build.xml

<project name="reconcile-hbase-annotations" default="compile" basedir="." >

	<description>Code to interact with hbase</description>

	<property name="antMavenBuildDir" value="../ant-maven-build"/>
	<import file="${antMavenBuildDir}/build.xml"/>
	
	<!--  =================================================================== -->
	<!--  run targets                     -->
	<!--  =================================================================== -->

	<!-- =================================================================== -->
	<!-- m/r query tasks -->
	<!-- =================================================================== -->
	<target name="checkAnnotations" depends="jars" description="check annotation strings. params: src">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${depJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.AnnotationChecker" />
			<arg value="-table=${table}" />
			<arg value="-source=${src}" />
		</exec>
	</target>

	<target name="stripBadAnnotations" depends="jars" description="strip bad annotation strings. params: src">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${depJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.BadAnnotationStripper" />
			<arg value="-source=${src}" />
		</exec>
	</target>

	
	<!-- ======================================================= -->
	<!-- annotation tasks -->
	<!-- ======================================================= -->
	<target name="baseAnnot" depends="hadoopJar" description="execute base NLP annotations over entire doc table">
		<echo>Filter Source : ${src}</echo>
		<echo>Filter HDFS Key List File : ${keyList}</echo>
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.BaseReconcileAnnotation" />
			<arg value="-source=${src}"/>
			<arg value="-keyList=${keyList}" />
		</exec>
		<sleep seconds="10" />
		<exec executable="${hadoop}">
			<arg value="job" />
			<arg value="-list" />
		</exec>
	</target>

	<target name="parAnnot" depends="hadoopJar" description="paragraph annotations over entire doc table">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="trinidad.hbase.mapreduce.annotation.ParagraphAnnotation" />
		</exec>
	</target>
	<target name="sentAnnot" depends="hadoopJar" description="sentence annotations over entire doc table">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="trinidad.hbase.mapreduce.annotation.SentenceAnnotation" />
		</exec>
	</target>
	<target name="tokenAnnot" depends="hadoopJar" description="token annotations over entire doc table">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="trinidad.hbase.mapreduce.annotation.TokenAnnotation" />
		</exec>
	</target>
	<target name="posAnnot" depends="hadoopJar" description="POS annotations over entire doc table">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="trinidad.hbase.mapreduce.annotation.POSAnnotation" />
		</exec>
	</target>

	<target name="neAnnot" depends="hadoopJar" description="execute named entity annotations over  doc table">
		<echo>Filter Source : ${src}</echo>
		<echo>Filter HDFS Key List File : ${keyList}</echo>
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.NEAnnotation" />
			<arg value="-source=${src}"/>
			<arg value="-keyList=${keyList}" />
		</exec>
	</target>

	<target name="parserAnnot" depends="hadoopJar" description="execute parser annotations over doc table ">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.ParserAnnotation" />
			<arg value="-source=${src}" />
			<arg value="-keyList=${keyList}" />
		</exec>
	</target>

	<target name="bnpAnnot" depends="hadoopJar" description="execute base NP annotations over doc table ">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.BaseNPAnnotation" />
			<arg value="-source=${src}" />
			<arg value="-keyList=${keyList}" />
		</exec>
	</target>

	<target name="corefAnnot" depends="hadoopJar" description="execute coref annotations over doc table ">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.CorefResolver" />
			<arg value="-source=${src}" />
			<arg value="-keyList=${keyList}" />
		</exec>
	</target>

	<target name="corefPairs" depends="hadoopJar" description="execute coref pair annotations over doc table ">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.annotation.CorefPairs" />
			<arg value="-source=${src}" />
			<arg value="-keyList=${keyList}" />
		</exec>
	</target>

	<target name="annotate" depends="hadoopJar" description="Run all neccessary annotators">
		<echo>Source name: ${src}</echo>
		<echo>Key List File: ${keyList}</echo>
		<antcall target="baseAnnot"/>
		<antcall target="neAnnot"/>
		<antcall target="parserAnnot"/>
		<antcall target="bnpAnnot"/>
		<antcall target="corefPairs"/>

	</target>

	<target name="printNPs" depends="hadoopJar" description="print the different types of treebank tag counts">
		<exec executable="${hadoop}" spawn="false">
			<arg value="jar" />
			<arg value="${hadoopJarFile}" />
			<arg value="reconcile.hbase.mapreduce.GenericMapper" />
			<arg value="-source=${src}" />
			<arg value="-keyList=${keyList}" />
			<arg value="-family=annotations" />
			<arg value="-family=text" />
			<arg value="-mapper=reconcile.hbase.mapper.PrintNPs" />
		</exec>
	</target>

</project>