view website/issues/extensions/spambayes.py @ 5725:6923225fd781

Handle UnicodeDecodeError in file class when file contents are not text (e.g. jpg). This shows up in python3 due to conversion from bytes to string. This allows binary file retrieval via the rest interface to complete.
author John Rouillard <rouilj@ieee.org>
date Wed, 22 May 2019 20:56:59 -0400
parents e46ce04d5bbc
children
line wrap: on
line source

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

import 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 = xmlrpc_.client.ServerProxy(spambayes_uri, verbose=False)
    try:
        server.train({'content':content}, tokens, {}, is_spam)
        return (True, None)
    except (socket.error, xmlrpc_.client.Error) as 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 "trainspam" in self.form:
            is_spam = True
        elif "trainham" in self.form:
            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/