Codalogic LMX - W3C XML Schema to C++ Binding Code Generator

Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Version 7.2
Document Revision 1
Copyright © 2003-2017 Codalogic Ltd.
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Table of Contents
Introduction.........................................................................................................................................................1
Quick Index.........................................................................................................................................................3
1 - Orientation.....................................................................................................................................................5
1.1 - What is LMX?.................................................................................................................................5
1.2 - Benefits of Using LMX...................................................................................................................5
1.3 - Download & Installation.................................................................................................................6
1.4 - Evaluating LMX..............................................................................................................................7
1.5 - A Simple Example...........................................................................................................................7
1.6 - Licensing & Purchase....................................................................................................................11
1.7 - Version History..............................................................................................................................12
1.8 - Supported XML Schema Features.................................................................................................20
2 - Quick Start...................................................................................................................................................23
2.1 - Code Generation............................................................................................................................23
2.1.1 - Code Generation Using the Windows Interface..................................................................23
2.1.2 - Code Generation Using the Windows (DOS) Command Line Version..............................24
2.1.3 - Code Generation Using the Linux Command Line Version................................................25
2.2 - C++ Compiling and Linking.........................................................................................................25
2.2.1 - Compiling the Runtime Supporting Software Source Code................................................26
2.2.2 - Compiling the Generated Code............................................................................................27
2.2.3 - Compiling Your Code to Use LMX Code...........................................................................27
2.3 - Unmarshaling (simple form).........................................................................................................28
2.4 - Marshaling (simple form)..............................................................................................................31
2.5 - Accesing the Data in the Generated Classes.................................................................................33
2.5.1 - Singular Empty Type C++ Interface....................................................................................33
2.5.2 - Optional Empty Type C++ Interface...................................................................................33
2.5.3 - Singular Simple Type C++ Interface...................................................................................34
2.5.4 - Optional Simple Type C++ Interface...................................................................................34
2.5.5 - Multiple Simple Type and List Simple Type C++ Interface...............................................35
2.5.6 - Singular Complex Type C++ Interface................................................................................36
2.5.6.1 - Singular Complex Type Usage Examples..................................................................37
2.5.7 - Optional Complex Type C++ Interface...............................................................................39
2.5.7.1 - Optional Complex Type Usage Examples.................................................................39
2.5.8 - Multiple Complex Type C++ Interface................................................................................39
2.5.8.1 - Multiple Complex Type Usage Examples..................................................................41
2.5.9 - Multiple xs:anyAttribute C++ Interface...............................................................................41
2.5.10 - Singular xs:any C++ Interface...........................................................................................42
2.5.11 - Optional xs:any C++ Interface...........................................................................................43
2.5.12 - Multiple xs:any C++ Interface...........................................................................................43
2.5.13 - Additional Polymorphic Methods......................................................................................44
2.5.13.1 - Finding the Identity of a Polymorphic Class............................................................45
2.5.13.2 - Polymorphic Cloning................................................................................................45
2.5.14 - Resetting the Class.............................................................................................................45
2.5.15 - Run-time Checking............................................................................................................46
2.6 - The Abbreviated Type Notation....................................................................................................47
2.6.1 - Abbreviated Type Notation Examples.................................................................................48
i
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Table of Contents
3 - In More Depth..............................................................................................................................................49
3.1 - Configuration using lmxuser.h......................................................................................................49
3.2 - Unmarshaling (advanced forms)...................................................................................................49
3.2.1 - Unmarshaling with Additional Control when Reading XML Input....................................51
3.2.2 - Unmarshalling Multiple XML Instances from the Same Source.........................................51
3.2.3 - Unmarshalling Schema Fragments......................................................................................52
3.3 - Unmarshalling a Sub-element within an XML Instance...............................................................53
3.4 - Unmarshalling Multiple Sub-elements within an XML Instance..................................................55
3.5 - Marshaling (advanced forms)........................................................................................................59
3.5.1 - Marshalling Multiple XML Instances to the Same Destination..........................................60
3.6 - Selecting the Input File Type (XSD, WSDL, DTD).....................................................................61
3.7 - The Command-line and Configuration File Format......................................................................61
3.8 - Command-line Flags.....................................................................................................................62
3.9 - Error Codes....................................................................................................................................74
3.10 - Generating makefiles...................................................................................................................77
3.11 - Use with Web Services................................................................................................................78
3.11.1 - Handling SOAP Messages.................................................................................................78
3.11.2 - HTTP Operations for Web Services..................................................................................79
3.11.2.1 - SOAP Message to Message Operations...................................................................80
3.11.2.2 - SOAP Object to Object Operations..........................................................................81
3.11.2.3 - Simple SOAP Operations.........................................................................................81
3.11.2.4 - RESTful Operations.................................................................................................82
3.12 - Naming of Methods and Variables..............................................................................................82
3.13 - Adapting LMX to Your Environment.........................................................................................84
3.13.1 - Modifying Schema Type to C++ Type Mapping...............................................................85
3.13.1.1 - Choosing Between Wide and Narrow Unicode Strings...........................................87
3.13.1.2 - Input/Output Converters...........................................................................................87
3.14 - Augmenting Generated Classes With Your Own Code..............................................................88
3.14.1 - Augmenting Generated Classes with Snippet Event Handlers..........................................88
3.15 - DTDs and Namespaces................................................................................................................91
3.16 - Debugging and Handling Errors..................................................................................................91
3.16.1 - Reporting Errors.................................................................................................................92
3.16.2 - Changing the Error Handling Behavior.............................................................................92
3.16.3 - Conditional Error Handling...............................................................................................93
3.16.4 - Collecting Debug Error Information when using Convenience Methods..........................93
3.16.5 - Debugging Support............................................................................................................95
3.17 - Adding External Character Set Transcoders...............................................................................96
3.18 - Adding Extra Namespace Information........................................................................................96
3.19 - Adding Schema Location Information........................................................................................97
3.20 - Disabling Output of XML Namespaces......................................................................................98
3.21 - Adding an XMLDecl to the XML output....................................................................................98
3.22 - Handling Multiple Schemas........................................................................................................99
3.22.1 - Handling Multiple Independent Schemas..........................................................................99
3.22.2 - Multiple Schemas that Share Common Schemas............................................................100
3.23 - Pattern Facet Handling Customization......................................................................................100
3.24 - Custom Pattern Facet Output Formatting..................................................................................103
3.25 - XML Output Format Customization.........................................................................................103
3.26 - Specifying Micro Formats.........................................................................................................105
ii
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Table of Contents
3 - In More Depth
3.27 - Allowing Unknown Items in an XML Instance........................................................................106
3.28 - mustUnderstand / Comprehension Required.............................................................................107
3.29 - Finding an XML Instance's Namespace....................................................................................108
3.30 - Setting a Decimal Value With a Float.......................................................................................109
3.31 - Test Framework Generation......................................................................................................109
3.32 - Strategies for Increasing Code Flexibility.................................................................................110
3.33 - LMX Extensions........................................................................................................................110
iii
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
iv
Introduction
This is the main documentation for the LMX™ XML Schema to C++ code generator. It is available as a single
HTML file for easy on-line viewing and viewing on Linux, as a Windows® HTML Help file (.chm) for
viewing on Windows, and as a PDF file for printing. A printed version of this manual can also be purchased.
If the LMX code generator seems like the tool you need (see What is LMX? to help you with that), the first
thing to do is download and install a copy. You can evaluate a limited functionality version of LMX without a
license. Having decided that LMX is what you need, you need to acquire a license file and store it on your PC.
You will then be able to generate code by following the Quick Start instructions. Later you may wish to adapt
LMX to better meet your requirements. Information on how to do this is covered in section 3 - In More Depth.
Introduction
1
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
2
Introduction
Quick Index
Quick links to popular topics:
1.7 - Version History
2 - Quick Start
2.1 - Code Generation
3.8 - Command-line Flags
2.2 - C++ Compiling and Linking
2.3 - Unmarshaling (simple form)
2.4 - Marshaling (simple form)
2.5 - Accesing the Data in the Generated Classes
3.14 - Augmenting Generated Classes With Your Own Code
3.11 - Use with Web Services
Quick Index
3
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
4
Quick Index
1 - Orientation
This section will give you an overview of what LMX does, and describes how to get ready to use LMX,
including how to install and license LMX.
1.1 - What is LMX?
LMX is a W3C XML Schema to C++ code generator. It takes a W3C XML Schema and generates C++ code
that allows you to read and write XML conforming to the Schema by interacting with the C++ classes/objects
generated by the tool.
It is primarily intended to be used in data oriented applications, but can also be used in document oriented
applications.
The code generator runs on a Windows or Linux PC, but the output is generated in C++ source code that is
portable to most platforms that support C++ templates and basic STL containers. (Contact us if the generated
code does not run on your platform.)
The code generator comes with its own lightweight XML pull parser and uses C++ ostreams for output. Thus
it is a complete XML solution that does not require additional components in order to interact with XML data.
1.2 - Benefits of Using LMX
• Quick and easy to manipulate XML data.
• Knowledge of XML and XML Schema not required - You can simply use the documentation in the
generated header file.
• The names of variables are checked at compile time - Simple typos in names are detected
automatically and early.
• No programming of tedious state code as required with SAX (and StAX) - LMX does all that in a
couple of seconds (or less!)
• Beneficial for both small and large schema.
• Easy to implement web services.
• Can read and generate XML fragments.
• Simple to use - A minimal number of source files so that you don't need to spend time managing
code.
• Data types can be customized using C++ language syntax - No need to learn complex code generator
configuration options, and less chance of introducing accidental differences between builds by using
the wrong configuration. (See table below.)
• Support for SOAP 'mustUnderstand' attribute detection.
• Easy to switch between wide character strings and 'narrow' character strings.
• Exception safe design.
• Portable code - Run on different platforms. Even build and test on one platform and run on another.
1 - Orientation
5
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
1.3 - Download & Installation
You can download the LMX code generator via http://www.codalogic.com/lmx/download.php#licensed.
To install on Windows simply run the installation program in the usual way.
On Linux, after extracting the file from the tarball you can either manually copy the files to appropriate
locations or use the install.pl script to install LMX. The latter can be used in interactive and
non-interactive mode. For an interactive install, simply execute the script in a terminal window, for example:
./install.pl
To enable a non-interactive install you can save the settings you want to use in a configuration file, an
example of which is:
# This is a comment
bin=/usr/local/bin
include=/usr/local/include/lmx-$version
lib32_static=/usr/local/lib/lmx-$version
lib32_shared=/usr/lib
lib64_static=/usr/local/lib64/lmx-$version
lib64_shared=/usr/lib64
doc=~/lmx/lmx-$version/doc
src=~/lmx/lmx-$version/src
examples=
license_ok=yes
Each specification has the form:
<category>=<install path>
There should be no leading white-space on a line. Comments may be preceded by a # character. Blank lines
and lines that contain all white-space are ignored.
Any occurrence of the string $version will be replaced with the value of the LMX version you are
installing. If an installation variable is set to the empty string (such as examples shown above) then that
category of items will not be installed. The script will ask you for any values that are not specified.
If the configuration settings are stored in a file such as my-install-info.txt, then installation using
those settings can be invoked using the command:
./install.pl -f my-install-info.txt
A simple way to create a configuration file is to do an interactive install and tell the script to save the results.
This can be done by doing:
./install.pl -o my-install-info.txt
You can operate the code generator with limited functionality for evaluation purposes without obtaining a
license.
6
1.3 - Download & Installation
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
1.4 - Evaluating LMX
To help with evaluation, LMX comes with a Windows based Evaluation Support Suite. This is a separate
installation packaged as part of the main installation. The Evaluation Support Suite can be installed during the
main LMX installation, or it can be installed at a later time by executing the Evaluation Support Suite
installation from the LMX folder on the Windows Start menu.
Additional information about the Evaluation Support Suite can be found in the LMX folder on the Windows
Start menu.
If you have your own schema and example XML instances, LMX can generate a simple test framework. This
consists of an additionally generated C++ file that exercises the main generated code. The test framework
code unmarshals an XML instance from a file, does a deep copy to a new set of objects, and then marshals the
result out as XML to a new file. You can then visually compare the original and generated files. To use this
feature, check the "Generate Test Framework File" box on the "Options" tab of WinLMX (or specify the
-tframework flag on the command-line versions).
If you would like to see examples of how to use the generated code, then the generated copy constructors (of
the form c_myClass::c_myClass( const c_myClass & rhs )) provide convenient examples.
A worked example is also available at http://www.codalogic.com/lmx/examples.php to further help with the
evaluation process.
1.5 - A Simple Example
We present here a simple example of the sort of code you'll write to use the LMX generated C++ code. While
short, the example does illustrate the majority of the aspects of using LMX generated code. We first present
the schema used by LMX to generate the code. In this case the generated code is stored in the files po.h and
po.cpp. This is followed by an example XML instance for the schema. Finally, the code used to interface to
the LMX generated classes is shown.
We have not included a description of this example, other than the comments that appear in the code. (Full
documentation on how to interface with LMX generated code appears in 2.5 - Accesing the Data in the
Generated Classes.) You should find however that this example is straightforward and intuitive to understand,
demonstrating that it is easy to learn how to use LMX. This makes your code faster to develop and easier to
understand.
Example code similar to this can be found in the examples sub-directory of the installation or at
http://www.codalogic.com/lmx/lmx-example.zip.
The schema:
<xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema">
<xsd:annotation>
<xsd:documentation xml:lang="en">
Purchase order schema for Example.com.
Copyright 2000 Example.com. All rights reserved.
</xsd:documentation>
</xsd:annotation>
1.4 - Evaluating LMX
7
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
<xsd:element name="purchaseOrder" type="PurchaseOrderType"/>
<xsd:element name="comment" type="xsd:string"/>
<xsd:complexType name="PurchaseOrderType">
<xsd:sequence>
<xsd:element name="shipTo" type="USAddress"/>
<xsd:element name="billTo" type="USAddress"/>
<xsd:element ref="comment" minOccurs="0"/>
<xsd:element name="items" type="Items"/>
</xsd:sequence>
<xsd:attribute name="orderDate" type="xsd:date"/>
</xsd:complexType>
<xsd:complexType name="USAddress">
<xsd:sequence>
<xsd:element name="name"
type="xsd:string"/>
<xsd:element name="street" type="xsd:string"/>
<xsd:element name="city"
type="xsd:string"/>
<xsd:element name="state" type="xsd:string"/>
<xsd:element name="zip"
type="xsd:decimal"/>
</xsd:sequence>
<xsd:attribute name="country" type="xsd:NMTOKEN"
fixed="US"/>
</xsd:complexType>
<xsd:complexType name="Items">
<xsd:sequence>
<xsd:element name="item" minOccurs="0" maxOccurs="unbounded">
<xsd:complexType>
<xsd:sequence>
<xsd:element name="productName" type="xsd:string"/>
<xsd:element name="quantity">
<xsd:simpleType>
<xsd:restriction base="xsd:positiveInteger">
<xsd:maxExclusive value="100"/>
</xsd:restriction>
</xsd:simpleType>
</xsd:element>
<xsd:element name="USPrice" type="xsd:decimal"/>
<xsd:element ref="comment"
minOccurs="0"/>
<xsd:element name="shipDate" type="xsd:date" minOccurs="0"/>
</xsd:sequence>
<xsd:attribute name="partNum" type="SKU" use="required"/>
</xsd:complexType>
</xsd:element>
</xsd:sequence>
</xsd:complexType>
<!-- Stock Keeping Unit, a code for identifying products -->
<xsd:simpleType name="SKU">
<xsd:restriction base="xsd:string">
<xsd:pattern value="\d{3}-[A-Z]{2}"/>
</xsd:restriction>
</xsd:simpleType>
</xsd:schema>
An XML Instance:
<?xml version='1.0'?>
<purchaseOrder orderDate='1999-10-20'>
8
1.5 - A Simple Example
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
<shipTo country='US'>
<name>Alice Smith</name>
<street>123 Maple Street</street>
<city>Mill Valley</city>
<state>CA</state>
<zip>90952</zip>
</shipTo>
<billTo country='US'>
<name>Robert Smith</name>
<street>8 Oak Avenue</street>
<city>Old Town</city>
<state>PA</state>
<zip>95819</zip>
</billTo>
<comment>Hurry, my lawn is going wild!</comment>
<items>
<item partNum='872-AA'>
<productName>Lawnmower</productName>
<quantity>1</quantity>
<USPrice>148.95</USPrice>
<comment>Confirm this is electric</comment>
</item>
<item partNum='926-AA'>
<productName>Baby Monitor</productName>
<quantity>1</quantity>
<USPrice>39.98</USPrice>
<shipDate>1999-05-21</shipDate>
</item>
</items>
</purchaseOrder>
And the example code to read the instance:
#include <iostream>
#include "po.h"
#include "lmxparse.h"
// For std::cout etc.
// Generated by LMX
// For the LMX parser
int main()
{
// Allocate a place to store the an error code returned by
// the unmarshaling operation.
lmx::elmx_error l_error;
// c_root is the class generated by LMX that will store the
// unmarshalled XML. This constructor unmarshals the data
// stored in the file "po.xml" and places any error code in
// l_error.
c_root l_po( "po.xml", &l_error );
if( l_error != lmx::ELMX_OK )
{
std::cout << "An error occurred while reading XML\n";
return 0;
}
//
//
//
//
Or we could have done:
c_root l_po;
lmx::elmx_error l_error = l_po.unmarshal( "po.xml" );
if( l_error ...
// Interrogate the parsed XML
//---------------------------
1.5 - A Simple Example
9
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
if( l_po.getchosen() == c_root::e_purchaseOrder )
{
// Make some intermediate variables so that components are
// easier to reference
const c_PurchaseOrderType *lp_pot = &l_po.get_purchaseOrder();
if( lp_pot->isset_orderDate() )
// If the orderDate is set...
{
std::cout << "Ordered on: " <<
lp_pot->get_orderDate() << "\n";
if( lp_pot->get_orderDate().get_year() <= 2000 )
std::cout <<
"This was ordered last century - It is now overdue!\n";
}
// Reference another part of the data structure
const c_Items *lp_items = &lp_pot->get_items();
float l_total_cost = 0;
// Print the total number of items
std::cout << "Number of items = " <<
lp_items->size_item() << "\n";
// Print out some information about each order item
for( size_t l_i=0; l_i<lp_items->size_item(); ++l_i )
{
std::cout << "Item " <<
(l_i + 1) << ": " <<
lp_items->get_item( l_i ).get_quantity() <<
" off - " <<
lmx::as_ascii( lp_items->get_item( l_i ).get_productName() ) <<
" (" <<
lmx::as_ascii( lp_items->get_item( l_i ).get_partNum() ) << ")";
// The comment is optional, so only print it out if it is present
if( lp_items->get_item( l_i ).isset_comment() )
std::cout << " [" <<
lmx::as_ascii( lp_items->get_item( l_i ).get_comment() ) << "]";
std::cout << " - $" <<
lp_items->get_item( l_i ).get_USPrice() << "\n";
l_total_cost +=
lp_items->get_item( l_i ).get_USPrice().get_scaled( 2 );
}
std::cout << "Total Cost: $" << l_total_cost/100 << "\n";
std::cout << "Ship to:
lmx::as_ascii(
std::cout << "Bill to:
lmx::as_ascii(
std::cout << "\n";
" <<
lp_pot->get_shipTo().get_name() ) << "\n";
" <<
lp_pot->get_billTo().get_name() ) << "\n";
// Modify the XML
//--------------c_root l_alt_po( l_po );
// We don't actually need to create
// a new instance before modifying
l_alt_po.get_purchaseOrder().get_items().append_item();
10
1.5 - A Simple Example
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
c_Items::c_item *lp_item =
&l_alt_po.get_purchaseOrder().get_items().back_item();
lp_item->set_productName( L"Fence" );
lp_item->set_quantity( 2 );
lp_item->set_partNum( L"100-AB" );
lp_item->set_USPrice( lmx::c_decimal( 12.95, 2U ) );
//
//
//
//
//
//
//
//
We can also do:
lp_item->set_USPrice( 12.95 );
// Although the schema needs to
// specify the fraction digits
// facet in order to get the number
// of decimal places correct!
lp_item->set_USPrice( lmx::c_decimal( 1295, 2 ) );
// e.g. 1295 / (10^2)
lp_item->set_USPrice( "12.95" );
lp_item->set_comment( L"Will this stop the baby getting on the lawn?" );
assert( lp_item->is_occurs_ok() );
// Check sufficient elements and
// attributes added
// Write the modified version of the XML to the file po-out.xml
//------------------------------------------------------------if( l_alt_po.marshal( "po-out.xml" ) == lmx::ELMX_OK )
std::cout << "Modified XML written successfully\n";
else
std::cout << "Error writing Modified XML\n";
}
return 0;
}
1.6 - Licensing & Purchase
For the purposes of evaluation, the LMX code generator can be used in a restricted mode without a license. In
the restricted mode, the code generator limits the number of XML objects that it will compile.
To fully use the product a license file must be acquired. This will be e-mailed to you when you purchase a
license.
To purchase a license, follow the links and instructions for purchasing at: http://www.codalogic.com/lmx/
When using a Windows version of LMX the license file can be placed in the same folder as the
lmx.exe/winlmx.exe executable (by default, C:\Program Files\LMX on 32-bit systems and
C:\Program Files (x86)\LMX on 64-bit systems), or in the folder specified by the
HKEY_CURRENT_USER\Software\Codalogic\LMX\dir registry key.
On Linux (and other Unix versions) you can either place the license file in one of the following directories;
$HOME/.linmx, $HOME/.settings/codalogic/linmx,
$HOME/.config/codalogic/linmx, or $HOME/.codalogic/linmx; or in an alternate directory
specified by setting and exporting the LMXDIR shell environment variable with the name of the directory
containing the license file (possibly in your login script such as .bash_profile). For example:1.6 - Licensing & Purchase
11
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
LMXDIR=$HOME/lmx
export LMXDIR
1.7 - Version History
7.2
• Added testing on VS 2015, VS 2017, g++ 5.4.0, g++ 6.3.0 and g++ 7.1.0
• Added c_xml_writer::set_convenience_options() and
c_xml_writer::get_convenience_options() to enable ability to configure whether
XMLDecl and 'standalone' are output when marshalling using the convenience methods. See
3.21 - Adding an XMLDecl to the XML output.
• Corrected default whitespace handling of xs:token and xs:normalizedString and their derived types.
• On a list type, calling clear_NAME() now marks a list as present.
• Added -fname-sep flag to specify separator character sequence for constructed file names.
• Added better support for using command-line wildcards to specify input XSD files.
• Added -method-getid and -method-has_id flags.
• Implemented on_lmx_unmarshal_outer_start() and
on_lmx_unmarshal_outer_end() snippet event hanlders.
• Added operator << ( std::ostream &, const c_binary & ) to facilitate output of raw binary data.
• Enhanced s_debug_error to allow customized descriptions for LMX error codes allocated for
user defined purposes. See 3.16.4 - Collecting Debug Error Information when using Convenience
Methods.
7.1
• Virtual methods added to c_xml_reader to allow conditional acceptance / rejection of unknown
attributes and elements in auto-versioning modes. See 3.27 - Allowing Unknown Items in an XML
Instance for more details.
• Modifications to allow for C++11 deprecation of std::auto_ptr.
• Added methods c_binary::get_as_hex_string() and
c_binary::get_as_base64_string().
• Enhanced support for non-C locales for converting strings to xs:float values.
• Added DECIMAL_INIT_MODE section to lmxuser.h to allow choice between initialising
xs:decimal values using C++ strings or doubles.
• Added lmx_assert_is_occurs_ok() macro to allow user control of what happens if
insufficient items are present for marshalling.
• Fixed application of pattern facets to a whole xs:list construct as opposed to its individual
members.
• Use of ptrdiff_t fixed to std::ptrdiff_t.
7
• To accommodate the increasing number of versions of Visual Studio and GCC we have decided to
only ship the source code for the Runtime Supporting Software and allow users to create their own
binaries for this code. This allows greater user flexibility and complete cross-platform solutions.
• Added ability to install external character transcoders. See 3.17 - Adding External Character Set
Transcoders.
• Added the -expose-storage option. This allows writing of more generic code.
12
1.7 - Version History
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• Improved handling of serialisation of floating point numbers on non-Windows platforms.
6.4
• Improved formatting of fractional seconds values in times and durations.
• is_occurs_ok() method can now collect better debugging information.
• Added generation of the check() method, effectively a deep version of the is_occurs_ok() method.
• Added -no-check-code flag.
6.3
• The Windows library naming convention has been changed from including the compiler version in the
name (e.g. vc11) to the IDE version (e.g. vs2012).
• Fixed HTML documentation of attribute cardinality.
• Reduced rounding errors when round-tripping float values through c_decimal.
• c_binary::operator [] const and at() const now return const unsigned char & to allow the return value
to be used in functions like fwrite() etc.
• Changed the internal variable naming convention.
• Classes that contain binary members now have added a non-const method for accessing the binary
member.
6.2
• Added the -enums-per-type flag option to generate separate C++ enums for each schema type
that has xs:enumeration facets defined.
• Added -narrow-strings flag to generate #defines suitable for instructing C++ compiler to use
narrow strings for Unicode strings.
• Added -prefix-enumeration and -prefix-enumeration-name to improve control of
naming of c++ enumerations associated with xs:enumeration facets.
• Added generation of const version of getany_attributes() method.
• Added facility to change whether seconds in types containing time are represented using C++ float or
double.
• Includes binaries for Microsoft VS 2012.
6.1
• The generated operator == () methods now take into account polymorphism of types.
• The generated code has been modified so that try/catch blocks are not required. Exception safety is
maintained.
• Added the -prefix-declash flag. See 3.22.1 - Handling Multiple Independent Schemas for more
details.
• Fixed Micro Format handling (see 3.26 - Specifying Micro Formats) when schema specifies a type to
be an xs:list type.
• If a project build environment defines LMX_XML_ALWAYS_ESCAPE_GT to be 1 when the
lmxtypes.cpp source code is compiled then > will always be escaped to &gt; (Professional
Edition required).
6
• Refactored the interface between the generated objects and the XML parser/writer.
• Restored the ability to allow globally stored objects that have string-based default values.
1.7 - Version History
13
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• Improved conditional automatic inclusion of xml.xsd namespace attributes.
5.7
• Improved support for choice of CamelCase or underscore_separated C++ names. See -naming flag.
5.6
• Convenience methods now take an optional pointer to an error object to allow capture of enhanced
error information. See 2.3 - Unmarshaling (simple form) for more information.
• Enhancements have been made to enable better control of attribute layout when formatting. See
3.25 - XML Output Format Customization for more information.
• Some concrete implementations of key template methods have been placed in the binary libraries to
reduce the size of binaries on some platforms.
5.5
• Enhancements to the Annotated XML Example (AXE) parser.
• Added -release-switch flag.
• Added on_lmx_is_occurs_ok() snippet event handler. (See 3.14.1 - Augmenting Generated
Classes with Snippet Event Handlers.)
• Added generation of assign_XYZ( size_t index, T * p ) methods for complexTypes.
• Added c_xml_reader::ignore_element(...) method mainly to be used with the
on_lmx_alt_unmarshal_element_XYZ() snippet event handler to enable ignoring the
contents of elements the user is not interested in.
5.4
• Added support for specifying the XML data format using Annotated XML Example (AXE™) format.
• Better support when using empty prefixes for C++ identifiers.
5.3
• Global setting of the XML output formatting is supported. (See 3.25 - XML Output Format
Customization.)
• Added on_lmx_alt_unmarshal_attribute_XYZ(),
on_lmx_alt_unmarshal_element_XYZ(),
on_lmx_alt_marshal_attribute_XYZ() and
on_lmx_alt_marshal_element_XYZ() snippet event handlers. (See 3.14.1 - Augmenting
Generated Classes with Snippet Event Handlers.)
• Added the c_xml_writer::disable_ns_map_output() method to disable output of the
xmlns namespace attributes. See 3.20 - Disabling Output of XML Namespaces for more.
• Added additional variants of the unmarshal_partial(...) template method. See
3.2.3 - Unmarshalling Schema Fragments.
• Documented the unmarshal_find() template functions. See 3.3 - Unmarshalling a Sub-element
within an XML Instance.
• Refactored internal marshalling and unmarshaling interface to c_xml_writer and
c_xml_reader.
5.2
14
1.7 - Version History
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• Added the -micro-format flag to allow support of micro formats. See 3.26 - Specifying Micro
Formats.
• Added generation of insert_XYZ( size_t index, T * p ) methods for complexTypes.
• Added generation of non-const get_XYZ(???) methods for compound simpleTypes such as
xs:date and micro formats to enable easier access.
• Fixed code generation bug for complexTypes containing fixed simpleContent.
5.1
• Added the -pattern-out flag to allow specification of custom output converters for data types
requiring xs:pattern specific output formatting. See 3.24 - Custom Pattern Facet Output Formatting.
• Changed name of .isset() methods on classes to .is_value_set() to avoid problems when other code
defines an isset() macro (e.g. HP aCC).
• Corrected handling of decimal number comparisons that contained 0s as final fractional digits.
• Add binaries for g++ 4.2.3.
• Includes support for Microsoft VS 2010.
• Enhanced Linux install.pl script to allow unattended install.
5
• Re-factored the inner workings of the generated code. The generated API to use the generated code
remains the same.
• Some short accessor methods can now be generated inline as part of the class definition rather than in
the .cpp file.
• Added -max-inline flag.
• Added -error-fast flag.
• Added -no-local-classes flag.
• Updated regex code derived libxml2 2.7.7.
4.2
• Added improved support for handling and subsequently unmarshalling contents of xs:any.
• Added on_lmx_unmarshal_attribute_XYZ() and
on_lmx_unmarshal_element_XYZ() snippet event handlers.
• Added option to specify the base class of a named class. See -class-base flag.
• Add option to be able to specify C++ include guard string. See -include-guard flag.
• Default C++ include guard text now includes the C++ namespace.
• Modified generated code to accommodate GCC version 4.3.3 and 4.4.1. warnings.
• Added a Perl based install script (install.pl) for Linux platforms.
4.1
• Added option to generate equality (and inequality) operators. See -eq flag.
• Added exec-pre-all, exec-post-all and exec-post-all-error flags.
• Added option to create a makefile fragment. See -makefile flag.
• Significantly sped up the compile time of some large schemas.
4.0
• GUI interface for Linux added, called glinmx.
1.7 - Version History
15
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• -expose-containers flag added, allowing more direct access to the containers (e.g. std::vector)
in the objects.
• Changed defaulting of LMX_NARROW_URI_STRINGS so that it is defaulted to the value of
LMX_NARROW_UNICODE_STRINGS rather than 0.
3.10
• Enhancements made so that string classes that have different method names to std::basic_string can be
used.
• Supporting Software and generated code changed so that GCC can be run with the
-Wunused-parameter flag without causing warnings.
• Wide-string based path names supported in generated convenience methods.
3.9
• Added the -naming YYY flag which specifies the naming convention to be used for names in the
generated code. If camel is specified, LMX will attempt to convert names generated in the code to
CamelCase. If underscore is specified, LMX will attempt to make names underscore separated.
• Added -file-ext-snippets YYY flag.
• Reworked the output converters code so that alternative output converters can be more readily used
with the Standard Edition of LMX.
• WinLMX now automatically checks to see if there are any new versions of LMX available.
3.8
• Added c_soap and c_winhttp classes to aid implementing web services. See 3.11 - Use with
Web Services for more information.
• Added the ability for users to customize the behavior of generated classes using snippet event
handlers. These call user defined methods during the marshal and unmarshal processes. See
3.14.1 - Augmenting Generated Classes with Snippet Event Handlers for more information.
• Added the -alt-xml-reader and -alt-xml-writer flags. See the flags section for more
information.
• To provide custom error feedback, primarily for snippet event handlers, the lmx::elmx_error
enumeration has had added to it the enumeration values ELMX_USER_DEFINED_1 to
ELMX_USER_DEFINED_9.
• Added -lmx-include-path flag to allow specifying the directory where the LMX header files
are stored.
• Added the -check-is-occurs-ok-on-marshal flag.
• Added tlmx_uri_string typedef so that URI string can have an independent type to Unicode
strings.
• Added #ifndef LMX_NO_WSTRING sections to the supporting software source code to allow
conditional removal of std::wstring from a build.
3.7.1
• Fixed handling of hierarchies of substitution groups.
• Auto-versioning of substitution groups now has an explicit option. Previously auto-versioning of
substitution groups was the default (but non-standard) behavior. See -autover-subst-groups
flag.
3.7
16
1.7 - Version History
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• Added options to specify the file extensions of the generated files. See -file-ext-cpp and
-file-ext-h flags.
• Added option to specify that the C++ enums associated with schema enumeration facets should be
local to a class. See -local-enums flag.
• Added options to specify that documentation from the schema (included in xs:documentation
elements) should be output in the .h and/or HTML file. See -doc-in-h and -doc-in-html flags.
• A flag has been added to cause any warnings generated during code generation to return the same
program error code returned by errors. See the -werror flag.
• The Windows DLLs now have version information in them.
• A bug associated with determining the used XML namespaces has been fixed.
• Added option to suppress generation of root class. See -no-root-class flag.
• Added option to not generate code for insert/delete/clear operations on items that can occur more than
once. See -no-container-ops flag.
3.6
• Added options to specify suffixes to be added to the code names of attributes, elements, types, and
groups. See -suffix-attribute, -suffix-element, -suffix-type and
-suffix-group flags.
• Corrected handling of xmlns="" idiom according to XML Namespaces 1.1.
• Added static debug_error object to aid the debugging process.
• Added one .cpp per schema file generation mode. See -cpp-per-schema flag.
• Added ability for developers to augment the generated classes with their own code. See -snippets
flag.
• More accurate HTML documentation generation.
3.5
• Added options to control whether only marshaling or only unmarshaling code is generated. See flags
-no-marshal and -no-unmarshal.
3.4
• Added polymorphic behavior for XSD extension and restriction types. Use no-polymorphic to
switch off polymorphic behavior.
3.3
• Adopted the Windows XP look-and-feel for WinLMX.
• Corrected some issues with simpleContent types.
• Changed company name.
3.2.5
• Fixed a dependency issue in the generated C++ code between C++ declarations and definitions for
attributes in restricted simple content.
3.2.4
• Improved support for QNames.
1.7 - Version History
17
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
3.2.3
• Improved support for multiple instances of simple type lists.
• Fixed handling of fractional seconds less than 10.0 in time, datetime and duration types.
3.2
• Further enhancements to WinLMX's method prototype viewing feature made.
• Improvements in the way xs:include files are handled.
• Corrected handling of multiple occurrences of information items with fixed or default value
constraints.
3.1
• Further enhancements to WinLMX's method prototype viewing feature made.
• Fixed an XML namespace handling issue.
• Modified c_read_file so that reading XML from files is faster in applications generated with Visual
Studio 2005.
• Added the ability to set schema location attributes in generated XML.
• Formerly the validation of the length facets for xs:string types was only correct if they were mapped
to std::wstring. This has now been corrected so that the validation is correct irrespective of whether
the xs:string types are mapped to std::wstring or std::string.
3.0
• WinLMX enhanced. The base names of variables and methods can now be edited using the interface.
• Editing of a schema can be invoked from WinLMX.
• The generated HTML documentation can be displayed from WinLMX.
3.0 Beta
• XML Parser refactored.
• Additional marshal and unmarshal methods generated to make it more convenient to marshal and
unmarshal to/from files and memory.
• Test framework output (-tframework) simplified.
• WinLMX enhanced.
• If multiple schemas are treated as global, each such schema has its own XML namespace URI to
namespace prefix map.
2.11
• Nested C++ namespaces can be specified.
• Removed use of global_event_map in 'global' unmarshal functions.
• Added provision for extension from xs:anyType.
• Improved parser efficiency.
• User has control over XML namespace prefixes used via -ns-prefix-map command-line flag.
2.10
• Primarily an upgrade to support the Basic Edition license.
• Includes code for VS2005 libraries.
18
1.7 - Version History
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• Enforce the "C" locale for string to float conversions in locale aware implementations.
• Unicode handling extended beyond the Basic Multi-lingual Plane to full Unicode range (0x-0x10ffff).
• Moved position of generated braces!
2.9
• Improved 'One .h/.cpp file per class' handling.
• insert_NAME() methods generated for interfacing with collections.
• Mapping of namespace prefix flags supported (see -ns-prefix-map).
• Improved mapping of punctuation characters in enumeration identifiers to C++ names.
• Unions are no longer anonymous, leading to improved portability on compilers such as HP-UX aCC.
2.8
• Can specify wide-char file names when opening XML files for reading.
• Added string converters for std::string to std::string and std::wstring to std::wstring. These
non-converters make it easier to operate in a environment where _TCHAR type mechanisms are being
used.
• Added features to help MS IntelliSense.
• Added work-arounds for IBM Visual Age Compiler.
2.7
• Code can be generated from an XML external DTD definition - see 3.6 - Selecting the Input File Type
(XSD, WSDL, DTD).
• A W3C Schema embedded in a WSDL file can be automatically parsed - see 3.6 - Selecting the Input
File Type (XSD, WSDL, DTD).
• Modified header file layout so that it is possible to switch between wide and narrow strings without
requiring the supporting software source code.
• Fixed a bug in the generated code that did a double memory allocation when calling c_class
&get_class( size_t index ), causing a memory leak.
• Fixed problem in the generated code when parsing XML instances from schemas that have multiple
global elements and have multiple namespaces associated with them.
• Added lmx_assert() macro to allow customisation of how assertions are handled. The macro is
currently mapped directly to assert(), but could be modified to allow throwing exceptions in release
code if desired.
• Character entities in entity definitions are now expanded immediately when first parsing, rather than
when the entity text is used.
2.6
• Modified constructor initialisation order to remove GCC -Wall warnings.
• Added at() and get( unsigned char [], size_t ) methods to c_binary.
• Added functions to convert between wide and narrow strings (and vice versa).
• Added some defaults to switch statements to address compiling with GCC -Wall.
• Fixed problem with single quote appearing in attribute values when parsing xs:any elements.
• Fixed problem with comments that contained no space after the <!-- sequence and contained a /
character (e.g. <!--My comment with / char -->).
2.5
1.7 - Version History
19
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
• Customized naming of methods now supported - see 3.12 - Naming of Methods and Variables.
• Union simpleType as base of simpleContent now supported.
• Improve recording of code generator options in generated code when WinLMX used.
• c_xml_reader destructor made virtual and unused variable removed from c_big_int::operator >.
2.4
• Added -multi-file flag to allow one class per file operation.
• WinLMX allows saving of default project settings for easier, more consistent project start up.
• Fixed issue with -output-defaults.
2.3
• Added -no-nested-classes flag to allow specifying that nested classes should not be used.
1.8 - Supported XML Schema Features
LMX XML C++ Databinder offers industry leading schema feature support for C++ to XML data binding as
shown in the following table:
XML Schema Feature
Schema Handling
Import
Include
Redefine
Chameleon Design
Qualified / Unqualified Elements
Qualified / Unqualified Attributes
Supported?
Yes
Yes
Yes
Yes
Yes
Yes
Types
Empty Types
simpleType
complexType / simpleContent
complexType / complexContent
Nillable
Simple Types & Facets
All Simple Types Supported
Lists
Unions
enumeration
Allow reading/writing using enumerated values
fractionDigits
length
maxExclusive
maxInclusive
20
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Partial1
Yes
Yes
Yes
Yes
Yes
Yes
1.8 - Supported XML Schema Features
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
minExclusive
minInclusive
maxLength
minLength
pattern
totalDigits
whiteSpace
List length facets
default values
fixed values
xs:anyAttribute
xs:anyAttribute namespace validation
xs:anyType
Automatic entity substitution (&amp; etc)
Expansion of DTD defined entities
complexType / complexContent
Sequence
Choice
All
Capture / control order of xs:all
Anonymous Compositors
xs:any
xs:any namespace validation
Mixed
Extension
Restriction
Polymorphic Extension using xsi:type
Polymorphic Restriction using xsi:type
Groups
Recursive Definitions
Cardinality
Optional
Mandatory
Multiple (0-n, 1-n, m-n, m-unbounded etc.)
Miscellaneous
Attribute Groups
Substitution Groups
Name clash prevention
UTF-8, UTF-16 (BE & LE), UCS2 (BE & LE), ISO-8859-1
Generate code from XML external DTDs
Generate code from Schemas embedded in WSDL files
1.8 - Supported XML Schema Features
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Partial2
Yes
Mostly3
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
21
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Notes:
1All XML schema Union simpleTypes are stored as strings.
2The contents of an XML schema Mixed type is stored as a string,
including any additional markup.
3Some corner cases that don't map cleanly to C++, such as an element that
has a child element that is restricted by using a type that is a restriction of
the original type, are not supported.
22
1.8 - Supported XML Schema Features
2 - Quick Start
Having installed and licensed your LMX product, you will want to generate code. This section gives a brief
introduction to this and will probably be sufficient for the majority of your use of the tool. For more detailed
information, see section 3 - In More Depth.
2.1 - Code Generation
The LMX code generator can be run on Windows® and Linux platforms. On the Windows platform LMX is
available as both a GUI and a command-line version.
The generated code is cross-platform.
2.1.1 - Code Generation Using the Windows Interface
The Windows® Interface version of LMX makes it easy to use the LMX tool. The Windows version is called
WinLMX.
When you start WinLMX, you will see a window displayed that contains a tabbed dialog. You can specify
what you want WinLMX to do by selecting the various tabs, and completing the dialog items that are
displayed.
The main aspect of configuration is selecting the files to be compiled, and the names of the output files. This
is configured using the left-most tab labelled 'Schema Files'. Enter the base schema file in the top-most edit
box, and the root of the output files (e.g. 'File' to generate 'File.h' and 'File.cpp') in the bottom-most edit box.
The files that the schema imports should be entered into the middle list box by pressing the associated 'Add...'
button.
LMX can parse both W3C Schema (XSD) and XML (external) DTD files. LMX can also locate and parse
W3C Schema definitions embedded in WSDL files. The parsing that LMX performs depends on the file
extension of the base schema file. See 3.6 - Selecting the Input File Type (XSD, WSDL, DTD) for more
information.
When you have completed configuring WinLMX you can save the configuration using the 'File' menu.
To generate code for a configuration, press the 'Compile...' button at the bottom of the window. This will
automatically select the 'Compilation Results' tab and display the results of the compilation.
Note that when a configuration file is loaded into WinLMX, the current working directory is set to the path of
the configuration file. This allows you to specify relative file names for the schema and output files.
In addition to using the 'File' menu to open configuration files, WinLMX also supports drag and drop.
The configuration files created by WinLMX can be used by the command-line version of LMX. This allows
you to develop configuration files using the windows version, and use that configuration in your build strategy
with the command-line version. (If the command-line version of LMX is called with a single argument that
ends in '.lmxprj' the tool assumes that it is a configuration file. Alternatively, the '-f' command-line option can
2 - Quick Start
23
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
be used to specify a configuration file.)
If WinLMX's default project settings are not suitable for your development environment, configure the
settings as you would like them and then select the 'File->Save Default New Project' menu item. These
settings will then be used when a new project is started, or a '.xsd' file is dragged over WinLMX.
When using a licensed version LMX needs to know the location of the license file. See 1.6 - Licensing &
Purchase for further information.
2.1.2 - Code Generation Using the Windows (DOS) Command Line
Version
The LMX code generator can be run from a (DOS) command line prompt. This is useful when LMX is used
as part of a batch build process such as a nightly build. The basic command line syntax is:
lmx [flags] primary-xsd-file *[++ additional-global-xsd-file] *[+ additional-library-xsd-file] outpu
The flags are optional and are described in 3.8 - Command-line Flags.
The primary-xsd-file is the name of the file that contains the main schema definition. See
3.6 - Selecting the Input File Type (XSD, WSDL, DTD) for more information on selecting between W3C
Schema files, WSDL files and XML external DTD files.
If the primary schema definition references other schemas, these additional schema files are specified on the
command line by including the + character, a space, and then the name of the file. Any number of additional
schema files can be specified in this way.
If the additional schema files contain global elements that you would like to treat as possible XML instance
document elements, then instead of using the + character in front of the schema name as above, use the ++
character sequence. This can be useful when you are generating code for two or more schemas that share
common imported schemas.
The last argument specifies the names of the C++ files into which the generated code is to be placed. The
LMX code generator will append .cpp to the name specified for the C++ source file, and append .h to the
name for the header file.
For example, the command line:
lmx my_xsd.xsd my_xsd_code
will compile the schema specified in my_xsd.xsd and generate the files my_xsd_code.h and
my_xsd_code.cpp.
If my_xsd.xsd refers to other schemas, an example command line would be:
lmx my_xsd.xsd + my_xsd_lib_1.xsd + my_xsd_lib_2.xsd my_xsd_code
This will combine the schemas specified in my_xsd.xsd, my_xsd_lib_1.xsd and
my_xsd_lib_2.xsd, and generate the files my_xsd_code.h and my_xsd_code.cpp.
24
2.1.1 - Code Generation Using the Windows Interface
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
If two schemas refer to a common schema, one way to generate code is to use a command line similar to:
lmx my_xsd_1.xsd ++ my_xsd_2.xsd + my_xsd_lib.xsd my_xsd_code
(N.B. notice the use of ++ rather than just +.) In this case, in addition to allowing global elements in
my_xsd_1.xsd to be document level elements in an XML instance, global elements in my_xsd_2.xsd
will also be treated as candidate document level elements.
LMX has support for command-line shells that provide wildcard expansion (LMX itself does not perform
wildcard expansion). A command-line of the form below will treat all files matching my_xsd_*.xsd as
additional-library-xsd-files. Only files not already mentioned on the command-line are treated
as additional-library-xsd-files. Therefore, in the example below, my_xsd_1.xsd will not be
treated as an additional-library-xsd-file even though it matches the wildcard, because it is
explicitly named earlier on the command-line.
lmx my_xsd_1.xsd + my_xsd_*.xsd my_xsd_code
When using a licensed version the LMX executable needs to know the location of the license file. See
1.6 - Licensing & Purchase for further information.
2.1.3 - Code Generation Using the Linux Command Line Version
The operation of the Linux command-line version is the same as the Windows (DOS) command-line version
as described in 2.1.2 - Code Generation Using the Windows (DOS) Command Line Version, except that the
executable is called 'linmx'. Hence the basic command-line syntax becomes:
linmx [flags] primary-xsd-file *[++ additional-global-xsd-file] *[+ additional-library-xsd-fi
For example:
linmx my_xsd.xsd my_xsd_code
When using a licensed version the LMX executable needs to know the location of the license file. See
1.6 - Licensing & Purchase for further information.
2.2 - C++ Compiling and Linking
An LMX project consists of three different types of source code.
• Code generated by the LMX code generator.
• The LMX Runtime Supporting Software. This is code common to all LMX projects and consists of
the low level XML parser and type classes. The .cpp source files for this code are located in the
supporting-software/src sub-directory of the installation, and the .h header files are located
in the supporting-software/include sub-directory.
• The code written by you that interfaces to the LMX generated code.
The simplest solution is to include the source code of all three types of code directly into your project. This is
the quickest and easiest solution for the evaluation phase. However, you may choose to compile the different
types of code separately into libraries or Windows DLLs. The following sections give guidance on how to do
2.1.2 - Code Generation Using the Windows (DOS) Command LineVersion
25
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
this, although note that there are many different ways in which you may choose to partition the code.
2.2.1 - Compiling the Runtime Supporting Software Source Code
As mentioned above, the Runtime Supporting Software is common to all LMX based projects. The .cpp
source files for this code are located in the supporting-software/src sub-directory of the installation,
and the .h header files are located in the supporting-software/include sub-directory.
The set of source files you need for your build will depend on your requirements:
• lmxparse.cpp contains the low-level XML parser and related code. Almost all LMX projects will
need this file.
• lmxtypes.cpp contains the LMX type classes. Most projects will use this file, although there is
the option to provide your own alternative.
• lmxregex.cpp and lmxunicode.cpp are needed if you have not disabled regular expression
handling in your build and you don't want to use your own variant of these (see 3.23 - Pattern Facet
Handling Customization).
• lmxsoap.cpp may be included if you wish to use LMX's SOAP functionality.
• lmxwinhttp.cpp can optionally be included if you wish to use LMX's Windows HTTP
functionality. (Note that lmxwinhttp.cpp is Windows specific and will not compile on other
platforms. We recommend using libraries such as Linux's libcurl on other platforms.)
During the build process, these files will require access to the Supporting Software include files, typically
located in the supporting-software/include sub-directory of the installation. (Note that you might
want to move the various files out of their default location when building a library so that you can more easily
place them under version control.)
The exact process of building the source code will depend on your platform. The simplest approach is to
directly include the various source files in with your main project or makefile. (Note that if you use the Visual
Studio DLL C Runtime, the LMX code assumes by default that ultimately you will put the supporting
software code in its own DLL. As a result, if you include it directly in your project you will get LNK4217:
locally defined symbol warnings (or similar). In initial evaluation you can either ignore these
warnings, or include LMX_WANT_DLL=0 in the Preprocessor Definitions of your project settings.)
The Windows version of LMX includes a batch file called build-libs.bat in the
supporting-software/src sub-directory of the installation that can be used to build the source code
into static .lib files and dynamic .dll files. This batch file should be run inside a Visual Studio Command
Prompt window. If you choose to build the files within the default c:\Program Files\LMX directory
location then you will need to run the Visual Studio Command Prompt as an Administrator. Read the
comments in build-libs.bat for further details on how to customize the build to your needs.
Note that to build the supporting software using build-libs.bat on Visual Studio 2015, 2017 & later,
you need to ensure that the Windows SDK component of the Visual Studio installation process has been
included. (If not you can modify your Visual Studio configuration by re-running the Visual Studio installer.)
26
2.2 - C++ Compiling and Linking
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
As you may notice from the build-libs.bat batch file, if you wish to build the Supporting Software
Source Code into your own DLL, you need to configure your project to define LMX_MAKE_DLL (i.e.
"Project->Properties->Configuration Properties->C/C++->Preprocessor->Preprocessor Definitions" includes
LMX_MAKE_DLL;.) and also define LMX_RT_SOURCE (the value is not important in this latter case). (For
versions prior to version 6.1 of LMX or if you only want part of the LMX Supporting Software in a DLL,
instead of defining LMX_RT_SOURCE, define LMX_PARSE_SOURCE, LMX_TYPES_SOURCE, and
LMX_REGEX_SOURCE.)
If the Supporting Software Source Code has been built into a DLL and you which to combine the DLL with
other code then you need to configure the project that references the DLL to define LMX_WANT_DLL to 1.
The Linux version of LMX includes a makefile called MakefileLMXlib that can be used as a starting point
for writing a makefile for building the Supporting Software into a static or shared library.
2.2.2 - Compiling the Generated Code
The code generated by LMX will need access to the Supporting Software include files, typically located in the
supporting-software/include sub-directory of the installation, in order to compile.
If you wish to build the generated code into a DLL, you need to configure your project to define
LMX_MAKE_DLL (i.e. "Project->Properties->Configuration Properties->C/C++->Preprocessor->Preprocessor
Definitions" includes LMX_MAKE_DLL;).
If the generated code has been built into a DLL and you which to combine the DLL with other code then you
need to configure the project that references the DLL to define LMX_WANT_GEN_IMPORT_DLL to 1.
2.2.3 - Compiling Your Code to Use LMX Code
To compile your code so that it can use the LMX generated code and Runtime Support Software your code
will need to have access to the generated header file. For example, if the LMX generated header file is
generated.h, then you will need to include the following lines in your .cpp file:
#include "generated.h"
If your code performs 'advanced' forms of marshaling and unmarshaling by calling the generated
marshal() and unmarshal() methods directly (see 3.5 - Marshaling (advanced forms) and
3.2 - Unmarshaling (advanced forms)), then you will need to include both the LMX generated .h file and the
LMX parser header file; lmxparse.h. For example:
#include "generated.h"
#include "lmxparse.h"
(If you choose to configure LMX to generate multiple header files (e.g. one header file per class), you should
substitute the line #include "generated.h" mentioned above as appropriate.)
The generated header file will need access to the Runtime Supporting Software header files, typically located
in the supporting-software/include sub-directory of the installation.
2.2.1 - Compiling the Runtime Supporting Software Source Code
27
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
If the Runtime Supporting Software Source Code has been built into a DLL and you wish to combine it with
your code then you need to configure your code sub-project to define LMX_WANT_DLL to 1.
If the generated code has been built into a DLL and you wish to combine it with your code then you need to
configure your code sub-project to define LMX_WANT_GEN_IMPORT_DLL to 1.
2.3 - Unmarshaling (simple form)
Unmarshaling is the process of reading in XML and converting it to C++ objects. LMX can parse UTF-8,
UTF-16 (big and little endian), ISO-10646-UCS-2 (big and little endian), ISO-8859-1 (aka Latin-1), and
US-ASCII. The source XML can either be in an in-memory data buffer, or a file.
For the class generated at the top of the class hierarchy (e.g. c_root) LMX generates convenience functions to
help the marshaling and unmarshaling process. These functions are not generated for other classes in order to
reduce generated code size. This section first presents the method for unmarshaling using these convenience
methods. Later in the section an alternative method of unmarshaling is presented, which can be used with
other classes. Additionally, advanced unmarshaling techniques are presented in 3.2 - Unmarshaling (advanced
forms).
To read XML from a std::string, code similar to the following can be used:
// Allocate somewhere to store any error code that the unmarshaling may
// produce
lmx::elmx_error result;
lmx::s_debug_error error_detail;
// Construct an object from the XML contained in the std::string
// xml_message_data_string and store the result code of the
// unmarshaling in the 'result' variable.
const c_generated_xsd_root_class top_object(
xml_message_data_string,
&result,
&error_detail );
// If the 'result' variable indicates that all is OK, interact with the
// object.
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
The error_detail argument is optional, so the following code is also valid if detailed error information is
not required:
lmx::elmx_error result;
const c_generated_xsd_root_class top_object(
xml_message_data_string,
&result );
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
To read XML from an in-memory buffer, where xml_message_data_buffer is of type const char
28
2.2.3 - Compiling Your Code to Use LMX Code
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
* or similar, code similar to the following can be used:
// Allocate somewhere to store any error code that the unmarshaling may
// produce
lmx::elmx_error result;
lmx::s_debug_error error_detail;
// Using error_detail is optional
// Construct an object from the XML contained in the memory located at
// xml_message_data_buffer and store the result code of the unmarshaling in
// the 'result' variable.
const c_generated_xsd_root_class top_object(
xml_message_data_buffer,
number_of_bytes_in_buffer,
&result,
&error_detail );
// error_detail is optional
// If the 'result' variable indicates that all is OK, interact with the
// object.
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
To read XML from a file, code similar to the following can be used (which doesn't use the option to retrieve
detailed error information):
// Allocate somewhere to store any error code that the unmarshaling may
// produce
lmx::elmx_error result;
// Construct an object from the XML contained in the file 'c:/myxml.xml'
// and store the result code of the unmarshaling in the 'result' variable.
const c_generated_xsd_root_class top_object( "c:/myxml.xml", &result );
// If the 'result' variable indicates that all is OK, interact with the
// object.
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
Convenience methods are also generated to unmarshal instances corresponding to xs:any constructs that
have previously been unmarshalled from an XML instance. This can be done using code similar to the
following:
// Allocate somewhere to store any error code that the unmarshaling may
// produce
lmx::elmx_error result;
// Construct an object from the XML corresponding to an xs:any instance
// unmarshalled from a previous unmarshaling run.
const c_generated_xsd_root_class top_object(
my_previously_unmarshaled_object.get_any(),
&result );
// If the 'result' variable indicates that all is OK, interact with the
// object.
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
2.3 - Unmarshaling (simple form)
29
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
where the get_any() method is described in 2.5.10 - Singular xs:any C++ Interface and 2.5.12 - Multiple
xs:any C++ Interface.
If it is not convenient to unmarshal at the time the object is constructed (for example, if the object is a static
global store of the application's user preferences) then the following methods can be used.
To read XML from a std::string, code similar to the following can be used:
// Create an instance of the class
c_generated_xsd_root_class top_object;
// ... additional code ...
// Use the unmarshal method, giving the std::string to read from
lmx::elmx_error result = top_object.unmarshal( xml_message_data_string );
// If unmarshaling was successful, interact with the object
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
To read XML from an in-memory buffer (of type const char * or similar), code similar to the following
can be used:
// Create an instance of the class
c_generated_xsd_root_class top_object;
// ... additional code ...
// Use the unmarshal method, giving the details of the memory to read from
lmx::elmx_error result = top_object.unmarshal( xml_message_data_buffer,
number_of_bytes_in_buffer );
// If unmarshaling was successful, interact with the object
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
To read XML from a file:
// Create an instance of the class
c_generated_xsd_root_class top_object;
// ... additional code ...
// Use the unmarshal method, giving the details of the file to read from
lmx::elmx_error result = top_object.unmarshal( "c:\\myxml.xml" );
// If unmarshaling was successful, interact with the object
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
If you wish to unmarshal from an object for which the convenience functions were not generated, then you
can use the unmarshal template functions that are defined in lmxparse.h. In this case the code form is:
- for XML in a std::string:
30
2.3 - Unmarshaling (simple form)
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
// Create an instance of the class
c_generated_class an_object;
// Use the template unmarshal method, giving the details of the memory to
// read from
lmx::elmx_error result = lmx::unmarshal( &an_object,
xml_message_data_string );
// If unmarshaling was successful, interrogate the object
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
- for XML in memory:
// Create an instance of the class
c_generated_class an_object;
// Use the template unmarshal method, giving the details of the memory to
// read from
lmx::elmx_error result = lmx::unmarshal( &an_object, xml_message_data_buffer,
number_of_bytes_in_buffer );
// If unmarshaling was successful, interrogate the object
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
- for XML in a file:
// Create an instance of the class
c_generated_class an_object;
// Use the unmarshal method, giving the details of the file to read from
lmx::elmx_error result = lmx::unmarshal( &an_object, "c:\\myxml.xml" );
// If unmarshaling was successful, interrogate the object
if( result == lmx::ELMX_OK )
{
// Work with unmarshalled data...
For information on how to diagnose errors, particularly during debugging, when using these convenience
methods, see 3.16.4 - Collecting Debug Error Information when using Convenience Methods.
As mentioned previously, additional unmarshaling techniques are described in 3.2 - Unmarshaling (advanced
forms).
2.4 - Marshaling (simple form)
Marshaling is the process converting the C++ object content into XML data.
For the class generated at the top of the class hierarchy (e.g. c_root) LMX generates convenience functions to
help the marshaling and unmarshaling process. These functions are not generated for other classes in order to
reduce generated code size. This section first presents the method for marshaling using these convenience
methods. Later in the section an alternative method of marshaling is presented, which can be used with other
classes. Additionally, advanced marshaling techniques are presented in 3.5 - Marshaling (advanced forms).
2.4 - Marshaling (simple form)
31
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
A typical section of marshaling code for writing the XML to a string looks as follows:
// Create an instance of the class
c_generated_xsd_root_class top_object;
// ... Populate and manipulate top_object ...
// Allocate a string in which to store the marshalled XML
std::string my_string;
// Marshal the object to the string
top_object.marshal( &my_string );
The equivalent code for writing to a file is as follows:
// Create an instance of the class
c_generated_xsd_root_class top_object;
// ... Populate and manipulate top_object ...
// Marshal the object to the file "c:\myxml.xml"
if( top_object.marshal( "c:\\myfile.xml" ) != lmx::ELMX_OK )
{
// Something went wrong!
}
If you wish to marshal an object for which the convenience functions are not generated, then you can use the
marshal template functions defined in lmxparse.h. In this case, the code for marshaling to string is:
// Create an instance of the class
c_generated_class an_object;
// ... Populate and manipulate an_object ...
// Allocate a string in which to store the marshalled XML
std::string my_string;
// Marshal the object to the string
lmx::marshal( an_object, &my_string );
and the code for marshaling to a file is:
// Create an instance of the class
c_generated_class an_object;
// ... Populate and manipulate an_object ...
// Marshal the object to the file "c:\myxml.xml"
if( lmx::marshal( an_object, "c:\\myfile.xml" ) != lmx::ELMX_OK )
{
// Something went wrong!
}
As in the unmarshalling case, for information on how to diagnose errors, particularly during debugging, when
using these convenience methods, see 3.16.4 - Collecting Debug Error Information when using Convenience
Methods.
32
2.4 - Marshaling (simple form)
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
As mentioned previously, additional marshaling techniques are described in 3.5 - Marshaling (advanced
forms).
2.5 - Accesing the Data in the Generated Classes
The C++ class interface to a generated item is independent of whether the item is an element or an attribute.
Hence the difference between these two forms is abstracted away. The term 'item' is used here to refer to
either an element or an attribute.
An interface to an item depends in one respect on whether an item is:
• empty.
• a simpleType.
• a complexType.
• xs:anyAttribute.
• xs:any.
The other aspect that affects the interface is whether the item is:
• Singular - minOccurs = maxOccurs = 1, or a required attribute.
• Optional - minOccurs = 0, maxOccurs = 1, or a normal attribute.
• Multiple - maxOccurs > 1 (or unbounded), or a list.
A further consideration for optional items is whether the item is in an xs:choice construct.
There are also special functions to interface with enumerated values, and additional functions to interface to
complex types that are polymorphic.
The following sections describe the interface for each of these combinations. Throughout the discussion, the
string NAME is used to represent the name given to the item.
An element of type xs:any will by default have the NAME part of the method name set to any. However, if
there is more than one xs:any element in a complex type, or another element is named any, numbers will be
appended to the name to provide differentiation.
2.5.1 - Singular Empty Type C++ Interface
No interface is generated.
2.5.2 - Optional Empty Type C++ Interface
If the item is not in an xs:choice, the following method is generated:
void set_NAME();
Marks the item NAME as present.
bool isset_NAME() const;
2.5 - Accesing the Data in the Generated Classes
33
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Returns true if the item NAME is present, and false otherwise.
void unset_NAME();
Marks the item NAME as not present.
If the item is allowed in an xs:choice, but not present, the getchosen() method returns <class
name>::e_choice_not_set.
2.5.3 - Singular Simple Type C++ Interface
Note that for the interface to list types, see Multiple Singular Types.
<return type> get_NAME() const;
Return either the value of item NAME or a reference to NAME depending on the particular type.
<non-const ref to type> get_NAME();
Generated if the type of NAME is a compound class such as for xs:date. Returns a non-const
reference to NAME.
void set_NAME( simple_type );
Set item NAME to the specified value.
If the type has enumerations, the following additional methods are generated:
elmx_enums getenum_NAME() const;
Returns the enumerated constant corresponding to the value of item NAME.
bool setenum_NAME( elmx_enums );
Sets the value of item NAME to the value corresponding to the specified enumerated constant. true is
returned if the value is correctly set, and false otherwise.
If a simple type is nillable, then it is treated as complex type / simple content. Thus the interaction with
nillable types is discussed below.
2.5.4 - Optional Simple Type C++ Interface
(Note: See Multiple Singular Types for the interface to list types.) The Optional Simple Type has the same
methods as defined for the Singular Simple Type, plus the following methods:
If the item is not in an xs:choice, the following methods are generated:
bool isset_NAME() const;
Returns true if the item NAME is present, and false otherwise.
void unset_NAME();
Marks the item NAME as not present.
If the item is allowed in an xs:choice, but not present, the getchosen() method returns <class
name>::e_choice_not_set.
34
2.5.2 - Optional Empty Type C++ Interface
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
2.5.5 - Multiple Simple Type and List Simple Type C++ Interface
This interface is used for elements that have a cardinality greater than 1, and elements or attributes with lists.
size_t size_NAME() const;
Returns the number of instances to type NAME.
<return type> get_NAME( size_t n ) const;
Returns the n-th value of item NAME. The first instance is n=0.
<non-const ref to type> get_NAME( size_t n );
Generated if the type of NAME is a compound class such as for xs:date. Returns a non-const
reference to the n-th value of item NAME. The first instance is n=0.
lmx::elmx_error set_NAME( size_t n, type );
Sets the instance at the n-th position of the collection of NAME items. The item previously at the n-th
position is over-written. If a value of n is specified that is larger than the size of the collection, the
append_NAME( type ) method is called. The first instance is n=0.
lmx::elmx_error append_NAME( type );
Appends an instance to the collection of NAME items.
lmx::elmx_error insert_NAME( size_t n, type );
Inserts an instance into the n-th position of the collection of NAME items. The item previously at the
n-th position is moved to the (n+1)-th position and so on. If a value of n is specified that is larger than
the size of the collection, the append_NAME( type ) method is called. The first instance is n=0.
void delete_NAME( size_t n );
Deletes the n-th value of item NAME. The first instance is n=0.
void clear_NAME();
Removes all items from the collection. In the case of a list item, calling this method can be used to
create an empty list if a list with no members is wanted.
If the type has enumerations, the following additional methods are generated:
elmx_enums getenum_NAME( size_t n ) const;
Returns the enumeration constant corresponding to the n-th instance of item NAME. The first instance
is n=0.
bool setenum_NAME( size_t n, elmx_enums set );
Sets the value of the item at the n-th position of the NAME collection to the value corresponding to the
specified enumeration constant. If a value of n is specified that is larger than the size of the collection,
the appendenum_NAME( type ) method is called. The first instance is n=0. true is returned if
the value is correctly set, and false otherwise.
bool appendenum_NAME( elmx_enums set );
Appends a value to the item NAME corresponding to the specified enumeration constant. true is
returned if the value is correctly set, and false otherwise.
bool insertenum_NAME( size_t n, elmx_enums set );
Inserts a value at the n-th position of the collection NAME corresponding to the specified enumeration
constant. If a value of n is specified that is larger than the size of the collection, the
2.5.5 - Multiple Simple Type and List Simple Type C++ Interface
35
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
appendenum_NAME( type ) method is called. The first instance is n=0. true is returned if the
value is correctly set, and false otherwise.
If the type is an optional list, the following additional methods are generated (this allows an empty list to be
present):
bool isset_NAME() const;
Returns true if the item NAME is present, and false otherwise.
void unset_NAME();
Marks the item NAME as not present.
2.5.6 - Singular Complex Type C++ Interface
<const ref to type> get_NAME() const;
This method provides read-only access to NAME. It is recommended that this method be used for
reading the item rather than the read/write variant.
<non-const ref to type> get_NAME();
This method allows read/write access to NAME by returning a reference.
If the complex type is an xs:choice, the following method is generated:
<class name>::elmx_chosen getchosen() const;
Returns the class specific enumerated value corresponding to the present choice.
For the marshaling operation, the appropriate choice is selected when the relevant set_NAME or non-const
get_NAME method among the choice's children is called. For example, if CHOICE is a choice, represented
by the class c_CHOICE, and OPTION is one of the elements within the choice, then if we have:
c_CHOICE &my_choice = ...;
we can select the desired option within the choice by doing:
my_choice.set_OPTION(...);
Similarly, if we do not immediately have a pointer or reference to c_CHOICE, we can set the desired option
by doing:
my_item.get_CHOICE().set_OPTION(...);
If OPTION is a complex type which has associated with it a non-const get method, then the act of calling that
non-const get method will cause the appropriate choice option to be selected. For example:
c_OPTION &my_option = my_item.get_CHOICE().get_OPTION(...);
or:
my_item.get_CHOICE().get_OPTION().set_an_option_child(12);
If the complex type is an xs:all, the following methods are generated:
36
2.5.6 - Singular Complex Type C++ Interface
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
<class name>::elmx_all getorder( size_t n ) const;
Returns the enumeration constant corresponding to the n-th present element in the xs:all construct.
The first element corresponds to n == 0.
size_t getorder( elmx_all enumeration_const ) const;
Returns the position within the xs:all construct of the element corresponding to the enumeration
constant enumeration_const. The function returns lmx::k_all_order_not_present if
the element is not present.
size_t getorder_NAME() const;
Returns the position of element NAME within the xs:all construct. The function returns
lmx::k_all_order_not_present if the element is not present.
size_t sizeorder() const;
Returns the number of elements present in the xs:all construct.
During marshaling, the order that the elements are output corresponds to the order in which they are written to
the object.
A nillable type will always have a class generated for it, even if it is a simple type. (In effect a nillable simple
type is treated as complex type / simple content.) If a type is nillable, the following functions will be
generated:void setnil_NAME();
Used to set the element to a nil value.
bool isnil_NAME() const;
Returns true is a value is nil, and false otherwise.
If a value is nillable, you should test isnil_NAME() prior to attempting to read the element's body value(s)
using get_NAME() etc.
If a type is polymorphic, the following method is generated:
void assign_NAME( c_NAME_TYPE *p_derived_type );
Sets the polymorphic type associated with item NAME to the object pointed to by p_derived_type,
which must be a derived type of c_NAME_TYPE. Once the pointer is assigned to the generated
object, the generated object takes responsibility for deleting the pointer on destruction. (Note: the
get_NAME() method can be used to access a polymorphic type, and no additional methods are
generated for this purpose.)
2.5.6.1 - Singular Complex Type Usage Examples
To read a child of NAME:
int an_int = top.get_NAME().get_CHILD();
Or, if you intend to read a lot of items from NAME, another possibility is:
const c_NAME & name_ref = top.get_NAME();
int an_int = name_ref.get_CHILD();
float a_float = name_ref.get_CHILD2();
2.5.6.1 - Singular Complex Type Usage Examples
37
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
To write to NAME:
top.get_NAME().set_CHILD( 12 );
Or:
c_NAME & name_ref = top.get_NAME();
name_ref.set_CHILD( 12 );
name_ref.set_CHILD2( 1.2 );
If the complex type is an xs:choice:
switch( top.getchosen() )
{
case c_top::e_my_int:
...
break;
case c_top::e_my_float:
...
break;
default:
assert(0);
}
An alternative to the above with better compile time checking is:
c_top::elmx_chosen chosen = top.getchosen();
if( chosen == c_top::e_my_int ) {
...
}
else if( chosen == c_top::e_my_float ) {
...
}
To set the choice option when marshaling, use one of the children's set or non-const get methods. For
example:
top.get_NAME().set_CHILD( 12 );
Or:
c_NAME & name_ref = top.get_NAME();
name_ref.set_CHILD( 12 );
If the complex type is an xs:all:
for( size_t i=0; i<top.sizeorder(); ++i )
{
switch( top.getorder(i) )
{
case c_top::e_my_int:
...
break;
case c_top::e_my_float:
...
break;
default:
assert(0);
38
2.5.6.1 - Singular Complex Type Usage Examples
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
}
}
2.5.7 - Optional Complex Type C++ Interface
An Optional Complex Type has the same methods as a Singular Complex type, plus the additional methods
specified for an Optional Simple Type.
If the complex type is an xs:choice, a getchosen() method is generated as described in Singular Complex
Type. If the optional xs:choice is not present, the getchosen() method returns <class
name>::e_choice_not_set.
2.5.7.1 - Optional Complex Type Usage Examples
To read an optional complex type:
if( top.isset_NAME() ) {
const c_NAME & name_ref = top.get_NAME();
int an_int = name_ref.get_CHILD();
float a_float = name_ref.get_CHILD2();
}
To write to an optional complex type:
c_NAME & name_ref = top.get_NAME();
name_ref.set_CHILD( 12 );
name_ref.set_CHILD2( 1.2 );
2.5.8 - Multiple Complex Type C++ Interface
Common to reading and writing:
size_t size_NAME() const;
Returns the number of instances of item NAME.
For reading:
<const ref to type> get_NAME( size_t n ) const;
Returns a const reference to the n-th instance of item NAME. This version should always be used when
reading from item NAME. The first instance is n=0.
For writing:
void append_NAME();
Appends a new instance of item NAME. This method is used in conjunction with the back_NAME()
method. For more information see the example below.
void append_NAME( <type> * p );
Appends the value pointed to by p to the collection of NAME items. The LMX generated code takes
ownership of the object pointed to by p when the method is called, and will delete it when the LMX
2.5.7 - Optional Complex Type C++ Interface
39
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
object is deleted. Use get_NAME( n ) to access and modify the inserted value.
<non-const ref to type> back_NAME();
Returns a reference to the instance at the back of the list of item NAME. This method is used in
conjunction with the append_NAME() method. For more information, see the example below.
void insert_NAME( size_t n );
Inserts an instance into the n-th position of the collection of NAME items. The item previously at the
n-th position is moved to the (n+1)-th position and so on. The first instance is n=0. If a value of n is
specified that is larger than the size of the collection, the append_NAME() method is called. Use
get_NAME( n ) to access and modify the inserted value.
void insert_NAME( size_t n, <type> * p );
Inserts the value pointed to by p into the n-th position of the collection of NAME items. The item
previously at the n-th position is moved to the (n+1)-th position and so on. The first instance is n=0. If
a value of n is specified that is larger than the size of the collection, the append_NAME( p )
method is called. The LMX generated code takes ownership of the object pointed to by p when the
method is called, and will delete it when the LMX object is deleted. Use get_NAME( n ) to access
and modify the inserted value.
<non-const ref to type> get_NAME( size_t n );
Returns a reference to the n-th instance of item NAME. The first instance is n=0. If such an instance
does not currently exist, an instance (and all intervening instances) will be created. For example, if m
instances already exist, and instance n is requested (where m < n) n - m instances will be created.
<non-const ref to type> assign_NAME( size_t n, <const ref to type> );
Does a deep copy of the input parameter to the n-th position of the collection. The first instance is
n=0. If such an instance does not currently exist, an instance (and all intervening instances) will be
created. For example, if m instances already exist, and instance n is requested (where m < n) n - m
instances will be created.
void delete_NAME( size_t n );
Deletes the n-th instance of item NAME. The first instance is n=0.
void clear_NAME();
Removes all items from the collection.
If NAME is an optional xs:list, it also has the methods defined for an Optional Complex Type.
If the complex type is an xs:choice, the following method is also generated:
<class name>::elmx_chosen getchosen( size_t n ) const;
Returns the class specific enumerated value corresponding to the chosen element in the n-th instance
of the xs:choice. n=0 specifies the first instance. Other than the n parameter to select the index, the
getchosen() method operates the same way as described in the Singular Complex Type section.
If a type is polymorphic, the following method is generated:
void append_NAME( c_NAME_TYPE *p_derived_type );
Appends the polymorphic type pointed to by p_derived_type to the item NAME, which must be a
derived type of c_NAME_TYPE. Once the pointer is assigned to the generated object, the generated
object takes responsibility for deleting the pointer on destruction. (Note: the get_NAME(size_t
n) method can be used to access a polymorphic type, and no additional methods are generated for
40
2.5.8 - Multiple Complex Type C++ Interface
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
this purpose.)
2.5.8.1 - Multiple Complex Type Usage Examples
For reading:
for( size_t i=0; i<top.size_NAME(); ++i )
do_something( top.get_NAME( i ) );
For writing, a new instance of the complex type is appended using the append_NAME method. The new
instance is then populated by repeated use of the back_NAME method. For example:
while( not_finished( &an_int, &a_float ) ) {
top.append_NAME();
top.back_NAME().set_int( an_int );
top.back_NAME().set_float( a_float );
}
2.5.9 - Multiple xs:anyAttribute C++ Interface
If xs:anyAttribute is specified, only the 'multiple' case is allowed.
When a document has instances xs:anyAttribute it may be necessary to add extra namespace information to
the c_xml_writer class. See 3.18 - Adding Extra Namespace Information for more information.
size_t sizeany_attribute() const;
Returns the number of xs:anyAttributes present.
void getany_attribute( size_t n,
std::string *p_namespace,
std::string *p_name,
std::string *p_value ) const;
Retrieves the n-th xs:anyAttribute. The namespace of the attribute (e.g. 'http://...') is stored in the
string pointed to by p_namespace, the name of the attribute is stored in the string pointed to by
p_name, and the value is stored in the string pointed to by p_value. n=0 refers to the first instance.
void appendany_attribute( const std::string &name,
const std::string &value );
Appends an xs:anyAttribute to the list of xs:anyAttributes. The name of the attribute is stored in
name, and the value in value. When written to the XML document, the value part will be surrounded
by quote marks, and standard XML entities will be escaped (e.g. & ' " etc).
void insertany_attribute( size_t n,
const std::string &name,
const std::string &value );
Inserts an xs:anyAttribute at the n-th position of the list of xs:anyAttributes. The item previously at
the n-th position is moved to the (n+1)-th position and so on. The first instance is n=0. If a value of n
is specified that is larger than the size of the collection, the appendany_attribute() method is
called. The name of the attribute is stored in name, and the value in value. When written to the XML
document, the value part will be surrounded by quote marks, and standard XML entities will be
escaped (e.g. & ' " etc).
2.5.8.1 - Multiple Complex Type Usage Examples
41
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
void deleteany_attribute( size_t n );
Deletes the n-th instance of xs:anyAttribute. n=0 refers to the first instance.
void clearany_attribute();
Removes all items from the collection.
2.5.10 - Singular xs:any C++ Interface
When a document has instances xs:any it may be necessary to add extra namespace information to the
c_xml_reader and c_xml_writer class. See 3.18 - Adding Extra Namespace Information for more
information.
To access the xs:any information, use the following methods:
const lmx::c_any_info & get_any() const;
lmx::c_any_info
& get_any();
Retrieves the lmx::c_any_info object corresponding to the data in the xs:any instance, where
lmx::c_any_info has the following public interface:
class c_any_info
{
public:
c_any_info();
// Compiler generated c_any_info( const c_any_info & ar_rhs ) is OK
// For writing during unmarshaling
c_any_info( const std::string &ar_namespace,
const std::string &ar_name,
const std::string &ar_value );
// For user writing xs:anyAttribute
c_any_info( const std::string &ar_name,
const std::string &ar_value )
// For user writing xs:any
c_any_info( const std::string &ar_value );
// Compiler generated operator = ( const c_any_info & ar_rhs ) is OK
const std::string & get_namespace() const;
const std::string & get_name() const;
const std::string & get_value() const;
std::string get_local_name() const;
void set_namespace( const std::string &ar_namespace );
void set_name( const std::string &ar_name );
void set_value( const std::string &ar_value );
const c_namespace_context & get_namespace_context() const;
c_namespace_context & get_namespace_context();
};
If the xs:any element corresponds to:
<nsp:myElement xmlns:nsp="http://codalogic.com/example"
etc....>...</nsp:myElement>
then:
◊ get_namespace() returns: http://codalogic.com/example
◊ get_name() returns: nsp:myElement
42
2.5.9 - Multiple xs:anyAttribute C++ Interface
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
◊ get_local_name() returns: myElement
◊ get_value() returns: <nsp:myElement
xmlns:nsp="http://codalogic.com/example"
etc....>...</nsp:myElement>.
void set_any( const lmx::tlmx_string &any );
Set the value of an xs:any element. The specified value must include the start tag and end tag, and
be UTF-8 encoded. The value is inserted into the output using a simple copy operation.
(See 2.5 - Accesing the Data in the Generated Classes for the naming of xs:any access methods.)
Additionally, the generated code supports a legacy mode interface to xs:any using the following interface:
void get_any( std::string *p_namespace,
std::string *p_name,
std::string *p_value ) const;
Retrieves the details of an xs:any element. If the xs:any element corresponds to:
<nsp:myElement xmlns:nsp="http://codalogic.com/example"
etc....>...</nsp:myElement>
then:
◊ *p_namespace is set to: http://codalogic.com/example
◊ *p_name is set to: nsp:myElement
◊ *p_value is set to: <nsp:myElement
xmlns:nsp="http://codalogic.com/example"
etc....>...</nsp:myElement>.
*p_namespace and/or *p_name may be set to 0 (NULL) if it is not desired to retrieve their
corresponding values.
2.5.11 - Optional xs:any C++ Interface
An Optional xs:any has the same interface methods as a Singular xs:any instance as described in
2.5.10 - Singular xs:any C++ Interface, plus the methods of any Optional Simple Type.
2.5.12 - Multiple xs:any C++ Interface
When a document has instances xs:any it may be necessary to add extra namespace information to the
c_xml_reader and c_xml_writer class. See 3.18 - Adding Extra Namespace Information for more
information.
size_t size_any() const;
Returns the number of instances of a particular xs:any.
const lmx::c_any_info & get_any( size_t n ) const;
lmx::c_any_info
& get_any( size_t n );
Retrieves a reference to the lmx::c_any_info object representing the n-th instance of a particular
xs:any. The first instance is n=0. lmx::c_any_info is described in 2.5.10 - Singular xs:any
2.5.10 - Singular xs:any C++ Interface
43
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
C++ Interface.
(See 2.5 - Accesing the Data in the Generated Classes for the naming of xs:any access methods.)
void append_any( const lmx::tlmx_string &any );
Appends an instance to a particular xs:any. The specified value must include the start tag and end
tag, and be UTF-8 encoded. The value is inserted into the output using a simple copy operation.
void insert_any( size_t n, const lmx::tlmx_string &any );
Inserts an instance to the n-th position of a particular xs:any. The first instance is n=0. The item
previously at the n-th position is moved to the (n+1)-th position and so on. If a value of n is specified
that is larger than the size of the collection, the append_any() method is called. The specified
value must include the start tag and end tag, and be UTF-8 encoded. The value is inserted into the
output using a simple copy operation.
void delete_any( size_t n );
Deletes the n-th instance of a particular xs:any. The first instance is n=0.
void clear_any();
Removes all items from the collection.
Additionally, the generated code supports a legacy mode interface to the xs:any using the following
interface:
void get_any( size_t n,
std::string *p_namespace,
std::string *p_name,
std::string *p_value ) const;
Retrieves the details of n-th element of a particular xs:any. On completion of the function, the
string pointed to by p_value will include the start tag and end tag. The first instance is n=0.
If the xs:any element corresponds to:
<nsp:myElement xmlns:nsp="http://codalogic.com/example"
etc....>...</nsp:myElement>
then:
◊ *p_namespace is set to: http://codalogic.com/example
◊ *p_name is set to: nsp:myElement
◊ *p_value is set to: <nsp:myElement
xmlns:nsp="http://codalogic.com/example"
etc....>...</nsp:myElement>.
*p_namespace and/or *p_name may be set to 0 (NULL) if it is not desired to retrieve their
corresponding values.
2.5.13 - Additional Polymorphic Methods
44
2.5.12 - Multiple xs:any C++ Interface
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
2.5.13.1 - Finding the Identity of a Polymorphic Class
One way to determine the type of a polymorphic class is to use C++'s RTTI. Additionally, LMX generates
methods and members that allow the identity of a class to be determined.
Each polymorphic class has the following public data member:
t_class_identity id
Returns a constant that is unique for the particular class.
Each polymorphic class has the following public methods:
t_class_identity getid() const;
A virtual function that returns the id constant that identifies the class type.
bool has_id( t_class_identity sought_id ) const;
A virtual function that returns true if the class hierarchy contains the type corresponding to sought_id.
For example:
c_base *p_base = &item.get_base();
if( p_base->getid() == c_derived::id )
{
c_derived *p_derived = dynamic_cast<c_derived *>( p_base );
}
Or:
c_base *p_base = &item.get_base();
if( p_base->has_id( c_derived::id ) )
{
c_derived *p_derived = dynamic_cast<c_derived *>( p_base );
// Do the common things specific to c_derived
}
if( p_base->has_id( c_more_derived::id ) )
{
c_more_derived *p_more_derived = dynamic_cast<c_more_derived *>( p_base );
// Do the common things specific to c_more_derived
}
2.5.13.2 - Polymorphic Cloning
To support polymorphic object duplication, LMX generates a virtual clone() method. This returns a pointer
to a deep copy of the object on which the clone() method is called. For example:
c_base *p_base = item.get_base().clone();
2.5.14 - Resetting the Class
Each class has a function that resets the object to the state it would be after construction. The prototype for
this function is:
void reset();
2.5.13.1 - Finding the Identity of a Polymorphic Class
45
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Resets the state of the object to that as if it had just been constructed.
2.5.15 - Run-time Checking
LMX generates two functions for each class that can be used to check whether sufficient attributes and
elements have been set for a class to be minimally valid.
bool is_occurs_ok( lmx::c_check_tracker * p_tracker_in = LMXNULL );
Tests whether sufficient elements and attributes have been set for the class to be valid.
is_occurs_ok() only checks members of its class and not members of any child or base classes.
Returns true if sufficient items have been set, and false otherwise.
bool check( lmx::c_check_tracker * p_tracker_in = LMXNULL );
Recursively calls the current classes is_occurs_ok() method and the check() methods of the
child classes offering a 'deep' version of is_occurs_ok(). Returns true if sufficient items have
been set, and false otherwise.
One way to use these functions is in a debug version assert statement, e.g.:
c_NAME & name_ref = top.get_NAME();
name_ref.set_CHILD( 12 );
name_ref.set_CHILD2( 1.2 );
assert( name_ref.is_occurs_ok() );
If the above call to is_occurs_ok() finds missing components it will populate the global instance of
lmx::c_check_exception called lmx::global_check_exception. If the #define
LMX_USE_CHECK_EXCEPTIONS is set to 1 or LMX_USE_CHECK_EXCEPTIONS is not set and
LMX_USE_EXCEPTIONS is set to 1 then is_occurs_ok() will throw a lmx::c_check_exception
exception if it finds missing components.
is_occurs_ok() can also be configured to populate a local instance of c_check_exception using the
following code:
lmx::c_check_exception check_exception;
lmx::c_check_tracker check_tracker( &check_exception );
if( ! name_ref.is_occurs_ok( &check_tracker ) )
{
// Look at check_exception for details of error
}
The check() method has similar behavior to the is_occurs_ok() method, but recursively calls the
child members versions of check() to give a deep version of is_occurs_ok().
To test the facets of simple type components LMX allows optional testing of facets when the various
set_NAME functions are called.
See section 3.16.5 - Debugging Support for more information on these features.
46
2.5.14 - Resetting the Class
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
2.6 - The Abbreviated Type Notation
In both the generated C++ header file and the HTML documentation file an abbreviated notation is used to
describe the various types. The intention is to allow you to work mainly from the generated C++ header file or
HTML file without having to refer to the actual schema definition, thus speeding up the development process.
This section describes the format.
The complete abbreviated type notation has the form:
name_in_schema --> type { facets } [ cardinality ]
The name_in_schema field is the name of the item from the schema (possibly made ASCII safe). If the
item is an xs:any element, then the field is set to {any}. If the item is the body of a Simple Content element,
the field is set to {body}.
The type is the type of the item. This may be one of the built-in schema types (e.g. xs:string) or the name of
a type defined elsewhere in the schema. If the type is a built-in schema type, its name is prefixed by xs:
irrespective of whether the schema namespace prefix has been set to xs.
The facets field specifies (some of) the facets applied to a type by a schema. The facets are wrapped in
curly braces and each facet description is separated by a comma. If there are no relevant facets, the facets
field and the enclosing curly braces are omitted. Currently this field captures pattern, min/max and length
facets. Enumeration facets are documented along with the methods for reading and writing items via
enumeration (e.g. getenum_NAME and setenum_NAME).
A pattern facet follows the Perl regular expression format without the beginning and end anchors. That is:
/pattern/
The min and max range facets have the form:
min_value <= x < max_value
If either the min or the max limits are not specified, then the corresponding leg of the facet description is
omitted. For example, if no min_value were specified, the following format would be used:
x < max_value
The comparison operator shown in the description (e.g. < or <=) depends on whether the facet is exclusive or
inclusive.
Length facets are documented for xs:string, xs:hexBinary and xs:base64Binary types where
specified. The facet is documented as:
min_length .. max_length
If the maximum length is unbounded, the following form is used:
min_length .. *
If the minimum length and the maximum length are the same, this is documented as:
2.6 - The Abbreviated Type Notation
47
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
length .. length
The cardinality field specifies how many instances of the item are allowed. If the field is absent (including
absence of the square brackets), one, and only one, occurrence of the item is allowed. If the field is present,
the general format is:
[ min .. max ]
where min specifies the minimum number of times the item can occur and max specifies the maximum
number of times an item can occur. If min and max are the same values, then the following form is used:
[ min ]
If the maximum number of occurrences is unbounded, then the following form is used:
[ min .. * ]
2.6.1 - Abbreviated Type Notation Examples
To illustrate the Abbreviated Type Notation, the following examples are presented:
e1 --> xs:unsignedShort
e1 is an unsigned short that can occur once, and only once.
e2 --> xs:unsignedShort [0..1]
e2 is an optional unsigned short. It can occur zero or one time.
e3 --> xs:unsignedShort { 0<=x<1000 }
e3 is an unsigned short that is constrained to be greater than or equal to 0 and less than 1000. It can occur
once, and only once.
e4 --> xs:unsignedShort { 0<=x<1000 }[0..1]
e4 is an unsigned short that is constrained to be greater than or equal to 0 and less than 1000. It can occur
zero or one time.
e5 --> xs:unsignedShort { 0<x<=1000 }[1..*]
e5 is an unsigned short that is constrained to be greater than 0 and less than or equal to 1000. It can occur one
or more times.
e6 --> xs:string { /a\d/,/b\d\d/ }[1..*]
e6 is a string that must match the pattern a\d or b\d\d (e.g. 'a1' or 'b12'). It can occur one or more times.
e7 --> xs:string { 1..5 }[1..*]
e7 is a string that can be between 1 and 5 characters long. It can occur one or more times.
48
2.6.1 - Abbreviated Type Notation Examples
3 - In More Depth
3.1 - Configuration using lmxuser.h
As already alluded to, in addition to configuring data binding behavior via the LMX executable, you can also
control the behavior by setting various C++ #defines and editing USER: sections in lmxuser.h.
For a number of features, the effect of lmxuser.h can be changed by setting C++ #defines in your C++
project or makefiles.
To make this easier, setting LMX_WANT_USER_DEFS in your project/makefile will cause lmxuser.h to
#include "lmxuser-defs.h" and you can set your other #defines in the included file.
You can also specify LMX_WANT_USER_DEFS_END, perhaps in the lmxuser-defs.h file, and this will
cause the file lmxuser-defs-end.h to be included at the end of lmxuser.h. This allows further
customization of LMX behavior using features that have been defined by lmxuser.h.
3.2 - Unmarshaling (advanced forms)
2.3 - Unmarshaling (simple form) describes the simple method for unmarshaling object content direct from
files and memory. In certain situations more flexibility is required, and the following more advanced
techniques may be required.
To read XML from an in-memory buffer, code similar to the following can be used:
lmx::c_xml_reader_memory reader( xml_message_data_buffer,
number_of_bytes_in_buffer );
c_generated_xsd_root_class top_object;
lmx::elmx_error error = top_object.unmarshal( reader );
if( error == lmx::ELMX_OK )
{
const c_generated_xsd_root_class & const_top_object = top_object;
...
To read XML from a file, code similar to the following can be used:
lmx::c_xml_reader_file reader( "c:\\myxml.xml" );
if( reader.is_open() )
{
c_generated_xsd_root_class top_object;
lmx::elmx_error error = top_object.unmarshal( reader );
if( error == lmx::ELMX_OK )
{
const c_generated_xsd_root_class & const_top_object = top_object;
...
3 - In More Depth
49
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Looking more closely at the in-memory case, we first create an object that will do low-level read operations
from memory. This is an instance of c_xml_reader_memory from the lmx namespace. The constructor
for this object takes a pointer to the buffer as the first argument and the number of valid bytes in the buffer as
the second argument.
lmx::c_xml_reader_memory reader( xml_message_data_buffer,
number_of_bytes_in_buffer );
We then create an instance of the top-level class generated by the LMX code generator. The name of this class
will differ. If there is only one global element, the relevant class will be that of the global element.
Alternatively, the name is either derived from the namespace prefix assigned to the schema's target
namespace, or is c_root. It is normally the last class defined in the .h file.
c_generated_xsd_root_class top_object;
To do the unmarshaling, the unmarshal method of the top level object is called, giving it a reference to the
low-level reader class. Note that you must always unmarshal into a freshly created object otherwise the results
may be unpredictable. The method returns an error code.
lmx::elmx_error error = top_object.unmarshal( reader );
If the returned error code is lmx::ELMX_OK then unmarshaling has been successful, and the data structure
can be interrogated. Section 3.9 - Error Codes lists other possible error codes.
if( error == lmx::ELMX_OK )
{
Having determined that the unmarshalled object is suitable for processing, it is advisable to create a const
version of the class reference to avoid accidentally modifying the various objects.
const c_generated_xsd_root_class & const_top_object = top_object;
...
Unmarshaling from a file is a similar process, however in this case the c_xml_reader_file class from
the lmx namespace is used as the type for the low-level reader. The constructor takes a single argument,
which is the name of the file from which to read.
lmx::c_xml_reader_file reader( "c:\\myxml.xml" );
Having created the c_xml_reader_file instance, it is necessary to test whether the file has been
successfully opened by calling the is_open method. (This method can also be called on an object of type
c_xml_reader_memory, but this always returns true.)
if( reader.is_open() )
{
If the file is successfully opened, the unmarshaling operation continues as for the from memory case, e.g.:
c_generated_xsd_root_class top_object;
lmx::elmx_error error = top_object.unmarshal( reader );
if( error == lmx::ELMX_OK )
{
50
3.2 - Unmarshaling (advanced forms)
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
const c_generated_xsd_root_class & const_top_object = top_object;
...
Note that during unmarshaling memory may be dynamically allocated, which may cause exceptions to be
thrown. Therefore it may be appropriate to include try/catch blocks at some level in your code.
3.2.1 - Unmarshaling with Additional Control when Reading XML Input
If additional control is required over the way XML is read, a further method for unmarshaling is available.
The interface class for reading XML input is c_read, which is defined in lmxparse.h. To implement
reading from different sources, or perform special processing while reading (such a reading from a socket or a
compressed file), implement a concrete class that derives from c_read and implements the relevant virtual
methods. Assuming such a concrete class is called c_my_reader, the code to unmarshal using this class
would be:
c_my_reader low_level_reader( ...relevant parameters... );
lmx::c_xml_reader reader( low_level_reader );
c_generated_xsd_root_class top_object;
lmx::elmx_error error = top_object.unmarshal( reader );
if( error == lmx::ELMX_OK )
{
const c_generated_xsd_root_class & const_top_object = top_object;
...
The key part is creating an instance of the concrete class and then passing it to an instance of
lmx::c_xml_reader. The rest of the unmarshaling operation is the same as described in
3.2 - Unmarshaling (advanced forms).
3.2.2 - Unmarshalling Multiple XML Instances from the Same Source
If you wish to read multiple XML instances from the same input source you can create a single low-level
reader (such as lmx::c_read_file, c_read_memory and c_read_string) and use that to construct
multiple instances of c_xml_reader from which to read each new XML instance. For example:
// Setup the input source
lmx::c_read_file file_reader( file_in );
for(;;)
{
// Create new c_xml_reader to read from the input source
lmx::c_xml_reader xml_reader( file_reader );
// Read the XML
lmx::elmx_error lmx_error;
c_MyElement item( xml_reader, &lmx_error );
// An error will be reported if there is no input XML
if( lmx_error != lmx::ELMX_OK )
break;
3.2.1 - Unmarshaling with Additional Control when Reading XML Input
51
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
// TODO: Process data in 'item'
}
See also 3.5.1 - Marshalling Multiple XML Instances to the Same Destination.
3.2.3 - Unmarshalling Schema Fragments
Typically an entry point for initiating an unmarshal process corresponds to a global element definition. If you
wish to unmarshal an item that corresponds to a global type you need to use the unmarshal_partial
template functions located in lmxparse.h. These have the prototypes:
template< class T >
elmx_error unmarshal_partial(
c_xml_reader & ar_reader,
T *ap_item,
const s_ns_map ac_ns_map[],
const std::string &ar_namespace,
const std::string &ar_local_name );
template< class T >
elmx_error unmarshal_partial(
T *ap_item,
const char ac_file_name[],
const s_ns_map ac_ns_map[],
const std::string &ar_namespace,
const std::string &ar_local_name );
template< class T >
elmx_error unmarshal_partial(
T *ap_item,
const wchar_t ac_file_name[],
const s_ns_map ac_ns_map[],
const std::string &ar_namespace,
const std::string &ar_local_name );
template< class T >
elmx_error unmarshal_partial(
T *ap_item,
const char *ap_memory,
size_t a_memory_size,
const s_ns_map ac_ns_map[],
const std::string &ar_namespace,
const std::string &ar_local_name );
template< class T >
elmx_error unmarshal_partial(
T *ap_item,
const std::string &ar_string,
const s_ns_map ac_ns_map[],
const std::string &ar_namespace,
const std::string &ar_local_name )
The first two read XML from a file, the third from an array of bytes, the fourth from a std::string and the last
one from an lmx::c_xml_reader object.
The parameters have the following meaning:
52
3.2.2 - Unmarshalling Multiple XML Instances from the Same Source
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
ar_reader
A reference to an LMX reader object.
ap_item
A pointer to the object that is will receive the unmarshalled data.
ac_file_name
The name of a file to be parsed.
ap_memory
A pointer to memory containing XML to be parsed.
a_memory_size
The number of bytes pointer to be ap_memory.
ar_string
A std::string containing XML to be parsed.
ac_ns_map
The generated namespace map for the schema to be unmarshalled. It can be found by looking in the
generated .h file and has the name ns_map_reader. You may need to include the C++ namespace
with the variable name, e.g. my_cpp_namespace::ns_map_reader.
ar_namespace
The XML namespace of the root element that is to be unmarshalled. If there is no namespace, set this
to "".
ar_local_name
The local part of the name of the root element to be unmarshalled, without any namespace or
namespace prefix information.
An example usage is:
c_my_object
my_object;
lmx::elmx_error error = lmx::unmarshal_partial(
"myFile.xml",
&my_object,
my_cpp_namespace::ns_map_reader,
"http://myDomain.com",
// or "" if no namespace
"localName" );
3.3 - Unmarshalling a Sub-element within an XML Instance
If you wish to only unmarshal a particular element within an XML instance you can use the
unmarshal_find template function located in lmxparse.h. This is available in the following forms:
template< class T >
bool /*is_found*/ unmarshal_find(
c_xml_reader & ar_reader,
elmx_error * ap_error,
T * ap_item,
const s_ns_map ac_ns_map[],
const std::string & ar_sought_namespace,
const std::string & ar_sought_local_name );
template< class T >
bool /*is_found*/ unmarshal_find(
elmx_error * ap_error,
T * ap_item,
const char ac_file_name[],
const s_ns_map ac_ns_map[],
const std::string & ar_namespace,
3.2.3 - Unmarshalling Schema Fragments
53
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
const std::string & ar_local_name );
template< class T >
bool /*is_found*/ unmarshal_find(
elmx_error * ap_error,
T * ap_item,
const wchar_t ac_file_name[],
const s_ns_map ac_ns_map[],
const std::string & ar_namespace,
const std::string & ar_local_name );
template< class T >
bool /*is_found*/ unmarshal_find(
elmx_error * ap_error,
T * ap_item,
const char * ap_memory,
size_t a_memory_size,
const s_ns_map ac_ns_map[],
const std::string & ar_namespace,
const std::string & ar_local_name );
template< class T >
bool /*is_found*/ unmarshal_find(
elmx_error * ap_error,
T * ap_item,
const std::string & ar_string,
const s_ns_map ac_ns_map[],
const std::string & ar_namespace,
const std::string & ar_local_name );
The parameters have the following meaning:
ar_reader
A reference to an LMX reader object.
ap_error
A pointer to where any error code is to be placed when unmarshalling is completed.
ap_item
A pointer to the object that is will receive the unmarshalled data.
ac_file_name
The name of a file to be parsed.
ap_memory
A pointer to memory containing XML to be parsed.
a_memory_size
The number of bytes pointer to be ap_memory.
ar_string
A std::string containing XML to be parsed.
ac_ns_map
The generated namespace map for the schema to be unmarshalled. It can be found by looking in the
generated .h file and has the name ns_map_reader. You may need to include the C++ namespace
with the variable name, e.g. my_cpp_namespace::ns_map_reader.
ar_namespace
The XML namespace of the root element that is to be unmarshalled. If there is no namespace, set this
to "".
ar_local_name
The local name of the root element to be unmarshalled, without any namespace or namespace prefix
information.
54
3.3 - Unmarshalling a Sub-element within an XML Instance
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
The function returns true if the specified element was found and false if not.
unmarshal_find() is available in the following variants:
An example usage is:
c_my_object my_object;
lmx::elmx_error error;
bool is_found = lmx::unmarshal_find(
&error,
&my_object,
"myfile.xml",
my_cpp_namespace::ns_map_reader,
"http://myDomain.com",
// or "" if no namespace
"localName" );
if( is_found && error != lmx::ELMX_OK )
{...}
3.4 - Unmarshalling Multiple Sub-elements within an XML
Instance
If you wish to only unmarshal a subset of elements within an XML instance you can use the
unmarshal_select functions located in lmxselected.h. These have the prototypes:
bool /*is_complete*/ unmarshal_select(
c_xml_reader & ar_reader,
lmx::elmx_error * ap_error,
const lmx::s_ns_map ac_ns_map[],
lmx::tc_elements_to_select ac_elements_to_select,
lmx::c_selected_element * ap_handler );
bool /*is_complete*/ unmarshal_select(
lmx::elmx_error * ap_error,
const char ac_file_name[],
const lmx::s_ns_map ac_ns_map[],
lmx::tc_elements_to_select ac_elements_to_select,
lmx::c_selected_element * ap_handler );
#if LMX_WANT_WIDE_FILE_NAMES == 1
bool /*is_complete*/ unmarshal_select(
lmx::elmx_error * ap_error,
const wchar_t ac_file_name[],
const lmx::s_ns_map ac_ns_map[],
lmx::tc_elements_to_select ac_elements_to_select,
lmx::c_selected_element * ap_handler );
#endif
bool /*is_complete*/ unmarshal_select(
lmx::elmx_error * ap_error,
const char * ap_memory,
size_t a_memory_size,
const lmx::s_ns_map ac_ns_map[],
lmx::tc_elements_to_select ac_elements_to_select,
lmx::c_selected_element * ap_handler );
bool /*is_complete*/ unmarshal_select(
3.4 - Unmarshalling Multiple Sub-elements within an XML Instance
55
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
lmx::elmx_error * ap_error,
const std::string & ar_string,
const lmx::s_ns_map ac_ns_map[],
lmx::tc_elements_to_select ac_elements_to_select,
lmx::c_selected_element * ap_handler );
The parameters have the following meaning:
ar_reader
A reference to an LMX reader object.
ap_error
A pointer to where any error code is to be placed when unmarshalling is completed.
ac_file_name
The name of a file to be parsed.
ap_memory
A pointer to memory containing XML to be parsed.
a_memory_size
The number of bytes pointer to be ap_memory.
ar_string
A std::string containing XML to be parsed.
ac_ns_map
The generated namespace map for the schema to be unmarshalled. It can be found by looking in the
generated .h file and has the name ns_map_reader. You may need to include the C++ namespace
with the variable name, e.g. my_cpp_namespace::ns_map_reader.
ac_elements_to_select
An array containing the namespaces and local names of the elements being selected. The final entry in
the array is marked by both fields being set to NULL.
ap_handler
A pointer to a base class of type lmx::c_selected_element that is called when a selected
element is found. This is described further below.
The function returns true if the operation was complete and false if not.
lmx::tc_elements_to_select is defined as:
typedef struct s_elements_to_select
{
const char *p_namespace;
const char *p_local_name;
} tc_elements_to_select[];
// Last member marked by NULL values
lmx::c_selected_element is defined as:
class c_selected_element
{
public:
enum elmx_unmarshal_select_result { E_STOP, E_PARSED, E_SKIP };
virtual elmx_unmarshal_select_result element_found(
c_xml_reader & ar_reader,
elmx_error *ap_error,
const std::string &ar_sought_namespace,
const std::string &ar_sought_local_name,
const std::string &ar_full_element_name ) = 0;
};
56
3.4 - Unmarshalling Multiple Sub-elements within an XML Instance
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
To use unmarshal_select() you define a class that derives from lmx::c_selected_element and
implements an element_found() virtual method.
You then give the unmarshal_select() function a list of element names that you want to process in an
instance of the lmx::tc_elements_to_select array.
Each time unmarshal_select() finds one of the elements you asked for it calls the
lmx::c_selected_element::element_found() virtual method and you can decide how to
process it.
lmx::c_selected_element::element_found() returns either
lmx::c_selected_element::E_PARSED to indicate that it parsed the element,
lmx::c_selected_element::E_SKIP to indicate that it did not parse the element and the
unmarshal_select() function should skip over it, or lmx::c_selected_element::E_STOP to
indicate that parsing should be stopped.
An example usage of unmarshal_select() is as follows:
#include
#include
#include
#include
#include
"selected.h"
"lmxparse.h"
<fstream>
<sstream>
<iostream>
#include "lmxselected.h"
// 1. Setup an array with a list of elements that we are interested in.
// Need namespace and local name. The end of the list is marked by all
// NULLs.
lmx::tc_elements_to_select mc_elements_to_select = {
{ "http://xml2cpp.com/selected.xsd", "stat" },
{ 0, 0 } };
// 2. Define a class that contains a callback that will be called when the
// sought element is found. This is derived from lmx::c_selected_element.
// This example is intended to just find the max and total values of the
// input.
class c_selected : public lmx::c_selected_element
{
private:
int count;
int max_height;
int total_height;
int max_width;
int total_width;
public:
c_selected() : count( 0 ), max_height( 0 ),
total_height( 0 ), max_width( 0 ), total_width( 0 )
{}
// 3. This is the virtual function that is called when the element is found.
virtual elmx_unmarshal_select_result
element_found(
lmx::c_xml_reader & ar_reader,
lmx::elmx_error *ap_error,
const std::string &ar_sought_namespace,
const std::string &ar_sought_local_name,
const std::string &ar_full_element_name );
3.4 - Unmarshalling Multiple Sub-elements within an XML Instance
57
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
void print( std::wostream &ar_os ) const
{
ar_os << L"Max Height: " << max_height<< L"\n" <<
L"Total Height: " << total_height<< L"\n" <<
L"Max Width: " << max_width<< L"\n" <<
L"Total Width: " << total_width << L"\n";
}
};
std::wostream & operator << ( std::wostream &ar_os, const c_selected &ar_selected )
{
ar_selected.print( ar_os );
return ar_os;
}
// 4. The implementation of the virtual function that is called when the
// element is found.
c_selected::elmx_unmarshal_select_result
c_selected::element_found(
lmx::c_xml_reader & ar_reader,
lmx::elmx_error *ap_error,
const std::string &ar_sought_namespace,
const std::string &ar_sought_local_name,
const std::string &ar_full_element_name )
{
// 5. Let's say we skip the second occurence of the element
if( ++count == 2 )
return c_selected_element::E_SKIP;
// 6. ...and stop at the 5th occurence
else if( count >= 5 )
return c_selected_element::E_STOP;
// 7. Unmarshal what we are interested in
c_statType l_stat;
*ap_error = l_stat.unmarshal( ar_reader, ar_full_element_name );
// 8. And if everything is all right, do something with it
if( *ap_error == lmx::ELMX_OK )
{
std::wcout << L"An element: Name=" << l_stat.get_name() <<
L", Height=" << l_stat.get_height() <<
L", Width=" << l_stat.get_width() << L"\n";
if( l_stat.get_height() > max_height )
max_height = l_stat.get_height();
total_height += l_stat.get_height();
if( l_stat.get_width() > max_width )
max_width = l_stat.get_width();
total_width += l_stat.get_width();
}
// 9. Say that we parsed the XML
return c_selected_element::E_PARSED;
}
// 10. Now to actually use what we have written...
int test_it( const char ac_file_in[] )
{
// 11. Create an instance of our callback class.
c_selected l_selected;
58
3.4 - Unmarshalling Multiple Sub-elements within an XML Instance
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
// 12. Create an LMX reader to read XML from a file
lmx::c_xml_reader_file l_reader( ac_file_in );
if( ! l_reader.is_open() )
return 1;
// 13. We need to be able to return an LMX error code
lmx::elmx_error l_error;
// 14. Actually call the LMX selective unmarsharshalling functionality
if( lmx::unmarshal_select(
&l_error,
// Place to store any LMX error code
ac_file_in,
// XML file name
ns_map_reader,
// List of namespaces relevant to
// this XML schema (created by LMX)
mc_elements_to_select, // List of sought element names
&l_selected ) )
// Object to be called when elements
// found
std::wcout << L"All the XML was parsed\n";
else
std::wcout << L"XML parsed was terminated before the end of the input\n";
// 15. Check if everything went OK
if( l_error != lmx::ELMX_OK )
{
std::wcout << L"Unable to unmarshal " << lmx::convert( ac_file_in ) << L"\n";
return 2;
}
// 16. Print out the callback classes collected results
std::wcout << l_selected;
return 0;
}
3.5 - Marshaling (advanced forms)
2.4 - Marshaling (simple form) describes the simple method for marshaling object content direct to files and
memory. In certain situations more flexibility is required, and the following more advanced techniques may
be required.
Marshaling is the process converting the C++ object content into XML data. An advanced section of
marshaling code for writing the XML to a string looks as follows:
c_generated_xsd_root_class top_object;
// ... Populate and manipulate top_object ...
std::ostringstream sos;
lmx::c_xml_writer writer( sos );
top_object.marshal( writer );
std::string string_out( sos.str() );
Breaking this down into its component parts, we first create an instance of the top-level class generated by the
LMX code generator. (See the description in section 2.3 - Unmarshaling (simple form) on the name given to
3.5 - Marshaling (advanced forms)
59
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
the top level class.) This is populated either by interacting with the object's methods (see 2.5 - Accesing the
Data in the Generated Classes), performing an unmarshaling operation as described in 2.3 - Unmarshaling
(simple form) or a combination of the two.
The first part of the marshaling operation is to create an instance of a class that is derived from
std::ostream. This is where the output will be written. The line below uses the class derived from
std::ostream by the C++ library for outputting to a string, but the line std::ofstream sos; could
equally be used for outputting to a file.
std::ostringstream sos;
Now create a low-level LMX XML writer object of type lmx::c_xml_writer, telling it to write using the
sos object.
lmx::c_xml_writer writer( sos );
Tell the generated top-level object to marshal itself using the writer object by calling the top-level object's
marshal method and giving it a reference to the writer low-level XML writer object:
top_object.marshal( writer );
Finally, for when writing to a string, get the output into a form where it is more readily usable:
std::string string_out( sos.str() );
string_out is then a std::string object containing the marshaled XML.
The equivalent code for writing to a file is as follows:
c_generated_xsd_root_class top_object;
// Open a file using the C++ class derived from ostream
// for writing to a file
std::ofstream fos( "c:\\myfile.xml" );
// If the file is opened successfully
if( fos.is_open() )
{
// Create an instance of an LMX writer class,
// giving it a reference to the opened file's
// stream
lmx::c_xml_writer writer( fos );
// Tell the top_object to marshal its contents
// using the LMX writer class
top_object.marshal( writer );
}
At the end of the above operations, the file c:\myfile.xml will contain the marshaled XML.
3.5.1 - Marshalling Multiple XML Instances to the Same Destination
If you wish to marshal multiple XML instances to the same output destination you can create a single C++
output stream and use that to construct multiple instances of c_xml_writer with which to write each new
60
3.5.1 - Marshalling Multiple XML Instances to the Same Destination
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
XML instance. For example:
// Setup output destination
std::ofstream fos( file_out );
if( fos.is_open() )
{
c_MyElement * p_item;
while( (p_item = get_item_data()) != 0 )
{
// Create a new c_xml_writer to the output destination
// (The 0 for the t_writer_options prevents the XMLDecl being output)
lmx::c_xml_writer xml_writer( fos, 0 );
// Write out the XML
p_item->marshal( xml_writer );
}
}
See also 3.2.2 - Unmarshalling Multiple XML Instances from the Same Source.
3.6 - Selecting the Input File Type (XSD, WSDL, DTD)
LMX can parse both W3C Schema (XSD) and XML (external) DTD files. LMX can also locate and parse
W3C Schema definitions embedded in WSDL files. The parsing that LMX performs depends on the file
extension of the base schema file.
If the file extension is '.wsdl', LMX with attempt to locate a schema within a WSDL file and parse that.
Note that LMX will only extract the first schema within a WSDL file. If multiple schema are present in a
WSDL file then it is necessary to manually extract the second and subsequent schemas into separate schema
files prior to parsing.
If the file extension is '.dtd', LMX will assume the file is an XML external DTD. When parsing DTD
definitions, the Imported files window has no effect. Files names referenced as external entities within a DTD
are opened relative to file location of the DTD file.
All other file extensions, including '.xsd' are assumed to be W3C Schema definitions.
3.7 - The Command-line and Configuration File Format
LMX and WinLMX can be configured using command-line flags and configuration files. The contents of a
configuration file is simply a number of command-line flags (See 3.8 - Command-line Flags). The layout of
the flags in the file (whether on the same line or different lines etc) is not significant, although one command
option per line is suggested. Comments can be included in a configuration file using a '#' character. Any
characters after the '#' character until the end of the line are ignored. For example:
# This is my LMX config file
# Created: today
# By: Me
-cpplines 2000
MyProject.xsd
# The schema for our project
3.6 - Selecting the Input File Type (XSD, WSDL, DTD)
61
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
MyProjectXML
# The root of the output files
When a set of command-line arguments are first received by either tool, it is subject to some special
processing:
• If a single command-line argument is present, and it ends with '.lmxprj' it is assumed to be the name
of a configuration file, and the tool reads and processes the file.
• In the case of the command-line tool, if a single argument is present, but it does not end in '.lmxprj',
the tool tries appending '.lmxprj' to the end of the name and tests whether a file of the resultant name
exists. If it does, the tool treats the resultant file as a configuration file and executes its contents. This
allows configuration files to be associated with particular schema files.
3.8 - Command-line Flags
The LMX code generator understands the following flags:-alt-xml-reader YYY
Changes the XML reader class that the unmarshal code uses to read XML. The default XML reader
class is lmx::c_xml_reader. Typically this flag would be used to specify a class that derives
from lmx::c_xml_reader and the derived class would contain extra data and methods to be used
in conjugation with custom event handlers during the unmarshal process.
-alt-xml-writer YYY
Changes the XML writer class that the marshal code uses to write XML. The default XML writer
class is lmx::c_xml_writer. Typically this flag would be used to specify a class that derives
from lmx::c_xml_writer and the derived class would contain extra data and methods to be used
in conjugation with custom event handlers during the marshal process.
-no-assign
Switches off the generation of operator = and copy constructor code. The code generated uses the
construct and swap pattern for exception safety. This can generate a lot of code and if the assignment
operator is not needed in your code you can have a smaller executable by specifying this flag.
-autover
This option allows the generated code to support limited automatic versioning of your schema. If this
option is selected, the parser will allow unknown elements to appear before an end tag. This allows
new elements to be added to a schema in subsequent versions, but still allow old installations of the
code to operate successfully. Note that selecting this option results in non-standard schema behavior.
-autover-subst-groups
This option allows the generated code to support automatic versioning of substitution groups. With
this option, if an unknown element is encountered in the place of a mandatory substitution group
member, then the element is skipped, the element is recorded as unknown, and no error is generated.
Without this option being specified an error is reported under these circumstances.
-braces-indent
Causes the braces ({}) to be indented.
-build YYY
62
3.7 - The Command-line and Configuration File Format
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Tells the code generator to generate code for the project described in configuration file YYY and
return. A configuration file may have multiple -build arguments allowing multiple projects to be built
from a single configuration file.
-cd YYY
Changes the current working directory to YYY. This allows the configuration file to specify relative
file names rather than absolute file names.
-no-check-code
Disables generation of the check() methods. See 2.5.15 - Run-time Checking.
-check-is-occurs-ok-on-marshal
Adds code to the marshal methods to check that is_occurs_ok() returns true. The
lmx::ELMX_OCCURRENCE_ERROR error is returned if is_occurs_ok() returns false. By
default this check is not done in order to maximize performance.
-class-base CLASS BASE
Specifies that the generated class named CLASS should be generated such that it derives from BASE.
Note the CLASS field should not include the class name prefix. Thus by default the option
-class-base MyClass MyBase will generate code of the form class c_MyClass :
public MyBase.
-cns YYY
Sets the C++ namespace to YYY. In other words, the generated code is wrapped in a namespace
YYY { ... } statement. By default the C++ namespace is derived from the XML namespace
prefix assigned to the primary schema's target namespace. Nested C++ namespaces can also be
specified. To do this use the form -cns YYY::ZZZ.
-no-cns
Indicates that no C++ namespace statements should be generated, even if a default namespace name
can be derived from the XML schema.
-no-container-ops
By default items that can have multiple instances will have container operations generated which
allow you to insert, delete or clear members of the container. Specifying this option disables the
generation of these methods.
-no-convenience
Switches off the generation of convenience functions that allow direct marshaling and unmarshaling
from files and strings.
-no-convenience-xs-any
Switches off the generation of convenience functions that allow direct unmarshaling from content
retrieved xs:any instances.
-cpp-compiler YYY
Indicates that code should be generated that is specific to the C++ compiler identified by YYY. The
only compiler that requires special support is the IBM Visual Age compiler. This is identified by
setting YYY to 'VisualAge'.
-cpp-per-class
3.8 - Command-line Flags
63
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Tells LMX to store the code for each class in a separate .h/.cpp file pair. -cpplines has no effect when
this option is specified. The -multi-file flag has the same effect. If neither -cpp-per-class or
-cpp-per-schema is specified, then a single .h and a single .cpp file are generated.
-cpp-per-schema
Tells LMX to generate one .cpp file for each schema. Note that the generated code for included
schemas is included in the same .cpp as the schema doing the including. This mode generates a single
.h file. -cpplines has no effect when this option is specified. If neither -cpp-per-class or
-cpp-per-schema is specified, then a single .h and a single .cpp file are generated.
-cpplines YYY
Some schemas can generate a large number of lines of code. This may cause a compiler problems if
all the lines are stored in a single file. This flag allows you to control the number of lines that are
stored in each .cpp file. During code generation, before the methods and data for a class is output,
LMX tests whether the number of lines in the .cpp file exceeds the value specified by YYY. If it
does, LMX closes the current file and starts outputting code to a new file. Files are differentiated by
appending a number to the end of the file name root; e.g. my_xsd_code2.cpp.
-doc-in-h
Causes the xs:documentation information from the schemas to be included in the generated .h
files.
-doc-in-html
Causes the xs:documentation information from the schemas to be included in the generated
HTML documentation.
-no-dynamic-cast
Forces use of a static_cast rather than a dynamic_cast when a down cast is required. This is less safe,
but may be more efficient and permits use of compilers that don't support C++'s RTTI.
-enums-global
Makes the C++ enums associated with schema xs:enumerations be all coalesced into a single global
C++ enum. This is the default behavior.
-enums-local
Generate separate C++ enums for each element or attribute having xs:enumeration facets within the
relevant class. This mode is deprecated and it is recommended to use -enums-per-type instead.
-enums-per-type
Generate separate C++ enums for each type that has xs:enumeration facets defined in the schema.
C++ enums associated with global schema types will be in C++ global space, and C++ enums
associated with local schema elements and attributes will be in the relevant C++ class.
-eq
Enables generation of operator ==() and operator !=() methods.
-error-fast
Normally when an error is detected LMX writes code that calls a common error handler method
giving you the opportunity to intercept the error and implement custom error handling. This also
provides a convenient place to put a breakpoint when debugging. When the -error-fast is
specified the error code is returned immediately without calling the handler. This results in faster,
64
3.8 - Command-line Flags
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
more compact code, but does not have the other benefits.
-no-ev
If a type such as a string is enumerated, by default LMX will generate code checks that a read XML
document matches one of the enumerations. Setting this flag turns off generation of such code.
-no-exceptions
Indicates that try/catch blocks should not be generated in C++ functions.
-exec-pre YYY
Specifies the name of a program or script that should be executed prior to starting to generate code
into a file. The script is given the name of the file that is about to be generated. For example, if LMX
is about to starting generating code into myproject.cpp, then, with this option, LMX will execute the
shell command YYY myproject.cpp.
-exec-pre-all YYY
Specifies the name of a program or script that should be executed prior to starting the code generation
process. The script is given the Output File Base Name for the project. For example, if LMX is about
to starting generating code for a project whose Out File Base name is myproject, then, with this
option, LMX will execute the shell command YYY myproject.
-exec-post YYY
Specifies the name of a program or script that should be executed when LMX has finished generating
code into a file. The script is given the name of the file that has been generated. For example, if LMX
finished generating code into myproject.h, then, with this option, LMX will execute the shell
command YYY myproject.h.
-exec-post-all YYY
Specifies the name of a program or script that should be executed when the code generation process is
completed. The script is given the Output File Base Name for the project. For example, if LMX has
completed code generation for a project whose Out File Base name is myproject, then, with this
option, LMX will execute the shell command YYY myproject. Note that the program/script will
not be executed if errors have occurred during the code generation process.
-exec-post-all-error YYY
Similar to -exec-post-all, except that the program/script is run when an error occurs during
code generation.
-expose-containers
Causes LMX to generate methods that return references to the containers (e.g. instances of
std::vector) used in the objects. This allows more fine-grained control of the containers. Specifying
-no-container-ops in conjunction with -expose-containers allows smaller code to be
generated while still maintaining the ability to manipulate the contents of the containers when
required.
-expose-storage
Generates methods that expose the underlying objects that store the data. Direct use of these objects
allows more generic handling of similar data in different parent objects.
-f YYY
3.8 - Command-line Flags
65
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Tells the code generator to add the configuration information contained in the file YYY to the current
configuration. The contents of the configuration file is the same as the command-line flags described
here.
-file-ext-cpp YYY
Set the extension for generated CPP files to YYY. The default is .cpp.
-file-ext-h YYY
Set the extension for generated header files to YYY. The default is .h.
-fname-sep YYY
Sets the character sequence used to separate parts of file names that are constructed from multiple
parts during multiple file generation modes to YYY. The default is -.
-file-ext-snippets YYY
By default a .h is checked to locate user specified snippets and snippet events. Using this option
specifies the file extension of files that are used to store snippets. This allows snippets to be stored in
a file other than the .h file, and means that the source file storing snippets will not be overwritten by
LMX. The separate snippets file should include zero or more sections of snippets consisting of both
the snippet markers and the additional user code.
-forcegen
Normally LMX will not generate code if errors are found when parsing the schema. Setting this flag
forces code generation even if errors are found. The results of setting this flag when errors are
encountered are unpredictable.
-no-gen
Tells LMX not to generate any code (even if there are no errors). This can be used when developing a
schema and code generation is not required. LMX can be used in evaluation mode with this flag set
without displaying any warning messages about licensing!
-no-gen-date
Normally, when writing a file, LMX will include a comment specifying the date when the file was
generated. This flag prevents this. This means that each time LMX is run it will generate the same
output, minimizing diffs in version control systems.
-no-html
Tells the tool NOT to create HTML documentation for the project.
-include-cpp YYY
Specifies that when generating a .cpp file, a line of the form #include "YYY" should be included
in the file. If you would like to use the #include <YYY> form, then include the <> brackets as part
of the parameter, e.g. do -include-cpp <YYY>. Multiple instances of this flag are allowed.
-include-h YYY
Specifies that when generating a .h file, a line of the form #include "YYY" should be included in
the file. If you would like to use the #include <YYY> form, then include the <> brackets as part of
the parameter, e.g. do -include-h <YYY>. Multiple instances of this flag are allowed.
-include-guard YYY
66
3.8 - Command-line Flags
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Sets the .h file include guard text to YYY rather that the default derived from the C++ namespace and
name of the file.
-lmx-cns YYY
Tells the code generator to use the C++ namespace YYY:: when referring to the LMX parser and
other similar items. This allows the user to define an alternative parser to that supplied with LMX.
-no-lmx-cns
Tells the code generator not to use a C++ namespace when referring to the LMX parser components
and other similar items.
-lmx-include-path YYY
Allows the path to be specified when code is generated to #include the LMX header files. To
avoid making assumptions about directory separator characters, LMX does not append a / or \
character if no such character is present. Thus the flag sequence -lmx-include-path YYY will
generate a line similar to #include "YYYlmxuser.h", and in general it will be necessary to
specify the directory separator as part of the parameter; for example -lmx-include-path YYY/.
If the LMX files are to be considered system files (i.e. so that code similar to #include
<YYY/lmxuser.h> is generated), then include the <> brackets as part of the parameter, e.g.
-lmx-include-path <YYY/>. If you want the LMX headers to be considered as system header
files, but don't want them in a sub-directory, do -lmx-include-path <>. This will generate
lines similar to #include <lmxuser.h>. Note that if the directory name contains spaces, then, as
for all such parameters, the parameter will need to be included in double quote marks, for example
-lmx-include-path "Y Y Y/". This is also the case if a system path contains spaces,
requiring a flag sequence of the form -lmx-include-path "<Y Y Y/>". In this case the "
characters will be stripped by the command-line processor, and won't appear in the generated code.
-lmx-types-cns YYY
Tells the code generator to use the C++ namespace YYY:: when referring to the LMX defined data
types. By default lmx:: is used. Changing this allows different C++ classes and types to be used to
represent the data objects in the XML schema.
-no-lmx-types-cns
Tells LMX not to generate a C++ namespace qualifier (such as lmx::) when referring to LMX defined
types. This allows you to define and control the types and classes used by the generated code.
-no-local-classes
By default LMX generates the unmarshal helper class as a local class in the unmarshal method. If
your compiler does not correctly support the access rights of local classes then use this flag to
generate the unmarshal helper class outside the unmarshal method.
-local-enums
Obsolete version of -enums-local flag.
-no-local-enums
Obsolete version of -enums-global flag.
-makefile
Generates a makefile fragment that contains makefile variables that specify the names of C++ files
that have been generated. See 3.10 - Generating makefiles for more information.
3.8 - Command-line Flags
67
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
-makefile-lib YYY
When creating a makefile, specifies the name of the library containing the LMX Supporting Software
Runtime Source Code. For example, -makefile-lib lmx410 will include a -llmx410 in the
makefile. See 3.10 - Generating makefiles for more information.
-markers
Indicates that comment based markers should be output in the code. This can facilitate post processing
of generated code, for example by using the -exec-post flag.
-no-marshal
Disables the generation of the methods associated with marshaling the C++ objects to XML. See also
-no-unmarshal.
-max-else-ifs YYY
Some sections of code have the form if() ... else if() ... else if() .... Some compilers have limitations on
how much nesting of this form can be used. This option allows specification of the maximum number
of else if() constructs. For constructs larger than YYY, the form if() ...; if() ...; is used. This is less
efficient in some cases.
-max-inline YYY
Some short methods can now be generated in-line with the class definition. This flag allows
specification of the maximum number of lines that are allowed to be generated in-line. The default is
3.
-micro-format PATH CLASS
Specifies that the simpleType item identified by PATH should be represented by a micro format using
C++ class CLASS. See 3.26 - Specifying Micro Formats for more information.
-method-getchosen YYY
Set the name of the getchosen() method to YYY.
-method-getid YYY
Set the name of the getid() method to YYY.
-method-getorder YYY
Set the name of the indexed getorder( index ) method to YYY.
-method-has_id YYY
Set the name of the has_id() method to YYY.
-method-is_occurs_ok YYY
Set the name of the is_occurs_ok() method to YYY.
-method-marshal YYY
Set the name of the marshal(...) method to YYY.
-method-reset YYY
Set the name of the reset() method to YYY.
-method-sizeorder YYY
Set the name of the sizeorder() method to YYY.
68
3.8 - Command-line Flags
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
-method-unmarshal YYY
Set the name of the unmarshal(...) method to YYY.
-multi-file
Same as the -cpp-per-class flag.
-names-dir YYY
Optionally specifies the directory that the names files are to be read from or saved to. This flag does
not enable names file usage by itself. Such behavior is enabled by using the -names-in and -names-out
flags. See 3.12 - Naming of Methods and Variables for more information.
-names-in
Use a names file to set the names of C++ variables in the generated code. The name of the names file
is derived from the schema file name by appending '.lmxnames-in.xml' to the file name. If the schema
is named mySchema.xsd, the names file looked for will be mySchema.xsd.lmxnames-in.xml. See the
section 3.12 - Naming of Methods and Variables for more information.
-names-out
Output the C++ names used in the generation to a names file. The name of the names file is derived
from the schema file name by appending '.lmxnames-out.xml' to the file name. If the schema is named
mySchema.xsd, the names file looked for will be mySchema.xsd.lmxnames-out.xml. See the section
3.12 - Naming of Methods and Variables for more information.
-naming YYY
Specifies the naming convention to be used for names in the generated code. If YYY corresponds to
camel LMX will attempt to convert base names generated in the code to CamelCase. If YYY
corresponds to underscore LMX will attempt to make base names underscore separated. If YYY
corresponds to camel-no-xform a set of name prefixes will be selected that are compatible with
the CamelCase naming convention, but base names derived from the schemas will not be transformed
to CamelCase. The absence of this option is equivalent to YYY being set to
underscore-no-xform.
-narrow-strings
Generates suitable #defines to instruct C++ compiler to use narrow strings for storing Unicode
strings.
-no-nested-classes
By default, C++ classes generated for types defined locally in a schema are nested in their parent class
to echo the structure of the original schema. For deeply nested schema (such as Russian Doll design)
this approach may not be desirable. This flag therefore allows you to specify that C++ classes should
not be nested.
-ns-map-base YYY
Specifies an alternate prefix name for the s_ns_map instances containing namespace mapping
information. This can be used to prevent name clashes when multiple instances of LMX generated
code are in the same project. While this option is one solution to this problem, the preferred solution
is to generate the code into separate C++ namespaces using the -cns flag.
-ns-prefix-map FROM TO
Specifies that if a schema defines a namespace prefix of FROM, the marshaling code should use a
namespace prefix of TO. For example, if the schema uses the namespace prefix tns, LMX can be
3.8 - Command-line Flags
69
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
instructed to use the default namespace for the marshalled XML using the flag sequence:
-ns-prefix-map tns ""
It is the responsibility of the user to ensure that mapped namespace prefixes do not overlap. Also,
FROM may specify the relevant namespace URI, such as
http://www.example.com/myschema.xsd. For example, the flag sequence:
-ns-prefix-map http://www.example.com/myschema.xsd myns
would cause elements and attributes in the http://www.example.com/myschema.xsd
namespace to use the namespace prefix myns.
-old-enums
Specifies that the old way of generating enum names should be used.
-output-defaults
Selecting this option results in the default values of attributes and elements being marshalled into the
output if the value is not set by the user access functions. This can be useful when debugging. This is
non-standard schema behavior.
-pattern-out PATTERN FUNCTION
Allows the specification of a custom output converter so that output can be formatted to conform to a
specified xs:pattern definition. See 3.24 - Custom Pattern Facet Output Formatting for more
information.
-no-polymorphic
Disable generation of polymorphic code for schema extensions and restrictions.
-prefix-anonymous YYY
Specifies that all anonymous class names are prefixed by YYY.
-prefix-append YYY
Set the prefix of the methods named append_{name}(...) to YYY{name}(...).
-prefix-appendenum YYY
Set the prefix of the methods named appendenum_{name}(...) to YYY{name}(...).
-prefix-assign YYY
Set the prefix of the methods named assign_{name}(...) to YYY{name}(...).
-prefix-back YYY
Set the prefix of the methods named back_{name}(...) to YYY{name}(...).
-prefix-class YYY
Set the prefix of class names to YYY.
-prefix-clear YYY
Set the prefix of the methods named clear_{name}(...) to YYY{name}(...).
-prefix-container YYY
Set the prefix of the methods named container_{name}(...) to YYY{name}(...).
-prefix-declash YYY
70
3.8 - Command-line Flags
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Specifies a prefix to use to avoid name clashes when including multiple sets of generated code into a
C++ project. We do not recommend using this flag. See 3.22.1 - Handling Multiple Independent
Schemas for more information.
-prefix-default YYY
Set the prefix of the default variables to YYY.
-prefix-delete YYY
Set the prefix of the methods named delete_{name}(...) to YYY{name}(...).
-prefix-enum YYY
Set the prefix of enum names to YYY.
-prefix-enumeration YYY
Sets the prefix used for the C++ enumerated values associated with xs:enumeration facets to
YYY. By default it is the same as that specified by the -prefix-enum flag. Note that for
backwards compatibility reasons this flag has no effect when a single global C++ enum is used to
represent xs:enumeration values.
-prefix-enumeration-name YYY
Sets the prefix used for the C++ enum types associated with xs:enumeration facets to YYY. By
default it is the same as that specified by the -prefix-enum flag. Note that for backwards
compatibility reasons this flag has no effect when a single global C++ enum is used to represent
xs:enumeration values.
-prefix-fraction-digits YYY
Set the prefix of the fraction_digits_{name} variables to YYY.
-prefix-get YYY
Set the prefix of the methods named get_{name}(...) to YYY{name}(...).
-prefix-getenum YYY
Set the prefix of the methods named getenum_{name}(...) to YYY{name}(...).
-prefix-getorder YYY
Set the prefix of the methods named getorder_{name}(...) to YYY{name}(...).
-prefix-inner YYY
Specifies that all inner class names are prefixed by YYY.
-prefix-insert YYY
Set the prefix of the methods named insert_{name}(...) to YYY{name}(...).
-prefix-insertenum YYY
Set the prefix of the methods named insertenum_{name}(...) to YYY{name}(...).
-prefix-isnil YYY
Set the prefix of the methods named isnil_{name}(...) to YYY{name}(...).
-prefix-isset YYY
Set the prefix of the methods named isset_{name}(...) to YYY{name}(...).
3.8 - Command-line Flags
71
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
-prefix-nil YYY
Set the prefix of the nil_{name} variables to YYY.
-prefix-present YYY
Set the prefix of the present_{name} variables to YYY.
-prefix-set YYY
Set the prefix of the methods named set_{name}(...) to YYY{name}(...).
-prefix-setenum YYY
Set the prefix of the methods named setenum_{name}(...) to YYY{name}(...).
-prefix-setnil YYY
Set the prefix of the methods named setnil_{name}(...) to YYY{name}(...).
-prefix-size YYY
Set the prefix of the methods named size_{name}(...) to YYY{name}(...).
-prefix-storage YYY
Set the prefix of the main stotage variable (m_{name}) to YYY.
-prefix-unset YYY
Set the prefix of the methods named unset_{name}(...) to YYY{name}(...).
-release-switch
Uses a switch statement when releasing members of a choice. This results in more code, but is more
portable.
-root YYY
Specifies that if the root class is generated, which by default is derived from the main schema's
namespace prefix, or called c_root, then it should be called c_YYY. Note that the
-prefix-class flag can be used to change the prefix of the class name.
-root-from-xsd-tns-prefix
Specifies that if the root class is generated, which by default is derived from the main schema's
namespace prefix, or called c_root, then it's name should be derived from the namespace prefix
associated with the schema's target namespace. Note that the -prefix-class flag can be used to
change the prefix of the class name.
-root-from-xsd-file
Specifies that if the root class is generated, which by default is derived from the main schema's
namespace prefix, or called c_root, then it's name should be derived from the filename of the main
schema. For example, if the main schema file is called mySchema.xsd, then the root class will be
called c_mySchema. Note that the -prefix-class flag can be used to change the prefix of the
class name.
-no-root-class
Prevents generation of a root class in the case where there is more than one global element.
-snippets
72
3.8 - Command-line Flags
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Causes LMX to output markers between which you can add snippets of your own code. See
3.14 - Augmenting Generated Classes With Your Own Code for more information.
-stdafx
Causes the line #include "stdafx.h" to be written into the generated code.
-soap
Generates additional code for when the generated results are used for SOAP transactions using LMX's
c_soap SOAP wrapper template class. See 3.11 - Use with Web Services for more information.
-suffix-attribute YYY
Specifies that all attribute names are suffixed by YYY.
-suffix-element YYY
Specifies that all element names are suffixed by YYY.
-suffix-type YYY
Specifies that all type names are suffixed by YYY.
-suffix-group YYY
Specifies that all group names are suffixed by YYY.
-suffix-inner YYY
Specifies that all inner class names are suffixed by YYY.
-suffix-anonymous YYY
Specifies that all anonymous class names are suffixed by YYY.
-no-summary
The error output of LMX uses the format used by Microsoft®* Visual Studio to allow easy integration.
To allow standalone operation, LMX will produce a summary of the errors and warnings encountered
(e.g. 0 error(s), 0 warning(s)). However, when used within an IDE, some IDEs produce this summary
themselves. Setting this flag tells LMX not to produce an error summary so that only one error
summary is produced. (*Microsoft is a Registered Trademark of Microsoft Corporation.)
-tframework
Tells LMX to create a separate .cpp file containing test code. See 3.31 - Test Framework Generation
for more information.
-no-unmarshal
Disables the generation of the methods associated with unmarshaling XML to C++ objects. See also
-no-marshal.
-user-base
Selecting this option results in all generated classes ultimately deriving from a user implemented base
class called lmx::c_user_base, defined in the USER: USER_BASE section of lmxuser.h.
This allows you to store common information in the generated class objects.
-no-warn-ur-type
Some texts recommend not using types specified as anyType or the ur-type. After all, this gives no
useful information to the validator. Consequently, by default, LMX gives a warning when a type is
3.8 - Command-line Flags
73
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
not specified for an item (thus defaulting to the ur-type). This flag switches off the warning generated
in this situation.
-werror
Causes warnings detected by LMX to be treated as errors when the final program return code is
evaluated.
Flags to support additional testing:
-eframework
Similar to -tframework, but generates more testing code.
-test-mem
When -eframework is used, this option generates code to ensure that memory is not lost when out
of memory errors occur.
-no-tmain
When -eframework is used, this option changes the name of the main() function. This permits
the inclusion of the test code into a larger piece of code.
3.9 - Error Codes
The marshal and unmarshal functions return an error value of type enum elmx_error. The error codes
have the following meanings:ELMX_OK
No errors - everything is OK.
ELMX_ALL_CARDINALITY_VIOLATION
The cardinality constraints of an xs:all construct have been violated. One possibility is that multiple
instances of an element are present.
ELMX_ATTRIBUTE_ALREADY_READ
Two (or more) attributes with the same name appear in the same start tag.
ELMX_ATTRIBUTES_IN_ANONYMOUS_COMPOSITOR
The read XML contained attributes in the start tag of an anonymous compositor. This should not
occur.
ELMX_BADLY_FORMED
The XML is badly formed.
ELMX_BAD_CHAR_ENCODING
Either the character encoding specified in the XML document, or the character encoding deduced
from the opening sequence of bytes in the document are not supported by the parser.
ELMX_BAD_CHOICE
A bad choice has been selected when you wrote data to the generated objects. Most likely a choice
was not selected when it should have been.
74
3.9 - Error Codes
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
ELMX_BAD_COMMENT
The format of an XML comment (e.g. <!--...-->) is invalid.
ELMX_BAD_DTD
The DTD in an XML document could not be read correctly.
ELMX_BAD_ELEMENT_END
An end tag was either expected and was not found, or the name in the end tag did not match the name
of the element in the start tag.
ELMX_BAD_END_OF_START_TAG
Failed to find a correct end of start tag when looked for.
ELMX_BAD_END_TAG
The format of an end tag is not syntactically correct from an XML viewpoint.
ELMX_BAD_XML_DECL
There is a problem with the XML decl spec. For example, it may be syntactically incorrect, in the
wrong place or multiple decl specs may appear.
ELMX_BAD_XML_VERSION
The XML version is not known.
ELMX_CHAR_ENCODING_MISMATCH
The sequence of bytes read from a data source are not legal for the character encoding specified in the
XML document.
ELMX_ELEMENT_NOT_FOUND
A sought element (typically the document element) was not found.
ELMX_ENTITY_NOT_FOUND
The XML contained an entity that does not have a definition.
ELMX_EXTERNAL_ENTITY
A reference to an external entity has been found. The parser does not resolve references to external
entities.
ELMX_FRACTION_DIGITS_EXCEEDED
The maximum number of fraction digits specified for a value by a schema have been exceeded.
ELMX_LENGTH_TOO_LONG
The length of a string or binary value is too long compared to the constraint set in the schema.
ELMX_LENGTH_TOO_SHORT
The length of a string or binary value is too short compared to the constraint set in the schema.
ELMX_MANDATORY_ELEMENT_MISSING
A mandatory element (minOccurs != 0) was not found.
ELMX_NILLED_ELEMENT_NOT_EMPTY
The body of a nilled element was not empty.
3.9 - Error Codes
75
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
ELMX_NO_FILE
Unable to open file.
ELMX_NO_PATTERN_MATCHED
The data value does not match any of the pattern facets specified for the item.
ELMX_NOT_WELL_FORMED
The document is not well formed in the XML sense.
ELMX_OCCURRENCE_ERROR
When marshalling, the number of occurrences of a element was not in the valid range.
ELMX_RECURSIVE_ENTITY_DEFINITION
An entity definition is recursive, and hence cannot be successfully expanded.
ELMX_REQUIRED_ATTRIBUTES_MISSING
One or more required attributes are missing.
ELMX_TOO_MANY_ITEMS
The number of instances of an element exceeds the constraint set by the schema's relevant maxOccurs
attribute.
ELMX_TOTAL_DIGITS_EXCEEDED
The maximum number of total digits specified for a value by the schema have been exceeded.
ELMX_TYPE_NOT_SET
An element that required its type to be specified using xsi:type did not have an xsi:type attribute.
ELMX_UNABLE_TO_READ_ATTRIBUTE_VALUE
Unable to read an attribute's value.
ELMX_UNABLE_TO_READ_ELEMENT_VALUE
Unable to read an element's value.
ELMX_UNDEFINED_ERROR
Used as an initialization value that can be later overwritten with a valid error code.
ELMX_UNEXPECTED_ALL_EVENT
Typically, an unknown element occurred in an xs:all construct.
ELMX_UNEXPECTED_ATTRIBUTE
An unknown attribute has been encountered.
ELMX_UNEXPECTED_ELEMENT_EVENT
Typically, an unknown element occurred in an xs:sequence or xs:choice construct.
ELMX_UNEXPECTED_EMPTY_ELEMENT
An element was empty when it should not have been.
ELMX_UNEXPECTED_ENTITY
An external entity was not declared as either PUBLIC or SYSTEM.
76
3.9 - Error Codes
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
ELMX_UNEXPECTED_EOM
The end of the message (or file) was encountered before the XML content was successfully parsed.
ELMX_UNKNOWN_XSI_TYPE
The type declared for an element using xsi:type is unknown.
ELMX_USER_DEFINED_1 to ELMX_USER_DEFINED_9
Error codes reserved for use by the user. These can be used when errors are returned from snippet
event handlers.
ELMX_VALUE_BAD_FORMAT
A value such as an integer, binary value, float and so on was not of a suitable format.
ELMX_VALUE_DOES_NOT_MATCH_FIXED_CONSTRAINT
The value of an element or attribute that is declared as fixed differs from the fixed value.
ELMX_VALUE_EXCEEDS_MAX
The value in the XML document is greater than the maximum specified by the schema.
ELMX_VALUE_EXCEEDS_MIN
The value in the XML document is less than the minimum specified by the schema.
ELMX_VALUE_NOT_ENUMERATED
The read value could not be matched against any of the allowed enumeration values.
3.10 - Generating makefiles
LMX has the ability to generate a makefile fragment suitable for inclusion into your own makefiles. This is
invoked using the -makefile command-line flag. The generated makefile fragment will be named
<output root>.makefile.
Additionally, if there is no file named makefile in the project output directory, then LMX will generate an
example makefile that demonstrates a possible use of the generated makefile fragment. You can then modify
this file according to your requirements as LMX will not overwrite it if it is already present.
An example of the generated makefile fragment is shown below. It contains a number of makefile variables
specifying the files used in the project. Note that it does not include any actual build instructions.
MYPROJECTCPPS = \
myproject.cpp
MYPROJECTTESTCPP = myproject-main.cpp
MYPROJECTOBJS = $(MYPROJECTCPPS:.cpp=.o)
MYPROJECTTESTOBJ = $(MYPROJECTTESTCPP:.cpp=.o)
MYPROJECTHS = $(MYPROJECTCPPS:.cpp=.h)
The expectation is that this makefile fragment will be included in another makefile written by the developer
that contains the actual build instructions, for example:
include myproject.makefile
3.10 - Generating makefiles
77
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
LMXLIB = -llmx # Modify using LMX -makefile-lib flag
.PHONY: all test clean
all: libmyproject.a
test: myproject-test
libmyproject.a: $(MYPROJECTOBJS)
-rm libmyproject.a
ar rcs libmyproject.a $(MYPROJECTOBJS)
myproject-test: $(MYPROJECTTESTOBJ) libmyproject.a
$(CXX) -o myproject-test $(MYPROJECTTESTOBJ) libmyproject.a $(LMXLIB)
%.o: %.cpp $(MYPROJECTHS)
$(CXX) -c $(CXXFLAGS) $(LMXCXXFLAGS) $(LMXFLAGS) $
<-o $@
clean:
rm $(MYPROJECTOBJS)
You can use the -makefile-lib flag to specify the name of the library containing the LMX Supporting
Software Runtime Source Code. For example, -makefile-lib lmx710 will include a -llmx710 in the
makefile.
Note that in order to build the test project, you will need to specify the -tframework command-line option,
or select the corresponding checkbox in the GUI interfaces when running LMX so that the necessary test
.cpp is generated.
3.11 - Use with Web Services
LMX supports a number of features that help developing web services. Specifically, LMX has a c_soap
template class that can be used to assemble the various components of a SOAP based web service, such as the
specific SOAP header, body and fault detail types.
LMX also has a c_winhttp class which provides HTTP client functionality for SOAP and REST style
client operations on the Windows platform. (A similar HTTP component for Linux platforms is due to be
included in a future release.)
Additionally LMX provides features for querying web service messages, such as that described in
3.28 - mustUnderstand / Comprehension Required.
3.11.1 - Handling SOAP Messages
To associate the various SOAP header, body and fault detail types that a SOAP messages can have, LMX
provides the c_soap template class. This is defined in lmxsoap.h. The code for this is built into the binary
versions of the LMX Supporting Software or available in source code form in the Professional edition. The
example at http://codalogic.com/lmx/download/?lmx-soap-example.zip shows a number of ways of using the
SOAP functionality.
The c_soap template takes up to three parameters, these being the class representing the SOAP bodies, the
class representing the SOAP headers and the class representing the fault detail. It is thus defined as:
78
3.11 - Use with Web Services
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
template< class Tbody, class Theader = c_soap_empty, class Tfault_detail = c_soap_empty >
class c_soap
{...}
For cases where a fault detail is required, but a header is not, a derivative of c_soap, called
c_soap_no_header, is defined as follows:
template< class Tbody, class Tfault_detail = c_soap_empty >
class c_soap_no_header
{...}
Unless stated otherwise, anything that applies to c_soap also applies to c_soap_no_header.
These classes store std::vectors of pointers to the respective body and header types, and a separate fault
class. The fault class stores a pointer to a fault detail object if required. The bodies() and headers()
methods return a reference the relevant std::vector of pointers and by using std::vector operations
such as push_back() and at() the overall SOAP message can be constructed or interrogated as required.
c_soap also contains a class for storing SOAP fault information, called c_soap_fault. This latter class
contains a std::vector of pointers to objects of the Tfault_detail template argument type.
c_soap_fault also contains the SOAP fault code, fault string and fault actor details if required, for which
there are appropriate getter and setter methods, e.g. get_fault_code(), set_fault_code(),
get_fault_string(), set_fault_string(), get_fault_actor(), and
set_fault_actor(). As the fault actor is optional, it's presence can be tested for using the
isset_fault_actor() method.
By default, the c_soap object deletes the objects that its std::vector containers point to when it is
destructed. If this behavior is not required, then the do_delete_on_destruct( false ); method
should be called before the object is destructed. If it is desired to have the c_soap object delete some objects
that are pointed to but not all, then the std::vector operations can be used to remove pointers to those
objects that it is desired to preserve prior to destructing the c_soap object.
If it is desired to marshal the SOAP message specified by a constructed c_soap instance, then one of
c_soap's marshal() methods can be called. There are a number of such methods which allow marshaling
to files, or memory for example.
Similarly, if a SOAP XML message is to be unmarshalled into the corresponding C++ objects, then one of
c_soap's unmarshal() methods can be called.
Finally, c_soap can also store a SOAPAction, accessible via the get_soap_action() and
set_soap_action() methods (located in c_soap_base). While SOAPAction is not part of a SOAP
message's XML contents, it can be convenient to group it together with the other SOAP information.
3.11.2 - HTTP Operations for Web Services
LMX includes the c_winhttp class (declared in lmxwinhttp.h) to provide HTTP client functionality for
the Windows platform. A derivation of c_winhttp called c_secure_winhttp provides secure HTTPS
client functionality. Unless stated otherwise, anything that applies to c_winhttp also applies to
c_secure_winhttp.
3.11.1 - Handling SOAP Messages
79
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
To allow customization of this functionality to your particular needs, these are provided in source code form
in both the Standard and Professional Editions of LMX. Similar HTTP/HTTPS client functionality for the
Linux platform is planned for a future release.
LMX does not provide HTTP server side functionality as the expectation is that a regular HTTP server such as
Apache or Microsoft IIS will provide this. LMX generated code can then be plugged into one of those
architectures in an appropriate way. (However, please let us know if you have different requirements.)
LMX HTTP (and HTTPS) operations consist of two main parts; creating a connection to a server and then
performing operations using the connection. The example mentioned above that is located at
http://codalogic.com/lmx/download/?lmx-soap-example.zip also includes examples of using the c_winhttp
class's functionality.
The server to connect to is specified using c_winhttp's constructor, which is:
c_winhttp( const std::string &ar_host,
size_t a_port = k_default_port,
bool a_is_secure = false );
Or that of c_secure_winhttp who's constructor is:
c_secure_winhttp( const std::string &ar_host,
size_t a_port = k_default_https_port );
The ar_host argument contains the name of the host being contacted. For example, if the desired operation
URI is http://example.com:8080/my/soap/service then ar_host should be set to
example.com.
c_winhttp allows a number of different types of web service operation to be performed once a connection
has been established. The following sections describe each of these operations.
3.11.2.1 - SOAP Message to Message Operations
Message to message SOAP operations are performed using the following method:
int soap_rpc( const std::string &ar_path,
const tlmx_uri_string &ar_soap_action,
const std::string &ar_request_xml,
std::string *ap_response_xml );
This operation sends the XML encoded SOAP message contained in ar_request_message to the server,
and places the reply into the std::string pointed to by ap_response_xml.
The ar_path argument contains the path part of the HTTP URI. For example, if the desired operation URI
is http://example.com:8080/my/soap/service then ar_path should be set to
/my/soap/service.
The ar_soap_action argument contains the SOAPAction.
The method returns 0 if the operation is successful and non-0 otherwise. The
get_error_description( int a_error ) method can be used to convert an error code into a text
based error description.
80
3.11.2 - HTTP Operations for Web Services
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
3.11.2.2 - SOAP Object to Object Operations
What LMX refers to as object to object SOAP operations are performed using the following method:
s_winhttp_error soap_rpc(
const std::string &ar_path,
const c_soap_base &ar_request,
c_soap_base *ap_response );
This operation marshals the objects aggregated by the c_soap object referenced by ar_request into an
XML message, and sends this to the server. The XML message returned from the server is unmarshalled and
placed in the c_soap object pointed to by ap_response.
The SOAPAction for the operation is retrieved from the c_soap object referenced by ar_request.
The ar_path argument has the same meaning as in the message to message operation case.
The method returns an instance of s_winhttp_error. As the method performs both LMX and HTTP
operations, this is an aggregate object capturing the various possible error conditions. If
s_winhttp_error's is_ok() method returns true then the operation was successful. If the operation
was unsuccessful, the operation phase at which the error occurred (e.g marshalling, HTTP transaction or
unmarshalling) can be determined by inspecting s_winhttp_error's phase member. Depending on the
phase at which the error occurred, either the lmx_error or http_error member can be inspected to find
out more specific information.
3.11.2.3 - Simple SOAP Operations
When a SOAP operation consists of only a single body part, without SOAP headers or faults, then a Simple
SOAP Operation can be performed using the following template function:
template< class Tin, class Tout>
s_winhttp_error simple_soap(
const std::string &ar_host,
size_t a_port,
const std::string &ar_path,
const tlmx_uri_string &ar_soap_action,
Tin *ap_in,
Tout **app_out = LMXNULL,
// Allows an empty SOAP body in the reply
bool a_is_secure = false )
Or, if it is not necessary to change the default port, the following template function:
template< class Tin, class Tout>
s_winhttp_error simple_soap(
const std::string &ar_host,
const std::string &ar_path,
const tlmx_uri_string &ar_soap_action,
Tin *ap_in,
Tout **app_out = LMXNULL,
// Allows an empty SOAP body in the reply
bool a_is_secure = false )
This is not a member of c_winhttp, but uses its services.
3.11.2.2 - SOAP Object to Object Operations
81
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
In these functions, the ar_host, a_port, ar_path, ar_soap_action, and a_is_secure have the
same meanings as described elsewhere in this section.
The ap_in argument takes a pointer to an object corresponding to the template argument Tin (whose code is
generated by LMX) that, when marshaled, will form the body of the SOAP request.
If the operation is successful, the function creates an unmarshalled object of type template argument Tout on
the heap. To be able to access this object, the calling function must have a pointer to a Tout object. A pointer
to this pointer is then supplied to simple_soap so that simple_soap can set its value. Note that if no
SOAP body is returned, the pointer to Tout in the calling function may be set to 0.
The return value of the function is an instance of s_winhttp_error, and is interpreted as described in the
object to object operation case above.
3.11.2.4 - RESTful Operations
REST style operations can be performed using the following c_winhttp member method:
int rest_query( const std::string &ar_path,
const std::string &ar_query,
std::string *ap_response_xml );
If the URI for the REST operation was http://example.com:8080/my/service?what=rest then
ar_path should be set to /my/service and ar_query should be set to what=rest. (The host is set
when the c_winhttp object is constructed.)
The XML response to the query is placed in the std::string pointed to by ap_response_xml. This
can subsequently be unmarshalled using LMX generated code.
If the method returns 0 then the operation was successful. A non-0 value indicates an error. If an error has
occurred, a description of the error can be obtained by calling the get_error_description( int
a_error ) method.
3.12 - Naming of Methods and Variables
LMX allows extensive customization of the variable and method names used in the generated code. There are
two parts to setting the names and variables in LMX; setting the prefixes and setting the base names. Both of
these operate independently.
The prefixes of methods and variables are configured using project configuration flags (also accessible via the
Windows UI). For example, if you wish the functions that are named by default as getenum_YYY() to be
getEnumYYY() then the project flag '-prefix-getenum' should be set to 'getEnum', either directly or
via the Windows UI. The prefixes set in this way have project wide scope. See the -prefix-??? flags in
section 3.8 - Command-line Flags for the set of prefixes that can be changed.
The base names of variables and methods (e.g. the YYY part in the examples above) can be automatically
converted from the form specified in the schema to CamelCase or underscore_separated either via the GUIs or
using the -naming camel or -naming underscore flag.
82
3.11.2.3 - Simple SOAP Operations
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
The base names of variables and methods can also be set by names files. LMX automatically performs name
collision avoidance, but if a schema is updated there is a small chance for a name to be assigned to a different
element/attribute in the new code to the one it was in the previous version. The names files address this
problem. (Note that in many scenarios such naming issues will not be a problem between versions of a
schema, so this feature need not be used in all cases.)
When the -names-out option is specified, LMX will generate a names file for each schema used in a
project. By default, the name of the names file will be the name of the schema appended with
'.lmxnames-out.xml'. For example, if the name of the schema file is 'mySchema.xsd', the generated
names file will be called 'mySchema.xsd.lmxnames-out.xml'. To avoid accidental overwriting, the
appended text on the names file needs to be modified to '.lmxnames-in.xml' before it can be used as an
input names file. An input names file will be read (for example, when processing the new version of the
schema) if the '-names-in' option is chosen.
Additionally, it is also possible to specify an alternative directory in which the names files are stored. This is
done using the -names-dir YYY flag; where YYY is the name of the directory (of folder) storing the
names files. This can allow the same schema to have different names files for different projects. In this case
the name of a names file is made up of the directory specified by the -names-dir flag, and the base name
of the schema file, suffixed by either the extension .lmxnames-out.xml or .lmxnames-out.xml. For
example, if -names-dir foo is specified and the schema file name is bar/mySchema.xsd, then the in
names file would be foo/mySchema.xsd.lmxnames-in.xml.
The names file also allows you to customize the names used in the code. If you are not able to use the
WinLMX version of LMX, the easiest way to do this is to run the code generator with the '-names-out'
option first, re-name the names file extension to '.lmxnames-in.xml', and edit the names in the file to
your convenience. When the naming is complete, run the code generator with the '-names-in' option
specified.
The schema for the names file is as follows:
<xs:schema targetNamespace="http://xml2cpp.com/lmx-customize.xsd"
elementFormDefault="qualified"
xmlns="http://xml2cpp.com/lmx-customize.xsd"
xmlns:xs="http://www.w3.org/2001/XMLSchema"
xmlns:clt="http://codalogic.com/xsdtypes">
<xs:element name='lmxCustomize'>
<xs:complexType>
<xs:sequence>
<xs:element name="element" type="item"
minOccurs="0" maxOccurs="unbounded"/>
<xs:element name="attribute" type="item"
minOccurs="0" maxOccurs="unbounded"/>
<xs:element name="group" type="item"
minOccurs="0" maxOccurs="unbounded"/>
<xs:element name="attributeGroup" type="item"
minOccurs="0" maxOccurs="unbounded"/>
<xs:element name="type" type="item"
minOccurs="0" maxOccurs="unbounded"/>
</xs:sequence>
<xs:attribute name="version" type="xs:token" default="1.0">
<xs:annotation><xs:documentation>
Use major.minor format
</xs:documentation></xs:annotation>
</xs:attribute>
3.12 - Naming of Methods and Variables
83
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
</xs:complexType>
</xs:element>
<xs:complexType name="item">
<xs:sequence>
<xs:element name="element" type="item"
minOccurs="0" maxOccurs="unbounded">
<xs:annotation><xs:documentation>
For naming child elements of complex type/complex
content.
</xs:documentation></xs:annotation>
</xs:element>
<xs:element name="attribute" type="item"
minOccurs="0" maxOccurs="unbounded"/>
</xs:sequence>
<xs:attribute name="name" type="xs:string">
<xs:annotation><xs:documentation>
Optional. The name of the item specified in the schema.
It may be absent if it is an anonymous type.
</xs:documentation></xs:annotation>
</xs:attribute>
<xs:attribute name="asVarName" type="clt:asciiString">
<xs:annotation><xs:documentation>
Optional - The name of the item when it is used as a
variable's name, as in: 'c_type var_name;' It must
be locally unique, but need not be globally unique.
</xs:documentation></xs:annotation>
</xs:attribute>
<xs:attribute name="asTypeName" type="clt:asciiString">
<xs:annotation><xs:documentation>
Optional. The name of the item when it is used as a
C++ type, e.g. 'c_type name;' It must be globally unique.
</xs:documentation></xs:annotation>
</xs:attribute>
</xs:complexType>
</xs:schema>
3.13 - Adapting LMX to Your Environment
In some cases you may find that the various classes and types included with LMX are not what you want.
LMX has been designed so that it is easy to change the types and classes used.
The majority of configuration options are specified in the lmxuser.h file. Each user configuration section is
marked with C++ comments of the form USER: A_SECTION.
There are a number of ways in which you can customize the features controlled by the lmxuser.h file. For
example, you can:
1. Modify lmxuser.h directly. This is the least recommended method because it means if a new
version of LMX is issued, then you may have to update a new copy of lmxuser.h, potentially
84
3.13 - Adapting LMX to Your Environment
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
introducing bugs through human error.
2. If only minor configuration is required, and the changes needed are of the more common type, the
build environment (such as project files or make files) can arrange for the appropriate #defines to
be set to get the desired configuration.
3. If more substantial configuration is required, then it is suggested that the various #defines are
included in their own .h file and that .h file in turn includes lmxuser.h. You can then use LMX's
-include-h flag (or "Advanced"->"Include Files in Generated.h Files..." menu option in
WinLMX) to include the file in the generated code.
4. As an alternative to the previous option, you can include your configuration changes in the files called
lmxuser-def.h and lmxuser-defs-end.h. lmxuser.h will pull in the file
lmxuser-def.h if your project specifies the definition LMX_WANT_USER_DEFS and pull in
lmxuser-defs-end.h if your project specifies LMX_WANT_USER_DEFS_END.
The last two methods are appropriate when complete sections of lmxuser.h need to be changed. It will be
observed that each section is included in its own conditional compilation guard block. If a particular feature
can not be changed using the #defines that LMX supports, then copy the entire section into the separate
include file mentioned in the latter methods, and edit it accordingly. Then #define the appropriate guard
value so that the default section in lmxuser.h is ignored.
3.13.1 - Modifying Schema Type to C++ Type Mapping
XML schema type to C++ type mapping is specified in the USER: BASIC_TYPES and USER:
COMPOUND_TYPES sections of the lmxuser.h file. The type mapping is achieved using simple C++
typedef statements to avoid you having to learn additional LMX code-generator configuration options.
Changing the types used is simply a matter of modifying the typedef statements. For example, if you wish
to use your own date class, you could define a mapping of:
typedef c_my_date
tc_date;
The following table describes the relationship between the schema types, the C++ typedef name used to
represent it, and the default C++ type that the typedef name is mapped to.
3.13.1 - Modifying Schema Type to C++ Type Mapping
85
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
XSD Type
C++ Intermediate
Type
Default C++ Type
boolean
bool
bool (This type can not be
byte
short
int
tlmx_int8
tlmx_int16
tlmx_int32
long
tlmx_int64
unsignedByte
unsignedShort
unsignedInt
tlmx_uns8
tlmx_uns16
tlmx_uns32
unsignedLong
tlmx_uns64
signed char
short
int
For MSVC: __int64
For GCC: long long
unsigned char
unsigned short
unsigned int
For MSVC: unsigned
__int64
For GCC: unsigned long
long
integer, nonPositiveInteger, negativeInteger,
tc_big_int
nonNegativeInteger, positiveInteger
decimal
tc_decimal
float
tlmx_float
double
tlmx_double
duration
tc_duration
dateTime
tc_datetime
time
tc_time
date
tc_date
gYearMonth
tc_gyearmonth
gYear
tc_gyear
gMonthDay
tc_gmonthday
gDay
tc_gday
gMonth
tc_gmonth
hexBinary, base64Binary
tc_binary
string, normalizedString, language, token, Name,
NCName, ID, IDREF, IDREFS, ENTITY, ENTITIES, tlmx_unicode_string
NMTOKEN, NMTOKENS, NOTATION
anyURI
tlmx_uri_string
QName
tc_qname
Table 5: XSD to C++ Type Mappings
customized)
c_big_int
c_decimal
float
double
c_duration
c_datetime
c_time
c_date
c_gyearmonth
c_gyear
c_gmonthday
c_gday
c_gmonth
c_binary
std::wstring
std::wstring
ct_qname<tlmx_uri_string>
Similarly tlmx_uri_string controls the type of string used for URIs. By default this is mapped to
std::wstring, but if you wish it to be mapped to std::string then arrange for the #define variable
LMX_NARROW_URI_STRINGS to be 1. If LMX_NARROW_URI_STRINGS is not specified, then it is
automatically given the same value as LMX_NARROW_UNICODE_STRINGS.
Note that by default xs:float is mapped to a C++ float. Such a single precision floating point number
has about 7.5 decimal places of precision; a precision that can not be readily expressed using C++'s output
86
3.13.1 - Modifying Schema Type to C++ Type Mapping
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
converters. Consequently, LMX outputs C++ floats using 7 decimal places of precision (as opposed to 8
places which typically results in sub-optimal formatting), thus losing the ability to express the entire precision
of the type. Therefore, if you require the full precision of the xs:float type it is recommended that you
map it to C++'s double type.
In addition to changing the C++ types used to represent the LMX types, LMX needs to know whether each
integer type is a type native to your C++ compiler (e.g. VC++'s __int64 is considered native), or a special
class. This is done using the USER: INT_NATIVE_FOREIGN section of the lmxuser.h file. This section
contains a number of macros specific to each integer type that either maps to a native value or a foreign value.
You should change the mapping of these macros according to your platform requirements.
See also 3.23 - Pattern Facet Handling Customization.
3.13.1.1 - Choosing Between Wide and Narrow Unicode Strings
Of particular note is the mapping of tlmx_unicode_string. By default this is mapped to
std::wstring. If you wish to use 'narrow' strings to store parsed strings, you can arrange your project C++
build configuration to define the #define variable LMX_NARROW_UNICODE_STRINGS to be 1.
Alternatively, you can directly modify the mapping in lmxuser.h to be std::string. All other changes
will be automatically taken care of by the C++ compiler. Or you can map tlmx_unicode_string to
some other string class.
3.13.1.2 - Input/Output Converters
As part of changing the type mappings used, it may be necessary to define additional input and output
converters. These are defined in the USER: TEMPLATE_INPUT_CONVERTERS, and USER:
TEMPLATE_OUTPUT_CONVERTERS sections of lmxuser.h.
All types are converted from their string form to their type form via the templated function v_to_o (short for
'value_to_object'.). If your type (such as a date class) has an operator = member that can convert a
std::string value to appropriate internal values, then the default v_to_o template function will take care of
this and no additional work is required. If your class does not support this, then you should provide a template
specialization of v_to_o that will do the conversion. Examples of such specializations are shown in the
USER: TEMPLATE_INPUT_CONVERTERS section. One option for such conversion that eases parsing is to
first convert the input to an appropriate LMX class, and then use the public members of that class to populate
your class.
For output, the LMX code generator arranges for each parameter to be wrapped in an instance of an as_xml
io manipulator. These manipulators are defined in the USER: TEMPLATE_OUTPUT_CONVERTERS. If your
class or type has an io operator (i.e. <<) that generates data in a suitable form for output in an XML document,
then the default template will be sufficient and no additional specialization is required. If this is not the case,
then a specialization of the output template is required. For example, a specialization for your type might
appear as:
template <>
std::ostream & operator << ( std::ostream & ar_os,
const c_as_xml<c_my_date> &ar_v )
{ ... }
3.13.1.1 - Choosing Between Wide and Narrow Unicode Strings
87
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Within the template specialization function, the quantity ar_v.p is a pointer to the item to be output. You
can look at some of the other specializations to see how to write a specialization.
3.14 - Augmenting Generated Classes With Your Own Code
LMX allows you to augment the generated classes with your own code. Specifying the -snippets flag will
cause LMX to output a pair of comments towards the beginning of each generated class in a .h file. The
markers have the form:
//:snippet:start YYY
//:snippet:end YYY
where YYY is the name of the class in which the markers are included. You can add your own code between
the markers. When code is re-generated, LMX first reads the original .h file and extracts the code between the
snippet markers. The code is then put back when the new code is generated. Non-header code such as method
definitions should be put in a separate .cpp file. You should not change the generated .cpp files.
As a precaution, it is recommended that you have a backup of the original .h file (for example, in a version
control system) before re-generating code to replace it.
If you wish to augment a class with lots of code it may be appropriate to consider defining a separate class,
and then including the declaration of an instance of that class in the snippets section. This further separates
your code and the generated code. To do this it may be necessary to use the -include-h to inform LMX of
a header file defining your class, and implement an on_lmx_construct() snippet event handler (see
3.14.1 - Augmenting Generated Classes with Snippet Event Handlers below) that can inform your separate
class of the this pointer (or a reference) of the generated class. For example, in this case the snippets section
for a class might appear as:
//:snippet:start YYY
c_my_details_class my_class;
void on_lmx_construct() { my_class.set_parent( this ); }
//:snippet:end YYY
As described above this procedure requires editing the generated .h file. If you do not wish to do this, the
-file-ext-snippets flag allows the specification of the file extension for files that store snippets. For
example, if -file-ext-snippets .lmxsnips is specified, LMX will look in foo.lmxsnips for
snippets rather than in foo.h.
3.14.1 - Augmenting Generated Classes with Snippet Event Handlers
The behavior of the LMX generated classes can be augmented by what are called snippet event handlers.
When generating code, LMX looks for the declaration of specially named functions in each class's snippets
section. If the name of a special function is found, then code is generated to call that function. The defined
snippet event handler functions are:
void on_lmx_construct();
void on_lmx_destruct();
lmx::elmx_error on_lmx_marshal_start( lmx::c_xml_writer & r_writer,
const char * p_element_name ) const;
88
3.13.1.2 - Input/Output Converters
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
lmx::elmx_error on_lmx_marshal_end( lmx::c_xml_writer & r_writer ) const;
// Currently return code for on_lmx_marshal_start_tag() is ignored
lmx::elmx_error on_lmx_marshal_start_tag( lmx::c_xml_writer & r_writer ) const;
lmx::elmx_error on_lmx_marshal_body_start( lmx::c_xml_writer & r_writer ) const;
lmx::elmx_error on_lmx_marshal_body_end( lmx::c_xml_writer & r_writer ) const;
lmx::elmx_error on_lmx_alt_marshal_attribute_XYZ( lmx::c_xml_writer & r_writer ) const;
lmx::elmx_error on_lmx_alt_marshal_element_XYZ( lmx::c_xml_writer & r_writer ) const;
bool on_lmx_is_occurs_ok() const;
lmx::elmx_error on_lmx_unmarshal_outer_start( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_outer_end( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_start( lmx::c_xml_reader & r_reader,
const std::string & r_name );
lmx::elmx_error on_lmx_unmarshal_start( lmx::c_xml_reader & r_reader,
const std::string & r_name );
lmx::elmx_error on_lmx_unmarshal_end( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_attributes_check(
lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_body_start( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_body_end( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_attribute_XYZ( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_alt_unmarshal_attribute_XYZ( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_unmarshal_element_XYZ( lmx::c_xml_reader & r_reader );
lmx::elmx_error on_lmx_alt_unmarshal_element_XYZ( lmx::c_xml_reader & r_reader );
The snippet event handler functions are called in the following situations:
on_lmx_construct()
Called immediately after the code generated to initialize the object. This allows initializing
snippet data.
on_lmx_destruct()
Called immediately before the code generated to destroy an object. This is intended to give
the opportunity to do special handling of the data before it is destroyed and destroy any data
introduced through snippets.
on_lmx_is_occurs_ok()
Called in place of the generated version of is_occurs_ok() at the start of the marshalling
operation.
on_lmx_marshal_start()
3.14.1 - Augmenting Generated Classes with Snippet Event Handlers
89
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
Called at the start of the object's main top-level marshaling method. It can be used to
implement additional user defined checks on the data before it is marshaled.
on_lmx_marshal_end()
Called at the end of the object's main top-level marshalling method just prior to the final
return statement.
on_lmx_marshal_start_tag()
Called while outputting attributes in an element's start tag. It can be used to output additional
attributes. (Note that the return code from this function is currently ignored. This situation
may change in future however.)
on_lmx_marshal_body_start()
Called at the start of an elements marshal_body() method.
on_lmx_marshal_body_end()
Called at the end of an elements marshal_body() method. It can be used to output
additional elements.
on_lmx_alt_marshal_attribute_XYZ()
Allows you to specify alternative marshalling behavior for attribute XYZ.
on_lmx_alt_marshal_element_XYZ()
Allows you to specify alternative marshalling behavior for element XYZ.
on_lmx_unmarshal_outer_start()
Called at the start of the outer-most, top-level object's main unmarshaling method.
on_lmx_unmarshal_outer_end()
Called at the end of the outer-most, top-level object's main unmarshaling method.
on_lmx_unmarshal_start()
Called at the start of the object's main top-level unmarshaling method.
on_lmx_unmarshal_end()
Called at the end of the object's main top-level unmarshalling method just prior to the final
return statement. It can be used to implement additional user defined checks on the data
immediately after it is unmarshaled (for example co-occurrence constraints).
on_lmx_unmarshal_attributes_check()
Called in the unmarshal_attributes_check() method.
on_lmx_unmarshal_body_start()
Called at the start of an elements unmarshal_body() method.
on_lmx_unmarshal_body_end()
Called at the end of an elements unmarshal_body() method.
on_lmx_unmarshal_attribute_XYZ()
Called when the attribute stored in the C++ variable m_XYZ has been unmarshalled and
stored in the object.
90
3.14.1 - Augmenting Generated Classes with Snippet Event Handlers
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
on_lmx_alt_unmarshal_attribute_XYZ()
Allows you to specify alternative unmarshalling behavior for attribute XYZ.
on_lmx_unmarshal_element_XYZ()
Called when and instance of the element stored in the C++ variable m_XYZ has been
unmarshalled and stored in the object.
on_lmx_alt_unmarshal_element_XYZ()
Allows you to specify alternative unmarshalling behavior for element XYZ.
It will be noticed that the marshal and unmarshal event handlers return an lmx::elmx_error value. If the
value returned is not lmx::ELMX_OK then the marshaling or unmarshaling operation is terminated and the
error code returned. To provide custom error feedback, the lmx::elmx_error enumeration contains
enumeration values ELMX_USER_DEFINED_1 to ELMX_USER_DEFINED_9.
Similarly, the marshal and unmarshal event handlers take as an argument a reference to the marshaling
operations lmx::c_xml_writer object, or the unmarshaling operations lmx::c_xml_reader object.
If the -alt-xml-writer or -alt-xml-reader command-line flags are used to change the names of
these two latter objects, then a reference to the specified objects will be passed. This gives the opportunity to
pass additional data into the event handlers.
One use of the on_lmx_alt_unmarshal_element_XYZ() snippet event handler is to enable the user
to ignore particular elements. The elmx_error c_xml_reader::ignore_element(...) method
is provided to help with the implementation of such a handler. For example:
lmx::elmx_error c_MyElement::on_lmx_alt_unmarshal_element_ChildElement(
lmx::c_xml_reader & ar_reader,
const lmx::s_event_map ac_event_map[] )
{
return ar_reader.ignore_element( ac_event_map );
}
3.15 - DTDs and Namespaces
DTDs do not explicitly support XML namespaces. However, LMX can extract namespace information from
DTDs via attribute definitions that contain the xmlns prefix and have a #FIXED specification. For example,
to declare that the namespace prefix 'ns' should be associated with the namespace
'http://xml2cpp.com/example', use the following attribute definition:
<!ATTLIST ns:MyElement xmlns:ns CDATA #FIXED "http://xml2cpp.com/dtd-test.html">
To set the default namespace, use:
<!ATTLIST MyElement xmlns CDATA #FIXED "http://xml2cpp.com/dtd-test.html">
3.16 - Debugging and Handling Errors
LMX is designed to be flexible about how run-time errors are handled. It also includes features to aid
debugging.
3.15 - DTDs and Namespaces
91
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
There are three categories of error that can occur. These are parse errors, write errors, and class access errors.
Any error that occurs in one of these categories is first passed to a category specific error handling function.
The functions are c_xml_reader::handle_error, c_xml_writer::handle_error (both in
lmxparse.h), and lmx_error in lmxuser.h respectively. These functions allow you to tailor the error
handling to your requirements.
3.16.1 - Reporting Errors
LMX includes a number of functions to help report errors to a user.
To convert an error code to a description, the global get_error_description function can be called.
The prototype for this function is:
const char *get_error_description( elmx_error a_error_code );
The error code is entered in a_error_code and the return value points to a const char string describing the
error.
c_xml_reader and c_xml_writer both include a get_error_message function. The prototypes for
the functions are:
virtual std::string & c_xml_reader::get_error_message( std::string *ap_message );
virtual std::string & c_xml_writer::get_error_message( std::string *ap_message );
An error message, including the name of the element or attribute in question, and the line number (if
applicable) will be inserted with a description into the string pointed to by ap_message. The function
returns a reference to *ap_message. The function is virtual to allow user customization of the message
format.
The line number that the error occurred on can be obtained by calling:
int c_xml_reader::get_line_no();
c_xml_writer does not keep track of line numbers due to its use of iostreams, and hence has no
corresponding function.
c_xml_reader and c_xml_writer allow the element or attribute name applicable to an error to be
obtained by calling:
const std::string & c_xml_reader::get_error_item_name();
const std::string & c_xml_writer::get_error_item_name();
3.16.2 - Changing the Error Handling Behavior
As supplied, the above functions support the 'return code' paradigm for reporting errors and the code
generated by LMX allows the return code to be propagated back to the original calling function.
You can select exception-based error reporting by setting the LMX_USE_USER_VALIDATE_EXCEPTIONS
and LMX_USE_EXCEPTIONS pre-processor constants to 1. The constants are defined in the USER:
CTRL_DEFS section of lmxuser.h. LMX_USE_USER_VALIDATE_EXCEPTIONS selects exceptions
92
3.16 - Debugging and Handling Errors
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
when validating user access to the objects and LMX_USE_EXCEPTIONS selects exceptions when marshaling
and unmarshaling. If exception-based error reporting is selected, exceptions are reported using instances of the
c_lmx_exception and c_lmx_reader_exception classes defined in lmxinternals.h.
c_lmx_reader_exception is used to signal exceptions from c_xml_reader, and
c_lmx_exception is used for all other exceptions. There public interface is as follows:
class c_lmx_exception
{
public:
c_lmx_exception( elmx_error a_error_code );
elmx_error get_error()
const;
const char *what() const throw();
virtual
};
class c_lmx_reader_exception : public c_lmx_exception
{
public:
c_lmx_reader_exception( elmx_error a_error_code,
int a_line );
int get_line() const;
};
If you wish to use an alternative error reporting strategy, then define new classes that derive from
c_xml_reader and c_xml_writer and define appropriate user_handle_error methods.
3.16.3 - Conditional Error Handling
The error handling structure used by LMX allows you to conditionally ignore errors. For example, if the XML
being parsed has been human generated, then for certain errors you may wish to continue parsing and
reporting errors after the first error has occurred. This can be done by modifying the
c_xml_reader::user_handle_error(...) and
c_xml_writer::user_handle_error(...) methods to return lmx::ELMX_OK rather than the
error that occurred. As c_xml_reader::user_handle_error(...) and
c_xml_writer::user_handle_error(...) are virtual methods an alternative strategy is to define
classes that extend c_xml_reader and c_xml_writer and define appropriate
user_handle_error(...) methods therein.
Note that care should be taken when adopting this strategy as it is not possible to proceed after certain errors
have occurred (such as an unexpected end of message). Hence it is recommended that your code explicitly
defines the errors that are being ignored rather than having a default of ignoring all errors and defining those
that are not ignored.
3.16.4 - Collecting Debug Error Information when using Convenience
Methods
The convenience methods for unmarshaling and marshaling (See 2.3 - Unmarshaling (simple form) and
2.4 - Marshaling (simple form)) are intentionally light-weight, and in their simplest form only return an error
code when an error occurs.
However, to aid debugging, in Debug mode the global debug_error object, which has the type
s_debug_error, is set with any information about the latest error.
3.16.2 - Changing the Error Handling Behavior
93
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
If use of a global object for debugging is not suitable, then the convenience methods can also be called with
an instance of s_debug_error that will be populated with any error information.
s_debug_error is declared in lmxinternals.h and has the following declaration:
struct s_debug_error
{
elmx_error error;
std::string item_name;
const char * p_prog_file;
int prog_line;
int xml_line;
LMX_PDECL s_debug_error();
LMX_PDECL void clear();
LMX_PDECL void set(
elmx_error error_in,
const std::string & r_item_name_in,
const char * p_prog_file_in,
int prog_line_in,
int xml_line_in = -1 );
LMX_PDECL const char * get_description() const;
LMX_PDECL const char * get_description(
const s_custom_error_description * p_custom_descriptions ) const;
LMX_PDECL std::string to_string() const;
LMX_PDECL std::string to_string(
const s_custom_error_description * p_custom_descriptions ) const;
LMX_PDECL std::ostream & to_stream( std::ostream & r_os_in ) const;
LMX_PDECL std::ostream & to_stream(
std::ostream & r_os_in,
const s_custom_error_description * p_custom_descriptions ) const;
};
LMX_PDECL std::ostream & operator << (
std::ostream &ar_os,
const s_debug_error &ar_debug_error );
LMX_PDECL extern s_debug_error debug_error;
#define LMX_OUTPUT_DEBUG_ERROR( xstream ) ...
The to_string() method returns the combined debug information as a suitable message in a string.
The to_stream() method outputs the combined debug information in a suitable message to the specified
stream.
The variants with the const s_custom_error_description * parameter allow setting custom
mappings of LMX error codes to descriptions. This is intended specifically for the LMX error codes of the
form ELMX_USER_DEFINED_?.
Mappings are specified by an array of structs of the form:
struct s_custom_error_description
{
elmx_error
code;
const char *
p_description;
};
94
3.16.4 - Collecting Debug Error Information when using ConvenienceMethods
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
For example:
lmx::s_custom_error_description custom_error_descriptions[] = {
{ lmx::ELMX_USER_DEFINED_1, "Out of range" },
{ lmx::ELMX_USER_DEFINED_2, "A must be present with B" },
{ lmx::ELMX_OK, LMXNULL }
};
Note the end of the array is identified by the entry { lmx::ELMX_OK, LMXNULL }.
If you want to disable debug_error being used in your code during Debug mode, then arrange for the
#define LMX_DEBUG_CAPTURE_ERROR to have the value 0. Conversely, if you want to force
debug_error to be in your code, even in Release mode (but see below), then arrange for the #define
LMX_DEBUG_CAPTURE_ERROR to have the value 1.
In Debug mode, or when LMX_DEBUG_CAPTURE_ERROR is set to 1, the LMX_OUTPUT_DEBUG_ERROR(
xstream ) macro will output the error information to the stream xstream. In Release mode and when
LMX_DEBUG_CAPTURE_ERROR is not set to 1 the LMX_OUTPUT_DEBUG_ERROR( xstream ) macro
has no effect.
The output operator function allows code such as:
std::cout << debug_error;
To access other debug information just access the data members directly.
If additional flexibility is required for error reporting, it is suggested that the Advanced forms of marshaling
and unmarshaling be used (See 3.5 - Marshaling (advanced forms) and 3.2 - Unmarshaling (advanced forms)).
3.16.5 - Debugging Support
The common error handling functions provided by LMX aid debugging support. Setting a break point in the
above-mentioned functions can facilitate detecting problems during parsing, writing or accessing the class
members. It is then a simple matter of tracing back through the stack to see why the error is being reported.
LMX also aids the development process when your code accesses the generated classes. You can choose
whether facet validation is done when you write data to a generated member functions by defining the
compiler definition LMX_USER_VALIDATE in the USER: CTRL_DEFS section of the lmxuser.h file. If this
directive is defined, any such errors that occur will first be reported to the lmx_error function in lmxuser.h,
and then optionally the return code of that function is returned to the caller. Alternatively, an exception can be
thrown, or an assert statement can be added to lmx_error. This feature supports the 'fail fast' paradigm
and allows errors to be detected close to where they occur.
LMX also generates functions called is_occurs_ok() and check() for each class. These test whether
the minimum cardinality requirements for the class have been satisfied. They thus allow you to test whether
you have set a sufficient number of the class members for the class to be valid. They test the cardinality
constraints for both elements and attributes. One way to use these functions is in an assert statement once
you have set all the appropriate members, e.g.:
c_NAME & name_ref = top.get_NAME();
name_ref.set_CHILD( 12 );
name_ref.set_CHILD2( 1.2 );
3.16.5 - Debugging Support
95
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
assert( name_ref.is_occurs_ok() );
See 2.5.15 - Run-time Checking for more information.
3.17 - Adding External Character Set Transcoders
The LMX XML parser natively supports UTF-8, UTF-16 (BE & LE), UCS2 (BE & LE), ISO-8859-1 and
US-ASCII. If support for additional character sets is required then these can be configured externally.
Examples of how to configure external transcoders are shown in the installed files
include/lmx-transcoder-example.h and src/lmx-transcoder-example.cpp.
There are two interface classes used to support external transcoders:
A const instance of a class implementing the lmx::c_external_transcoder_factory interface is
used to create instances of the classes that actually do the character transcoding. When a new transcoder
object is required LMX calls the interface's get_transcoder( c_read & r_reader_in,
c_error & r_error_in, c_get_as_utf8::e_encoding_mode
initial_encoding_mode_in, const char * p_encoding_name_in ) method. This should
return a pointer to a transcoder object, or NULL if no suitable transcoder is found. LMX will delete the
returned transcoder object when it is no longer required.
The class implementing the lmx::c_external_transcoder_factory interface can be specified to
LMX on a permanent basis using the static
c_xml_reader::set_default_external_transcoder_factory() method or on a per-parsing
session basis by calling the set_external_transcoder_factory() method on the relevant instance
of the c_xml_reader class (or derivative thereof).
Each transcoder should implement the lmx::c_external_transcoder interface class and implement
the c_external_transcoder::s_result read() method. The
c_external_transcoder::s_result object includes the character just read in the c member
variable, and optionally a pointer to a NULL-terminated array of UTF-8 encoded bytes that LMX should read
before asking the transcoder for new characters in the mp_utf8_expansion member variable. If no such
additional UTF-8 bytes are present then s_result.mp_utf8_expansion should be set to NULL. If
there is no input available then the transcoder class should return c_read::k_eof in s_result.c. If an
error occurs then this should be signaled to the c_error error class referenced by r_error_in and
c_read::k_eof returned in s_result.c.
3.18 - Adding Extra Namespace Information
When a document contains xs:anyAttribute or xs:any, it may be necessary to provide extra
namespace information to the unmarshaling and marshaling operations.
If a section of xs:any XML has been extracted while unmarshaling it is possible that the start tag of the value
part of the instance does not include appropriate namespace information. Without suitable namespace
information it is not possible to unmarshal its contents using another module of unmarshaling code. To
remedy this, c_xml_reader allows additional namespace information to be added to it prior to initiating
96
3.17 - Adding External Character Set Transcoders
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
unmarshaling. The functions that support this are:
void c_xml_reader::add_namespace_mapping( const std::string & r_ns_prefix, const std::string
void c_xml_reader::add_default_namespace_mapping( const std::string & r_namespace );
Where:
ar_ns_prefix
is the namespace prefix to be assigned to the namespace.
ar_namespace
is the actual namespace URI, e.g. "http://codalogic.com/example".
add_default_namespace_mapping( namespace ); is effectively an alias for
add_namespace_mapping( "", namespace );.
An example of adding extra namespace information is:
lmx::c_xml_reader l_reader( l_low_reader );
l_reader.add_namespace_mapping( "alt", "http://codalogic.com/example" );
c_MyElement l_item;
lmx::elmx_error l_error = l_item.unmarshal( l_reader );
Equally, when writing XML that contains xs:anyAttribute or xs:any additional namespace
information may need to be added to the c_xml_writer class. This is supported by the following methods:
void c_xml_writer::add_namespace( const std::string & r_ns_prefix, const std::string & r_namespac
void c_xml_writer::add_default_namespace( const std::string & r_namespace );
Where:
ar_ns_prefix
is the namespace prefix to be assigned to the namespace.
ar_namespace
is the actual namespace URI, e.g. "http://codalogic.com/example".
Again, add_default_namespace( namespace ); is an alias for add_namespace( "",
namespace );.
An example of their use is:
lmx::c_xml_writer l_writer( l_out_stream );
l_writer.add_namespace( "alt", "http://codalogic.com/example" );
l_item.marshal( l_writer );
3.19 - Adding Schema Location Information
You can specify to the c_xml_writer class that you want schema location information to be output in the
generated XML. This will cause either the xsi:schemaLocation or
xsi:noNamespaceSchemaLocation attributes to be output in the marshaled code. This is done using
3.18 - Adding Extra Namespace Information
97
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
the following methods:
void c_xml_writer::add_schema_location( const std::string & r_ns_uri, const std::string & r_location
void c_xml_writer::add_no_namespace_schema_location( const std::string & r_location );
Where:
ar_ns_uri
is the namespace URI associated with the schema, e.g. "http://codalogic.com/example".
ar_location
is the location hint for the actual schema file, e.g. "http://codalogic.com/schemas/example.xsd".
add_schema_location() specifies the location for a schema that has a namespace URI associated with
it.
add_no_namespace_schema_location() specifies the location of a schema that has no namespace
associated with it.
Multiple calls can be made to add_schema_location() to specify the locations of multiple schemas
with namespaces.
Only one call to add_no_namespace_schema_location() may be made.
If add_schema_location() is called, then add_no_namespace_schema_location() must not
be called, and vice versa.
3.20 - Disabling Output of XML Namespaces
When generating XML fragments that will be inserted by other means into other XML instances it may be
appropriate to not have the generated XML include the xmlns namespace attributes. This can be achieved by
calling the disable_ns_map_output() method on an instance of a c_xml_writer object. For
example:
c_MyObject myObject;
std::ostringstream sos;
lmx::c_xml_writer writer( sos );
writer.disable_ns_map_output();
myObject.marshal( writer );
// N.B.
This requires using the advanced forms of marshalling as described in 3.5 - Marshaling (advanced forms).
3.21 - Adding an XMLDecl to the XML output
By default when using the Convenience methods for marshalling (2.3 - Unmarshaling (simple form)), the
XMLDecl is output (the part that looks like <?xml version="1.0"?>). This can be configured on a
global basis using the c_xml_writer::set_convenience_options( t_writer_options )
and c_xml_writer::get_convenience_options() methods, where. t_writer_options is
defined as:
98
3.19 - Adding Schema Location Information
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
typedef int t_writer_options;
static const t_writer_options include_xml_decl;
static const t_writer_options include_standalone;
The t_writer_options can be combined by using the C++ OR operator (e.g. include_xml_decl |
include_standalone). An XMLDecl is output when the include_xml_decl option is included. The
standalone attribute can be included in the XMLDecl by specifying the include_standalone option.
Use of the include_standalone option implies include_xml_decl, and so it is not necessary to
specify include_xml_decl when selecting the standalone attribute. To disable output of an XMLDecl,
use set_convenience_options( 0 ).
To enable instance by instance configuration of whether the XMLDecl is output it is recommended to use the
advanced forms of marshalling (3.5 - Marshaling (advanced forms)) rather than the above methods.
By default c_xml_writer does not output the optional XMLDecl.
You can instruct c_xml_writer to output the XMLDecl by using the constructor that takes a
t_writer_options argument. This has the form:
c_xml_writer( std::ostream &ar_os,
t_writer_options a_options,
const char *ap_tab = "\t",
const char *ap_nl = "\n" );
Thus an example of how to construct an instance of c_xml_writer that outputs the XMLDecl is:
lmx::c_xml_writer l_writer( l_out_stream,
lmx::c_xml_writer::include_xml_decl );
An example of selecting an XMLDecl that contains the standalone attribute is:
lmx::c_xml_writer l_writer( l_out_stream,
lmx::c_xml_writer::include_standalone );
3.22 - Handling Multiple Schemas
3.22.1 - Handling Multiple Independent Schemas
If your project includes multiple independent schemas (or independent sets of schema) then we recommend
generating the code for each independent schema set separately and instructing LMX to generate the code into
separate C++ namespaces using the -cns YYY flag or via the WinLMX "Generated Code C++ Namespace"
group box on the "C++ Names & Namespaces" tab. For example, if you have schemas s1.xsd and s2.xsd
you could run LMX and instruct it to generate code for s1.xsd into C++ namespace s1 (using -cns s1)
and then run LMX again and instruct it to generate code for s2.xsd into the s2 C++ namespace (using
-cns s2). In your C++ files that #include the generated code you can use statements such as "using
namespace s1;" and "using namespace s2;" to avoid qualifying each use of a variable with a
namespace prefix.
While the above is our recommended method for handling this situation, if it is not acceptable in your
environment, you can use the -prefix-declash YYY flag to avoid name clashes when you come to link
3.21 - Adding an XMLDecl to the XML output
99
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
your generated files. In this scenario you could use -prefix-declash s1_ when generating code for
s1.xsd and -prefix-declash s2_ when generating code for s2.xsd.
3.22.2 - Multiple Schemas that Share Common Schemas
A common scenario is when two or more schemas share one of more common imported or included schemas.
In this case problems occur if the schemas are compiled separately due to duplicate code being generated.
To address this problem, all schemas that share common schemas should have code generated for them in a
single LMX compilation operation. As an example, assume that schema1.xsd and schema2.xsd both
share definitions from common.xsd. Rather than compiling schema1.xsd and schema2.xsd
separately, schema1.xsd and schema2.xsd should be compiled together. Using the command-line
interface, this can be done using a DOS command-line such as:
lmx
schema1.xsd
++
schema2.xsd
+
common.xsd
outbase
Or a Linux command-line such as:
linmx
schema1.xsd
++
schema2.xsd
+
common.xsd
outbase
From 2.1.2 - Code Generation Using the Windows (DOS) Command Line Version it can be seen that the +
sign informs LMX of a schema file that contains additional definitions that need to be imported. However,
global elements in such a schema file are not treated as possible root elements for the XML instances. The ++
sign similarly informs LMX of a schema file that contains additional definitions, but in this case the global
elements are considered to be possible root elements for a valid XML instance. Hence, specifying ++
schema2.xsd on the command-line will ensure that the global elements in schema2.xsd will be treated
as valid XML instance root elements, giving the desired effect.
The same results can be achieved using WinLMX. In this case, on the Schema Files tab, enter
schema1.xsd in the Base Schema File edit box. Enter schema2.xsd and common.xsd in the
Additional Imported Schema Files section using the relevant Add... button. Then select
schema2.xsd in the listview and press the Edit... button. In the dialog box that appears, check the
Treat as Global Schema checkbox and press OK. When the dialog box closes a small globe symbol
will be shown next to schema2.xsd's name. This signifies that global elements in the schema will be
treated as valid root elements in XML instances. Compile as usual.
3.23 - Pattern Facet Handling Customization
LMX includes a default regular expression library for pattern facet validation (which is built into the binaries).
You can disable testing pattern facets by arranging for the #define LMX_VALIDATE_PATTERNS to be
set to 0. In this case all input strings will be considered to be valid against their schema patterns.
LMX also allows you to specify an alternative pattern-matching library. This is configured by setting the
tc_pattern typedef in lmxuser.h. Simply modify the typedef statement so that your
pattern-matching class is used instead of the default supplied with LMX. Your pattern-matching class must
include as a minimum the following public interfaces:
class c_my_pattern_class
100
3.22.1 - Handling Multiple Independent Schemas
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
{
public:
c_my_pattern_class( const char *ap_pattern );
bool is_match( const std::string &ar_v ) const;
bool is_match( const std::wstring &ar_v ) const;
};
In both cases the function is_match() should return true if the string in ar_v matches the pattern
specified by ap_pattern when the class was constructed, and false otherwise. The pattern specified by
ap_pattern is UTF-8 encoded.
Note that the is_match( const std::wstring &ar_v ) variant is used for facet testing when you
set data using the generated classes accessor methods. If you do not use facet testing in this phase (see
LMX_USER_VALIDATE in 3.16.5 - Debugging Support), then you don't need to implement this member.
Alternatively, if you use accessor side facet testing, but don't want to test pattern facets, you can arrange for
this function to always return true.
A simple approach to implementing the is_match( const std::wstring &ar_v ) method is to
use LMX's convert() functions to convert the wide string to a narrow string and then call is_match(
const std::string &ar_v ).
If you only have simple pattern facets and you don't want a full regular expression engine in your application
(perhaps due to platform constraints), then one approach is to define a class that, based on the string that is
passed to it in its constructor (specifying the desired pattern), sets a pointer to a function that will test for
matches against that pattern. The class will then call the function pointed to by the pointer when a test is
required. The code below demonstrates such an example:
#include <cassert>
#include <cstdlib>
#include <iostream>
// Utility functions called by the match functions
// (Note: C++'s isdigit() and isupper() are not so good on 8-bit chars...)
// N.B. A reference to the pointer is passed in so the pointer is updated
// in the calling function
size_t scan_digits( const char * & arp_string )
{
size_t l_size;
for( l_size = 0; *arp_string >= '0' && *arp_string <= '9'; ++l_size )
++arp_string;
return l_size;
}
size_t scan_upper_letters( const char * & arp_string )
{
size_t l_size;
for( l_size = 0; *arp_string >= 'A' && *arp_string <= 'Z'; ++l_size )
++arp_string;
return l_size;
}
size_t scan_range( const char * & arp_string,
size_t (*pf_test)( const char * & arp_string ),
size_t a_min,
size_t a_max=~0 )
{
size_t l_n_matches = pf_test( arp_string );
3.23 - Pattern Facet Handling Customization
101
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
if( l_n_matches >= a_min && l_n_matches <= a_max )
return true;
return false;
}
// The functions implementing the matches we want
bool is_ip_address( const std::string &ar_v )
{
const char *lp_v = ar_v.c_str();
for( size_t l_i=0; l_i<3; ++ l_i )
{
if( ! scan_range( lp_v, scan_digits, 1, 3 ) ||
*lp_v++ != '.' )
return false;
}
if( ! scan_range( lp_v, scan_digits, 1, 3 ) ||
*lp_v != '\0' )
return false;
return true;
}
bool is_sku( const std::string &ar_v )
{
const char *lp_v = ar_v.c_str();
if( scan_range( lp_v, scan_upper_letters, 3, 3 ) &&
*lp_v++ == '-' &&
scan_range( lp_v, scan_digits, 2, 2 ) &&
*lp_v == '\0' )
return true;
return false;
}
// The actual class doing the matching
class c_my_simple_patterns
{
private:
// Pointer to function that will do the match
bool (*pf_matcher)( const std::string &ar_v );
public:
c_my_simple_patterns( const char *ap_pattern ) : pf_matcher( 0 )
{
// Select the right match function based on the input pattern
if( strcmp( ap_pattern, "([0-9]{1,3}\\.){3}[0-9]{1,3}" ) == 0 )
pf_matcher = &is_ip_address;
else if( strcmp( ap_pattern, "[A-Z]{3}-[0-9]{2}" ) == 0 )
pf_matcher = &is_sku;
else
assert( 0 );
}
bool is_match( const std::string &ar_v ) const
{
// Just call the right match function
if( pf_matcher )
return pf_matcher( ar_v );
return false;
}
bool is_match( const std::wstring &ar_v ) const
{
std::string l_narrow_string;
102
3.23 - Pattern Facet Handling Customization
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
//lmx::convert( &l_narrow_string, ar_v ); // Include lmxuser.h for this
return is_match( l_narrow_string );
}
};
// main() function demonstrating usage
// (Note that LMX usually generates all the code to generate the match class
// instances and invoke the tests.)
int main(int argc, char* argv[])
{
c_my_simple_patterns l_ip_addr( "([0-9]{1,3}\\.){3}[0-9]{1,3}" );
std::cout << "Is IP 10.0.0.1 (true): " <<
l_ip_addr.is_match( "10.0.0.1" ) << "\n";
std::cout << "Is IP 1000.0.0.1 (false): " <<
l_ip_addr.is_match( "1000.0.0.1" ) << "\n";
std::cout << "Is IP 10.0.0.1.0 (false): " <<
l_ip_addr.is_match( "10.0.0.1.0" ) << "\n";
std::cout << "Is IP 10.0.0. (false): " <<
l_ip_addr.is_match( "10.0.0." ) << "\n";
std::cout << "Is IP 10.0.0 (false): " <<
l_ip_addr.is_match( "10.0.0" ) << "\n";
c_my_simple_patterns l_sku( "[A-Z]{3}-[0-9]{2}" );
std::cout << "Is SKU ABC-12 (true): " <<
l_sku.is_match( "ABC-12" ) << "\n";
std::cout << "Is SKU ABCD-12 (false): " <<
l_sku.is_match( "ABCD-12" ) << "\n";
std::cout << "Is SKU ABC-12A (false): " <<
l_sku.is_match( "ABC-12A" ) << "\n";
std::cout << "Is SKU ABC.12 (false): " <<
l_sku.is_match( "ABC.12" ) << "\n";
std::cout << "Is SKU 1AB-12 (false): " <<
l_sku.is_match( "1AB-12" ) << "\n";
return 0;
}
3.24 - Custom Pattern Facet Output Formatting
Occasionally an xs:pattern facet implies XML formatting of a data type that differs from the default
representation. For example, an integer type may require leading zeros.
To accommodate this you can use the -pattern-out flag. The format of this flag is -pattern-out
PATTERN FUNCTION, where PATTERN is the xs:pattern specified in the schema and FUNCTION is
the name of the function you have specified to output the type correctly.
If the data type has multiple xs:patterns specified for it then all the xs:pattern values should be
concatenated together using the | character as a separator into one pattern string to form the flag's PATTERN
argument. This may require quotes to be used on the command-line.
3.25 - XML Output Format Customization
LMX allows you to control the format of the generated XML. Currently you can control the character(s) used
at the end of a 'line', and the character(s) used for indenting. This capability allows you to control whether you
3.24 - Custom Pattern Facet Output Formatting
103
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
want the most compact XML format, or a format that is more attractive and readable when displayed in a
viewer/editor.
The format can be controlled at compile-time, globally at run-time and per-unmarshalling process at run-time.
The default format can be defined at compile-time in the USER: XML_WRITER_FORMAT section of
lmxuser.h by setting the LMX_WRITER_DEFAULT_TAB, LMX_WRITER_DEFAULT_NL,
LMX_WRITER_DEFAULT_ATTRIBUTE_TAB and LMX_WRITER_DEFAULT_ATTRIBUTE_NL #defines
to appropriate values.
LMX_WRITER_DEFAULT_TAB and LMX_WRITER_DEFAULT_NL control the formatting of elements and
LMX_WRITER_DEFAULT_ATTRIBUTE_TAB and LMX_WRITER_DEFAULT_ATTRIBUTE_NL control
the formatting of attributes.
If not explicitly set, the values of LMX_WRITER_DEFAULT_TAB and LMX_WRITER_DEFAULT_NL are
defaulted to "\t" and "\n" respectively.
If not explicitly set, the values of LMX_WRITER_DEFAULT_ATTRIBUTE_TAB and
LMX_WRITER_DEFAULT_ATTRIBUTE_NL are defaulted to the values of LMX_WRITER_DEFAULT_TAB
and LMX_WRITER_DEFAULT_NL respectively.
Setting all values to "" will result in the smallest XML. Setting LMX_WRITER_DEFAULT_TAB to " " and
LMX_WRITER_DEFAULT_NL to "\n" will result in shallower nesting.
The default format can be defined at run-time using the c_xml_writer static methods
c_xml_writer::set_default_tab( const std::string & ar_default_tab ),
c_xml_writer::set_default_nl( const std::string & ar_default_nl ),
c_xml_writer::set_default_attribute_tab( const std::string & ar_default_
attribute_tab ) and c_xml_writer::set_default_attribute_nl( const
std::string & ar_default_ attribute_nl ) methods.
set_default_attribute_tab() and set_default_attribute_nl() control the default
formatting of attributes. For backwards compatibility reasons, if the element and attribute formatting
characters are the same when set_default_tab() and set_default_nl() are called, they will alter
both the element and attribute formatting characters. Conversely, if the element and attribute formatting
characters are different at the time the methods are called then they will only change the respective element
formatting characters. For this reason it may be appropriate to ensure that set_default_tab() and
set_default_nl() are called prior to set_default_attribute_tab() and
set_default_attribute_nl() if independent formatting is required.
If you do not wish to allow these two methods to be effective then you can set the define
LMX_WRITER_LOCK_FORMAT to 1.
The formatting is controlled per-unmarshalling process via the constructor of the c_xml_writer class. This
mechanism overrides the defaults set by the above methods. The prototype for this is:
c_xml_writer( std::ostream
const char *
const char *
const char *
const char *
104
& ar_os,
ap_tab = LMXNULL,
ap_nl = LMXNULL,
ap_attribute_tab = LMXNULL,
ap_attribute_nl = LMXNULL );
3.25 - XML Output Format Customization
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
The format is specified by setting ap_tab, ap_nl ap_attribute_tab and ap_attribute_nl to
point to suitable character strings.
If ap_attribute_tab is absent, then, if present, the value of ap_tab is used to control attribute
formatting. If ap_tab is absent, then the default attribute formatting character sequence is used.
Similarly, if ap_attribute_nl is absent, then, if present, the value of ap_nl is used to control attribute
formatting. If ap_nl is absent, then the default attribute formatting character sequence is used.
As above, for the most compact representation, you may wish to construct the writer using the arguments:
c_xml_writer l_writer( os, "", "" );
If the tab character nests the XML output too deeply, you might wish to choose the following form:
c_xml_writer l_writer( os, "
", "\n" );
Note that not all character combinations used in the constructor will result in legal XML being generated.
To make use of per-unmarshalling process formatting feature you will need to use the advanced forms of
unmarshalling as described in 3.5 - Marshaling (advanced forms).
3.26 - Specifying Micro Formats
A micro format is a simple type that has multiple values encoded into a single string. XML Schema's
xs:dateTime family of types are micro formats, encoding multiple integer types such as the year, date of
month and hour into a single string. You might choose to define a GPS coordinate or complex number (with
real and imaginary parts) as a micro format.
You can tell LMX to treat a simpleType xs:string as a micro format using the -micro-format flag. This has the
form:
-micro-format PATH CLASS
Here, PATH specifies the item(s) in the schema to which the micro format applies. It has the format:
/? (<name>/)* <name>
Note that only type specifications, either as an independently defined simpleType or a simpleType definition
embedded in an element definition, can be assigned a micro format. An element that links to a separately
defined simpleType can not have a micro format specification that is independent of the simpleType to which
it links.
CLASS is the fully C++ namespace-qualified name of the C++ class used to represent the micro format.
An example micro format specification might be:
-micro-format position c_gps_coordinate
By default, it is assumed that the class that represents the micro format has a static method of the form:
3.26 - Specifying Micro Formats
105
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
static bool is_valid( const std::string & ar_v );
This will be used to test whether received XML values conform to the micro format. If such a method does
not exist then you should specify a template specialization for the lmx::s_is_valid< T > template class located
in the MICRO_FORMAT_VALIDATORS section of lmxuser.h to do this; for example:
template<>
struct s_is_valid< c_gps_coordinate >
{
static bool is_valid( const std::string & ar_v )
{ return ...; }
};
You may wish to specify this in a separate .h file in which case you can use the -include-h flag to
specify the file.
It is also assumed by default that that the micro format class has an assignment operator (operator =) that
takes a const std::string reference to convert the XML input into the object, and an output stream
operator (operator << or non-class equivalent) that will generate appropriate XML. If this is not the case
then see 3.13.1.2 - Input/Output Converters.
3.27 - Allowing Unknown Items in an XML Instance
A commonly requested feature is to allow auto-versioning of XML schemas. This allows code generated
using older versions of an XML schema to accept XML instances that contain attributes and elements
specified in a newer version of an XML schema, without generating errors. This enables the XML format to
be extended, or versioned, over time as a product s feature set grows, in a simple and effective way.
LMX can generate code using two modes of versioning:
The first mode is enabled using the -autover flag. This will ignore any unknown attributes. In the case of
an xs:sequence construct it will ignore any unknown child elements appended to the end of a sequence of
elements. However, it will not ignore unknown elements positioned within the known set of elements. In the
case of an xs:choice construct, an unknown element is allowed (and ignored) in place of the elements
specified in the original XML schema.
The second mode is enabled using the -autover2 flag (and is accessible in the GUI interfaces under the
Advanced -> Non-Standard Features... menu option). This mode enhances the first mode by
removing the restriction that the unknown elements must be located at the end of an xs:sequence
construct, so that any unknown child element within an xs:sequence construct will be ignored. Its
behavior when handling attributes and xs:choice constructs is the same as for the first mode.
To enable the code to be selective about which attributes and elements it ignores, the c_xml_reader class
has the following two virtual functions, which are called when an unknown attribute or element is found when
one of the auto-versioning modes is enabled:
virtual elmx_error user_found_unknown_attribute(
const std::string & r_attribute_name_in,
const c_string_stack & r_element_name_stack_in,
const char * p_prog_file_in,
int prog_line_in )
106
3.27 - Allowing Unknown Items in an XML Instance
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
and:
virtual elmx_error user_found_unknown_element(
const std::string & r_element_name_in,
const c_string_stack & r_element_name_stack_in,
const char * p_prog_file_in,
int prog_line_in )
Where:
r_attribute_name_in
The name of the unknown attribute, including namespace prefix.
r_element_name_in
The name of the unknown element, including namespace prefix.
r_element_name_stack_in
Reference to a class that contains the name of the element in which the unknown attribute or element
is located. To convert this into a string like "/foo/bar", do:
r_element_name_stack_in.to_string().
p_prog_file_in
The name of the C++ program file associated with the unknown attribute or element.
prog_line_in
The line number within the C++ program file associated with the unknown attribute or element.
Parsing will continue if ELMX_OK is returned, and terminate (yielding the specified error code) if any other
error code is returned.
3.28 - mustUnderstand / Comprehension Required
Some specifications allow extensions to specify that they must be understood, or the XML instance should not
be parsed. There are a number of ways to do this. One way is to include an attribute in the extension
indicating that such comprehension is required. SOAP takes this approach. To accommodate this approach,
LMX provides the has_must_understand function. The prototype for this function is:
elmx_must_understand has_must_understand( c_xml_reader & ar_reader,
const std::string &ar_must_understand_namespace,
const std::string &ar_must_understand_name,
t_must_understand_test_function a_function = bool_must_understand );
The return code for the function is of the type:enum elmx_must_understand { EMU_OK, EMU_ERROR, EMU_MUST_UNDERSTAND };
Where:EMU_OK
indicates that a must understand attribute was not found, and no errors were encountered
3.28 - mustUnderstand / Comprehension Required
107
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
EMU_ERROR
indicates that an error was encountered during parsing
EMU_MUST_UNDERSTAND
indicates that an attribute specifying that understanding is required was found.
The parameters to has_must_understand have the following meaning:ar_reader
is a low level reader object as used in 2.3 - Unmarshaling (simple form)
ar_must_understand_namespace
is the namespace of the attribute specifying must understand
ar_must_understand_name
is the local name of the attribute specifying must understand
a_function
is an optional pointer to a function that indicates whether the attribute value specifies that
comprehension is required or not. If this parameter is not present, then it is assumed that the attribute
value is of type xs:boolean. The prototype for this function is:
elmx_must_understand <must understand function name>( const std::string & ar_value );
It should return EMU_MUST_UNDERSTAND if the extension must be understood and EMU_OK
otherwise.
An example for SOAP might be:if( has_must_understand( reader,
"http://www.w3.org/2002/12/soap-envelope",
"mustUnderstand" ) == EMU_OK )
// Process the block
Note that it is very important to call the has_must_understand() function at the correct place during
XML instance processing, otherwise undesired results will occur. The blocks that contain
mustUnderstand() attributes are expected to occur as part of xs:any elements contained in an instance.
If information about an instance fragment contained in xs:any is understood, then it should be parsed
without calling has_must_understand(), even if it has further xs:any elements that may not be
understood. When a fragment from an xs:any element is found which is not understood, then
has_must_understand() can be called to determine whether it must be understood. If it must be
understood, then the whole XML instance must be discarded, otherwise the fragment can be ignored and
processing can continue.
3.29 - Finding an XML Instance's Namespace
If your code must interpret XML instances from multiple schemas, you may wish to find the namespace of the
root element of the XML instance in order to determine which parser needs to be used to unmarshal the
instance. LMX provides the get_instance_namespace function for this purpose. The function has the
following prototype:108
3.29 - Finding an XML Instance's Namespace
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
bool get_instance_namespace( c_xml_reader & ar_reader, std::string *ap_namespace_out );
Where:
ar_reader
is a low level reader object created as in the unmarshaling examples above (see 2.3 - Unmarshaling
(simple form))
ap_namespace_out
is a pointer to a string where the namespace is to be placed if it is found.
The function returns true if the namespace is found, and false otherwise.
For example:std::string instance_namespace;
if( lmx::get_instance_namespace( reader, &instance_namespace ) )
{
std::cout << "The instance namespace is: " << instance_namespace << "\n";
}
3.30 - Setting a Decimal Value With a Float
C++ provides no native support for the XML schema xs:decimal type. A convenient way to interface with
such values is via C++ float values (although note that rounding errors may cause undesirable results in some
situations). The LMX c_decimal class provides methods to read decimal values as floats. However, using a
float to set a c_decimal type may require knowledge of the fraction digits facet which an instance of
c_decimal will not have. To allow setting a c_decimal value with a fraction digits facet via a float LMX
generates an extra function in this case of the form:set_NAME( double in );
As such a method may not be required in all builds, the code for this is generated in a conditionally compiled
block. The code is included in the compilation if the #define:
LMX_WANT_DECIMAL_FLOAT_SET
is set to 1 in the USER: CTRL_DEFS section of lmxuser.h (the default case) and excluded otherwise.
3.31 - Test Framework Generation
To help with evaluation and testing, LMX can generate a separate .cpp file containing code to exercise the
generated marshaling and unmarshaling code and a main function. The name of this file is <output
root>-main.cpp. The generated test code unmarshals to objects a specified XML instance from a file,
assigns the object to a new object (using both assignment and construction) and then marshals the result back
out to a named file. If run under Windows, the test code will also check for memory leaks that occur during
the test process - there should be none though. You can then compare the two files to check that the round
tripping has been successful. For XML instances that do not use multiple namespaces or entity substitutions,
the xmlcomp.exe program supplied in the evaluation support suite can be used to automate the comparison.
3.30 - Setting a Decimal Value With a Float
109
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
3.32 - Strategies for Increasing Code Flexibility
One aspect of using LMX generated code is that the code you write can end up being very sensitive to
changes in the re-generated code after making small changes to a schema. This section describes some ways
to write code that is more flexible.
One option is to generate code using the -no-nested-classes flag. If a schema has nested complex
types, this effectively flattens it out. So rather than ending up with classes like:
class c_foo {
class c_bar {
class c_baz {};
}; };
you end up with classes like:
class c_baz {};
class c_bar {};
class c_foo {};
This avoids naming types such as c_foo::c_bar::c_baz.
Avoiding long chains of accessors also helps. Instead, first taking a pointer to a type at the end of a long chain
of accessors makes the code easier to maintain. So instead of:
item.get_foo().get_bar().get_baz( 0 );
item.get_foo().get_bar().get_baz( 1 );
do:
c_bar * p_bar = &item.get_foo().get_bar();
p_bar->get_baz( 0 );
p_bar->get_baz( 1 );
Even better is to put the 'baz' code in a separate function, e.g.:
handle_baz( item.get_foo().get_bar() );
Making handle_baz() a template can make the code even more generic. For example:
template< class T >
void handle_baz( T * p_t )
{
p_t->get_baz( 0 );
p_t->get_baz( 1 );
}
Then it doesn't matter what type handle_baz() is called with.
3.33 - LMX Extensions
LMX supports some additional built-in data types. These types are defined in the namespace
110
3.32 - Strategies for Increasing Code Flexibility
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
http://codalogic.com/xsdtypes. Note that use of these additional built-in types is suitable for
internal proprietary schema, but is not recommended for schema that are made public.
The type asciiString defines a string that consists only of ASCII characters, and by default is mapped to
std::string. The existence of this type allows easy co-existence of Unicode and ASCII values using a more
natural representation in C++ code. An example of how to define the type is:
<xs:attribute xmlns:clt="http://codalogic.com/xsdtypes"
name="OwnerName" type="clt:asciiString"/>
3.33 - LMX Extensions
111
Codalogic LMX - W3C XML Schema to C++ Binding Code Generator
112
3.33 - LMX Extensions
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

advertising