def __init__(self, tweet):
self.created_at = tweet["created_at"]
self.id_str = tweet["id_str"]
self.entities = Entities(tweet["entities"])
self.external_sources = {"instagram": False, "foursquare": False, "swarm": False, "path": False}
for url_data in self.entities.urls:
match_external_sources = re.findall('instagram|swarmapp|foursquare|path', url_data['expanded_url'], re.I)
if match_external_sources != []:
self.external_sources[match_external_sources[0]] = True
self.in_reply_to_user_id_str = tweet["in_reply_to_user_id_str"]
self.contributors = tweet["contributors"]
self.text = tweet["text"]
self.retweet_count = tweet["retweet_count"]
self.in_reply_to_status_id_str = tweet["in_reply_to_status_id_str"]
self.id = tweet["id"]
self.in_reply_to_user_id = tweet["in_reply_to_user_id"]
# Place
# ISSUES:
# 1 - Sometimes location checkin coordinates is not where the location really is.
# 2 - Sometimes a specific name, a city name or country.
# 3 - Match location names with coordinates?
# 4 - Check if location is within bbox of each city? (Belem in first case)
if tweet["place"] is not None:
self.place = TwitterPlace(tweet["place"])
coordinates = self.place.bounding_box["coordinates"]
# If coordinates for latitutde and longitude are the same, get those as the coordinates of the location.
if len(set([coordinates[0][0][1], coordinates[0][1][1], coordinates[0][2][1], coordinates[0][3][1]])) == 1 and len(set([coordinates[0][0][0], coordinates[0][1][0], coordinates[0][2][0], coordinates[0][3][0]])) == 1:
self.location_coordinates = (coordinates[0][0][1], coordinates[0][0][0])
# Else, take the mean of those as coordinates of a location (centroid).
else:
mean_lat = (coordinates[0][0][1] + coordinates[0][2][1]) / 2
mean_lon = (coordinates[0][0][0] + coordinates[0][1][0]) / 2
self.location_coordinates = (mean_lat, mean_lon)
# User
if tweet["user"] != None:
self.user = TwitterUser(tweet["user"])
self.in_reply_to_screen_name = tweet["in_reply_to_screen_name"]
# Treat source field
self.source = tweet["source"]
html = BeautifulSoup(self.source, "lxml")
tag = html.findAll("a")[0]
self.source = tag.contents[0]
def reply_to_user(self):
'''
Returns TwitterUser object if Status is a direct response to a user, otherwise returns None.
'''
if self.data['in_reply_to_user_id']:
for user_hash in self.mentions:
if user_hash['id'] == self.data['in_reply_to_user_id']:
user = TwitterUser(user_hash)
return user
def deserialize_user(raw_tweet):
"""
:param raw_tweet:
:return TwitterUser:
"""
user = TwitterUser()
full_tweet = json.loads(raw_tweet)
try:
user.user_id = full_tweet['user']['id']
user.user_name = full_tweet['user']['name']
user.screen_name = full_tweet['user']['screen_name']
user.location = full_tweet['user']['location']
user.followers = full_tweet['user']['followers_count']
except KeyError as ke:
log.warning('Could not map user: '.format(ke))
return user
def recipient(self):
'''Returns a TwitterUser object for the DM's receipient'''
recipient_hash = self.data['recipient']
recipient = TwitterUser(recipient_hash)
return recipient
def sender(self):
'''Returns a TwitterUser object for the DM's sender'''
sender_hash = self.data['sender']
sender = TwitterUser(sender_hash)
return sender
def recipient(self):
'''Returns a TwitterUser object for the user being followed'''
recipient_hash = self.data['target']
recipient = TwitterUser(recipient_hash)
return recipient
def sender(self):
'''Returns a TwitterUser object for the follower'''
sender_hash = self.data['source']
sender = TwitterUser(sender_hash)
return sender
def sender(self):
sender_hash = self.data['user']
sender = TwitterUser(sender_hash)
return sender