fix: json parse error in docker #600 #601

feat: add options to set voice bot

README.md

@@ -54,6 +54,9 @@
 
 > 项目中使用的对话模型是 davinci，计费方式是约每 750 字 (包含请求和回复) 消耗 $0.02，图片生成是每张消耗 $0.016，账号创建有免费的 $18 额度 (更新3.25: 最新注册的已经无免费额度了)，使用完可以更换邮箱重新注册。
 
+#### 1.1 ChapGPT service On Azure
+一种替换以上的方法是使用Azure推出的[ChatGPT service](https://azure.microsoft.com/en-in/products/cognitive-services/openai-service/)。它host在公有云Azure上，因此不需要VPN就可以直接访问。不过目前仍然处于preview阶段。新用户可以通过Try Azure for free来薅一段时间的羊毛
+
 
 ### 2.运行环境
 
@@ -94,7 +97,7 @@ pip3 install --upgrade openai
 # config.json文件内容示例
 { 
   "open_ai_api_key": "YOUR API KEY",                          # 填入上面创建的 OpenAI API KEY
-  "model": "gpt-3.5-turbo",                                   # 模型名称
+  "model": "gpt-3.5-turbo",                                   # 模型名称。当use_azure_chatgpt为true时，其名称为Azure上model deployment名称
   "proxy": "127.0.0.1:7890",                                  # 代理客户端的ip和端口
   "single_chat_prefix": ["bot", "@bot"],                      # 私聊时文本需要包含该前缀才能触发机器人回复
   "single_chat_reply_prefix": "[bot] ",                       # 私聊时自动回复的前缀，用于区分真人
@@ -104,7 +107,8 @@ pip3 install --upgrade openai
   "image_create_prefix": ["画", "看", "找"],                   # 开启图片回复的前缀
   "conversation_max_tokens": 1000,                            # 支持上下文记忆的最多字符数
   "speech_recognition": false,                                # 是否开启语音识别
-  "character_desc": "你是ChatGPT, 一个由OpenAI训练的大型语言模型, 你旨在回答并解决人们的任何问题，并且可以使用多种语言与人交流。",  # 人格描述
+  "use_azure_chatgpt": false,                                 # 是否使用Azure ChatGPT service代替openai ChatGPT service. 当设置为true时需要设置 open_ai_api_base，如 https://xxx.openai.azure.com/
+  "character_desc": "你是ChatGPT, 一个由OpenAI训练的大型语言模型, 你旨在回答并解决人们的任何问题，并且可以使用多种语言与人交流。",  # 人格描述,
 }
 ```
 **配置说明：**
@@ -173,6 +177,14 @@ nohup python3 app.py & tail -f nohup.out          # 在后台运行程序并通
 
 参考文档 [Docker部署](https://github.com/limccn/chatgpt-on-wechat/wiki/Docker%E9%83%A8%E7%BD%B2)   (Contributed by [limccn](https://github.com/limccn))。
 
+### 4. Railway部署
+[Use with Railway](#use-with-railway)(PaaS, Free, Stable, ✅Recommended)
+> Railway offers $5 (500 hours) of runtime per month
+1. Click the [Railway](https://railway.app/) button to go to the Railway homepage
+2. Click the `Start New Project` button.
+3. Click the `Deploy from Github repo` button.
+4. Choose your repo (you can fork this repo firstly)
+5. Set environment variable to override settings in config-template.json, such as: model, open_ai_api_base, open_ai_api_key, use_azure_chatgpt etc.
 
 ## 常见问题
 

app.py

@@ -5,7 +5,8 @@ from channel import channel_factory
 from common.log import logger
 
 from plugins import *
-if __name__ == '__main__':
+
+def run():
     try:
         # load config
         config.load_config()
@@ -21,3 +22,6 @@ if __name__ == '__main__':
     except Exception as e:
         logger.error("App startup failed!")
         logger.exception(e)
+
+if __name__ == '__main__':
+    run()

bot/bot_factory.py

@@ -24,4 +24,9 @@ def create_bot(bot_type):
         # OpenAI 官方对话模型API
         from bot.openai.open_ai_bot import OpenAIBot
         return OpenAIBot()
+
+    elif bot_type == const.CHATGPTONAZURE:
+        # Azure chatgpt service https://azure.microsoft.com/en-in/products/cognitive-services/openai-service/
+        from bot.chatgpt.chat_gpt_bot import AzureChatGPTBot
+        return AzureChatGPTBot()
     raise RuntimeError

bot/chatgpt/chat_gpt_bot.py

@@ -76,6 +76,16 @@ class ChatGPTBot(Bot):
             reply = Reply(ReplyType.ERROR, 'Bot不支持处理{}类型的消息'.format(context.type))
             return reply
 
+    def compose_args(self):
+        return {
+            "model": conf().get("model") or "gpt-3.5-turbo",  # 对话模型的名称
+            "temperature":conf().get('temperature', 0.9),  # 值在[0,1]之间，越大表示回复越具有不确定性
+            # "max_tokens":4096,  # 回复最大的字符数
+            "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
+        }
+
     def reply_text(self, session, session_id, retry_count=0) -> dict:
         '''
         call openai's ChatCompletion to get the answer
@@ -88,13 +98,7 @@ class ChatGPTBot(Bot):
             if conf().get('rate_limit_chatgpt') and not self.tb4chatgpt.get_token():
                 return {"completion_tokens": 0, "content": "提问太快啦，请休息一下再问我吧"}
             response = openai.ChatCompletion.create(
-                model= conf().get("model") or "gpt-3.5-turbo",  # 对话模型的名称
-                messages=session,
-                temperature=conf().get('temperature', 0.9),  # 值在[0,1]之间，越大表示回复越具有不确定性
-                #max_tokens=4096,  # 回复最大的字符数
-                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
+                messages=session, **self.compose_args()
             )
             # logger.info("[ChatGPT] reply={}, total_tokens={}".format(response.choices[0]['message']['content'], response["usage"]["total_tokens"]))
             return {"total_tokens": response["usage"]["total_tokens"],
@@ -150,6 +154,19 @@ class ChatGPTBot(Bot):
             return False, str(e)
 
 
+class AzureChatGPTBot(ChatGPTBot):
+    def __init__(self):
+        super().__init__()
+        openai.api_type = "azure"
+        openai.api_version = "2023-03-15-preview"
+
+    def compose_args(self):
+        args = super().compose_args()
+        args["engine"] = args["model"]
+        del(args["model"])
+        return args
+
+
 class SessionManager(object):
     def __init__(self):
         if conf().get('expires_in_seconds'):

bridge/bridge.py

@@ -13,12 +13,14 @@ class Bridge(object):
     def __init__(self):
         self.btype={
             "chat": const.CHATGPT,
-            "voice_to_text": "openai",
-            "text_to_voice": "baidu"
+            "voice_to_text": conf().get("voice_to_text", "openai"),
+            "text_to_voice": conf().get("text_to_voice", "baidu")
         }
         model_type = conf().get("model")
         if model_type in ["text-davinci-003"]:
             self.btype['chat'] = const.OPEN_AI
+        if conf().get("use_azure_chatgpt"):
+            self.btype['chat'] = const.CHATGPTONAZURE
         self.bots={}
 
     def get_bot(self,typename):

common/const.py

@@ -1,4 +1,5 @@
 # bot_type
 OPEN_AI = "openAI"
 CHATGPT = "chatGPT"
-BAIDU = "baidu"
+BAIDU = "baidu"
+CHATGPTONAZURE = "chatGPTOnAzure"

config-template.json

@@ -2,6 +2,7 @@
   "open_ai_api_key": "YOUR API KEY",
   "model": "gpt-3.5-turbo",
   "proxy": "",
+  "use_azure_chatgpt": false,
   "single_chat_prefix": ["bot", "@bot"],
   "single_chat_reply_prefix": "[bot] ",
   "group_chat_prefix": ["@bot"],

config.py

@@ -4,17 +4,112 @@ import json
 import os
 from common.log import logger
 
-config = {}
+# 将所有可用的配置项写在字典里
+available_setting ={
+    #openai api配置
+    "open_ai_api_key": "", # openai api key
+    "open_ai_api_base": "https://api.openai.com/v1", # openai apibase，当use_azure_chatgpt为true时，需要设置对应的api base
+    "proxy": "", # openai使用的代理
+    "model": "gpt-3.5-turbo", # chatgpt模型， 当use_azure_chatgpt为true时，其名称为Azure上model deployment名称
+    "use_azure_chatgpt": False, # 是否使用azure的chatgpt
+
+    #Bot触发配置
+    "single_chat_prefix": ["bot", "@bot"], # 私聊时文本需要包含该前缀才能触发机器人回复
+    "single_chat_reply_prefix": "[bot] ", # 私聊时自动回复的前缀，用于区分真人
+    "group_chat_prefix": ["@bot"], # 群聊时包含该前缀则会触发机器人回复
+    "group_chat_reply_prefix": "", # 群聊时自动回复的前缀
+    "group_chat_keyword": [], # 群聊时包含该关键词则会触发机器人回复
+    "group_at_off": False, # 是否关闭群聊时@bot的触发
+    "group_name_white_list": ["ChatGPT测试群", "ChatGPT测试群2"], # 开启自动回复的群名称列表
+    "group_name_keyword_white_list": [], # 开启自动回复的群名称关键词列表
+    "group_chat_in_one_session": ["ChatGPT测试群"], # 支持会话上下文共享的群名称
+    "image_create_prefix": ["画", "看", "找"], # 开启图片回复的前缀
+    
+    #chatgpt会话参数
+    "expires_in_seconds": 3600, # 无操作会话的过期时间
+    "character_desc": "你是ChatGPT, 一个由OpenAI训练的大型语言模型, 你旨在回答并解决人们的任何问题，并且可以使用多种语言与人交流。", # 人格描述
+    "conversation_max_tokens": 1000, # 支持上下文记忆的最多字符数
+            
+    #chatgpt限流配置
+    "rate_limit_chatgpt": 20, # chatgpt的调用频率限制
+    "rate_limit_dalle": 50, # openai dalle的调用频率限制
+
+
+    #chatgpt api参数 参考https://platform.openai.com/docs/api-reference/chat/create
+    "temperature": 0.9,
+    "top_p": 1,
+    "frequency_penalty": 0,
+    "presence_penalty": 0,
+
+    #语音设置
+    "speech_recognition": False, # 是否开启语音识别
+    "voice_reply_voice": False, # 是否使用语音回复语音，需要设置对应语音合成引擎的api key
+    "voice_to_text": "openai", # 语音识别引擎，支持openai和google
+    "text_to_voice": "baidu", # 语音合成引擎，支持baidu和google
+
+    # baidu api的配置， 使用百度语音识别和语音合成时需要
+    'baidu_app_id': "",
+    'baidu_api_key': "",
+    'baidu_secret_key': "",
+
+    #服务时间限制，目前支持itchat
+    "chat_time_module": False, # 是否开启服务时间限制
+    "chat_start_time": "00:00", # 服务开始时间
+    "chat_stop_time": "24:00", # 服务结束时间
+
+    # itchat的配置
+    "hot_reload": False, # 是否开启热重载
+
+    # wechaty的配置
+    "wechaty_puppet_service_token": "", # wechaty的token
+
+    # chatgpt指令自定义触发词
+    "clear_memory_commands": ['#清除记忆'], # 重置会话指令
+
+
+}
+
+class Config(dict):
+    def __getitem__(self, key):
+        if key not in available_setting:
+            raise Exception("key {} not in available_setting".format(key))
+        return super().__getitem__(key)
+
+    def __setitem__(self, key, value):
+        if key not in available_setting:
+            raise Exception("key {} not in available_setting".format(key))
+        return super().__setitem__(key, value)
+
+    def get(self, key, default=None):
+        try :
+            return self[key]
+        except KeyError as e:
+            return default
+        except Exception as e:
+            raise e
+    
+config = Config()
 
 def load_config():
     global config
     config_path = "./config.json"
     if not os.path.exists(config_path):
-        raise Exception('配置文件不存在，请根据config-template.json模板创建config.json文件')
+        logger.info('配置文件不存在，将使用config-template.json模板')
+        config_path = "./config-template.json"
 
     config_str = read_file(config_path)
+    logger.debug("[INIT] config str: {}".format(config_str))
+
     # 将json字符串反序列化为dict类型
-    config = json.loads(config_str)
+    config = Config(json.loads(config_str))
+
+    # override config with environment variables.
+    # Some online deployment platforms (e.g. Railway) deploy project from github directly. So you shouldn't put your secrets like api key in a config file, instead use environment variables to override the default config.
+    for name, value in os.environ.items():
+        if name in available_setting:
+            logger.info("[INIT] override config by environ args: {}={}".format(name, value))
+            config[name] = value
+
     logger.info("[INIT] load config: {}".format(config))
 
 

docker/Dockerfile.alpine

@@ -35,7 +35,7 @@ ADD ./entrypoint.sh /entrypoint.sh
 
 RUN chmod +x /entrypoint.sh \
     && adduser -D -h /home/noroot -u 1000 -s /bin/bash noroot \
-    && chown noroot:noroot ${BUILD_PREFIX}
+    && chown -R noroot:noroot ${BUILD_PREFIX}
 
 USER noroot
 

docker/entrypoint.sh

@@ -84,7 +84,7 @@ if [ "$CHARACTER_DESC" != "" ] ; then
 fi
 
 if [ "$EXPIRES_IN_SECONDS" != "" ] ; then
-    sed -i "s/\"expires_in_seconds\".*$/\"expires_in_seconds\": $EXPIRES_IN_SECONDS/" $CHATGPT_ON_WECHAT_CONFIG_PATH
+    sed -i "s/\"expires_in_seconds\".*$/\"expires_in_seconds\": $EXPIRES_IN_SECONDS,/" $CHATGPT_ON_WECHAT_CONFIG_PATH
 fi
 
 # go to prefix dir

main.py

@@ -0,0 +1,5 @@
+# entry point for online railway deployment
+from app import run
+
+if __name__ == '__main__':
+	run()

plugins/plugin_manager.py

@@ -32,15 +32,15 @@ class PluginManager:
         return wrapper
 
     def save_config(self):
-        with open("plugins/plugins.json", "w", encoding="utf-8") as f:
+        with open("./plugins/plugins.json", "w", encoding="utf-8") as f:
             json.dump(self.pconf, f, indent=4, ensure_ascii=False)
 
     def load_config(self):
         logger.info("Loading plugins config...")
 
         modified = False
-        if os.path.exists("plugins/plugins.json"):
-            with open("plugins/plugins.json", "r", encoding="utf-8") as f:
+        if os.path.exists("./plugins/plugins.json"):
+            with open("./plugins/plugins.json", "r", encoding="utf-8") as f:
                 pconf = json.load(f)
                 pconf['plugins'] = SortedDict(lambda k,v: v["priority"],pconf['plugins'],reverse=True)
         else:
@@ -53,7 +53,7 @@ class PluginManager:
 
     def scan_plugins(self):
         logger.info("Scaning plugins ...")
-        plugins_dir = "plugins"
+        plugins_dir = "./plugins"
         for plugin_name in os.listdir(plugins_dir):
             plugin_path = os.path.join(plugins_dir, plugin_name)
             if os.path.isdir(plugin_path):
@@ -61,7 +61,7 @@ class PluginManager:
                 main_module_path = os.path.join(plugin_path, plugin_name+".py")
                 if os.path.isfile(main_module_path):
                     # 导入插件
-                    import_path = "{}.{}.{}".format(plugins_dir, plugin_name, plugin_name)
+                    import_path = "plugins.{}.{}".format(plugin_name, plugin_name)
                     try:
                         main_module = importlib.import_module(import_path)
                     except Exception as e: