def process_friend_batch(user, page, api):
"""
Create User object for each friend in batches of 100.
Return batch as list of friend objects.
Parameters:
-----------
User object
A page of 100 friend ids (Twitter rate limit for user/lookup is 100 ids)
API object
Output:
-------
A list of 100 friend objects.
"""
batch = []
friend_objs = user.lookup_friends(f_ids=page)
for f in friend_objs:
friend = User(api, user_id=f.id, central_user=user.central_user)
friend.num_followers = f.followers_count
friend.screen_name = f.screen_name
batch.append(friend)
return batch
def display_friends(screen_name):
"""
For a given user, select top 50 most influential friends (by number of
followers) and send JSON back to the front end for visualization.
Parameters:
----------
A user's screen name.
Output:
------
JSON of friends' screen names, number of followers, and scores.
TODO: Refactor this route into 2 separate routes, in order to be able to
update the front end more frequently.
TODO: Add the searched user to friendlist, to be displayed in d3.
"""
api = connect_to_API()
# initialize the searched user as user object
user = User(api, user_id=screen_name, central_user=screen_name)
# will contain list of friend objects
friendlist = []
# store friends, followers and scores as objects inside "children" list
friend_scores = {"name": user.user_id, "children": []}
# unpickle classifier and vectorizer used for scoring timelines
with open(PATH_TO_CLASSIFIER, "rb") as f:
classifier = pickle.load(f)
with open(PATH_TO_VECTORIZER, "rb") as f:
vectorizer = pickle.load(f)
try:
friends_ids = user.get_friends_ids()
for page in user.paginate_friends(friends_ids, 100):
friends = process_friend_batch(user, page, api)
friendlist.extend(friends)
print friendlist
if len(friendlist) > MAX_NUM_FRIENDS:
friendlist = get_top_influencers(friendlist, MAX_NUM_FRIENDS)
for friend in friendlist:
timeline = friend.get_timeline(MAX_NUM_TWEETS)
friend.score = friend.score_user(timeline, vectorizer, classifier)
friend_scores["children"].append({"name": friend.screen_name,
"size": friend.num_followers,
"score": friend.score})
json_scores = json.dumps(friend_scores)
return json_scores
except tweepy.TweepError as e:
error_message = handle_error(e)
flash(error_message)
return redirect(url_for("index"))
"""
# initialize tweepy api object with auth, OAuth
TWITTER_API_KEY = os.environ.get("TWITTER_API_KEY")
TWITTER_SECRET_KEY = os.environ.get("TWITTER_SECRET_KEY")
TWITTER_ACCESS_TOKEN = os.environ.get("TWITTER_ACCESS_TOKEN")
TWITTER_SECRET_TOKEN = os.environ.get("TWITTER_SECRET_TOKEN")
auth = tweepy.OAuthHandler(TWITTER_API_KEY, TWITTER_SECRET_KEY, secure=True)
auth.set_access_token(TWITTER_ACCESS_TOKEN, TWITTER_SECRET_TOKEN)
api = tweepy.API(auth, cache=None, retry_count=2)
return api
api = connect_to_API()
user = User(api, "bookstein", "bookstein")
# get timeline
t0 = time.time()
timeline = user.get_timeline(20)
t1 = time.time()
print "TIME TO GET TIMELINE: ", t1 - t0
# get friends
t0 = time.time()
friends_ids = user.get_friends_ids()
t1 = time.time()
print "TIME TO GET FRIENDS:", t1 - t0
# paginate and lookup friends
t0 = time.time()