def remove_all_tweets():
"""Empties the database"""
try:
db = next(get_db())
delete_all_tweets(db)
finally:
db.close()
def remove_tweets(ids: List[int]):
"""
Deletes the specified tweets from the database
Parameters:
ids (list): A list containing the ids of the tweets to be deleted
"""
for id in tqdm(ids):
try:
db = next(get_db())
delete_tweet(db, id)
finally:
db.close()
def new_tweets_only(tweet_df: pd.DataFrame) -> int:
"""
Puts tweets into the database. Gets rid of duplicates so that all new tweets can be added at
the same time saving significant amounts of time.
"""
tweets = [
TweetCreate(
text=tweet_df.iloc[i]["text"],
user=tweet_df.iloc[i]["user"],
search_query=tweet_df.iloc[i]["search_query"],
favorite_count=tweet_df.iloc[i]["favorite_count"],
retweet_count=tweet_df.iloc[i]["retweet_count"],
follower_count=tweet_df.iloc[i]["follower_count"],
media=tweet_df.iloc[i]["media"],
is_reply=tweet_df.iloc[i]["is_reply"],
is_retweet=tweet_df.iloc[i]["is_retweet"],
has_mentions=tweet_df.iloc[i]["has_mentions"],
is_funny=tweet_df.iloc[i]["is_funny"],
)
for i in range(len(tweet_df))
]
try:
# Gets a list containing the text of all tweets in the database
db = next(get_db())
tweet_text = [tweet.text for tweet in read_tweets(db, 0, 1000000)]
db.close()
# Creates a new list containing only the tweets not already in the database
new_tweets = [tweet for tweet in tweets if tweet.text not in tweet_text]
num_tweets = len(new_tweets)
db = next(get_db())
# Adds all of the new tweets as a batch
print("Tweets filtered. Posting to database...")
create_tweets(db, new_tweets)
except Exception as e:
print(e)
finally:
db.close()
return num_tweets # , tweet_text, new_tweets, tweet_df
def get_tweet(tweet_id: int) -> models.Tweet:
"""
Gets the tweet with the specified id from the database
Parameters:
tweet_id (int): the id of the tweet to query from the database
Returns:
(models.Tweet): A tweet object from the database with the specified id
"""
try:
db = next(get_db())
tweet = read_tweet(db, tweet_id)
finally:
db.close()
return tweet
def get_tweets(count: int = 100) -> List[models.Tweet]:
"""
Gets the first n=count tweets from the database
Parameters:
count (int): the number of tweets to query from the database
Returns:
(list): A list containing the first n=count tweets from the database
"""
try:
db = next(get_db())
tweets = read_tweets(db, limit=count)
finally:
db.close()
return tweets
def get_query() -> query.Query:
"""Returns a sqlalchemy.orm.query.Query object for custom queries"""
db = next(get_db())
return custom_query(db)