used by the ACL system
internally to efficiently store your users' permissions in the database, and perform access checks using
extremely fast bitmask operations.
Extensibility
The above permission map is by no means static, and theoretically could be completely replaced at will.
However, it should cover most problems you encounter, and for interoperability with other bundles, you
are encouraged to stick to the meaning envisaged for them.
Post Authorization Decisions
Post authorization decisions are made after a secure method has been invoked, and typically involve the
domain object which is returned by such a method. After invocation providers also allow to modify, or
filter the domain object before it is returned.
PDF brought to you by
generated on December 3, 2014
Chapter 91: How to Use advanced ACL Concepts | 335
Due to current limitations of the PHP language, there are no post-authorization capabilities build into the
core Security component. However, there is an experimental JMSSecurityExtraBundle4 which adds these
capabilities. See its documentation for further information on how this is accomplished.
Process for Reaching Authorization Decisions
The ACL class provides two methods for determining whether a security identity has the required
bitmasks, isGranted and isFieldGranted. When the ACL receives an authorization request through
one of these methods, it delegates this request to an implementation of PermissionGrantingStrategy5.
This allows you to replace the way access decisions are reached without actually modifying the ACL class
itself.
The PermissionGrantingStrategy first checks all your object-scope ACEs. If none is applicable, the
class-scope ACEs will be checked. If none is applicable, then the process will be repeated with the ACEs
of the parent ACL. If no parent ACL exists, an exception will be thrown.
4. https://github.com/schmittjoh/JMSSecurityExtraBundle
5. http://api.symfony.com/master/Symfony/Component/Security/Acl/Domain/PermissionGrantingStrategy.html
PDF brought to you by
generated on December 3, 2014
Chapter 91: How to Use advanced ACL Concepts | 336
Chapter 92
How to Force HTTPS or HTTP for different URLs
You can force areas of your site to use the HTTPS protocol in the security config. This is done through the
access_control rules using the requires_channel option. For example, if you want to force all URLs
starting with /secure to use HTTPS then you could use the following configuration:
Listing 92-1
1 access_control:
2
- { path: ^/secure, roles: ROLE_ADMIN, requires_channel: https }
The login form itself needs to allow anonymous access, otherwise users will be unable to authenticate.
To force it to use HTTPS you can still use access_control rules by using the
IS_AUTHENTICATED_ANONYMOUSLY role:
Listing 92-2
1 access_control:
2
- { path: ^/login, roles: IS_AUTHENTICATED_ANONYMOUSLY, requires_channel: https }
It is also possible to specify using HTTPS in the routing configuration, see How to Force Routes to always
Use HTTPS or HTTP for more details.
PDF brought to you by
generated on December 3, 2014
Chapter 92: How to Force HTTPS or HTTP for different URLs | 337
Chapter 93
How to Restrict Firewalls to a Specific Request
When using the Security component, you can create firewalls that match certain request options. In most
cases, matching against the URL is sufficient, but in special cases you can further restrict the initialization
of a firewall against other options of the request.
You can use any of these restrictions individually or mix them together to get your desired firewall
configuration.
Restricting by Pattern
This is the default restriction and restricts a firewall to only be initialized if the request URL matches the
configured pattern.
Listing 93-1
1 # app/config/security.yml
2
3 # ...
4 security:
5
firewalls:
6
secured_area:
7
pattern: ^/admin
8
# ...
The pattern is a regular expression. In this example, the firewall will only be activated if the URL starts
(due to the ^ regex character) with /admin. If the URL does not match this pattern, the firewall will not
be activated and subsequent firewalls will have the opportunity to be matched for this request.
PDF brought to you by
generated on December 3, 2014
Chapter 93: How to Restrict Firewalls to a Specific Request | 338
Restricting by Host
New in version 2.4: Support for restricting security firewalls to a specific host was introduced in
Symfony 2.4.
If matching against the pattern only is not enough, the request can also be matched against host. When
the configuration option host is set, the firewall will be restricted to only initialize if the host from the
request matches against the configuration.
Listing 93-2
1 # app/config/security.yml
2
3 # ...
4 security:
5
firewalls:
6
secured_area:
7
host: ^admin\.example\.com$
8
# ...
The host (like the pattern) is a regular expression. In this example, the firewall will only be activated
if the host is equal exactly (due to the ^ and $ regex characters) to the hostname admin.example.com. If
the hostname does not match this pattern, the firewall will not be activated and subsequent firewalls will
have the opportunity to be matched for this request.
Restricting by HTTP Methods
New in version 2.5: Support for restricting security firewalls to specific HTTP methods was
introduced in Symfony 2.5.
The configuration option methods restricts the initialization of the firewall to the provided HTTP
methods.
Listing 93-3
1 # app/config/security.yml
2
3 # ...
4 security:
5
firewalls:
6
secured_area:
7
methods: [GET, POST]
8
# ...
In this example, the firewall will only be activated if the HTTP method of the request is either GET or
POST. If the method is not in the array of the allowed methods, the firewall will not be activated and
subsequent firewalls will again have the opportunity to be matched for this request.
PDF brought to you by
generated on December 3, 2014
Chapter 93: How to Restrict Firewalls to a Specific Request | 339
Chapter 94
How to Customize your Form Login
Using a form login for authentication is a common, and flexible, method for handling authentication in
Symfony. Pretty much every aspect of the form login can be customized. The full, default configuration
is shown in the next section.
Form Login Configuration Reference
To see the full form login configuration reference, see SecurityBundle Configuration ("security"). Some of
the more interesting options are explained below.
Redirecting after Success
You can change where the login form redirects after a successful login using the various config options.
By default the form will redirect to the URL the user requested (i.e. the URL which triggered the login
form being shown). For example, if the user requested http://www.example.com/admin/post/18/edit,
then after they successfully log in, they will eventually be sent back to http://www.example.com/admin/
post/18/edit. This is done by storing the requested URL in the session. If no URL is present in the
session (perhaps the user went directly to the login page), then the user is redirected to the default page,
which is / (i.e. the homepage) by default. You can change this behavior in several ways.
As mentioned, by default the user is redirected back to the page originally requested. Sometimes,
this can cause problems, like if a background Ajax request "appears" to be the last visited URL,
causing the user to be redirected there. For information on controlling this behavior, see How to
Change the default Target Path Behavior.
Changing the default Page
First, the default page can be set (i.e. the page the user is redirected to if no previous page was stored in
the session). To set it to the default_security_target route use the following config:
Listing 94-1
PDF brought to you by
generated on December 3, 2014
Chapter 94: How to Customize your Form Login | 340
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
form_login:
6
# ...
7
default_target_path: default_security_target
Now, when no URL is set in the session, users will be sent to the default_security_target route.
Always Redirect to the default Page
You can make it so that users are always redirected to the default page regardless of what URL they had
requested previously by setting the always_use_default_target_path option to true:
Listing 94-2
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
form_login:
6
# ...
7
always_use_default_target_path: true
Using the Referring URL
In case no previous URL was stored in the session, you may wish to try using the HTTP_REFERER instead,
as this will often be the same. You can do this by setting use_referer to true (it defaults to false):
Listing 94-3
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
form_login:
6
# ...
7
use_referer:
true
Control the Redirect URL from inside the Form
You can also override where the user is redirected to via the form itself by including a hidden field with
the name _target_path. For example, to redirect to the URL defined by some account route, use the
following:
Listing 94-4
1
2
3
4
5
6
7
8
9
10
11
{# src/Acme/SecurityBundle/Resources/views/Security/login.html.twig #}
{% if error %}
<div>{{ error.message }}</div>
{% endif %}
<form action="{{ path('login_check') }}" method="post">
<label for="username">Username:</label>
<input type="text" id="username" name="_username" value="{{ last_username }}" />
<label for="password">Password:</label>
<input type="password" id="password" name="_password" />
PDF brought to you by
generated on December 3, 2014
Chapter 94: How to Customize your Form Login | 341
12
13
<input type="hidden" name="_target_path" value="account" />
14
15
<input type="submit" name="login" />
16 </form>
Now, the user will be redirected to the value of the hidden form field. The value attribute can be a relative
path, absolute URL, or a route name. You can even change the name of the hidden form field by changing
the target_path_parameter option to another value.
Listing 94-5
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
form_login:
6
target_path_parameter: redirect_url
Redirecting on Login Failure
In addition to redirecting the user after a successful login, you can also set the URL that the user should
be redirected to after a failed login (e.g. an invalid username or password was submitted). By default, the
user is redirected back to the login form itself. You can set this to a different route (e.g. login_failure)
with the following config:
Listing 94-6
1 # app/config/security.yml
2 security:
3
firewalls:
4
main:
5
form_login:
6
# ...
7
failure_path: login_failure
PDF brought to you by
generated on December 3, 2014
Chapter 94: How to Customize your Form Login | 342
Chapter 95
How to Secure any Service or Method in your
Application
In the security chapter, you can see how to secure a controller by requesting the security.context
service from the Service Container and checking the current user's role:
Listing 95-1
1
2
3
4
5
6
7
8
9
10
11
// ...
use Symfony\Component\Security\Core\Exception\AccessDeniedException;
public function helloAction($name)
{
if (false === $this->get('security.context')->isGranted('ROLE_ADMIN')) {
throw new AccessDeniedException();
}
// ...
}
You can also secure any service in a similar way by injecting the security.context service into it. For
a general introduction to injecting dependencies into services see the Service Container chapter of the
book. For example, suppose you have a NewsletterManager class that sends out emails and you want to
restrict its use to only users who have some ROLE_NEWSLETTER_ADMIN role. Before you add security, the
class looks something like this:
Listing 95-2
1
2
3
4
5
6
7
8
9
10
// src/Acme/HelloBundle/Newsletter/NewsletterManager.php
namespace Acme\HelloBundle\Newsletter;
class NewsletterManager
{
public function sendNewsletter()
{
// ... where you actually do the work
}
PDF brought to you by
generated on December 3, 2014
Chapter 95: How to Secure any Service or Method in your Application | 343
11
12
13 }
// ...
Your goal is to check the user's role when the sendNewsletter() method is called. The first step towards
this is to inject the security.context service into the object. Since it won't make sense not to perform
the security check, this is an ideal candidate for constructor injection, which guarantees that the security
context object will be available inside the NewsletterManager class:
Listing 95-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
namespace Acme\HelloBundle\Newsletter;
use Symfony\Component\Security\Core\SecurityContextInterface;
class NewsletterManager
{
protected $securityContext;
public function __construct(SecurityContextInterface $securityContext)
{
$this->securityContext = $securityContext;
}
// ...
}
Then in your service configuration, you can inject the service:
Listing 95-4
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 parameters:
3
newsletter_manager.class: Acme\HelloBundle\Newsletter\NewsletterManager
4
5 services:
6
newsletter_manager:
7
class:
"%newsletter_manager.class%"
8
arguments: ["@security.context"]
The injected service can then be used to perform the security check when the sendNewsletter() method
is called:
Listing 95-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
namespace Acme\HelloBundle\Newsletter;
use Symfony\Component\Security\Core\Exception\AccessDeniedException;
use Symfony\Component\Security\Core\SecurityContextInterface;
// ...
class NewsletterManager
{
protected $securityContext;
public function __construct(SecurityContextInterface $securityContext)
{
$this->securityContext = $securityContext;
}
public function sendNewsletter()
{
PDF brought to you by
generated on December 3, 2014
Chapter 95: How to Secure any Service or Method in your Application | 344
18
19
20
21
22
23
24
25
26 }
if (false === $this->securityContext->isGranted('ROLE_NEWSLETTER_ADMIN')) {
throw new AccessDeniedException();
}
// ...
}
// ...
If the current user does not have the ROLE_NEWSLETTER_ADMIN, they will be prompted to log in.
Securing Methods Using Annotations
You can also secure method calls in any service with annotations by using the optional
JMSSecurityExtraBundle1 bundle. This bundle is not included in the Symfony Standard Distribution, but
you can choose to install it.
To enable the annotations functionality, tag the service you want to secure with the
security.secure_service tag (you can also automatically enable this functionality for all services, see
the sidebar below):
Listing 95-6
1 # src/Acme/HelloBundle/Resources/config/services.yml
2
3 # ...
4 services:
5
newsletter_manager:
6
# ...
7
tags:
8
- { name: security.secure_service }
You can then achieve the same results as above using an annotation:
Listing 95-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
namespace Acme\HelloBundle\Newsletter;
use JMS\SecurityExtraBundle\Annotation\Secure;
// ...
class NewsletterManager
{
/**
* @Secure(roles="ROLE_NEWSLETTER_ADMIN")
*/
public function sendNewsletter()
{
// ...
}
// ...
}
1. https://github.com/schmittjoh/JMSSecurityExtraBundle
PDF brought to you by
generated on December 3, 2014
Chapter 95: How to Secure any Service or Method in your Application | 345
The annotations work because a proxy class is created for your class which performs the security
checks. This means that, whilst you can use annotations on public and protected methods, you
cannot use them with private methods or methods marked final.
The JMSSecurityExtraBundle also allows you to secure the parameters and return values of methods. For
more information, see the JMSSecurityExtraBundle2 documentation.
Activating the Annotations Functionality for all Services
When securing the method of a service (as shown above), you can either tag each service
individually, or activate the functionality for all services at once. To do so, set the
secure_all_services configuration option to true:
Listing 95-8
1 # app/config/config.yml
2 jms_security_extra:
3
# ...
4
secure_all_services: true
The disadvantage of this method is that, if activated, the initial page load may be very slow
depending on how many services you have defined.
2. https://github.com/schmittjoh/JMSSecurityExtraBundle
PDF brought to you by
generated on December 3, 2014
Chapter 95: How to Secure any Service or Method in your Application | 346
Chapter 96
How to Create a custom User Provider
Part of Symfony's standard authentication process depends on "user providers". When a user submits a
username and password, the authentication layer asks the configured user provider to return a user object
for a given username. Symfony then checks whether the password of this user is correct and generates a
security token so the user stays authenticated during the current session. Out of the box, Symfony has
an "in_memory" and an "entity" user provider. In this entry you'll see how you can create your own user
provider, which could be useful if your users are accessed via a custom database, a file, or - as shown in
this example - a web service.
Create a User Class
First, regardless of where your user data is coming from, you'll need to create a User class that represents
that data. The User can look however you want and contain any data. The only requirement is that
the class implements UserInterface1. The methods in this interface should therefore be defined in the
custom user class: getRoles()2, getPassword()3, getSalt()4, getUsername()5, eraseCredentials()6.
It may also be useful to implement the EquatableInterface7 interface, which defines a method to check
if the user is equal to the current user. This interface requires an isEqualTo()8 method.
This is how your WebserviceUser class looks in action:
Listing 96-1
1
2
3
4
5
6
// src/Acme/WebserviceUserBundle/Security/User/WebserviceUser.php
namespace Acme\WebserviceUserBundle\Security\User;
use Symfony\Component\Security\Core\User\UserInterface;
use Symfony\Component\Security\Core\User\EquatableInterface;
1. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html
2. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getRoles()
3. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getPassword()
4. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getSalt()
5. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#getUsername()
6. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserInterface.html#eraseCredentials()
7. http://api.symfony.com/master/Symfony/Component/Security/Core/User/EquatableInterface.html
8. http://api.symfony.com/master/Symfony/Component/Security/Core/User/EquatableInterface.html#isEqualTo()
PDF brought to you by
generated on December 3, 2014
Chapter 96: How to Create a custom User Provider | 347
7 class WebserviceUser implements UserInterface, EquatableInterface
8 {
9
private $username;
10
private $password;
11
private $salt;
12
private $roles;
13
14
public function __construct($username, $password, $salt, array $roles)
15
{
16
$this->username = $username;
17
$this->password = $password;
18
$this->salt = $salt;
19
$this->roles = $roles;
20
}
21
22
public function getRoles()
23
{
24
return $this->roles;
25
}
26
27
public function getPassword()
28
{
29
return $this->password;
30
}
31
32
public function getSalt()
33
{
34
return $this->salt;
35
}
36
37
public function getUsername()
38
{
39
return $this->username;
40
}
41
42
public function eraseCredentials()
43
{
44
}
45
46
public function isEqualTo(UserInterface $user)
47
{
48
if (!$user instanceof WebserviceUser) {
49
return false;
50
}
51
52
if ($this->password !== $user->getPassword()) {
53
return false;
54
}
55
56
if ($this->salt !== $user->getSalt()) {
57
return false;
58
}
59
60
if ($this->username !== $user->getUsername()) {
61
return false;
62
}
63
64
return true;
PDF brought to you by
generated on December 3, 2014
Chapter 96: How to Create a custom User Provider | 348
65
66 }
}
If you have more information about your users - like a "first name" - then you can add a firstName field
to hold that data.
Create a User Provider
Now that you have a User class, you'll create a user provider, which will grab user information from some
web service, create a WebserviceUser object, and populate it with data.
The user provider is just a plain PHP class that has to implement the UserProviderInterface9, which
requires three methods to be defined: loadUserByUsername($username), refreshUser(UserInterface
$user), and supportsClass($class). For more details, see UserProviderInterface10.
Here's an example of how this might look:
Listing 96-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
// src/Acme/WebserviceUserBundle/Security/User/WebserviceUserProvider.php
namespace Acme\WebserviceUserBundle\Security\User;
use
use
use
use
Symfony\Component\Security\Core\User\UserProviderInterface;
Symfony\Component\Security\Core\User\UserInterface;
Symfony\Component\Security\Core\Exception\UsernameNotFoundException;
Symfony\Component\Security\Core\Exception\UnsupportedUserException;
class WebserviceUserProvider implements UserProviderInterface
{
public function loadUserByUsername($username)
{
// make a call to your webservice here
$userData = ...
// pretend it returns an array on success, false if there is no user
if ($userData) {
$password = '...';
// ...
return new WebserviceUser($username, $password, $salt, $roles);
}
throw new UsernameNotFoundException(
sprintf('Username "%s" does not exist.', $username)
);
}
public function refreshUser(UserInterface $user)
{
if (!$user instanceof WebserviceUser) {
throw new UnsupportedUserException(
sprintf('Instances of "%s" are not supported.', get_class($user))
);
}
9. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserProviderInterface.html
10. http://api.symfony.com/master/Symfony/Component/Security/Core/User/UserProviderInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 96: How to Create a custom User Provider | 349
38
39
40
41
42
43
44
45 }
return $this->loadUserByUsername($user->getUsername());
}
public function supportsClass($class)
{
return $class === 'Acme\WebserviceUserBundle\Security\User\WebserviceUser';
}
Create a Service for the User Provider
Now you make the user provider available as a service:
Listing 96-3
1
2
3
4
5
6
7
# src/Acme/WebserviceUserBundle/Resources/config/services.yml
parameters:
webservice_user_provider.class:
Acme\WebserviceUserBundle\Security\User\WebserviceUserProvider
services:
webservice_user_provider:
class: "%webservice_user_provider.class%"
The real implementation of the user provider will probably have some dependencies or
configuration options or other services. Add these as arguments in the service definition.
Make sure the services file is being imported. See Importing Configuration with imports for details.
Modify security.yml
Everything comes together in your security configuration. Add the user provider to the list of providers
in the "security" section. Choose a name for the user provider (e.g. "webservice") and mention the id of
the service you just defined.
Listing 96-4
1 # app/config/security.yml
2 security:
3
providers:
4
webservice:
5
id: webservice_user_provider
Symfony also needs to know how to encode passwords that are supplied by website users, e.g. by filling
in a login form. You can do this by adding a line to the "encoders" section in your security configuration:
Listing 96-5
1 # app/config/security.yml
2 security:
PDF brought to you by
generated on December 3, 2014
Chapter 96: How to Create a custom User Provider | 350
3
4
encoders:
Acme\WebserviceUserBundle\Security\User\WebserviceUser: sha512
The value here should correspond with however the passwords were originally encoded when creating
your users (however those users were created). When a user submits their password, the salt value
is appended to the password and then encoded using this algorithm before being compared to the
hashed password returned by your getPassword() method. Additionally, depending on your options,
the password may be encoded multiple times and encoded to base64.
Specifics on how Passwords are Encoded
Symfony uses a specific method to combine the salt and encode the password before comparing it
to your encoded password. If getSalt() returns nothing, then the submitted password is simply
encoded using the algorithm you specify in security.yml. If a salt is specified, then the following
value is created and then hashed via the algorithm:
$password.'{'.$salt.'}';
If your external users have their passwords salted via a different method, then you'll need to do
a bit more work so that Symfony properly encodes the password. That is beyond the scope of
this entry, but would include sub-classing MessageDigestPasswordEncoder and overriding the
mergePasswordAndSalt method.
Additionally, the hash, by default, is encoded multiple times and encoded to base64. For specific
details, see MessageDigestPasswordEncoder11. To prevent this, configure it in your configuration
file:
Listing 96-6
1 # app/config/security.yml
2 security:
3
encoders:
4
Acme\WebserviceUserBundle\Security\User\WebserviceUser:
5
algorithm: sha512
6
encode_as_base64: false
7
iterations: 1
11. https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Security/Core/Encoder/MessageDigestPasswordEncoder.php
PDF brought to you by
generated on December 3, 2014
Chapter 96: How to Create a custom User Provider | 351
Chapter 97
How to Create a Custom Form Password
Authenticator
Imagine you want to allow access to your website only between 2pm and 4pm UTC. Before Symfony 2.4,
you had to create a custom token, factory, listener and provider. In this entry, you'll learn how to do this
for a login form (i.e. where your user submits their username and password). Before Symfony 2.6, you
had to use the password encoder to authenticate the user password.
The Password Authenticator
New in version 2.4: The SimpleFormAuthenticatorInterface interface was introduced in
Symfony 2.4.
New in version 2.6: The UserPasswordEncoderInterface interface was introduced in Symfony
2.6.
First, create a new class that implements SimpleFormAuthenticatorInterface1. Eventually, this will
allow you to create custom logic for authenticating the user:
Listing 97-1
1
2
3
4
5
6
7
// src/Acme/HelloBundle/Security/TimeAuthenticator.php
namespace Acme\HelloBundle\Security;
use
use
use
use
Symfony\Component\HttpFoundation\Request;
Symfony\Component\Security\Core\Authentication\SimpleFormAuthenticatorInterface;
Symfony\Component\Security\Core\Authentication\Token\TokenInterface;
Symfony\Component\Security\Core\Authentication\Token\UsernamePasswordToken;
1. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/SimpleFormAuthenticatorInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 97: How to Create a Custom Form Password Authenticator | 352
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
use
use
use
use
Symfony\Component\Security\Core\Encoder\UserPasswordEncoderInterface;
Symfony\Component\Security\Core\Exception\AuthenticationException;
Symfony\Component\Security\Core\Exception\UsernameNotFoundException;
Symfony\Component\Security\Core\User\UserProviderInterface;
class TimeAuthenticator implements SimpleFormAuthenticatorInterface
{
private $encoder;
public function __construct(UserPasswordEncoderInterface $encoder)
{
$this->encoder = $encoder;
}
public function authenticateToken(TokenInterface $token, UserProviderInterface
$userProvider, $providerKey)
{
try {
$user = $userProvider->loadUserByUsername($token->getUsername());
} catch (UsernameNotFoundException $e) {
throw new AuthenticationException('Invalid username or password');
}
$passwordValid = $this->encoder->isPasswordValid($user, $token->getCredentials());
if ($passwordValid) {
$currentHour = date('G');
if ($currentHour < 14 || $currentHour > 16) {
throw new AuthenticationException(
'You can only log in between 2 and 4!',
100
);
}
return new UsernamePasswordToken(
$user,
$user->getPassword(),
$providerKey,
$user->getRoles()
);
}
throw new AuthenticationException('Invalid username or password');
}
public function supportsToken(TokenInterface $token, $providerKey)
{
return $token instanceof UsernamePasswordToken
&& $token->getProviderKey() === $providerKey;
}
public function createToken(Request $request, $username, $password, $providerKey)
{
return new UsernamePasswordToken($username, $password, $providerKey);
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 97: How to Create a Custom Form Password Authenticator | 353
How it Works
Great! Now you just need to setup some Configuration. But first, you can find out more about what each
method in this class does.
1) createToken
When Symfony begins handling a request, createToken() is called, where you create a TokenInterface2
object that contains whatever information you need in authenticateToken() to authenticate the user
(e.g. the username and password).
Whatever token object you create here will be passed to you later in authenticateToken().
2) supportsToken
After Symfony calls createToken(), it will then call supportsToken() on your class (and any other
authentication listeners) to figure out who should handle the token. This is just a way to allow several
authentication mechanisms to be used for the same firewall (that way, you can for instance first try to
authenticate the user via a certificate or an API key and fall back to a form login).
Mostly, you just need to make sure that this method returns true for a token that has been created by
createToken(). Your logic should probably look exactly like this example.
3) authenticateToken
If supportsToken returns true, Symfony will now call authenticateToken(). Your job here is to check
that the token is allowed to log in by first getting the User object via the user provider and then, by
checking the password and the current time.
The "flow" of how you get the User object and determine whether or not the token is valid (e.g.
checking the password), may vary based on your requirements.
Ultimately, your job is to return a new token object that is "authenticated" (i.e. it has at least 1 role set on
it) and which has the User object inside of it.
Inside this method, the password encoder is needed to check the password's validity:
Listing 97-2
1 $passwordValid = $this->encoder->isPasswordValid($user, $token->getCredentials());
This is a service that is already available in Symfony and it uses the password algorithm that is configured
in the security configuration (e.g. security.yml) under the encoders key. Below, you'll see how to inject
that into the TimeAuthenticator.
Configuration
Now, configure your TimeAuthenticator as a service:
Listing 97-3
1 # app/config/config.yml
2 services:
3
# ...
2. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/Token/TokenInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 97: How to Create a Custom Form Password Authenticator | 354
4
5
6
7
time_authenticator:
class:
Acme\HelloBundle\Security\TimeAuthenticator
arguments: ["@security.password_encoder"]
Then, activate it in the firewalls section of the security configuration using the simple_form key:
Listing 97-4
1 # app/config/security.yml
2 security:
3
# ...
4
5
firewalls:
6
secured_area:
7
pattern: ^/admin
8
# ...
9
simple_form:
10
authenticator: time_authenticator
11
check_path:
login_check
12
login_path:
login
The simple_form key has the same options as the normal form_login option, but with the additional
authenticator key that points to the new service. For details, see Form Login Configuration.
If creating a login form in general is new to you or you don't understand the check_path or login_path
options, see How to Customize your Form Login.
PDF brought to you by
generated on December 3, 2014
Chapter 97: How to Create a Custom Form Password Authenticator | 355
Chapter 98
How to Authenticate Users with API Keys
Nowadays, it's quite usual to authenticate the user via an API key (when developing a web service for
instance). The API key is provided for every request and is passed as a query string parameter or via an
HTTP header.
The API Key Authenticator
New in version 2.4: The SimplePreAuthenticatorInterface interface was introduced in Symfony
2.4.
Authenticating a user based on the Request information should be done via a pre-authentication
mechanism. The SimplePreAuthenticatorInterface1 allows you to implement such a scheme really
easily.
Your exact situation may differ, but in this example, a token is read from an apikey query parameter, the
proper username is loaded from that value and then a User object is created:
Listing 98-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/HelloBundle/Security/ApiKeyAuthenticator.php
namespace Acme\HelloBundle\Security;
use
use
use
use
use
use
use
Symfony\Component\Security\Core\Authentication\SimplePreAuthenticatorInterface;
Symfony\Component\Security\Core\Authentication\Token\TokenInterface;
Symfony\Component\Security\Core\Exception\AuthenticationException;
Symfony\Component\Security\Core\Authentication\Token\PreAuthenticatedToken;
Symfony\Component\HttpFoundation\Request;
Symfony\Component\Security\Core\User\UserProviderInterface;
Symfony\Component\Security\Core\Exception\BadCredentialsException;
class ApiKeyAuthenticator implements SimplePreAuthenticatorInterface
{
protected $userProvider;
1. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/SimplePreAuthenticatorInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 356
15
16
public function __construct(ApiKeyUserProvider $userProvider)
17
{
18
$this->userProvider = $userProvider;
19
}
20
21
public function createToken(Request $request, $providerKey)
22
{
23
// look for an apikey query parameter
24
$apiKey = $request->query->get('apikey');
25
26
// or if you want to use an "apikey" header, then do something like this:
27
// $apiKey = $request->headers->get('apikey');
28
29
if (!$apiKey) {
30
throw new BadCredentialsException('No API key found');
31
}
32
33
return new PreAuthenticatedToken(
34
'anon.',
35
$apiKey,
36
$providerKey
37
);
38
}
39
40
public function authenticateToken(TokenInterface $token, UserProviderInterface
41 $userProvider, $providerKey)
42
{
43
$apiKey = $token->getCredentials();
44
$username = $this->userProvider->getUsernameForApiKey($apiKey);
45
46
if (!$username) {
47
throw new AuthenticationException(
48
sprintf('API Key "%s" does not exist.', $apiKey)
49
);
50
}
51
52
$user = $this->userProvider->loadUserByUsername($username);
53
54
return new PreAuthenticatedToken(
55
$user,
56
$apiKey,
57
$providerKey,
58
$user->getRoles()
59
);
60
}
61
62
public function supportsToken(TokenInterface $token, $providerKey)
63
{
64
return $token instanceof PreAuthenticatedToken && $token->getProviderKey() ===
65 $providerKey;
}
}
Once you've configured everything, you'll be able to authenticate by adding an apikey parameter to the
query string, like http://example.com/admin/foo?apikey=37b51d194a7513e45b56f6524f2d51f2.
The authentication process has several steps, and your implementation will probably differ:
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 357
1. createToken
Early in the request cycle, Symfony calls createToken(). Your job here is to create a token object
that contains all of the information from the request that you need to authenticate the user (e.g. the
apikey query parameter). If that information is missing, throwing a BadCredentialsException2 will
cause authentication to fail.
2. supportsToken
After Symfony calls createToken(), it will then call supportsToken() on your class (and any other
authentication listeners) to figure out who should handle the token. This is just a way to allow several
authentication mechanisms to be used for the same firewall (that way, you can for instance first try to
authenticate the user via a certificate or an API key and fall back to a form login).
Mostly, you just need to make sure that this method returns true for a token that has been created by
createToken(). Your logic should probably look exactly like this example.
3. authenticateToken
If supportsToken() returns true, Symfony will now call authenticateToken(). One key part is the
$userProvider, which is an external class that helps you load information about the user. You'll learn
more about this next.
In this specific example, the following things happen in authenticateToken():
1. First, you use the $userProvider to somehow look up the $username that corresponds to the
$apiKey;
2. Second, you use the $userProvider again to load or create a User object for the $username;
3. Finally, you create an authenticated token (i.e. a token with at least one role) that has the proper
roles and the User object attached to it.
The goal is ultimately to use the $apiKey to find or create a User object. How you do this (e.g. query a
database) and the exact class for your User object may vary. Those differences will be most obvious in
your user provider.
The User Provider
The $userProvider can be any user provider (see How to Create a custom User Provider). In this
example, the $apiKey is used to somehow find the username for the user. This work is done in a
getUsernameForApiKey() method, which is created entirely custom for this use-case (i.e. this isn't a
method that's used by Symfony's core user provider system).
The $userProvider might look something like this:
Listing 98-2
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/Acme/HelloBundle/Security/ApiKeyUserProvider.php
namespace Acme\HelloBundle\Security;
use
use
use
use
Symfony\Component\Security\Core\User\UserProviderInterface;
Symfony\Component\Security\Core\User\User;
Symfony\Component\Security\Core\User\UserInterface;
Symfony\Component\Security\Core\Exception\UnsupportedUserException;
class ApiKeyUserProvider implements UserProviderInterface
{
public function getUsernameForApiKey($apiKey)
{
// Look up the username based on the token in the database, via
2. http://api.symfony.com/master/Symfony/Component/Security/Core/Exception/BadCredentialsException.html
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 358
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44 }
// an API call, or do something entirely different
$username = ...;
return $username;
}
public function loadUserByUsername($username)
{
return new User(
$username,
null,
// the roles for the user - you may choose to determine
// these dynamically somehow based on the user
array('ROLE_USER')
);
}
public function refreshUser(UserInterface $user)
{
// this is used for storing authentication in the session
// but in this example, the token is sent in each request,
// so authentication can be stateless. Throwing this exception
// is proper to make things stateless
throw new UnsupportedUserException();
}
public function supportsClass($class)
{
return 'Symfony\Component\Security\Core\User\User' === $class;
}
Read the dedicated article to learn how to create a custom user provider.
The logic inside getUsernameForApiKey() is up to you. You may somehow transform the API key (e.g.
37b51d) into a username (e.g. jondoe) by looking up some information in a "token" database table.
The same is true for loadUserByUsername(). In this example, Symfony's core User3 class is simply
created. This makes sense if you don't need to store any extra information on your User object (e.g.
firstName). But if you do, you may instead have your own user class which you create and populate here
by querying a database. This would allow you to have custom data on the User object.
Finally, just make sure that supportsClass() returns true for User objects with the same class as
whatever user you return in loadUserByUsername(). If your authentication is stateless like in this
example (i.e. you expect the user to send the API key with every request and so you don't save the login to
the session), then you can simply throw the UnsupportedUserException exception in refreshUser().
If you do want to store authentication data in the session so that the key doesn't need to be sent on
every request, see Storing Authentication in the Session.
3. http://api.symfony.com/master/Symfony/Component/Security/Core/User/User.html
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 359
Handling Authentication Failure
In order for your ApiKeyAuthentication to correctly display a 403 http status when either bad
credentials
or
authentication
fails
you
will
need
to
implement
the
AuthenticationFailureHandlerInterface4 on your Authenticator. This will provide a method
onAuthenticationFailure which you can use to create an error Response.
Listing 98-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/Acme/HelloBundle/Security/ApiKeyAuthenticator.php
namespace Acme\HelloBundle\Security;
use
use
use
use
use
Symfony\Component\Security\Core\Authentication\SimplePreAuthenticatorInterface;
Symfony\Component\Security\Core\Exception\AuthenticationException;
Symfony\Component\Security\Http\Authentication\AuthenticationFailureHandlerInterface;
Symfony\Component\HttpFoundation\Response;
Symfony\Component\HttpFoundation\Request;
class ApiKeyAuthenticator implements SimplePreAuthenticatorInterface,
AuthenticationFailureHandlerInterface
{
// ...
public function onAuthenticationFailure(Request $request, AuthenticationException
$exception)
{
return new Response("Authentication Failed.", 403);
}
}
Configuration
Once you have your ApiKeyAuthentication all setup, you need to register it as a service and use it in
your security configuration (e.g. security.yml). First, register it as a service. This assumes that you have
already setup your custom user provider as a service called your_api_key_user_provider (see How to
Create a custom User Provider).
Listing 98-4
1 # app/config/config.yml
2 services:
3
# ...
4
5
apikey_authenticator:
6
class:
Acme\HelloBundle\Security\ApiKeyAuthenticator
7
arguments: ["@your_api_key_user_provider"]
Now, activate it in the firewalls section of your security configuration using the simple_preauth key:
Listing 98-5
1 # app/config/security.yml
2 security:
3
# ...
4
5
firewalls:
6
secured_area:
7
pattern: ^/admin
4. http://api.symfony.com/master/Symfony/Component/Security/Http/Authentication/AuthenticationFailureHandlerInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 360
8
9
10
stateless: true
simple_preauth:
authenticator: apikey_authenticator
That's it! Now, your ApiKeyAuthentication should be called at the beginning of each request and your
authentication process will take place.
The stateless configuration parameter prevents Symfony from trying to store the authentication
information in the session, which isn't necessary since the client will send the apikey on each request. If
you do need to store authentication in the session, keep reading!
Storing Authentication in the Session
So far, this entry has described a situation where some sort of authentication token is sent on every
request. But in some situations (like an OAuth flow), the token may be sent on only one request. In this
case, you will want to authenticate the user and store that authentication in the session so that the user
is automatically logged in for every subsequent request.
To make this work, first remove the stateless key from your firewall configuration or set it to false:
Listing 98-6
1 # app/config/security.yml
2 security:
3
# ...
4
5
firewalls:
6
secured_area:
7
pattern: ^/admin
8
stateless: false
9
simple_preauth:
10
authenticator: apikey_authenticator
Even though the token is being stored in the session, the credentials - in this case the API key (i.e. $token>getCredentials()) - are not stored in the session for security reasons. To take advantage of the session,
update ApiKeyAuthenticator to see if the stored token has a valid User object that can be used:
Listing 98-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/Acme/HelloBundle/Security/ApiKeyAuthenticator.php
// ...
class ApiKeyAuthenticator implements SimplePreAuthenticatorInterface
{
// ...
public function authenticateToken(TokenInterface $token, UserProviderInterface
$userProvider, $providerKey)
{
$apiKey = $token->getCredentials();
$username = $this->userProvider->getUsernameForApiKey($apiKey);
// User is the Entity which represents your user
$user = $token->getUser();
if ($user instanceof User) {
return new PreAuthenticatedToken(
$user,
$apiKey,
$providerKey,
$user->getRoles()
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 361
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
);
}
if (!$username) {
throw new AuthenticationException(
sprintf('API Key "%s" does not exist.', $apiKey)
);
}
$user = $this->userProvider->loadUserByUsername($username);
return new PreAuthenticatedToken(
$user,
$apiKey,
$providerKey,
$user->getRoles()
);
}
// ...
}
Storing authentication information in the session works like this:
1. At the end of each request, Symfony serializes the token object (returned from
authenticateToken()), which also serializes the User object (since it's set on a property on the
token);
2. On the next request the token is deserialized and the deserialized User object is passed to the
refreshUser() function of the user provider.
The second step is the important one: Symfony calls refreshUser() and passes you the user object that
was serialized in the session. If your users are stored in the database, then you may want to re-query
for a fresh version of the user to make sure it's not out-of-date. But regardless of your requirements,
refreshUser() should now return the User object:
Listing 98-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/HelloBundle/Security/ApiKeyUserProvider.php
// ...
class ApiKeyUserProvider implements UserProviderInterface
{
// ...
public function refreshUser(UserInterface $user)
{
// $user is the User that you set in the token inside authenticateToken()
// after it has been deserialized from the session
// you might use $user to query the database for a fresh user
// $id = $user->getId();
// use $id to make a query
// if you are *not* reading from a database and are just creating
// a User object (like in this example), you can just return it
return $user;
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 362
You'll also want to make sure that your User object is being serialized correctly. If your User object
has private properties, PHP can't serialize those. In this case, you may get back a User object that
has a null value for each property. For an example, see How to Load Security Users from the
Database (the Entity Provider).
Only Authenticating for Certain URLs
This entry has assumed that you want to look for the apikey authentication on every request. But in
some situations (like an OAuth flow), you only really need to look for authentication information once
the user has reached a certain URL (e.g. the redirect URL in OAuth).
Fortunately, handling this situation is easy: just check to see what the current URL is before creating the
token in createToken():
Listing 98-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/Acme/HelloBundle/Security/ApiKeyAuthenticator.php
// ...
use Symfony\Component\Security\Http\HttpUtils;
use Symfony\Component\HttpFoundation\Request;
class ApiKeyAuthenticator implements SimplePreAuthenticatorInterface
{
protected $userProvider;
protected $httpUtils;
public function __construct(ApiKeyUserProviderInterface $userProvider, HttpUtils
$httpUtils)
{
$this->userProvider = $userProvider;
$this->httpUtils = $httpUtils;
}
public function createToken(Request $request, $providerKey)
{
// set the only URL where we should look for auth information
// and only return the token if we're at that URL
$targetUrl = '/login/check';
if (!$this->httpUtils->checkRequestPath($request, $targetUrl)) {
return;
}
// ...
}
}
This uses the handy HttpUtils5 class to check if the current URL matches the URL you're looking for.
In this case, the URL (/login/check) has been hardcoded in the class, but you could also inject it as the
third constructor argument.
Next, just update your service configuration to inject the security.http_utils service:
Listing 98-10
5. http://api.symfony.com/master/Symfony/Component/Security/Http/HttpUtils.html
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 363
1 # app/config/config.yml
2 services:
3
# ...
4
5
apikey_authenticator:
6
class:
Acme\HelloBundle\Security\ApiKeyAuthenticator
7
arguments: ["@your_api_key_user_provider", "@security.http_utils"]
That's it! Have fun!
PDF brought to you by
generated on December 3, 2014
Chapter 98: How to Authenticate Users with API Keys | 364
Chapter 99
How to Create a custom Authentication
Provider
Creating a custom authentication system is hard, and this entry will walk you through that process.
But depending on your needs, you may be able to solve your problem in a simpler way using these
documents:
• How to Create a Custom Form Password Authenticator
• How to Authenticate Users with API Keys
If you have read the chapter on Security, you understand the distinction Symfony makes between
authentication and authorization in the implementation of security. This chapter discusses the core
classes involved in the authentication process, and how to implement a custom authentication provider.
Because authentication and authorization are separate concepts, this extension will be user-provider
agnostic, and will function with your application's user providers, may they be based in memory, a
database, or wherever else you choose to store them.
Meet WSSE
The following chapter demonstrates how to create a custom authentication provider for WSSE
authentication. The security protocol for WSSE provides several security benefits:
1. Username / Password encryption
2. Safe guarding against replay attacks
3. No web server configuration required
WSSE is very useful for the securing of web services, may they be SOAP or REST.
There is plenty of great documentation on WSSE1, but this article will focus not on the security protocol,
but rather the manner in which a custom protocol can be added to your Symfony application. The basis
1. http://www.xml.com/pub/a/2003/12/17/dive.html
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 365
of WSSE is that a request header is checked for encrypted credentials, verified using a timestamp and
nonce2, and authenticated for the requested user using a password digest.
WSSE also supports application key validation, which is useful for web services, but is outside the
scope of this chapter.
The Token
The role of the token in the Symfony security context is an important one. A token represents the user
authentication data present in the request. Once a request is authenticated, the token retains the user's
data, and delivers this data across the security context. First, you'll create your token class. This will allow
the passing of all relevant information to your authentication provider.
Listing 99-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
// src/Acme/DemoBundle/Security/Authentication/Token/WsseUserToken.php
namespace Acme\DemoBundle\Security\Authentication\Token;
use Symfony\Component\Security\Core\Authentication\Token\AbstractToken;
class WsseUserToken extends AbstractToken
{
public $created;
public $digest;
public $nonce;
public function __construct(array $roles = array())
{
parent::__construct($roles);
// If the user has roles, consider it authenticated
$this->setAuthenticated(count($roles) > 0);
}
public function getCredentials()
{
return '';
}
}
The WsseUserToken class extends the Security component's AbstractToken3 class, which provides
basic token functionality. Implement the TokenInterface4 on any class to use as a token.
The Listener
Next, you need a listener to listen on the security context. The listener is responsible for fielding
requests to the firewall and calling the authentication provider. A listener must be an instance of
2. http://en.wikipedia.org/wiki/Cryptographic_nonce
3. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/Token/AbstractToken.html
4. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/Token/TokenInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 366
ListenerInterface5. A security listener should handle the GetResponseEvent6 event, and set an
authenticated token in the security context if successful.
Listing 99-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
// src/Acme/DemoBundle/Security/Firewall/WsseListener.php
namespace Acme\DemoBundle\Security\Firewall;
use
use
use
use
use
use
use
Symfony\Component\HttpFoundation\Response;
Symfony\Component\HttpKernel\Event\GetResponseEvent;
Symfony\Component\Security\Http\Firewall\ListenerInterface;
Symfony\Component\Security\Core\Exception\AuthenticationException;
Symfony\Component\Security\Core\SecurityContextInterface;
Symfony\Component\Security\Core\Authentication\AuthenticationManagerInterface;
Acme\DemoBundle\Security\Authentication\Token\WsseUserToken;
class WsseListener implements ListenerInterface
{
protected $securityContext;
protected $authenticationManager;
public function __construct(SecurityContextInterface $securityContext,
AuthenticationManagerInterface $authenticationManager)
{
$this->securityContext = $securityContext;
$this->authenticationManager = $authenticationManager;
}
public function handle(GetResponseEvent $event)
{
$request = $event->getRequest();
$wsseRegex = '/UsernameToken Username="([^"]+)", PasswordDigest="([^"]+)",
Nonce="([^"]+)", Created="([^"]+)"/';
if (!$request->headers->has('x-wsse') || 1 !== preg_match($wsseRegex,
$request->headers->get('x-wsse'), $matches)) {
return;
}
$token = new WsseUserToken();
$token->setUser($matches[1]);
$token->digest
$token->nonce
$token->created
= $matches[2];
= $matches[3];
= $matches[4];
try {
$authToken = $this->authenticationManager->authenticate($token);
$this->securityContext->setToken($authToken);
return;
} catch (AuthenticationException $failed) {
// ... you might log something here
// To deny the authentication clear the token. This will redirect to the login
page.
// Make sure to only clear your token, not those of other authentication
listeners.
// $token = $this->securityContext->getToken();
5. http://api.symfony.com/master/Symfony/Component/Security/Http/Firewall/ListenerInterface.html
6. http://api.symfony.com/master/Symfony/Component/HttpKernel/Event/GetResponseEvent.html
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 367
55
// if ($token instanceof WsseUserToken && $this->providerKey ===
56 $token->getProviderKey()) {
57
//
$this->securityContext->setToken(null);
58
// }
59
// return;
60
}
61
// By default deny authorization
$response = new Response();
$response->setStatusCode(Response::HTTP_FORBIDDEN);
$event->setResponse($response);
}
}
New in version 2.4: Support for HTTP status code constants was introduced in Symfony 2.4.
This listener checks the request for the expected X-WSSE header, matches the value returned for the
expected WSSE information, creates a token using that information, and passes the token on to the
authentication manager. If the proper information is not provided, or the authentication manager throws
an AuthenticationException7, a 403 Response is returned.
A class not used above, the AbstractAuthenticationListener8 class, is a very useful base class
which provides commonly needed functionality for security extensions. This includes maintaining
the token in the session, providing success / failure handlers, login form URLs, and more. As
WSSE does not require maintaining authentication sessions or login forms, it won't be used for this
example.
Returning prematurely from the listener is relevant only if you want to chain authentication
providers (for example to allow anonymous users). If you want to forbid access to anonymous
users and have a nice 403 error, you should set the status code of the response before returning.
The Authentication Provider
The authentication provider will do the verification of the WsseUserToken. Namely, the provider will
verify the Created header value is valid within five minutes, the Nonce header value is unique within five
minutes, and the PasswordDigest header value matches with the user's password.
Listing 99-3
1
2
3
4
5
6
7
8
// src/Acme/DemoBundle/Security/Authentication/Provider/WsseProvider.php
namespace Acme\DemoBundle\Security\Authentication\Provider;
use
Symfony\Component\Security\Core\Authentication\Provider\AuthenticationProviderInterface;
use Symfony\Component\Security\Core\User\UserProviderInterface;
use Symfony\Component\Security\Core\Exception\AuthenticationException;
use Symfony\Component\Security\Core\Exception\NonceExpiredException;
7. http://api.symfony.com/master/Symfony/Component/Security/Core/Exception/AuthenticationException.html
8. http://api.symfony.com/master/Symfony/Component/Security/Http/Firewall/AbstractAuthenticationListener.html
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 368
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
use Symfony\Component\Security\Core\Authentication\Token\TokenInterface;
use Acme\DemoBundle\Security\Authentication\Token\WsseUserToken;
class WsseProvider implements AuthenticationProviderInterface
{
private $userProvider;
private $cacheDir;
public function __construct(UserProviderInterface $userProvider, $cacheDir)
{
$this->userProvider = $userProvider;
$this->cacheDir
= $cacheDir;
}
public function authenticate(TokenInterface $token)
{
$user = $this->userProvider->loadUserByUsername($token->getUsername());
if ($user && $this->validateDigest($token->digest, $token->nonce, $token->created,
$user->getPassword())) {
$authenticatedToken = new WsseUserToken($user->getRoles());
$authenticatedToken->setUser($user);
return $authenticatedToken;
}
throw new AuthenticationException('The WSSE authentication failed.');
}
/**
* This function is specific to Wsse authentication and is only used to help this
example
*
* For more information specific to the logic here, see
* https://github.com/symfony/symfony-docs/pull/3134#issuecomment-27699129
*/
protected function validateDigest($digest, $nonce, $created, $secret)
{
// Check created time is not in the future
if (strtotime($created) > time()) {
return false;
}
// Expire timestamp after 5 minutes
if (time() - strtotime($created) > 300) {
return false;
}
// Validate that the nonce is *not* used in the last 5 minutes
// if it has, this could be a replay attack
if (file_exists($this->cacheDir.'/'.$nonce) &&
file_get_contents($this->cacheDir.'/'.$nonce) + 300 > time()) {
throw new NonceExpiredException('Previously used nonce detected');
}
// If cache directory does not exist we create it
if (!is_dir($this->cacheDir)) {
mkdir($this->cacheDir, 0777, true);
}
file_put_contents($this->cacheDir.'/'.$nonce, time());
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 369
68
69
70
71
72
73
74
75
// Validate Secret
$expected = base64_encode(sha1(base64_decode($nonce).$created.$secret, true));
return $digest === $expected;
}
public function supports(TokenInterface $token)
{
return $token instanceof WsseUserToken;
}
}
The AuthenticationProviderInterface9 requires an authenticate method on the user token,
and a supports method, which tells the authentication manager whether or not to use this
provider for the given token. In the case of multiple providers, the authentication manager will
then move to the next provider in the list.
The Factory
You have created a custom token, custom listener, and custom provider. Now you need to tie them all
together. How do you make a unique provider available for every firewall? The answer is by using a
factory. A factory is where you hook into the Security component, telling it the name of your provider
and any configuration options available for it. First, you must create a class which implements
SecurityFactoryInterface10.
Listing 99-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
// src/Acme/DemoBundle/DependencyInjection/Security/Factory/WsseFactory.php
namespace Acme\DemoBundle\DependencyInjection\Security\Factory;
use Symfony\Component\DependencyInjection\ContainerBuilder;
use Symfony\Component\DependencyInjection\Reference;
use Symfony\Component\DependencyInjection\DefinitionDecorator;
use Symfony\Component\Config\Definition\Builder\NodeDefinition;
use
Symfony\Bundle\SecurityBundle\DependencyInjection\Security\Factory\SecurityFactoryInterface;
class WsseFactory implements SecurityFactoryInterface
{
public function create(ContainerBuilder $container, $id, $config, $userProvider,
$defaultEntryPoint)
{
$providerId = 'security.authentication.provider.wsse.'.$id;
$container
->setDefinition($providerId, new
DefinitionDecorator('wsse.security.authentication.provider'))
->replaceArgument(0, new Reference($userProvider))
;
$listenerId = 'security.authentication.listener.wsse.'.$id;
$listener = $container->setDefinition($listenerId, new
DefinitionDecorator('wsse.security.authentication.listener'));
9. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/Provider/AuthenticationProviderInterface.html
10. http://api.symfony.com/master/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/SecurityFactoryInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 370
26
27
28
29
30
31
32
33
34
35
36
37
38
39
return array($providerId, $listenerId, $defaultEntryPoint);
}
public function getPosition()
{
return 'pre_auth';
}
public function getKey()
{
return 'wsse';
}
public function addConfiguration(NodeDefinition $node)
{
}
}
The SecurityFactoryInterface11 requires the following methods:
• create method, which adds the listener and authentication provider to the DI container for
the appropriate security context;
• getPosition method, which must be of type pre_auth, form, http, and remember_me and
defines the position at which the provider is called;
• getKey method which defines the configuration key used to reference the provider in the
firewall configuration;
• addConfiguration method, which is used to define the configuration options underneath the
configuration key in your security configuration. Setting configuration options are explained
later in this chapter.
A class not used in this example, AbstractFactory12, is a very useful base class which provides
commonly needed functionality for security factories. It may be useful when defining an
authentication provider of a different type.
Now that you have created a factory class, the wsse key can be used as a firewall in your security
configuration.
You may be wondering "why do you need a special factory class to add listeners and providers to
the dependency injection container?". This is a very good question. The reason is you can use your
firewall multiple times, to secure multiple parts of your application. Because of this, each time your
firewall is used, a new service is created in the DI container. The factory is what creates these new
services.
Configuration
It's time to see your authentication provider in action. You will need to do a few things in order to
make this work. The first thing is to add the services above to the DI container. Your factory class above
makes reference to service ids that do not exist yet: wsse.security.authentication.provider and
wsse.security.authentication.listener. It's time to define those services.
11. http://api.symfony.com/master/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/SecurityFactoryInterface.html
12. http://api.symfony.com/master/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/AbstractFactory.html
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 371
Listing 99-5
1 # src/Acme/DemoBundle/Resources/config/services.yml
2 services:
3
wsse.security.authentication.provider:
4
class: Acme\DemoBundle\Security\Authentication\Provider\WsseProvider
5
arguments: ["", "%kernel.cache_dir%/security/nonces"]
6
7
wsse.security.authentication.listener:
8
class: Acme\DemoBundle\Security\Firewall\WsseListener
9
arguments: ["@security.context", "@security.authentication.manager"]
Now that your services are defined, tell your security context about your factory in your bundle class:
Listing 99-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/DemoBundle/AcmeDemoBundle.php
namespace Acme\DemoBundle;
use Acme\DemoBundle\DependencyInjection\Security\Factory\WsseFactory;
use Symfony\Component\HttpKernel\Bundle\Bundle;
use Symfony\Component\DependencyInjection\ContainerBuilder;
class AcmeDemoBundle extends Bundle
{
public function build(ContainerBuilder $container)
{
parent::build($container);
$extension = $container->getExtension('security');
$extension->addSecurityListenerFactory(new WsseFactory());
}
}
You are finished! You can now define parts of your app as under WSSE protection.
Listing 99-7
1 security:
2
firewalls:
3
wsse_secured:
4
pattern:
/api/.*
5
stateless: true
6
wsse:
true
Congratulations! You have written your very own custom security authentication provider!
A little Extra
How about making your WSSE authentication provider a bit more exciting? The possibilities are endless.
Why don't you start by adding some sparkle to that shine?
Configuration
You can add custom options under the wsse key in your security configuration. For instance, the time
allowed before expiring the Created header item, by default, is 5 minutes. Make this configurable, so
different firewalls can have different timeout lengths.
You will first need to edit WsseFactory and define the new option in the addConfiguration method.
Listing 99-8
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 372
1 class WsseFactory implements SecurityFactoryInterface
2 {
3
// ...
4
5
public function addConfiguration(NodeDefinition $node)
6
{
7
$node
8
->children()
9
->scalarNode('lifetime')->defaultValue(300)
10
->end();
11
}
12 }
Now, in the create method of the factory, the $config argument will contain a lifetime key, set
to 5 minutes (300 seconds) unless otherwise set in the configuration. Pass this argument to your
authentication provider in order to put it to use.
Listing 99-9
1 class WsseFactory implements SecurityFactoryInterface
2 {
3
public function create(ContainerBuilder $container, $id, $config, $userProvider,
4 $defaultEntryPoint)
5
{
6
$providerId = 'security.authentication.provider.wsse.'.$id;
7
$container
8
->setDefinition($providerId,
9
new DefinitionDecorator('wsse.security.authentication.provider'))
10
->replaceArgument(0, new Reference($userProvider))
11
->replaceArgument(2, $config['lifetime']);
12
// ...
13
}
14
15
// ...
}
You'll also need to add a third argument to the wsse.security.authentication.provider
service configuration, which can be blank, but will be filled in with the lifetime in the factory. The
WsseProvider class will also now need to accept a third constructor argument - the lifetime - which
it should use instead of the hard-coded 300 seconds. These two steps are not shown here.
The lifetime of each WSSE request is now configurable, and can be set to any desirable value per firewall.
Listing 99-10
1 security:
2
firewalls:
3
wsse_secured:
4
pattern:
/api/.*
5
stateless: true
6
wsse:
{ lifetime: 30 }
The rest is up to you! Any relevant configuration items can be defined in the factory and consumed or
passed to the other classes in the container.
PDF brought to you by
generated on December 3, 2014
Chapter 99: How to Create a custom Authentication Provider | 373
Chapter 100
Using pre Authenticated Security Firewalls
A lot of authentication modules are already provided by some web servers, including Apache. These
modules generally set some environment variables that can be used to determine which user is accessing
your application. Out of the box, Symfony supports most authentication mechanisms. These requests
are called pre authenticated requests because the user is already authenticated when reaching your
application.
X.509 Client Certificate Authentication
When using client certificates, your webserver is doing all the authentication process itself. With Apache,
for example, you would use the SSLVerifyClient Require directive.
Enable the x509 authentication for a particular firewall in the security configuration:
Listing 100-1
1 # app/config/security.yml
2 security:
3
firewalls:
4
secured_area:
5
pattern: ^/
6
x509:
7
provider: your_user_provider
By default, the firewall provides the SSL_CLIENT_S_DN_Email variable to the user provider, and sets the
SSL_CLIENT_S_DN as credentials in the PreAuthenticatedToken1. You can override these by setting the
user and the credentials keys in the x509 firewall configuration respectively.
1. http://api.symfony.com/master/Symfony/Component/Security/Core/Authentication/Token/PreAuthenticatedToken.html
PDF brought to you by
generated on December 3, 2014
Chapter 100: Using pre Authenticated Security Firewalls | 374
An authentication provider will only inform the user provider of the username that made the
request. You will need to create (or use) a "user provider" that is referenced by the provider
configuration parameter (your_user_provider in the configuration example). This provider will
turn the username into a User object of your choice. For more information on creating or
configuring a user provider, see:
• How to Create a custom User Provider
• How to Load Security Users from the Database (the Entity Provider)
REMOTE_USER Based Authentication
New in version 2.6: REMOTE_USER pre authenticated firewall was introduced in Symfony 2.6.
A lot of authentication modules, like auth_kerb for Apache provide the username using the REMOTE_USER
environment variable. This variable can be trusted by the application since the authentication happened
before the request reached it.
To configure Symfony using the REMOTE_USER environment variable, simply enable the corresponding
firewall in your security configuration:
Listing 100-2
1 # app/config/security.yml
2 security:
3
firewalls:
4
secured_area:
5
pattern: ^/
6
remote_user:
7
provider: your_user_provider
The firewall will then provide the REMOTE_USER environment variable to your user provider. You can
change the variable name used by setting the user key in the remote_user firewall configuration.
Just like for X509 authentication, you will need to configure a "user provider". See the note previous
note for more information.
PDF brought to you by
generated on December 3, 2014
Chapter 100: Using pre Authenticated Security Firewalls | 375
Chapter 101
How to Change the default Target Path
Behavior
By default, the Security component retains the information of the last request URI in a session variable
named _security.main.target_path (with main being the name of the firewall, defined in
security.yml). Upon a successful login, the user is redirected to this path, as to help them continue from
the last known page they visited.
In some situations, this is not ideal. For example, when the last request URI was an XMLHttpRequest
which returned a non-HTML or partial HTML response, the user is redirected back to a page which the
browser cannot render.
To get around this behavior, you would simply need to extend the ExceptionListener class and override
the default method named setTargetPath().
First, override the security.exception_listener.class parameter in your configuration file. This can
be done from your main configuration file (in app/config) or from a configuration file being imported
from a bundle:
Listing 101-1
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 parameters:
3
# ...
4
security.exception_listener.class: Acme\HelloBundle\Security\Firewall\ExceptionListener
Next, create your own ExceptionListener:
Listing 101-2
1
2
3
4
5
6
7
8
9
10
// src/Acme/HelloBundle/Security/Firewall/ExceptionListener.php
namespace Acme\HelloBundle\Security\Firewall;
use Symfony\Component\HttpFoundation\Request;
use Symfony\Component\Security\Http\Firewall\ExceptionListener as BaseExceptionListener;
class ExceptionListener extends BaseExceptionListener
{
protected function setTargetPath(Request $request)
{
PDF brought to you by
generated on December 3, 2014
Chapter 101: How to Change the default Target Path Behavior | 376
11
12
13
14
15
16
17
18
19
20 }
//
//
//
if
Do not save target path for XHR requests
You can add any more logic here you want
Note that non-GET requests are already ignored
($request->isXmlHttpRequest()) {
return;
}
parent::setTargetPath($request);
}
Add as much or as little logic here as required for your scenario!
PDF brought to you by
generated on December 3, 2014
Chapter 101: How to Change the default Target Path Behavior | 377
Chapter 102
Using CSRF Protection in the Login Form
When using a login form, you should make sure that you are protected against CSRF (Cross-site request
forgery1). The Security component already has built-in support for CSRF. In this article you'll learn how
you can use it in your login form.
Login CSRF attacks are a bit less well-known. See Forging Login Requests2 if you're curious about
more details.
Configuring CSRF Protection
First, configure the Security component so it can use CSRF protection. The Security component needs a
CSRF token provider. You can set this to use the default provider available in the Form component:
Listing 102-1
1 # app/config/security.yml
2 security:
3
firewalls:
4
secured_area:
5
# ...
6
form_login:
7
# ...
8
csrf_provider: form.csrf_provider
The Security component can be configured further, but this is all information it needs to be able to use
CSRF in the login form.
1. http://en.wikipedia.org/wiki/Cross-site_request_forgery
2. http://en.wikipedia.org/wiki/Cross-site_request_forgery#Forging_login_requests
PDF brought to you by
generated on December 3, 2014
Chapter 102: Using CSRF Protection in the Login Form | 378
Rendering the CSRF field
Now that Security component will check for the CSRF token, you have to add a hidden field to the
login form containing the CSRF token. By default, this field is named _csrf_token. That hidden field
must contain the CSRF token, which can be generated by using the csrf_token function. That function
requires a token ID, which must be set to authenticate when using the login form:
Listing 102-2
1
2
3
4
5
6
7
8
9
10
11
12
{# src/Acme/SecurityBundle/Resources/views/Security/login.html.twig #}
{# ... #}
<form action="{{ path('login_check') }}" method="post">
{# ... the login fields #}
<input type="hidden" name="_csrf_token"
value="{{ csrf_token('authenticate') }}"
>
<button type="submit">login</button>
</form>
After this, you have protected your login form against CSRF attacks.
You can change the name of the field by setting csrf_parameter and change the token ID by
setting intention in your configuration:
Listing 102-3
1 # app/config/security.yml
2 security:
3
firewalls:
4
secured_area:
5
# ...
6
form_login:
7
# ...
8
csrf_parameter: _csrf_security_token
9
intention: a_private_string
PDF brought to you by
generated on December 3, 2014
Chapter 102: Using CSRF Protection in the Login Form | 379
Chapter 103
How to Choose the Password Encoder
Algorithm Dynamically
New in version 2.5: Named encoders were introduced in Symfony 2.5.
Usually, the same password encoder is used for all users by configuring it to apply to all instances of a
specific class:
Listing 103-1
1 # app/config/security.yml
2 security:
3
# ...
4
encoders:
5
Symfony\Component\Security\Core\User\User: sha512
Another option is to use a "named" encoder and then select which encoder you want to use dynamically.
In the previous example, you've set the sha512 algorithm for Acme\UserBundle\Entity\User. This may
be secure enough for a regular user, but what if you want your admins to have a stronger algorithm, for
example bcrypt. This can be done with named encoders:
Listing 103-2
1 # app/config/security.yml
2 security:
3
# ...
4
encoders:
5
harsh:
6
algorithm: bcrypt
7
cost: 15
PDF brought to you by
generated on December 3, 2014
Chapter 103: How to Choose the Password Encoder Algorithm Dynamically | 380
This creates an encoder named harsh. In order for a User instance to use it, the class must implement
EncoderAwareInterface1. The interface requires one method - getEncoderName - which should return
the name of the encoder to use:
Listing 103-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/UserBundle/Entity/User.php
namespace Acme\UserBundle\Entity;
use Symfony\Component\Security\Core\User\UserInterface;
use Symfony\Component\Security\Core\Encoder\EncoderAwareInterface;
class User implements UserInterface, EncoderAwareInterface
{
public function getEncoderName()
{
if ($this->isAdmin()) {
return 'harsh';
}
return null; // use the default encoder
}
}
1. http://api.symfony.com/master/Symfony/Component/Security/Core/Encoder/EncoderAwareInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 103: How to Choose the Password Encoder Algorithm Dynamically | 381
Chapter 104
How to Use the Serializer
Serializing and deserializing to and from objects and different formats (e.g. JSON or XML) is a very
complex topic. Symfony comes with a Serializer Component, which gives you some tools that you can
leverage for your solution.
In fact, before you start, get familiar with the serializer, normalizers and encoders by reading the Serializer
Component. You should also check out the JMSSerializerBundle1, which expands on the functionality
offered by Symfony's core serializer.
Activating the Serializer
New in version 2.3: The Serializer has always existed in Symfony, but prior to Symfony 2.3, you
needed to build the serializer service yourself.
The serializer service is not available by default. To turn it on, activate it in your configuration:
Listing 104-1
1 # app/config/config.yml
2 framework:
3
# ...
4
serializer:
5
enabled: true
Adding Normalizers and Encoders
Once enabled, the serializer service will be available in the container and will be loaded with two
encoders (JsonEncoder2 and XmlEncoder3) but no normalizers, meaning you'll need to load your own.
1. http://jmsyst.com/bundles/JMSSerializerBundle
2. http://api.symfony.com/master/Symfony/Component/Serializer/Encoder/JsonEncoder.html
3. http://api.symfony.com/master/Symfony/Component/Serializer/Encoder/XmlEncoder.html
PDF brought to you by
generated on December 3, 2014
Chapter 104: How to Use the Serializer | 382
You can load normalizers and/or encoders by tagging them as serializer.normalizer and serializer.encoder.
It's also possible to set the priority of the tag in order to decide the matching order.
Here is an example on how to load the GetSetMethodNormalizer4:
Listing 104-2
1 # app/config/config.yml
2 services:
3
get_set_method_normalizer:
4
class: Symfony\Component\Serializer\Normalizer\GetSetMethodNormalizer
5
tags:
6
- { name: serializer.normalizer }
The GetSetMethodNormalizer5 is broken by design. As soon as you have a circular object graph,
an infinite loop is created when calling the getters. You're encouraged to add your own normalizers
that fit your use-case.
4. http://api.symfony.com/master/Symfony/Component/Serializer/Normalizer/GetSetMethodNormalizer.html
5. http://api.symfony.com/master/Symfony/Component/Serializer/Normalizer/GetSetMethodNormalizer.html
PDF brought to you by
generated on December 3, 2014
Chapter 104: How to Use the Serializer | 383
Chapter 105
How to Create an Event Listener
Symfony has various events and hooks that can be used to trigger custom behavior in your application.
Those events are thrown by the HttpKernel component and can be viewed in the KernelEvents1 class.
To hook into an event and add your own custom logic, you have to create a service that will act as an
event listener on that event. In this entry, you will create a service that will act as an Exception Listener,
allowing you to modify how exceptions are shown by your application. The KernelEvents::EXCEPTION
event is just one of the core kernel events:
Listing 105-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/Acme/DemoBundle/EventListener/AcmeExceptionListener.php
namespace Acme\DemoBundle\EventListener;
use Symfony\Component\HttpKernel\Event\GetResponseForExceptionEvent;
use Symfony\Component\HttpFoundation\Response;
use Symfony\Component\HttpKernel\Exception\HttpExceptionInterface;
class AcmeExceptionListener
{
public function onKernelException(GetResponseForExceptionEvent $event)
{
// You get the exception object from the received event
$exception = $event->getException();
$message = sprintf(
'My Error says: %s with code: %s',
$exception->getMessage(),
$exception->getCode()
);
// Customize your response object to display the exception details
$response = new Response();
$response->setContent($message);
// HttpExceptionInterface is a special type of exception that
// holds status code and header details
if ($exception instanceof HttpExceptionInterface) {
1. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelEvents.html
PDF brought to you by
generated on December 3, 2014
Chapter 105: How to Create an Event Listener | 384
27
28
29
30
31
32
33
34
35
36 }
$response->setStatusCode($exception->getStatusCode());
$response->headers->replace($exception->getHeaders());
} else {
$response->setStatusCode(Response::HTTP_INTERNAL_SERVER_ERROR);
}
// Send the modified response object to the event
$event->setResponse($response);
}
New in version 2.4: Support for HTTP status code constants was introduced in Symfony 2.4.
Each event receives a slightly different type of $event object. For the kernel.exception event, it
is GetResponseForExceptionEvent2. To see what type of object each event listener receives, see
KernelEvents3.
Now that the class is created, you just need to register it as a service and notify Symfony that it is a
"listener" on the kernel.exception event by using a special "tag":
Listing 105-2
1 # app/config/config.yml
2 services:
3
kernel.listener.your_listener_name:
4
class: Acme\DemoBundle\EventListener\AcmeExceptionListener
5
tags:
6
- { name: kernel.event_listener, event: kernel.exception, method:
onKernelException }
There is an additional tag option priority that is optional and defaults to 0. This value can be
from -255 to 255, and the listeners will be executed in the order of their priority (highest to lowest).
This is useful when you need to guarantee that one listener is executed before another.
Request Events, Checking Types
New in version 2.4: The isMasterRequest() method was introduced in Symfony 2.4. Prior, the
getRequestType() method must be used.
A single page can make several requests (one master request, and then multiple sub-requests), which is
why when working with the KernelEvents::REQUEST event, you might need to check the type of the
request. This can be easily done as follow:
Listing 105-3
2. http://api.symfony.com/master/Symfony/Component/HttpKernel/Event/GetResponseForExceptionEvent.html
3. http://api.symfony.com/master/Symfony/Component/HttpKernel/KernelEvents.html
PDF brought to you by
generated on December 3, 2014
Chapter 105: How to Create an Event Listener | 385
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/Acme/DemoBundle/EventListener/AcmeRequestListener.php
namespace Acme\DemoBundle\EventListener;
use Symfony\Component\HttpKernel\Event\GetResponseEvent;
use Symfony\Component\HttpKernel\HttpKernel;
class AcmeRequestListener
{
public function onKernelRequest(GetResponseEvent $event)
{
if (!$event->isMasterRequest()) {
// don't do anything if it's not the master request
return;
}
// ...
}
}
Two types of request are available in the HttpKernelInterface4 interface:
HttpKernelInterface::MASTER_REQUEST and HttpKernelInterface::SUB_REQUEST.
Debugging Event Listeners
You can find out what listeners are registered in the event dispatcher using the console. To show all
events and their listeners, run:
Listing 105-4
1 $ php app/console debug:event-dispatcher
You can get registered listeners for a particular event by specifying its name:
Listing 105-5
1 $ php app/console debug:event-dispatcher kernel.exception
4. http://api.symfony.com/master/Symfony/Component/HttpKernel/HttpKernelInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 105: How to Create an Event Listener | 386
Chapter 106
How to Work with Scopes
This entry is all about scopes, a somewhat advanced topic related to the Service Container. If you've ever
gotten an error mentioning "scopes" when creating services, then this entry is for you.
If you are trying to inject the request service, the simple solution is to inject the request_stack
service instead and access the current Request by calling the getCurrentRequest()1 method (see
Injecting the Request). The rest of this entry talks about scopes in a theoretical and more advanced
way. If you're dealing with scopes for the request service, simply inject request_stack.
Understanding Scopes
The scope of a service controls how long an instance of a service is used by the container. The
DependencyInjection component provides two generic scopes:
• container (the default one): The same instance is used each time you request it from this
container.
• prototype: A new instance is created each time you request the service.
The ContainerAwareHttpKernel2 also defines a third scope: request. This scope is tied to the request,
meaning a new instance is created for each subrequest and is unavailable outside the request (for instance
in the CLI).
An Example: Client Scope
Other than the request service (which has a simple solution, see the above note), no services in the
default Symfony2 container belong to any scope other than container and prototype. But for the
purposes of this entry, imagine there is another scope client and a service client_configuration that
belongs to it. This is not a common situation, but the idea is that you may enter and exit multiple client
scopes during a request, and each has its own client_configuration service.
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/RequestStack.html#getCurrentRequest()
2. http://api.symfony.com/master/Symfony/Component/HttpKernel/DependencyInjection/ContainerAwareHttpKernel.html
PDF brought to you by
generated on December 3, 2014
Chapter 106: How to Work with Scopes | 387
Scopes add a constraint on the dependencies of a service: a service cannot depend on services from
a narrower scope. For example, if you create a generic my_foo service, but try to inject the
client_configuration service, you will receive a ScopeWideningInjectionException3 when
compiling the container. Read the sidebar below for more details.
Scopes and Dependencies
Imagine you've configured a my_mailer service. You haven't configured the scope of the service,
so it defaults to container. In other words, every time you ask the container for the my_mailer
service, you get the same object back. This is usually how you want your services to work.
Imagine, however, that you need the client_configuration service in your my_mailer service,
maybe because you're reading some details from it, such as what the "sender" address should be.
You add it as a constructor argument. There are several reasons why this presents a problem:
• When requesting my_mailer, an instance of my_mailer (called MailerA here) is created
and the client_configuration service ( called ConfigurationA here) is passed to it.
Life is good!
• Your application now needs to do something with another client, and you've
architected your application in such a way that you handle this by entering a new
client_configuration scope and setting a new client_configuration service into
the container. Call this ConfigurationB.
• Somewhere in your application, you once again ask for the my_mailer service. Since
your service is in the container scope, the same instance (MailerA) is just re-used. But
here's the problem: the MailerA instance still contains the old ConfigurationA object,
which is now not the correct configuration object to have (ConfigurationB is now the
current client_configuration service). This is subtle, but the mis-match could cause
major problems, which is why it's not allowed.
So, that's the reason why scopes exist, and how they can cause problems. Keep reading
to find out the common solutions.
A service can of course depend on a service from a wider scope without any issue.
Using a Service from a Narrower Scope
There are several solutions to the scope problem:
• A) Use setter injection if the dependency is synchronized (see A) Using a Synchronized
Service);
• B) Put your service in the same scope as the dependency (or a narrower one). If you depend on
the client_configuration service, this means putting your new service in the client scope
(see B) Changing the Scope of your Service);
• C) Pass the entire container to your service and retrieve your dependency from the container
each time you need it to be sure you have the right instance -- your service can live in the
default container scope (see C) Passing the Container as a Dependency of your Service).
Each scenario is detailed in the following sections.
3. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Exception/ScopeWideningInjectionException.html
PDF brought to you by
generated on December 3, 2014
Chapter 106: How to Work with Scopes | 388
A) Using a Synchronized Service
New in version 2.3: Synchronized services were introduced in Symfony 2.3.
Both injecting the container and setting your service to a narrower scope have drawbacks. Assume first
that the client_configuration service has been marked as synchronized:
Listing 106-1
1 # app/config/config.yml
2 services:
3
client_configuration:
4
class:
Acme\HelloBundle\Client\ClientConfiguration
5
scope:
client
6
synchronized: true
7
synthetic:
true
8
# ...
Now, if you inject this service using setter injection, there are no drawbacks and everything works
without any special code in your service or in your definition:
Listing 106-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/HelloBundle/Mail/Mailer.php
namespace Acme\HelloBundle\Mail;
use Acme\HelloBundle\Client\ClientConfiguration;
class Mailer
{
protected $clientConfiguration;
public function setClientConfiguration(ClientConfiguration $clientConfiguration = null)
{
$this->clientConfiguration = $clientConfiguration;
}
public function sendEmail()
{
if (null === $this->clientConfiguration) {
// throw an error?
}
// ... do something using the client configuration here
}
}
Whenever the client scope is active, the service container will automatically call the
setClientConfiguration() method when the client_configuration service is set in the container.
You might have noticed that the setClientConfiguration() method accepts null as a valid value
for the client_configuration argument. That's because when leaving the client scope, the
client_configuration instance can be null. Of course, you should take care of this possibility in your
code. This should also be taken into account when declaring your service:
Listing 106-3
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 services:
3
my_mailer:
PDF brought to you by
generated on December 3, 2014
Chapter 106: How to Work with Scopes | 389
4
5
6
class: Acme\HelloBundle\Mail\Mailer
calls:
- [setClientConfiguration, ["@?client_configuration="]]
B) Changing the Scope of your Service
Changing the scope of a service should be done in its definition. This example assumes that the Mailer
class has a __construct function whose first argument is the ClientConfiguration object:
Listing 106-4
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 services:
3
my_mailer:
4
class: Acme\HelloBundle\Mail\Mailer
5
scope: client
6
arguments: ["@client_configuration"]
C) Passing the Container as a Dependency of your Service
Setting the scope to a narrower one is not always possible (for instance, a twig extension must be in the
container scope as the Twig environment needs it as a dependency). In these cases, you can pass the
entire container into your service:
Listing 106-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/Acme/HelloBundle/Mail/Mailer.php
namespace Acme\HelloBundle\Mail;
use Symfony\Component\DependencyInjection\ContainerInterface;
class Mailer
{
protected $container;
public function __construct(ContainerInterface $container)
{
$this->container = $container;
}
public function sendEmail()
{
$request = $this->container->get('client_configuration');
// ... do something using the client configuration here
}
}
Take care not to store the client configuration in a property of the object for a future call of the
service as it would cause the same issue described in the first section (except that Symfony cannot
detect that you are wrong).
The service config for this class would look something like this:
Listing 106-6
1 # src/Acme/HelloBundle/Resources/config/services.yml
2 parameters:
3
# ...
PDF brought to you by
generated on December 3, 2014
Chapter 106: How to Work with Scopes | 390
4
my_mailer.class: Acme\HelloBundle\Mail\Mailer
5
6 services:
7
my_mailer:
8
class:
"%my_mailer.class%"
9
arguments: ["@service_container"]
10
# scope: container can be omitted as it is the default
Injecting the whole container into a service is generally not a good idea (only inject what you need).
PDF brought to you by
generated on December 3, 2014
Chapter 106: How to Work with Scopes | 391
Chapter 107
How to Work with Compiler Passes in Bundles
Compiler passes give you an opportunity to manipulate other service definitions that have been registered
with the service container. You can read about how to create them in the components section "Compiling
the Container". To register a compiler pass from a bundle you need to add it to the build method of the
bundle definition class:
Listing 107-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/MailerBundle/AcmeMailerBundle.php
namespace Acme\MailerBundle;
use Symfony\Component\HttpKernel\Bundle\Bundle;
use Symfony\Component\DependencyInjection\ContainerBuilder;
use Acme\MailerBundle\DependencyInjection\Compiler\CustomCompilerPass;
class AcmeMailerBundle extends Bundle
{
public function build(ContainerBuilder $container)
{
parent::build($container);
$container->addCompilerPass(new CustomCompilerPass());
}
}
One of the most common use-cases of compiler passes is to work with tagged services (read more about
tags in the components section "Working with Tagged Services"). If you are using custom tags in a bundle
then by convention, tag names consist of the name of the bundle (lowercase, underscores as separators),
followed by a dot, and finally the "real" name. For example, if you want to introduce some sort of
"transport" tag in your AcmeMailerBundle, you should call it acme_mailer.transport.
PDF brought to you by
generated on December 3, 2014
Chapter 107: How to Work with Compiler Passes in Bundles | 392
Chapter 108
Session Proxy Examples
The session proxy mechanism has a variety of uses and this example demonstrates two common uses.
Rather than injecting the session handler as normal, a handler is injected into the proxy and registered
with the session storage driver:
Listing 108-1
1
2
3
4
5
6
use Symfony\Component\HttpFoundation\Session\Session;
use Symfony\Component\HttpFoundation\Session\Storage\NativeSessionStorage;
use Symfony\Component\HttpFoundation\Session\Storage\Handler\PdoSessionHandler;
$proxy = new YourProxy(new PdoSessionHandler());
$session = new Session(new NativeSessionStorage(array(), $proxy));
Below, you'll learn two real examples that can be used for YourProxy: encryption of session data and
readonly guest sessions.
Encryption of Session Data
If you wanted to encrypt the session data, you could use the proxy to encrypt and decrypt the session as
required:
Listing 108-2
1 use Symfony\Component\HttpFoundation\Session\Storage\Proxy\SessionHandlerProxy;
2
3 class EncryptedSessionProxy extends SessionHandlerProxy
4 {
5
private $key;
6
7
public function __construct(\SessionHandlerInterface $handler, $key)
8
{
9
$this->key = $key;
10
11
parent::__construct($handler);
12
}
13
14
public function read($id)
PDF brought to you by
generated on December 3, 2014
Chapter 108: Session Proxy Examples | 393
15
16
17
18
19
20
21
22
23
24
25
26
27 }
{
$data = parent::read($id);
return mcrypt_decrypt(\MCRYPT_3DES, $this->key, $data);
}
public function write($id, $data)
{
$data = mcrypt_encrypt(\MCRYPT_3DES, $this->key, $data);
return parent::write($id, $data);
}
Readonly Guest Sessions
There are some applications where a session is required for guest users, but where there is no particular
need to persist the session. In this case you can intercept the session before it is written:
Listing 108-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
use Foo\User;
use Symfony\Component\HttpFoundation\Session\Storage\Proxy\SessionHandlerProxy;
class ReadOnlyGuestSessionProxy extends SessionHandlerProxy
{
private $user;
public function __construct(\SessionHandlerInterface $handler, User $user)
{
$this->user = $user;
parent::__construct($handler);
}
public function write($id, $data)
{
if ($this->user->isGuest()) {
return;
}
return parent::write($id, $data);
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 108: Session Proxy Examples | 394
Chapter 109
Making the Locale "Sticky" during a User's
Session
Prior to Symfony 2.1, the locale was stored in a session attribute called _locale. Since 2.1, it is stored in
the Request, which means that it's not "sticky" during a user's request. In this article, you'll learn how to
make the locale of a user "sticky" so that once it's set, that same locale will be used for every subsequent
request.
Creating a LocaleListener
To simulate that the locale is stored in a session, you need to create and register a new event listener.
The listener will look something like this. Typically, _locale is used as a routing parameter to signify the
locale, though it doesn't really matter how you determine the desired locale from the request:
Listing 109-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/Acme/LocaleBundle/EventListener/LocaleListener.php
namespace Acme\LocaleBundle\EventListener;
use Symfony\Component\HttpKernel\Event\GetResponseEvent;
use Symfony\Component\HttpKernel\KernelEvents;
use Symfony\Component\EventDispatcher\EventSubscriberInterface;
class LocaleListener implements EventSubscriberInterface
{
private $defaultLocale;
public function __construct($defaultLocale = 'en')
{
$this->defaultLocale = $defaultLocale;
}
public function onKernelRequest(GetResponseEvent $event)
{
$request = $event->getRequest();
PDF brought to you by
generated on December 3, 2014
Chapter 109: Making the Locale "Sticky" during a User's Session | 395
20
if (!$request->hasPreviousSession()) {
21
return;
22
}
23
24
// try to see if the locale has been set as a _locale routing parameter
25
if ($locale = $request->attributes->get('_locale')) {
26
$request->getSession()->set('_locale', $locale);
27
} else {
28
// if no explicit locale has been set on this request, use one from the session
29
$request->setLocale($request->getSession()->get('_locale',
30 $this->defaultLocale));
31
}
32
}
33
34
public static function getSubscribedEvents()
35
{
36
return array(
37
// must be registered before the default Locale listener
38
KernelEvents::REQUEST => array(array('onKernelRequest', 17)),
39
);
40
}
}
Then register the listener:
Listing 109-2
1 services:
2
acme_locale.locale_listener:
3
class: Acme\LocaleBundle\EventListener\LocaleListener
4
arguments: ["%kernel.default_locale%"]
5
tags:
6
- { name: kernel.event_subscriber }
That's it! Now celebrate by changing the user's locale and seeing that it's sticky throughout the request.
Remember, to get the user's locale, always use the Request::getLocale1 method:
Listing 109-3
1
2
3
4
5
6
7
// from a controller...
use Symfony\Component\HttpFoundation\Request;
public function indexAction(Request $request)
{
$locale = $request->getLocale();
}
1. http://api.symfony.com/master/Symfony/Component/HttpFoundation/Request.html#getLocale()
PDF brought to you by
generated on December 3, 2014
Chapter 109: Making the Locale "Sticky" during a User's Session | 396
Chapter 110
Configuring the Directory where Session Files
are Saved
By default, the Symfony Standard Edition uses the global php.ini values for session.save_handler
and session.save_path to determine where to store session data. This is because of the following
configuration:
Listing 110-1
1 # app/config/config.yml
2 framework:
3
session:
4
# handler_id set to null will use default session handler from php.ini
5
handler_id: ~
With this configuration, changing where your session metadata is stored is entirely up to your php.ini
configuration.
However, if you have the following configuration, Symfony will store the session data in files in the
cache directory %kernel.cache_dir%/sessions. This means that when you clear the cache, any current
sessions will also be deleted:
Listing 110-2
1 # app/config/config.yml
2 framework:
3
session: ~
Using a different directory to save session data is one method to ensure that your current sessions aren't
lost when you clear Symfony's cache.
Using a different session save handler is an excellent (yet more complex) method of session
management available within Symfony. See Configuring Sessions and Save Handlers for a discussion
of session save handlers. There is also an entry in the cookbook about storing sessions in the
database.
To change the directory in which Symfony saves session data, you only need change the framework
configuration. In this example, you will change the session directory to app/sessions:
PDF brought to you by
generated on December 3, 2014
Chapter 110: Configuring the Directory where Session Files are Saved | 397
Listing 110-3
1 # app/config/config.yml
2 framework:
3
session:
4
handler_id: session.handler.native_file
5
save_path: "%kernel.root_dir%/sessions"
PDF brought to you by
generated on December 3, 2014
Chapter 110: Configuring the Directory where Session Files are Saved | 398
Chapter 111
Bridge a legacy Application with Symfony
Sessions
New in version 2.3: The ability to integrate with a legacy PHP session was introduced in Symfony
2.3.
If you're integrating the Symfony full-stack Framework into a legacy application that starts the session
with session_start(), you may still be able to use Symfony's session management by using the PHP
Bridge session.
If the application has sets it's own PHP save handler, you can specify null for the handler_id:
Listing 111-1
1 framework:
2
session:
3
storage_id: session.storage.php_bridge
4
handler_id: ~
Otherwise, if the problem is simply that you cannot avoid the application starting the session with
session_start(), you can still make use of a Symfony based session save handler by specifying the save
handler as in the example below:
Listing 111-2
1 framework:
2
session:
3
storage_id: session.storage.php_bridge
4
handler_id: session.handler.native_file
PDF brought to you by
generated on December 3, 2014
Chapter 111: Bridge a legacy Application with Symfony Sessions | 399
If the legacy application requires its own session save-handler, do not override this. Instead set
handler_id: ~. Note that a save handler cannot be changed once the session has been started. If
the application starts the session before Symfony is initialized, the save-handler will have already
been set. In this case, you will need handler_id: ~. Only override the save-handler if you are sure
the legacy application can use the Symfony save-handler without side effects and that the session
has not been started before Symfony is initialized.
For more details, see Integrating with Legacy Sessions.
PDF brought to you by
generated on December 3, 2014
Chapter 111: Bridge a legacy Application with Symfony Sessions | 400
Chapter 112
Limit Session Metadata Writes
New in version 2.4: The ability to limit session metadata writes was introduced in Symfony 2.4.
The default behavior of PHP session is to persist the session regardless of whether the session data has
changed or not. In Symfony, each time the session is accessed, metadata is recorded (session created/last
used) which can be used to determine session age and idle time.
If for performance reasons you wish to limit the frequency at which the session persists, this feature can
adjust the granularity of the metadata updates and persist the session less often while still maintaining
relatively accurate metadata. If other session data is changed, the session will always persist.
You can tell Symfony not to update the metadata "session last updated" time until a certain amount
of time has passed, by setting framework.session.metadata_update_threshold to a value in seconds
greater than zero:
Listing 112-1
1 framework:
2
session:
3
metadata_update_threshold: 120
PHP default's behavior is to save the session whether it has been changed or not. When using
framework.session.metadata_update_threshold Symfony will wrap the session handler
(configured at framework.session.handler_id) into the WriteCheckSessionHandler. This will
prevent any session write if the session was not modified.
Be aware that if the session is not written at every request, it may be garbage collected sooner than
usual. This means that your users may be logged out sooner than expected.
PDF brought to you by
generated on December 3, 2014
Chapter 112: Limit Session Metadata Writes | 401
Chapter 113
How Symfony2 Differs from Symfony1
The Symfony2 framework embodies a significant evolution when compared with the first version of the
framework. Fortunately, with the MVC architecture at its core, the skills used to master a symfony1
project continue to be very relevant when developing in Symfony2. Sure, app.yml is gone, but routing,
controllers and templates all remain.
This chapter walks through the differences between symfony1 and Symfony2. As you'll see, many tasks
are tackled in a slightly different way. You'll come to appreciate these minor differences as they promote
stable, predictable, testable and decoupled code in your Symfony2 applications.
So, sit back and relax as you travel from "then" to "now".
Directory Structure
When looking at a Symfony2 project - for example, the Symfony2 Standard Edition1 - you'll notice a very
different directory structure than in symfony1. The differences, however, are somewhat superficial.
The app/ Directory
In symfony1, your project has one or more applications, and each lives inside the apps/ directory
(e.g. apps/frontend). By default in Symfony2, you have just one application represented by the app/
directory. Like in symfony1, the app/ directory contains configuration specific to that application. It also
contains application-specific cache, log and template directories as well as a Kernel class (AppKernel),
which is the base object that represents the application.
Unlike symfony1, almost no PHP code lives in the app/ directory. This directory is not meant to house
modules or library files as it did in symfony1. Instead, it's simply the home of configuration and other
resources (templates, translation files).
The src/ Directory
Put simply, your actual code goes here. In Symfony2, all actual application-code lives inside a bundle
(roughly equivalent to a symfony1 plugin) and, by default, each bundle lives inside the src directory.
1. https://github.com/symfony/symfony-standard
PDF brought to you by
generated on December 3, 2014
Chapter 113: How Symfony2 Differs from Symfony1 | 402
In that way, the src directory is a bit like the plugins directory in symfony1, but much more flexible.
Additionally, while your bundles will live in the src/ directory, third-party bundles will live somewhere
in the vendor/ directory.
To get a better picture of the src/ directory, first think of the structure of a symfony1 application. First,
part of your code likely lives inside one or more applications. Most commonly these include modules,
but could also include any other PHP classes you put in your application. You may have also created a
schema.yml file in the config directory of your project and built several model files. Finally, to help with
some common functionality, you're using several third-party plugins that live in the plugins/ directory.
In other words, the code that drives your application lives in many different places.
In Symfony2, life is much simpler because all Symfony2 code must live in a bundle. In the pretend
symfony1 project, all the code could be moved into one or more plugins (which is a very good practice,
in fact). Assuming that all modules, PHP classes, schema, routing configuration, etc. were moved into a
plugin, the symfony1 plugins/ directory would be very similar to the Symfony2 src/ directory.
Put simply again, the src/ directory is where your code, assets, templates and most anything else specific
to your project will live.
The vendor/ Directory
The vendor/ directory is basically equivalent to the lib/vendor/ directory in symfony1, which was the
conventional directory for all vendor libraries and bundles. By default, you'll find the Symfony2 library
files in this directory, along with several other dependent libraries such as Doctrine2, Twig and Swift
Mailer. 3rd party Symfony2 bundles live somewhere in the vendor/.
The web/ Directory
Not much has changed in the web/ directory. The most noticeable difference is the absence of the
css/, js/ and images/ directories. This is intentional. Like with your PHP code, all assets should also
live inside a bundle. With the help of a console command, the Resources/public/ directory of each
bundle is copied or symbolically-linked to the web/bundles/ directory. This allows you to keep assets
organized inside your bundle, but still make them available to the public. To make sure that all bundles
are available, run the following command:
Listing 113-1
1 $ php app/console assets:install web
This command is the Symfony2 equivalent to the symfony1 plugin:publish-assets command.
Autoloading
One of the advantages of modern frameworks is never needing to worry about requiring files. By making
use of an autoloader, you can refer to any class in your project and trust that it's available. Autoloading
has changed in Symfony2 to be more universal, faster, and independent of needing to clear your cache.
In symfony1, autoloading was done by searching the entire project for the presence of PHP class files
and caching this information in a giant array. That array told symfony1 exactly which file contained each
class. In the production environment, this caused you to need to clear the cache when classes were added
or moved.
PDF brought to you by
generated on December 3, 2014
Chapter 113: How Symfony2 Differs from Symfony1 | 403
In Symfony2, a tool named Composer2 handles this process. The idea behind the autoloader is simple:
the name of your class (including the namespace) must match up with the path to the file containing that
class. Take the FrameworkExtraBundle from the Symfony2 Standard Edition as an example:
Listing 113-2
1
2
3
4
5
6
7
8
9
namespace Sensio\Bundle\FrameworkExtraBundle;
use Symfony\Component\HttpKernel\Bundle\Bundle;
// ...
class SensioFrameworkExtraBundle extends Bundle
{
// ...
}
The
file
itself
lives
at
vendor/sensio/framework-extra-bundle/Sensio/Bundle/
FrameworkExtraBundle/SensioFrameworkExtraBundle.php. As you can see, the second part of the
path follows the namespace of the class. The first part is equal to the package name of the
SensioFrameworkExtraBundle.
The namespace, Sensio\Bundle\FrameworkExtraBundle, and package name, sensio/frameworkextra-bundle, spells out the directory that the file should live in (vendor/sensio/framework-extrabundle/Sensio/Bundle/FrameworkExtraBundle/). Composer can then look for the file at this specific
place and load it very fast.
If
the
file
did
not
live
at
this
exact
location,
you'd
receive
a
Class
"Sensio\Bundle\FrameworkExtraBundle\SensioFrameworkExtraBundle" does not exist. error. In
Symfony2, a "class does not exist" error means that the namespace of the class and physical location do
not match. Basically, Symfony2 is looking in one exact location for that class, but that location doesn't
exist (or contains a different class). In order for a class to be autoloaded, you never need to clear your
cache in Symfony2.
As mentioned before, for the autoloader to work, it needs to know that the Sensio namespace lives in
the vendor/sensio/framework-extra-bundle directory and that, for example, the Doctrine namespace
lives in the vendor/doctrine/orm/lib/ directory. This mapping is entirely controlled by Composer.
Each third-party library you load through Composer has its settings defined and Composer takes care of
everything for you.
For this to work, all third-party libraries used by your project must be defined in the composer.json file.
If you look at the HelloController from the Symfony2 Standard Edition you can see that it lives
in the Acme\DemoBundle\Controller namespace. Yet, the AcmeDemoBundle is not defined in your
composer.json file. Nonetheless are the files autoloaded. This is because you can tell composer to
autoload files from specific directories without defining a dependency:
Listing 113-3
"autoload": {
"psr-0": { "": "src/" }
}
This means that if a class is not found in the vendor directory, Composer will search in the src
directory before throwing a "class does not exist" exception. Read more about configuring the Composer
autoloader in the Composer documentation3.
Using the Console
In symfony1, the console is in the root directory of your project and is called symfony:
2. http://getcomposer.org
3. http://getcomposer.org/doc/04-schema.md#autoload
PDF brought to you by
generated on December 3, 2014
Chapter 113: How Symfony2 Differs from Symfony1 | 404
Listing 113-4
1 $ php symfony
In Symfony2, the console is now in the app sub-directory and is called console:
Listing 113-5
1 $ php app/console
Applications
In a symfony1 project, it is common to have several applications: one for the frontend and one for the
backend for instance.
In a Symfony2 project, you only need to create one application (a blog application, an intranet
application, ...). Most of the time, if you want to create a second application, you might instead create
another project and share some bundles between them.
And if you need to separate the frontend and the backend features of some bundles, you can create
sub-namespaces for controllers, sub-directories for templates, different semantic configurations, separate
routing configurations, and so on.
Of course, there's nothing wrong with having multiple applications in your project, that's entirely up to
you. A second application would mean a new directory, e.g. my_app/, with the same basic setup as the
app/ directory.
Read the definition of a Project, an Application, and a Bundle in the glossary.
Bundles and Plugins
In a symfony1 project, a plugin could contain configuration, modules, PHP libraries, assets and anything
else related to your project. In Symfony2, the idea of a plugin is replaced by the "bundle". A bundle is
even more powerful than a plugin because the core Symfony2 framework is brought in via a series of
bundles. In Symfony2, bundles are first-class citizens that are so flexible that even core code itself is a
bundle.
In symfony1, a plugin must be enabled inside the ProjectConfiguration class:
Listing 113-6
1
2
3
4
5
6
// config/ProjectConfiguration.class.php
public function setup()
{
// some plugins here
$this->enableAllPluginsExcept(array(...));
}
In Symfony2, the bundles are activated inside the application kernel:
Listing 113-7
1 // app/AppKernel.php
2 public function registerBundles()
3 {
4
$bundles = array(
5
new Symfony\Bundle\FrameworkBundle\FrameworkBundle(),
PDF brought to you by
generated on December 3, 2014
Chapter 113: How Symfony2 Differs from Symfony1 | 405
6
7
8
9
10
11
12 }
new Symfony\Bundle\TwigBundle\TwigBundle(),
...,
new Acme\DemoBundle\AcmeDemoBundle(),
);
return $bundles;
Routing (routing.yml) and Configuration (config.yml)
In symfony1, the routing.yml and app.yml configuration files were automatically loaded inside any
plugin. In Symfony2, routing and application configuration inside a bundle must be included manually.
For example, to include a routing resource from a bundle called AcmeDemoBundle, you can do the
following:
Listing 113-8
1 # app/config/routing.yml
2 _hello:
3
resource: "@AcmeDemoBundle/Resources/config/routing.yml"
This will load the routes found in the Resources/config/routing.yml file of the AcmeDemoBundle.
The special @AcmeDemoBundle is a shortcut syntax that, internally, resolves to the full path to that bundle.
You can use this same strategy to bring in configuration from a bundle:
Listing 113-9
1 # app/config/config.yml
2 imports:
3
- { resource: "@AcmeDemoBundle/Resources/config/config.yml" }
In Symfony2, configuration is a bit like app.yml in symfony1, except much more systematic. With
app.yml, you could simply create any keys you wanted. By default, these entries were meaningless and
depended entirely on how you used them in your application:
# some app.yml file from symfony1
2 all:
3
email:
4
from_address: [email protected]
Listing 113-10 1
In Symfony2, you can also create arbitrary entries under the parameters key of your configuration:
Listing 113-11 1
2
parameters:
email.from_address: [email protected]
You can now access this from a controller, for example:
Listing 113-12 1
public function helloAction($name)
2 {
3
$fromAddress = $this->container->getParameter('email.from_address');
4 }
In reality, the Symfony2 configuration is much more powerful and is used primarily to configure objects
that you can use. For more information, see the chapter titled "Service Container".
PDF brought to you by
generated on December 3, 2014
Chapter 113: How Symfony2 Differs from Symfony1 | 406
Chapter 114
How to Inject Variables into all Templates (i.e.
global Variables)
Sometimes you want a variable to be accessible to all the templates you use. This is possible inside your
app/config/config.yml file:
Listing 114-1
1 # app/config/config.yml
2 twig:
3
# ...
4
globals:
5
ga_tracking: UA-xxxxx-x
Now, the variable ga_tracking is available in all Twig templates:
Listing 114-2
1 <p>The google tracking code is: {{ ga_tracking }}</p>
It's that easy!
Using Service Container Parameters
You can also take advantage of the built-in Service Parameters system, which lets you isolate or reuse the
value:
Listing 114-3
1 # app/config/parameters.yml
2 parameters:
3
ga_tracking: UA-xxxxx-x
Listing 114-4
1 # app/config/config.yml
2 twig:
3
globals:
4
ga_tracking: "%ga_tracking%"
PDF brought to you by
generated on December 3, 2014
Chapter 114: How to Inject Variables into all Templates (i.e. global Variables) | 407
The same variable is available exactly as before.
Referencing Services
Instead of using static values, you can also set the value to a service. Whenever the global variable is
accessed in the template, the service will be requested from the service container and you get access to
that object.
The service is not loaded lazily. In other words, as soon as Twig is loaded, your service is
instantiated, even if you never use that global variable.
To define a service as a global Twig variable, prefix the string with @. This should feel familiar, as it's the
same syntax you use in service configuration.
Listing 114-5
1 # app/config/config.yml
2 twig:
3
# ...
4
globals:
5
user_management: "@acme_user.user_management"
Using a Twig Extension
If the global variable you want to set is more complicated - say an object - then you won't be able to use
the above method. Instead, you'll need to create a Twig Extension and return the global variable as one of
the entries in the getGlobals method.
PDF brought to you by
generated on December 3, 2014
Chapter 114: How to Inject Variables into all Templates (i.e. global Variables) | 408
Chapter 115
How to Use and Register Namespaced Twig
Paths
Usually, when you refer to a template, you'll use the MyBundle:Subdir:filename.html.twig format (see
Template Naming and Locations).
Twig also natively offers a feature called "namespaced paths", and support is built-in automatically for all
of your bundles.
Take the following paths as an example:
Listing 115-1
1 {% extends "AcmeDemoBundle::layout.html.twig" %}
2 {% include "AcmeDemoBundle:Foo:bar.html.twig" %}
With namespaced paths, the following works as well:
Listing 115-2
1 {% extends "@AcmeDemo/layout.html.twig" %}
2 {% include "@AcmeDemo/Foo/bar.html.twig" %}
Both paths are valid and functional by default in Symfony.
As an added bonus, the namespaced syntax is faster.
Registering your own Namespaces
You can also register your own custom namespaces. Suppose that you're using some third-party library
that includes Twig templates that live in vendor/acme/foo-bar/templates. First, register a namespace
for this directory:
Listing 115-3
PDF brought to you by
generated on December 3, 2014
Chapter 115: How to Use and Register Namespaced Twig Paths | 409
1 # app/config/config.yml
2 twig:
3
# ...
4
paths:
5
"%kernel.root_dir%/../vendor/acme/foo-bar/templates": foo_bar
The registered namespace is called foo_bar, which refers to the vendor/acme/foo-bar/templates
directory. Assuming there's a file called sidebar.twig in that directory, you can use it easily:
Listing 115-4
1 {% include '@foo_bar/sidebar.twig' %}
Multiple Paths per Namespace
You can also assign several paths to the same template namespace. The order in which paths are
configured is very important, because Twig will always load the first template that exists, starting from
the first configured path. This feature can be used as a fallback mechanism to load generic templates
when the specific template doesn't exist.
Listing 115-5
1 # app/config/config.yml
2 twig:
3
# ...
4
paths:
5
"%kernel.root_dir%/../vendor/acme/themes/theme1": theme
6
"%kernel.root_dir%/../vendor/acme/themes/theme2": theme
7
"%kernel.root_dir%/../vendor/acme/themes/common": theme
Now, you can use the same @theme namespace to refer to any template located in the previous three
directories:
Listing 115-6
1 {% include '@theme/header.twig' %}
PDF brought to you by
generated on December 3, 2014
Chapter 115: How to Use and Register Namespaced Twig Paths | 410
Chapter 116
How to Use PHP instead of Twig for Templates
Symfony defaults to Twig for its template engine, but you can still use plain PHP code if you want. Both
templating engines are supported equally in Symfony. Symfony adds some nice features on top of PHP to
make writing templates with PHP more powerful.
Rendering PHP Templates
If you want to use the PHP templating engine, first, make sure to enable it in your application
configuration file:
Listing 116-1
1 # app/config/config.yml
2 framework:
3
# ...
4
templating:
5
engines: ['twig', 'php']
You can now render a PHP template instead of a Twig one simply by using the .php extension in the
template name instead of .twig. The controller below renders the index.html.php template:
Listing 116-2
1
2
3
4
5
6
7
8
9
10
// src/Acme/HelloBundle/Controller/HelloController.php
// ...
public function indexAction($name)
{
return $this->render(
'AcmeHelloBundle:Hello:index.html.php',
array('name' => $name)
);
}
You
can
also
use
the
@Template1
AcmeHelloBundle:Hello:index.html.php template:
shortcut
to
render
the
default
1. http://symfony.com/doc/current/bundles/SensioFrameworkExtraBundle/annotations/view`
PDF brought to you by
generated on December 3, 2014
Chapter 116: How to Use PHP instead of Twig for Templates | 411
Listing 116-3
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/HelloBundle/Controller/HelloController.php
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Template;
// ...
/**
* @Template(engine="php")
*/
public function indexAction($name)
{
return array('name' => $name);
}
Enabling the php and twig template engines simultaneously is allowed, but it will produce an
undesirable side effect in your application: the @ notation for Twig namespaces will no longer be
supported for the render() method:
Listing 116-4
1 public function indexAction()
2 {
3
// ...
4
5
// namespaced templates will no longer work in controllers
6
$this->render('@Acme/Default/index.html.twig');
7
8
// you must use the traditional template notation
9
$this->render('AcmeBundle:Default:index.html.twig');
10 }
Listing 116-5
1
2
3
4
5
{# inside a Twig template, namespaced templates work as expected #}
{{ include('@Acme/Default/index.html.twig') }}
{# traditional template notation will also work #}
{{ include('AcmeBundle:Default:index.html.twig') }}
Decorating Templates
More often than not, templates in a project share common elements, like the well-known header and
footer. In Symfony, this problem is thought about differently: a template can be decorated by another
one.
The index.html.php template is decorated by layout.html.php, thanks to the extend() call:
Listing 116-6
1 <!-- src/Acme/HelloBundle/Resources/views/Hello/index.html.php -->
2 <?php $view->extend('AcmeHelloBundle::layout.html.php') ?>
3
4 Hello <?php echo $name ?>!
The AcmeHelloBundle::layout.html.php notation sounds familiar, doesn't it? It is the same notation
used to reference a template. The :: part simply means that the controller element is empty, so the
corresponding file is directly stored under views/.
Now, have a look at the layout.html.php file:
PDF brought to you by
generated on December 3, 2014
Chapter 116: How to Use PHP instead of Twig for Templates | 412
Listing 116-7
1
2
3
4
5
6
<!-- src/Acme/HelloBundle/Resources/views/layout.html.php -->
<?php $view->extend('::base.html.php') ?>
<h1>Hello Application</h1>
<?php $view['slots']->output('_content') ?>
The layout is itself decorated by another one (::base.html.php). Symfony supports multiple decoration
levels: a layout can itself be decorated by another one. When the bundle part of the template name is
empty, views are looked for in the app/Resources/views/ directory. This directory stores global views
for your entire project:
Listing 116-8
1
2
3
4
5
6
7
8
9
10
11
<!-- app/Resources/views/base.html.php -->
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title><?php $view['slots']->output('title', 'Hello Application') ?></title>
</head>
<body>
<?php $view['slots']->output('_content') ?>
</body>
</html>
For both layouts, the $view['slots']->output('_content') expression is replaced by the content
of the child template, index.html.php and layout.html.php respectively (more on slots in the next
section).
As you can see, Symfony provides methods on a mysterious $view object. In a template, the $view
variable is always available and refers to a special object that provides a bunch of methods that makes the
template engine tick.
Working with Slots
A slot is a snippet of code, defined in a template, and reusable in any layout decorating the template. In
the index.html.php template, define a title slot:
Listing 116-9
1
2
3
4
5
6
<!-- src/Acme/HelloBundle/Resources/views/Hello/index.html.php -->
<?php $view->extend('AcmeHelloBundle::layout.html.php') ?>
<?php $view['slots']->set('title', 'Hello World Application') ?>
Hello <?php echo $name ?>!
The base layout already has the code to output the title in the header:
<!-- app/Resources/views/base.html.php -->
2 <head>
3
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
4
<title><?php $view['slots']->output('title', 'Hello Application') ?></title>
5 </head>
Listing 116-10 1
The output() method inserts the content of a slot and optionally takes a default value if the slot is not
defined. And _content is just a special slot that contains the rendered child template.
PDF brought to you by
generated on December 3, 2014
Chapter 116: How to Use PHP instead of Twig for Templates | 413
For large slots, there is also an extended syntax:
Listing 116-11 1
<?php $view['slots']->start('title') ?>
2
Some large amount of HTML
3 <?php $view['slots']->stop() ?>
Including other Templates
The best way to share a snippet of template code is to define a template that can then be included into
other templates.
Create a hello.html.php template:
<!-- src/Acme/HelloBundle/Resources/views/Hello/hello.html.php -->
2 Hello <?php echo $name ?>!
Listing 116-12 1
And change the index.html.php template to include it:
<!-- src/Acme/HelloBundle/Resources/views/Hello/index.html.php -->
2 <?php $view->extend('AcmeHelloBundle::layout.html.php') ?>
3
4 <?php echo $view->render('AcmeHelloBundle:Hello:hello.html.php', array('name' => $name)) ?>
Listing 116-13 1
The render() method evaluates and returns the content of another template (this is the exact same
method as the one used in the controller).
Embedding other Controllers
And what if you want to embed the result of another controller in a template? That's very useful when
working with Ajax, or when the embedded template needs some variable not available in the main
template.
If you create a fancy action, and want to include it into the index.html.php template, simply use the
following code:
Listing 116-14
1
2
3
4
5
6
7
<!-- src/Acme/HelloBundle/Resources/views/Hello/index.html.php -->
<?php echo $view['actions']->render(
new
\Symfony\Component\HttpKernel\Controller\ControllerReference('AcmeHelloBundle:Hello:fancy',
array(
'name' => $name,
'color' => 'green',
))
) ?>
Here, the AcmeHelloBundle:Hello:fancy string refers to the fancy action of the Hello controller:
Listing 116-15
1 // src/Acme/HelloBundle/Controller/HelloController.php
2
3 class HelloController extends Controller
4 {
PDF brought to you by
generated on December 3, 2014
Chapter 116: How to Use PHP instead of Twig for Templates | 414
5
6
7
8
9
10
11
12
13
14
15
16
17 }
public function fancyAction($name, $color)
{
// create some object, based on the $color variable
$object = ...;
return $this->render('AcmeHelloBundle:Hello:fancy.html.php', array(
'name'
=> $name,
'object' => $object
));
}
// ...
But where is the $view['actions'] array element defined? Like $view['slots'], it's called a template
helper, and the next section tells you more about those.
Using Template Helpers
The Symfony templating system can be easily extended via helpers. Helpers are PHP objects that provide
features useful in a template context. actions and slots are two of the built-in Symfony helpers.
Creating Links between Pages
Speaking of web applications, creating links between pages is a must. Instead of hardcoding URLs in
templates, the router helper knows how to generate URLs based on the routing configuration. That way,
all your URLs can be easily updated by changing the configuration:
Listing 116-16 1
<a href="<?php echo $view['router']->generate('hello', array('name' => 'Thomas')) ?>">
2
Greet Thomas!
3 </a>
The generate() method takes the route name and an array of parameters as arguments. The route name
is the main key under which routes are referenced and the parameters are the values of the placeholders
defined in the route pattern:
# src/Acme/HelloBundle/Resources/config/routing.yml
2 hello: # The route name
3
path:
/hello/{name}
4
defaults: { _controller: AcmeHelloBundle:Hello:index }
Listing 116-17 1
Using Assets: Images, JavaScripts and Stylesheets
What would the Internet be without images, JavaScripts, and stylesheets? Symfony provides the assets
tag to deal with them easily:
Listing 116-18
1 <link href="<?php echo $view['assets']->getUrl('css/blog.css') ?>" rel="stylesheet"
2 type="text/css" />
3
<img src="<?php echo $view['assets']->getUrl('images/logo.png') ?>" />
PDF brought to you by
generated on December 3, 2014
Chapter 116: How to Use PHP instead of Twig for Templates | 415
The assets helper's main purpose is to make your application more portable. Thanks to this helper,
you can move the application root directory anywhere under your web root directory without changing
anything in your template's code.
Profiling Templates
By using the stopwatch helper, you are able to time parts of your template and display it on the timeline
of the WebProfilerBundle:
Listing 116-19 1
<?php $view['stopwatch']->start('foo') ?>
2 ... things that get timed
3 <?php $view['stopwatch']->stop('foo') ?>
If you use the same name more than once in your template, the times are grouped on the same line
in the timeline.
Output Escaping
When using PHP templates, escape variables whenever they are displayed to the user:
Listing 116-20 1
<?php echo $view->escape($var) ?>
By default, the escape() method assumes that the variable is outputted within an HTML context. The
second argument lets you change the context. For instance, to output something in a JavaScript script,
use the js context:
Listing 116-21 1
<?php echo $view->escape($var, 'js') ?>
PDF brought to you by
generated on December 3, 2014
Chapter 116: How to Use PHP instead of Twig for Templates | 416
Chapter 117
How to Write a custom Twig Extension
The main motivation for writing an extension is to move often used code into a reusable class like adding
support for internationalization. An extension can define tags, filters, tests, operators, global variables,
functions, and node visitors.
Creating an extension also makes for a better separation of code that is executed at compilation time and
code needed at runtime. As such, it makes your code faster.
Before writing your own extensions, have a look at the Twig official extension repository1.
Create the Extension Class
This cookbook describes how to write a custom Twig extension as of Twig 1.12. If you are using
an older version, please read Twig extensions documentation legacy2.
To get your custom functionality you must first create a Twig Extension class. As an example you'll create
a price filter to format a given number into price:
Listing 117-1
1
2
3
4
5
6
7
8
// src/Acme/DemoBundle/Twig/AcmeExtension.php
namespace Acme\DemoBundle\Twig;
class AcmeExtension extends \Twig_Extension
{
public function getFilters()
{
return array(
1. https://github.com/twigphp/Twig-extensions
2. http://twig.sensiolabs.org/doc/advanced_legacy.html#creating-an-extension
PDF brought to you by
generated on December 3, 2014
Chapter 117: How to Write a custom Twig Extension | 417
9
new \Twig_SimpleFilter('price', array($this, 'priceFilter')),
10
);
11
}
12
13
public function priceFilter($number, $decimals = 0, $decPoint = '.', $thousandsSep =
14 ',')
15
{
16
$price = number_format($number, $decimals, $decPoint, $thousandsSep);
17
$price = '$'.$price;
18
19
return $price;
20
}
21
22
public function getName()
23
{
24
return 'acme_extension';
25
}
}
Along with custom filters, you can also add custom functions and register global variables.
Register an Extension as a Service
Now you must let the Service Container know about your newly created Twig Extension:
Listing 117-2
1 # src/Acme/DemoBundle/Resources/config/services.yml
2 services:
3
acme.twig.acme_extension:
4
class: Acme\DemoBundle\Twig\AcmeExtension
5
tags:
6
- { name: twig.extension }
Keep in mind that Twig Extensions are not lazily loaded. This means that there's a higher chance
that
you'll
get
a
ServiceCircularReferenceException3
or
a
4
ScopeWideningInjectionException if any services (or your Twig Extension in this case) are
dependent on the request service. For more information take a look at How to Work with Scopes.
Using the custom Extension
Using your newly created Twig Extension is no different than any other:
Listing 117-3
1 {# outputs $5,500.00 #}
2 {{ '5500'|price }}
Passing other arguments to your filter:
3. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Exception/ServiceCircularReferenceException.html
4. http://api.symfony.com/master/Symfony/Component/DependencyInjection/Exception/ScopeWideningInjectionException.html
PDF brought to you by
generated on December 3, 2014
Chapter 117: How to Write a custom Twig Extension | 418
Listing 117-4
1 {# outputs $5500,2516 #}
2 {{ '5500.25155'|price(4, ',', '') }}
Learning further
For a more in-depth look into Twig Extensions, please take a look at the Twig extensions documentation5.
5. http://twig.sensiolabs.org/doc/advanced.html#creating-an-extension
PDF brought to you by
generated on December 3, 2014
Chapter 117: How to Write a custom Twig Extension | 419
Chapter 118
How to Render a Template without a custom
Controller
Usually, when you need to create a page, you need to create a controller and render a template from
within that controller. But if you're rendering a simple template that doesn't need any data passed
into it, you can avoid creating the controller entirely, by using the built-in
FrameworkBundle:Template:template controller.
For example, suppose you want to render a AcmeBundle:Static:privacy.html.twig template, which
doesn't require that any variables are passed to it. You can do this without creating a controller:
Listing 118-1
1 acme_privacy:
2
path: /privacy
3
defaults:
4
_controller: FrameworkBundle:Template:template
5
template:
'AcmeBundle:Static:privacy.html.twig'
The FrameworkBundle:Template:template controller will simply render whatever template you've
passed as the template default value.
You can of course also use this trick when rendering embedded controllers from within a template. But
since the purpose of rendering a controller from within a template is typically to prepare some data in
a custom controller, this is probably only useful if you'd like to cache this page partial (see Caching the
static Template).
Listing 118-2
1 {{ render(url('acme_privacy')) }}
Caching the static Template
Since templates that are rendered in this way are typically static, it might make sense to cache them.
Fortunately, this is easy! By configuring a few other variables in your route, you can control exactly how
your page is cached:
PDF brought to you by
generated on December 3, 2014
Chapter 118: How to Render a Template without a custom Controller | 420
Listing 118-3
1 acme_privacy:
2
path: /privacy
3
defaults:
4
_controller:
5
template:
6
maxAge:
7
sharedAge:
FrameworkBundle:Template:template
'AcmeBundle:Static:privacy.html.twig'
86400
86400
The maxAge and sharedAge values are used to modify the Response object created in the controller. For
more information on caching, see HTTP Cache.
There is also a private variable (not shown here). By default, the Response will be made public, as long
as maxAge or sharedAge are passed. If set to true, the Response will be marked as private.
PDF brought to you by
generated on December 3, 2014
Chapter 118: How to Render a Template without a custom Controller | 421
Chapter 119
How to Simulate HTTP Authentication in a
Functional Test
If your application needs HTTP authentication, pass the username and password as server variables to
createClient():
Listing 119-1
1 $client = static::createClient(array(), array(
2
'PHP_AUTH_USER' => 'username',
3
'PHP_AUTH_PW'
=> 'pa$$word',
4 ));
You can also override it on a per request basis:
Listing 119-2
1 $client->request('DELETE', '/post/12', array(), array(), array(
2
'PHP_AUTH_USER' => 'username',
3
'PHP_AUTH_PW'
=> 'pa$$word',
4 ));
When your application is using a form_login, you can simplify your tests by allowing your test
configuration to make use of HTTP authentication. This way you can use the above to authenticate in
tests, but still have your users log in via the normal form_login. The trick is to include the http_basic
key in your firewall, along with the form_login key:
Listing 119-3
1 # app/config/config_test.yml
2 security:
3
firewalls:
4
your_firewall_name:
5
http_basic: ~
PDF brought to you by
generated on December 3, 2014
Chapter 119: How to Simulate HTTP Authentication in a Functional Test | 422
Chapter 120
How to Simulate Authentication with a Token
in a Functional Test
Authenticating requests in functional tests might slow down the suite. It could become an issue especially
when form_login is used, since it requires additional requests to fill in and submit the form.
One of the solutions is to configure your firewall to use http_basic in the test environment as explained
in How to Simulate HTTP Authentication in a Functional Test. Another way would be to create a token
yourself and store it in a session. While doing this, you have to make sure that an appropriate cookie is
sent with a request. The following example demonstrates this technique:
Listing 120-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
// src/Acme/DemoBundle/Tests/Controller/DemoControllerTest.php
namespace Acme\DemoBundle\Tests\Controller;
use Symfony\Bundle\FrameworkBundle\Test\WebTestCase;
use Symfony\Component\BrowserKit\Cookie;
use Symfony\Component\Security\Core\Authentication\Token\UsernamePasswordToken;
class DemoControllerTest extends WebTestCase
{
private $client = null;
public function setUp()
{
$this->client = static::createClient();
}
public function testSecuredHello()
{
$this->logIn();
$crawler = $this->client->request('GET', '/demo/secured/hello/Fabien');
$this->assertTrue($this->client->getResponse()->isSuccessful());
$this->assertGreaterThan(0, $crawler->filter('html:contains("Hello
Fabien")')->count());
PDF brought to you by
generated on December 3, 2014
Chapter 120: How to Simulate Authentication with a Token in a Functional Test | 423
26
27
28
29
30
31
32
33
34
35
36
37
38
39
}
private function logIn()
{
$session = $this->client->getContainer()->get('session');
$firewall = 'secured_area';
$token = new UsernamePasswordToken('admin', null, $firewall, array('ROLE_ADMIN'));
$session->set('_security_'.$firewall, serialize($token));
$session->save();
$cookie = new Cookie($session->getName(), $session->getId());
$this->client->getCookieJar()->set($cookie);
}
}
The technique described in How to Simulate HTTP Authentication in a Functional Test is cleaner
and therefore the preferred way.
PDF brought to you by
generated on December 3, 2014
Chapter 120: How to Simulate Authentication with a Token in a Functional Test | 424
Chapter 121
How to Test the Interaction of several Clients
If you need to simulate an interaction between different clients (think of a chat for instance), create
several clients:
Listing 121-1
1
2
3
4
5
6
7
8
9
10
// ...
$harry = static::createClient();
$sally = static::createClient();
$harry->request('POST', '/say/sally/Hello');
$sally->request('GET', '/messages');
$this->assertEquals(Response::HTTP_CREATED, $harry->getResponse()->getStatusCode());
$this->assertRegExp('/Hello/', $sally->getResponse()->getContent());
New in version 2.4: Support for HTTP status code constants was introduced in Symfony 2.4.
This works except when your code maintains a global state or if it depends on a third-party library that
has some kind of global state. In such a case, you can insulate your clients:
Listing 121-2
1
2
3
4
5
6
7
8
9
10
11
// ...
$harry = static::createClient();
$sally = static::createClient();
$harry->insulate();
$sally->insulate();
$harry->request('POST', '/say/sally/Hello');
$sally->request('GET', '/messages');
PDF brought to you by
generated on December 3, 2014
Chapter 121: How to Test the Interaction of several Clients | 425
12 $this->assertEquals(Response::HTTP_CREATED, $harry->getResponse()->getStatusCode());
13 $this->assertRegExp('/Hello/', $sally->getResponse()->getContent());
Insulated clients transparently execute their requests in a dedicated and clean PHP process, thus avoiding
any side-effects.
As an insulated client is slower, you can keep one client in the main process, and insulate the other
ones.
PDF brought to you by
generated on December 3, 2014
Chapter 121: How to Test the Interaction of several Clients | 426
Chapter 122
How to Use the Profiler in a Functional Test
It's highly recommended that a functional test only tests the Response. But if you write functional tests
that monitor your production servers, you might want to write tests on the profiling data as it gives you
a great way to check various things and enforce some metrics.
The Symfony Profiler gathers a lot of data for each request. Use this data to check the number of database
calls, the time spent in the framework, etc. But before writing assertions, enable the profiler and check
that the profiler is indeed available (it is enabled by default in the test environment):
Listing 122-1
1 class HelloControllerTest extends WebTestCase
2 {
3
public function testIndex()
4
{
5
$client = static::createClient();
6
7
// Enable the profiler for the next request (it does nothing if the profiler is
8 not available)
9
$client->enableProfiler();
10
11
$crawler = $client->request('GET', '/hello/Fabien');
12
13
// ... write some assertions about the Response
14
15
// Check that the profiler is enabled
16
if ($profile = $client->getProfile()) {
17
// check the number of requests
18
$this->assertLessThan(
19
10,
20
$profile->getCollector('db')->getQueryCount()
21
);
22
23
// check the time spent in the framework
24
$this->assertLessThan(
25
500,
26
$profile->getCollector('time')->getDuration()
27
);
}
PDF brought to you by
generated on December 3, 2014
Chapter 122: How to Use the Profiler in a Functional Test | 427
28
29 }
}
If a test fails because of profiling data (too many DB queries for instance), you might want to use the Web
Profiler to analyze the request after the tests finish. It's easy to achieve if you embed the token in the error
message:
Listing 122-2
1 $this->assertLessThan(
2
30,
3
$profile->getCollector('db')->getQueryCount(),
4
sprintf(
5
'Checks that query count is less than 30 (token %s)',
6
$profile->getToken()
7
)
8 );
The profiler store can be different depending on the environment (especially if you use the SQLite
store, which is the default configured one).
The profiler information is available even if you insulate the client or if you use an HTTP layer for
your tests.
Read the API for built-in data collectors to learn more about their interfaces.
Speeding up Tests by not Collecting Profiler Data
To avoid collecting data in each test you can set the collect parameter to false:
Listing 122-3
1 # app/config/config_test.yml
2
3 # ...
4 framework:
5
profiler:
6
enabled: true
7
collect: false
In this way only tests that call $client->enableProfiler() will collect data.
PDF brought to you by
generated on December 3, 2014
Chapter 122: How to Use the Profiler in a Functional Test | 428
Chapter 123
How to Test Code that Interacts with the
Database
If your code interacts with the database, e.g. reads data from or stores data into it, you need to adjust
your tests to take this into account. There are many ways how to deal with this. In a unit test, you can
create a mock for a Repository and use it to return expected objects. In a functional test, you may need
to prepare a test database with predefined values to ensure that your test always has the same data to
work with.
If you want to test your queries directly, see How to Test Doctrine Repositories.
Mocking the Repository in a Unit Test
If you want to test code which depends on a Doctrine repository in isolation, you need to mock the
Repository. Normally you inject the EntityManager into your class and use it to get the repository. This
makes things a little more difficult as you need to mock both the EntityManager and your repository
class.
It is possible (and a good idea) to inject your repository directly by registering your repository as a
factory service. This is a little bit more work to setup, but makes testing easier as you only need to
mock the repository.
Suppose the class you want to test looks like this:
Listing 123-1
1 namespace Acme\DemoBundle\Salary;
2
3 use Doctrine\Common\Persistence\ObjectManager;
4
PDF brought to you by
generated on December 3, 2014
Chapter 123: How to Test Code that Interacts with the Database | 429
5 class SalaryCalculator
6 {
7
private $entityManager;
8
9
public function __construct(ObjectManager $entityManager)
10
{
11
$this->entityManager = $entityManager;
12
}
13
14
public function calculateTotalSalary($id)
15
{
16
$employeeRepository =
17 $this->entityManager->getRepository('AcmeDemoBundle::Employee');
18
$employee = $employeeRepository->find($id);
19
20
return $employee->getSalary() + $employee->getBonus();
21
}
}
Since the ObjectManager gets injected into the class through the constructor, it's easy to pass a mock
object within a test:
Listing 123-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
use Acme\DemoBundle\Salary\SalaryCalculator;
class SalaryCalculatorTest extends \PHPUnit_Framework_TestCase
{
public function testCalculateTotalSalary()
{
// First, mock the object to be used in the test
$employee = $this->getMock('\Acme\DemoBundle\Entity\Employee');
$employee->expects($this->once())
->method('getSalary')
->will($this->returnValue(1000));
$employee->expects($this->once())
->method('getBonus')
->will($this->returnValue(1100));
// Now, mock the repository so it returns the mock of the employee
$employeeRepository = $this->getMockBuilder('\Doctrine\ORM\EntityRepository')
->disableOriginalConstructor()
->getMock();
$employeeRepository->expects($this->once())
->method('find')
->will($this->returnValue($employee));
// Last, mock the EntityManager to return the mock of the repository
$entityManager =
$this->getMockBuilder('\Doctrine\Common\Persistence\ObjectManager')
->disableOriginalConstructor()
->getMock();
$entityManager->expects($this->once())
->method('getRepository')
->will($this->returnValue($employeeRepository));
$salaryCalculator = new SalaryCalculator($entityManager);
$this->assertEquals(2100, $salaryCalculator->calculateTotalSalary(1));
}
}
PDF brought to you by
generated on December 3, 2014
Chapter 123: How to Test Code that Interacts with the Database | 430
In this example, you are building the mocks from the inside out, first creating the employee which gets
returned by the Repository, which itself gets returned by the EntityManager. This way, no real class is
involved in testing.
Changing Database Settings for Functional Tests
If you have functional tests, you want them to interact with a real database. Most of the time you want
to use a dedicated database connection to make sure not to overwrite data you entered when developing
the application and also to be able to clear the database before every test.
To do this, you can specify a database configuration which overwrites the default configuration:
Listing 123-3
1 # app/config/config_test.yml
2 doctrine:
3
# ...
4
dbal:
5
host:
localhost
6
dbname:
testdb
7
user:
testdb
8
password: testdb
Make sure that your database runs on localhost and has the defined database and user credentials set up.
PDF brought to you by
generated on December 3, 2014
Chapter 123: How to Test Code that Interacts with the Database | 431
Chapter 124
How to Test Doctrine Repositories
Unit testing Doctrine repositories in a Symfony project is not recommended. When you're dealing with
a repository, you're really dealing with something that's meant to be tested against a real database
connection.
Fortunately, you can easily test your queries against a real database, as described below.
Functional Testing
If you need to actually execute a query, you will need to boot the kernel to get a valid connection. In this
case, you'll extend the KernelTestCase, which makes all of this quite easy:
Listing 124-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/Acme/StoreBundle/Tests/Entity/ProductRepositoryFunctionalTest.php
namespace Acme\StoreBundle\Tests\Entity;
use Symfony\Bundle\FrameworkBundle\Test\KernelTestCase;
class ProductRepositoryFunctionalTest extends KernelTestCase
{
/**
* @var \Doctrine\ORM\EntityManager
*/
private $em;
/**
* {@inheritDoc}
*/
public function setUp()
{
self::bootKernel();
$this->em = static::$kernel->getContainer()
->get('doctrine')
->getManager()
;
}
PDF brought to you by
generated on December 3, 2014
Chapter 124: How to Test Doctrine Repositories | 432
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43 }
public function testSearchByCategoryName()
{
$products = $this->em
->getRepository('AcmeStoreBundle:Product')
->searchByCategoryName('foo')
;
$this->assertCount(1, $products);
}
/**
* {@inheritDoc}
*/
protected function tearDown()
{
parent::tearDown();
$this->em->close();
}
PDF brought to you by
generated on December 3, 2014
Chapter 124: How to Test Doctrine Repositories | 433
Chapter 125
How to Customize the Bootstrap Process
before Running Tests
Sometimes when running tests, you need to do additional bootstrap work before running those tests. For
example, if you're running a functional test and have introduced a new translation resource, then you will
need to clear your cache before running those tests. This cookbook covers how to do that.
First, add the following file:
Listing 125-1
1
2
3
4
5
6
7
8
9
10
// app/tests.bootstrap.php
if (isset($_ENV['BOOTSTRAP_CLEAR_CACHE_ENV'])) {
passthru(sprintf(
'php "%s/console" cache:clear --env=%s --no-warmup',
__DIR__,
$_ENV['BOOTSTRAP_CLEAR_CACHE_ENV']
));
}
require __DIR__.'/bootstrap.php.cache';
Replace the test bootstrap
tests.bootstrap.php:
Listing 125-2
file
bootstrap.php.cache
in
app/phpunit.xml.dist
with
<!-- app/phpunit.xml.dist -->
<!-- ... -->
<phpunit
...
bootstrap = "tests.bootstrap.php"
>
Now, you can define in your phpunit.xml.dist file which environment you want the cache to be
cleared:
Listing 125-3
PDF brought to you by
generated on December 3, 2014
Chapter 125: How to Customize the Bootstrap Process before Running Tests | 434
1 <!-- app/phpunit.xml.dist -->
2 <php>
3
<env name="BOOTSTRAP_CLEAR_CACHE_ENV" value="test"/>
4 </php>
This now becomes an environment variable (i.e. $_ENV) that's available in the custom bootstrap file
(tests.bootstrap.php).
PDF brought to you by
generated on December 3, 2014
Chapter 125: How to Customize the Bootstrap Process before Running Tests | 435
Chapter 126
How to Create a custom Validation Constraint
You can create a custom constraint by extending the base constraint class, Constraint1. As an example
you're going to create a simple validator that checks if a string contains only alphanumeric characters.
Creating the Constraint Class
First you need to create a Constraint class and extend Constraint2:
Listing 126-1
1
2
3
4
5
6
7
8
9
10
11
12
// src/Acme/DemoBundle/Validator/Constraints/ContainsAlphanumeric.php
namespace Acme\DemoBundle\Validator\Constraints;
use Symfony\Component\Validator\Constraint;
/**
* @Annotation
*/
class ContainsAlphanumeric extends Constraint
{
public $message = 'The string "%string%" contains an illegal character: it can only
contain letters or numbers.';
}
The @Annotation annotation is necessary for this new constraint in order to make it available for
use in classes via annotations. Options for your constraint are represented as public properties on
the constraint class.
1. http://api.symfony.com/master/Symfony/Component/Validator/Constraint.html
2. http://api.symfony.com/master/Symfony/Component/Validator/Constraint.html
PDF brought to you by
generated on December 3, 2014
Chapter 126: How to Create a custom Validation Constraint | 436
Creating the Validator itself
As you can see, a constraint class is fairly minimal. The actual validation is performed by another
"constraint validator" class. The constraint validator class is specified by the constraint's validatedBy()
method, which includes some simple default logic:
Listing 126-2
1
2
3
4
5
// in the base Symfony\Component\Validator\Constraint class
public function validatedBy()
{
return get_class($this).'Validator';
}
In other words, if you create a custom Constraint (e.g. MyConstraint), Symfony will automatically look
for another class, MyConstraintValidator when actually performing the validation.
The validator class is also simple, and only has one required method validate():
Listing 126-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/Acme/DemoBundle/Validator/Constraints/ContainsAlphanumericValidator.php
namespace Acme\DemoBundle\Validator\Constraints;
use Symfony\Component\Validator\Constraint;
use Symfony\Component\Validator\ConstraintValidator;
class ContainsAlphanumericValidator extends ConstraintValidator
{
public function validate($value, Constraint $constraint)
{
if (!preg_match('/^[a-zA-Za0-9]+$/', $value, $matches)) {
// If you're using the new 2.5 validation API (you probably are!)
$this->context->buildViolation($constraint->message)
->setParameter('%string%', $value)
->addViolation();
// If you're using the old 2.4 validation API
/*
$this->context->addViolation(
$constraint->message,
array('%string%' => $value)
);
*/
}
}
}
Inside validate, you don't need to return a value. Instead, you add violations to the validator's context
property and a value will be considered valid if it causes no violations. The buildViolation method
takes
the
error
message
as
its
argument
and
returns
an
instance
of
ConstraintViolationBuilderInterface3. The addViolation method call finally adds the violation to
the context.
New in version 2.5: The buildViolation method was added in Symfony 2.5. For usage examples
with older Symfony versions, see the corresponding versions of this documentation page.
3. http://api.symfony.com/master/Symfony/Component/Validator/Violation/ConstraintViolationBuilderInterface.html
PDF brought to you by
generated on December 3, 2014
Chapter 126: How to Create a custom Validation Constraint | 437
Using the new Validator
Using custom validators is very easy, just as the ones provided by Symfony itself:
Listing 126-4
1 # src/Acme/BlogBundle/Resources/config/validation.yml
2 Acme\DemoBundle\Entity\AcmeEntity:
3
properties:
4
name:
5
- NotBlank: ~
6
- Acme\DemoBundle\Validator\Constraints\ContainsAlphanumeric: ~
If your constraint contains options, then they should be public properties on the custom Constraint class
you created earlier. These options can be configured like options on core Symfony constraints.
Constraint Validators with Dependencies
If your constraint validator has dependencies, such as a database connection, it will need to be configured
as a service in the dependency injection container. This service must include the
validator.constraint_validator tag and an alias attribute:
Listing 126-5
1 services:
2
validator.unique.your_validator_name:
3
class: Fully\Qualified\Validator\Class\Name
4
tags:
5
- { name: validator.constraint_validator, alias: alias_name }
Your constraint class should now use this alias to reference the appropriate validator:
Listing 126-6
1 public function validatedBy()
2 {
3
return 'alias_name';
4 }
As mentioned above, Symfony will automatically look for a class named after the constraint, with
Validator appended. If your constraint validator is defined as a service, it's important that you override
the validatedBy() method to return the alias used when defining your service, otherwise Symfony won't
use the constraint validator service, and will instantiate the class instead, without any dependencies
injected.
Class Constraint Validator
Beside validating a class property, a constraint can have a class scope by providing a target in its
Constraint class:
Listing 126-7
1 public function getTargets()
2 {
3
return self::CLASS_CONSTRAINT;
4 }
With this, the validator validate() method gets an object as its first argument:
Listing 126-8
1 class ProtocolClassValidator extends ConstraintValidator
2 {
3
public function validate($protocol, Constraint $constraint)
PDF brought to you by
generated on December 3, 2014
Chapter 126: How to Create a custom Validation Constraint | 438
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22 }
{
if ($protocol->getFoo() != $protocol->getBar()) {
// If you're using the new 2.5 validation API (you probably are!)
$this->context->buildViolation($constraint->message)
->atPath('foo')
->addViolation();
// If you're using the old 2.4 validation API
/*
$this->context->addViolationAt(
'foo',
$constraint->message,
array(),
null
);
*/
}
}
Note that a class constraint validator is applied to the class itself, and not to the property:
Listing 126-9
1 # src/Acme/BlogBundle/Resources/config/validation.yml
2 Acme\DemoBundle\Entity\AcmeEntity:
3
constraints:
4
- Acme\DemoBundle\Validator\Constraints\ContainsAlphanumeric: ~
PDF brought to you by
generated on December 3, 2014
Chapter 126: How to Create a custom Validation Constraint | 439
Chapter 127
How to Use PHP's built-in Web Server
New in version 2.6: The ability to run the server as a background process was introduced in
Symfony 2.6.
Since PHP 5.4 the CLI SAPI comes with a built-in web server1. It can be used to run your PHP applications
locally during development, for testing or for application demonstrations. This way, you don't have to
bother configuring a full-featured web server such as Apache or Nginx.
The built-in web server is meant to be run in a controlled environment. It is not designed to be
used on public networks.
Starting the Web Server
Running a Symfony application using PHP's built-in web server is as easy as executing the server:start
command:
Listing 127-1
1 $ php app/console server:start
This starts the web server at localhost:8000 in the background that serves your Symfony application.
By default, the web server listens on port 8000 on the loopback device. You can change the socket passing
an IP address and a port as a command-line argument:
Listing 127-2
1 $ php app/console server:run 192.168.0.1:8080
1. http://www.php.net/manual/en/features.commandline.webserver.php
PDF brought to you by
generated on December 3, 2014
Chapter 127: How to Use PHP's built-in Web Server | 440
You can use the server:status command to check if a web server is listening on a certain socket:
Listing 127-3
1 $ php app/console server:status
2
3 $ php app/console server:status 192.168.0.1:8080
The first command shows if your Symfony application will be server through localhost:8000, the
second one does the same for 192.168.0.1:8080.
Before Symfony 2.6, the server:run command was used to start the built-in web server. This
command is still available and behaves slightly different. Instead of starting the server in the
background, it will block the current terminal until you terminate it (this is usually done by
pressing Ctrl and C).
Using the built-in Web Server from inside a Virtual Machine
If you want to use the built-in web server from inside a virtual machine and then load the site from
a browser on your host machine, you'll need to listen on the 0.0.0.0:8000 address (i.e. on all IP
addresses that are assigned to the virtual machine):
Listing 127-4
1 $ php app/console server:start 0.0.0.0:8000
You should NEVER listen to all interfaces on a computer that is directly accessible from
the Internet. The built-in web server is not designed to be used on public networks.
Command Options
The built-in web server expects a "router" script (read about the "router" script on php.net2) as an
argument. Symfony already passes such a router script when the command is executed in the prod or in
the dev environment. Use the --router option in any other environment or to use another router script:
Listing 127-5
1 $ php app/console server:start --env=test --router=app/config/router_test.php
If your application's document root differs from the standard directory layout, you have to pass the
correct location using the --docroot option:
Listing 127-6
1 $ php app/console server:start --docroot=public_html
Stopping the Server
When you are finished, you can simply stop the web server using the server:stop command:
Listing 127-7
2. http://php.net/manual/en/features.commandline.webserver.php#example-401
PDF brought to you by
generated on December 3, 2014
Chapter 127: How to Use PHP's built-in Web Server | 441
1 $ php app/console server:stop
Like with the start command, if you omit the socket information, Symfony will stop the web server bound
to localhost:8000. Just pass the socket information when the web server listens to another IP address
or to another port:
Listing 127-8
1 $ php app/console server:stop 192.168.0.1:8080
PDF brought to you by
generated on December 3, 2014
Chapter 127: How to Use PHP's built-in Web Server | 442
Chapter 128
How to Create a SOAP Web Service in a
Symfony Controller
Setting up a controller to act as a SOAP server is simple with a couple tools. You must, of course, have
the PHP SOAP1 extension installed. As the PHP SOAP extension can not currently generate a WSDL, you
must either create one from scratch or use a 3rd party generator.
There are several SOAP server implementations available for use with PHP. Zend SOAP2 and
NuSOAP3 are two examples. Although the PHP SOAP extension is used in these examples, the
general idea should still be applicable to other implementations.
SOAP works by exposing the methods of a PHP object to an external entity (i.e. the person using the
SOAP service). To start, create a class - HelloService - which represents the functionality that you'll
expose in your SOAP service. In this case, the SOAP service will allow the client to call a method called
hello, which happens to send an email:
Listing 128-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/SoapBundle/Services/HelloService.php
namespace Acme\SoapBundle\Services;
class HelloService
{
private $mailer;
public function __construct(\Swift_Mailer $mailer)
{
$this->mailer = $mailer;
}
public function hello($name)
{
1. http://php.net/manual/en/book.soap.php
2. http://framework.zend.com/manual/en/zend.soap.server.html
3. http://sourceforge.net/projects/nusoap
PDF brought to you by
generated on December 3, 2014
Chapter 128: How to Create a SOAP Web Service in a Symfony Controller | 443
15
16
17
18
19
20
21
22
23
24
25 }
$message = \Swift_Message::newInstance()
->setTo('[email protected]')
->setSubject('Hello Service')
->setBody($name . ' says hi!');
$this->mailer->send($message);
return 'Hello, '.$name;
}
Next, you can train Symfony to be able to create an instance of this class. Since the class sends an e-mail,
it's been designed to accept a Swift_Mailer instance. Using the Service Container, you can configure
Symfony to construct a HelloService object properly:
Listing 128-2
1 # app/config/config.yml
2 services:
3
hello_service:
4
class: Acme\SoapBundle\Services\HelloService
5
arguments: ["@mailer"]
Below is an example of a controller that is capable of handling a SOAP request. If indexAction() is
accessible via the route /soap, then the WSDL document can be retrieved via /soap?wsdl.
Listing 128-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
namespace Acme\SoapBundle\Controller;
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
use Symfony\Component\HttpFoundation\Response;
class HelloServiceController extends Controller
{
public function indexAction()
{
$server = new \SoapServer('/path/to/hello.wsdl');
$server->setObject($this->get('hello_service'));
$response = new Response();
$response->headers->set('Content-Type', 'text/xml; charset=ISO-8859-1');
ob_start();
$server->handle();
$response->setContent(ob_get_clean());
return $response;
}
}
Take note of the calls to ob_start() and ob_get_clean(). These methods control output buffering4
which allows you to "trap" the echoed output of $server->handle(). This is necessary because Symfony
expects your controller to return a Response object with the output as its "content". You must also
remember to set the "Content-Type" header to "text/xml", as this is what the client will expect. So, you
use ob_start() to start buffering the STDOUT and use ob_get_clean() to dump the echoed output
into the content of the Response and clear the output buffer. Finally, you're ready to return the Response.
4. http://php.net/manual/en/book.outcontrol.php
PDF brought to you by
generated on December 3, 2014
Chapter 128: How to Create a SOAP Web Service in a Symfony Controller | 444
Below is an example calling the service using a NuSOAP5 client. This example assumes that the
indexAction in the controller above is accessible via the route /soap:
Listing 128-4
1 $client = new \Soapclient('http://example.com/app.php/soap?wsdl', true);
2
3 $result = $client->call('hello', array('name' => 'Scott'));
An example WSDL is below.
Listing 128-5
1 <?xml version="1.0" encoding="ISO-8859-1"?>
2 <definitions xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/"
3
xmlns:xsd="http://www.w3.org/2001/XMLSchema"
4
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
5
xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/"
6
xmlns:tns="urn:arnleadservicewsdl"
7
xmlns:soap="http://schemas.xmlsoap.org/wsdl/soap/"
8
xmlns:wsdl="http://schemas.xmlsoap.org/wsdl/"
9
xmlns="http://schemas.xmlsoap.org/wsdl/"
10
targetNamespace="urn:helloservicewsdl">
11
12
<types>
13
<xsd:schema targetNamespace="urn:hellowsdl">
14
<xsd:import namespace="http://schemas.xmlsoap.org/soap/encoding/" />
15
<xsd:import namespace="http://schemas.xmlsoap.org/wsdl/" />
16
</xsd:schema>
17
</types>
18
19
<message name="helloRequest">
20
<part name="name" type="xsd:string" />
21
</message>
22
23
<message name="helloResponse">
24
<part name="return" type="xsd:string" />
25
</message>
26
27
<portType name="hellowsdlPortType">
28
<operation name="hello">
29
<documentation>Hello World</documentation>
30
<input message="tns:helloRequest"/>
31
<output message="tns:helloResponse"/>
32
</operation>
33
</portType>
34
35
<binding name="hellowsdlBinding" type="tns:hellowsdlPortType">
36
<soap:binding style="rpc" transport="http://schemas.xmlsoap.org/soap/http"/>
37
<operation name="hello">
38
<soap:operation soapAction="urn:arnleadservicewsdl#hello" style="rpc"/>
39
40
<input>
41
<soap:body use="encoded" namespace="urn:hellowsdl"
42
encodingStyle="http://schemas.xmlsoap.org/soap/encoding/"/>
43
</input>
44
45
<output>
46
<soap:body use="encoded" namespace="urn:hellowsdl"
47
encodingStyle="http://schemas.xmlsoap.org/soap/encoding/"/>
48
</output>
5. http://sourceforge.net/projects/nusoap
PDF brought to you by
generated on December 3, 2014
Chapter 128: How to Create a SOAP Web Service in a Symfony Controller | 445
49
</operation>
50
</binding>
51
52
<service name="hellowsdl">
53
<port name="hellowsdlPort" binding="tns:hellowsdlBinding">
54
<soap:address location="http://example.com/app.php/soap" />
55
</port>
56
</service>
57 </definitions>
PDF brought to you by
generated on December 3, 2014
Chapter 128: How to Create a SOAP Web Service in a Symfony Controller | 446
Chapter 129
How to Create and Store a Symfony Project in
Git
Though this entry is specifically about Git, the same generic principles will apply if you're storing
your project in Subversion.
Once you've read through Creating Pages in Symfony and become familiar with using Symfony, you'll nodoubt be ready to start your own project. In this cookbook article, you'll learn the best way to start a new
Symfony project that's stored using the Git1 source control management system.
Initial Project Setup
To get started, you'll need to download Symfony and initialize your local git repository:
1. Download the Symfony Standard Edition2 using Composer:
Listing 129-1
1 $ php composer.phar create-project symfony/framework-standard-edition path/ '~2.5'
Add the -vvv flag to see everything that Composer is doing - this is especially useful on a
slow connection where it may seem that nothing is happening.
Composer will now download the Standard Distribution along with all of the required vendor
libraries. For more information about downloading Symfony using Composer, see Installing
Symfony using Composer3.
1. http://git-scm.com/
2. http://symfony.com/download
3. http://symfony.com/doc/current/book/installation.html#option-1-composer
PDF brought to you by
generated on December 3, 2014
Chapter 129: How to Create and Store a Symfony Project in Git | 447
2. Initialize your Git repository:
Listing 129-2
1 $ git init
3. Add all of the initial files to Git:
Listing 129-3
1 $ git add .
As you might have noticed, not all files that were downloaded by Composer in step
1, have been staged for commit by Git. Certain files and folders, such as the project's
dependencies (which are managed by Composer), parameters.yml (which contains
sensitive information such as database credentials), log and cache files and dumped
assets (which are created automatically by your project), should not be committed in
Git. To help you prevent committing those files and folders by accident, the Standard
Distribution comes with a file called .gitignore, which contains a list of files and folders
that Git should ignore.
You may also want to create a .gitignore file that can be used system-wide. This allows
you to exclude files/folders for all your projects that are created by your IDE or operating
system. For details, see GitHub .gitignore4.
4. Create an initial commit with your started project:
Listing 129-4
1 $ git commit -m "Initial commit"
At this point, you have a fully-functional Symfony project that's correctly committed to Git. You can
immediately begin development, committing the new changes to your Git repository.
You can continue to follow along with the Creating Pages in Symfony chapter to learn more about how
to configure and develop inside your application.
The Symfony Standard Edition comes with some example functionality. To remove the sample
code, follow the instructions in the "How to Remove the AcmeDemoBundle" article.
Managing Vendor Libraries with composer.json
How Does it Work?
Every Symfony project uses a group of third-party "vendor" libraries. One way or another the goal is to
download these files into your vendor/ directory and, ideally, to give you some sane way to manage the
exact version you need for each.
By default, these libraries are downloaded by running a php composer.phar install "downloader"
binary. This composer.phar file is from a library called Composer5 and you can read more about
installing it in the Installation chapter.
4. https://help.github.com/articles/ignoring-files
5. http://getcomposer.org/
PDF brought to you by
generated on December 3, 2014
Chapter 129: How to Create and Store a Symfony Project in Git | 448
The composer.phar file reads from the composer.json file at the root of your project. This is an JSONformatted file, which holds a list of each of the external packages you need, the version to be downloaded
and more. The composer.phar file also reads from a composer.lock file, which allows you to pin each
library to an exact version. In fact, if a composer.lock file exists, the versions inside will override those
in composer.json. To upgrade your libraries to new versions, run php composer.phar update.
If you want to add a new package to your application, run the composer require command:
Listing 129-5
1 $ php composer.phar require doctrine/doctrine-fixtures-bundle
To learn more about Composer, see GetComposer.org6:
It's important to realize that these vendor libraries are not actually part of your repository. Instead, they're
simply un-tracked files that are downloaded into the vendor/. But since all the information needed to
download these files is saved in composer.json and composer.lock (which are stored in the repository),
any other developer can use the project, run php composer.phar install, and download the exact same
set of vendor libraries. This means that you're controlling exactly what each vendor library looks like,
without needing to actually commit them to your repository.
So, whenever a developer uses your project, they should run the php composer.phar install script to
ensure that all of the needed vendor libraries are downloaded.
Upgrading Symfony
Since Symfony is just a group of third-party libraries and third-party libraries are entirely controlled
through composer.json and composer.lock, upgrading Symfony means simply upgrading each of
these files to match their state in the latest Symfony Standard Edition.
Of course, if you've added new entries to composer.json, be sure to replace only the original parts
(i.e. be sure not to also delete any of your custom entries).
Vendors and Submodules
Instead of using the composer.json system for managing your vendor libraries, you may instead choose
to use native git submodules7. There is nothing wrong with this approach, though the composer.json
system is the official way to solve this problem and probably much easier to deal with. Unlike Git
submodules, Composer is smart enough to calculate which libraries depend on which other libraries.
Storing your Project on a remote Server
You now have a fully-functional Symfony project stored in Git. However, in most cases, you'll also want
to store your project on a remote server both for backup purposes, and so that other developers can
collaborate on the project.
The easiest way to store your project on a remote server is via a web-based hosting service like GitHub8
or Bitbucket9. Of course, there are more services out there, you can start your research with a comparison
of hosting services10.
6. http://getcomposer.org/
7. http://git-scm.com/book/en/Git-Tools-Submodules
8. https://github.com/
9. https://bitbucket.org/
10. http://en.wikipedia.org/wiki/Comparison_of_open-source_software_hosting_facilities
PDF brought to you by
generated on December 3, 2014
Chapter 129: How to Create and Store a Symfony Project in Git | 449
Alternatively, you can store your Git repository on any server by creating a barebones repository11 and
then pushing to it. One library that helps manage this is Gitolite12.
11. http://git-scm.com/book/en/Git-Basics-Getting-a-Git-Repository
12. https://github.com/sitaramc/gitolite
PDF brought to you by
generated on December 3, 2014
Chapter 129: How to Create and Store a Symfony Project in Git | 450
Chapter 130
How to Create and Store a Symfony Project in
Subversion
This entry is specifically about Subversion, and based on principles found in How to Create and
Store a Symfony Project in Git.
Once you've read through Creating Pages in Symfony and become familiar with using Symfony, you'll
no-doubt be ready to start your own project. The preferred method to manage Symfony projects is using
Git1 but some prefer to use Subversion2 which is totally fine!. In this cookbook article, you'll learn how to
manage your project using SVN3 in a similar manner you would do with Git4.
This is a method to tracking your Symfony project in a Subversion repository. There are several
ways to do and this one is simply one that works.
The Subversion Repository
For this article it's assumed that your repository layout follows the widespread standard structure:
Listing 130-1
1 myproject/
2
branches/
3
tags/
4
trunk/
1. http://git-scm.com/
2. http://subversion.apache.org/
3. http://subversion.apache.org/
4. http://git-scm.com/
PDF brought to you by
generated on December 3, 2014
Chapter 130: How to Create and Store a Symfony Project in Subversion | 451
Most Subversion hosting should follow this standard practice. This is the recommended layout in
Version Control with Subversion5 and the layout used by most free hosting (see Subversion Hosting
Solutions).
Initial Project Setup
To get started, you'll need to download Symfony and get the basic Subversion setup:
1. Download the Symfony Standard Edition6 with or without vendors.
2. Unzip/untar the distribution. It will create a folder called Symfony with your new project
structure, config files, etc. Rename it to whatever you like.
3. Checkout the Subversion repository that will host this project. Suppose it is hosted on Google
code7 and called myproject:
Listing 130-2
1 $ svn checkout http://myproject.googlecode.com/svn/trunk myproject
4. Copy the Symfony project files in the Subversion folder:
Listing 130-3
1 $ mv Symfony/* myproject/
5. Now, set the ignore rules. Not everything should be stored in your Subversion repository. Some
files (like the cache) are generated and others (like the database configuration) are meant to be
customized on each machine. This makes use of the svn:ignore property, so that specific files
can be ignored.
Listing 130-4
1
2
3
4
5
6
7
8
9
10
11
12
$ cd myproject/
$ svn add --depth=empty app app/cache app/logs app/config web
$
$
$
$
$
svn
svn
svn
svn
svn
propset
propset
propset
propset
propset
svn:ignore
svn:ignore
svn:ignore
svn:ignore
svn:ignore
"vendor" .
"bootstrap*" app/
"parameters.yml" app/config/
"*" app/cache/
"*" app/logs/
$ svn propset svn:ignore "bundles" web
$ svn ci -m "commit basic Symfony ignore list (vendor, app/bootstrap*, app/config/
parameters.yml, app/cache/*, app/logs/*, web/bundles)"
6. The rest of the files can now be added and committed to the project:
Listing 130-5
1 $ svn add --force .
2 $ svn ci -m "add basic Symfony Standard 2.X.Y"
7. Copy
app/config/parameters.yml
to
app/config/parameters.yml.dist.
The
parameters.yml file is ignored by svn (see above) so that machine-specific settings like database
passwords aren't committed. By creating the parameters.yml.dist file, new developers can
quickly clone the project, copy this file to parameters.yml, customize it, and start developing.
5. http://svnbook.red-bean.com/
6. http://symfony.com/download
7. http://code.google.com/hosting/
PDF brought to you by
generated on December 3, 2014
Chapter 130: How to Create and Store a Symfony Project in Subversion | 452
8. Finally, download all of the third-party vendor libraries by executing Composer. For details, see
Updating Vendors.
If you rely on any "dev" versions, then Git may be used to install those libraries, since there is no
archive available for download.
At this point, you have a fully-functional Symfony project stored in your Subversion repository. The
development can start with commits in the Subversion repository.
You can continue to follow along with the Creating Pages in Symfony chapter to learn more about how
to configure and develop inside your application.
The Symfony Standard Edition comes with some example functionality. To remove the sample
code, follow the instructions in the "How to Remove the AcmeDemoBundle" article.
Managing Vendor Libraries with composer.json
How Does it Work?
Every Symfony project uses a group of third-party "vendor" libraries. One way or another the goal is to
download these files into your vendor/ directory and, ideally, to give you some sane way to manage the
exact version you need for each.
By default, these libraries are downloaded by running a php composer.phar install "downloader"
binary. This composer.phar file is from a library called Composer8 and you can read more about
installing it in the Installation chapter.
The composer.phar file reads from the composer.json file at the root of your project. This is an JSONformatted file, which holds a list of each of the external packages you need, the version to be downloaded
and more. The composer.phar file also reads from a composer.lock file, which allows you to pin each
library to an exact version. In fact, if a composer.lock file exists, the versions inside will override those
in composer.json. To upgrade your libraries to new versions, run php composer.phar update.
If you want to add a new package to your application, run the composer require command:
Listing 130-6
1 $ php composer.phar require doctrine/doctrine-fixtures-bundle
To learn more about Composer, see GetComposer.org9:
It's important to realize that these vendor libraries are not actually part of your repository. Instead, they're
simply un-tracked files that are downloaded into the vendor/. But since all the information needed to
download these files is saved in composer.json and composer.lock (which are stored in the repository),
any other developer can use the project, run php composer.phar install, and download the exact same
set of vendor libraries. This means that you're controlling exactly what each vendor library looks like,
without needing to actually commit them to your repository.
So, whenever a developer uses your project, they should run the php composer.phar install script to
ensure that all of the needed vendor libraries are downloaded.
8. http://getcomposer.org/
9. http://getcomposer.org/
PDF brought to you by
generated on December 3, 2014
Chapter 130: How to Create and Store a Symfony Project in Subversion | 453
Upgrading Symfony
Since Symfony is just a group of third-party libraries and third-party libraries are entirely controlled
through composer.json and composer.lock, upgrading Symfony means simply upgrading each of
these files to match their state in the latest Symfony Standard Edition.
Of course, if you've added new entries to composer.json, be sure to replace only the original parts
(i.e. be sure not to also delete any of your custom entries).
Subversion Hosting Solutions
The biggest difference between Git10 and SVN11 is that Subversion needs a central repository to work. You
then have several solutions:
• Self hosting: create your own repository and access it either through the filesystem or the
network. To help in this task you can read Version Control with Subversion.
• Third party hosting: there are a lot of serious free hosting solutions available like GitHub12,
Google code13, SourceForge14 or Gna15. Some of them offer Git hosting as well.
10. http://git-scm.com/
11. http://subversion.apache.org/
12. https://github.com/
13. http://code.google.com/hosting/
14. http://sourceforge.net/
15. http://gna.org/
PDF brought to you by
generated on December 3, 2014
Chapter 130: How to Create and Store a Symfony Project in Subversion | 454
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement