lucene/sandbox/contributions/webcrawler-LARM/build.xml

<?xml version="1.0"?>

<project name="webcrawler-LARM" default="build" basedir=".">

  <property file="${basedir}/build.properties" />

  <property name="name"           value="webcrawler_LARM"/>
  <property name="version"        value="0.5"/>
  <property name="final.name"     value="${name}-${version}"/>
  <property name="debug"          value="on"/>

  <property name="src.dir"        value="./src"/>
  <property name="lib.dir"        value="./libs"/>
  <property name="logs.dir"       value="./logs"/>
  <property name="cache.dir"      value="./cachingqueue"/>
  <property name="build.dir"      value="./build"/>

  <property name="src.httpclient" value="${lib.dir}/HTTPClient.zip"/>
  <property name="build.classes"  value="${build.dir}/src"/>
  <property name="build.src"      value="${build.dir}/src"/>
  <property name="build.encoding" value="ISO-8859-1"/>

  <property name="threads"        value="15"/>

  <!-- Build classpath -->
  <path id="classpath">
    <pathelement location="${build.classes}"/>
    <pathelement location="${lucene.jar}"/>
    <pathelement location="${oro.jar}"/>
    <fileset dir="${lib.dir}">
        <include name="*.jar" />
     </fileset>
   </path>

  <path id="run.classpath">
    <pathelement location="${build.dir}/${final.name}.jar"/>
    <pathelement location="${lucene.jar}"/>
    <pathelement location="${oro.jar}"/>
    <fileset dir="${lib.dir}">
      <include name="*.jar" />
    </fileset>
  </path>



  <!-- ================================================================== -->
  <!-- I N I T                                                            -->
  <!-- ================================================================== -->


  <target name="init">
    <available
      property="lucene.present"
      classname="org.apache.lucene.document.Document"
      classpath="${lucene.jar}"
    />
  </target>

  <target name="lucene_check" depends="init" unless="lucene.present">
    <echo>
      ##################################################################
      Lucene not found.
      Lucene Home: ${lucene.jar}
      ##################################################################
    </echo>
  </target>


  <!-- ================================================================== -->
  <!-- B U I L D                                                          -->
  <!-- ================================================================== -->
  <target name="build" depends="init,lucene_check"
    description="-> builds jar file">
    <mkdir dir="${build.dir}"/>
    <mkdir dir="${build.classes}"/>
    <mkdir dir="${build.src}"/>
    <unzip src="${src.httpclient}" dest="${build.src}" overwrite="false"/>
    <javac
      encoding="${build.encoding}"
      srcdir="${src.dir}:${build.src}"
      excludes="**/CVS/*"
      destdir="${build.classes}"
      debug="${debug}">
      <classpath refid="classpath"/>
    </javac>
    <jar
      jarfile="${build.dir}/${final.name}.jar"
      basedir="${build.classes}"
    />
  </target>

  <!-- ================================================================== -->
  <!-- Check Syntax for Run Task                                          -->
  <!-- ================================================================== -->
  <target name="checksyntax" unless="start">
    <echo>
    use run with the following syntax

    ant run -Dstart=&lt;URL&gt; -Drestrictto=&lt;Pattern&gt; [-Dthreads=&lt;Thread Count&gt;]

    default value for threads is 15
    </echo>
  </target>

  <!-- ================================================================== -->
  <!-- R U N                                                              -->
  <!-- ================================================================== -->
  <target name="run" depends="build, checksyntax" if="start"
    description="-> runs command-line version of the crawler">
    <delete dir="${logs.dir}"/>
    <mkdir dir="${logs.dir}"/>
    <java
      classname="de.lanlab.larm.fetcher.FetcherMain"
      fork="yes">
      <jvmarg value="-server"/>
      <jvmarg value="-Xmx400mb"/>
      <arg value="-start"/>
      <arg value="${start}"/>
      <arg value="-restrictto"/>
      <arg value="${restrictto}"/>
      <arg value="-threads"/>
      <arg value="${threads}"/>
      <classpath refid="run.classpath"/>
    </java>
  </target>

  <!-- ================================================================== -->
  <!-- C L E A N   L A S T   R U N                                        -->
  <!-- ================================================================== -->
  <target name="cleanlastrun"
    description="-> cleans files created by each run of the crawler">
    <delete dir="${logs.dir}"/>
    <delete dir="${cache.dir}"/>
  </target>

  <!-- ================================================================== -->
  <!-- C L E A N   A L L                                                  -->
  <!-- ================================================================== -->
  <target name="cleanall" depends="cleanlastrun"
    description="-> cleans all build and run files">
    <delete dir="${build.dir}"/>
  </target>
</project>