Ugh -- sounds like you've entered dependency hell.
Though, I think the one shred of good news here is that it seems to only
have a dependency conflict in one place in our codebase.
It looks like (at a glance) if our WordFilter can be re-written to no
longer need the org.textmining project, you *might* be OK (i.e.
hopefully it wouldn't snowball on you). But, that would require finding
a Word document text extractor that is as good as (or better than) that
'org.textmining' one, and then hoping it doesn't cause another
dependency conflict. Not sure of any alternative Word text extractors,
off the top of my head, but maybe others know of one?
- Tim
On 10/6/2010 5:51 AM, Keith Gilbertson wrote:
> Thanks, Tim. That helped me to understand. I put the version numbers of the
> dependency in the parent pom.xml ('dspace-src/pom.xml') and left the version
> numbers out of 'dspace-src/dspace-api/pom.xml'.
>
> So then I found another thing I didn't look at closely enough. The
> WordFilter doesn't use poi directly, but the org.textmining project that it
> uses depends on that old version of POI. To confuse things more, the old
> versions of poi had groupId 'poi', and the new versions have groupId
> 'org.apache.poi'.
> I can convince Maven to forget about the old version of the POIi library by
> making this exclusion change in the parent pom:
> <dependency>
> <groupId>org.textmining</groupId>
> <artifactId>tm-extractors</artifactId>
> <version>0.4</version>
> <exclusions>
> <exclusion>
> <groupId>poi</groupId>
> <artifactId>poi</artifactId>
> </exclusion>
> </exclusions>
> </dependency>
> <dependency>
>
> Then only the new version, org.apache.poi/poi/3.6 is included in the project.
> Unfortunately, the org.textmining extractors really do need that version of
> POI. The PowerPointFilter works, but I've broken the WordFilter:
>
> Exception:
> org.apache.poi.poifs.filesystem.POIFSFileSystem.getRoot()Lorg/apache/poi/poifs/filesystem/DirectoryEntry;
> java.lang.NoSuchMethodError:
> org.apache.poi.poifs.filesystem.POIFSFileSystem.getRoot()Lorg/apache/poi/poifs/filesystem/DirectoryEntry;
> at
> org.textmining.text.extraction.WordExtractor.extractText(WordExtractor.java:51)
> at
> org.dspace.app.mediafilter.WordFilter.getDestinationStream(WordFilter.java:95)
>
> I have two programs that share the same classpath, but need different
> versions of the same library.
>
> I could rewrite the WordFilter so that it no longer uses the org.textmining
> package which needs the old library, but I keep thinking that the more I try
> to "fix" stuff, the more I'm likely to break:
>
> http://www.nypost.com/p/news/local/brooklyn/rat_bastards_f5onjzgcqxm0fu3RFz3ySL
>
>
>
> On Oct 5, 2010, at 4:09 PM, Tim Donohue wrote:
>
>> Hi Keith,
>>
>> Simply put, it's because you were accidentally looking in the wrong pom.xml
>> :) There's many of them sprinkled through the DSpace codebase, and they all
>> inherit many of their settings from one main pom.xml.
>>
>> So, you noticed that the 'dspace-api/pom.xml' file included a dependency for
>> "poi". But, if you look closely, that dependency doesn't list a<version>.
>> This is because, for DSpace, we manage all the versions of dependencies in
>> one parent pom.xml (which is loaded via the<parent> tag within the
>> dspace-aip/pom.xml).
>>
>> Now, take a look at the [dspace-src]/pom.xml. This is the main Parent
>> pom.xml for dspace (with an artifactid of 'dspace-parent')
>>
>> http://scm.dspace.org/svn/repo/dspace/trunk/pom.xml
>>
>> This is the pom.xml which actually lists the versions of every dependency
>> used by the various APIs of DSpace. If you search in this pom.xml, you'll
>> find this entry:
>>
>> <dependency>
>> <groupId>poi</groupId>
>> <artifactId>poi</artifactId>
>> <version>2.5.1-final-20040804</version>
>> </dependency>
>>
>> That's where the 2.5.1 version is sneaking in. If you make your necessary
>> changes to this pom.xml, everything should act as you expect it to. So, just
>> undo your changes in 'dspace-src/dspace-api/pom.xml', and instead make those
>> changes to 'dspace-src/pom.xml'
>>
>> I hope that helps!
>>
>> - Tim
>>
>> On 10/5/2010 2:36 PM, Keith Gilbertson wrote:
>>> Hi,
>>>
>>> I've been experimenting with a Media Filter for text extraction from
>>> PowerPoint files. It's based on the Apache POI libraries, as was suggested
>>> by others in a previous thread.
>>>
>>> It uses the poi, poi-scratchpad, and poi-ooxml artifacts, in version 3.6,
>>> the latest release version from Apache. I haven't done much with Maven,
>>> and am not sure how to tell it which libraries I need.
>>>
>>> This bit was already in the dspace-api/pom.xml file:
>>> <dependency>
>>> -<groupId>poi</groupId>
>>> -<artifactId>poi</artifactId>
>>> -</dependency>
>>>
>>>
>>> I removed it, because I wanted the latest version of the libraries. Then,
>>> I added these dependencies to the bottom of the file:
>>>
>>> +<dependency>
>>> +<groupId>org.apache.poi</groupId>
>>> +<artifactId>poi</artifactId>
>>> +<version>3.6</version>
>>> +</dependency>
>>> +<dependency>
>>> +<groupId>org.apache.poi</groupId>
>>> +<artifactId>poi-scratchpad</artifactId>
>>> +<version>3.6</version>
>>> +</dependency>
>>> +<dependency>
>>> +<groupId>org.apache.poi</groupId>
>>> +<artifactId>poi-ooxml</artifactId>
>>> +<version>3.6</version>
>>> +</dependency>
>>>
>>> Somehow Maven magically found the correct versions of the dependencies, and
>>> everything built fine. When I deployed DSpace and looked in the lib
>>> directory, there were two versions of the main poi library there:
>>>
>>> poi-2.5.1-final-20040804.jar
>>> poi-3.6.jar
>>> poi-ooxml-3.6.jar
>>> poi-ooxml-schemas-3.6.jar
>>> poi-scratchpad-3.6.jar
>>>
>>> I couldn't figure out why the poi-2.5.1 version was still there, or find
>>> anything that actually used it. So, in the interest of doing some quick
>>> testing, I just deleted it.
>>>
>>> Can someone give a hand on how to do this properly? I'm trying to tell the
>>> build process to find and use only version 3.6 of poi.
>>>
>>> Thank you!
>>> --keith
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Beautiful is writing same markup. Internet Explorer 9 supports
>>> standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2& L3.
>>> Spend less time writing and rewriting code and more time creating great
>>> experiences on the web. Be a part of the beta today.
>>> http://p.sf.net/sfu/beautyoftheweb
>>> _______________________________________________
>>> DSpace-tech mailing list
>>> DSpace-tech@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/dspace-tech
>
------------------------------------------------------------------------------
Beautiful is writing same markup. Internet Explorer 9 supports
standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3.
Spend less time writing and rewriting code and more time creating great
experiences on the web. Be a part of the beta today.
http://p.sf.net/sfu/beautyoftheweb
_______________________________________________
DSpace-tech mailing list
DSpace-tech@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dspace-tech
