[Zope-Checkins] CVS: Zope/lib/python/Products/PluginIndexes/TextIndexNG - TextIndexNG.py:1.2.2.30

[Andreas Jung]

Update of /cvs-repository/Zope/lib/python/Products/PluginIndexes/TextIndexNG
In directory cvs.zope.org:/tmp/cvs-serv8996

Modified Files:
      Tag: ajung-textindexng-branch
	TextIndexNG.py 
Log Message:
added support for exact phrase searches


=== Zope/lib/python/Products/PluginIndexes/TextIndexNG/TextIndexNG.py 1.2.2.29 => 1.2.2.30 ===
         for docId in res[0]:
             bucket[ docId ] = 1 
-
-        print bucket
+        
         return res[0], (self.id, )
 
 
@@ -566,6 +565,7 @@
         txI = self.txIntersection
         txU = self.txUnion
         txN = self.txNear
+        txQ = self.txQuote
 
         debug('Query: ',q)
 
@@ -630,7 +630,6 @@
         if not self.useSimilarity:
             raise TextIndexNGException, 'Similarity search is not enabled'
 
-
         # Lookup list of wordIds (usually should contain only *one*)
         wids = self._PROX_LEX.get(word)
         debug("\tWids: ", wids)
@@ -692,10 +691,21 @@
         debug("\treturn: ",r)
 
         return r
-        
 
-    def txNear(self, *sets):
+
+    def txQuote(self, *sets):
+        """ quote is a special case of near search where the near-
+            distance is one and we only search to the right
+        """
+        return apply(self.txNear,sets,{'distance':1,'bidirectional':1}) 
+
+
+    def txNear(self, *sets, **kw):
         """ perform near search on results sets """
+
+
+        distance = kw.get('distance',self.nearDistance)
+        bidirectional = kw.get('bidirectional',1)
         
         # One resultset consists of an IISet() or documentIds and 
         # tuple whose first element is the word (from LexiconLookup())
@@ -707,6 +717,8 @@
         r = self.txIntersection(*sets)
         docIds = r.docIds()
 
+        debug('txNear (%s)' % str(kw))
+
         # Now we determine for every document the positions of all
         # the words inside the document. then we compare all the positions
         # to determine neighbourship
@@ -718,12 +730,13 @@
 
         for docId in docIds:
 
+            # the posMap is a list of tuples(word,IISet[positions])
             posMap = self.positionsFromDocumentLookup(docId, words)
 
-            debug("Position Map for NEAR:")
-            for item in posMap: debug("\t%s" % str(item))
+            debug("\tPosition Map for NEAR:")
+            for item in posMap: debug("\t\t%s" % str(item))
 
-            if len(self.checkPositionMap(posMap,self.nearDistance)) > 0:
+            if len(self.checkPositionMap(posMap, distance, bidirectional)) > 0:
                 res_docIds.insert(docId)
 
         r = ResultSet( res_docIds,  words)       
@@ -732,7 +745,7 @@
         return r
 
         
-    def checkPositionMap(self, posMap, nearDistance):
+    def checkPositionMap(self, posMap, nearDistance, bidirectional=1):
         """ check if a PositionMap represents a valid match for
             a near search
         """
@@ -762,7 +775,7 @@
 
             num = 0        
             for word,posLst in posMap:
-                keys = posLst.keys( pos-nearDistance, pos+nearDistance)
+                keys = posLst.keys( pos-bidirectional*nearDistance, pos+nearDistance)
 
                 if len(keys)>0: num+=1