Changeset 17736

Timestamp:

Oct 28, 2023, 11:29:42 PM (9 months ago)

Author:

Dirk Stöcker

Message:

remove usage of two intermediate classes for spambayes, create proper interface for admin.py, include the few spambayes classes we need directly from the inofficial incomplete python3 port at ​https://github.com/mpwillson/spambayes3

Location:

plugins/trunk/spam-filter

Files:

• setup.py (modified) (1 diff)
• spambayes (added)
• spambayes/Options.py (added)
• spambayes/OptionsClass.py (added)
• spambayes/__init__.py (added)
• spambayes/chi2.py (added)
• spambayes/classifier.py (added)
• spambayes/mboxutils.py (added)
• spambayes/safepickle.py (added)
• spambayes/tokenizer.py (added)
• tracspamfilter/admin.py (modified) (3 diffs)
• tracspamfilter/filters/bayes.py (modified) (9 diffs)
• tracspamfilter/templates/admin_bayes.html (modified) (1 diff)

plugins/trunk/spam-filter/setup.py

@@ -52 +52 @@
     extras_require={
         'dns': ['dnspython>=1.3.5'],
-        'spambayes': ['spambayes'],
+        'spambayes': ['spambayes'],
         'pillow': ['pillow>=2'],
         'account': ['TracAccountManager >= 0.4'],

plugins/trunk/spam-filter/tracspamfilter/admin.py

@@ -446 +446 @@
 
         bayes = BayesianFilterStrategy(self.env)
-        hammie = bayes._get_hammie()
         data = {}
 
@@ -459 +458 @@
                 data['content'] = bayes_content
                 try:
-                    data['score'] = hammie.score(bayes_content.encode('utf-8'))
+                    data['score'] = )
                 except Exception as e:
                     self.log.warning('Bayes test failed: %s', e, exc_info=True)
@@ -488 +487 @@
                 req.redirect(req.href.admin(cat, page))
         ratio = ''
-        nspam = hammie.bayes.nspam
-        nham = hammie.bayes.nham
+        nspam, nham = bayes.getnumbers()
         if nham and nspam:
             if nspam > nham:

plugins/trunk/spam-filter/tracspamfilter/filters/bayes.py

@@ -22 +22 @@
 from tracspamfilter.api import IFilterStrategy, N_
 
-from spambayes.hammie import Hammie
-from spambayes.storage import SQLClassifier
-
+from spambayes.classifier import Classifier
+from spambayes.tokenizer import tokenize
 
 class BayesianFilterStrategy(Component):
@@ -51 +50 @@
 
     def test(self, req, author, content, ip):
-        hammie = self._get_hammie()
-        nspam = hammie.bayes.nspam
-        nham = hammie.bayes.nham
+        ()
+        nspam = bayes.nspam
+        nham = bayes.nham
         if author is not None:
             testcontent = author + '\n' + content
@@ -71 +70 @@
                           "large, results may be bad.")
 
-        score = hammie.score(testcontent)
+        score = )
         self.log.debug("SpamBayes reported spam probability of %s", score)
         points = -int(round(self.karma_points * (score * 2 - 1)))
@@ -79 +78 @@
                     ("%3.2f" % (score * 100)))
 
+
+
+
+
+
+
+
+
     def train(self, req, author, content, ip, spam=True):
-        # Split tokens at null characters by tokenizer in spambayes.hammie
+        # Split tokens at null characters by tokenizer
         content = content.replace('\x00', ' ')
         if author is not None:
@@ -89 +96 @@
                       spam and "spam" or "ham")
 
-        hammie = self._get_hammie()
-        hammie.train(testcontent, spam)
-        hammie.store()
+        ()
+        , spam)
+        .store()
         return 1
 
@@ -102 +109 @@
     # Internal methods
 
-    def _get_hammie(self):
-        try:  # 1.0
-            return Hammie(TracDbClassifier(self.env, self.log))
-        except TypeError:  # 1.1
-            return Hammie(TracDbClassifier(self.env, self.log), 'c')
-
-    def _get_numbers(self):
-        hammie = self._get_hammie()
-        return hammie.nspam, hammie.nham
+    def _get_bayes(self):
+        return TracDbClassifier(self.env, self.log)
 
     # used by admin panel
@@ -140 +140 @@
 
 
-class TracDbClassifier(SQLClassifier):
-    # FIXME: This thing is incredibly slow
-
+class TracDbClassifier(Classifier):
     def __init__(self, env_db, log):
+
         self.env_db = env_db
         self.log = log
-        self.nham = None
-        self.nspam = None
-        SQLClassifier.__init__(self, 'Trac')
+        self.statekey = 'saved state'
+        self.load()
 
     def load(self):
@@ -157 +155 @@
         else:  # new database
             self.nspam = self.nham = 0
+
+
+
 
     def _sanitize(self, text):
@@ -224 +225 @@
             words.append(word)
         return words
+
+
+
+
+
+
+
+
+
+

plugins/trunk/spam-filter/tracspamfilter/templates/admin_bayes.html

@@ -119 +119 @@
           <strong>${_("Error: %(error)s") % {'error': error}}</strong>
           # else:
-          <strong>${_("Score: %(score)s%%") % {'score': round(score * 100, 2)}}</strong>
+          <strong>${_("Score: %(score)s%%") % {'score': }}</strong>
           # endif
         </div>