start=time.time()
# User information:
names = []
screen_names = []
id_str = []
location = []
description = []
followers_count = []
friends_count = []
statuses_count = []
created_at = []
# Tweet information:
tweet_id = []
tweet_txt = []
is_retweet = []
mentions = []
hashtags = []
retweet_user = []
for i in Å_tweets:
# User information:
names.append(i._json["user"]["name"])
screen_names.append(i._json["user"]["screen_name"])
id_str.append(i._json["user"]["id"])
location.append(i._json["user"]["location"])
description.append(i._json["user"]["description"])
followers_count.append(i._json["user"]["followers_count"])
friends_count.append(i._json["user"]["friends_count"])
statuses_count.append(i._json["user"]["statuses_count"])
created_at.append(i._json["user"]["created_at"])
# Tweet information:
tweet_id.append(i._json["id"])
# Creating tweet_txt, mentions and hashtags depending on whether it is a retweet or not
m = []
htags = []
if "retweeted_status" in i._json.keys():
is_retweet.append(True)
tweet_txt.append(i._json["retweeted_status"]["full_text"])
retweet_user.append(i._json["retweeted_status"]["user"]["screen_name"])
# append mentions and hashtags from original post (in order to get all)
for mention in i._json["retweeted_status"]["entities"]["user_mentions"]:
m.append(mention["screen_name"])
for htag in i._json["retweeted_status"]["entities"]["hashtags"]:
htags.append(htag["text"])
else:
is_retweet.append(False)
retweet_user.append(np.nan)
tweet_txt.append(i._json["full_text"])
for mention in i._json["entities"]["user_mentions"]:
m.append(mention["screen_name"])
for htag in i._json["entities"]["hashtags"]:
htags.append(htag["text"])
mentions.append(m)
hashtags.append(htags)
twitter_dictionary = {
"name": names,
"screen_name": screen_names,
"id_str": id_str,
"location": location,
"description": description,
"followers_count": followers_count,
"friends_count": friends_count,
"statuses_count": statuses_count,
"created_at": created_at,
"tweet_id": tweet_id,
"is_retweet": is_retweet,
"tweet_txt": tweet_txt,
"mentions" : mentions,
"hashtags" : hashtags,
'retweet_user' : retweet_user
}
end=time.time()
print("Adventure done. Time taken =", end-start, "seconds")
ÅU_df = pd.DataFrame(twitter_dictionary)
ÅU_df.head()