[Zope-Checkins] CVS: Zope/lib/python/Products/PluginIndexes/TextIndexNG - interactiveDemo.py:1.1.2.3

Andreas Jung andreas@digicool.com
Thu, 14 Feb 2002 17:56:44 -0500


Update of /cvs-repository/Zope/lib/python/Products/PluginIndexes/TextIndexNG
In directory cvs.zope.org:/tmp/cvs-serv10759

Modified Files:
      Tag: ajung-textindexng-branch
	interactiveDemo.py 
Log Message:
- some code cleanup
- added --old option to run script against old TextIndex for 
  performances measurements


=== Zope/lib/python/Products/PluginIndexes/TextIndexNG/interactiveDemo.py 1.1.2.2 => 1.1.2.3 ===
 
 
-def index_directory(dirname, verbose, timed):
+def index_directory(dirname, verbose, timed, old):
 
     if not dirname: raise RuntimeError,'no directory name'
 
@@ -35,29 +35,29 @@
     ex.useStemmer           = None
     ex.useOperator          = 'and'
     ex.lexicon              = None
-    ex.useGlobbing          = 0
+    ex.useGlobbing          = 1
     ex.nearDistance         = 5
     ex.useSimilarity        = 0
-    ex.stopWords = None
+    ex.stopWords            = 'default_en.txt'
+#    ex.characterMapping     = [('a','b')]
 
     CAT = Catalog.Catalog("cat")
     CAT.aq_parent = TO('aq_parent')
 
-    TI = TextIndexNG.TextIndexNG('text',ex, caller = CAT)
-    if verbose: TI.debugOn()
-    else:       TI.debugOff()
-    TI.timed_statistics = timed
+    if old:
+        TI = TextIndex.TextIndex('text',ex, caller = CAT)
+    else:
+        TI = TextIndexNG.TextIndexNG('text',ex, caller = CAT, verbose=verbose)
+
+        if verbose: TI.debugOn()
+        else:       TI.debugOff()
+        TI.timed_statistics = timed
+
 
     CAT.addIndex('text',TI)
     CAT.addColumn('text')
     CAT.addColumn('path')
 
-    t1 = TO ('this text is a text')
-    t2 = TO ('the quick brown fox jumps over the lazy dog because the dog is quick and jumps quick') 
-
-    CAT.catalogObject(t1 , 't1')
-    CAT.catalogObject(t1 , 't2')
-
     files = os.listdir(dirname)
     files.sort()
 
@@ -76,8 +76,10 @@
         T = TO(data,fname)
         CAT.catalogObject(T,fname)
 
+    diff = time.time() - ts
     print "%d files, total size: %d" % (len(files), bytes)
-    print "Indexing time: %5.3lf" % (time.time() - ts)
+    print "Indexing time: %5.3lf" % diff
+    print 'Indexingspeed: %5.3lf KB/sec' % (1.0*bytes/diff/1024.0)
     for x in dir(ex):
         print "%25s = %s" % (x,getattr(ex,x))
 
@@ -107,19 +109,21 @@
 if __name__== '__main__':
 
     opts,args = getopt.getopt(sys.argv[1:],'hd:',['help','directory=',\
-            'verbose','timed'])
+            'verbose','timed','old'])
 
     directory = None
     verbose = 0
     timed   = 0 
+    old     = 0
 
     for k,v in opts:
         if k in ['-h','--help']:        usage(); sys.exit(1)
         if k in ['-d','--directory']:   directory = v
         if k in ['--verbose']:          verbose = 1
         if k in ['--timed']:            timed = 1
+        if k in ['--old']:              old = 1
 
     
-    cat = index_directory(directory, verbose, timed)
+    cat = index_directory(directory, verbose, timed, old)
     interactive_mode(cat)