Mercurial > p > roundup > code
annotate website/issues/extensions/spambayes.py @ 5108:67fad01d2009
issue2550653: xapian search, stemming is not working
This is a partial fix for the issue. It does make stemming work
(so searching for silent will also return docs with silently in
them). However to do this we need to lowercase the text so the
porter stemmer will work. This means capitalization is not
preserved.
Tests in test/test_indexer for xapian backend all pass.
David Wolever (wolever) did the work.
| author | John Rouillard <rouilj@ieee.org> |
|---|---|
| date | Mon, 27 Jun 2016 22:10:45 -0400 |
| parents | ca692423e401 |
| children | 198b6e810c67 |
| rev | line source |
|---|---|
|
4024
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
1 import re, math |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
2 from roundup.cgi.actions import Action |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
3 from roundup.cgi.exceptions import * |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
4 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
5 import xmlrpclib, socket |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
6 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
7 REVPAT = re.compile(r'(r[0-9]+\b|rev(ision)? [0-9]+\b)') |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
8 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
9 def extract_classinfo(db, classname, nodeid): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
10 node = db.getnode(classname, nodeid) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
11 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
12 authorage = node['creation'].timestamp() - \ |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
13 db.getnode('user', node.get('author', node.get('creator')))['creation'].timestamp() |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
14 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
15 authorid = node.get('author', node.get('creator')) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
16 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
17 content = db.getclass(classname).get(nodeid, 'content') |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
18 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
19 tokens = ["klass:%s" % classname, |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
20 "author:%s" % authorid, |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
21 "authorage:%d" % int(math.log(authorage)), |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
22 "hasrev:%s" % (REVPAT.search(content) is not None)] |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
23 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
24 return (content, tokens) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
25 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
26 def train_spambayes(db, content, tokens, is_spam): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
27 spambayes_uri = db.config.detectors['SPAMBAYES_URI'] |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
28 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
29 server = xmlrpclib.ServerProxy(spambayes_uri, verbose=False) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
30 try: |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
31 server.train({'content':content}, tokens, {}, is_spam) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
32 return (True, None) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
33 except (socket.error, xmlrpclib.Error), e: |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
34 return (False, str(e)) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
35 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
36 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
37 class SpambayesClassify(Action): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
38 permissionType = 'SB: May Classify' |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
39 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
40 def handle(self): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
41 (content, tokens) = extract_classinfo(self.db, |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
42 self.classname, self.nodeid) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
43 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
44 if self.form.has_key("trainspam"): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
45 is_spam = True |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
46 elif self.form.has_key("trainham"): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
47 is_spam = False |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
48 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
49 (status, errmsg) = train_spambayes(self.db, content, tokens, |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
50 is_spam) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
51 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
52 node = self.db.getnode(self.classname, self.nodeid) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
53 props = {} |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
54 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
55 if status: |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
56 if node.get('spambayes_misclassified', False): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
57 props['spambayes_misclassified'] = True |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
58 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
59 props['spambayes_score'] = 1.0 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
60 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
61 s = " SPAM" |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
62 if not is_spam: |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
63 props['spambayes_score'] = 0.0 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
64 s = " HAM" |
|
4880
ca692423e401
Different approach to fix XSS in issue2550817
Ralf Schlatterbeck <rsc@runtux.com>
parents:
4024
diff
changeset
|
65 self.client.add_ok_message(self._('Message classified as') + s) |
|
4024
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
66 else: |
|
4880
ca692423e401
Different approach to fix XSS in issue2550817
Ralf Schlatterbeck <rsc@runtux.com>
parents:
4024
diff
changeset
|
67 self.client.add_error_message(self._('Unable to classify message, got error:') + errmsg) |
|
4024
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
68 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
69 klass = self.db.getclass(self.classname) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
70 klass.set(self.nodeid, **props) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
71 self.db.commit() |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
72 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
73 def sb_is_spam(obj): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
74 cutoff_score = float(obj._db.config.detectors['SPAMBAYES_SPAM_CUTOFF']) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
75 try: |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
76 score = obj['spambayes_score'] |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
77 except KeyError: |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
78 return False |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
79 return score >= cutoff_score |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
80 |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
81 def init(instance): |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
82 instance.registerAction("spambayes_classify", SpambayesClassify) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
83 instance.registerUtil('sb_is_spam', sb_is_spam) |
|
c2d0d3e9099d
svn repository setup
Stefan Seefeld <stefan@users.sourceforge.net>
parents:
diff
changeset
|
84 |
