Kombu 基础

本文地址：http://blog.csdn.net/spch2008/article/details/11530007

AMQP协议：

Producer产生消息，将消息赋予路由信息，发送给exchange。

queue存储消息，并将消息发送Consumer

bindings queues与exchange通过路由信息进行绑定。发送过来的消息带有路由信息，exchange提取路由信息和queues与exchange绑定的路由信息匹配，

                  匹配成功后，将消息发给相应队列。

consumer 从queues中提取信息，并消耗。

exchange  接收消息，并路由到相应队列。

Kombu代码示例：

producer.py

#!/usr/bin/python

from kombu.entity import Exchange
from kombu.messaging import Producer
from kombu.connection import Connection

connection = Connection('amqp://guest:[email protected]:5672//')
channel = connection.channel()

media_exchange = Exchange('media', 'direct', channel)
producer = Producer(channel, exchange=media_exchange, routing_key='video')

producer.publish({'name': '/tmp/lolcat1.avi', 'size': 1301013})

思路很简单：

1. 创建连接

2. 获得一个channel

3. 创建一个exchange，名字为‘media’，类型为‘direct’。

4. 创建一个Producer，指定exchange与routing_key，这样，发送信息的时候填充路由信息，发送到指定exchange。也可以在publish的时候指定。

5. 发送消息。

consumer.py

#!/usr/bin/python

from kombu.entity import Exchange, Queue
from kombu.messaging import Consumer
from kombu.connection import Connection



def process_media(body, message):
    print body
    message.ack()

connection = Connection('amqp://guest:[email protected]:5672//')
channel = connection.channel()

media_exchange = Exchange('media', 'direct', channel)
video_queue = Queue('video', exchange=media_exchange, routing_key='video', channel=channel)


consumer =  Consumer(channel, queues=[video_queue], callbacks=[process_media]) 
consumer.consume()

while True:
    connection.drain_events()
consumer.cancel()

思路很明确：

1. 创建连接

2. 获得channel

3. 创建exchange

4. 创建队列并与exchange绑定

5. 创建Consumer

6. consume() 向server注册，表明现在可以接收信息啦。server可以向该consumer发送消息。

7. drain_events阻塞程序，等待消息到来。当消息到来时，会执行回调函数process_media

8. cancel()通知server不要向该consumer发送任何消息啦。

疑惑：

1. channel是什么

    When producers and consumers connects to the broker using a TCP socket after authenticating the connection they establish a channel where AMQP commands are sent. The channel is a virtual path inside a TCP connection between this is very useful because there can be multiple channels inside the TCP connection each channels is identified using an unique ID. 

2. process_media的参数message是什么？为什么要调用message.ack()?

         打印message，输出kombu.transport.pyamqplib.Message object at 0x28d2b90。于是定位到Message类。     

class Message(base.Message):
    """A message received by the broker.
    .. attribute:: body  The message body.
    .. attribute:: channel The channel instance the message was received on.
    """

    def __init__(self, channel, msg, **kwargs):
        props = msg.properties
        super(Message, self).__init__(channel,
                body=msg.body,
                delivery_tag=msg.delivery_tag,
                content_type=props.get("content_type"),
                content_encoding=props.get("content_encoding"),
                delivery_info=msg.delivery_info,
                properties=msg.properties,
                headers=props.get("application_headers"),
                **kwargs)

    可以通过message取得想要的信息。其次，为什么要调用ack呢？

    如果队列要求回复确认信息，那么只有当用户返回一个确认信息，表明该message已经收到，message才会从队列中消失，否则，message一直存放在队列中。

       

3. drain_events做了些什么？

 kombu.transport.pyamqlib.Connection.drain_events

def drain_events(self, allowed_methods=None, timeout=None):
        """Wait for an event on any channel."""
        return self.wait_multi(self.channels.values(), timeout=timeout)

def wait_multi(self, channels, allowed_methods=None, timeout=None):
	"""Wait for an event on a channel."""
	chanmap = dict((chan.channel_id, chan) for chan in channels)
	chanid, method_sig, args, content = self._wait_multiple(
			chanmap.keys(), allowed_methods, timeout=timeout)

	channel = chanmap[chanid]

	amqp_method = self._method_override.get(method_sig) or \
					channel._METHOD_MAP.get(method_sig, None)
					
	return amqp_method(channel, args)

amqp_method即为回调函数。

ampqlib.client_0_8.channel.py.Channel.

 _METHOD_MAP = {
        (60, 60): _basic_deliver,
        (60, 71): _basic_get_ok,
        }

根据method_sig取得相应函数

def _basic_deliver(self, args, msg):
       
        consumer_tag = args.read_shortstr()
        func = self.callbacks.get(consumer_tag, None)
        if func is not None:
            func(msg)

从channel中的callback中获得用户注册的回调函数。明明在创建consumer的时候注册的回调函数，为什么回调函数会在channel中呢？

Comsumer类的comsume函数：

def consume(self, no_ack=None):
        
    self._basic_consume(T, no_ack=no_ack, nowait=False)
			
									
def _basic_consume(self, queue, consumer_tag=None,
            no_ack=no_ack, nowait=True):

    queue.consume(tag, self._receive_callback,
                  no_ack=no_ack, nowait=nowait)


def _receive_callback(self, message):

    self.receive(decoded, message)
            
                                
def receive(self, body, message):
    
    callbacks = self.callbacks
    if not callbacks:
        raise NotImplementedError("Consumer does not have any callback")
    [callback(body, message) for callback in callbacks]

注册的回调函数被包装了一下。新的回调函数传入了queue.consume

def consume(self, consumer_tag='', callback=None, no_ack=None,
            nowait=False):
        
    return self.channel.basic_consume(queue=self.name,
                                      no_ack=no_ack,
                                      consumer_tag=consumer_tag or '',
                                      callback=callback,
                                      nowait=nowait)

回调函数进入channel中。

总结一下：drain_events等待队列中的消息，当取得消息后，会调用回调函数进行相应处理。

4. 从connection获得Producer与Consumer

    有一个简便方法，可以直接从connection中获得Producer或者Consumer。

consumer = connection.Consumer(queues=[video_queue], callbacks=[process_media])

完……