<?xml version="1.0" encoding="utf-8"?>
<!--

Put this below 'config' in solr/conf/solrconfig.xml

	<requestHandler name="/dataimport" class="org.apache.solr.handler.dataimport.DataImportHandler">
		<lst name="defaults">
			<str name="config">/tmp/dataimport-wordpress.xml</str>
		</lst>
	</requestHandler>


This file is dataimport-wordpress.xml

-->

<dataConfig>
<dataSource type="JdbcDataSource" driver="com.mysql.jdbc.Driver" url="jdbc:mysql://localhost/wp" user="root" password="" readOnly="true" batchSize="-1" />
    <document name="wordpress">

	<!-- Iterate over each Wordpress tablename prefix found in the information schema -->
	<entity
		name="blogtables"
		recursive="true"
		query="SELECT REPLACE(TABLE_NAME, '_posts', '') AS blog FROM information_schema.TABLES WHERE TABLE_SCHEMA=DATABASE() AND TABLE_NAME NOT LIKE 'wp_nwt_%' AND TABLE_NAME LIKE 'wp_%_posts'"
		rootEntity="false"
		>


		<!-- This is our root entity and each row returned will index one document -->
		<entity rootEntity="true"
			name="blog"
			pk="ID"
			query="SELECT * FROM ${blogtables.blog}_posts WHERE post_type='post' AND post_status='publish' AND post_password=''"
			transformer="HTMLStripTransformer,TemplateTransformer,NumberFormatTransformer">

			<!-- XXX: This should be $home/$post_date/$post_name/ -->
			<field name="url" column="guid" />

			<!-- XXX: Will always be 0 -->
			<entity query="SELECT IF(post_status='publish', 0, 1) AS deleted FROM ${blogtables.blog}_posts WHERE ID=${blog.ID}">
				<field column="deleted" />
			</entity>

			<field column="source" template="${blogtables.blog}" />
			<field name="sourceID" column="ID" formatStyle="number"/>

			<field name="pubdate" column="post_date_gmt" />
			<field name="moddate" column="post_modified_gmt" />
			<entity query="SELECT CONCAT(option_value, '/') AS section FROM ${blogtables.blog}_options WHERE option_name='home'">
				<field name="homesection" column="section" />
				<field name="sections" column="section" />
			</entity>
			<field name="headline" column="post_title" />
			<field name="body" column="post_content" stripHTML="true" />

			<entity query="SELECT display_name FROM ${blogtables.blog}_users WHERE ID=${blog.post_author}">
				<field name="authors" column="display_name" />
			</entity>

		</entity> <!-- /name: blog -->

	</entity> <!-- /name: blogtables -->

    </document> <!-- /name: wordpress -->

</dataConfig>