Although MPXJ can be downloaded as a complete package from Maven, GitHub and SourceForge, the development of the library continues between releases and is driven by user requests for new functionality and bug fixes being applied to existing features. Many MPXJ users will work with and ship software based on intermediate versions of MPXJ built from the code on GitHub in order to take advantage of these enhancements before they become available in an official release. This approach is supported by the fact that code is only pushed to the master branch on GitHub if the suite of regression tests have been completed successfully: therefore the quality of the code taken from GitHub at any point can normally be guaranteed to be as good as that in an official release.
In order to take advantage of MPXJ functionality from GitHub, you will need to understand how to build the library, whether you are using it in the form of a Java JAR or a .NET DLL. The following sections explain how to do this.
Obtaining the source
The first step in the process of building your own version of MPXJ is to obtain the latest source from GitHub. Instructions for cloning the repository can be found on this page.
Building the Java JAR
MPXJ is built using Maven. Once you have a cloned copy of the MPXJ repository,
you may wish to update the
version attributes in
pom.xml. This will ensure that there is no confusion between the version of
MPXJ you build and the official distributions.
If you have a copy of Maven installed, you can issue the following command to build MPXJ:
mvn -DskipTests=true -Dmaven.javadoc.skip=true -Dsource.skip=true package
This will generate the
mpxj.jar for you in the Maven target directory, and
copies MPXJ's dependencies to the
lib directory. Note that for convenience
this skips running the unit tests, javadoc generation and source packaging.
If you are using Maven to manage dependencies for your own project, you can install your newly built version of MPXJ in a local Maven repository:
mvn -DskipTests=true -Dmaven.javadoc.skip=true -Dsource.skip=true install
Building the .NET DLLs
Building the .NET versions of MPXJ uses an Ant script to first run Maven to create the Java version, then run IKVM to create a .Net Framework and a .Net Core version.
- Download .Net Framework version of IKVM and unzip the files into a convenient directory
- Change directory to the MPXJ folder.
- Edit the
build.xmlfile and ensure that the property named
ikvm.net45.diris set to point to the location where you have unzipped IKVM.
You can now issue the following command:
The Ant script will recognise that IKVM is present and build the .NET Framework
version of MPXJ, with the results found in the
Generating the JAXB code
In order to read and write various XML file formats, MPXJ relies on code
generated by the JAXB tool
xjc from the XML schema for each file format.
Normally you will not need to regenerate this source, but if you are changing
the JAXB implementation, or modifying the use of JAXB in some way, then you may
need to regenerate this code.
Where I have created an XML schema to support a particular file format, I have
included it in the MPXJ distribution in the
jaxb directory. For XML schemas
published by product vendors, I have included a note on the home page indicating
where these can be located.
If you obtain a copy of the XML schema file you want to work with, you can
update the JAXB source using the
xjc target found in the ant
file. Note that the
xjc target is platform specific, you will need to change
the name of
xjc tool to be
xjc.sh depending on
your operating system. You will also need to set the properties indicated in
build.xml to tell it where to find
xjc and the XML schema file. If you are
only regenerating source for one of the XML schemas, you can comment out the
others in the Ant script to avoid unnecessary work.