I’ve gotten a little farther. I discovered that the FO generation process produces different results from run inside Oxygen than when run outside Oxygen using the libraries in the fop/lib dir. In particular, this bit of XSLT: <math xmlns="http://www.w3.org/1998/Math/MathML" > results in <math xmlns="http://www.w3.org/1998/Math/MathML"

In the FO instance produced from within Oxygen, but: <math> When run outside Oxygen. That was causing FOP to fail to render the math even when the JEuclid libraries were in place. Correcting the XSLT to use a prefix on the <math> element solved that problem, but the question remains of why the XSLT serialization behavior is different between Oxygen and not-Oxygen (it implies that Oxygen is using a newer or fixed version of some library). I modified my local startCmd.sh as follows: #— Start of FOP stuff # Classpath entries for FOP: NEW_CLASSPATH="$DITA_DIR/plugins/org.dita.pdf2/fop/build/fop.jar:$NEW_CLASS PATH" DIRLIBS=$DITA_DIR/plugins/org.dita.pdf2/fop/lib/*.jar for i in ${DIRLIBS} do # if the directory is empty, then it will return the input string # this is stupid, so case for it if [ "$i" != "${DIRLIBS}" ] ; then NEW_CLASSPATH="$i"${pathSepChar}$NEW_CLASSPATH fi done #— End of FOP stuff And with that class path, I can do this: ant -Dtranstype=pdf -Dargs.input=/Users/ekimber/workspace_40/temp/MathML-Test/MathML-Test.ditam ap "-Dretain.topic.fo=yes" "-Dclean.temp=no" And get an FOP-generated PDF with rendered equations. Cheers, E. -- Eliot Kimber Senior Solutions Architect "Bringing Strategy, Content, and Technology Together" Main: 512.554.9368 www.reallysi.com www.rsuitecms.com On 1/7/14, 5:04 AM, "Oxygen XML Editor Support" <support@oxygenxml.com> wrote:

Hi Eliot,

The jar files that you added in the pdf2/fop/lib dir are not automatically loaded by the ANT process started from command line. I think that is true also for the PDF generation step (the FO -> PDF transformation) performed by FOP, regardless how the transformation was started. The JEuclid jar files have to be added to the classpath and that is what the -lib parameters do in the Oxygen command that runs the DITA transformation in the Oxygen application.

FOP delegates MathML processing to JEuclid (which is a FOP plugin) if it finds it in the classpath. If FOP does not find a FOP plugin for the MathML content you get the error about unrecognized MathML markup. JEuclid can handle inline MathML markup in the FO document exactly as you have it (inside <fo:instream-foreign-object> elements) so you should just add the two JEuclid jar files explicitly to the ANT classpath, for example as -lib parameters to the Java command that starts the ANT process.

Best regards, Sorin

http://www.oxygenxml.com

Eliot Kimber wrote:

...

Using Oxygen 15.1 with the 1.6.3 OT with FOP 1.1 manually installed into the pdf2 plugin, and given DITA docs with MathML and a plugin that outputs the MathML as markup within fo:instream-foreign-object elements, if I run the OT PDF transform from within Oxygen, then FOP renders the MathML nicely.

If I run the same transform from the command line from within a startCmd.sh-created environment, e.g.:

bash-3.2$ ant -lib "/Applications/oxygen/lib/xml-apis.jar" -lib "/Applications/oxygen/lib/xercesImpl.jar" -lib "/Applications/oxygen/frameworks/dita/DITA-OT" -lib "/Applications/oxygen/lib/commons-io-1.3.1.jar" -lib "/Applications/oxygen/lib/commons-logging-1.1.1.jar" -lib "/Applications/oxygen/lib/log4j.jar" -lib "/Applications/oxygen/frameworks/dita/DITA-OT/lib/dost-patches.jar" -lib "/Applications/oxygen/frameworks/dita/DITA-OT/lib" -lib "/Applications/oxygen/frameworks/dita/DITA-OT/lib/dost.jar" -lib "/Applications/oxygen/frameworks/dita/DITA-OT/lib/commons-codec-1.4.jar" -lib "/Applications/oxygen/lib/jing.jar" -lib "/Applications/oxygen/lib/saxon9ee.jar" -lib "/Applications/oxygen/lib/saxon.jar" -lib "/Applications/oxygen/lib/xmlgraphics-commons-1.5.jar" -lib "/Applications/oxygen/lib/fop.jar" -lib "/Applications/oxygen/lib/batik-all-1.7.jar" -lib "/Applications/oxygen/lib/xml-apis-ext.jar" -lib "/Applications/oxygen/lib/avalon-framework-4.2.0.jar" -lib "/Applications/oxygen/lib/jeuclid-core.jar" -lib "/Applications/oxygen/lib/jeuclid-fop.jar" -lib "/Applications/oxygen/lib/jh.jar" -lib "/Applications/oxygen/frameworks/dita/DITA-OT/lib/jsearch.jar" -lib "/Applications/oxygen/lib/lucene-analyzers-common-4.0.0.jar" -lib "/Applications/oxygen/lib/lucene-core-4.0.0.jar" -lib

"/Applications/oxygen/frameworks/dita/DITA-OT/plugins/com.oxygenxml.webhe lp /lib/xhtml-indexer.jar" -f "/Applications/oxygen/frameworks/dita/DITA-OT/build.xml" "-Dtranstype=pdf2" "-Dbasedir=/Users/ekimber/workspace_40/omics-plugin/temp/MathML-Test"

"-Doutput.dir=/Users/ekimber/workspace_40/omics-plugin/temp/MathML-Test/o ut /pdf" "-Dwebhelp.trial.license=no"

"-Ddita.temp.dir=/Users/ekimber/workspace_40/omics-plugin/temp/MathML-Tes t/ temp/pdf" "-Dpdf.formatter=fop" "-Ddita.dir=/Applications/oxygen/frameworks/dita/DITA-OT" "-Dretain.topic.fo=yes" "-Dclean.temp=no" "-Dxep.failOnError=true"

"-Dargs.input=/Users/ekimber/workspace_40/omics-plugin/temp/MathML-Test/M at hML-Test.ditamap" "-DbaseJVMArgLine=-Xmx384m"

The process runs and I don’t get any FOP messages about the math, but the equations are not rendered.

I noticed that Oxygen includes the jeuclid-core and jeuclid-fop jars and I tried copying those into the pdf2/fop/lib dir.

If I run the normal Toolkit process without the Oxygen-defined -lib options (meaning it’s using the class path defined by the startCmd.sh) then FOP complains about the MathML markup as being unrecognized, even with the JEuclid libraries present (which suggests that the Toolkit’s Ant process is not automatically loading the libraries).

Likewise, if I run the fop shell agains the FO file I get the same MathML complaints.

I verified that both the Oxygen-supplied FOP jar and the one in my OT are 1.1, e.g.:

contrext01:fop ekimber$ java -jar /Applications/oxygen/lib/fop.jar -v FOP Version 1.1

contrext01:fop ekimber$ java -jar build/fop.jar -v FOP Version 1.1

So I’m at a bit of a loss as to how Oxygen is getting the result I want and what I need to do to get that same result outside of Oxygen.

What am I missing?

Thanks,

Eliot

————— Eliot Kimber, Owner Contrext, LLC http://contrext.com

_______________________________________________ oXygen-user mailing list oXygen-user@oxygenxml.com http://www.oxygenxml.com/mailman/listinfo/oxygen-user

Hi Eliot, The DITA OT bundled with Oxygen contains our own custom MathML plugin:

OXYGEN_INSTALL_DIR\frameworks\dita\DITA-OT\plugins\mathml

which adds our XSL override of the template:

<xsl:template match="*[contains(@class,' topic/foreign ') or contains(@class,' topic/unknown ')]" > <xsl:apply-templates select="mml:math" xmlns:mml="http://www.w3.org/1998/Math/MathML" mode="copyMathML"/> <xsl:apply-templates select="*[contains(@class,' topic/object ')][@type='DITA-foreign']"/> </xsl:template>

Probably when you were running your plugin with the DITA OT bundled with Oxygen our own MathML processing plugin's XSL had a higher import priority and handled the namespace correctly so your XSLT template was probably not called at all. When you tested in an external DITA OT, your plugin's XSLT template was used instead and it had that bug which did not handle the namespace correctly. And indeed, the first step to debug the issue would have been to check the "topic.fo" contents. Regards, Radu Radu Coravu <oXygen/> XML Editor, Schema Editor and XSLT Editor/Debugger http://www.oxygenxml.com On 1/8/2014 7:03 PM, Eliot Kimber wrote:

OK, final answer. It appears that all my problems were caused by the serialization of <math> without the namespace declaration.

With that corrected, then simply putting the JEuclid jars in the fop/lib dir is sufficient to get the equation rendering, as it should be. I tested both a clean 1.8.M2 install and my previously-tested 1.6.3 install (I also inspected the FOP-related Ant scripts in the PDF2 transform and there’s no difference between them that would affect this).

So the lesson is make sure your FO files are really correct before spending hours trying to track down a non-existent class path configuration problem . . . .

Cheers,

Eliot

--