kh-fataoui / xdocreport

Automatically exported from code.google.com/p/xdocreport
0 stars 0 forks source link

Can not find DOCX converters in OSGi container. #159

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
Hi,

I have a felix osgi container that has the following relevant bundles active 
and fragments resolved.

-> find itext
START LEVEL 60
   ID   State         Level  Name
[  15] [Active     ] [   30] Lowagie iText (2.0.8)
-> find xdocreport
START LEVEL 60
   ID   State         Level  Name
[  52] [Active     ] [   30] fr.opensagres.xdocreport.converter (0.9.8)
[  53] [Active     ] [   30] fr.opensagres.xdocreport.core (0.9.8)
[  54] [Active     ] [   30] fr.opensagres.xdocreport.document (0.9.8)
[  55] [Resolved   ] [   30] fr.opensagres.xdocreport.document.docx (0.9.8)
[  56] [Active     ] [   30] fr.opensagres.xdocreport.template (0.9.8)
[  57] [Resolved   ] [   30] fr.opensagres.xdocreport.template.freemarker 
(0.9.8)

I have some code that in another active bundle that calls

  Options options = Options.getFrom(DocumentKind.DOCX).getTo(ConverterTypeTo.PDF).via(ConverterTypeVia.ITEXT);
  report.convert(context, options, fos);

To convert a docx file to pds via itext.

However when report.convert(..) is called there is an exception.

2012-10-02 17:30:28,366 159360 [Thread-14] ERROR 
fr.opensagres.xdocreport.core.registry.AbstractRegistry  - Cannot find 
converters from=DOCX
2012-10-02 17:30:28,381 159375 [Thread-14] ERROR 
com.experian.eda.component.workflow.test.runtime.builder.WorkflowTestRuntimeBuil
der  - Document Template component Accept:encountered content which could not 
be converted correctly
fr.opensagres.xdocreport.converter.XDocConverterException: Cannot find 
converters from=DOCX
    at fr.opensagres.xdocreport.converter.ConverterRegistry.internalFindConverter(ConverterRegistry.java:121)
    at fr.opensagres.xdocreport.converter.ConverterRegistry.findConverter(ConverterRegistry.java:74)
    at fr.opensagres.xdocreport.document.AbstractXDocReport.getConverter(AbstractXDocReport.java:645)
    at fr.opensagres.xdocreport.document.AbstractXDocReport.convert(AbstractXDocReport.java:662)

Which implies that no DOCX converters have been found.

However the the relevent xdocreport bundles are active and so is iText.

Looking at the code it is trying to instantiate the xdoc converter which 
depends on the itext bundle using ServiceLoader

http://code.google.com/p/xdocreport/source/browse/core/fr.opensagres.xdocreport.
core/src/main/java/fr/opensagres/xdocreport/core/registry/AbstractRegistry.java?
name=xdocreport-0.9.8

Now following comment in the code indicates that it should work in an OSGi 
container.
// getClass().getClassLoader() to work under OSGi context
Iterator<Discovery> discoveries = ServiceRegistry.lookupProviders( 
registryType, getClass().getClassLoader() );

This page 
http://code.google.com/p/xdocreport/wiki/DocxReportingJavaMainConverter 
indicates that we should be using itext 2.1.7 but the most recent osgi itext 
far that i can find in itext 2.0.8. If I were to use 2.1.7 i would have to wrap 
itext myself.

There are no error messages that indicate that itext 2.0.8 shouldn't work.

Could you tell me if this is a bug, or I have I missed something and just need 
to do some more configuration to get it to work in osgi container.

Regards
Phil Wilkinson.

Original issue reported on code.google.com by wilkinso...@gmail.com on 3 Oct 2012 at 11:52

GoogleCodeExporter commented 9 years ago
Hi Phil,

At first I know that our documentation is very bad about our converter. As soon 
as we will release 1.0.0, I will update it.
https://oss.sonatype.org/content/repositories/snapshots/fr/opensagres/xdocreport
/) 
I suggest you to use 1.0.0-SNAPSHOT (use maven for that or download at hand on 
the for docx converter because teh converter is very very lot improved (I'm 
improving again). The docx converter 0.9.8 is very bad.

Lik eexplained at 
http://code.google.com/p/xdocreport/wiki/DocxReportingJavaMainConverter you 
need iText and our iText extension, POI, our POI converter and frament 
fr.opensagres.xdocreport.converter.docx.xwpf which register the POI converter 
in the ServiceRegistry.

Pay attention, if you wish to use 1.0.0-SNAPSHOT you need : 

1) Docx Converter JARs

fr.opensagres.xdocreport.converter.docx.xwpf => 
https://oss.sonatype.org/content/repositories/snapshots/fr/opensagres/xdocreport
/fr.opensagres.xdocreport.converter.docx.xwpf/1.0.0-SNAPSHOT/

2) Docx Converter JARs

org.apache.poi.xwpf.converters.core => 
https://oss.sonatype.org/content/repositories/snapshots/fr/opensagres/xdocreport
/org.apache.poi.xwpf.converters.core/1.0.0-SNAPSHOT/

org.apache.poi.xwpf.converters.pdf=> 
https://oss.sonatype.org/content/repositories/snapshots/fr/opensagres/xdocreport
/org.apache.poi.xwpf.converters.pdf/1.0.0-SNAPSHOT/

fr.opensagres.xdocreport.itext.extension => 
https://oss.sonatype.org/content/repositories/snapshots/fr/opensagres/xdocreport
/fr.opensagres.xdocreport.itext.extension/1.0.0-SNAPSHOT/

4) POI JARs

I know it's complicated for the moment, but once release will be done it will 
have zip with docx converter with the whole JARs.

But if you use maven it's will be easy to use 1.0.0-SNAPSHOT

Regards Angelo

Original comment by angelo.z...@gmail.com on 3 Oct 2012 at 12:32

GoogleCodeExporter commented 9 years ago
Thanks for you're reply Angelo.

Can I just ask you to be specific, will 0.9.8 work in an OSGi felix container 
converting docx to pdf via itext, and what are the runtime 
bundles\fragments\jars it needs to work.

(are they....

Lowagie iText (2.0.8)
fr.opensagres.xdocreport.converter (0.9.8)
fr.opensagres.xdocreport.core (0.9.8)
fr.opensagres.xdocreport.document (0.9.8)
fr.opensagres.xdocreport.document.docx (0.9.8)
fr.opensagres.xdocreport.template (0.9.8)
fr.opensagres.xdocreport.template.freemarker (0.9.8)
fr.opensagres.xdocreport.converter.docx.xwpf (0.9.8)
fr.opensagres.xdocreport.itext.extension (0.9.8)
org.apache.poi.xwpf.converters.core (0.9.8)
org.apache.poi.xwpf.converters.pdf (0.9.8)

.. are there more runtime dependencies?)

if 0.9.8 doesn't work then will 1.0.0?
and if we need to work with 1.0.0, when will it be released?

The itext bundle, can I use 2.0.8 from springsource, or do I have to wrap it 
myself, or possibly put it in the osgi containers lib directory so it goes on 
the osgi container class path (the nuclear option :( ).

Really sorry for all the questions, but we've spent upwards of 30 hours trying 
to get get this working, and my boss is telling me to look at at possible libre 
office solution. but i feel we are really close to getting it xdocreport 
working.

Regards
Phil W.

Original comment by wilkinso...@gmail.com on 3 Oct 2012 at 2:26

GoogleCodeExporter commented 9 years ago
>Can I just ask you to be specific, will 0.9.8 work in an OSGi felix container 
>converting docx to pdf via itext, and what are the runtime 
bundles\fragments\jars it >needs to work.

It should work on felix because we have JUnit Test which use OSGi container 
(felix perhaps? It's Pascal who has developped that).

In 0.9.8, it's org.apache.poi.xwpf.converters (0.9.8) and not 

org.apache.poi.xwpf.converters.core (1.0.0)
org.apache.poi.xwpf.converters.pdf (1.0.0)

>if 0.9.8 doesn't work then will 1.0.0?
it should work.

>and if we need to work with 1.0.0, when will it be released?
I don't know when we do that (we develop XDocReport on our spare time and I 
would like improve more docx->pdf converter).

As I said you, if you maven it will be easy for you (it will download the whole 
dependencies)

> The itext bundle, can I use 2.0.8 from springsource
I don't know.

Pascal do you know that? 

I suggest you to download this sample zip which contains the whole JARs for 
docx reporting with velocity (just remove velocity jars because it seems that 
you use Freemarker) and convert PDF/XHTML =>

http://code.google.com/p/xdocreport/downloads/detail?name=docxandvelocity.conver
ters-0.9.8-sample.zip&can=2&q=#makechanges

Regards Angelo

Original comment by angelo.z...@gmail.com on 3 Oct 2012 at 2:37

GoogleCodeExporter commented 9 years ago
Hi,

The good news is that I’ve got the xdocreport working creating a pdf with 
itext.
The bad news is that I had to modify your docx.xwpf-0.9.8 bundle to get it 
working.

To get it working I took your list of jars from the zip file 
(docxandvelocity.converters-0.9.8-sample.zip) excluding the velocity and adding 
freemaker. [thanks for providing that by the way]

I used the following bundles in the OSGi container (in brackets if not already 
an osgi bundle)

commons-codec-1.5.jar (com.springsource.org.apache.commons.codec-1.4.0.jar)
commons-collections-3.2.1.jar 
(com.springsource.org.apache.commons.collections-3.2.1.jar)
commons-lang-2.4.jar (com.springsource.org.apache.commons.lang-2.6.0.jar)
dom4j-1.6.1.jar (com.springsource.org.dom4j-1.6.1.jar)
fr.opensagres.xdocreport.converter-0.9.8.jar
fr.opensagres.xdocreport.converter.docx.xwpf-0.9.8.jar 
(modified-fr.opensagres.xdocreport.converter.docx.xwpf-0.9.8- modified 
MANIFEST.MF to import-package fr.opensagres.xdocreport.core.utils)
fr.opensagres.xdocreport.core-0.9.8.jar
fr.opensagres.xdocreport.document-0.9.8.jar
fr.opensagres.xdocreport.document.docx-0.9.8.jar
fr.opensagres.xdocreport.itext.extension-0.9.8.jar
fr.opensagres.xdocreport.template-0.9.8.jar
fr.opensagres.xdocreport.template.freemarker-0.9.8.jar
itext-2.1.7.jar (wrapper-itext-2.1.7.jar - created custom shade wrapper)
ooxml-schemas-1.1.jar (poi-osgi-wrapper-3.8.jar - created custom shade wrapper)
org.apache.poi.xwpf.converter-0.9.8.jar
oro-2.0.8.jar (com.springsource.org.apache.oro-2.0.8.jar)
poi-3.8.jar (poi-osgi-wrapper-3.8.jar - created custom shade wrapper)
poi-ooxml-3.8.jar (poi-osgi-wrapper-3.8.jar - created custom shade wrapper)
stax-api-1.0.1.jar (com.springsource.javax.xml.stream-1.0.1.jar)
xml-apis-1.0.b2.jar (com.springsource.org.apache.xmlcommons-1.3.4.jar)
xmlbeans-2.3.0.jar (modified-com.springsource.org.apache.xmlbeans-2.4.0.jar 
modified MANIFEST.MF to import-package org.w3c.dom)

Note that I had to modify 
fr.opensagres.xdocreport.converter.docx.xwpf-0.9.8.jar to add an import-package 
for fr.opensagres.xdocreport.core.utils, in 
fr.opensagres.xdocreport.converter.docx.xwpf  as it needed StringUtils.

Caused by: java.lang.NoClassDefFoundError: 
fr/opensagres/xdocreport/core/utils/StringUtils
    at fr.opensagres.xdocreport.converter.docx.poi.itext.XWPF2PDFViaITextConverter.toPDFViaITextOptions(XWPF2PDFViaITextConverter.java:93)
    at fr.opensagres.xdocreport.converter.docx.poi.itext.XWPF2PDFViaITextConverter.convert(XWPF2PDFViaITextConverter.java:66)
    at fr.opensagres.xdocreport.document.AbstractXDocReport.convert(AbstractXDocReport.java:685)
    ... 34 more
Caused by: java.lang.ClassNotFoundException: 
fr.opensagres.xdocreport.core.utils.StringUtils
    at org.apache.felix.framework.ModuleImpl.findClassOrResourceByDelegation(ModuleImpl.java:814)

If you could comment on if you think this is valid and whether you could modify 
this yourselves and when then that would be great, else I’ll just have to 
create a patched version and deploy to our maven repository.

I also had to modify com.springsource.org.apache.xmlbeans-2.4.0.jar modified 
MANIFEST.MF to import-package org.w3c.dom to prevent..
Caused by: java.lang.NoClassDefFoundError
    at org.apache.xmlbeans.XmlBeans.class$(XmlBeans.java:43)
    at org.apache.xmlbeans.XmlBeans.buildNodeMethod(XmlBeans.java:195)
    at org.apache.xmlbeans.XmlBeans.buildNodeToCursorMethod(XmlBeans.java:232)
    at org.apache.xmlbeans.XmlBeans.<clinit>(XmlBeans.java:131)
    at org.openxmlformats.schemas.wordprocessingml.x2006.main.DocumentDocument$Factory.parse(Unknown Source)
    at org.apache.poi.xwpf.usermodel.XWPFDocument.onDocumentRead(XWPFDocument.java:134)
    at org.apache.poi.POIXMLDocument.load(POIXMLDocument.java:159)
    at org.apache.poi.xwpf.usermodel.XWPFDocument.<init>(XWPFDocument.java:123)
    at fr.opensagres.xdocreport.converter.docx.poi.itext.XWPF2PDFViaITextConverter.convert(XWPF2PDFViaITextConverter.java:65)
    at fr.opensagres.xdocreport.document.AbstractXDocReport.convert(AbstractXDocReport.java:685)
    ... 34 more
Caused by: java.lang.ClassNotFoundException: org.w3c.dom.Node
    at org.apache.felix.framework.ModuleImpl.findClassOrResourceByDelegation(ModuleImpl.java:814)

As this is just a wrapper, I will create my own wrapper, re-wrapping the 
original xmlbeans library with the required org.w3c.dom import.

I had to wrap all the apache.poi libraries as I couldn't locate osgi bundles 
for these. (poi-3.8.jar, poi-ooxml-3.8.jar and ooxml-schemas-1.1.jar all 
wrapped in my wrapper ooxml-schemas-1.1.jar).
(attached poi.pom.xml).

I had to wrap the itext libraries to get the 2.1.7 version of itext, as I could 
only locate a springsource version at 2.0.8.
(attached itext.pom.xml)

I had to ensure that org.w3c.dom was being exported from the java runtime via 
the felix container by adding org.w3c.dom to the 
org.osgi.framework.system.packages property in felix.properties.

Finally I had to increase the permgen heap size from its default (64m) to 128m 
by adding -XX:MaxPermSize=128m to the batch file we use to lanch the osgi 
container, to prevent..
fr.opensagres.xdocreport.converter.XDocConverterException: 
java.lang.OutOfMemoryError: PermGen space
    at fr.opensagres.xdocreport.document.AbstractXDocReport.convert(AbstractXDocReport.java:720)
    at com.experian.eda.component.workflow.runtime.engine.handler.DocumentTemplateActivityBehaviour.mergeDoc(DocumentTemplateActivityBehaviour.java:286)
    at com.experian.eda.component.workflow.runtime.engine.handler.DocumentTemplateActivityBehaviour.executeInternal(DocumentTemplateActivityBehaviour.java:187)
    at com.experian.eda.component.workflow.runtime.engine.handler.AbstractActivityBehaviour.execute(AbstractActivityBehaviour.java:57)

Regards,
Philip Wilkinson.

Original comment by wilkinso...@gmail.com on 10 Oct 2012 at 9:09

Attachments:

GoogleCodeExporter commented 9 years ago
Hi Phil,

Wow congratulation for your work. I don't understand why you had had so many 
problems because we have a JUnit which works with OSGi?

@Pascal :I think we should provide several zip sample like we have provided 
(reporting+freemarker, reporting+velocity, reporting+converter+freemarker for 
docx and odt, etc) with OSGi. Those zip will contain a basic sample with 
Activator which generate report and contains the whole JARs with well 
MANIFEST.MF.

What do you think about that?
If you are OK, could you start this task (if you have time). I could help you 
if you wish.

Regards Angelo

Original comment by angelo.z...@gmail.com on 10 Oct 2012 at 10:02

GoogleCodeExporter commented 9 years ago
[deleted comment]
GoogleCodeExporter commented 9 years ago
Hi,
your junit tests that I think you are refereing to  are here.
http://code.google.com/searchframe#cLr_AEi-h-g/integrationtests/fr.opensagres.xd
ocreport.osgi.integrationtests/src/test/java/fr/opensagres/xdocreport/osgi/integ
rationtests/ConverterTest.java
You use pax exam to wrap all the 3rd party jars.
pax exam uses bnd but from what I can see specifies no parameters so everything 
is imported and everything is exported.
I've had problems in past with this where I've had conflicts between different 
bundles exporting the same packages, which should be internal. So I use a maven 
shade plugin and are more explicit about what gets exported and what gets 
imported.
I think your tests work because, they are not complete, i.e. they don't use all 
the jars I needed, they don't produce a pdf, and they run in a dedicated 
container whose only purpose is to run the tests (so no other possibility of 
conflict).
Also I wanted to use springsource provided wrappers where ever possible (not 
on-the fly bnd created 3rd party bundles) as it makes the whole configuration 
simpler.
Regards,
Philip Wilkinson.

Original comment by wilkinso...@gmail.com on 10 Oct 2012 at 11:57

GoogleCodeExporter commented 9 years ago
Hi Philip,

> your junit tests that I think you are refereing to  are here.
http://code.google.com/searchframe#cLr_AEi-h-g/integrationtests/fr.opensagres.xd
ocreport.osgi.integrationtests/src/test/java/fr/opensagres/xdocreport/osgi/integ
rationtests/ConverterTest.java

yes it that.

Many thank's for your great information. 
I tell me if you will be OK to improve our OSGi JUnit and avoid having your big 
problems that you have done. Are you interested to help us?

Many thanks for your help.

Regards Angelo

Original comment by angelo.z...@gmail.com on 11 Oct 2012 at 10:17

GoogleCodeExporter commented 9 years ago
Has this been installed anywhere on a Maven Repo, in its working form?

Original comment by sethhels...@googlemail.com on 15 Jul 2013 at 10:04

GoogleCodeExporter commented 9 years ago
Is it again this problem with 1.0.5-SNAPSHOT?

Original comment by angelo.z...@gmail.com on 7 Aug 2014 at 9:05