def test_is_english(self):
"""
Test that the sentiment class can classify text as english.
"""
sentimentanalysis = Sentimentanalysis()
tweet = 'This is a dumb tweet'
score = sentimentanalysis.evaluatetweet(tweet)
self.assertTrue(score > 0.9)
def test_special_characters(self):
"""
Test that sentiment class will not break on special characters.
"""
sentimentanalysis = Sentimentanalysis()
strange_tweet = 'ÅÚÍÎÏ””ŒER""øøø…weqrwqerqw wwcrqwe lchrjqekwchr http://asf.adfasdfæ…ø圜åÅ'
sentimentanalysis.afinnsentiment(strange_tweet)
sentimentanalysis.labmtsentiment(strange_tweet)
# If we made it this far we passed the test
self.assertTrue(True)
def dump(method='afinn'):
"""
Do sentiment analysis on all tweets for all companies and
write the results to a file.
"""
sentimentanalysis = Sentimentanalysis()
sent_method = None
if method == 'afinn':
sent_method = sentimentanalysis.afinnsentiment
elif method == 'labmt':
sent_method = sentimentanalysis.labmtsentiment
for stock_symbol, company in settings.STOCK_SYMBOL_MAPPINGS.items():
print 'Company: ' + str(company)
data = tweets.tweets(stock_symbol)
sentiment = {}
filename = filename_for_company(method, company)
print 'Filename: ' + str(filename)
try:
sentiment = json.loads(open(filename).read())
except:
print 'No file data'
for key in data:
docs = data[key]
first = True
count = 0
for doc in docs:
count += 1
print 'Iterating %d of %d' % (count, len(docs))
tweetdate = datetime.datetime.fromtimestamp(doc['timestamp']).strftime('%Y-%m-%d %H:%M:%S').split(' ')[0]
if first:
sentiment[tweetdate] = 0
first = False
tweet = doc['tweet']
isenglish = sentimentanalysis.evaluatetweet(tweet)
if isenglish > 0.8:
sentval = sent_method(tweet)
if sentiment.get(tweetdate, 0) == 0:
sentiment[tweetdate] = sentval
else:
sentiment[tweetdate] = (sentiment.get(tweetdate, 0) + sentval)/2
with io.open(filename, 'wb') as outfile:
json.dump(sentiment, outfile)
print 'Saved file: ' + filename
