def schedule_task_with_lock(self, task):
"""Crawler scheduler filters tasks according to task type"""
if not task.get('enable'):
return None
task_queue = task.get('task_queue')
if task_queue not in self.task_queues:
return None
conn = get_redis_conn()
task_name = task.get('name')
interval = task.get('interval')
urls = task.get('resource')
lock_indentifier = acquire_lock(conn, task_name)
if not lock_indentifier:
return False
pipe = conn.pipeline(True)
try:
now = int(time.time())
pipe.hget(TIMER_RECORDER, task_name)
r = pipe.execute()[0]
if not r or (now - int(r.decode('utf-8'))) >= interval * 60:
pipe.lpush(task_queue, *urls)
pipe.hset(TIMER_RECORDER, task_name, now)
pipe.execute()
# scheduler_logger.info('crawler task {} has been stored into redis successfully'.format(task_name))
return True
else:
return None
finally:
release_lock(conn, task_name, lock_indentifier)
def schedule_task_with_lock(self, task):
"""Crawler scheduler filters tasks according to task type"""
if not task.get('enable'):
return None
task_queue = task.get('task_queue')
if task_queue not in self.task_queues:
return None
conn = get_redis_conn()
task_name = task.get('name')
internal = task.get('internal')
urls = task.get('resource')
lock_indentifier = acquire_lock(conn, task_name)
if not lock_indentifier:
return False
pipe = conn.pipeline(True)
try:
now = int(time.time())
pipe.hget(TIMER_RECORDER, task_name)
r = pipe.execute()[0]
if not r or (now - int(r.decode('utf-8'))) >= internal * 60:
pipe.lpush(task_queue, *urls)
pipe.hset(TIMER_RECORDER, task_name, now)
pipe.execute()
# scheduler_logger.info('crawler task {} has been stored into redis successfully'.format(task_name))
return True
else:
return None
finally:
release_lock(conn, task_name, lock_indentifier)
def get_lock(self, conn, task):
if not task.get('enable'):
return None
task_queue = task.get('task_queue')
if task_queue not in self.task_queues:
return None
task_name = task.get('name')
lock_indentifier = acquire_lock(conn, task_name)
return lock_indentifier
def get_lock(self, conn, task):
if not task.get('enable'):
return None
task_queue = task.get('task_queue')
if task_queue not in self.task_queues:
return None
task_name = task.get('name')
lock_indentifier = acquire_lock(conn, task_name)
return lock_indentifier
def schedule_task_with_lock(self, task):
"""Validator scheduler filters tasks according to task name
since it's task name stands for task type"""
if not task.get('enable'):
return None
task_queue = task.get('task_queue')
if task_queue not in self.task_queues:
return None
conn = get_redis_conn()
internal = task.get('internal')
task_name = task.get('name')
resource_queue = task.get('resource')
lock_indentifier = acquire_lock(conn, task_name)
if not lock_indentifier:
return False
pipe = conn.pipeline(True)
try:
now = int(time.time())
pipe.hget(TIMER_RECORDER, task_name)
pipe.zrevrangebyscore(resource_queue, '+inf', '-inf')
r, proxies = pipe.execute()
if not r or (now - int(r.decode('utf-8'))) >= internal * 60:
if not proxies:
print('fetched no proxies from task {}'.format(task_name))
return None
pipe.sadd(task_queue, *proxies)
pipe.hset(TIMER_RECORDER, task_name, now)
pipe.execute()
print(
'validator task {} has been stored into redis successfully'
.format(task_name))
return True
else:
return None
finally:
release_lock(conn, task_name, lock_indentifier)
def schedule_task_with_lock(self, task):
"""Validator scheduler filters tasks according to task name
since its task name stands for task type"""
if not task.get('enable'):
return None
task_queue = task.get('task_queue')
if task_queue not in self.task_queues:
return None
conn = get_redis_conn()
interval = task.get('interval')
task_name = task.get('name')
resource_queue = task.get('resource')
lock_indentifier = acquire_lock(conn, task_name)
if not lock_indentifier:
return False
pipe = conn.pipeline(True)
try:
now = int(time.time())
pipe.hget(TIMER_RECORDER, task_name)
pipe.zrevrangebyscore(resource_queue, '+inf', '-inf')
r, proxies = pipe.execute()
if not r or (now - int(r.decode('utf-8'))) >= interval * 60:
if not proxies:
# scheduler_logger.warning('fetched no proxies from task {}'.format(task_name))
print('fetched no proxies from task {}'.format(task_name))
return None
pipe.sadd(task_queue, *proxies)
pipe.hset(TIMER_RECORDER, task_name, now)
pipe.execute()
# scheduler_logger.info('validator task {} has been stored into redis successfully'.format(task_name))
return True
else:
return None
finally:
release_lock(conn, task_name, lock_indentifier)