def is_tweet_valid(self, tweet):
if not tweet or "delete" in tweet:
logger.debug("Empty tweet - skipping")
return False
if not "lang" in tweet or tweet["lang"] != "en":
logger.debug("Non EN - skipping")
return False
if not "text" in tweet or tweet["text"].startswith("RT"):
logger.debug("RE-Tweet found - skipping")
return False
folded_text = TwitterMixin.word_map(tweet["text"]).split()
if "__h__" in folded_text and "__s__" in folded_text:
logger.debug("Tweet with double emoicons found - skipping")
return False
return True
def is_tweet_valid(self, tweet):
if not tweet or 'delete' in tweet:
logger.debug('Empty tweet - skipping')
return False
if not 'lang' in tweet or tweet['lang'] != 'en':
logger.debug('Non EN - skipping')
return False
if not 'text' in tweet or tweet['text'].startswith('RT'):
logger.debug('RE-Tweet found - skipping')
return False
folded_text = TwitterMixin.word_map(tweet['text']).split()
if '__h__' in folded_text and '__s__' in folded_text:
logger.debug('Tweet with double emoicons found - skipping')
return False
return True
def is_tweet_valid(self, tweet):
if not tweet or 'delete' in tweet:
logger.debug('Empty tweet - skipping')
return False
if not 'lang' in tweet or tweet['lang'] != 'en':
logger.debug('Non EN - skipping')
return False
if not 'text' in tweet or tweet['text'].startswith('RT'):
logger.debug('RE-Tweet found - skipping')
return False
folded_text = TwitterMixin.word_map(tweet['text']).split()
if '__h__' in folded_text and '__s__' in folded_text:
logger.debug('Tweet with double emoicons found - skipping')
return False
return True
def test_word_map(self):
text = "hello :) :( not xxx :)"
result = TwitterMixin.word_map(text).split()
expect = "hello __h__ __s__ __not__ xxx __h__".split()
self.assertEqual(expect, result)
def test_word_map(self):
text = "hello :) :( not xxx :)"
result = TwitterMixin.word_map(text).split()
expect = "hello __h__ __s__ __not__ xxx __h__".split()
self.assertEqual(expect, result)
