[Python-checkins] r68654 - in tracker/instances/spambayes_integration: detectors/spambayes.py extensions/spambayes.py

[martin.v.loewis]

Author: martin.v.loewis
Date: Sat Jan 17 12:36:35 2009
New Revision: 68654

Log:
Issue 215: add hasrev token.


Modified:
   tracker/instances/spambayes_integration/detectors/spambayes.py
   tracker/instances/spambayes_integration/extensions/spambayes.py

Modified: tracker/instances/spambayes_integration/detectors/spambayes.py
==============================================================================
--- tracker/instances/spambayes_integration/detectors/spambayes.py	(original)
+++ tracker/instances/spambayes_integration/detectors/spambayes.py	Sat Jan 17 12:36:35 2009
@@ -3,9 +3,12 @@
 import socket
 import time
 import math
+import re
 
 from roundup.exceptions import Reject
 
+REVPAT = re.compile(r'(r[0-9]+\b|rev(ision)? [0-9]+\b)')
+
 def extract_classinfo(db, klass, nodeid, newvalues):
     if None == nodeid:
         node = newvalues
@@ -28,7 +31,8 @@
 
     tokens = ["klass:%s" % klass.classname,
               "author:%s" % authorid,
-              "authorage:%d" % int(math.log(authorage))]
+              "authorage:%d" % int(math.log(authorage)),
+              "hasrev:%s" % REVPAT.search(content) is not None]
 
 
     return (content, tokens)

Modified: tracker/instances/spambayes_integration/extensions/spambayes.py
==============================================================================
--- tracker/instances/spambayes_integration/extensions/spambayes.py	(original)
+++ tracker/instances/spambayes_integration/extensions/spambayes.py	Sat Jan 17 12:36:35 2009
@@ -3,15 +3,20 @@
 
 import xmlrpclib, socket
 
+REVPAT = re.compile(r'(r[0-9]+\b|rev(ision)? [0-9]+\b)')
+
 def extract_classinfo(db, classname, nodeid):
     node = db.getnode(classname, nodeid)
 
     authorage = node['creation'].timestamp() - \
                 db.getnode('user', node.get('author', node.get('creator')))['creation'].timestamp()
 
+    authorid = node.get('author', node.get('creator')
+
     tokens = ["klass:%s" % classname,
-              "author:%s" % node.get('author', node.get('creator')),
-              "authorage:%d" % int(authorage)]
+              "author:%s" % authorid,
+              "authorage:%d" % int(math.log(authorage)),
+              "hasrev:%s" % REVPAT.search(content) is not None]
 
     klass = db.getclass(classname)
     return (klass.get(nodeid, 'content'), tokens)