# -*- coding: utf-8 -*- from __future__ import absolute_import from __future__ import with_statement import socket import threading import time from itertools import count from kombu.entity import Exchange, Queue from kombu.messaging import Consumer, Producer from celery import states from celery.backends.base import BaseDictBackend from celery.exceptions import TimeoutError class BacklogLimitExceeded(Exception): """Too much state history to fast-forward.""" def repair_uuid(s): # Historically the dashes in UUIDS are removed from AMQ entity names, # but there is no known reason to. Hopefully we'll be able to fix # this in v3.0. return "%s-%s-%s-%s-%s" % (s[:8], s[8:12], s[12:16], s[16:20], s[20:]) class AMQPBackend(BaseDictBackend): """Publishes results by sending messages.""" Exchange = Exchange Queue = Queue Consumer = Consumer Producer = Producer BacklogLimitExceeded = BacklogLimitExceeded def __init__(self, connection=None, exchange=None, exchange_type=None, persistent=None, serializer=None, auto_delete=True, expires=None, connection_max=None, **kwargs): super(AMQPBackend, self).__init__(**kwargs) conf = self.app.conf self._connection = connection self.queue_arguments = {} self.persistent = (conf.CELERY_RESULT_PERSISTENT if persistent is None else persistent) delivery_mode = persistent and "persistent" or "transient" exchange = exchange or conf.CELERY_RESULT_EXCHANGE exchange_type = exchange_type or conf.CELERY_RESULT_EXCHANGE_TYPE self.exchange = self.Exchange(name=exchange, type=exchange_type, delivery_mode=delivery_mode, durable=self.persistent, auto_delete=auto_delete) self.serializer = serializer or conf.CELERY_RESULT_SERIALIZER self.auto_delete = auto_delete self.expires = (conf.CELERY_AMQP_TASK_RESULT_EXPIRES if expires is None else expires) if self.expires is not None: self.expires = self.prepare_expires(self.expires) # x-expires requires RabbitMQ 2.1.0 or higher. self.queue_arguments["x-expires"] = int(self.expires * 1000) self.connection_max = (connection_max or conf.CELERY_AMQP_TASK_RESULT_CONNECTION_MAX) self.mutex = threading.Lock() def _create_binding(self, task_id): name = task_id.replace("-", "") return self.Queue(name=name, exchange=self.exchange, routing_key=name, durable=self.persistent, auto_delete=self.auto_delete, queue_arguments=self.queue_arguments) def _create_producer(self, task_id, channel): self._create_binding(task_id)(channel).declare() return self.Producer(channel, exchange=self.exchange, routing_key=task_id.replace("-", ""), serializer=self.serializer) def _create_consumer(self, bindings, channel): return self.Consumer(channel, bindings, no_ack=True) def _publish_result(self, connection, task_id, meta): # cache single channel if hasattr(connection, "_result_producer_chan") and \ connection._result_producer_chan is not None and \ connection._result_producer_chan.connection is not None: channel = connection._result_producer_chan else: channel = connection._result_producer_chan = connection.channel() self._create_producer(task_id, channel).publish(meta) def revive(self, channel): pass def _store_result(self, task_id, result, status, traceback=None, max_retries=20, interval_start=0, interval_step=1, interval_max=1): """Send task return value and status.""" with self.mutex: with self.app.pool.acquire(block=True) as conn: def errback(error, delay): conn._result_producer_chan = None print("Couldn't send result for %r: %r. Retry in %rs." % ( task_id, error, delay)) send = conn.ensure(self, self._publish_result, max_retries=max_retries, errback=errback, interval_start=interval_start, interval_step=interval_step, interval_max=interval_max) send(conn, task_id, {"task_id": task_id, "status": status, "result": self.encode_result(result, status), "traceback": traceback}) return result def get_task_meta(self, task_id, cache=True): return self.poll(task_id) def wait_for(self, task_id, timeout=None, cache=True, propagate=True, **kwargs): cached_meta = self._cache.get(task_id) if cache and cached_meta and \ cached_meta["status"] in states.READY_STATES: meta = cached_meta else: try: meta = self.consume(task_id, timeout=timeout) except socket.timeout: raise TimeoutError("The operation timed out.") state = meta["status"] if state == states.SUCCESS: return meta["result"] elif state in states.PROPAGATE_STATES: if propagate: raise self.exception_to_python(meta["result"]) return meta["result"] else: return self.wait_for(task_id, timeout, cache) def poll(self, task_id, backlog_limit=100): with self.app.pool.acquire_channel(block=True) as (_, channel): binding = self._create_binding(task_id)(channel) binding.declare() latest, acc = None, None for i in count(): # fast-forward latest, acc = acc, binding.get(no_ack=True) if not acc: break if i > backlog_limit: raise self.BacklogLimitExceeded(task_id) if latest: payload = self._cache[task_id] = latest.payload return payload elif task_id in self._cache: # use previously received state. return self._cache[task_id] return {"status": states.PENDING, "result": None} def drain_events(self, connection, consumer, timeout=None, now=time.time): wait = connection.drain_events results = {} def callback(meta, message): if meta["status"] in states.READY_STATES: uuid = repair_uuid(message.delivery_info["routing_key"]) results[uuid] = meta consumer.callbacks[:] = [callback] time_start = now() while 1: # Total time spent may exceed a single call to wait() if timeout and now() - time_start >= timeout: raise socket.timeout() wait(timeout=timeout) if results: # got event on the wanted channel. break self._cache.update(results) return results def consume(self, task_id, timeout=None): with self.app.pool.acquire_channel(block=True) as (conn, channel): binding = self._create_binding(task_id) with self._create_consumer(binding, channel) as consumer: return self.drain_events(conn, consumer, timeout).values()[0] def get_many(self, task_ids, timeout=None, **kwargs): with self.app.pool.acquire_channel(block=True) as (conn, channel): ids = set(task_ids) cached_ids = set() for task_id in ids: try: cached = self._cache[task_id] except KeyError: pass else: if cached["status"] in states.READY_STATES: yield task_id, cached cached_ids.add(task_id) ids ^= cached_ids bindings = [self._create_binding(task_id) for task_id in task_ids] with self._create_consumer(bindings, channel) as consumer: while ids: r = self.drain_events(conn, consumer, timeout) ids ^= set(r.keys()) for ready_id, ready_meta in r.iteritems(): yield ready_id, ready_meta def reload_task_result(self, task_id): raise NotImplementedError( "reload_task_result is not supported by this backend.") def reload_taskset_result(self, task_id): """Reload taskset result, even if it has been previously fetched.""" raise NotImplementedError( "reload_taskset_result is not supported by this backend.") def save_taskset(self, taskset_id, result): raise NotImplementedError( "save_taskset is not supported by this backend.") def restore_taskset(self, taskset_id, cache=True): raise NotImplementedError( "restore_taskset is not supported by this backend.") def delete_taskset(self, taskset_id): raise NotImplementedError( "delete_taskset is not supported by this backend.")