Full text indexing with Solr
Hi, When Oak uses Solr then do we send the complete binary to Solr for full text indexing or we extract the content on Oak side and send the extracted content. And if send the complete binary content do we send it inline or it is first uploaded to Solr and reference to that is passed? Chetan Mehrotra
Re: Full text indexing with Solr
Hi Chetan, I think that currently the complete binary is sent and the on the Solr side you have the ability to choose which field to use for indexing and searching properties of Binary type via the OakSolrConfiguration#getFieldNameFor(Type propertyType) [1] method. Currently the default configuration and implementation use a Solr binary dynamic field of type so that a binary property called propname is indexed in a Solr field called propname_bin of type BinaryField [2], however my plan for it is to implement some dedicated analyzers that use Apache Tika to extract the text and index that instead (or too). Regards, Tommaso [1] : http://svn.apache.org/repos/asf/jackrabbit/oak/trunk/oak-solr-core/src/main/java/org/apache/jackrabbit/oak/plugins/index/solr/OakSolrConfiguration.java [2] : http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/BinaryField.html 2013/9/24 Chetan Mehrotra > Hi, > > When Oak uses Solr then do we send the complete binary to Solr for > full text indexing or we extract the content on Oak side and send the > extracted content. > > And if send the complete binary content do we send it inline or it is > first uploaded to Solr and reference to that is passed? > > Chetan Mehrotra >
Re: Full text indexing with Solr
Thanks for the details Tommaso. Would look at the code for implementation details. Chetan Mehrotra On Wed, Sep 25, 2013 at 2:33 AM, Tommaso Teofili wrote: > Hi Chetan, > > I think that currently the complete binary is sent and the on the Solr side > you have the ability to choose which field to use for indexing and > searching properties of Binary type via the > OakSolrConfiguration#getFieldNameFor(Type propertyType) [1] method. > > Currently the default configuration and implementation use a Solr binary > dynamic field of type so that a binary property called propname is indexed > in a Solr field called propname_bin of type BinaryField [2], however my > plan for it is to implement some dedicated analyzers that use Apache Tika > to extract the text and index that instead (or too). > > Regards, > Tommaso > > [1] : > http://svn.apache.org/repos/asf/jackrabbit/oak/trunk/oak-solr-core/src/main/java/org/apache/jackrabbit/oak/plugins/index/solr/OakSolrConfiguration.java > [2] : > http://lucene.apache.org/solr/4_4_0/solr-core/org/apache/solr/schema/BinaryField.html > > > > 2013/9/24 Chetan Mehrotra > >> Hi, >> >> When Oak uses Solr then do we send the complete binary to Solr for >> full text indexing or we extract the content on Oak side and send the >> extracted content. >> >> And if send the complete binary content do we send it inline or it is >> first uploaded to Solr and reference to that is passed? >> >> Chetan Mehrotra >>