python队列通信：rabbitMQ使用

电玩女神 2022-06-05 11:11 167阅读 0赞

**（一）、前言**  
为什么引入消息队列？  
1.程序解耦  
2.提升性能  
3.降低多业务逻辑复杂度

**（二）、python操作rabbit mq**  
rabbitmq配置安装基本使用参见上节文章，不再复述。  
若想使用python操作rabbitmq，需安装pika模块，直接pip安装：  
pip install pika

1.最简单的rabbitmq producer端与consumer端对话：  
producer：

#Author :ywq
    
    import pika
    auth=pika.PlainCredentials('ywq','qwe')  #save auth indo
    
    connection = pika.BlockingConnection(pika.ConnectionParameters(
                   '192.168.0.158',5672,'/',auth))  #connect to rabbit
    
    channel = connection.channel()   #create channel
    
    channel.queue_declare(queue='hello')   #declare queue
    
    #n RabbitMQ a message can never be sent directly to the queue, it always needs to go through an exchange.
    channel.basic_publish(exchange='',
                          routing_key='hello',
                          body='Hello World!')    #the body is the msg content
    print(" [x] Sent 'Hello World!'")
    connection.close()

consumer：

#Author :ywq
    import pika
    auth=pika.PlainCredentials('ywq','qwe') #auth info
    
    connection = pika.BlockingConnection(pika.ConnectionParameters(
                   '192.168.0.158',5672,'/',auth))   #connect to rabbit
    
    channel = connection.channel()          #create channel
    
    
    channel.queue_declare(queue='hello')      #decalre queue
    
    def callback(ch, method, properties, body):
        print(" [x] Received %r" % body)
    
    
    channel.basic_consume(callback,
                          queue='hello',
                          no_ack=True)
    
    print(' [*] Waiting for messages. To exit press CTRL+C')
    channel.start_consuming()

消息传递消费过程中，可以在rabbit web管理页面实时查看队列消息信息。  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI1MTgwNjU2MzU5]

2.持久化的消息队列，避免宕机等意外情况造成消息队列丢失。  
consumer端无需改变，在producer端代码内加上两个属性，分别使消息持久化、队列持久化，只选其一还是会出现消息丢失，必须同时开启：

delivery_mode=2  #make msg persisdent
    durable=True

属性插入位置见如下代码（producer端）：

#Author :ywq
    import pika,sys
    auth_info=pika.PlainCredentials('ywq','qwe')
    
    connection=pika.BlockingConnection(pika.ConnectionParameters(
                 '192.168.0.158',5672,'/',auth_info
            ))
    
    channel=connection.channel()
    
    channel.queue_declare(queue='test1',durable=True)  #durable=Ture, make queue persistent
    
    
    msg=''.join(sys.argv[1:]) or 'Hello'
    
    channel.basic_publish(
        exchange='',
        routing_key='test1',
        body=msg,
        properties=pika.BasicProperties(
                delivery_mode=2  #make msg persisdent
        )
    )
    
    
    print('Send done:',msg)
    connection.close()

3.公平分发  
在多consumer的情况下，默认rabbit是轮询发送消息的，但有的consumer消费速度快，有的消费速度慢，为了资源使用更平衡，引入ack确认机制。consumer消费完消息后会给rabbit发送ack，一旦未ack的消息数量超过指定允许的数量，则不再往该consumer发送，改为发送给其他consumer。

producer端代码不用改变，需要给consumer端代码插入两个属性：

channel.basic_qos(prefetch_count= *)   #define the max  non_ack_count
    channel.basic_ack(delivery_tag=deliver.delivery_tag)  #send ack to rabbitmq

属性插入位置见如下代码（consumer端）：

#Author :ywq
    import pika,time
    auth_info=pika.PlainCredentials('ywq','qwe')
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
        )
    )
    
    channel=connection.channel()
    
    channel.queue_declare(queue='test2',durable=True)
    
    def callback(chann,deliver,properties,body):
        print('Recv:',body)
        time.sleep(5)
        chann.basic_ack(delivery_tag=deliver.delivery_tag)  #send ack to rabbit
    
    channel.basic_qos(prefetch_count=1)
    '''
    注意，no_ack=False  注意，这里的no_ack类型仅仅是告诉rabbit该消费者队列是否返回ack，若要返回ack，需要在callback内定义
    
    prefetch_count=1,未ack的msg数量超过1个，则此consumer不再接受msg，此配置需写在channel.basic_consume上方，否则会造成non_ack情况出现。
    '''
    
    channel.basic_consume(
        callback,
        queue='test2'
    )
    
    
    channel.start_consuming()

三、消息发布/订阅  
上方的几种模式都是producer端发送一次，则consumer端接收一次，能不能实现一个producer发送，多个关联的consumer同时接收呢？of course，rabbit支持消息发布订阅，共支持三种模式，通过组件exchange转发器，实现3种模式：  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI1MTgyNDMwMDA3]

fanout: 所有bind到此exchange的queue都可以接收消息，类似广播。  
direct: 通过routingKey和exchange决定的哪个唯一的queue可以接收消息，推送给绑定了该queue的consumer，类似组播。  
topic:所有符合routingKey(此时可以是一个表达式)的routingKey所bind的queue可以接收消息，类似前缀列表匹配路由。

1.fanout  
publish端（producer）：

#Author :ywq
    import pika,sys,time
    auth_info=pika.PlainCredentials('ywq','qwe')
    
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
        )
    )
    
    channel=connection.channel()
    
    channel.exchange_declare(exchange='hello',
                             exchange_type='fanout'
                             )
    
    msg=''.join(sys.argv[1:]) or 'Hello world %s' %time.time()
    
    channel.basic_publish(
        exchange='hello',
        routing_key='',
        body=msg,
        properties=pika.BasicProperties(
            delivery_mode=2
        )
    )
    
    print('send done')
    connection.close()

subscribe端（consumer）：

#Author :ywq
    import pika
    auth_info=pika.PlainCredentials('ywq','qwe')
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
        )
    )
    
    channel=connection.channel()
    
    channel.exchange_declare(
        exchange='hello',
        exchange_type='fanout'
    )
    
    random_num=channel.queue_declare(exclusive=True)  #随机与rabbit建立一个queue，comsumer断开后，该queue立即删除释放
    queue_name=random_num.method.queue
    channel.basic_qos(prefetch_count=1)
    
    channel.queue_bind(
        queue=queue_name,
        exchange='hello'
    )
    
    def callback(chann,deliver,properties,body):
        print('Recv:',body)
        chann.basic_ack(delivery_tag=deliver.delivery_tag)  #send ack to rabbit
    
    
    channel.basic_consume(
        callback,
        queue=queue_name,
    )
    
    channel.start_consuming()

实现producer一次发送，多个关联consumer接收。  
使用exchange模式时：  
1.producer端不再申明queue，直接申明exchange  
2.consumer端仍需绑定队列并指定exchange来接收message  
3.consumer最好创建随机queue，使用完后立即释放。  
随机队列名在web下可以检测到：  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI1MTgzNzUwNTYw]

2.direct  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTMzMjMxMDIw]  
使用exchange同时consumer有选择性的接收消息。队列绑定关键字，producer将数据根据关键字发送到消息exchange，exchange根据 关键字 判定应该将数据发送至指定队列，consumer相应接收。即在fanout基础上增加了routing key.

producer：

#Author :ywq
    import pika,sys
    auth_info=pika.PlainCredentials('ywq','qwe')
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
        )
    )
    
    channel=connection.channel()
    channel.exchange_declare(exchange='direct_log',
                          exchange_type='direct',
                          )
    while True:
        route_key=input('Input  routing key:')
    
        msg=''.join(sys.argv[1:]) or 'Hello'
    
        channel.basic_publish(
            exchange='direct_log',
            routing_key=route_key,
            body=msg,
            properties=pika.BasicProperties(
                delivery_mode=2
            )
        )
    
    connection.close()

consumer：

#Author :ywq
    import pika,sys
    auth_info=pika.PlainCredentials('ywq','qwe')
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
    ))
    
    channel=connection.channel()
    
    channel.exchange_declare(
        exchange='direct_log',
        exchange_type='direct'
    )
    
    queue_num=channel.queue_declare(exclusive=True)
    queue_name=queue_num.method.queue
    
    route_key=input('Input routing key:')
    
    
    channel.queue_bind(
        queue=queue_name,
        exchange='direct_log',
        routing_key=route_key
    )
    
    def callback(chann,deliver,property,body):
        print('Recv:[level:%s],[msg:%s]' %(route_key,body))
        chann.basic_ack(delivery_tag=deliver.delivery_tag)
    
    
    channel.basic_qos(prefetch_count=1)
    channel.basic_consume(
        callback,
        queue=queue_name
    )
    
    channel.start_consuming()

同时开启多个consumer，其中两个接收notice，两个接收warning，运行效果如下：  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTM0NDE2NzY4]

3.topic  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTM1NDAxMjQ2]  
相较于direct，topic能实现模糊匹配式工作方式（在consumer端指定匹配方式），只要routing key包含指定的关键字，则将该msg发往绑定的queue上。

rabbitmq通配符规则：  
符号“\#”匹配一个或多个词，符号“\*”匹配一个词。因此“abc.\#”能够匹配到“abc.m.n”，但是“abc.\*\*‘’ 只会匹配到“abc.m”。‘.’号为分割符。使用通配符匹配时必须使用‘.’号分割。

producer：

#Author :ywq
    import pika,sys
    auth_info=pika.PlainCredentials('ywq','qwe')
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
        )
    )
    
    channel=connection.channel()
    channel.exchange_declare(exchange='topic_log',
                          exchange_type='topic',
                          )
    while True:
        route_key=input('Input  routing key:')
    
        msg=''.join(sys.argv[1:]) or 'Hello'
    
        channel.basic_publish(
            exchange='topic_log',
            routing_key=route_key,
            body=msg,
            properties=pika.BasicProperties(
                delivery_mode=2
            )
        )
    
    connection.close()

consumer：

#Author :ywq
    import pika,sys
    auth_info=pika.PlainCredentials('ywq','qwe')
    connection=pika.BlockingConnection(pika.ConnectionParameters(
        '192.168.0.158',5672,'/',auth_info
    ))
    
    channel=connection.channel()
    
    channel.exchange_declare(
        exchange='topic_log',
        exchange_type='topic'
    )
    
    queue_num=channel.queue_declare(exclusive=True)
    queue_name=queue_num.method.queue
    
    route_key=input('Input routing key:')
    
    
    channel.queue_bind(
        queue=queue_name,
        exchange='topic_log',
        routing_key=route_key
    )
    
    def callback(chann,deliver,property,body):
        print('Recv:[type:%s],[msg:%s]' %(route_key,body))
        chann.basic_ack(delivery_tag=deliver.delivery_tag)
    
    
    channel.basic_qos(prefetch_count=1)
    channel.basic_consume(
        callback,
        queue=queue_name
    )
    
    channel.start_consuming()

运行效果：  
![这里写图片描述][aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTQzNDQ3NTgz]

rabbitmq三种publish/subscribe模型简单介绍完毕，  
以上！

[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI1MTgwNjU2MzU5]: /images/20220605/7a3abfefdaf34562981bed07a0ff9e40.png
[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI1MTgyNDMwMDA3]: /images/20220605/a0b82b1a2fb247509f707d0f21c76e72.png
[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI1MTgzNzUwNTYw]: /images/20220605/e4ab6f8ec4ac4faa8568edcb6da3f6a3.png
[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTMzMjMxMDIw]: /images/20220605/5acb43ad5ae74cb490670cb5f9e5900e.png
[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTM0NDE2NzY4]: /images/20220605/c4a5e2ab4d8942f482f4d9ddd99b865f.png
[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTM1NDAxMjQ2]: /images/20220605/9ae450774fae49e786ab630fe71be9a9.png
[aHR0cDovL2ltZy5ibG9nLmNzZG4ubmV0LzIwMTcxMTI2MTQzNDQ3NTgz]: /images/20220605/d4b5b592832043279b7171789fbb5eb1.png