Merge pull request #1488 from yy1781051483/master

add xunfei v3.0

bot/linkai/link_ai_bot.py

@@ -7,13 +7,12 @@ import requests
 
 from bot.bot import Bot
 from bot.chatgpt.chat_gpt_session import ChatGPTSession
-from bot.openai.open_ai_image import OpenAIImage
 from bot.session_manager import SessionManager
 from bridge.context import Context, ContextType
 from bridge.reply import Reply, ReplyType
 from common.log import logger
 from config import conf, pconf
-
+import threading
 
 class LinkAIBot(Bot):
     # authentication failed
@@ -47,10 +46,10 @@ class LinkAIBot(Bot):
         :param retry_count: 当前递归重试次数
         :return: 回复
         """
-        if retry_count >= 2:
+        if retry_count > 2:
             # exit from retry 2 times
             logger.warn("[LINKAI] failed after maximum number of retry times")
-            return Reply(ReplyType.ERROR, "请再问我一次吧")
+            return Reply(ReplyType.TEXT, "请再问我一次吧")
 
         try:
             # load config
@@ -64,7 +63,7 @@ class LinkAIBot(Bot):
             session_id = context["session_id"]
 
             session = self.sessions.session_query(query, session_id)
-            model = conf().get("model") or "gpt-3.5-turbo"
+            model = conf().get("model")
             # remove system message
             if session.messages[0].get("role") == "system":
                 if app_code or model == "wenxin":
@@ -104,6 +103,10 @@ class LinkAIBot(Bot):
                     knowledge_suffix = self._fetch_knowledge_search_suffix(response)
                     if knowledge_suffix:
                         reply_content += knowledge_suffix
+                # image process
+                if response["choices"][0].get("img_urls"):
+                    thread = threading.Thread(target=self._send_image, args=(context.get("channel"), context, response["choices"][0].get("img_urls")))
+                    thread.start()
                 return Reply(ReplyType.TEXT, reply_content)
 
             else:
@@ -118,7 +121,7 @@ class LinkAIBot(Bot):
                     logger.warn(f"[LINKAI] do retry, times={retry_count}")
                     return self._chat(query, context, retry_count + 1)
 
-                return Reply(ReplyType.ERROR, "提问太快啦，请休息一下再问我吧")
+                return Reply(ReplyType.TEXT, "提问太快啦，请休息一下再问我吧")
 
         except Exception as e:
             logger.exception(e)
@@ -262,3 +265,14 @@ class LinkAIBot(Bot):
                 return suffix
         except Exception as e:
             logger.exception(e)
+
+
+    def _send_image(self, channel, context, image_urls):
+        if not image_urls:
+            return
+        try:
+            for url in image_urls:
+                reply = Reply(ReplyType.IMAGE_URL, url)
+                channel.send(reply, context)
+        except Exception as e:
+            logger.error(e)

bot/xunfei/xunfei_spark_bot.py

@@ -40,10 +40,11 @@ class XunFeiBot(Bot):
         self.app_id = conf().get("xunfei_app_id")
         self.api_key = conf().get("xunfei_api_key")
         self.api_secret = conf().get("xunfei_api_secret")
-        # 默认使用v2.0版本，1.5版本可设置为 general
+        # 默认使用v3.0版本，2.0版本可设置为generalv2，  1.5版本可设置为 general
         self.domain = "generalv2"
-        # 默认使用v2.0版本，1.5版本可设置为 "ws://spark-api.xf-yun.com/v1.1/chat"
-        self.spark_url = "ws://spark-api.xf-yun.com/v2.1/chat"
+        # 默认使用v3.0版本，1.5版本可设置为 "ws://spark-api.xf-yun.com/v1.1/chat"，
+        # 2.0版本可设置为 "ws://spark-api.xf-yun.com/v2.1/chat"
+        self.spark_url = "ws://spark-api.xf-yun.com/v3.1/chat"
         self.host = urlparse(self.spark_url).netloc
         self.path = urlparse(self.spark_url).path
         # 和wenxin使用相同的session机制
@@ -56,7 +57,8 @@ class XunFeiBot(Bot):
             request_id = self.gen_request_id(session_id)
             reply_map[request_id] = ""
             session = self.sessions.session_query(query, session_id)
-            threading.Thread(target=self.create_web_socket, args=(session.messages, request_id)).start()
+            threading.Thread(target=self.create_web_socket,
+                             args=(session.messages, request_id)).start()
             depth = 0
             time.sleep(0.1)
             t1 = time.time()
@@ -83,20 +85,27 @@ class XunFeiBot(Bot):
                     depth += 1
                     continue
             t2 = time.time()
-            logger.info(f"[XunFei-API] response={reply_map[request_id]}, time={t2 - t1}s, usage={usage}")
-            self.sessions.session_reply(reply_map[request_id], session_id, usage.get("total_tokens"))
+            logger.info(
+                f"[XunFei-API] response={reply_map[request_id]}, time={t2 - t1}s, usage={usage}"
+            )
+            self.sessions.session_reply(reply_map[request_id], session_id,
+                                        usage.get("total_tokens"))
             reply = Reply(ReplyType.TEXT, reply_map[request_id])
             del reply_map[request_id]
             return reply
         else:
-            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            reply = Reply(ReplyType.ERROR,
+                          "Bot不支持处理{}类型的消息".format(context.type))
             return reply
 
     def create_web_socket(self, prompt, session_id, temperature=0.5):
         logger.info(f"[XunFei] start connect, prompt={prompt}")
         websocket.enableTrace(False)
         wsUrl = self.create_url()
-        ws = websocket.WebSocketApp(wsUrl, on_message=on_message, on_error=on_error, on_close=on_close,
+        ws = websocket.WebSocketApp(wsUrl,
+                                    on_message=on_message,
+                                    on_error=on_error,
+                                    on_close=on_close,
                                     on_open=on_open)
         data_queue = queue.Queue(1000)
         queue_map[session_id] = data_queue
@@ -108,7 +117,8 @@ class XunFeiBot(Bot):
         ws.run_forever(sslopt={"cert_reqs": ssl.CERT_NONE})
 
     def gen_request_id(self, session_id: str):
-        return session_id + "_" + str(int(time.time())) + "" + str(random.randint(0, 100))
+        return session_id + "_" + str(int(time.time())) + "" + str(
+            random.randint(0, 100))
 
     # 生成url
     def create_url(self):
@@ -122,22 +132,21 @@ class XunFeiBot(Bot):
         signature_origin += "GET " + self.path + " HTTP/1.1"
 
         # 进行hmac-sha256进行加密
-        signature_sha = hmac.new(self.api_secret.encode('utf-8'), signature_origin.encode('utf-8'),
+        signature_sha = hmac.new(self.api_secret.encode('utf-8'),
+                                 signature_origin.encode('utf-8'),
                                  digestmod=hashlib.sha256).digest()
 
-        signature_sha_base64 = base64.b64encode(signature_sha).decode(encoding='utf-8')
+        signature_sha_base64 = base64.b64encode(signature_sha).decode(
+            encoding='utf-8')
 
         authorization_origin = f'api_key="{self.api_key}", algorithm="hmac-sha256", headers="host date request-line", ' \
                                f'signature="{signature_sha_base64}"'
 
-        authorization = base64.b64encode(authorization_origin.encode('utf-8')).decode(encoding='utf-8')
+        authorization = base64.b64encode(
+            authorization_origin.encode('utf-8')).decode(encoding='utf-8')
 
         # 将请求的鉴权参数组合为字典
-        v = {
-            "authorization": authorization,
-            "date": date,
-            "host": self.host
-        }
+        v = {"authorization": authorization, "date": date, "host": self.host}
         # 拼接鉴权参数，生成url
         url = self.spark_url + '?' + urlencode(v)
         # 此处打印出建立连接时候的url,参考本demo的时候可取消上方打印的注释，比对相同参数时生成的url与自己代码生成的url是否一致
@@ -190,11 +199,15 @@ def on_close(ws, one, two):
 # 收到websocket连接建立的处理
 def on_open(ws):
     logger.info(f"[XunFei] Start websocket, session_id={ws.session_id}")
-    thread.start_new_thread(run, (ws,))
+    thread.start_new_thread(run, (ws, ))
 
 
 def run(ws, *args):
-    data = json.dumps(gen_params(appid=ws.appid, domain=ws.domain, question=ws.question, temperature=ws.temperature))
+    data = json.dumps(
+        gen_params(appid=ws.appid,
+                   domain=ws.domain,
+                   question=ws.question,
+                   temperature=ws.temperature))
     ws.send(data)
 
 
@@ -212,7 +225,8 @@ def on_message(ws, message):
         content = choices["text"][0]["content"]
         data_queue = queue_map.get(ws.session_id)
         if not data_queue:
-            logger.error(f"[XunFei] can't find data queue, session_id={ws.session_id}")
+            logger.error(
+                f"[XunFei] can't find data queue, session_id={ws.session_id}")
             return
         reply_item = ReplyItem(content)
         if status == 2:

bridge/bridge.py

@@ -18,7 +18,7 @@ class Bridge(object):
             "text_to_voice": conf().get("text_to_voice", "google"),
             "translate": conf().get("translate", "baidu"),
         }
-        model_type = conf().get("model")
+        model_type = conf().get("model") or const.GPT35
         if model_type in ["text-davinci-003"]:
             self.btype["chat"] = const.OPEN_AI
         if conf().get("use_azure_chatgpt", False):

channel/chat_channel.py

@@ -175,6 +175,7 @@ class ChatChannel(Channel):
             if e_context.is_break():
                 context["generate_breaked_by"] = e_context["breaked_by"]
             if context.type == ContextType.TEXT or context.type == ContextType.IMAGE_CREATE:  # 文字和图片消息
+                context["channel"] = e_context["channel"]
                 reply = super().build_reply_content(context.content, context)
             elif context.type == ContextType.VOICE:  # 语音消息
                 cmsg = context["msg"]

common/const.py

@@ -8,6 +8,7 @@ LINKAI = "linkai"
 CLAUDEAI = "claude"
 
 # model
+GPT35 = "gpt-3.5-turbo"
 GPT4 = "gpt-4"
 GPT4_TURBO_PREVIEW = "gpt-4-1106-preview"
 GPT4_VISION_PREVIEW = "gpt-4-vision-preview"

config-template.json

@@ -1,7 +1,7 @@
 {
   "channel_type": "wx",
+  "model": "",
   "open_ai_api_key": "YOUR API KEY",
-  "model": "gpt-3.5-turbo",
   "text_to_image": "dall-e-2",
   "voice_to_text": "openai",
   "text_to_voice": "openai",
@@ -28,8 +28,7 @@
   "speech_recognition": true,
   "group_speech_recognition": false,
   "voice_reply_voice": false,
-  "tts_voice_id": "alloy",
-  "conversation_max_tokens": 1000,
+  "conversation_max_tokens": 2500,
   "expires_in_seconds": 3600,
   "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。",
   "temperature": 0.7,

config.py

@@ -16,7 +16,7 @@ available_setting = {
     "open_ai_api_base": "https://api.openai.com/v1",
     "proxy": "",  # openai使用的代理
     # chatgpt模型， 当use_azure_chatgpt为true时，其名称为Azure上model deployment名称
-    "model": "gpt-3.5-turbo",  # 还支持 gpt-3.5-turbo-16k, gpt-4, wenxin, xunfei
+    "model": "gpt-3.5-turbo",  # 还支持 gpt-4, gpt-4-turbo, wenxin, xunfei
     "use_azure_chatgpt": False,  # 是否使用azure的chatgpt
     "azure_deployment_id": "",  # azure 模型部署名称
     "azure_api_version": "",  # azure api版本
@@ -52,7 +52,7 @@ available_setting = {
     "top_p": 1,
     "frequency_penalty": 0,
     "presence_penalty": 0,
-    "request_timeout": 60,  # chatgpt请求超时时间，openai接口默认设置为600，对于难问题一般需要较长时间
+    "request_timeout": 180,  # chatgpt请求超时时间，openai接口默认设置为600，对于难问题一般需要较长时间
     "timeout": 120,  # chatgpt重试超时时间，在这个时间内，将会自动重试
     # Baidu 文心一言参数
     "baidu_wenxin_model": "eb-instant",  # 默认使用ERNIE-Bot-turbo模型

plugins/godcmd/godcmd.py

@@ -266,14 +266,16 @@ class Godcmd(Plugin):
                     if not isadmin and not self.is_admin_in_group(e_context["context"]):
                         ok, result = False, "需要管理员权限执行"
                     elif len(args) == 0:
-                        ok, result = True, "当前模型为: " + str(conf().get("model"))
+                        model = conf().get("model") or const.GPT35
+                        ok, result = True, "当前模型为: " + str(model)
                     elif len(args) == 1:
                         if args[0] not in const.MODEL_LIST:
                             ok, result = False, "模型名称不存在"
                         else:
                             conf()["model"] = self.model_mapping(args[0])
                             Bridge().reset_bot()
-                            ok, result = True, "模型设置为: " + str(conf().get("model"))
+                            model = conf().get("model") or const.GPT35
+                            ok, result = True, "模型设置为: " + str(model)
                 elif cmd == "id":
                     ok, result = True, user
                 elif cmd == "set_openai_api_key":

plugins/linkai/README.md

@@ -25,7 +25,8 @@
     "summary": {
         "enabled": true,              # 文档总结和对话功能开关
         "group_enabled": true,        # 是否支持群聊开启
-        "max_file_size": 5000        # 文件的大小限制，单位KB，默认为5M，超过该大小直接忽略
+        "max_file_size": 5000,        # 文件的大小限制，单位KB，默认为5M，超过该大小直接忽略
+        "type": ["FILE", "SHARING", "IMAGE"]  # 支持总结的类型，分别表示 文件、分享链接、图片
     }
 }
 ```
@@ -99,7 +100,7 @@
 
 #### 使用
 
-功能开启后，向机器人发送 **文件** 或 **分享链接卡片** 即可生成摘要，进一步可以与文件或链接的内容进行多轮对话。
+功能开启后，向机器人发送 **文件**、 **分享链接卡片**、**图片** 即可生成摘要，进一步可以与文件或链接的内容进行多轮对话。如果需要关闭某种类型的内容总结，设置 `summary`配置中的type字段即可。
 
 #### 限制
 

voice/linkai/linkai_voice.py

@@ -25,9 +25,12 @@ class LinkAIVoice(Voice):
             if not conf().get("text_to_voice") or conf().get("voice_to_text") == "openai":
                 model = const.WHISPER_1
             if voice_file.endswith(".amr"):
-                mp3_file = os.path.splitext(voice_file)[0] + ".mp3"
-                audio_convert.any_to_mp3(voice_file, mp3_file)
-                voice_file = mp3_file
+                try:
+                    mp3_file = os.path.splitext(voice_file)[0] + ".mp3"
+                    audio_convert.any_to_mp3(voice_file, mp3_file)
+                    voice_file = mp3_file
+                except Exception as e:
+                    logger.warn(f"[LinkVoice] amr file transfer failed, directly send amr voice file: {format(e)}")
             file = open(voice_file, "rb")
             file_body = {
                 "file": file
@@ -46,7 +49,7 @@ class LinkAIVoice(Voice):
             logger.info(f"[LinkVoice] voiceToText success, text={text}, file name={voice_file}")
         except Exception as e:
             logger.error(e)
-            reply = Reply(ReplyType.ERROR, "我暂时还无法听清您的语音，请稍后再试吧~")
+            return None
         return reply
 
     def textToVoice(self, text):
@@ -75,5 +78,5 @@ class LinkAIVoice(Voice):
                 return None
         except Exception as e:
             logger.error(e)
-            reply = Reply(ReplyType.ERROR, "遇到了一点小问题，请稍后再问我吧")
-        return reply
+            # reply = Reply(ReplyType.ERROR, "遇到了一点小问题，请稍后再问我吧")
+            return None