Question

I need to test if certain documents match a query before actually indexing them. How would you do this? One of the possibilities I’m thinking of is running a plain lucene index on memory (ramdisk?) and follow a index -> test query -> delete loop for every new document I have before sending it to the actual Solr server.

Can anyone think of a better solution for this problem?

Thanks a lot.

Update:

Looks like this could be a good starting point: http://www.lucenetutorial.com/lucene-in-5-minutes.html

Answer 1

Since Solr allows transactions / commits you can actually index them and before you do commit do state a delete query which removes all non matching documents.

/**
 * @author Omnaest
 */
public class SolrSimpleIndexingTest
{
  protected SolrServer solrServer = newSolrServerInstance();

  @Test
  public void testSolr() throws IOException,
                        SolrServerException
  {

    {
      SolrInputDocument solrInputDocument = new SolrInputDocument();
      {
        solrInputDocument.addField( "id", "0" );
        solrInputDocument.addField( "text", "test1" );
      }
      this.solrServer.add( solrInputDocument );
    }
    {
      SolrInputDocument solrInputDocument = new SolrInputDocument();
      {
        solrInputDocument.addField( "id", "1" );
        solrInputDocument.addField( "text", "test2" );
      }
      this.solrServer.add( solrInputDocument );
    }
    this.solrServer.deleteByQuery( "text:([* TO *] -test2)" );
    this.solrServer.commit();

    /*
     * Now your index does only contain the document with id=1 !!
     */

    QueryResponse queryResponse = this.solrServer.query( new SolrQuery().setQuery( "*:*" ) );
    SolrDocumentList solrDocumentList = queryResponse.getResults();

    assertEquals( 1, solrDocumentList.size() );
    assertEquals( "1", solrDocumentList.get( 0 ).getFieldValue( "id" ) );
  }

  /**
   * @return
   */
  private static CommonsHttpSolrServer newSolrServerInstance()
  {
    try
    {
      return new CommonsHttpSolrServer( "http://localhost:8983/solr" );
    }
    catch ( MalformedURLException e )
    {
      e.printStackTrace();
      fail();
    }
    return null;
  }
}