[Bug 990690] Filter EichGate keywords locally.

mozilla · Apr 7, 2014 · 607601e · 607601e
1 parent ffcc9e3
commit 607601e
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 9 deletions.
diff --git a/kitsune/customercare/cron.py b/kitsune/customercare/cron.py
@@ -31,6 +31,22 @@
 log = logging.getLogger('k.twitter')
 
 
+def get_word_blacklist_regex():
+    """
+    Make a regex that looks kind of like r'\b(foo|bar|baz)\b'.
+
+    This is a function so that it isn't calculated at import time,
+    and so can be tested more easily.
+
+    This doesn't use raw strings (r'') because the "mismatched" parens
+    were confusing my syntax highlighter, which was confusing me.
+    """
+    return re.compile(
+        '\\b(' +
+        '|'.join(map(re.escape, settings.CC_WORD_BLACKLIST)) +
+        ')\\b')
+
+
 @cronjobs.register
 def collect_tweets():
     # Don't (ab)use the twitter API from dev and stage.
@@ -45,15 +61,8 @@ def collect_tweets():
                     settings.TWITTER_ACCESS_TOKEN_SECRET)
 
         search_options = {
-            'q': ('firefox OR #fxinput OR @firefoxbrasil OR #firefoxos'
-              ' OR @firefox_es -brendan -eich -ceo -leadership -protest'
-              ' -marriage -boycott -boycottfirefox -opposition -political'
-              ' -gay -political -tolerance -speech -censor -censorship'
-              ' -resign -resignation -equal -equality -intolerant'
-              ' -intolerance -StandWithMozilla -StandWithFirefox -freedom'
-              '  -bigot -bigots -bigoted -liberal -hypocrite -hypocrites'
-              ' -hypocritical -harassment -beliefs -leftists -bullying'
-              ' -bully -homophobic -homophobia'),
+            'q': ('firefox OR #fxinput OR @firefoxbrasil OR #firefoxos '
+                  'OR @firefox_es'),
             'count': settings.CC_TWEETS_PERPAGE,  # Items per page.
             'result_type': 'recent',  # Retrieve tweets by date.
         }
@@ -172,6 +181,13 @@ def _filter_tweet(item, allow_links=False):
         statsd.incr('customercare.tweet.rejected.user')
         return None
 
+    # Exclude problem words
+    match = get_word_blacklist_regex().search(text)
+    if match:
+        bad_word = match.group(1)
+        statsd.incr('customercare.tweet.rejected.blacklist_word.' + bad_word)
+        return None
+
     return item
 
 

diff --git a/kitsune/customercare/tests/test_cron.py b/kitsune/customercare/tests/test_cron.py
@@ -101,6 +101,15 @@ def test_username_and_tweet_contain_firefox(self):
         self.tweet['text'] = 'My Firefox crashes :-( Any advice?'
         assert _filter_tweet(self.tweet) is not None
 
+    @override_settings(CC_WORD_BLACKLIST=['foo'])
+    def test_word_blacklist(self):
+        # Full words are blocked.
+        self.tweet['text'] = 'the word "foo" should be blocked.'
+        assert _filter_tweet(self.tweet) is None
+        # Substrings aren't blocked.
+        self.tweet['text'] = 'but "food" should not be blocked.'
+        assert _filter_tweet(self.tweet) is not None
+
 
 class GetOldestTweetTestCase(TestCase):
 

diff --git a/kitsune/settings.py b/kitsune/settings.py
@@ -763,6 +763,30 @@ def read_only_mode(env):
 CC_REPLIES_GOAL = 175  # Goal # of replies in 24 hours.
 CC_TWEETS_DAYS = 7  # Limit tweets to those from the last 7 days.
 CC_BANNED_USERS = ['lucasbytegenius']  # Twitter handles banned from AoA
+# If any of these words show up in a tweet, it probably isn't
+# actionable, so don't add it to the AoA.
+CC_WORD_BLACKLIST = [
+    '$1000',
+    'boycott',
+    '#boycottfirefox',
+    'brendan',
+    'ceo',
+    'civil',
+    'cupid',
+    'donation',
+    'eich',
+    'leadership',
+    'lgbt',
+    'marriage',
+    '#mozillagate',
+    'opposition',
+    'political',
+    'prop8',
+    'proposition',
+    'protest',
+    'rights',
+]
+
 
 TWITTER_COOKIE_SECURE = True
 TWITTER_CONSUMER_KEY = ''