Merge branch 'zhayujie:master' into chenzk

README.md

@@ -5,7 +5,7 @@
 最新版本支持的功能如下：
 
 - [x] **多端部署：** 有多种部署方式可选择且功能完备，目前已支持个人微信，微信公众号和企业微信应用等部署方式
-- [x] **基础对话：** 私聊及群聊的消息智能回复，支持多轮会话上下文记忆，支持 GPT-3, GPT-3.5, GPT-4, claude, 文心一言, 讯飞星火
+- [x] **基础对话：** 私聊及群聊的消息智能回复，支持多轮会话上下文记忆，支持 GPT-3.5, GPT-4, claude, 文心一言, 讯飞星火
 - [x] **语音识别：** 可识别语音消息，通过文字或语音回复，支持 azure, baidu, google, openai等多种语音模型
 - [x] **图片生成：** 支持图片生成 和 图生图（如照片修复），可选择 Dell-E, stable diffusion, replicate, midjourney模型
 - [x] **丰富插件：** 支持个性化插件扩展，已实现多角色切换、文字冒险、敏感词过滤、聊天记录总结等插件
@@ -27,7 +27,7 @@ Demo made by [Visionn](https://www.wangpc.cc/)
 <img width="240" src="./docs/images/contact.jpg">
 
 # 更新日志
->**2023.09.01：** 接入讯飞星火，claude机器人
+>**2023.09.01：** 增加 [企微个人号](#1385) 通道，[claude](1388)，讯飞星火模型
 
 >**2023.08.08：** 接入百度文心一言模型，通过 [插件](https://github.com/zhayujie/chatgpt-on-wechat/tree/master/plugins/linkai) 支持 Midjourney 绘图
 
@@ -176,26 +176,6 @@ pip3 install azure-cognitiveservices-speech
 + `linkai_api_key`: LinkAI Api Key，可在 [控制台](https://chat.link-ai.tech/console/interface) 创建
 + `linkai_app_code`: LinkAI 应用code，选填
 
-**6.wenxin配置 (可选 model 为 wenxin 时生效)**
-
-+ `baidu_wenxin_api_key`: 文心一言官网api key。
-+ `baidu_wenxin_secret_key`: 文心一言官网secret key。
-
-
-**6.Claude配置 (可选 model 为 claude 时生效)**
-
-+ `claude_api_cookie`: claude官网聊天界面复制完整 cookie 字符串。
-+ `claude_uuid`: 可以指定对话id，默认新建对话实体。
-
-
-**7.xunfei配置 (可选 model 为 xunfei 时生效)**
-
-+ `xunfei_app_id`: 讯飞星火app id。
-+ `xunfei_api_key`: 讯飞星火 api key。
-+ `xunfei_api_secret`: 讯飞星火 secret。
-
-
-
 **本说明文档可能会未及时更新，当前所有可选的配置项均在该[`config.py`](https://github.com/zhayujie/chatgpt-on-wechat/blob/master/config.py)中列出。**
 
 ## 运行

bot/claude/claude_ai_bot.py

@@ -27,6 +27,7 @@ class ClaudeAIBot(Bot, OpenAIImage):
         }
         else:
             self.proxies = None
+        self.error = ""
         self.org_uuid = self.get_organization_id()
 
     def generate_uuid(self):
@@ -34,32 +35,6 @@ class ClaudeAIBot(Bot, OpenAIImage):
         random_uuid_str = str(random_uuid)
         formatted_uuid = f"{random_uuid_str[0:8]}-{random_uuid_str[9:13]}-{random_uuid_str[14:18]}-{random_uuid_str[19:23]}-{random_uuid_str[24:]}"
         return formatted_uuid
-
-    def get_uuid(self):
-        if conf().get("claude_uuid") != None:
-            self.con_uuid = conf().get("claude_uuid")
-        else:
-            con_uuid = self.generate_uuid()
-            self.create_new_chat(con_uuid)
-
-    def get_organization_id(self):
-        url = "https://claude.ai/api/organizations"
-        headers = {
-            'User-Agent':
-                'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/115.0',
-            'Accept-Language': 'en-US,en;q=0.5',
-            'Referer': 'https://claude.ai/chats',
-            'Content-Type': 'application/json',
-            'Sec-Fetch-Dest': 'empty',
-            'Sec-Fetch-Mode': 'cors',
-            'Sec-Fetch-Site': 'same-origin',
-            'Connection': 'keep-alive',
-            'Cookie': f'{self.claude_api_cookie}'
-        }
-        response = requests.get(url, headers=headers,impersonate="chrome110",proxies=self.proxies)
-        res = json.loads(response.text)
-        uuid = res[0]['uuid']
-        return uuid
         
     def reply(self, query, context: Context = None) -> Reply:
         if context.type == ContextType.TEXT:
@@ -90,20 +65,38 @@ class ClaudeAIBot(Bot, OpenAIImage):
             'Cookie': f'{self.claude_api_cookie}'
         }
         try:
-            response = requests.get(url, headers=headers,impersonate="chrome110",proxies =self.proxies )
+            response = requests.get(url, headers=headers, impersonate="chrome110", proxies =self.proxies, timeout=400)
             res = json.loads(response.text)
             uuid = res[0]['uuid']
         except:
-            print(response.text)
+            if "App unavailable" in response.text:
+                logger.error("IP error: The IP is not allowed to be used on Claude")
+                self.error = "ip所在地区不被claude支持"
+            elif "Invalid authorization" in response.text:
+                logger.error("Cookie error: Invalid authorization of claude, check cookie please.")
+                self.error = "无法通过claude身份验证，请检查cookie"
+            return None
         return uuid
 
     def conversation_share_check(self,session_id):
+        if conf().get("claude_uuid") is not None and conf().get("claude_uuid") != "":
+            con_uuid = conf().get("claude_uuid")
+            return con_uuid
         if session_id not in self.con_uuid_dic:
             self.con_uuid_dic[session_id] = self.generate_uuid()
             self.create_new_chat(self.con_uuid_dic[session_id])
         return self.con_uuid_dic[session_id]
 
+    def check_cookie(self):
+        flag = self.get_organization_id()
+        return flag
+
     def create_new_chat(self, con_uuid):
+        """
+        新建claude对话实体
+        :param con_uuid: 对话id
+        :return:
+        """
         url = f"https://claude.ai/api/organizations/{self.org_uuid}/chat_conversations"
         payload = json.dumps({"uuid": con_uuid, "name": ""})
         headers = {
@@ -121,7 +114,7 @@ class ClaudeAIBot(Bot, OpenAIImage):
             'Sec-Fetch-Site': 'same-origin',
             'TE': 'trailers'
         }
-        response = requests.post(url, headers=headers, data=payload,impersonate="chrome110", proxies= self.proxies)
+        response = requests.post(url, headers=headers, data=payload, impersonate="chrome110", proxies=self.proxies, timeout=400)
         # Returns JSON of the newly created conversation information
         return response.json()
         
@@ -140,8 +133,12 @@ class ClaudeAIBot(Bot, OpenAIImage):
 
         try:
             session_id = context["session_id"]
+            if self.org_uuid is None:
+                return Reply(ReplyType.ERROR, self.error)
+
             session = self.sessions.session_query(query, session_id)
             con_uuid = self.conversation_share_check(session_id)
+
             model = conf().get("model") or "gpt-3.5-turbo"
             # remove system message
             if session.messages[0].get("role") == "system":
@@ -193,11 +190,18 @@ class ClaudeAIBot(Bot, OpenAIImage):
                         completions.append(data['completion'])
 
                 reply_content = ''.join(completions)
-                logger.info(f"[CLAUDE] reply={reply_content}, total_tokens=invisible")
 
+                if "rate limi" in reply_content:
+                    logger.error("rate limit error: The conversation has reached the system speed limit and is synchronized with Cladue. Please go to the official website to check the lifting time")
+                    return Reply(ReplyType.ERROR, "对话达到系统速率限制，与cladue同步，请进入官网查看解除限制时间")
+                logger.info(f"[CLAUDE] reply={reply_content}, total_tokens=invisible")
                 self.sessions.session_reply(reply_content, session_id, 100)
                 return Reply(ReplyType.TEXT, reply_content)
             else:
+                flag = self.check_cookie()
+                if flag == None:
+                    return Reply(ReplyType.ERROR, self.error)
+
                 response = res.json()
                 error = response.get("error")
                 logger.error(f"[CLAUDE] chat failed, status_code={res.status_code}, "

bot/linkai/link_ai_bot.py

@@ -23,6 +23,7 @@ class LinkAIBot(Bot, OpenAIImage):
     def __init__(self):
         super().__init__()
         self.sessions = SessionManager(ChatGPTSession, model=conf().get("model") or "gpt-3.5-turbo")
+        self.args = {}
 
     def reply(self, query, context: Context = None) -> Reply:
         if context.type == ContextType.TEXT:
@@ -72,7 +73,7 @@ class LinkAIBot(Bot, OpenAIImage):
             body = {
                 "app_code": app_code,
                 "messages": session.messages,
-                "model": model,     # 对话模型的名称, 支持 gpt-3.5-turbo, gpt-3.5-turbo-16k, gpt-4, wenxin
+                "model": model,     # 对话模型的名称, 支持 gpt-3.5-turbo, gpt-3.5-turbo-16k, gpt-4, wenxin, xunfei
                 "temperature": conf().get("temperature"),
                 "top_p": conf().get("top_p", 1),
                 "frequency_penalty": conf().get("frequency_penalty", 0.0),  # [-2,2]之间，该值越大则更倾向于产生不同的内容
@@ -114,3 +115,68 @@ class LinkAIBot(Bot, OpenAIImage):
             time.sleep(2)
             logger.warn(f"[LINKAI] do retry, times={retry_count}")
             return self._chat(query, context, retry_count + 1)
+
+    def reply_text(self, session: ChatGPTSession, app_code="", retry_count=0) -> dict:
+        if retry_count >= 2:
+            # exit from retry 2 times
+            logger.warn("[LINKAI] failed after maximum number of retry times")
+            return {
+                "total_tokens": 0,
+                "completion_tokens": 0,
+                "content": "请再问我一次吧"
+            }
+
+        try:
+            body = {
+                "app_code": app_code,
+                "messages": session.messages,
+                "model": conf().get("model") or "gpt-3.5-turbo",  # 对话模型的名称, 支持 gpt-3.5-turbo, gpt-3.5-turbo-16k, gpt-4, wenxin, xunfei
+                "temperature": conf().get("temperature"),
+                "top_p": conf().get("top_p", 1),
+                "frequency_penalty": conf().get("frequency_penalty", 0.0),  # [-2,2]之间，该值越大则更倾向于产生不同的内容
+                "presence_penalty": conf().get("presence_penalty", 0.0),  # [-2,2]之间，该值越大则更倾向于产生不同的内容
+            }
+            if self.args.get("max_tokens"):
+                body["max_tokens"] = self.args.get("max_tokens")
+            headers = {"Authorization": "Bearer " +  conf().get("linkai_api_key")}
+
+            # do http request
+            base_url = conf().get("linkai_api_base", "https://api.link-ai.chat")
+            res = requests.post(url=base_url + "/v1/chat/completions", json=body, headers=headers,
+                                timeout=conf().get("request_timeout", 180))
+            if res.status_code == 200:
+                # execute success
+                response = res.json()
+                reply_content = response["choices"][0]["message"]["content"]
+                total_tokens = response["usage"]["total_tokens"]
+                logger.info(f"[LINKAI] reply={reply_content}, total_tokens={total_tokens}")
+                return {
+                    "total_tokens": total_tokens,
+                    "completion_tokens": response["usage"]["completion_tokens"],
+                    "content": reply_content,
+                }
+
+            else:
+                response = res.json()
+                error = response.get("error")
+                logger.error(f"[LINKAI] chat failed, status_code={res.status_code}, "
+                             f"msg={error.get('message')}, type={error.get('type')}")
+
+                if res.status_code >= 500:
+                    # server error, need retry
+                    time.sleep(2)
+                    logger.warn(f"[LINKAI] do retry, times={retry_count}")
+                    return self.reply_text(session, app_code, retry_count + 1)
+
+                return {
+                    "total_tokens": 0,
+                    "completion_tokens": 0,
+                    "content": "提问太快啦，请休息一下再问我吧"
+                }
+
+        except Exception as e:
+            logger.exception(e)
+            # retry
+            time.sleep(2)
+            logger.warn(f"[LINKAI] do retry, times={retry_count}")
+            return self.reply_text(session, app_code, retry_count + 1)

channel/wechatmp/wechatmp_channel.py

@@ -20,7 +20,7 @@ from common.log import logger
 from common.singleton import singleton
 from common.utils import split_string_by_utf8_length
 from config import conf
-from voice.audio_convert import any_to_mp3
+from voice.audio_convert import any_to_mp3, split_audio
 
 # If using SSL, uncomment the following lines, and modify the certificate path.
 # from cheroot.server import HTTPServer
@@ -162,13 +162,28 @@ class WechatMPChannel(ChatChannel):
                         file_name = os.path.basename(file_path)
                         file_type = "audio/mpeg"
                     logger.info("[wechatmp] file_name: {}, file_type: {} ".format(file_name, file_type))
-                    # support: <2M, <60s, AMR\MP3
-                    response = self.client.media.upload("voice", (file_name, open(file_path, "rb"), file_type))
-                    logger.debug("[wechatmp] upload voice response: {}".format(response))
+                    media_ids = []
+                    duration, files = split_audio(file_path, 60 * 1000)
+                    if len(files) > 1:
+                        logger.info("[wechatmp] voice too long {}s > 60s , split into {} parts".format(duration / 1000.0, len(files)))
+                    for path in files:
+                        # support: <2M, <60s, AMR\MP3
+                        response = self.client.media.upload("voice", (os.path.basename(path), open(path, "rb"), file_type))
+                        logger.debug("[wechatcom] upload voice response: {}".format(response))
+                        media_ids.append(response["media_id"])
+                        os.remove(path)
                 except WeChatClientException as e:
                     logger.error("[wechatmp] upload voice failed: {}".format(e))
                     return
-                self.client.message.send_voice(receiver, response["media_id"])
+
+                try:
+                    os.remove(file_path)
+                except Exception:
+                    pass
+
+                for media_id in media_ids:
+                    self.client.message.send_voice(receiver, media_id)
+                    time.sleep(1)
                 logger.info("[wechatmp] Do send voice to {}".format(receiver))
             elif reply.type == ReplyType.IMAGE_URL:  # 从网络下载图片
                 img_url = reply.content

channel/wework/wework_channel.py

@@ -16,6 +16,7 @@ from channel.wework.wework_message import WeworkMessage
 from common.singleton import singleton
 from common.log import logger
 from common.time_check import time_checker
+from common.utils import compress_imgfile, fsize
 from config import conf
 from channel.wework.run import wework
 from channel.wework import run
@@ -38,12 +39,25 @@ def download_and_compress_image(url, filename, quality=30):
         os.makedirs(directory)
 
     # 下载图片
-    response = requests.get(url)
-    image = Image.open(io.BytesIO(response.content))
-
-    # 压缩图片
-    image_path = os.path.join(directory, f"{filename}.jpg")
-    image.save(image_path, "JPEG", quality=quality)
+    pic_res = requests.get(url, stream=True)
+    image_storage = io.BytesIO()
+    for block in pic_res.iter_content(1024):
+        image_storage.write(block)
+
+    # 检查图片大小并可能进行压缩
+    sz = fsize(image_storage)
+    if sz >= 10 * 1024 * 1024:  # 如果图片大于 10 MB
+        logger.info("[wework] image too large, ready to compress, sz={}".format(sz))
+        image_storage = compress_imgfile(image_storage, 10 * 1024 * 1024 - 1)
+        logger.info("[wework] image compressed, sz={}".format(fsize(image_storage)))
+
+    # 将内存缓冲区的指针重置到起始位置
+    image_storage.seek(0)
+
+    # 读取并保存图片
+    image = Image.open(image_storage)
+    image_path = os.path.join(directory, f"{filename}.png")
+    image.save(image_path, "png")
 
     return image_path
 
@@ -213,6 +227,9 @@ class WeworkChannel(ChatChannel):
     @time_checker
     @_check
     def handle_single(self, cmsg: ChatMessage):
+        if cmsg.from_user_id == cmsg.to_user_id:
+            # ignore self reply
+            return
         if cmsg.ctype == ContextType.VOICE:
             if not conf().get("speech_recognition"):
                 return

plugins/linkai/README.md

@@ -4,15 +4,15 @@
 
 ## 插件配置
 
-将 `plugins/linkai` 目录下的 `config.json.template` 配置模板复制为最终生效的 `config.json`:
+将 `plugins/linkai` 目录下的 `config.json.template` 配置模板复制为最终生效的 `config.json`。 (如果未配置则会默认使用`config.json.template`模板中配置，功能默认关闭，可通过指令进行开启)。
 
 以下是配置项说明：
 
 ```bash
 {
     "group_app_map": {            # 群聊 和 应用编码 的映射关系
-        "测试群1": "default",      # 表示在名称为 "测试群1" 的群聊中将使用app_code 为 default 的应用
-        "测试群2": "Kv2fXJcH"
+        "测试群名称1": "default",      # 表示在名称为 "测试群名称1" 的群聊中将使用app_code 为 default 的应用
+        "测试群名称2": "Kv2fXJcH"
     },
     "midjourney": {
         "enabled": true,          # midjourney 绘画开关
@@ -51,6 +51,8 @@
 
 ### 2.Midjourney绘画功能
 
+若未配置 `plugins/linkai/config.json`，默认会关闭画图功能，直接使用 `$mj open` 可基于默认配置直接使用mj画图。
+
 指令格式：
 
 ```
@@ -69,7 +71,9 @@
 "$mjr 11055927171882"
 ```
 
-注：
-1. 开启 `use_image_create_prefix` 配置后可直接复用全局画图触发词，以"画"开头便可以生成图片。
-2. 提示词内容中包含敏感词或者参数格式错误可能导致绘画失败，生成失败不消耗积分
-3. 使用 `$mj open` 和 `$mj close` 指令可以快速打开和关闭绘图功能
+注意事项：
+1. 使用 `$mj open` 和 `$mj close` 指令可以快速打开和关闭绘图功能
+2. 海外环境部署请将 `img_proxy` 设置为 `False`
+3. 开启 `use_image_create_prefix` 配置后可直接复用全局画图触发词，以"画"开头便可以生成图片。
+4. 提示词内容中包含敏感词或者参数格式错误可能导致绘画失败，生成失败不消耗积分
+5. 若未收到图片可能有两种可能，一种是收到了图片但微信发送失败，可以在后台日志查看有没有获取到图片url，一般原因是受到了wx限制，可以稍后重试或更换账号尝试；另一种情况是图片提示词存在疑似违规，mj不会直接提示错误但会在画图后删掉原图导致程序无法获取，这种情况不消耗积分。

plugins/linkai/config.json.template

@@ -1,7 +1,7 @@
 {
     "group_app_map": {
-        "测试群1": "default",
-        "测试群2": "Kv2fXJcH"
+        "测试群名1": "default",
+        "测试群名2": "Kv2fXJcH"
     },
     "midjourney": {
         "enabled": true,

plugins/linkai/linkai.py

@@ -18,6 +18,9 @@ class LinkAI(Plugin):
         super().__init__()
         self.handlers[Event.ON_HANDLE_CONTEXT] = self.on_handle_context
         self.config = super().load_config()
+        if not self.config:
+            # 未加载到配置，使用模板中的配置
+            self.config = self._load_config_template()
         if self.config:
             self.mj_bot = MJBot(self.config.get("midjourney"))
         logger.info("[LinkAI] inited")
@@ -70,7 +73,7 @@ class LinkAI(Plugin):
                 is_open = False
             conf()["use_linkai"] = is_open
             bridge.Bridge().reset_bot()
-            _set_reply_text(f"知识库功能已{tips_text}", e_context, level=ReplyType.INFO)
+            _set_reply_text(f"LinkAI对话功能{tips_text}", e_context, level=ReplyType.INFO)
             return
 
         if len(cmd) == 3 and cmd[1] == "app":
@@ -139,6 +142,17 @@ class LinkAI(Plugin):
         help_text += f"\n\"{trigger_prefix}mjv 11055927171882 2\"\n\"{trigger_prefix}mjr 11055927171882\""
         return help_text
 
+    def _load_config_template(self):
+        logger.debug("No LinkAI plugin config.json, use plugins/linkai/config.json.template")
+        try:
+            plugin_config_path = os.path.join(self.path, "config.json.template")
+            if os.path.exists(plugin_config_path):
+                with open(plugin_config_path, "r", encoding="utf-8") as f:
+                    plugin_conf = json.load(f)
+                    plugin_conf["midjourney"]["enabled"] = False
+                    return plugin_conf
+        except Exception as e:
+            logger.exception(e)
 
 # 静态方法
 def _is_admin(e_context: EventContext) -> bool:

plugins/linkai/midjourney.py

@@ -96,7 +96,7 @@ class MJBot:
                 return TaskType.VARIATION
             elif cmd_list[0].lower() == f"{trigger_prefix}mjr":
                 return TaskType.RESET
-        elif context.type == ContextType.IMAGE_CREATE and self.config.get("use_image_create_prefix"):
+        elif context.type == ContextType.IMAGE_CREATE and self.config.get("use_image_create_prefix") and self.config.get("enabled"):
             return TaskType.GENERATE
 
     def process_mj_task(self, mj_type: TaskType, e_context: EventContext):

plugins/plugin.py

@@ -15,8 +15,8 @@ class Plugin:
         """
         # 优先获取 plugins/config.json 中的全局配置
         plugin_conf = pconf(self.name)
-        if not plugin_conf or not conf().get("use_global_plugin_config"):
-            # 全局配置不存在 或者 未开启全局配置开关，则获取插件目录下的配置
+        if not plugin_conf:
+            # 全局配置不存在，则获取插件目录下的配置
             plugin_config_path = os.path.join(self.path, "config.json")
             if os.path.exists(plugin_config_path):
                 with open(plugin_config_path, "r", encoding="utf-8") as f:

requirements-optional.txt

@@ -23,7 +23,6 @@ web.py
 wechatpy
 
 # chatgpt-tool-hub plugin
---extra-index-url https://pypi.python.org/simple
 chatgpt_tool_hub==0.4.6
 
 # xunfei spark