view 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
line wrap: on
line source

import re, math
from roundup.cgi.actions import Action
from roundup.cgi.exceptions import *

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'))

    content = db.getclass(classname).get(nodeid, 'content')

    tokens = ["klass:%s" % classname,
              "author:%s" % authorid,
              "authorage:%d" % int(math.log(authorage)),
              "hasrev:%s" % (REVPAT.search(content) is not None)]

    return (content, tokens)

def train_spambayes(db, content, tokens, is_spam):
    spambayes_uri = db.config.detectors['SPAMBAYES_URI']

    server = xmlrpclib.ServerProxy(spambayes_uri, verbose=False)
    try:
        server.train({'content':content}, tokens, {}, is_spam)
        return (True, None)
    except (socket.error, xmlrpclib.Error), e:
        return (False, str(e))


class SpambayesClassify(Action):
    permissionType = 'SB: May Classify'
    
    def handle(self):
        (content, tokens) = extract_classinfo(self.db,
                                              self.classname, self.nodeid)

        if self.form.has_key("trainspam"):
            is_spam = True
        elif self.form.has_key("trainham"):
            is_spam = False

        (status, errmsg) = train_spambayes(self.db, content, tokens,
                                           is_spam)

        node = self.db.getnode(self.classname, self.nodeid)
        props = {}

        if status:
            if node.get('spambayes_misclassified', False):
                props['spambayes_misclassified'] = True

            props['spambayes_score'] = 1.0
            
            s = " SPAM"
            if not is_spam:
                props['spambayes_score'] = 0.0
                s = " HAM"
            self.client.add_ok_message(self._('Message classified as') + s)
        else:
            self.client.add_error_message(self._('Unable to classify message, got error:') + errmsg)

        klass = self.db.getclass(self.classname)
        klass.set(self.nodeid, **props)
        self.db.commit()

def sb_is_spam(obj):
    cutoff_score = float(obj._db.config.detectors['SPAMBAYES_SPAM_CUTOFF'])
    try:
        score = obj['spambayes_score']
    except KeyError:
        return False
    return score >= cutoff_score

def init(instance):
    instance.registerAction("spambayes_classify", SpambayesClassify)
    instance.registerUtil('sb_is_spam', sb_is_spam)
    

Roundup Issue Tracker: http://roundup-tracker.org/