python操作RabbitMQ

 RabbitMQ介绍

RabbitMQ是一个由erlang开发的AMQP（Advanced Message Queue）的开源实现的产品，RabbitMQ是一个消息代理，从“生产者”接收消息并传递消息至“消费者”，期间可根据规则路由、缓存、持久化消息。“生产者”也即message发送者以下简称P，相对应的“消费者”乃message接受者以下简称C，message通过queue由P到C，queue存在于RabbitMQ，可存储尽可能多的message，多个P可向同一queue发送message，多个C可从同一个queue接收message

• 内部架构：

   

• 说明

  • Message （消息）：RabbitMQ 转发的二进制对象，包括Headers（头）、Properties （属性）和 Data （数据），其中数据部分不是必要的。Producer（生产者）： 消息的生产者，负责产生消息并把消息发到交换机

  • Exhange的应用。

    • Consumer （消费者）：使用队列 Queue 从 Exchange 中获取消息的应用。
    • Exchange （交换机）：负责接收生产者的消息并把它转到到合适的队列

    • Queue （队列）：一个存储Exchange 发来的消息的缓冲，并将消息主动发送给Consumer，或者 Consumer 主动来获取消息。见 1.4 部分的描述。

    • Binding （绑定）：队列 和 交换机 之间的关系。Exchange 根据消息的属性和 Binding 的属性来转发消息。绑定的一个重要属性是 binding_key。

    • Connection （连接）和 Channel （通道）：生产者和消费者需要和 RabbitMQ 建立 TCP 连接。一些应用需要多个connection，为了节省TCP 连接，可以使用 Channel，它可以被认为是一种轻型的共享 TCP 连接的连接。连接需要用户认证，并且支持 TLS (SSL)。连接需要显式关闭。

 参考：RabbitMQ百度百科

RabbitMQ安装

 centos安装：

http://www.rabbitmq.com/install-rpm.html 

windows安装：

http://www.rabbitmq.com/install-windows.html

安装python rabbitMQ module：

pip install pika
or
easy_install pika
or
源码

https://pypi.python.org/pypi/pika

Python操作RabbitMQ

1.实现简单消息队列

一个Product向queue发送一个message，一个Client从该queue接收message并打印

远程连接rabbitmq server，需要配置权限：

首先在rabbitmq server上创建一个用户

sudo rabbitmqctl  add_user alex alex3714　

同时还要配置权限，允许从外面访问

sudo rabbitmqctl set_permissions -p / alex ".*" ".*" ".*"

set_permissions [-p vhost] {user} {conf} {write} {read}

 

vhost

The name of the virtual host to which to grant the user access, defaulting to /.

user

The name of the user to grant access to the specified virtual host.

conf

A regular expression matching resource names for which the user is granted configure permissions.

write

A regular expression matching resource names for which the user is granted write permissions.

read

A regular expression matching resource names for which the user is granted read permissions.

• 发消息 product 

import pika

credentials = pika.PlainCredentials('alex','alex3714')
# 凭证

connection = pika.BlockingConnection(pika.ConnectionParameters(host='192.168.152.134',port=5672,credentials=credentials))
# 定义连接池

channel = connection.channel()
# 生成连接通道

channel.queue_declare(queue='test')
# 声明队列以向其发送消息

channel.basic_publish(exchange='',routing_key='test',body='Hello World!')
# 注意当未定义exchange时，routing_key需和queue的值保持一致

print('send success msg to rabbitmq')
connection.close()
# 关闭连接

• 收消息，client 

import pika

credentials = pika.PlainCredentials('alex','alex3714')
# 凭证

connection = pika.BlockingConnection(pika.ConnectionParameters(host='192.168.152.134',port=5672,credentials=credentials))
# 连接参数

channel = connection.channel()
# 生成连接通道

channel.queue_declare(queue='test')
# 声明队列。之所以消费者也需要声明队列，是为了防止生产者未声明队列，导致运行报错。


def callback(ch, method, properties, body):
    """
    回调函数,处理从rabbitmq中取出的消息
    :param ch: 通道
    :param method: 方法
    :param properties: 属性
    :param body: 内容
    :return: 接收到得信息
    """
    print("[x] Received %r" % body)

    # print(ch,method,properties,body)
    """
    <pika.adapters.blocking_connection.BlockingChannel object at 0x0000000002F1DB70>
    <Basic.Deliver(['consumer_tag=ctag1.3c1d688587c447e5ac3a72ea99e98cac', 'delivery_tag=1', 'exchange=', 'redelivered=False', 'routing_key=test'])>
     <BasicProperties> b'Hello World!'
    """


channel.basic_consume(callback, queue='test', no_ack=True)
# no_ack 表示不需要发送ack。默认是False，表示开启状态。

print('[*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()
# 开始监听，接收消息

执行效果：

#product端：
send success msg to rabbitmq

#client端：
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'Hello World!'

• 消息确认

当客户端从队列中取出消息之后，可能需要一段时间才能处理完成，如果在这个过程中，客户端出错了，异常退出了，而数据还没用处理完成，那么非常不幸，这段数据就丢失了，因为rabbitmq默认会把此消息标记为已完成，然后从队列中移除。
消息确认是客户端从rabbitmq中取出消息，并处理完成之后，会发送一个ack告诉rabbitmq，消息处理完成，当rabbitmq收到客户端的获取消息请求之后，或标记为处理中，当再次收到ack之后，才会标记为已完成，然后从队列中删除。当rabbitmq检测到客户端和自己断开链接之后，还没收到ack，则会重新将消息放回消息队列，交给下一个客户端处理，保证消息不丢失，也就是说，RabbitMQ给了客户端足够长的时间来做数据处理。

在客户端使用no_ack来标记是否需要发送ack，默认是False，开启状态。

product向rabbitmq发送两条消息：

import pika

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672, ))     #定义连接池
channel = connection.channel()          #声明队列以向其发送消息消息

channel.queue_declare(queue='test')
channel.basic_publish(exchange='', routing_key='test', body='1')
channel.basic_publish(exchange='', routing_key='test', body='2')
channel.basic_publish(exchange='', routing_key='test', body='3')
print('send success msg to rabbitmq')
connection.close()   #关闭连接

客户端接受消息，不发送ack

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test')


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(5)
    #ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息




channel.basic_consume(callback,queue='test',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

执行结果，发现消息并没有从队列中删除

第一次执行：
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'1'
 [x] Received b'2'
 [x] Received b'3'
第二次执行：
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'1'
 [x] Received b'2'
 [x] Received b'3'

加入ack之后：

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test')


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(5)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息




channel.basic_consume(callback,queue='test',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

运行结果：发现第二次运行队列中已经没有消息

第一次：
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'1'
 [x] Received b'2'
 [x] Received b'3
 第二次：
  [*] Waiting for messages. To exit press CTRL+C

• 改变消息获取顺序

默认消息队列里的数据是按照顺序被消费者拿走，例如：消费者1 去队列中获取 奇数 序列的任务，消费者1去队列中获取 偶数 序列的任务。

channel.basic_qos(prefetch_count=1) 表示谁来谁取，不再按照奇偶数排列
默认情况：使用product往队列中放10个数字

import pika

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672, ))     #定义连接池
channel = connection.channel()          #声明队列以向其发送消息消息

channel.queue_declare(queue='test')
for i in range(10):
    channel.basic_publish(exchange='', routing_key='test', body=str(i))
    print('send success msg[%s] to rabbitmq' %i)
connection.close()   #关闭连接

运行结果；

send success msg[1] to rabbitmq
send success msg[2] to rabbitmq
send success msg[3] to rabbitmq
send success msg[4] to rabbitmq
send success msg[5] to rabbitmq
send success msg[6] to rabbitmq
send success msg[7] to rabbitmq
send success msg[8] to rabbitmq
send success msg[9] to rabbitmq

客户端1收消息：

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test')


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    #time.sleep(5)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息




channel.basic_consume(callback,queue='test',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

运行结果：

 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'0'
 [x] Received b'2'
 [x] Received b'4'
 [x] Received b'6'
 [x] Received b'8'

客户端2收消息：和client1的区别是加了一个sleep(1)

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test')


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息
    
channel.basic_consume(callback,queue='test',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

执行结果：

 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'1'
 [x] Received b'3'
 [x] Received b'5'
 [x] Received b'7'
 [x] Received b'9'

在两个客户端里加入channel.basic_qos(prefetch_count=1)参数

客户端1:

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test')


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
   #＃time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息

channel.basic_qos(prefetch_count=1)  #添加不按顺序分配消息的参数
channel.basic_consume(callback,queue='test',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

执行效果：

[*] Waiting for messages. To exit press CTRL+C
 [x] Received b'0'
 [x] Received b'2'
 [x] Received b'3'
 [x] Received b'4'
 [x] Received b'5'
 [x] Received b'6'
 [x] Received b'7'
 [x] Received b'8'
 [x] Received b'9'

客户端2:

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test')


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息



channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue='test',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

执行结果：

[*] Waiting for messages. To exit press CTRL+C
[x] Received b'1'

发现，加入channel.basic_qos(prefetch_count=1)参数之后，客户端2由于sleep了1s，所以只拿到了一个消息，其他的消息都被client1拿到了

• 消息持久化 消息确认机制使得客户端在崩溃的时候，服务端消息不丢失，但是如果rabbitmq奔溃了呢？该如何保证队列中的消息不丢失？ 此就需要product在往队列中push消息的时候，告诉rabbitmq，此队列中的消息需要持久化，用到的参数：durable=True，再次强调，Producer和client都应该去创建这个queue，尽管只有一个地方的创建是真正起作用的：

channel.basic_publish(exchange='',  
                      routing_key="test",  
                      body=message,  
                      properties=pika.BasicProperties(  
                         delivery_mode = 2, # make message persistent  
                      ))  

具体代码：

product端：

import pika

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672, ))     #定义连接池
channel = connection.channel()          #声明队列以向其发送消息消息

channel.queue_declare(queue='test_persistent',durable=True)
for i in range(10):
    channel.basic_publish(exchange='', routing_key='test_persistent', body=str(i),properties=pika.BasicProperties(delivery_mode=2))
    print('send success msg[%s] to rabbitmq' %i)
connection.close()   #关闭连接

client端：

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.queue_declare(queue='test_persistent',durable=True)


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    #time.sleep(5)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息



channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue='test_persistent',no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息
注意：client端也需配置durable=True，否则将报下面错误：

pika.exceptions.ChannelClosed: (406, "PRECONDITION_FAILED - parameters for queue 'test_persistent' in vhost '/' not equivalent")

配置完之后，发现product往rabbitmq端push消息之后，重启rabbitmq，消息依然存在

[root@dns ~]# rabbitmqctl list_queues
Listing queues ...
abc 0
abcd    0
hello2  300
test    0
test1   20
test_persistent 10
...done.
[root@dns ~]# /etc/init.d/rabbitmq-server restart
Restarting rabbitmq-server: SUCCESS
rabbitmq-server.
[root@dns ~]# rabbitmqctl list_queues
Listing queues ...
abc 0
abcd    0
hello2  300
test1   20
test_persistent 10
...done.

参考文档：参考文档：http://www.rabbitmq.com/tutorials/tutorial-two-python.html

2.使用Exchanges：

exchanges主要负责从product那里接受push的消息，根据product定义的规则，投递到queue中，是product和queue的中间件

• Exchange 类型

  • direct 关键字类型
  • topic 模糊匹配类型
  • fanout 广播类型

• 使用fanout实现发布订阅者模型

发布订阅和简单的消息队列区别在于，发布订阅会将消息发送给所有的订阅者，而消息队列中的数据被消费一次便消失。所以，RabbitMQ实现发布和订阅时，会为每一个订阅者创建一个队列，而发布者发布消息时，会将消息放置在所有相关队列中
订阅者：

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.exchange_declare(exchange='test123',type='fanout')  #定义一个exchange ,类型为fanout
rest = channel.queue_declare(exclusive=True)   #创建一个随机队列,并启用exchange
queue_name = rest.method.queue          #获取队列名
channel.queue_bind(exchange='test123',queue=queue_name)   #将随机队列名和exchange进行绑定


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息



channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue=queue_name,no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

发布者：

import pika

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672, ))     #定义连接池
channel = connection.channel()          #声明队列以向其发送消息消息

channel.exchange_declare(exchange='test123',type='fanout')
for i in range(10):
    channel.basic_publish(exchange='test123', routing_key='', body=str(i),properties=pika.BasicProperties(delivery_mode=2))
    print('send success msg[%s] to rabbitmq' %i)
connection.close()   #关闭连接

注意：
需先定义订阅者，启动订阅者，否则发布者publish到一个不存在的exchange是被禁止的。如果没有queue bindings exchange的话，msg是被丢弃的。

• 使用direct 实现根据关键字发布消息

消息发布订阅者模型是发布者发布一条消息，所有订阅者都可以收到，现在rabbitmq还支持根据关键字发送，在发送消息的时候使用routing_key参数指定关键字，rabbitmq的exchange会判断routing_key的值，然后只将消息转发至匹配的队列，注意，此时需要订阅者先创建队列

配置参数为exchange的type＝direct，然后定义routing_key即可

订阅者1： 订阅error,warning,info

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.exchange_declare(exchange='test321',type='direct')  #定义一个exchange ,类型为fanout
rest = channel.queue_declare(exclusive=True)   #创建一个随机队列,并启用exchange
queue_name = rest.method.queue          #获取队列名

severities = ['error','warning','info']   #定义三个routing_key

for severity in severities:
    channel.queue_bind(exchange='test321', routing_key=severity,queue=queue_name)


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息


channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue=queue_name,no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

订阅者2：订阅error,warning

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.exchange_declare(exchange='test321',type='direct')  #定义一个exchange ,类型为fanout
rest = channel.queue_declare(exclusive=True)   #创建一个随机队列,并启用exchange
queue_name = rest.method.queue          #获取队列名

severities = ['error','warning']   #定义两个routing_key

for severity in severities:
    channel.queue_bind(exchange='test321', routing_key=severity,queue=queue_name)


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息



channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue=queue_name,no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

发布者：

import pika

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672, ))     #定义连接池
channel = connection.channel()          #声明队列以向其发送消息消息

channel.exchange_declare(exchange='test321',type='direct')
channel.basic_publish(exchange='test321', routing_key='info', body='info msg',properties=pika.BasicProperties(delivery_mode=2))  #发送info msg到 info routing_key
channel.basic_publish(exchange='test321', routing_key='error', body='error msg',properties=pika.BasicProperties(delivery_mode=2)) #发送error msg到 error routing_key

print('send success msg[] to rabbitmq')
connection.close()   #关闭连接**

效果：发现订阅者1和订阅者2都收到error消息，但是只有订阅者1收到了info消息

订阅者1:
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'info msg'
 [x] Received b'error msg'
订阅者2:
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'error msg'

• 使用topic实现模糊匹配发布消息

direct实现了根据自定义的routing_key来标示不同的queue，使用topic可以让队列绑定几个模糊的关键字，之后发送者将数据发送到exchange，exchange将传入”路由值“和 ”关键字“进行匹配，匹配成功，则将数据发送到指定队列

# 表示可以匹配 0 个 或 多个 单词
*  表示只能匹配 一个 单词

如：
fuzj.test 和fuzj.test.test
fuzj.# 会匹配到 fuzj.test 和fuzj.test.test
fuzj.* 只会匹配到fuzj.test

订阅者1: 使用#匹配

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.exchange_declare(exchange='test333',type='topic')  #定义一个exchange ,类型为fanout
rest = channel.queue_declare(exclusive=True)   #创建一个随机队列,并启用exchange
queue_name = rest.method.queue          #获取队列名

channel.queue_bind(exchange='test333', routing_key='test.#',queue=queue_name)


def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息



channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue=queue_name,no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

订阅者2：使用＊匹配

import pika
import time

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672))
channel = connection.channel()

channel.exchange_declare(exchange='test333',type='topic')  #定义一个exchange ,类型为fanout
rest = channel.queue_declare(exclusive=True)   #创建一个随机队列,并启用exchange
queue_name = rest.method.queue          #获取队列名

channel.queue_bind(exchange='test333', routing_key='test.*',queue=queue_name)

def callback(ch, method, properties, body):
    '''回调函数,处理从rabbitmq中取出的消息'''
    print(" [x] Received %r" % body)
    time.sleep(1)
    ch.basic_ack(delivery_tag = method.delivery_tag)  #发送ack消息



channel.basic_qos(prefetch_count=1)
channel.basic_consume(callback,queue=queue_name,no_ack=False)
print(' [*] Waiting for messages. To exit press CTRL+C')
channel.start_consuming()    #开始监听 接受消息

发布者：

import pika

connection = pika.BlockingConnection(pika.ConnectionParameters(
    host='127.0.0.1', port=5672, ))     #定义连接池
channel = connection.channel()          #声明队列以向其发送消息消息

channel.exchange_declare(exchange='test333',type='topic')
channel.basic_publish(exchange='test333', routing_key='test.123', body='test.123 msg',properties=pika.BasicProperties(delivery_mode=2))
channel.basic_publish(exchange='test333', routing_key='test.123.321', body=' test.123.321 msg',properties=pika.BasicProperties(delivery_mode=2))

print('send success msg[] to rabbitmq')
connection.close()   #关闭连接

输出效果：

订阅者1:
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'test.123 msg'
 [x] Received b' test.123.321 msg'

订阅者2:
 [*] Waiting for messages. To exit press CTRL+C
 [x] Received b'test.123 msg'

• 实现RPC

• 过程：

• 客户端 Client 设置消息的 routing key 为 Service 的队列 op_q，设置消息的 reply-to 属性为返回的 response 的目标队列 reponse_q，设置其 correlation_id 为以随机UUID，然后将消息发到 exchange。比如channel.basic_publish(exchange='', routing_key='op_q', properties=pika.BasicProperties(reply_to = reponse_q, correlation_id = self.corr_id),body=request)

• Exchange 将消息转发到 Service 的 op_q

• Service 收到该消息后进行处理，然后将response 发到 exchange，并设置消息的 routing_key 为原消息的 reply_to 属性，以及设置其 correlation_id 为原消息的 correlation_id 。
  
ch.basic_publish(exchange='', routing_key=props.reply_to, properties=pika.BasicProperties(correlation_id = props.correlation_id), body=str(response))

• Exchange 将消息转发到 reponse_q

• Client 逐一接受 response_q 中的消息，检查消息的 correlation_id 是否为等于它发出的消息的correlation_id，是的话表明该消息为它需要的response。

•  代码实现:
  • 服务端：

    import pika
    import subprocess
    connection = pika.BlockingConnection(pika.ConnectionParameters(
        host='127.0.0.1', port=5672, ))       #定义连接池
    
    channel = connection.channel()    #创建通道
    
    channel.queue_declare(queue='rpc_queue')            #创建rpc_queue队列
    
    def operating(arg):
        p = subprocess.Popen(arg, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)   #执行系统命令
        res = p.stdout.read()       #取出标准输出
        if not res:                 #判断是否有执行结果
            responses_msg = p.stderr.read()         #没有执行结果则取出标准错误输出
        else:
            responses_msg = res
        return responses_msg
    
    def on_request(ch, method, props, body):
        command = str(body,encoding='utf-8')
        print(" [.] start Processing command : %s" % command)
        response_msg = operating(command)          #调用函数执行命令
        ch.basic_publish(exchange='',
                         routing_key=props.reply_to,
                         properties=pika.BasicProperties(correlation_id = props.correlation_id),body=str(response_msg))
        ch.basic_ack(delivery_tag = method.delivery_tag)
    
    
    channel.basic_qos(prefetch_count=1)       #消息不平均分配,谁取谁得
    channel.basic_consume(on_request, queue='rpc_queue')    #监听队列
    
    print(" [x] Awaiting RPC requests")
    channel.start_consuming()

  • 客户端

    import pika
    import uuid
    import time
    
    class FibonacciRpcClient(object):
        def __init__(self):
            self.connection = pika.BlockingConnection(pika.ConnectionParameters(
            host='127.0.0.1',port=5672,))     #定义连接池
    
            self.channel = self.connection.channel()        #创建通道
    
            result = self.channel.queue_declare(exclusive=True,auto_delete=True)  #创建客户端短接受服务端回应消息的队列,exclusive=True表示只队列只允许当前链接进行连接,auto_delete=True表示自动删除
            self.callback_queue = result.method.queue     #获取队列名称
    
            self.channel.basic_consume(self.on_response, no_ack=True,
                                       queue=self.callback_queue)            #从队列中获取消息
    
        def on_response(self, ch, method, props, body):
            if self.corr_id == props.correlation_id:     #判断
                self.response = body
    
        def call(self, n):
            self.response = None
            self.corr_id = str(uuid.uuid4())
            self.channel.basic_publish(exchange='',
                                       routing_key='rpc_queue',
                                       properties=pika.BasicProperties(
                                             reply_to = self.callback_queue,   #回应消息的队列
                                             correlation_id = self.corr_id, #correlation id可以理解为请求的唯一标识码
                                             ),
                                       body=str(n))
            while self.response is None:        #不断从自己监听的队列里取消息,直到取到消息
                self.connection.process_data_events()
            return self.response.decode()
    
    fibonacci_rpc = FibonacciRpcClient()
    
    print(" [x] Requesting server" )
    time.sleep(0.1)
    while True:
        command = input('>> ')
        response = fibonacci_rpc.call(command)
        print(" [.] Get %r 
    " % response)