Changeset 16912

Timestamp:

2008-08-20 15:06:13 (5 months 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.

Files:

• indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneQuery.java (modified) (4 diffs)
• indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/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);
-term_docs.freq()
+0
-                int match_docs = 0;
-
-
-
-
+
+
+
+
+
-                        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);
-
+
+
-                }
-            }
@@ -202 +218 @@
-                // Output the matching documents
-                for (int i = start_results; (i <= hits.scoreDocs.length && i <= end_results); i++) {
-
-
+
+
+
+
-                }
-            }

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

@@ -110 +110 @@
-    }
-
-
+, int termfreq
-    {
-
+, termfreq
-    }
-    
@@ -225 +225 @@
-        public String id_ = "";
-        public float rank_ = 0;
-
-
+
+
+
-        {
-            id_ = id;
-            rank_ = rank;
+            termfreq_ = termfreq;
-        }
-
-        public String toString()
-        {
-
+ (" + termfreq_ + ")
-        }
-
-        public String toXMLString()
-        {
-
+termfreq=\"" + termfreq_ + "\" 
-        }
-    }