[Archive-access-cvs] SF.net SVN: archive-access: [2345] trunk/archive-access/projects/nutchwax/ archive/src/plugin/query-nutchwax/src/java/org/archive/nutchwax/query/ DateQueryFilter.java

[<bi...@us...>]

Revision: 2345
          http://archive-access.svn.sourceforge.net/archive-access/?rev=2345&view=rev
Author:   binzino
Date:     2008-06-30 13:38:36 -0700 (Mon, 30 Jun 2008)

Log Message:
-----------
Changed logic to that the RangeQuery has a boost of 0.0f and is always
required.  This is necessary for Nutch to auto-convert the RangeQuery
into a RangeFilter.

Added class-level JavaDoc.

Modified Paths:
--------------
    trunk/archive-access/projects/nutchwax/archive/src/plugin/query-nutchwax/src/java/org/archive/nutchwax/query/DateQueryFilter.java

Modified: trunk/archive-access/projects/nutchwax/archive/src/plugin/query-nutchwax/src/java/org/archive/nutchwax/query/DateQueryFilter.java
===================================================================
--- trunk/archive-access/projects/nutchwax/archive/src/plugin/query-nutchwax/src/java/org/archive/nutchwax/query/DateQueryFilter.java	2008-06-29 03:07:43 UTC (rev 2344)
+++ trunk/archive-access/projects/nutchwax/archive/src/plugin/query-nutchwax/src/java/org/archive/nutchwax/query/DateQueryFilter.java	2008-06-30 20:38:36 UTC (rev 2345)
@@ -37,7 +37,28 @@
 import org.apache.nutch.searcher.QueryFilter;
 
 /** 
- * 
+ * <p>
+ *   Filter on a date or date range.  This filter assumes the dates
+ *   are in field named "date" and adhere to the IA 14-digit date
+ *   format: YYYYMMDDHHMMSS
+ * </p>
+ * <p>
+ *   Date values in the query can have less than the full 14-digit
+ *   precision.  In that case, they are converted into a range over
+ *   given precision.  For example, "date:2007" is automagically
+ *   converted into "date[20070000000000-20079999999999]".
+ * </p>
+ * <p>
+ *   NOTE: In order for this filter to take advantage of the Nutch
+ *   auto-magic conversion of RangeQuery into RangeFilter, we have to
+ *   create the RangeQuery with:
+ *   <ul>
+ *   <li>occur = BooleanClause.Occur.MUST</li>
+ *   <li>boost = 0.0f;</li>
+ *   </ul>
+ *   These are the two conditions that Nutch's LuceneQueryOptimizer
+ *   checks before doing a RangeQuery-&gt;RangeFilter conversion.
+ * </p>
  */
 public class DateQueryFilter implements QueryFilter
 {
@@ -122,17 +143,14 @@
         return ;
       }
 
+    // Otherwise make it plain-old TermQuery to match the exact date.
     TermQuery term = new TermQuery( new Term( FIELD, date ) );
     
-    // Set boost on term?
-    // term.setBoolst( boost );
+    // Not strictly required since this is a TermQuery and not a
+    // RangeQuery, but we use the same 0.0f boost for consistency.
+    term.setBoost( 0.0f );
 
-    output.add( term, 
-                ( clause.isProhibited()
-                  ? BooleanClause.Occur.MUST_NOT
-                  : ( clause.isRequired()
-                      ? BooleanClause.Occur.MUST
-                      : BooleanClause.Occur.SHOULD ) ) );
+    output.add( term, BooleanClause.Occur.MUST );
   }
 
   private void doRangeQuery( BooleanQuery output, Clause clause, String lower, String upper )
@@ -143,16 +161,11 @@
     RangeQuery range = new RangeQuery( new Term( FIELD, lower ),
                                        new Term( FIELD, upper ),
                                        true );
-    
-    // Set boost on range query?
-    // range.setBoost( boost );
 
-    output.add( range,
-                ( clause.isProhibited()
-                  ? BooleanClause.Occur.MUST_NOT
-                  : ( clause.isRequired()
-                      ? BooleanClause.Occur.MUST
-                      : BooleanClause.Occur.SHOULD ) ) );
+    // Required for LuceneQueryOptimizer to convert to RangeFilter.
+    range.setBoost( 0.0f );
+
+    output.add( range, BooleanClause.Occur.MUST );
   }
 
 


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.