Changeset 18464 for indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneIndexer.java

Timestamp:

2009-02-05T14:02:25+13:00 (15 years ago)

Author:

davidb

Message:

Support for reindexing documents

File:

• indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneIndexer.java (modified) (5 diffs)

indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneIndexer.java

@@ -294 +294 @@
     public void startDocument() throws SAXException 
     {
-        if (mode_.equals("index")) {
-        println("Starting to index " + file_id_);
-        print("[");
-        }
+        println("Starting to process " + file_id_);
+        print("[");
     }
 
     public void endDocument() throws SAXException 
     {
-        if (mode_.equals("index")) {
-        println("]");
-        println("... indexing finished.");
-        }
+        println("]");
+        println("... processing finished.");
     }
 
@@ -311 +307 @@
         throws SAXException 
     {
-        if (mode_.equals("index")) {
-        path_ = appendPathLink(path_, qName, atts);
+        path_ = appendPathLink(path_, qName, atts);
+        
+        if (qName.equals(doc_tag_level_)) {
+        mode_ = atts.getValue("gs2:mode");
+
+        pushOnStack(); // start new doc
+        current_node_ = qName;
         
-        if (qName.equals(doc_tag_level_)) {
-            pushOnStack(); // start new doc
-            current_node_ = qName;
-            
-            String node_id = atts.getValue("gs2:id");
-            print(" " + qName + ": " + node_id );
-            current_doc_.add(new Field("nodeID", node_id, Field.Store.YES, Field.Index.UN_TOKENIZED));
-            
-            current_doc_oid_ = atts.getValue("gs2:docOID");
-            current_doc_.add(new Field("docOID", current_doc_oid_, Field.Store.YES, Field.Index.UN_TOKENIZED));
-        }
+        String node_id = atts.getValue("gs2:id");
+        print(" " + qName + ": " + node_id + " (" + mode_ + ")" );
+        current_doc_.add(new Field("nodeID", node_id, Field.Store.YES, Field.Index.UN_TOKENIZED));
         
-        if (isIndexable(atts)) {
-            indexable_current_node_ = qName;    
-        }
-        else {
-            indexable_current_node_ = "";
-        }
+        current_doc_oid_ = atts.getValue("gs2:docOID");
+        current_doc_.add(new Field("docOID", current_doc_oid_, Field.Store.YES, Field.Index.UN_TOKENIZED));
+        }
+        
+        if (isIndexable(atts)) {
+        indexable_current_node_ = qName;    
+        }
+        else {
+        indexable_current_node_ = "";
         }
     }
@@ -350 +346 @@
     public void endElement(String uri, String localName, String qName) throws SAXException 
     {
-        if (mode_.equals("index")) {
+        if (mode_.equals("delete")) {
+        try {
+            deleteDocument(current_doc_oid_);
+        }
+        catch (java.io.IOException e) {
+            e.printStackTrace();
+        }
+        }
+        else if (mode_.equals("index") || mode_.equals("reindex")) {
         if (qName.equals(indexable_current_node_))
             {
@@ -367 +371 @@
         if (qName.equals(doc_tag_level_)) {
             try {
+            // perhaps this is more efficient if addDocument()
+            // used for "index" and updateDocument() for reindex
             writer_.updateDocument(new Term("docOID", current_doc_oid_), current_doc_, analyzer_);
             } 
@@ -376 +382 @@
         
         path_ = removePathLink(path_);
-        }
-        else if (mode_.equals("delete")) {
-        if (qName.equals("Delete")) {
-            try {
-            deleteDocument(current_contents_);
-            current_contents_ = "";
-            }
-            catch (java.io.IOException e) {
-            e.printStackTrace();
-            }
-        }
         }
     }