RabbitMQ中交换机的消息分发机制

RabbitMQ是一个消息代理，它接受和转发消息，是一个由 Erlang 语言开发的遵循AMQP协议的开源实现。在RabbitMQ中生产者不会将消息直接发送到队列当中，而是将消息直接发送到交换机(exchange)，交换机用来接受生产者发送的消息并将这些消息发送给绑定的队列，即：生产者-->交换机-->队列。

在RabbitMQ中最主要的三种交换机：1. fanout(广播交换机)  2. direct(直连交换机)  3. topic(话题交换机)

1. fanout(广播交换机)

 fanout会将接受到的所有消息广播到它所绑定的所有队列当中(每个消费者都会收到所有的消息)，对于广播交换机，消息路由键routing_key和队列绑定键routing_key的作用都会被忽略。

fanout生产者：

 

import pika


class RabbitProducer(object):
    """
    与RabbitMq服务器建立连接
    """

    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 声明一个exchange交换机,交换机的类型为fanout广播.
        self.channel.exchange_declare(
            exchange='fanout_exchange', exchange_type='fanout', durable=True
        )

    def send_msg(self, message):
        """
        routing_key:绑定的key
        :param message:
        :return:
        """
        self.channel.basic_publish(
            exchange='fanout_exchange',
            routing_key='',  # 因为exchange的类型为fanout,所以routing_key的数值在这里将被忽略
            body=message,
            properties=pika.BasicProperties(
                delivery_mode=2,
                # 消息进行持久化(防止服务器挂掉.)===> 如果没有queue绑定到这个exchange交换机,这个参数是没有的.
            ))

    def close(self):
        self.conn.close()


if __name__ == "__main__":
    rabbit_producer = RabbitProducer()
    for i in range(10):
        message = 'hello world {}!'.format(i)
        rabbit_producer.send_msg(message)

 

消费者consumer1：

 

import pika
import uuid


class RabbitConsumer(object):
    """
    fanout 消费者1
    """

    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 声明一个队列queue_consumer1,并进行持久化(防止服务器挂掉),exclusive设置为false
        self.channel.queue_declare(
            exclusive=False, durable=True, queue='queue_consumer1'
        )

        # 声明一个exhange交换机,其类型为fanout广播类型  与生产者的交换机一致
        self.channel.exchange_declare(
            exchange='fanout_exchange', exchange_type='fanout', durable=True
        )

        # 将队列queue_consumer1与该exchange交换机进行绑定
        self.channel.queue_bind(exchange='fanout_exchange', queue='queue_consumer1')

    def call_back(self, method, body):
        """
        消费者对消息进行确认,防止消费者挂掉.
        :param method:
        :param body:
        :return:
        """
        self.channel.basic_ack(delivery_tag=method.delivery_tag)
        print('接收到的消息为:{}'.format(str(body)))

    def receive_msg(self):
        print('consumer1开始接受消息...')
        # 当上一条消息未确认时，会告知RabbitMQ不要再发送消息给这个消费者了 可以控制流量
        self.channel.basic_qos(prefetch_count=1)
        self.channel.basic_consume(
            consumer_callback=self.call_back,
            queue='queue_consumer1',
            no_ack=False,  # 消费者对消息进行确认,防止消费者挂掉
            consumer_tag=str(uuid.uuid4())
        )

    def consume(self):
        self.receive_msg()
        self.channel.start_consuming()


if __name__ == '__main__':
    rabbit_consumer = RabbitConsumer()
    rabbit_consumer.consume()

 

消费者consumer2：

import pika
import uuid


class RabbitConsumer(object):
    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 声明一个队列queue_consumer2,并进行持久化(防止服务器挂掉),exclusive设置为false
        self.channel.queue_declare(
            exclusive=False, durable=True, queue='queue_consumer2'
        )

        # T声明一个exhange交换机,其类型为fanout广播类型
        self.channel.exchange_declare(
            exchange='fanout_exchange', exchange_type='fanout', durable=True
        )

        # 将队列queue_consumer2与该exchange交换机进行绑定
        self.channel.queue_bind(exchange='fanout_exchange', queue='queue_consumer2')

    def call_back(self, method, body):
        """
        消费者对消息进行确认,防止消费者挂掉.
        :param method:
        :param body:
        :return:
        """
        self.channel.basic_ack(delivery_tag=method.delivery_tag)
        print('接收到的消息为:{}'.format(str(body)))

    def receive_msg(self):
        print('consumer2开始接受消息...')
        self.channel.basic_consume(
            consumer_callback=self.call_back,
            queue='queue_consumer2',
            no_ack=False,
            consumer_tag=str(uuid.uuid4())
        )

    def consume(self):
        self.receive_msg()
        self.channel.start_consuming()


if __name__ == '__main__':
    rabbit_consumer = RabbitConsumer()
    rabbit_consumer.consume()

fanout会将接受到的所有消息广播到消费者consumer1和消费者consumer2，交换机的缺陷：它只能无意识的播放，不够灵活地控制消息广播给指定的消费者

 2. direct(直连交换机)

对于direct，根据绑定键判定应该将数据发送至哪个队列，消息进入队列，其绑定秘钥（routing_key）与消息的路由秘钥要完全匹配，当exchange使用相同的绑定秘钥（routing_key）去绑定多个队列也是合法的，在这种情况下direct exchange的效果等同于fanout exchange，交换机会将消息广播到所有匹配的队列当中。

direct生产者：

import pika


class RabbitProducer(object):
    """
    与RabbitMq服务器建立连接
    """

    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 声明一个exchange交换机,交换机的类型为direct
        self.channel.exchange_declare(
            exchange='direct_exchange', exchange_type='direct', durable=True
        )

    def send_msg(self, routing_key, message):
        """
        :param routing_key: 消息的路由键 本例中为routing_info
        :param message: 生成者发送的消息
        :return:
        """
        self.channel.basic_publish(
            exchange='direct_exchange',
            routing_key=routing_key,
            body=message,
            properties=pika.BasicProperties(
                delivery_mode=2,
                # 消息进行持久化(防止服务器挂掉.)===> 如果没有queue绑定到这个exchange交换机,这个参数是没有的.
            ))

    def close(self):
        self.conn.close()


if __name__ == "__main__":
    rabbit_producer = RabbitProducer()
    routing_key = 'routing_info'
    for i in range(10):
        message = 'hello world {}!'.format(i)
        print('生产者发送的消息为:{}'.format(message))
        rabbit_producer.send_msg(routing_key, message)

direct消费者：

import pika
import uuid


class RabbitConsumer(object):
    """
    消费者(订阅者)
    """

    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 消息持久化
        self.channel.queue_declare(
            exclusive=False, durable=True, queue='task_queue'
        )

        # 交换机类型为direct.
        self.channel.exchange_declare(
            exchange='direct_exchange', exchange_type='direct', durable=True
        )

        # 将队列与该exchange交换机进行绑定
        routing_keys = ['routing_info', 'aaa']
        for routing_key in routing_keys:
            self.channel.queue_bind(
                exchange='direct_exchange', queue='task_queue', routing_key=routing_key
            )  # 如果生产者发生消息的routing_key与消费者绑定队列的routing_key相同则成功发送

    def call_back(self, channel, method, properties, body):
        """
        消费者对消息进行确认,防止消费者挂掉
        :param channel:
        :param method:
        :param properties:
        :param body:
        :return:
        """
        self.channel.basic_ack(delivery_tag=method.delivery_tag)
        print('接收到的消息为:{}'.format(str(body)))

    def receive_msg(self):
        print('开始接受消息...')
        self.channel.basic_qos(prefetch_count=1)  # TODO 告诉RabbitMQ,不要向我发送新的消息.
        self.channel.basic_consume(
            consumer_callback=self.call_back,
            queue='task_queue',
            no_ack=False,
            consumer_tag=str(uuid.uuid4())
        )

    def consume(self):
        self.receive_msg()
        self.channel.start_consuming()


if __name__ == '__main__':
    rabbit_consumer = RabbitConsumer()
    rabbit_consumer.consume()

direct直连交换机相当于是fanout的升级版，当消费者的队列绑定的秘钥routing_key与生产者的routing_key相同时，消费者就会收到消息；当所有消费者的队列所绑定的routing_key都一样且与生产者相同时，就相当于fanout交换机

  3. topic(话题交换机)

direct(直连交换机)虽然相当于fanout的升级版，但它仍然有局限性，它不能根据多个标准进行路由;topic(话题交换机)可以很好地解决这一问题：
(1) 如果消息的路由秘钥与队列的绑定秘钥符合匹配规则，topic就会将消息发送到相应的队列当中

(2) 对于绑定键(routing_key)有两个特殊的情况： * （星号）可以代替一个单词，＃（散列）可以替代零个或多个单词

(3) 对于发送到topic交换机消息的routing_key如果包含特殊字符，只能是由"."分割的单词表，如("zhangsan.lisi")

topic 生产者：

import pika


class RabbitProducer(object):
    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 声明交换机,交换机的类型为topic
        self.channel.exchange_declare(
            exchange='logs_topic', exchange_type='topic', durable=True
        )

    def send_msg(self, routing_key, message):
        """
        :param routing_key: 消息的路由键
        :param message: 生成者发送的消息
        :return:
        """
        self.channel.basic_publish(
            exchange='logs_topic',
            routing_key=routing_key,
            body=message,
            properties=pika.BasicProperties(
                delivery_mode=2,
                # 消息进行持久化
            ))

    def close(self):
        self.conn.close()


if __name__ == "__main__":
    rabbit_producer = RabbitProducer()
    routing_keys = ['info', "debug", "a.debug.b", "a.info.b"]
    for routing_key in routing_keys:
        message = 'hello world! {}'.format(routing_key)
        print('生产者发送的消息为:{}'.format(message))
        rabbit_producer.send_msg(routing_key, message)

topic 消费者1 --> 实现fanout交换机：

"""
当topic交换机使用#绑定键绑定队列时，此时topic交换机就会将消息广播到所有的队列当中，
不管消息的路由秘钥如何，此时topic交换机的效果等同于fanout:发送所有消息都会接受到
"""
import pika
import uuid


class RabbitConsumer(object):
    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 消息持久化
        self.channel.queue_declare(
            exclusive=False, durable=True, queue='task_queue'
        )

        # 声明交换机,其类型为topic
        self.channel.exchange_declare(
            exchange='logs_topic', exchange_type='topic', durable=True
        )

        # 将队列与该交换机进行绑定
        routing_keys = ['#']  # 使用#绑定键时,它将接受所有的消息,同fanout效果一样.
        for routing_key in routing_keys:
            self.channel.queue_bind(
                exchange='logs_topic', queue='task_queue', routing_key=routing_key
            )

    def call_back(self, channel, method, properties, body):
        """
        消费者对消息进行确认,防止消费者挂掉
        :param channel:
        :param method:
        :param properties:
        :param body:
        :return:
        """
        self.channel.basic_ack(delivery_tag=method.delivery_tag)
        print('接收到的消息为:{}'.format(str(body)))

    def receive_msg(self):
        print('开始接受消息...')
        self.channel.basic_qos(prefetch_count=1)
        self.channel.basic_consume(
            consumer_callback=self.call_back,
            queue='task_queue',
            no_ack=False,  # 消费者对消息进行确认
            consumer_tag=str(uuid.uuid4())
        )

    def consume(self):
        self.receive_msg()
        self.channel.start_consuming()


if __name__ == '__main__':
    rabbit_consumer = RabbitConsumer()
    rabbit_consumer.consume()

topic 消费者2 --> 实现direct交换机：

"""
当topic交换机没有使用*和#匹配符绑定键绑定队列时，此时topic交换机的效果等同于direct，
会收到key相匹配的消息  如：info debug
"""
import pika
import uuid


class RabbitConsumer(object):
    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 消息持久化
        self.channel.queue_declare(
            exclusive=False, durable=True, queue='work_queue'
        )

        #
        # 声明交换机,其类型为topic
        self.channel.exchange_declare(
            exchange='logs_topic', exchange_type='topic', durable=True
        )

        # 将队列与交换机进行绑定
        routing_keys = ['info', 'debug']
        for routing_key in routing_keys:
            self.channel.queue_bind(
                exchange='logs_topic', queue='work_queue', routing_key=routing_key
            )

    def call_back(self, channel, method, properties, body):
        """
        消费者对消息进行确认,防止消费者挂掉
        :param channel:
        :param method:
        :param properties:
        :param body:
        :return:
        """
        self.channel.basic_ack(delivery_tag=method.delivery_tag)
        print('接收到的消息为:{}'.format(str(body)))

    def receive_msg(self):
        print('开始接受消息...')
        self.channel.basic_qos(prefetch_count=1)
        self.channel.basic_consume(
            consumer_callback=self.call_back,
            queue='work_queue',
            no_ack=False,  # 消费者对消息进行确认
            consumer_tag=str(uuid.uuid4())
        )

    def consume(self):
        self.receive_msg()
        self.channel.start_consuming()


if __name__ == '__main__':
    rabbit_consumer = RabbitConsumer()
    rabbit_consumer.consume()

topic 消费者3 --> 实现*.x.* 消息匹配：

"""
匹配任意点分割的单词 生产者发送的：a.debug.b 则匹配了'*.debug.*'
生产者发送的：a.info.b 则匹配了'*.info.*'
"""
import pika
import uuid


class RabbitConsumer(object):
    def __init__(self):
        self.conn = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost', port=5672)
        )
        self.channel = self.conn.channel()

        # 消息持久化
        self.channel.queue_declare(
            exclusive=False, durable=True, queue='other_queue'
        )

        # 声明交换机,其类型为topic
        self.channel.exchange_declare(
            exchange='logs_topic', exchange_type='topic', durable=True
        )

        # 将队列与交换机进行绑定
        routing_keys = ['*.info.*', '*.debug.*', 'dfdf*']
        for routing_key in routing_keys:
            self.channel.queue_bind(
                exchange='logs_topic', queue='other_queue', routing_key=routing_key
            )

    def call_back(self, channel, method, properties, body):
        """
        消费者对消息进行确认,防止消费者挂掉
        :param channel:
        :param method:
        :param properties:
        :param body:
        :return:
        """
        self.channel.basic_ack(delivery_tag=method.delivery_tag)
        print('接收到的消息为:{}'.format(str(body)))

    def receive_msg(self):
        print('开始接受消息...')
        self.channel.basic_qos(prefetch_count=1)
        self.channel.basic_consume(
            consumer_callback=self.call_back,
            queue='other_queue',
            no_ack=False,  # 消费者对消息进行确认
            consumer_tag=str(uuid.uuid4())
        )

    def consume(self):
        self.receive_msg()
        self.channel.start_consuming()


if __name__ == '__main__':
    rabbit_consumer = RabbitConsumer()
    rabbit_consumer.consume()

topic消费者执行结果：

消费者1：

消费者2：

消费者3：