tutorial,

MetaPuppetForPython - 用Python写Wechaty程序

Tom Tom Follow Mar 07, 2020 · 17 mins read
MetaPuppetForPython - 用Python写Wechaty程序

一个支持Wechaty调用的Python框架

MetaPuppetForPython

特征:

  • 你可以只会python
  • 支持Wechaty全部协议,全部接口
  • 兼容同步/异步编程
  • 可拓展至其他app/社交平台

声明:作者是更加支持native的python-wechaty的哈,看好@Huan@wj-Mcat的工作! 不过貌似开发还需要一段时间,有需要的童鞋可以暂时用MetaPuppetForPython。

Talk is cheap, show me the code

初级:Hello Human

目标:更改微信签名 要点:你可以调用类似a_server.change_self_signature()的函数来主动调用Wechaty的各种接口,从而主动查找联系人,发信息等。 你可以在SocketServerCoreSweetSocketServer中找到更多类似的函数。 但他们其实都是只语法糖(下文会介绍原理),你也可以继承SweetSocketServer,然后根据自己的需要设计自己的函数。

# -*- coding: utf-8 -*-
import time

from MetaPuppet.core.SweetSocketServer import SweetSocketServer
from MetaPuppet.core.TestBot import TestBot

if __name__ == '__main__':
    # init
    a_bot = TestBot(name='test')
    a_server = SweetSocketServer(
        robot=a_bot,
        num_async_threads=1,
        debug_mode=True
    )
    a_server.run()

    print(
        'Please make sure the client is connected '
        'before run the following codes'
    )
    time.sleep(20)

    #  -------------edit following code for simple tasks-----------------------
    a_server.change_self_signature('Hello Human!')

初级:a simple bot

目标:一个简易机器人。自动接受加人邀请,并问好Hello Human!。自动回复,内容为接收到信息的反序。 要点:你可以继承RobotBase然后override _process_message()_process_friend_invitation()函数。 这些函数在收到相关微信信息后会被自动调用,你可以把机器人自动回复的逻辑放在这个类中。 通过这种方式,你可以处理所有的被动请求。结合Hello Human!的例子,你可以主动发出请求。 到此,你已经可以实现全部的业务逻辑。

# -*- coding: utf-8 -*-
import time

from MetaPuppet.core.SweetSocketServer import SweetSocketServer
from MetaPuppet.core.RobotBase import RobotBase

class MyBot(RobotBase):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

    async def _process_message(self, message, verbose=False):
        #  -------------edit following code for simple tasks-----------------------
        return_msg = None
        if 'payload' in message and 'text' in message['payload']:
            return_msg = message['payload']['text'][::-1]
        return return_msg

    async def _process_friend_invitation(self, message, verbose=False):
        return_msg = {
            'wx_msg_type': 'TEXT',
            'path': 'Hello Human!',
        }
        return return_msg

if __name__ == '__main__':
    # init
    a_bot = MyBot(name='test')
    a_server = SweetSocketServer(
        robot=a_bot,
        num_async_threads=1,
        debug_mode=True
    )
    a_server.run()

    print(
        'Please make sure the client is connected '
        'before run the following codes'
    )
    time.sleep(20)

    #  -------------edit following code for simple tasks-----------------------
    # nothing needed here in this example
    # however, you can put your other backend code here

中级:语法糖

目标:自定义函数:发送文本给特定联系人/群聊 要点:MetaPuppetForPython实际上是用python将ts的代码片段转发给node并编译运行(详见原理部分),所以若已有的语法糖不能满足需求,你需要定义自己的函数。 由于ts的代码和python其实比较相似,并且往往只需要写很少的ts的代码,主要业务代码还是python,所以自定义函数的难度应该不大。 你可以继承SweetSocketServer,然后在自定义函数中调用exec_wx_function()exec_one_wx_function()执行文本格式的ts代码。 如果你需要使用第三方的ts库,可以在socket_client/src的相关代码中声明。

同步代码的语法糖:

from MetaPuppet.core.SweetSocketServer import SweetSocketServer

class ExtendedSocketServer(SweetSocketServer):

    # definition for other functions/variables
    ...

    def send_wx_msg_text(self, text, username, chat_type):
        ts_code = '''
            let say_content = `{}`
            const a_contact = bot.{}.load('{}')
            await a_contact.say(say_content)
        '''.format(text, self.all_chat_type[chat_type.lower()], username)
        self.exec_wx_function(
            ts_code=ts_code,
            need_return=False,
        )

异步代码的语法糖(只需要在函数的定义前加上async,在调用异步函数前面加上await):

from MetaPuppet.core.SweetSocketServer import SweetSocketServer

class ExtendedSocketServer(SweetSocketServer):

    # definition for other functions/variables
    ...

    async def async_send_wx_msg_text(self, text, username, chat_type):
        ts_code = '''
            let say_content = `{}`
            const a_contact = bot.{}.load('{}')
            await a_contact.say(say_content)
        '''.format(text, self.all_chat_type[chat_type.lower()], username)
        await self.async_exec_wx_function(
            ts_code=ts_code,
            need_return=False,
        )

中级:同步/异步

目标:通过使用异步编程加速代码 要点:同步和异步的区别是在等待响应时前者是真的等待,后者是会缓存当前程序,执行其他异步程序,收到响应时恢复当前程序继续执行。 需要等待的情况有:网络通信等待回复,等待其他某个进程结束,人为使用等待/睡眠等函数。 所以,对于大多数看到这篇帖子的人而言,异步代码加速的主要是网络通信的部分。 如果同步代码本身很慢,将其改为异步并不能减少其执行时间。 基于MetaPuppetForPython的原理,对于频繁调用Wechaty的函数,建议使用异步编程,如默认的_process_message()就是异步函数。 在_process_message()中,若需要经常调用Wechaty函数,建议使用异步版本的语法糖;若仅是偶尔调用,使用同步版本亦可。 当然,为避免所有函数都需要异步定义,建议在设计层面将Wechaty的相关调用放在比较上层的位置,甚至与核心的同步代码相互独立。 对于大部分业务代码,只是偶尔调用Wechaty函数,可以直接使用同步风格的语法糖,并不影响效率。

异步:

# -*- coding: utf-8 -*-
import threading
import time

from MetaPuppet.core.SweetSocketServer import SweetSocketServer
from MetaPuppet.core.RobotBase import RobotBase
from MetaPuppet.core.time_classes import Time
from MetaPuppet.core.utils import run_coroutine_in_new_thread

class MyBot(RobotBase):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

    async def _process_message(self, message, verbose=False):
        #  -------------edit following code for simple tasks-----------------------
        return_msg = None
        if 'payload' in message and 'text' in message['payload']:
            return_msg = message['payload']['text'][::-1]
        return return_msg

    async def _process_friend_invitation(self, message, verbose=False):
        return_msg = {
            'wx_msg_type': 'TEXT',
            'path': 'Hello Human!',
        }
        return return_msg

async def async_foo(server):
    #  -------------edit following code for simple tasks-----------------------
    rooms = await server.async_exec_one_wx_function(
        func_name='bot.Room.findAll',
        func_paras=[],
        need_return=True,
    )
    print('Time:', Time())
    print('async_foo', threading.currentThread().getName())
    if rooms is not None:
        print('async: len(rooms)', len(rooms))
        print(rooms[0])
    else:
        print('async: rooms not found')

if __name__ == '__main__':
    # init
    a_bot = MyBot(name='test')
    a_server = SweetSocketServer(
        robot=a_bot,
        num_async_threads=1,
        debug_mode=True
    )
    a_server.run()

    print(
        'Please make sure the client is connected '
        'before run the following codes'
    )
    time.sleep(20)

    #  -------------edit following code for simple tasks-----------------------
    # async version
    # better to use async version because sync version might block io
    run_coroutine_in_new_thread(
        async_foo(a_server)
    )

同步:


...

def bar(server):
    contacts = server.exec_one_wx_function(
        func_name='bot.Contact.findAll',
        func_paras=[],
        need_return=True,
    )
    print('Time:', Time())
    print('bar', threading.currentThread().getName())
    if contacts is not None:
        print('sync: len(contacts)', len(contacts))
        print(contacts[0])
    else:
        print('sync: contacts not found')

if __name__ == '__main__':
    ...

    #  -------------edit following code for simple tasks-----------------------
    # sync version
    # sync code is simpler, though not recommended for heavy load
    # you can use it if you don't need to use many async functions
    # (e.g. async_exec_one_wx_function) at the same time and also
    # (1) the task here is short-term and light-load
    # or (2) you have a powerful computer
    bar(a_server)

中级:线程

目标:通过使用多线程加速代码 要点:上文中提到,异步主要是提升io方面的效率。 对于需要较长计算时间的代码本身,异步帮助不大。 此时可以使用多线程来避免堵塞其他任务。 SweetSocketServer会默认开放一个线程进行消息的自动回复,需要的情况下可以设置初始化参数num_async_threads增加线程数。


...

def bar(server):
    contacts = server.exec_one_wx_function(
        func_name='bot.Contact.findAll',
        func_paras=[],
        need_return=True,
    )
    print('Time:', Time())
    print('bar', threading.currentThread().getName())
    if contacts is not None:
        print('sync: len(contacts)', len(contacts))
        print(contacts[0])
    else:
        print('sync: contacts not found')

if __name__ == '__main__':
    ...

    #  -------------edit following code for simple tasks-----------------------
    # sync version
    # sync code is simpler, though not recommended for heavy load
    # you can use it if you don't need to use many async functions
    # (e.g. async_exec_one_wx_function) at the same time and also
    # (1) the task here is short-term and light-load
    # or (2) you have a powerful computer
    # however, pay attention that you should run the task here in a new thread
    # it is not good to run it in the main thread
    threading.Thread(
        target=bar,
        args=(a_server,)
    ).start()

高级:拓展至其他app/社交平台

目标:使用MetaPuppetForPython框架与web app通信 要点:MetaPuppetForPython将机器人本身作为服务器,将Wechaty作为客户,但其实可以有多个客户同时与机器人进行交互,可以是微博,QQ,或者网页应用等。 你可以继承SweetSocketServer并override process_socket_message()process_custom_message()以处理来自其他app的信息。

服务器:

# -*- coding: utf-8 -*-
import time
from MetaPuppet.core.SweetSocketServer import SweetSocketServer
from MetaPuppet.core.RobotBase import RobotBase

class ExtendedSocketServer(SweetSocketServer):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        self.webgui_room = 'webgui'

    async def process_socket_message(self, sid, message, verbose=False):
        sender = message['sender']
        if sender.startswith('wx_'):
            room_name = self.wx_room
        elif sender.startswith('webgui_'):
            room_name = self.webgui_room
        else:
            room_name = sender

        text = message.get('text', '')
        if text == 'CONNECTED':
            print('{}: sender, sid, room_name'.format(text), sender, sid, room_name)
            self.sio.enter_room(sid, room_name)
            self.add_room(
                sender=sender,
                room_name=room_name,
            )
        else:
            pass

    async def process_custom_message(self, sid, message, verbose=False):
        sender = message['sender']
        msg_type = message.get('type', '')
        if sender.startswith('webgui_'):
            if msg_type == 'CHAT_INFO':
                print('webgui info received', message)
                await self.robot.process_webgui_chat_message(message, verbose=verbose)
                pass

class MyBot(RobotBase):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

    async def _process_message(self, message, verbose=False):
        #  -------------edit following code for simple tasks-----------------------
        return_msg = None
        if 'payload' in message and 'text' in message['payload']:
            return_msg = message['payload']['text'][::-1]
        return return_msg

    async def _process_friend_invitation(self, message, verbose=False):
        return_msg = {
            'wx_msg_type': 'TEXT',
            'path': 'Hello Human!',
        }
        return return_msg

    async def process_webgui_chat_message(self, message, verbose=False):
        return_msg = await self._process_webgui_chat_message(message, verbose=verbose)
        return return_msg

    async def _process_webgui_chat_message(self, message, verbose=False):
        print('message', message)
        return None

if __name__ == '__main__':
    # init
    a_bot = MyBot(name='test')
    a_server = ExtendedSocketServer(
        robot=a_bot,
        num_async_threads=1,
        debug_mode=True
    )
    a_server.run()

    print(
        'Please make sure the client is connected '
        'before run the following codes'
    )
    time.sleep(20)

     # -------------edit following code for simple tasks-----------------------

Web app作为客户端(示例中的任务不需要性能,可以直接用同步代码):

import socketio
import json
import os

class WebClient(object):
    def __init__(self,
                 config_path):
        self.config_path = config_path
        self.config = self.read_config(config_path=self.config_path)
        self.sio = socketio.Client()
        self.create_response_functions()

    def read_config(self, config_path):
        # get config
        with open(config_path, 'r') as fr:
            config = json.load(fr)
        return config

    def run(self):
        self.sio.connect('http://{}:{}'.format(
                self.config['server']['host'],
                self.config['server']['port'],
            ))
        # use sio.wait if no other process keeping the program living
        # self.sio.wait()
        ...

    def create_response_functions(self):
        @self.sio.on('message')
        def process_msg_from_server(message):
            ...

        @self.sio.on('connect')
        def on_connect():
            print('CONNECTED')
            msg = {
                'type': 'SOCKET_INFO',
                'text': 'CONNECTED',
            }
            self.send_msg_to_server(msg)

        @self.sio.on('disconnect')
        def on_disconnect():
            print('DISCONNECTED')

    def send_msg_to_server(self, msg):
        msg_to_send = {
            'sender': 'webgui_client',
            'status': 'NORMAL',
        }
        if isinstance(msg, str):
            msg_to_send['text'] = msg
        elif isinstance(msg, dict):
            msg_to_send.update(msg)
        else:
            msg_to_send['status'] = 'ERROR'

        self.sio.emit(
            'message',
            msg_to_send,
        )

a_webgui = WebClient(
    config_path=os.path.join(os.path.dirname(__file__), '../config.json')
)
# run is a dead loop if sio.wait() called, so we should call lastly
a_webgui.run()

原理

MetaPuppetForPython
MetaPuppetForPython的核心是通过socket双向通信,实现业务逻辑(server)与第三方业务(client)的交互,包括信息的收发以及接口的调用等。 所以,在这个框架下,与Wechaty的架构类比,server是Wechaty的TS代码,client是Wechaty调用的各个puppet。 MetaPuppetForPython默认包括完全的server和运行Wechaty的client代码,针对其他app(weibo, qq, web, etc.)的client需要自定义。 对于想要一个python版本的Wechaty的童鞋来说,默认的代码已经够用。

针对Wechaty而言,server使用python-socketio,client使用socket.io-client,从而建立起双向异步通信。 主要的业务逻辑在server端,使用Python控制。 在需要调用Wechaty时,使用Python将文本形式的小段ts代码发送给client,client将其编译运行。 client端包含少量的ts代码,用于编译来自server的代码,和响应微信的请求(新消息,好友请求等),一般不需要更改。

基于此原理,可以使用Python调用Wechaty的任意原生代码,所以理论上可以兼容Wechaty的所有协议和接口函数。 已实现async_exec_wx_function()用于运行代码块,和async_exec_one_wx_function()用于运行单个函数。

安装

git clone https://github.com/quantumFlame/MetaPuppetForPython.git
cd MetaPuppetForPython
pip install .

cd socket_client
# install node.js
# https://github.com/nodesource/distributions/blob/master/README.md
# Using Ubuntu
curl -sL https://deb.nodesource.com/setup_13.x | sudo -E bash -
sudo apt-get install -y nodejs
npm install -g ts-node
npm install -g typescript  
sudo apt-get install autoconf
sudo apt-get install libtool

npm install
# or
# rm -rf node_modules package-lock.json
# npm install wechaty@latest
# npm install wechaty-puppet-padplus@next
# npm install qrcode-terminal
# npm install socket.io-client
# npm install @types/socket.io-client
# npm install @types/node
# npm install other packages if needed...

运行

# start client in one terminal
cd socket_client
# (before you run, you need a wechaty token and
# create the config.json file following
# the example config.example.json)
ts-node src/wechaty_actions.ts

# start server in another terminal
python example/hello_word.py

Keys to remember

  • Extend RobotBase and modify _process_message() to reply to various wechat messages.

  • Compile your management tasks as async_foo() and call with run_coroutine_in_new_thread().

  • If you don’t like async, you can also run the sync version functions in new thread (see more details in example/hello_world.py).

作者: Tom 代码: MetaPuppetForPython 首发于博客: 用Python写Wechaty程序 文本协议: CC BY-NC-SA 3.0 CN

Join Newsletter
Get the latest news right in your inbox. We never spam!
Written by Tom
A student curious about and interested in many things I don't know.