Menu
▾
▴
[Archive-access-cvs] SF.net SVN: archive-access: [1626] trunk/archive-access/projects/nutchwax
|
From: <sta...@us...> - 2007-03-22 00:07:39
|
Revision: 1626
http://archive-access.svn.sourceforge.net/archive-access/?rev=1626&view=rev
Author: stack-sf
Date: 2007-03-21 17:07:39 -0700 (Wed, 21 Mar 2007)
Log Message:
-----------
Moved third-party checkout under nutchwax-thirdparty module.
Let go of maven1.
* nutchwax-core/pom.xml
* src/java/overview.html
* src/plugin/build-plugin.xml
* nutchwax-thirdparty/pom.xml
* nutchwax-webapp/src/main/assembly/assemble-war.xml
* .classpath
* nutchwax-job/src/main/assembly/assemble-job.xml
* nutchwax-webapp/pom.xml
Reference new third-party location.
* project.properties
* maven.xml
* project.xml
Let go of maven1
Modified Paths:
--------------
trunk/archive-access/projects/nutchwax/.classpath
trunk/archive-access/projects/nutchwax/nutchwax-core/pom.xml
trunk/archive-access/projects/nutchwax/nutchwax-job/src/main/assembly/assemble-job.xml
trunk/archive-access/projects/nutchwax/nutchwax-thirdparty/pom.xml
trunk/archive-access/projects/nutchwax/nutchwax-webapp/pom.xml
trunk/archive-access/projects/nutchwax/nutchwax-webapp/src/main/assembly/assemble-war.xml
trunk/archive-access/projects/nutchwax/src/java/overview.html
trunk/archive-access/projects/nutchwax/src/plugin/build-plugin.xml
Removed Paths:
-------------
trunk/archive-access/projects/nutchwax/maven.xml
trunk/archive-access/projects/nutchwax/project.properties
trunk/archive-access/projects/nutchwax/project.xml
Property Changed:
----------------
trunk/archive-access/projects/nutchwax/
trunk/archive-access/projects/nutchwax/nutchwax-thirdparty/
Property changes on: trunk/archive-access/projects/nutchwax
___________________________________________________________________
Name: svn:externals
- third-party/nutch -r 508238 http://svn.apache.org/repos/asf/lucene/nutch/trunk
Modified: trunk/archive-access/projects/nutchwax/.classpath
===================================================================
--- trunk/archive-access/projects/nutchwax/.classpath 2007-03-21 23:27:40 UTC (rev 1625)
+++ trunk/archive-access/projects/nutchwax/.classpath 2007-03-22 00:07:39 UTC (rev 1626)
@@ -9,234 +9,234 @@
<classpathentry kind="src" path="src/plugin/query-host/src/java"/>
<classpathentry kind="src" path="src/plugin/query-title/src/java"/>
<classpathentry kind="src" path="src/plugin/query-wax/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/index-basic/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/index-more/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/languageidentifier/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/languageidentifier/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/lib-regex-filter/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/lib-regex-filter/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-ext/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-ext/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-html/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-html/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-js/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-pdf/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-pdf/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/parse-text/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/query-basic/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/query-more/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/query-site/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/query-url/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/query-url/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/scoring-opic/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/summary-basic/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/summary-lucene/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-automaton/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-automaton/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-prefix/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-regex/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-regex/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-suffix/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlfilter-suffix/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlnormalizer-basic/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlnormalizer-basic/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlnormalizer-pass/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlnormalizer-pass/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlnormalizer-regex/src/java"/>
- <classpathentry kind="src" path="third-party/nutch/src/plugin/urlnormalizer-regex/src/test"/>
- <classpathentry kind="src" path="third-party/nutch/src/test"/>
- <classpathentry kind="lib" path="third-party/nutch/build/clustering-carrot2/clustering-carrot2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/creativecommons/creativecommons.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/index-basic/index-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/index-more/index-more.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/language-identifier/language-identifier.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-http/lib-http.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-jakarta-poi/poi-3.0-alpha1-20050704.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-jakarta-poi/poi-scratchpad-3.0-alpha1-20050704.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-log4j/log4j-1.2.11.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-lucene-analyzers/lucene-analyzers-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-nekohtml/nekohtml-0.9.4.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-parsems/lib-parsems.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-regex-filter/lib-regex-filter.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-xml/jaxen-core.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-xml/jaxen-jdom.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-xml/jdom.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-xml/saxpath.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/lib-xml/xercesImpl.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/microformats-reltag/microformats-reltag.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/nutch-0.9-dev.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/nutch-extensionpoints/nutch-extensionpoints.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/ontology/ontology.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-ext/parse-ext.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-html/parse-html.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-js/parse-js.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-msexcel/parse-msexcel.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-mspowerpoint/parse-mspowerpoint.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-msword/parse-msword.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-oo/parse-oo.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-pdf/parse-pdf.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-rss/parse-rss.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-swf/parse-swf.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-text/parse-text.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/parse-zip/parse-zip.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/carrot2-filter-lingo.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/carrot2-local-core.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/carrot2-snowball-stemmers.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/carrot2-util-common.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/carrot2-util-tokenizer.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/clustering-carrot2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/commons-collections-3.1-patched.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/commons-pool-1.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/Jama-1.0.1-patched.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/clustering-carrot2/violinstrings-1.0.2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/creativecommons/creativecommons.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/index-basic/index-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/index-more/index-more.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/language-identifier/language-identifier.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-http/lib-http.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-jakarta-poi/poi-3.0-alpha1-20050704.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-jakarta-poi/poi-scratchpad-3.0-alpha1-20050704.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-log4j/log4j-1.2.11.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-lucene-analyzers/lucene-analyzers-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-nekohtml/nekohtml-0.9.4.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-parsems/lib-parsems.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-regex-filter/lib-regex-filter.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-xml/jaxen-core.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-xml/jaxen-jdom.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-xml/jdom.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-xml/saxpath.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/lib-xml/xercesImpl.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/microformats-reltag/microformats-reltag.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/nutch-extensionpoints/nutch-extensionpoints.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/ontology/commons-logging-1.0.3.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/ontology/icu4j_2_6_1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/ontology/jena-2.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/ontology/ontology.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-ext/parse-ext.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-html/parse-html.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-html/tagsoup-1.0rc3.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-js/parse-js.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-msexcel/parse-msexcel.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-mspowerpoint/parse-mspowerpoint.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-msword/parse-msword.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-oo/parse-oo.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-pdf/parse-pdf.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-pdf/PDFBox-0.7.2-log4j.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-rss/commons-feedparser-0.6-fork.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-rss/parse-rss.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-rss/xmlrpc-1.2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-swf/javaswf.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-swf/parse-swf.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-text/parse-text.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/parse-zip/parse-zip.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/protocol-file/protocol-file.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/protocol-ftp/commons-net-1.2.0-dev.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/protocol-ftp/protocol-ftp.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/protocol-http/protocol-http.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/protocol-httpclient/protocol-httpclient.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/query-basic/query-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/query-more/query-more.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/query-site/query-site.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/query-url/query-url.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/scoring-opic/scoring-opic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/subcollection/subcollection.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/summary-basic/summary-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/summary-lucene/lucene-highlighter-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/summary-lucene/summary-lucene.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlfilter-automaton/automaton.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlfilter-automaton/urlfilter-automaton.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlfilter-prefix/urlfilter-prefix.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlfilter-regex/urlfilter-regex.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlfilter-suffix/urlfilter-suffix.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlnormalizer-basic/urlnormalizer-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlnormalizer-pass/urlnormalizer-pass.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/plugins/urlnormalizer-regex/urlnormalizer-regex.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/protocol-file/protocol-file.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/protocol-ftp/protocol-ftp.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/protocol-http/protocol-http.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/protocol-httpclient/protocol-httpclient.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/query-basic/query-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/query-more/query-more.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/query-site/query-site.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/query-url/query-url.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/scoring-opic/scoring-opic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/subcollection/subcollection.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/summary-basic/summary-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/summary-lucene/summary-lucene.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlfilter-automaton/urlfilter-automaton.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlfilter-prefix/urlfilter-prefix.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlfilter-regex/urlfilter-regex.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlfilter-suffix/urlfilter-suffix.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlnormalizer-basic/urlnormalizer-basic.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlnormalizer-pass/urlnormalizer-pass.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/build/urlnormalizer-regex/urlnormalizer-regex.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/lib/commons-beanutils.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/lib/commons-collections-3.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/lib/commons-digester.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/lib/jstl.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/lib/standard.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/lib/struts.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/contrib/web2/plugins/web-caching-oscache/lib/oscache-2.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/commons-cli-2.0-SNAPSHOT.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/commons-codec-1.3.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/commons-httpclient-3.0.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/commons-lang-2.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/commons-logging-1.0.4.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/commons-logging-api-1.0.4.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/hadoop-0.10.1-core.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jakarta-oro-2.0.7.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jets3t.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jetty-5.1.4.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jetty-ext/ant.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jetty-ext/commons-el.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jetty-ext/jasper-compiler.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jetty-ext/jasper-runtime.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/jetty-ext/jsp-api.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/junit-3.8.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/log4j-1.2.13.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/lucene-core-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/lucene-misc-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/pmd-ext/jakarta-oro-2.0.8.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/pmd-ext/jaxen-1.1-beta-7.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/pmd-ext/pmd-3.6.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/servlet-api.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/taglibs-i18n.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/xerces-2_6_2-apis.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/lib/xerces-2_6_2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/carrot2-filter-lingo.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/carrot2-local-core.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/carrot2-snowball-stemmers.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/carrot2-util-common.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/carrot2-util-tokenizer.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/commons-collections-3.1-patched.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/commons-pool-1.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/Jama-1.0.1-patched.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/clustering-carrot2/lib/violinstrings-1.0.2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-jakarta-poi/lib/poi-3.0-alpha1-20050704.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-jakarta-poi/lib/poi-scratchpad-3.0-alpha1-20050704.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-log4j/lib/log4j-1.2.11.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-lucene-analyzers/lib/lucene-analyzers-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-nekohtml/lib/nekohtml-0.9.4.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-xml/lib/jaxen-core.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-xml/lib/jaxen-jdom.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-xml/lib/jdom.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-xml/lib/saxpath.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/lib-xml/lib/xercesImpl.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/ontology/lib/commons-logging-1.0.3.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/ontology/lib/icu4j_2_6_1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/ontology/lib/jena-2.1.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/parse-html/lib/tagsoup-1.0rc3.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/parse-pdf/lib/PDFBox-0.7.2-log4j.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/parse-rss/lib/commons-feedparser-0.6-fork.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/parse-rss/lib/xmlrpc-1.2.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/parse-swf/lib/javaswf.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/protocol-ftp/lib/commons-net-1.2.0-dev.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/summary-lucene/lib/lucene-highlighter-2.0.0.jar"/>
- <classpathentry kind="lib" path="third-party/nutch/src/plugin/urlfilter-automaton/lib/automaton.jar"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/index-basic/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/index-more/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/languageidentifier/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/languageidentifier/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/lib-regex-filter/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/lib-regex-filter/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-ext/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-ext/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-html/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-html/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-js/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-pdf/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-pdf/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/parse-text/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/query-basic/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/query-more/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/query-site/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/query-url/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/query-url/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/scoring-opic/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/summary-basic/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/summary-lucene/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-automaton/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-automaton/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-prefix/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-regex/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-regex/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-suffix/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-suffix/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlnormalizer-basic/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlnormalizer-basic/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlnormalizer-pass/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlnormalizer-pass/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlnormalizer-regex/src/java"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/plugin/urlnormalizer-regex/src/test"/>
+ <classpathentry kind="src" path="nutchwax-thirdparty/nutch/src/test"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/clustering-carrot2/clustering-carrot2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/creativecommons/creativecommons.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/index-basic/index-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/index-more/index-more.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/language-identifier/language-identifier.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-http/lib-http.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-jakarta-poi/poi-3.0-alpha1-20050704.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-jakarta-poi/poi-scratchpad-3.0-alpha1-20050704.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-log4j/log4j-1.2.11.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-lucene-analyzers/lucene-analyzers-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-nekohtml/nekohtml-0.9.4.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-parsems/lib-parsems.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-regex-filter/lib-regex-filter.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-xml/jaxen-core.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-xml/jaxen-jdom.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-xml/jdom.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-xml/saxpath.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/lib-xml/xercesImpl.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/microformats-reltag/microformats-reltag.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/nutch-0.9-dev.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/nutch-extensionpoints/nutch-extensionpoints.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/ontology/ontology.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-ext/parse-ext.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-html/parse-html.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-js/parse-js.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-msexcel/parse-msexcel.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-mspowerpoint/parse-mspowerpoint.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-msword/parse-msword.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-oo/parse-oo.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-pdf/parse-pdf.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-rss/parse-rss.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-swf/parse-swf.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-text/parse-text.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/parse-zip/parse-zip.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/carrot2-filter-lingo.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/carrot2-local-core.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/carrot2-snowball-stemmers.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/carrot2-util-common.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/carrot2-util-tokenizer.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/clustering-carrot2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/commons-collections-3.1-patched.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/commons-pool-1.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/Jama-1.0.1-patched.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/clustering-carrot2/violinstrings-1.0.2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/creativecommons/creativecommons.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/index-basic/index-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/index-more/index-more.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/language-identifier/language-identifier.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-http/lib-http.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-jakarta-poi/poi-3.0-alpha1-20050704.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-jakarta-poi/poi-scratchpad-3.0-alpha1-20050704.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-log4j/log4j-1.2.11.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-lucene-analyzers/lucene-analyzers-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-nekohtml/nekohtml-0.9.4.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-parsems/lib-parsems.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-regex-filter/lib-regex-filter.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-xml/jaxen-core.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-xml/jaxen-jdom.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-xml/jdom.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-xml/saxpath.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/lib-xml/xercesImpl.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/microformats-reltag/microformats-reltag.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/nutch-extensionpoints/nutch-extensionpoints.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/ontology/commons-logging-1.0.3.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/ontology/icu4j_2_6_1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/ontology/jena-2.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/ontology/ontology.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-ext/parse-ext.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-html/parse-html.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-html/tagsoup-1.0rc3.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-js/parse-js.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-msexcel/parse-msexcel.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-mspowerpoint/parse-mspowerpoint.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-msword/parse-msword.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-oo/parse-oo.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-pdf/parse-pdf.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-pdf/PDFBox-0.7.2-log4j.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-rss/commons-feedparser-0.6-fork.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-rss/parse-rss.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-rss/xmlrpc-1.2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-swf/javaswf.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-swf/parse-swf.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-text/parse-text.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/parse-zip/parse-zip.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/protocol-file/protocol-file.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/protocol-ftp/commons-net-1.2.0-dev.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/protocol-ftp/protocol-ftp.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/protocol-http/protocol-http.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/protocol-httpclient/protocol-httpclient.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/query-basic/query-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/query-more/query-more.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/query-site/query-site.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/query-url/query-url.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/scoring-opic/scoring-opic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/subcollection/subcollection.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/summary-basic/summary-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/summary-lucene/lucene-highlighter-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/summary-lucene/summary-lucene.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlfilter-automaton/automaton.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlfilter-automaton/urlfilter-automaton.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlfilter-prefix/urlfilter-prefix.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlfilter-regex/urlfilter-regex.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlfilter-suffix/urlfilter-suffix.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlnormalizer-basic/urlnormalizer-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlnormalizer-pass/urlnormalizer-pass.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/plugins/urlnormalizer-regex/urlnormalizer-regex.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/protocol-file/protocol-file.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/protocol-ftp/protocol-ftp.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/protocol-http/protocol-http.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/protocol-httpclient/protocol-httpclient.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/query-basic/query-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/query-more/query-more.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/query-site/query-site.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/query-url/query-url.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/scoring-opic/scoring-opic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/subcollection/subcollection.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/summary-basic/summary-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/summary-lucene/summary-lucene.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlfilter-automaton/urlfilter-automaton.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlfilter-prefix/urlfilter-prefix.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlfilter-regex/urlfilter-regex.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlfilter-suffix/urlfilter-suffix.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlnormalizer-basic/urlnormalizer-basic.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlnormalizer-pass/urlnormalizer-pass.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build/urlnormalizer-regex/urlnormalizer-regex.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/lib/commons-beanutils.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/lib/commons-collections-3.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/lib/commons-digester.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/lib/jstl.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/lib/standard.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/lib/struts.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/contrib/web2/plugins/web-caching-oscache/lib/oscache-2.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/commons-cli-2.0-SNAPSHOT.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/commons-codec-1.3.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/commons-httpclient-3.0.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/commons-lang-2.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/commons-logging-1.0.4.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/commons-logging-api-1.0.4.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/hadoop-0.10.1-core.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jakarta-oro-2.0.7.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jets3t.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jetty-5.1.4.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jetty-ext/ant.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jetty-ext/commons-el.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jetty-ext/jasper-compiler.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jetty-ext/jasper-runtime.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/jetty-ext/jsp-api.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/junit-3.8.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/log4j-1.2.13.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/lucene-core-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/lucene-misc-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/pmd-ext/jakarta-oro-2.0.8.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/pmd-ext/jaxen-1.1-beta-7.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/pmd-ext/pmd-3.6.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/servlet-api.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/taglibs-i18n.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/xerces-2_6_2-apis.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/lib/xerces-2_6_2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/carrot2-filter-lingo.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/carrot2-local-core.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/carrot2-snowball-stemmers.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/carrot2-util-common.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/carrot2-util-tokenizer.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/commons-collections-3.1-patched.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/commons-pool-1.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/Jama-1.0.1-patched.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/clustering-carrot2/lib/violinstrings-1.0.2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-jakarta-poi/lib/poi-3.0-alpha1-20050704.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-jakarta-poi/lib/poi-scratchpad-3.0-alpha1-20050704.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-log4j/lib/log4j-1.2.11.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-lucene-analyzers/lib/lucene-analyzers-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-nekohtml/lib/nekohtml-0.9.4.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-xml/lib/jaxen-core.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-xml/lib/jaxen-jdom.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-xml/lib/jdom.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-xml/lib/saxpath.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/lib-xml/lib/xercesImpl.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/ontology/lib/commons-logging-1.0.3.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/ontology/lib/icu4j_2_6_1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/ontology/lib/jena-2.1.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/parse-html/lib/tagsoup-1.0rc3.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/parse-pdf/lib/PDFBox-0.7.2-log4j.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/parse-rss/lib/commons-feedparser-0.6-fork.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/parse-rss/lib/xmlrpc-1.2.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/parse-swf/lib/javaswf.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/protocol-ftp/lib/commons-net-1.2.0-dev.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/summary-lucene/lib/lucene-highlighter-2.0.0.jar"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/src/plugin/urlfilter-automaton/lib/automaton.jar"/>
<classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER"/>
<classpathentry kind="con" path="org.maven.ide.eclipse.MAVEN2_CLASSPATH_CONTAINER"/>
- <classpathentry kind="lib" path="third-party/nutch/build"/>
- <classpathentry kind="lib" path="third-party/nutch/conf"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/build"/>
+ <classpathentry kind="lib" path="nutchwax-thirdparty/nutch/conf"/>
<classpathentry kind="output" path="target"/>
</classpath>
Deleted: trunk/archive-access/projects/nutchwax/maven.xml
===================================================================
--- trunk/archive-access/projects/nutchwax/maven.xml 2007-03-21 23:27:40 UTC (rev 1625)
+++ trunk/archive-access/projects/nutchwax/maven.xml 2007-03-22 00:07:39 UTC (rev 1626)
@@ -1,181 +0,0 @@
-<?xml version="1.0"?>
-<project
- xmlns:j="jelly:core"
- xmlns:define="jelly:define"
- xmlns:doc="doc"
- xmlns:artifact="artifact"
- xmlns:util="jelly:util"
- xmlns:maven="jelly:maven"
- xmlns:ant="jelly:ant">
-
- <goal name="site:update-sourceforge" description="Update sf.">
- <exec executable="rsync" >
- <arg value="--quiet" />
- <arg value="--archive" />
- <arg value="--rsh=ssh" />
- <arg
- value="${maven.build.dir}/docs/"/>
- <arg value="${maven.username}@archive-access.sf.net:/home/groups/a/ar/archive-access/htdocs/projects/nutchwax/" />
- </exec>
- </goal>
-
- <preGoal name="xdoc:jelly-transform">
- <attainGoal name="faq" />
- <attainGoal name="docbook"/>
- </preGoal>
-
- <postGoal name="xdoc:jelly-transform">
- <!--Overwrite the maven stylesheets because the do that
- different greys on odd and even rows in tables and its messing up
- my home page-->
- <echo message="Copying stylesheet ${basedir}/xdocs/style/maven-theme.css to ${maven.build.dir}/docs/style/" />
- <copy todir="${maven.build.dir}/docs/style/"
- file="${basedir}/xdocs/style/maven-theme.css" overwrite="true" />
- </postGoal>
-
- <postGoal name="site:generate" >
- <copy todir="${maven.build.dir}/docs/images">
- <fileset dir="${basedir}/src/images" >
- <include name="**/*" />
- </fileset>
- </copy>
- </postGoal>
-
- <!--Call out to ant goals.-->
- <goal name="ant:war"><ant:ant target="war" /></goal>
- <goal name="ant:clean" ><ant:ant target="clean" /></goal>
- <goal name="ant:jar"><ant:ant target="jar" /></goal>
- <goal name="ant:job" prereqs="ant:war"><ant:ant target="job" /></goal>
- <goal name="ant:compile"><ant:ant target="compile" /></goal>
- <goal name="ant:compile-plugins"><ant:ant target="compile-plugins" /></goal>
- <goal name="ant:init"><ant:ant target="init" /></goal>
- <preGoal name="clean" >
- <!--Maven in 1.0.2, the maven on build box, doesn't support
- the 'available' argument used in the ant clean target. Uncomment until
- we update the build box to use a later maven.
-
- Note: The build box scripts are doing this build of the nutchwax jar.
- Remove this special handling when we move to new maven.
-
- I just tired replacing the ant jar with a new one but that fails;
- maven has hardcoded dependencies on old 1.5.x ant.
-
- <attainGoal name="ant:clean" />
- -->
- </preGoal>
-
- <goal name="jar:jar"><!--Block building of jar--></goal>
-
- <postGoal name="dist:build-setup">
- <ant:available file="${basedir}/target/nutchwax.jar"
- property="job.jar.exists"/>
- <ant:fail
- message="Must run ant 'jar' and 'war' targets before maven dist"
- unless="job.jar.exists" />
- <!--Maven in 1.0.2, the maven on build box, doesn't support
- the 'available' argument used in the ant clean target. Uncomment until
- we update the build box to use a later maven.
- <attainGoal name="ant:war" />
- <attainGoal name="ant:job" />
- -->
- </postGoal>
-
- <postGoal name="dist:prepare-bin-filesystem">
- <echo>[nutchwax] dist:prepare-bin-filesystem postGoal</echo>
- <attainGoal name="docbook" />
- <copy todir="${maven.dist.bin.assembly.dir}/bin"
- filtering="true" overwrite="true" >
- <fileset dir="${basedir}/bin" />
- </copy>
- <copy todir="${maven.dist.bin.assembly.dir}">
- <fileset dir="${basedir}/target/">
- <include name="nutchwax.war"/>
- <include name="nutchwax.jar"/>
- </fileset>
- </copy>
- <!--Rename the nutchwax.jar to include a version-->
- <ant:rename src="${maven.dist.bin.assembly.dir}/nutchwax.jar"
- dest="${maven.dist.bin.assembly.dir}/nutchwax-${pom.currentVersion}.jar" />
- <attainGoal name="copy_docbook" />
- <!--Copy over the wax-default.xml and hadoop-site.xml.
- They're referred to in documentation.
- -->
- <copy todir="${maven.dist.bin.assembly.dir}/docs">
- <fileset dir="${basedir}/conf" >
- <include name="hadoop-site.xml.template"/>
- <include name="wax-default.xml"/>
- </fileset>
- </copy>
- <!--Copy over the parse-pdf.sh so available outside of
- nutchwax.jar bundle for those running standalone mode
- -->
- <copy todir="${maven.dist.bin.assembly.dir}/bin">
- <fileset dir="${basedir}/src/plugin/parse-waxext/bin" >
- <include name="parse-pdf.sh"/>
- </fileset>
- </copy>
- </postGoal>
-
- <postGoal name="dist:prepare-src-filesystem">
- <echo>[nutchwax] dist:prepare-src-filesystem postGoal</echo>
- <copy todir="${maven.dist.src.assembly.dir}/bin"
- filtering="true" overwrite="true" >
- <fileset dir="${basedir}/bin" />
- </copy>
- <copy todir="${maven.dist.src.assembly.dir}/" overwrite="true" >
- <fileset dir="${basedir}" >
- <include name="README.txt" />
- </fileset>
- </copy>
- <copy todir="${maven.dist.src.assembly.dir}/conf" overwrite="true" >
- <fileset dir="${basedir}/conf" />
- </copy>
- <copy todir="${maven.dist.src.assembly.dir}/xdocs" overwrite="true" >
- <fileset dir="${basedir}/xdocs" />
- </copy>
- <copy todir="${maven.dist.src.assembly.dir}/lib" overwrite="true" >
- <fileset dir="${basedir}/lib" />
- </copy>
- </postGoal>
-
- <goal name="docbook">
- <!--Check that the jimi jars are present. If not, skip docbook.
- -->
- <j:set var="sDocbookSupportPresent" value="false" />
- <util:available file="${maven.repo.local}/jimi/jars" >
- <j:set var="sDocbookSupportPresent" value="true"/>
- </util:available>
- <j:if test="${sDocbookSupportPresent == 'true'}">
- <attainGoal name="sdocbook:generate-html"/>
- <!--<attainGoal name="sdocbook:generate-pdf"/>
- -->
- </j:if>
- <!-- else -->
- <j:if test="${sDocbookSupportPresent != 'true'}">
- <echo
- message="sdocbook SUN jimi jar not present. Skipping docbook build." />
- </j:if>
- </goal>
-
- <goal name="copy_docbook">
- <!--Copies docbooks under dist docs and under webapps. This goal runs
- after these dirs have been setup.
- -->
- <property name="docbook.assembled"
- value="${maven.build.dir}/docs/articles/" />
- <j:set var="docbookPresent" value="false" />
- <util:available file="${docbook.assembled}" >
- <j:set var="docbookPresent" value="true"/>
- </util:available>
- <j:if test="${docbookPresent == 'true'}">
- <echo message="Copying over docbook" />
- <!--Copy under docs in binary build-->
- <mkdir dir="${maven.dist.bin.assembly.dir}/docs/articles/" />
- <copy todir="${maven.dist.bin.assembly.dir}/docs/articles/"
- verbose="true">
- <fileset dir="${maven.build.dir}/docs/articles/" />
- </copy>
- </j:if>
- </goal>
-
-</project>
Modified: trunk/archive-access/projects/nutchwax/nutchwax-core/pom.xml
===================================================================
--- trunk/archive-access/projects/nutchwax/nutchwax-core/pom.xml 2007-03-21 23:27:40 UTC (rev 1625)
+++ trunk/archive-access/projects/nutchwax/nutchwax-core/pom.xml 2007-03-22 00:07:39 UTC (rev 1626)
@@ -27,8 +27,8 @@
<source>1.5</source>
<target>1.5</target>
<!--
- <compilerArgument> -verbose -cp ../third-party/nutch/build/classes</compilerArgument>
- <compilerArgument> -verbose -classpath ../third-party/nutch/build/classes</compilerArgument>
+ <compilerArgument> -verbose -cp ../nutchwax-thirdparty/nutch/build/classes</compilerArgument>
+ <compilerArgument> -verbose -classpath ../nutchwax-thirdparty/nutch/build/classes</compilerArgument>
-->
</configuration>
</plugin>
Modified: trunk/archive-access/projects/nutchwax/nutchwax-job/src/main/assembly/assemble-job.xml
===================================================================
--- trunk/archive-access/projects/nutchwax/nutchwax-job/src/main/assembly/assemble-job.xml 2007-03-21 23:27:40 UTC (rev 1625)
+++ trunk/archive-access/projects/nutchwax/nutchwax-job/src/main/assembly/assemble-job.xml 2007-03-22 00:07:39 UTC (rev 1626)
@@ -34,7 +34,7 @@
</includes>
</fileSet>
<fileSet>
- <directory>../third-party/nutch/build/plugins</directory>
+ <directory>../nutchwax-thirdparty/nutch/build/plugins</directory>
<outputDirectory>/plugins</outputDirectory>
<includes>
<include>analysis-*/**</include>
@@ -54,7 +54,7 @@
</excludes>
</fileSet>
<fileSet>
- <directory>../third-party/nutch/conf</directory>
+ <directory>../nutchwax-thirdparty/nutch/conf</directory>
<outputDirectory>/</outputDirectory>
<includes>
<include>mime-types.xml</include>
@@ -64,7 +64,7 @@
</includes>
</fileSet>
<fileSet>
- <directory>../third-party/nutch/lib</directory>
+ <directory>../nutchwax-thirdparty/nutch/lib</directory>
<outputDirectory>/lib</outputDirectory>
<includes>
<include>commons-lang*</include>
@@ -81,7 +81,7 @@
class can be inside of a jar.
<fileSet>
- <directory>../third-party/nutch/build</directory>
+ <directory>../nutchwax-thirdparty/nutch/build</directory>
<outputDirectory>/lib</outputDirectory>
<includes>
<include>nutch*jar</include>
@@ -89,7 +89,7 @@
</fileSet>
-->
<fileSet>
- <directory>../third-party/nutch/build/classes</directory>
+ <directory>../nutchwax-thirdparty/nutch/build/classes</directory>
<outputDirectory>/</outputDirectory>
</fileSet>
</fileSets>
Property changes on: trunk/archive-access/projects/nutchwax/nutchwax-thirdparty
___________________________________________________________________
Name: svn:externals
+ nutch -r 508238 http://svn.apache.org/repos/asf/lucene/nutch/trunk
Modified: trunk/archive-access/projects/nutchwax/nutchwax-thirdparty/pom.xml
===================================================================
--- trunk/archive-access/projects/nutchwax/nutchwax-thirdparty/pom.xml 2007-03-21 23:27:40 UTC (rev 1625)
+++ trunk/archive-access/projects/nutchwax/nutchwax-thirdparty/pom.xml 2007-03-22 00:07:39 UTC (rev 1626)
@@ -38,7 +38,7 @@
we can invoke it from eclipse.
-->
<echo>Building nutch third-party dependency (jar)</echo>
- <ant dir="../third-party/nutch" target="jar" inheritAll="false" >
+ <ant dir="nutch" target="jar" inheritAll="false" >
<property name="build.compiler" value="extJavac" />
</ant>
<!--Copy over the nutch classes to target/classes so they
@@ -46,10 +46,10 @@
has on its classpath when it goes to build subsequent modules).
-->
<copy todir="target/classes" overwrite="true">
- <fileset dir="../third-party/nutch/build/classes" />
+ <fileset dir="nutch/build/classes" />
</copy>
<echo>Building nutch third-party dependency (plugins)</echo>
- <ant dir="../third-party/nutch" target="compile-plugins" inheritAll="false" >
+ <ant dir="nutch" target="compile-plugins" inheritAll="false" >
<property name="build.compiler" value="extJavac" />
</ant>
</tasks>
@@ -64,7 +64,7 @@
<configuration>
<tasks>
<echo>Cleaning nutch third-party dependency</echo>
- <ant dir="../third-party/nutch" target="clean" inheritAll="false" >
+ <ant dir="nutch" target="clean" inheritAll="false" >
...
[truncated message content] |
