def _convert_row_to_post(self, row):
# [site, social_id, username_hash, comment_time, comment_tokens]
print("\rInsert post to DataFrame {0}/{1}".format(
self._current_row, len(self.posts_csv_df)),
end="")
self._current_row += 1
date = datetime.datetime.fromtimestamp(row['comment_time'])
post = Post()
claim_id = compute_author_guid_by_author_name(str(row['social_id']))
post.post_id = str(
compute_post_guid(row['site'] + str(claim_id),
row['username_hash'], date_to_str(date)))
post.content = str(row['comment_tokens'])
post.author = str(row['username_hash'])
post.author_guid = str(row['username_hash'])
post.domain = str(row['site'])
post.date = date
self._posts.append(post)
claim_tweet_connection = Claim_Tweet_Connection()
claim_tweet_connection.claim_id = str(claim_id)
claim_tweet_connection.post_id = str(post.post_id)
self._claim_tweet_connections.append(claim_tweet_connection)
if self._current_row % self._max_posts_without_save == 0:
self._save_posts_and_connections()
def fill_claim_post_connection_using_keywords(self):
self.build_word_post_dict_for_trec()
claim_tweets_connections = []
claims = self._db.get_claims()
claim_id_keywords_dict = self._db.get_claim_id_keywords_dict_by_connection_type(
'hill_climbing')
for claim in claims:
claim_keywords_list = claim_id_keywords_dict[claim.claim_id]
for claim_keywords in claim_keywords_list:
posts = self.get_posts_from_word_post_dict(
claim_keywords, claim)
claim_tweets_connections += [
Claim_Tweet_Connection(claim_id=claim.claim_id,
post_id=p.post_id) for p in posts
]
if len(claim_tweets_connections) > 10000:
self._db.add_claim_tweet_connections_fast(
claim_tweets_connections)
claim_tweets_connections = []
self._db.add_claim_tweet_connections_fast(claim_tweets_connections)
del claim_tweets_connections
def _add_claim_tweet_connection(self, claim_id, post_id):
connection = Claim_Tweet_Connection()
connection.claim_id = claim_id
connection.post_id = post_id
self._db.add_claim_connections([connection])
pass
def get_claim_tweet_connection(self, claim_id, post_id):
claim_tweet_connection = Claim_Tweet_Connection()
claim_tweet_connection.claim_id = claim_id
claim_tweet_connection.post_id = post_id
return claim_tweet_connection
def _create_claim_post_connection(self, original_claim_id, post_id):
claim_post_connection = Claim_Tweet_Connection()
claim_post_connection.claim_id = original_claim_id
claim_post_connection.post_id = post_id
return claim_post_connection
