Changeset 16912 for indexers/trunk

Timestamp:

2008-08-20T15:06:13+12:00 (16 years ago)

Author:

mdewsnip

Message:

Changes made by Richard Managh at DL Consulting Ltd for returning document-level term frequency totals, slightly modified to work with the latest Greenstone.

Location:

indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper

Files:

• GS2LuceneQuery.java (modified) (4 diffs)
• LuceneQueryResult.java (modified) (2 diffs)

indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneQuery.java

@@ -141 +141 @@
         HashSet terms = new HashSet();
         query.extractTerms(terms);
+
+        HashMap doc_term_freq_map = new HashMap();
         
         Iterator iter = terms.iterator();
@@ -149 +151 @@
         // Get the term frequency over all the documents
         TermDocs term_docs = reader.termDocs(term);
-        int term_freq = term_docs.freq();
+        int term_freq = 0;
         int match_docs = 0;
-        if (term_freq != 0) match_docs++;
-        while (term_docs.next()) {
-            term_freq += term_docs.freq();
-            if (term_docs.freq()!= 0) {
+        while (term_docs.next())
+        {
+            if (term_docs.freq() != 0)
+            {
+            term_freq += term_docs.freq();
             match_docs++;
+
+            // Calculate the document-level term frequency as well
+            Integer lucene_doc_num_obj = new Integer(term_docs.doc());
+            int doc_term_freq = 0;
+                        if (doc_term_freq_map.containsKey(lucene_doc_num_obj))
+            {
+                doc_term_freq = ((Integer) doc_term_freq_map.get(lucene_doc_num_obj)).intValue();
+            }
+            doc_term_freq += term_docs.freq();
+
+            doc_term_freq_map.put(lucene_doc_num_obj, new Integer(doc_term_freq));
             }
         }
@@ -186 +200 @@
 
         for (int i = start_results; i <= hits.length(); i++) {
+            int lucene_doc_num = hits.id(i - 1);
             Document doc = hits.doc(i - 1);
-            lucene_query_result.addDoc(doc.get("nodeID").trim(), hits.score(i-1));
+            int doc_term_freq = ((Integer) doc_term_freq_map.get(new Integer(lucene_doc_num))).intValue();
+            lucene_query_result.addDoc(doc.get("nodeID").trim(), hits.score(i-1), doc_term_freq);
         }
         }
@@ -202 +218 @@
         // Output the matching documents
         for (int i = start_results; (i <= hits.scoreDocs.length && i <= end_results); i++) {
-            Document doc = reader.document(hits.scoreDocs[i - 1].doc);
-            lucene_query_result.addDoc(doc.get("nodeID").trim(), hits.scoreDocs[i-1].score);
+            int lucene_doc_num = hits.scoreDocs[i - 1].doc;
+            Document doc = reader.document(lucene_doc_num);
+            int doc_term_freq = ((Integer) doc_term_freq_map.get(new Integer(lucene_doc_num))).intValue();
+            lucene_query_result.addDoc(doc.get("nodeID").trim(), hits.scoreDocs[i-1].score, doc_term_freq);
         }
         }

indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/LuceneQueryResult.java

@@ -110 +110 @@
     }
 
-    public void addDoc(String id, float rank)
+    public void addDoc(String id, float rank, int termfreq)
     {
-    docs_.add(new DocInfo(id, rank));
+    docs_.add(new DocInfo(id, rank, termfreq));
     }
     
@@ -225 +225 @@
     public String id_ = "";
     public float rank_ = 0;
-
-    public DocInfo (String id, float rank)
+    public int termfreq_ = 0;
+
+    public DocInfo (String id, float rank, int termfreq)
     {
         id_ = id;
         rank_ = rank;
+        termfreq_ = termfreq;
     }
 
     public String toString()
     {
-        return "" + id_ + " (" + rank_ + ")";
+        return "" + id_ + " (" + rank_ + ") (" + termfreq_ + ")";
     }
 
     public String toXMLString()
     {
-        return "<Match id=\"" + id_ + "\" rank=\"" + rank_ + "\" />";
+        return "<Match id=\"" + id_ + "\" rank=\"" + rank_ + "\" termfreq=\"" + termfreq_ + "\" />";
     }
     }