def get_not_finished_session_tasks_count(scrap_session_id: str) -> int:
queries = [
'''SELECT COUNT(*) FROM twint_distributed_tasks.SearchTweetScrapTasks
WHERE scrap_session_id=%s AND finished IS NULL''',
'''SELECT COUNT(*) FROM twint_distributed_tasks.UserDetailsScrapTasks
WHERE scrap_session_id=%s AND finished IS NULL''',
'''SELECT COUNT(*) FROM twint_distributed_tasks.UserTweetScrapTasks
WHERE scrap_session_id=%s AND finished IS NULL''',
'''SELECT COUNT(*) FROM twint_distributed_tasks.UserFollowersScrapTasks
WHERE scrap_session_id=%s AND finished IS NULL''',
'''SELECT COUNT(*) FROM twint_distributed_tasks.UserFollowingScrapTasks
WHERE scrap_session_id=%s AND finished IS NULL''',
'''SELECT COUNT(*) FROM twint_distributed_tasks.UserFavoritesScrapTasks
WHERE scrap_session_id=%s AND finished IS NULL'''
]
return sum([
execute_sql_query(query, [scrap_session_id]).to_numpy()[0]
for query in queries
])
def get_all_tasks():
return execute_sql_query(
'''SELECT task_id, username, created, finished, queue_name, scrap_session_name
FROM twint_distributed_tasks.UserFavoritesScrapTasks t
JOIN twint_distributed_tasks.ScrapSession s ON t.scrap_session_id = s.scrap_session_id''',
[])
def get_all_by_username(username: str):
return execute_sql_query(
'SELECT * FROM twint_distributed_tasks.UserFavoritesScrapTasks WHERE username=%s',
[username])
def get_session_id(task_id: str) -> str:
return execute_sql_query(
'SELECT * FROM twint_distributed_tasks.UserFavoritesScrapTasks WHERE task_id=%s',
[task_id])['scrap_session_id'].to_numpy()[0]
def get_task_id_sub_task_id(sub_task_id: str) -> str:
return execute_sql_query(
'SELECT * FROM twint_distributed_tasks.UserTweetScrapSubTasks WHERE sub_task_id=%s',
[sub_task_id]
)['task_id'].to_numpy()[0]
def get_all_not_finished_sub_tasks_by_task_id(task_id: str):
return execute_sql_query(
'SELECT * FROM twint_distributed_tasks.UserTweetScrapSubTasks WHERE task_id=%s AND finished IS NULL',
[task_id])
def get_all_sessions():
return execute_sql_query(
'SELECT * FROM twint_distributed_tasks.ScrapSession', [])
def get_scrap_session_name_by_id(scrap_session_name: str) -> Optional[str]:
values = list(
execute_sql_query(
'SELECT * FROM twint_distributed_tasks.ScrapSession WHERE scrap_session_id=%s',
[scrap_session_name])['scrap_session_name'].to_numpy())
return values[0] if len(values) > 0 else None
def get_all_tasks():
return execute_sql_query(
'''SELECT task_id, phrase, since, until, language, created, finished, queue_name, scrap_session_name
FROM twint_distributed_tasks.SearchTweetScrapTasks t
JOIN twint_distributed_tasks.ScrapSession s ON t.scrap_session_id = s.scrap_session_id''',
[])
def get_all_tasks_by_username(phrase: str):
return execute_sql_query(
'SELECT * FROM twint_distributed_tasks.SearchTweetScrapTasks WHERE phrase=%s',
[phrase]
)
