Просмотр исходного кода

refactor: using one processing logic in chat channel

develop
lanvent 1 год назад
Родитель
Сommit
02cd553990
7 измененных файлов: 159 добавлений и 339 удалений
  1. +1
    -0
      channel/channel.py
  2. +18
    -18
      channel/chat_channel.py
  3. +4
    -10
      channel/wechat/wechat_channel.py
  4. +1
    -1
      channel/wechat/wechat_message.py
  5. +99
    -304
      channel/wechat/wechaty_channel.py
  6. +26
    -6
      channel/wechat/wechaty_message.py
  7. +10
    -0
      voice/audio_convert.py

+ 1
- 0
channel/channel.py Просмотреть файл

@@ -20,6 +20,7 @@ class Channel(object):
"""
raise NotImplementedError

# 统一的发送函数,每个Channel自行实现,根据reply的type字段发送不同类型的消息
def send(self, reply: Reply, context: Context):
"""
send message to user


+ 18
- 18
channel/chat_channel.py Просмотреть файл

@@ -1,6 +1,7 @@



import os
import re
import time
from common.expired_dict import ExpiredDict
@@ -10,7 +11,10 @@ from bridge.context import *
from config import conf
from common.log import logger
from plugins import *

try:
from voice.audio_convert import any_to_wav
except Exception as e:
pass

# 抽象类, 它包含了与消息通道无关的通用处理逻辑
class ChatChannel(Channel):
@@ -30,11 +34,13 @@ class ChatChannel(Channel):
context['origin_ctype'] = ctype
# context首次传入时,receiver是None,根据类型设置receiver
first_in = 'receiver' not in context

# 群名匹配过程,设置session_id和receiver
if first_in: # context首次传入时,receiver是None,根据类型设置receiver
config = conf()
cmsg = context['msg']
if cmsg.from_user_id == self.user_id:
logger.debug("[WX]self message skipped")
return None
if context["isgroup"]:
group_name = cmsg.other_user_nickname
group_id = cmsg.other_user_id
@@ -47,14 +53,13 @@ class ChatChannel(Channel):
if any([group_name in group_chat_in_one_session, 'ALL_GROUP' in group_chat_in_one_session]):
session_id = group_id
else:
return
return None
context['session_id'] = session_id
context['receiver'] = group_id
else:
context['session_id'] = cmsg.other_user_id
context['receiver'] = cmsg.other_user_id


# 消息内容匹配过程,并处理content
if ctype == ContextType.TEXT:
if first_in and "」\n- - - - - - -" in content: # 初次匹配 过滤引用消息
@@ -99,10 +104,6 @@ class ChatChannel(Channel):


return context
# 统一的发送函数,每个Channel自行实现,根据reply的type字段发送不同类型的消息
def send(self, reply: Reply, context: Context, retry_cnt = 0):
raise NotImplementedError

# 处理消息 TODO: 如果wechaty解耦,此处逻辑可以放置到父类
def _handle(self, context: Context):
@@ -128,22 +129,21 @@ class ChatChannel(Channel):
if context.type == ContextType.TEXT or context.type == ContextType.IMAGE_CREATE: # 文字和图片消息
reply = super().build_reply_content(context.content, context)
elif context.type == ContextType.VOICE: # 语音消息
msg = context['msg']
msg.prepare()
mp3_path = context.content
# mp3转wav
wav_path = os.path.splitext(mp3_path)[0] + '.wav'
cmsg = context['msg']
cmsg.prepare()
file_path = context.content
wav_path = os.path.splitext(file_path)[0] + '.wav'
try:
mp3_to_wav(mp3_path=mp3_path, wav_path=wav_path)
any_to_wav(file_path, wav_path)
except Exception as e: # 转换失败,直接使用mp3,对于某些api,mp3也可以识别
logger.warning("[WX]mp3 to wav error, use mp3 path. " + str(e))
wav_path = mp3_path
logger.warning("[WX]any to wav error, use raw path. " + str(e))
wav_path = file_path
# 语音识别
reply = super().build_voice_to_text(wav_path)
# 删除临时文件
try:
os.remove(file_path)
os.remove(wav_path)
os.remove(mp3_path)
except Exception as e:
logger.warning("[WX]delete temp file error: " + str(e))

@@ -204,7 +204,7 @@ class ChatChannel(Channel):
logger.error('[WX] sendMsg error: {}'.format(e))
if retry_cnt < 2:
time.sleep(3+3*retry_cnt)
self._send(reply, context, retry_cnt+1)
self._send(reply, context, retry_cnt+1)




+ 4
- 10
channel/wechat/wechat_channel.py Просмотреть файл

@@ -8,28 +8,22 @@ import os
import requests
import io
import time
import json
from channel.chat_channel import ChatChannel
from channel.wechat.wechat_message import *
from common.singleton import singleton
from common.log import logger
from lib import itchat
import json
from lib.itchat.content import *
from bridge.reply import *
from bridge.context import *
from channel.channel import Channel
from concurrent.futures import ThreadPoolExecutor
from common.log import logger
from config import conf
from common.time_check import time_checker
from common.expired_dict import ExpiredDict
from plugins import *
try:
from voice.audio_convert import mp3_to_wav
except Exception as e:
pass
thread_pool = ThreadPoolExecutor(max_workers=8)


def thread_pool_callback(worker):
worker_exception = worker.exception()
if worker_exception:
@@ -122,7 +116,7 @@ class WechatChannel(ChatChannel):
@time_checker
@_check
def handle_text(self, cmsg : ChatMessage):
logger.debug("[WX]receive text msg: " + json.dumps(cmsg._rawmsg, ensure_ascii=False))
logger.debug("[WX]receive text msg: {}, cmsg={}".format(json.dumps(cmsg._rawmsg, ensure_ascii=False), cmsg))
context = self._compose_context(ContextType.TEXT, cmsg.content, isgroup=False, msg=cmsg)
if context:
thread_pool.submit(self._handle, context).add_done_callback(thread_pool_callback)
@@ -130,7 +124,7 @@ class WechatChannel(ChatChannel):
@time_checker
@_check
def handle_group(self, cmsg : ChatMessage):
logger.debug("[WX]receive group msg: " + json.dumps(cmsg._rawmsg, ensure_ascii=False))
logger.debug("[WX]receive group msg: {}, cmsg={}".format(json.dumps(cmsg._rawmsg, ensure_ascii=False), cmsg))
context = self._compose_context(ContextType.TEXT, cmsg.content, isgroup=True, msg=cmsg)
if context:
thread_pool.submit(self._handle, context).add_done_callback(thread_pool_callback)


+ 1
- 1
channel/wechat/wechat_message.py Просмотреть файл

@@ -44,7 +44,7 @@ class WeChatMessage(ChatMessage):
self.from_user_nickname = self.other_user_nickname
if self.other_user_id == self.to_user_id:
self.to_user_nickname = self.other_user_nickname
except KeyError as e:
except KeyError as e: # 处理偶尔没有对方信息的情况
logger.warn("[WX]get other_user_id failed: " + str(e))
if self.from_user_id == user_id:
self.other_user_id = self.to_user_id


+ 99
- 304
channel/wechat/wechaty_channel.py Просмотреть файл

@@ -4,21 +4,32 @@
wechaty channel
Python Wechaty - https://github.com/wechaty/python-wechaty
"""
import base64
from concurrent.futures import ThreadPoolExecutor
import os
import time
import asyncio
from typing import Optional, Union
from bridge.context import Context, ContextType
from wechaty_puppet import MessageType, FileBox, ScanStatus # type: ignore
from bridge.context import Context
from wechaty_puppet import FileBox
from wechaty import Wechaty, Contact
from wechaty.user import Message, MiniProgram, UrlLink
from channel.channel import Channel
from wechaty.user import Message
from bridge.reply import *
from bridge.context import *
from channel.chat_channel import ChatChannel
from channel.wechat.wechaty_message import WechatyMessage
from common.log import logger
from common.tmp_dir import TmpDir
from config import conf
from voice.audio_convert import sil_to_wav, mp3_to_sil

class WechatyChannel(Channel):
try:
from voice.audio_convert import mp3_to_sil
except Exception as e:
pass

thread_pool = ThreadPoolExecutor(max_workers=8)
def thread_pool_callback(worker):
worker_exception = worker.exception()
if worker_exception:
logger.exception("Worker return exception: {}".format(worker_exception))
class WechatyChannel(ChatChannel):

def __init__(self):
pass
@@ -28,312 +39,96 @@ class WechatyChannel(Channel):

async def main(self):
config = conf()
# 使用PadLocal协议 比较稳定(免费web协议 os.environ['WECHATY_PUPPET_SERVICE_ENDPOINT'] = '127.0.0.1:8080')
token = config.get('wechaty_puppet_service_token')
os.environ['WECHATY_PUPPET_SERVICE_TOKEN'] = token
global bot
bot = Wechaty()

bot.on('scan', self.on_scan)
bot.on('login', self.on_login)
bot.on('message', self.on_message)
await bot.start()
self.bot = Wechaty()
self.bot.on('login', self.on_login)
self.bot.on('message', self.on_message)
await self.bot.start()

async def on_login(self, contact: Contact):
self.user_id = contact.contact_id
self.name = contact.name
logger.info('[WX] login user={}'.format(contact))

async def on_scan(self, status: ScanStatus, qr_code: Optional[str] = None,
data: Optional[str] = None):
pass
# contact = self.Contact.load(self.contact_id)
# logger.info('[WX] scan user={}, scan status={}, scan qr_code={}'.format(contact, status.name, qr_code))
# print(f'user <{contact}> scan status: {status.name} , 'f'qr_code: {qr_code}')
# 统一的发送函数,每个Channel自行实现,根据reply的type字段发送不同类型的消息
def send(self, reply: Reply, context: Context):
receiver_id = context['receiver']
loop = asyncio.get_event_loop()
if context['isgroup']:
receiver = asyncio.run_coroutine_threadsafe(self.bot.Room.find(receiver_id),loop).result()
else:
receiver = asyncio.run_coroutine_threadsafe(self.bot.Contact.find(receiver_id),loop).result()
msg = None
if reply.type == ReplyType.TEXT:
msg = reply.content
asyncio.run_coroutine_threadsafe(receiver.say(msg),loop).result()
logger.info('[WX] sendMsg={}, receiver={}'.format(reply, receiver))
elif reply.type == ReplyType.ERROR or reply.type == ReplyType.INFO:
msg = reply.content
asyncio.run_coroutine_threadsafe(receiver.say(msg),loop).result()
logger.info('[WX] sendMsg={}, receiver={}'.format(reply, receiver))
elif reply.type == ReplyType.VOICE:
voiceLength = None
if reply.content.endswith('.mp3'):
mp3_file = reply.content
sil_file = os.path.splitext(mp3_file)[0] + '.sil'
voiceLength = mp3_to_sil(mp3_file, sil_file)
try:
os.remove(mp3_file)
except Exception as e:
pass
elif reply.content.endswith('.sil'):
sil_file = reply.content
else:
raise Exception('voice file must be mp3 or sil format')
# 发送语音
t = int(time.time())
msg = FileBox.from_file(sil_file, name=str(t) + '.sil')
if voiceLength is not None:
msg.metadata['voiceLength'] = voiceLength
asyncio.run_coroutine_threadsafe(receiver.say(msg),loop).result()
try:
os.remove(sil_file)
except Exception as e:
pass
logger.info('[WX] sendVoice={}, receiver={}'.format(reply.content, receiver))
elif reply.type == ReplyType.IMAGE_URL: # 从网络下载图片
img_url = reply.content
t = int(time.time())
msg = FileBox.from_url(url=img_url, name=str(t) + '.png')
asyncio.run_coroutine_threadsafe(receiver.say(msg),loop).result()
logger.info('[WX] sendImage url={}, receiver={}'.format(img_url,receiver))
elif reply.type == ReplyType.IMAGE: # 从文件读取图片
image_storage = reply.content
image_storage.seek(0)
t = int(time.time())
msg = FileBox.from_base64(base64.b64encode(image_storage.read()), str(t) + '.png')
asyncio.run_coroutine_threadsafe(receiver.say(msg),loop).result()
logger.info('[WX] sendImage, receiver={}'.format(receiver))

async def on_message(self, msg: Message):
"""
listen for message event
"""
from_contact = msg.talker() # 获取消息的发送者
to_contact = msg.to() # 接收人
room = msg.room() # 获取消息来自的群聊. 如果消息不是来自群聊, 则返回None
from_user_id = from_contact.contact_id
to_user_id = to_contact.contact_id # 接收人id
# other_user_id = msg['User']['UserName'] # 对手方id
content = msg.text()
mention_content = await msg.mention_text() # 返回过滤掉@name后的消息
match_prefix = self.check_prefix(content, conf().get('single_chat_prefix'))
# conversation: Union[Room, Contact] = from_contact if room is None else room

if room is None and msg.type() == MessageType.MESSAGE_TYPE_TEXT:
if not msg.is_self() and match_prefix is not None:
# 好友向自己发送消息
if match_prefix != '':
str_list = content.split(match_prefix, 1)
if len(str_list) == 2:
content = str_list[1].strip()

img_match_prefix = self.check_prefix(content, conf().get('image_create_prefix'))
if img_match_prefix:
content = content.split(img_match_prefix, 1)[1].strip()
await self._do_send_img(content, from_user_id)
else:
await self._do_send(content, from_user_id)
elif msg.is_self() and match_prefix:
# 自己给好友发送消息
str_list = content.split(match_prefix, 1)
if len(str_list) == 2:
content = str_list[1].strip()
img_match_prefix = self.check_prefix(content, conf().get('image_create_prefix'))
if img_match_prefix:
content = content.split(img_match_prefix, 1)[1].strip()
await self._do_send_img(content, to_user_id)
else:
await self._do_send(content, to_user_id)
elif room is None and msg.type() == MessageType.MESSAGE_TYPE_AUDIO:
if not msg.is_self(): # 接收语音消息
# 下载语音文件
voice_file = await msg.to_file_box()
silk_file = TmpDir().path() + voice_file.name
await voice_file.to_file(silk_file)
logger.info("[WX]receive voice file: " + silk_file)
# 将文件转成wav格式音频
wav_file = os.path.splitext(silk_file)[0] + '.wav'
sil_to_wav(silk_file, wav_file)
# 语音识别为文本
query = super().build_voice_to_text(wav_file).content
# 交验关键字
match_prefix = self.check_prefix(query, conf().get('single_chat_prefix'))
if match_prefix is not None:
if match_prefix != '':
str_list = query.split(match_prefix, 1)
if len(str_list) == 2:
query = str_list[1].strip()
# 返回消息
if conf().get('voice_reply_voice'):
await self._do_send_voice(query, from_user_id)
else:
await self._do_send(query, from_user_id)
else:
logger.info("[WX]receive voice check prefix: " + 'False')
# 清除缓存文件
os.remove(wav_file)
os.remove(silk_file)
elif room and msg.type() == MessageType.MESSAGE_TYPE_TEXT:
# 群组&文本消息
room_id = room.room_id
room_name = await room.topic()
from_user_id = from_contact.contact_id
from_user_name = from_contact.name
is_at = await msg.mention_self()
content = mention_content
config = conf()
match_prefix = (is_at and not config.get("group_at_off", False)) \
or self.check_prefix(content, config.get('group_chat_prefix')) \
or self.check_contain(content, config.get('group_chat_keyword'))
# Wechaty判断is_at为True,返回的内容是过滤掉@之后的内容;而is_at为False,则会返回完整的内容
# 故判断如果匹配到自定义前缀,则返回过滤掉前缀+空格后的内容,用于实现类似自定义+前缀触发生成AI图片的功能
prefixes = config.get('group_chat_prefix')
for prefix in prefixes:
if content.startswith(prefix):
content = content.replace(prefix, '', 1).strip()
break
if ('ALL_GROUP' in config.get('group_name_white_list') or room_name in config.get(
'group_name_white_list') or self.check_contain(room_name, config.get(
'group_name_keyword_white_list'))) and match_prefix:
img_match_prefix = self.check_prefix(content, conf().get('image_create_prefix'))
if img_match_prefix:
content = content.split(img_match_prefix, 1)[1].strip()
await self._do_send_group_img(content, room_id)
else:
await self._do_send_group(content, room_id, room_name, from_user_id, from_user_name)
elif room and msg.type() == MessageType.MESSAGE_TYPE_AUDIO:
# 群组&语音消息
room_id = room.room_id
room_name = await room.topic()
from_user_id = from_contact.contact_id
from_user_name = from_contact.name
is_at = await msg.mention_self()
config = conf()
# 是否开启语音识别、群消息响应功能、群名白名单符合等条件
if config.get('group_speech_recognition') and (
'ALL_GROUP' in config.get('group_name_white_list') or room_name in config.get(
'group_name_white_list') or self.check_contain(room_name, config.get(
'group_name_keyword_white_list'))):
# 下载语音文件
voice_file = await msg.to_file_box()
silk_file = TmpDir().path() + voice_file.name
await voice_file.to_file(silk_file)
logger.info("[WX]receive voice file: " + silk_file)
# 将文件转成wav格式音频
wav_file = os.path.splitext(silk_file)[0] + '.wav'
sil_to_wav(silk_file, wav_file)
# 语音识别为文本
query = super().build_voice_to_text(wav_file).content
# 校验关键字
match_prefix = self.check_prefix(query, config.get('group_chat_prefix')) \
or self.check_contain(query, config.get('group_chat_keyword'))
# Wechaty判断is_at为True,返回的内容是过滤掉@之后的内容;而is_at为False,则会返回完整的内容
if match_prefix is not None:
# 故判断如果匹配到自定义前缀,则返回过滤掉前缀+空格后的内容,用于实现类似自定义+前缀触发生成AI图片的功能
prefixes = config.get('group_chat_prefix')
for prefix in prefixes:
if query.startswith(prefix):
query = query.replace(prefix, '', 1).strip()
break
# 返回消息
img_match_prefix = self.check_prefix(query, conf().get('image_create_prefix'))
if img_match_prefix:
query = query.split(img_match_prefix, 1)[1].strip()
await self._do_send_group_img(query, room_id)
elif config.get('voice_reply_voice'):
await self._do_send_group_voice(query, room_id, room_name, from_user_id, from_user_name)
else:
await self._do_send_group(query, room_id, room_name, from_user_id, from_user_name)
else:
logger.info("[WX]receive voice check prefix: " + 'False')
# 清除缓存文件
os.remove(wav_file)
os.remove(silk_file)

async def send(self, message: Union[str, Message, FileBox, Contact, UrlLink, MiniProgram], receiver):
logger.info('[WX] sendMsg={}, receiver={}'.format(message, receiver))
if receiver:
contact = await bot.Contact.find(receiver)
await contact.say(message)

async def send_group(self, message: Union[str, Message, FileBox, Contact, UrlLink, MiniProgram], receiver):
logger.info('[WX] sendMsg={}, receiver={}'.format(message, receiver))
if receiver:
room = await bot.Room.find(receiver)
await room.say(message)

async def _do_send(self, query, reply_user_id):
try:
if not query:
return
context = Context(ContextType.TEXT, query)
context['session_id'] = reply_user_id
reply_text = super().build_reply_content(query, context).content
if reply_text:
await self.send(conf().get("single_chat_reply_prefix") + reply_text, reply_user_id)
except Exception as e:
logger.exception(e)

async def _do_send_voice(self, query, reply_user_id):
try:
if not query:
return
context = Context(ContextType.TEXT, query)
context['session_id'] = reply_user_id
reply_text = super().build_reply_content(query, context).content
if reply_text:
# 转换 mp3 文件为 silk 格式
mp3_file = super().build_text_to_voice(reply_text).content
silk_file = os.path.splitext(mp3_file)[0] + '.sil'
voiceLength = mp3_to_sil(mp3_file, silk_file)
# 发送语音
t = int(time.time())
file_box = FileBox.from_file(silk_file, name=str(t) + '.sil')
file_box.metadata = {'voiceLength': voiceLength}
await self.send(file_box, reply_user_id)
# 清除缓存文件
os.remove(mp3_file)
os.remove(silk_file)
except Exception as e:
logger.exception(e)
async def _do_send_img(self, query, reply_user_id):
try:
if not query:
return
context = Context(ContextType.IMAGE_CREATE, query)
img_url = super().build_reply_content(query, context).content
if not img_url:
return
# 图片下载
# pic_res = requests.get(img_url, stream=True)
# image_storage = io.BytesIO()
# for block in pic_res.iter_content(1024):
# image_storage.write(block)
# image_storage.seek(0)

# 图片发送
logger.info('[WX] sendImage, receiver={}'.format(reply_user_id))
t = int(time.time())
file_box = FileBox.from_url(url=img_url, name=str(t) + '.png')
await self.send(file_box, reply_user_id)
except Exception as e:
logger.exception(e)

async def _do_send_group(self, query, group_id, group_name, group_user_id, group_user_name):
if not query:
cmsg = await WechatyMessage(msg)
except NotImplementedError as e:
logger.debug('[WX] {}'.format(e))
return
context = Context(ContextType.TEXT, query)
group_chat_in_one_session = conf().get('group_chat_in_one_session', [])
if ('ALL_GROUP' in group_chat_in_one_session or \
group_name in group_chat_in_one_session or \
self.check_contain(group_name, group_chat_in_one_session)):
context['session_id'] = str(group_id)
else:
context['session_id'] = str(group_id) + '-' + str(group_user_id)
reply_text = super().build_reply_content(query, context).content
if reply_text:
reply_text = '@' + group_user_name + ' ' + reply_text.strip()
await self.send_group(conf().get("group_chat_reply_prefix", "") + reply_text, group_id)

async def _do_send_group_voice(self, query, group_id, group_name, group_user_id, group_user_name):
if not query:
return
context = Context(ContextType.TEXT, query)
group_chat_in_one_session = conf().get('group_chat_in_one_session', [])
if ('ALL_GROUP' in group_chat_in_one_session or \
group_name in group_chat_in_one_session or \
self.check_contain(group_name, group_chat_in_one_session)):
context['session_id'] = str(group_id)
else:
context['session_id'] = str(group_id) + '-' + str(group_user_id)
reply_text = super().build_reply_content(query, context).content
if reply_text:
reply_text = '@' + group_user_name + ' ' + reply_text.strip()
# 转换 mp3 文件为 silk 格式
mp3_file = super().build_text_to_voice(reply_text).content
silk_file = os.path.splitext(mp3_file)[0] + '.sil'
voiceLength = mp3_to_sil(mp3_file, silk_file)
# 发送语音
t = int(time.time())
file_box = FileBox.from_file(silk_file, name=str(t) + '.silk')
file_box.metadata = {'voiceLength': voiceLength}
await self.send_group(file_box, group_id)
# 清除缓存文件
os.remove(mp3_file)
os.remove(silk_file)

async def _do_send_group_img(self, query, reply_room_id):
try:
if not query:
return
context = Context(ContextType.IMAGE_CREATE, query)
img_url = super().build_reply_content(query, context).content
if not img_url:
return
# 图片发送
logger.info('[WX] sendImage, receiver={}'.format(reply_room_id))
t = int(time.time())
file_box = FileBox.from_url(url=img_url, name=str(t) + '.png')
await self.send_group(file_box, reply_room_id)
except Exception as e:
logger.exception(e)
def check_prefix(self, content, prefix_list):
for prefix in prefix_list:
if content.startswith(prefix):
return prefix
return None
def check_contain(self, content, keyword_list):
if not keyword_list:
return None
for ky in keyword_list:
if content.find(ky) != -1:
return True
return None
logger.exception('[WX] {}'.format(e))
return
logger.debug('[WX] message:{}'.format(cmsg))
room = msg.room() # 获取消息来自的群聊. 如果消息不是来自群聊, 则返回None
isgroup = room is not None
ctype = cmsg.ctype
context = self._compose_context(ctype, cmsg.content, isgroup=isgroup, msg=cmsg)
if context:
logger.info('[WX] receiveMsg={}, context={}'.format(cmsg, context))
thread_pool.submit(self._handle_loop, context, asyncio.get_event_loop()).add_done_callback(thread_pool_callback)

def _handle_loop(self,context,loop):
asyncio.set_event_loop(loop)
self._handle(context)

+ 26
- 6
channel/wechat/wechaty_message.py Просмотреть файл

@@ -1,3 +1,5 @@
import asyncio
import re
from wechaty import MessageType
from bridge.context import ContextType
from channel.chat_message import ChatMessage
@@ -17,7 +19,6 @@ class aobject(object):

async def __init__(self):
pass

class WechatyMessage(ChatMessage, aobject):

async def __init__(self, wechaty_msg: Message):
@@ -36,13 +37,23 @@ class WechatyMessage(ChatMessage, aobject):
self.ctype = ContextType.VOICE
voice_file = await wechaty_msg.to_file_box()
self.content = TmpDir().path() + voice_file.name # content直接存临时目录路径
self._prepare_fn = lambda: voice_file.to_file(self.content)

def func():
loop = asyncio.get_event_loop()
asyncio.run_coroutine_threadsafe(voice_file.to_file(self.content),loop).result()
self._prepare_fn = func
else:
raise NotImplementedError("Unsupported message type: {}".format(wechaty_msg.type()))
from_contact = wechaty_msg.talker() # 获取消息的发送者
self.from_user_id = from_contact.contact_id
self.from_user_nickname = from_contact.name

# group中的from和to,wechaty跟itchat含义不一样
# wecahty: from是消息实际发送者, to:所在群
# itchat: 如果是你发送群消息,from和to是你自己和所在群,如果是别人发群消息,from和to是所在群和你自己
# 但这个差别不影响逻辑,group中只使用到:1.用from来判断是否是自己发的,2.actual_user_id来判断实际发送用户
if self.is_group:
self.to_user_id = room.room_id
@@ -52,14 +63,23 @@ class WechatyMessage(ChatMessage, aobject):
self.to_user_id = to_contact.contact_id
self.to_user_nickname = to_contact.name

if wechaty_msg.is_self():
if self.is_group or wechaty_msg.is_self(): # 如果是群消息,other_user设置为群,如果是私聊消息,而且自己发的,就设置成对方。
self.other_user_id = self.to_user_id
self.other_user_nickname = self.to_user_nickname
else:
self.other_user_id = self.from_user_id
self.other_user_nickname = self.from_user_nickname

if self.is_group:

if self.is_group: # wechaty群聊中,实际发送用户就是from_user
self.is_at = await wechaty_msg.mention_self()
self.actual_user_id = self.other_user_id
self.actual_user_nickname = self.other_user_nickname
if not self.is_at: # 有时候复制粘贴的消息,不算做@,但是内容里面会有@xxx,这里做一下兼容
name = wechaty_msg.wechaty.user_self().name
pattern = f'@{name}(\u2005|\u0020)'
if re.search(pattern,self.content):
logger.debug(f'wechaty message {self.msg_id} include at')
self.is_at = True

self.actual_user_id = self.from_user_id
self.actual_user_nickname = self.from_user_nickname

+ 10
- 0
voice/audio_convert.py Просмотреть файл

@@ -21,6 +21,16 @@ def mp3_to_wav(mp3_path, wav_path):
audio = AudioSegment.from_mp3(mp3_path)
audio.export(wav_path, format="wav")

def any_to_wav(any_path, wav_path):
"""
把任意格式转成wav文件
"""
if any_path.endswith('.wav'):
return
if any_path.endswith('.sil') or any_path.endswith('.silk') or any_path.endswith('.slk'):
return sil_to_wav(any_path, wav_path)
audio = AudioSegment.from_file(any_path)
audio.export(wav_path, format="wav")

def pcm_to_silk(pcm_path, silk_path):
"""


Загрузка…
Отмена
Сохранить