Changeset 27359 for main/trunk/greenstone2/common-src

Timestamp:

2013-05-16T15:33:15+12:00 (11 years ago)

Author:

kjdon

Message:

sort fields are now separate from index fields. index fields will be like <TI index=1> and sort fields will be like <byTI index=1 tokenize=0>

File:

• main/trunk/greenstone2/common-src/indexers/lucene-gs/src/org/greenstone/LuceneWrapper3/GS2LuceneIndexer.java (modified) (4 diffs)

main/trunk/greenstone2/common-src/indexers/lucene-gs/src/org/greenstone/LuceneWrapper3/GS2LuceneIndexer.java

@@ -188 +188 @@
     String current_doc_oid_ = "";
     String indexable_current_node_ = "";
+      boolean tokenize = true;
     String current_contents_ = "";
 
@@ -330 +331 @@
         if (isIndexable(atts)) {
         indexable_current_node_ = qName;    
+        if (isTokenized(atts)) {
+          tokenize = true;
+        } else {
+          tokenize = false;
+        }
         }
         else {
@@ -335 +341 @@
         }
     }
+
+      public static boolean isTokenized(Attributes atts) {
+    boolean tokenize = true;
+    String tok = atts.getValue("tokenize");
+    if (tok!=null && tok.equals("0")) {
+      tokenize = false;
+    }
+    return tokenize;
+      }
 
     public static boolean isIndexable(Attributes atts) 
@@ -362 +377 @@
         if (qName.equals(indexable_current_node_))
             {
+              if (tokenize) {
             current_doc_.add(new Field(qName, current_contents_, Field.Store.NO, Field.Index.ANALYZED, Field.TermVector.YES));
-            // The byXX fields are used for sorting search results
-            // We don't want to do that for Text or AllFields fields
-            // They need to be untokenised for sorting
-            if (!qName.equals("TX") && !qName.equals("ZZ"))
-                {
-                current_doc_.add(new Field("by" + qName, current_contents_, Field.Store.NO, Field.Index.NOT_ANALYZED, Field.TermVector.NO));
-                }
+              } else {
+            current_doc_.add(new Field(qName, current_contents_, Field.Store.NO, Field.Index.NOT_ANALYZED, Field.TermVector.NO));
+              }
+            // // The byXX fields are used for sorting search results
+            // // We don't want to do that for Text or AllFields fields
+            // // They need to be untokenised for sorting
+            // if (!qName.equals("TX") && !qName.equals("ZZ"))
+            //     {
+            //  current_doc_.add(new Field("by" + qName, current_contents_, Field.Store.NO, Field.Index.NOT_ANALYZED, Field.TermVector.NO));
+            //     }
             
             current_contents_ = "";