spam filter for publishing suggestions
[wolnelektury.git] / src / suggest / models.py
index 87454a2..b499ee8 100644 (file)
@@ -2,6 +2,9 @@
 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
+import re
+from datetime import timedelta
+
 from django.db import models
 from django.contrib.auth.models import User
 from django.utils.translation import ugettext_lazy as _
@@ -36,5 +39,35 @@ class PublishingSuggestion(models.Model):
         verbose_name = _('publishing suggestion')
         verbose_name_plural = _('publishing suggestions')
 
+    def is_spam(self):
+        suggestion_text = (self.books or self.audiobooks).strip(' \r\n,')
+        # similar = PublishingSuggestion.objects.filter(
+        #     books__in=('', suggestion_text), audiobooks__in=('', suggestion_text))
+        similar = PublishingSuggestion.objects.filter(books=self.books, audiobooks=self.audiobooks).exclude(pk=self.pk)
+        http = 'http' in suggestion_text
+        spam = False
+        if re.search(r'([^\W\d_])\1\1\1', suggestion_text):
+            # same letter repetition outside URL
+            spam = True
+        elif re.search(r'[^\W\d_]\d|\d[^\W\d_]', suggestion_text) and not http:
+            # string of letters and digits outside URL
+            spam = True
+        elif re.search(r'[^\W\d_]{17}', suggestion_text):
+            # long string of letters (usually gibberish)
+            spam = True
+        elif ' ' not in suggestion_text:
+            # single word - usually spam
+            spam = True
+        elif len(suggestion_text) < 11:
+            # too short
+            spam = True
+        elif similar.filter(created_at__range=(self.created_at - timedelta(1), self.created_at)):
+            # the same suggestion within 24h
+            spam = True
+        elif similar.filter(ip=self.ip):
+            # the same suggestion from the same IP
+            spam = True
+        return spam
+
     def __unicode__(self):
         return unicode(self.created_at)