Name=Nutch name=nutch version=1.0 final.name=${name}-${version} year=2006 basedir = ./ src.dir = ./src/java lib.dir = ./lib conf.dir = ./conf plugins.dir = ./src/plugin docs.dir = ./docs docs.src = ${basedir}/src/web xmlcatalog.dir = ${basedir}/src/xmlcatalog build.dir = ./build build.classes = ${build.dir}/classes build.webapps = ${build.dir}/webapps build.plugins = ${build.dir}/plugins build.docs = ${build.dir}/docs build.javadoc = ${build.docs}/api build.encoding = UTF-8 test.src.dir = ./src/test test.build.dir = ${build.dir}/test test.build.data = ${test.build.dir}/data test.build.classes = ${test.build.dir}/classes test.build.javadoc = ${test.build.dir}/docs/api javacc.home=/usr/java/javacc web.src.dir = ./src/web src.webapps = ./src/webapps # Proxy Host and Port to use for building JavaDoc javadoc.proxy.host=-J-DproxyHost= javadoc.proxy.port=-J-DproxyPort= javadoc.link.java=http://java.sun.com/j2se/1.4.2/docs/api/ javadoc.link.lucene=http://jakarta.apache.org/lucene/docs/api/ javadoc.link.hadoop=http://lucene.apache.org/hadoop/docs/api/ javadoc.packages=org.apache.nutch.* dist.dir=${build.dir}/${final.name} javac.debug=on javac.optimize=on javac.deprecation=off javac.version= 1.5 # # Plugins API # plugins.api=\ org.apache.nutch.protocol.http.api*:\ org.apache.nutch.urlfilter.api*:\ org.apache.nutch.parse.ms* # # Protocol Plugins # plugins.protocol=\ org.apache.nutch.protocol.file*:\ org.apache.nutch.protocol.ftp*:\ org.apache.nutch.protocol.http*:\ org.apache.nutch.protocol.httpclient* # # URL Filter Plugins # plugins.urlfilter=\ org.apache.nutch.urlfilter.automaton*:\ org.apache.nutch.urlfilter.prefix*:\ org.apache.nutch.urlfilter.regex* # # Scoring Plugins # plugins.scoring=\ org.apache.nutch.scoring.opic* # # Parse Plugins # plugins.parse=\ org.apache.nutch.parse.ext*:\ org.apache.nutch.parse.html*:\ org.apache.nutch.parse.js:\ org.apache.nutch.parse.mp3*:\ org.apache.nutch.parse.msexcel*:\ org.apache.nutch.parse.mspowerpoint*:\ org.apache.nutch.parse.msword*:\ org.apache.nutch.parse.oo*:\ org.apache.nutch.parse.pdf*:\ org.apache.nutch.parse.rtf*:\ org.apache.nutch.parse.rss*:\ org.apache.nutch.parse.swf*:\ org.apache.nutch.parse.text:\ org.apache.nutch.parse.zip # # Analysis Plugins # plugins.analysis=\ # ${plugin.analysis-de}:\ # ${plugin.analysis-fr} # # Indexing Filter Plugins # plugins.index=\ org.apache.nutch.indexer.basic*:\ org.apache.nutch.indexer.more* # # Query Filter Plugins # plugins.query=\ org.apache.nutch.searcher.basic*:\ org.apache.nutch.searcher.more*:\ org.apache.nutch.searcher.site*:\ org.apache.nutch.searcher.url* # # Ontology Plugins # plugins.ontology=\ org.apache.nutch.ontology.jena* # # Online Clusterer Plugins # plugins.clustering=\ org.apache.nutch.clustering.carrot2* # # Summary Plugins # plugins.summary=\ org.apache.nutch.summary.basic*:\ org.apache.nutch.summary.lucene* # # Misc. Plugins # # (gathers plugins that cannot be dispatched # in any category, mainly because they contains # many extension points) # plugins.misc=\ org.apache.nutch.analysis.lang*:\ org.apache.nutch.microformats.reltag*:\ org.creativecommons.nutch*