def search_and_save_users(self):
language = 'en'
#use the lines below if you want to target another language besides english
language = raw_input("Which language do you want to target 'es'espanish, 'en'english: ")
language = language.lower()
query = raw_input("Which words do you want to search? ")
max_tweets = int(input("How many tweets do you want to retrieve? "))
tweets = self.tw.search_tweets(query, language, max_tweets)
list_user=[]
list_hashtag=[]
list_tweets =[]
print('users found saved to the database: ')
for elem in tweets:
list_user.append({'user_id': elem.user.id, 'screen_name': elem.user.screen_name, 'description': elem.user.description})
list_hashtag.append({'user_of_hashtag': elem.user.id, 'hashtag_text': query})
list_tweets.append({'user_tweets': elem.user.id, 'tweet_message': elem.text, 'created_date': elem.created_at})
for itemr, itemh, itemt in zip(list_user, list_hashtag, list_tweets):
try:
#The following is the same as the short version Recruited.create(**itemr)
#a= Recruited.create(user_id = itemr['user_id'],screen_name = itemr['screen_name'],description=itemr['description'])
a = Recruited.create(**itemr)
b = Hashtag.create(**itemh)
c = Tweets.create(**itemt)
a.save()
b.save()
c.save()
except IntegrityError:
pass
return list_user
def search_and_save_users(self):
language = 'en'
#use the lines below if you want to target another language besides english
#language = input("Which language do you want to target 'es'espanish, 'en'english: ")
#language = language.lower()
query = input("Which words do you want to search ?")
max_tweets = int(input("How many tweets do you want to retrieve? "))
searched_tweets = [
status for status in tweepy.Cursor(
twitter.api.search, q=query, lang=language).items(max_tweets)
]
list_user = []
list_hashtag = []
list_tweets = []
print('users found saved to the database: ')
for elem in searched_tweets:
#print(elem.user.screen_name)
#print(elem.user.id)
#print(elem.user.description)
#print(elem.text)
#print(elem.created_at)
#print(query)
list_user.append({
'user_id': elem.user.id,
'screen_name': elem.user.screen_name,
'description': elem.user.description
})
list_hashtag.append({
'user_of_hashtag': elem.user.id,
'hashtag_text': query
})
list_tweets.append({
'user_tweets': elem.user.id,
'tweet_message': elem.text,
'created_date': elem.created_at
})
print(elem.user.id)
for itemr, itemh, itemt in zip(list_user, list_hashtag, list_tweets):
try:
#The following is the same as the short version Recruited.create(**itemr)
#a= Recruited.create(user_id = itemr['user_id'],screen_name = itemr['screen_name'],description=itemr['description'])
a = Recruited.create(**itemr)
b = Hashtag.create(**itemh)
c = Tweets.create(**itemt)
a.save()
b.save()
c.save()
except IntegrityError:
pass
return list_user
def retrieve_users(self, target_hashtag_idx, the_hashtags):
"""
#retrieve all users that have used the selected hashtags and that have not received tweets recently (last 24 hours)
:return: the list of selected users
"""
now = datetime.datetime.now()
now = datetime.datetime.now()
pop_user = []
final_list = []
print(target_hashtag_idx)
the_target_hashtag = the_hashtags[target_hashtag_idx - 1]
# this query gets the user_ids of those users that tweeted the selected hashtag
users = Hashtag.select(Hashtag.user_of_hashtag).where(
Hashtag.hashtag_text == the_target_hashtag)
users_list = []
for item in users:
users_list.append(item.user_of_hashtag_id)
print(users_list)
selected_user = []
# this query selects the datetime of the users that where retrieved with the last query (users who tweeted certain hashtag)
for item in users_list:
selected_user = SentDate.select(
SentDate.user_sent, SentDate.date_tweet_sent,
SentDate.tweet_sent_message).where(SentDate.user_sent == item)
# if the user has been sent a tweet in the last 24 hours skip it
print("selected users")
for item in selected_user:
print(item.user_sent_id)
for item in selected_user:
# print("printing selected user")
# print (item.date_tweet_sent)
# print(item.tweet_sent_message)
# print(item.user_sent_id)
date_retrieved = datetime.datetime.strptime(
item.date_tweet_sent, '%Y-%m-%d %H:%M:%S.%f')
print('sent', date_retrieved)
print('48 hours', now - datetime.timedelta(hours=48))
if (now - datetime.timedelta(hours=48)) < date_retrieved:
print('Not have passed 48 hours, cannot send tweet')
if item.user_sent_id not in pop_user:
pop_user.append(item.user_sent_id)
print("user to pop: ", pop_user)
# only keeps the users that who didn't receive a tweet in the last 48 hours
print('user_list', users_list)
for elem in pop_user:
users_list = [value for value in users_list if value != elem]
print("final list", users_list)
return users_list
def get_hashtags(self):
# Option 1: Retrieve the tweet to send from a previously filled table in the database
# Option 2: Create the tweet to send on the fly (easier to do)
"""
#retrieve all the hashtags from database
:return: the lis of the hashtags
"""
hashtag = Hashtag.select(Hashtag.hashtag_text).distinct()
hashtag_list = []
for item in hashtag:
hashtag_list.append(item.hashtag_text)
return hashtag_list
def get_hashtags(self):
# Option 1: Retrieve the tweet to send from a previously filled table in the database
# Option 2: Create the tweet to send on the fly (easier to do)
"""
#retrieve all the hashtags from database
:return: the lis of the hashtags
"""
hashtag = Hashtag.select(Hashtag.hashtag_text).distinct()
hashtag_list = []
for item in hashtag:
hashtag_list.append(item.hashtag_text)
return hashtag_list
def retrieve_users(self, target_hashtag_idx, the_hashtags):
"""
#retrieve all users that have used the selected hashtags and that have not received tweets recently (last 24 hours)
:return: the list of selected users
"""
now = datetime.datetime.now()
now = datetime.datetime.now()
pop_user = []
final_list = []
print(target_hashtag_idx)
the_target_hashtag = the_hashtags[target_hashtag_idx - 1]
# this query gets the user_ids of those users that tweeted the selected hashtag
users = Hashtag.select(Hashtag.user_of_hashtag).where(Hashtag.hashtag_text == the_target_hashtag)
users_list = []
for item in users:
users_list.append(item.user_of_hashtag_id)
print(users_list)
selected_user = []
# this query selects the datetime of the users that where retrieved with the last query (users who tweeted certain hashtag)
for item in users_list:
selected_user = SentDate.select(SentDate.user_sent, SentDate.date_tweet_sent, SentDate.tweet_sent_message).where(SentDate.user_sent == item)
# if the user has been sent a tweet in the last 24 hours skip it
print("selected users")
for item in selected_user:
print(item.user_sent_id)
for item in selected_user:
# print("printing selected user")
# print (item.date_tweet_sent)
# print(item.tweet_sent_message)
# print(item.user_sent_id)
date_retrieved = datetime.datetime.strptime(item.date_tweet_sent, '%Y-%m-%d %H:%M:%S.%f')
print('sent', date_retrieved)
print('48 hours', now - datetime.timedelta(hours=48))
if (now - datetime.timedelta(hours=48)) < date_retrieved:
print('Not have passed 48 hours, cannot send tweet')
if item.user_sent_id not in pop_user:
pop_user.append(item.user_sent_id)
print("user to pop: ", pop_user)
# only keeps the users that who didn't receive a tweet in the last 48 hours
print('user_list', users_list)
for elem in pop_user:
users_list = [value for value in users_list if value != elem]
print("final list", users_list)
return users_list