def parse_twitter(raw):
MAX_TWEETS = 30
count = 0
response = []
for tweet in raw['statuses']:
try:
# only return <30 english tweets with no sensitive content
if (tweet['lang'] == 'en') and \
count < MAX_TWEETS and tweet['possibly_sensitive'] == False:
count += 1
selection = []
selection.append({'date': tweet['created_at']})
text = tweet['text']
if sensitive_text(text) == True:
# sensitive word -> exclude
continue
else:
selection.append({'text': text})
hashtags = []
for tag in tweet['entities']['hashtags']:
hashtags.append(tag['text'])
selection.append({'hashtags': hashtags})
try:
image_url = tweet['entities']['media'][0]['media_url']
selection.append({'image_url': image_url})
except:
pass # no image
response.append(selection)
except:
pass #possible sensitive content
return response
def test_sensitive_text():
input_text = "Car, is; better*than\nugly#car#bus1"
assert sensitive_text(input_text) == False # "test_sensitive_text test 1"
input_text = "sadfsdaf, is; betsdfasfter*than\sex#car#bus2"
assert sensitive_text(input_text) == True # "test_sensitive_text test 2"
input_text = "asdfasdf, is; bettasdfer*than.ass#car#bus3"
assert sensitive_text(input_text) == True # "test_sensitive_text test 3"
input_text = "asdfdfs, is; betasdfter*than.any#ass#bus4"
assert sensitive_text(input_text) == True # "test_sensitive_text test 4"
input_text = "sfasfsdfdsaf, is; besdftter*than.any# ass#bus5"
assert sensitive_text(input_text) == True # "test_sensitive_text test 5"
input_text = "Be, is; better*thsfan.any# ass #bus6"
assert sensitive_text(input_text) == True # "test_sensitive_text test 6"
input_text = "Basdf, is; ?ass?bettsadfer*than.any# ok #bus7"
assert sensitive_text(input_text) == True # "test_sensitive_text test 7"
input_text = "Basdfsdaf, is; @ass?besafdtter*than.any# ok #bus8"
assert sensitive_text(input_text) == True # "test_sensitive_text test 8"
input_text = "# #"
assert sensitive_text(input_text) == False # "test_sensitive_text test 9"
input_text = "Basdfsdaf, is; @mas?besafdtter*massage.any# ok #bus8"
assert sensitive_text(input_text) == True # "test_sensitive_text test 8"
def test_sensitive_text():
input_text = 'Car, is; better*than\nugly#car#bus1'
assert(sensitive_text(input_text) == False) # "test_sensitive_text test 1"
input_text = 'sadfsdaf, is; betsdfasfter*than\sex#car#bus2'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 2"
input_text = 'asdfasdf, is; bettasdfer*than.ass#car#bus3'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 3"
input_text = 'asdfdfs, is; betasdfter*than.any#ass#bus4'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 4"
input_text = 'sfasfsdfdsaf, is; besdftter*than.any# ass#bus5'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 5"
input_text = 'Be, is; better*thsfan.any# ass #bus6'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 6"
input_text = 'Basdf, is; ?ass?bettsadfer*than.any# ok #bus7'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 7"
input_text = 'Basdfsdaf, is; @ass?besafdtter*than.any# ok #bus8'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 8"
input_text = '# #'
assert(sensitive_text(input_text) == False) # "test_sensitive_text test 9"
input_text = 'Basdfsdaf, is; @mas?besafdtter*massage.any# ok #bus8'
assert(sensitive_text(input_text) == True) # "test_sensitive_text test 8"