chore: the bot directory was changed to models

models/ali/ali_qwen_bot.py

@@ -0,0 +1,214 @@
+# encoding:utf-8
+
+import json
+import time
+from typing import List, Tuple
+
+import openai
+import openai.error
+import broadscope_bailian
+from broadscope_bailian import ChatQaMessage
+
+from models.bot import Bot
+from models.ali.ali_qwen_session import AliQwenSession
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from common import const
+from config import conf, load_config
+
+class AliQwenBot(Bot):
+    def __init__(self):
+        super().__init__()
+        self.api_key_expired_time = self.set_api_key()
+        self.sessions = SessionManager(AliQwenSession, model=conf().get("model", const.QWEN))
+
+    def api_key_client(self):
+        return broadscope_bailian.AccessTokenClient(access_key_id=self.access_key_id(), access_key_secret=self.access_key_secret())
+
+    def access_key_id(self):
+        return conf().get("qwen_access_key_id")
+
+    def access_key_secret(self):
+        return conf().get("qwen_access_key_secret")
+
+    def agent_key(self):
+        return conf().get("qwen_agent_key")
+
+    def app_id(self):
+        return conf().get("qwen_app_id")
+
+    def node_id(self):
+        return conf().get("qwen_node_id", "")
+
+    def temperature(self):
+        return conf().get("temperature", 0.2 )
+
+    def top_p(self):
+        return conf().get("top_p", 1)
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context.type == ContextType.TEXT:
+            logger.info("[QWEN] query={}".format(query))
+
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[QWEN] session query={}".format(session.messages))
+
+            reply_content = self.reply_text(session)
+            logger.debug(
+                "[QWEN] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[QWEN] reply {} used 0 tokens.".format(reply_content))
+            return reply
+
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: AliQwenSession, retry_count=0) -> dict:
+        """
+        call bailian's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            prompt, history = self.convert_messages_format(session.messages)
+            self.update_api_key_if_expired()
+            # NOTE 阿里百炼的call()函数未提供temperature参数，考虑到temperature和top_p参数作用相同，取两者较小的值作为top_p参数传入，详情见文档 https://help.aliyun.com/document_detail/2587502.htm
+            response = broadscope_bailian.Completions().call(app_id=self.app_id(), prompt=prompt, history=history, top_p=min(self.temperature(), self.top_p()))
+            completion_content = self.get_completion_content(response, self.node_id())
+            completion_tokens, total_tokens = self.calc_tokens(session.messages, completion_content)
+            return {
+                "total_tokens": total_tokens,
+                "completion_tokens": completion_tokens,
+                "content": completion_content,
+            }
+        except Exception as e:
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if isinstance(e, openai.error.RateLimitError):
+                logger.warn("[QWEN] RateLimitError: {}".format(e))
+                result["content"] = "提问太快啦，请休息一下再问我吧"
+                if need_retry:
+                    time.sleep(20)
+            elif isinstance(e, openai.error.Timeout):
+                logger.warn("[QWEN] Timeout: {}".format(e))
+                result["content"] = "我没有收到你的消息"
+                if need_retry:
+                    time.sleep(5)
+            elif isinstance(e, openai.error.APIError):
+                logger.warn("[QWEN] Bad Gateway: {}".format(e))
+                result["content"] = "请再问我一次"
+                if need_retry:
+                    time.sleep(10)
+            elif isinstance(e, openai.error.APIConnectionError):
+                logger.warn("[QWEN] APIConnectionError: {}".format(e))
+                need_retry = False
+                result["content"] = "我连接不到你的网络"
+            else:
+                logger.exception("[QWEN] Exception: {}".format(e))
+                need_retry = False
+                self.sessions.clear_session(session.session_id)
+
+            if need_retry:
+                logger.warn("[QWEN] 第{}次重试".format(retry_count + 1))
+                return self.reply_text(session, retry_count + 1)
+            else:
+                return result
+
+    def set_api_key(self):
+        api_key, expired_time = self.api_key_client().create_token(agent_key=self.agent_key())
+        broadscope_bailian.api_key = api_key
+        return expired_time
+
+    def update_api_key_if_expired(self):
+        if time.time() > self.api_key_expired_time:
+            self.api_key_expired_time = self.set_api_key()
+
+    def convert_messages_format(self, messages) -> Tuple[str, List[ChatQaMessage]]:
+        history = []
+        user_content = ''
+        assistant_content = ''
+        system_content = ''
+        for message in messages:
+            role = message.get('role')
+            if role == 'user':
+                user_content += message.get('content')
+            elif role == 'assistant':
+                assistant_content = message.get('content')
+                history.append(ChatQaMessage(user_content, assistant_content))
+                user_content = ''
+                assistant_content = ''
+            elif role =='system':
+                system_content += message.get('content')
+        if user_content == '':
+            raise Exception('no user message')
+        if system_content != '':
+            # NOTE 模拟系统消息，测试发现人格描述以"你需要扮演ChatGPT"开头能够起作用，而以"你是ChatGPT"开头模型会直接否认
+            system_qa = ChatQaMessage(system_content, '好的，我会严格按照你的设定回答问题')
+            history.insert(0, system_qa)
+        logger.debug("[QWEN] converted qa messages: {}".format([item.to_dict() for item in history]))
+        logger.debug("[QWEN] user content as prompt: {}".format(user_content))
+        return user_content, history
+
+    def get_completion_content(self, response, node_id):
+        if not response['Success']:
+            return f"[ERROR]\n{response['Code']}:{response['Message']}"
+        text = response['Data']['Text']
+        if node_id == '':
+            return text
+        # TODO: 当使用流程编排创建大模型应用时，响应结构如下，最终结果在['finalResult'][node_id]['response']['text']中，暂时先这么写
+        # {
+        #     'Success': True,
+        #     'Code': None,
+        #     'Message': None,
+        #     'Data': {
+        #         'ResponseId': '9822f38dbacf4c9b8daf5ca03a2daf15',
+        #         'SessionId': 'session_id',
+        #         'Text': '{"finalResult":{"LLM_T7islK":{"params":{"modelId":"qwen-plus-v1","prompt":"${systemVars.query}${bizVars.Text}"},"response":{"text":"作为一个AI语言模型，我没有年龄，因为我没有生日。\n我只是一个程序，没有生命和身体。"}}}}',
+        #         'Thoughts': [],
+        #         'Debug': {},
+        #         'DocReferences': []
+        #     },
+        #     'RequestId': '8e11d31551ce4c3f83f49e6e0dd998b0',
+        #     'Failed': None
+        # }
+        text_dict = json.loads(text)
+        completion_content =  text_dict['finalResult'][node_id]['response']['text']
+        return completion_content
+
+    def calc_tokens(self, messages, completion_content):
+        completion_tokens = len(completion_content)
+        prompt_tokens = 0
+        for message in messages:
+            prompt_tokens += len(message["content"])
+        return completion_tokens, prompt_tokens + completion_tokens

models/ali/ali_qwen_session.py

@@ -0,0 +1,62 @@
+from models.session_manager import Session
+from common.log import logger
+
+"""
+    e.g.
+    [
+        {"role": "system", "content": "You are a helpful assistant."},
+        {"role": "user", "content": "Who won the world series in 2020?"},
+        {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."},
+        {"role": "user", "content": "Where was it played?"}
+    ]
+"""
+
+class AliQwenSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="qianwen"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        self.reset()
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "user":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens, len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+def num_tokens_from_messages(messages, model):
+    """Returns the number of tokens used by a list of messages."""
+    # 官方token计算规则："对于中文文本来说，1个token通常对应一个汉字；对于英文文本来说，1个token通常对应3至4个字母或1个单词"
+    # 详情请产看文档：https://help.aliyun.com/document_detail/2586397.html
+    # 目前根据字符串长度粗略估计token数，不影响正常使用
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["content"])
+    return tokens

models/baidu/baidu_unit_bot.py

@@ -0,0 +1,36 @@
+# encoding:utf-8
+
+import requests
+
+from models.bot import Bot
+from bridge.reply import Reply, ReplyType
+
+
+# Baidu Unit对话接口 (可用, 但能力较弱)
+class BaiduUnitBot(Bot):
+    def reply(self, query, context=None):
+        token = self.get_token()
+        url = "https://aip.baidubce.com/rpc/2.0/unit/service/v3/chat?access_token=" + token
+        post_data = (
+            '{"version":"3.0","service_id":"S73177","session_id":"","log_id":"7758521","skill_ids":["1221886"],"request":{"terminal_id":"88888","query":"'
+            + query
+            + '", "hyper_params": {"chat_custom_bot_profile": 1}}}'
+        )
+        print(post_data)
+        headers = {"content-type": "application/x-www-form-urlencoded"}
+        response = requests.post(url, data=post_data.encode(), headers=headers)
+        if response:
+            reply = Reply(
+                ReplyType.TEXT,
+                response.json()["result"]["context"]["SYS_PRESUMED_HIST"][1],
+            )
+            return reply
+
+    def get_token(self):
+        access_key = "YOUR_ACCESS_KEY"
+        secret_key = "YOUR_SECRET_KEY"
+        host = "https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id=" + access_key + "&client_secret=" + secret_key
+        response = requests.get(host)
+        if response:
+            print(response.json())
+            return response.json()["access_token"]

models/baidu/baidu_wenxin.py

@@ -0,0 +1,120 @@
+# encoding:utf-8
+
+import requests
+import json
+from common import const
+from models.bot import Bot
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf
+from models.baidu.baidu_wenxin_session import BaiduWenxinSession
+
+BAIDU_API_KEY = conf().get("baidu_wenxin_api_key")
+BAIDU_SECRET_KEY = conf().get("baidu_wenxin_secret_key")
+
+class BaiduWenxinBot(Bot):
+
+    def __init__(self):
+        super().__init__()
+        wenxin_model = conf().get("baidu_wenxin_model")
+        self.prompt_enabled = conf().get("baidu_wenxin_prompt_enabled")
+        if self.prompt_enabled:
+            self.prompt = conf().get("character_desc", "")
+            if self.prompt == "":
+                logger.warn("[BAIDU] Although you enabled model prompt, character_desc is not specified.")
+        if wenxin_model is not None:
+            wenxin_model = conf().get("baidu_wenxin_model") or "eb-instant"
+        else:
+            if conf().get("model") and conf().get("model") == const.WEN_XIN:
+                wenxin_model = "completions"
+            elif conf().get("model") and conf().get("model") == const.WEN_XIN_4:
+                wenxin_model = "completions_pro"
+
+        self.sessions = SessionManager(BaiduWenxinSession, model=wenxin_model)
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context and context.type:
+            if context.type == ContextType.TEXT:
+                logger.info("[BAIDU] query={}".format(query))
+                session_id = context["session_id"]
+                reply = None
+                if query == "#清除记忆":
+                    self.sessions.clear_session(session_id)
+                    reply = Reply(ReplyType.INFO, "记忆已清除")
+                elif query == "#清除所有":
+                    self.sessions.clear_all_session()
+                    reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+                else:
+                    session = self.sessions.session_query(query, session_id)
+                    result = self.reply_text(session)
+                    total_tokens, completion_tokens, reply_content = (
+                        result["total_tokens"],
+                        result["completion_tokens"],
+                        result["content"],
+                    )
+                    logger.debug(
+                        "[BAIDU] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(session.messages, session_id, reply_content, completion_tokens)
+                    )
+
+                    if total_tokens == 0:
+                        reply = Reply(ReplyType.ERROR, reply_content)
+                    else:
+                        self.sessions.session_reply(reply_content, session_id, total_tokens)
+                        reply = Reply(ReplyType.TEXT, reply_content)
+                return reply
+            elif context.type == ContextType.IMAGE_CREATE:
+                ok, retstring = self.create_img(query, 0)
+                reply = None
+                if ok:
+                    reply = Reply(ReplyType.IMAGE_URL, retstring)
+                else:
+                    reply = Reply(ReplyType.ERROR, retstring)
+                return reply
+
+    def reply_text(self, session: BaiduWenxinSession, retry_count=0):
+        try:
+            logger.info("[BAIDU] model={}".format(session.model))
+            access_token = self.get_access_token()
+            if access_token == 'None':
+                logger.warn("[BAIDU] access token 获取失败")
+                return {
+                    "total_tokens": 0,
+                    "completion_tokens": 0,
+                    "content": 0,
+                    }
+            url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/" + session.model + "?access_token=" + access_token
+            headers = {
+                'Content-Type': 'application/json'
+            }
+            payload = {'messages': session.messages, 'system': self.prompt} if self.prompt_enabled else {'messages': session.messages}
+            response = requests.request("POST", url, headers=headers, data=json.dumps(payload))
+            response_text = json.loads(response.text)
+            logger.info(f"[BAIDU] response text={response_text}")
+            res_content = response_text["result"]
+            total_tokens = response_text["usage"]["total_tokens"]
+            completion_tokens = response_text["usage"]["completion_tokens"]
+            logger.info("[BAIDU] reply={}".format(res_content))
+            return {
+                "total_tokens": total_tokens,
+                "completion_tokens": completion_tokens,
+                "content": res_content,
+            }
+        except Exception as e:
+            need_retry = retry_count < 2
+            logger.warn("[BAIDU] Exception: {}".format(e))
+            need_retry = False
+            self.sessions.clear_session(session.session_id)
+            result = {"total_tokens": 0, "completion_tokens": 0, "content": "出错了: {}".format(e)}
+            return result
+
+    def get_access_token(self):
+        """
+        使用 AK，SK 生成鉴权签名（Access Token）
+        :return: access_token，或是None(如果错误)
+        """
+        url = "https://aip.baidubce.com/oauth/2.0/token"
+        params = {"grant_type": "client_credentials", "client_id": BAIDU_API_KEY, "client_secret": BAIDU_SECRET_KEY}
+        return str(requests.post(url, params=params).json().get("access_token"))

models/baidu/baidu_wenxin_session.py

@@ -0,0 +1,53 @@
+from models.session_manager import Session
+from common.log import logger
+
+"""
+    e.g.  [
+        {"role": "user", "content": "Who won the world series in 2020?"},
+        {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."},
+        {"role": "user", "content": "Where was it played?"}
+    ]
+"""
+
+
+class BaiduWenxinSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="gpt-3.5-turbo"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        # 百度文心不支持system prompt
+        # self.reset()
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) >= 2:
+                self.messages.pop(0)
+                self.messages.pop(0)
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens, len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+
+def num_tokens_from_messages(messages, model):
+    """Returns the number of tokens used by a list of messages."""
+    tokens = 0
+    for msg in messages:
+        # 官方token计算规则暂不明确： "大约为 token数为 "中文字 + 其他语种单词数 x 1.3"
+        # 这里先直接根据字数粗略估算吧，暂不影响正常使用，仅在判断是否丢弃历史会话的时候会有偏差
+        tokens += len(msg["content"])
+    return tokens

models/bot.py

@@ -0,0 +1,17 @@
+"""
+Auto-replay chat robot abstract class
+"""
+
+
+from bridge.context import Context
+from bridge.reply import Reply
+
+
+class Bot(object):
+    def reply(self, query, context: Context = None) -> Reply:
+        """
+        bot auto-reply content
+        :param req: received message
+        :return: reply content
+        """
+        raise NotImplementedError

models/bot_factory.py

@@ -0,0 +1,76 @@
+"""
+channel factory
+"""
+from common import const
+
+
+def create_bot(bot_type):
+    """
+    create a bot_type instance
+    :param bot_type: bot type code
+    :return: bot instance
+    """
+    if bot_type == const.BAIDU:
+        # 替换Baidu Unit为Baidu文心千帆对话接口
+        # from models.baidu.baidu_unit_bot import BaiduUnitBot
+        # return BaiduUnitBot()
+        from models.baidu.baidu_wenxin import BaiduWenxinBot
+        return BaiduWenxinBot()
+
+    elif bot_type == const.CHATGPT:
+        # ChatGPT 网页端web接口
+        from models.chatgpt.chat_gpt_bot import ChatGPTBot
+        return ChatGPTBot()
+
+    elif bot_type == const.OPEN_AI:
+        # OpenAI 官方对话模型API
+        from models.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 models.chatgpt.chat_gpt_bot import AzureChatGPTBot
+        return AzureChatGPTBot()
+
+    elif bot_type == const.XUNFEI:
+        from models.xunfei.xunfei_spark_bot import XunFeiBot
+        return XunFeiBot()
+
+    elif bot_type == const.LINKAI:
+        from models.linkai.link_ai_bot import LinkAIBot
+        return LinkAIBot()
+
+    elif bot_type == const.CLAUDEAI:
+        from models.claude.claude_ai_bot import ClaudeAIBot
+        return ClaudeAIBot()
+    elif bot_type == const.CLAUDEAPI:
+        from models.claudeapi.claude_api_bot import ClaudeAPIBot
+        return ClaudeAPIBot()
+    elif bot_type == const.QWEN:
+        from models.ali.ali_qwen_bot import AliQwenBot
+        return AliQwenBot()
+    elif bot_type == const.QWEN_DASHSCOPE:
+        from models.dashscope.dashscope_bot import DashscopeBot
+        return DashscopeBot()
+    elif bot_type == const.GEMINI:
+        from models.gemini.google_gemini_bot import GoogleGeminiBot
+        return GoogleGeminiBot()
+
+    elif bot_type == const.ZHIPU_AI:
+        from models.zhipuai.zhipuai_bot import ZHIPUAIBot
+        return ZHIPUAIBot()
+
+    elif bot_type == const.MOONSHOT:
+        from models.moonshot.moonshot_bot import MoonshotBot
+        return MoonshotBot()
+    
+    elif bot_type == const.MiniMax:
+        from models.minimax.minimax_bot import MinimaxBot
+        return MinimaxBot()
+
+    elif bot_type == const.MODELSCOPE:
+        from models.modelscope.modelscope_bot import ModelScopeBot
+        return ModelScopeBot()
+
+
+    raise RuntimeError

models/chatgpt/chat_gpt_bot.py

@@ -0,0 +1,262 @@
+# encoding:utf-8
+
+import time
+import json
+
+import openai
+import openai.error
+import requests
+from common import const
+from models.bot import Bot
+from models.openai_compatible_bot import OpenAICompatibleBot
+from models.chatgpt.chat_gpt_session import ChatGPTSession
+from models.openai.open_ai_image import OpenAIImage
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from common.token_bucket import TokenBucket
+from config import conf, load_config
+from models.baidu.baidu_wenxin_session import BaiduWenxinSession
+
+# OpenAI对话模型API (可用)
+class ChatGPTBot(Bot, OpenAIImage, OpenAICompatibleBot):
+    def __init__(self):
+        super().__init__()
+        # set the default api_key
+        openai.api_key = conf().get("open_ai_api_key")
+        if conf().get("open_ai_api_base"):
+            openai.api_base = conf().get("open_ai_api_base")
+        proxy = conf().get("proxy")
+        if proxy:
+            openai.proxy = proxy
+        if conf().get("rate_limit_chatgpt"):
+            self.tb4chatgpt = TokenBucket(conf().get("rate_limit_chatgpt", 20))
+        conf_model = conf().get("model") or "gpt-3.5-turbo"
+        self.sessions = SessionManager(ChatGPTSession, model=conf().get("model") or "gpt-3.5-turbo")
+        # o1相关模型不支持system prompt，暂时用文心模型的session
+
+        self.args = {
+            "model": conf_model,  # 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": 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
+            "request_timeout": conf().get("request_timeout", None),  # 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": conf().get("request_timeout", None),  # 重试超时时间，在这个时间内，将会自动重试
+        }
+        # 部分模型暂不支持一些参数，特殊处理
+        if conf_model in [const.O1, const.O1_MINI, const.GPT_5, const.GPT_5_MINI, const.GPT_5_NANO]:
+            remove_keys = ["temperature", "top_p", "frequency_penalty", "presence_penalty"]
+            for key in remove_keys:
+                self.args.pop(key, None)  # 如果键不存在，使用 None 来避免抛出错、
+            if conf_model in [const.O1, const.O1_MINI]:  # o1系列模型不支持系统提示词，使用文心模型的session
+                self.sessions = SessionManager(BaiduWenxinSession, model=conf().get("model") or const.O1_MINI)
+
+    def get_api_config(self):
+        """Get API configuration for OpenAI-compatible base class"""
+        return {
+            'api_key': conf().get("open_ai_api_key"),
+            'api_base': conf().get("open_ai_api_base"),
+            'model': conf().get("model", "gpt-3.5-turbo"),
+            'default_temperature': conf().get("temperature", 0.9),
+            'default_top_p': conf().get("top_p", 1.0),
+            'default_frequency_penalty': conf().get("frequency_penalty", 0.0),
+            'default_presence_penalty': conf().get("presence_penalty", 0.0),
+        }
+    
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context.type == ContextType.TEXT:
+            logger.info("[CHATGPT] query={}".format(query))
+
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[CHATGPT] session query={}".format(session.messages))
+
+            api_key = context.get("openai_api_key")
+            model = context.get("gpt_model")
+            new_args = None
+            if model:
+                new_args = self.args.copy()
+                new_args["model"] = model
+            # if context.get('stream'):
+            #     # reply in stream
+            #     return self.reply_text_stream(query, new_query, session_id)
+
+            reply_content = self.reply_text(session, api_key, args=new_args)
+            logger.debug(
+                "[CHATGPT] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[CHATGPT] reply {} used 0 tokens.".format(reply_content))
+            return reply
+
+        elif context.type == ContextType.IMAGE_CREATE:
+            ok, retstring = self.create_img(query, 0)
+            reply = None
+            if ok:
+                reply = Reply(ReplyType.IMAGE_URL, retstring)
+            else:
+                reply = Reply(ReplyType.ERROR, retstring)
+            return reply
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: ChatGPTSession, api_key=None, args=None, retry_count=0) -> dict:
+        """
+        call openai's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            if conf().get("rate_limit_chatgpt") and not self.tb4chatgpt.get_token():
+                raise openai.error.RateLimitError("RateLimitError: rate limit exceeded")
+            # if api_key == None, the default openai.api_key will be used
+            if args is None:
+                args = self.args
+            response = openai.ChatCompletion.create(api_key=api_key, messages=session.messages, **args)
+            # logger.debug("[CHATGPT] response={}".format(response))
+            logger.info("[ChatGPT] reply={}, total_tokens={}".format(response.choices[0]['message']['content'], response["usage"]["total_tokens"]))
+            return {
+                "total_tokens": response["usage"]["total_tokens"],
+                "completion_tokens": response["usage"]["completion_tokens"],
+                "content": response.choices[0]["message"]["content"],
+            }
+        except Exception as e:
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if isinstance(e, openai.error.RateLimitError):
+                logger.warn("[CHATGPT] RateLimitError: {}".format(e))
+                result["content"] = "提问太快啦，请休息一下再问我吧"
+                if need_retry:
+                    time.sleep(20)
+            elif isinstance(e, openai.error.Timeout):
+                logger.warn("[CHATGPT] Timeout: {}".format(e))
+                result["content"] = "我没有收到你的消息"
+                if need_retry:
+                    time.sleep(5)
+            elif isinstance(e, openai.error.APIError):
+                logger.warn("[CHATGPT] Bad Gateway: {}".format(e))
+                result["content"] = "请再问我一次"
+                if need_retry:
+                    time.sleep(10)
+            elif isinstance(e, openai.error.APIConnectionError):
+                logger.warn("[CHATGPT] APIConnectionError: {}".format(e))
+                result["content"] = "我连接不到你的网络"
+                if need_retry:
+                    time.sleep(5)
+            else:
+                logger.exception("[CHATGPT] Exception: {}".format(e))
+                need_retry = False
+                self.sessions.clear_session(session.session_id)
+
+            if need_retry:
+                logger.warn("[CHATGPT] 第{}次重试".format(retry_count + 1))
+                return self.reply_text(session, api_key, args, retry_count + 1)
+            else:
+                return result
+
+class AzureChatGPTBot(ChatGPTBot):
+    def __init__(self):
+        super().__init__()
+        openai.api_type = "azure"
+        openai.api_version = conf().get("azure_api_version", "2023-06-01-preview")
+        self.args["deployment_id"] = conf().get("azure_deployment_id")
+
+    def create_img(self, query, retry_count=0, api_key=None):
+        text_to_image_model = conf().get("text_to_image")
+        if text_to_image_model == "dall-e-2":
+            api_version = "2023-06-01-preview"
+            endpoint = conf().get("azure_openai_dalle_api_base","open_ai_api_base")
+            # 检查endpoint是否以/结尾
+            if not endpoint.endswith("/"):
+                endpoint = endpoint + "/"
+            url = "{}openai/images/generations:submit?api-version={}".format(endpoint, api_version)
+            api_key = conf().get("azure_openai_dalle_api_key","open_ai_api_key")
+            headers = {"api-key": api_key, "Content-Type": "application/json"}
+            try:
+                body = {"prompt": query, "size": conf().get("image_create_size", "256x256"),"n": 1}
+                submission = requests.post(url, headers=headers, json=body)
+                operation_location = submission.headers['operation-location']
+                status = ""
+                while (status != "succeeded"):
+                    if retry_count > 3:
+                        return False, "图片生成失败"
+                    response = requests.get(operation_location, headers=headers)
+                    status = response.json()['status']
+                    retry_count += 1
+                image_url = response.json()['result']['data'][0]['url']
+                return True, image_url
+            except Exception as e:
+                logger.error("create image error: {}".format(e))
+                return False, "图片生成失败"
+        elif text_to_image_model == "dall-e-3":
+            api_version = conf().get("azure_api_version", "2024-02-15-preview")
+            endpoint = conf().get("azure_openai_dalle_api_base","open_ai_api_base")
+            # 检查endpoint是否以/结尾
+            if not endpoint.endswith("/"):
+                endpoint = endpoint + "/"
+            url = "{}openai/deployments/{}/images/generations?api-version={}".format(endpoint, conf().get("azure_openai_dalle_deployment_id","text_to_image"),api_version)
+            api_key = conf().get("azure_openai_dalle_api_key","open_ai_api_key")
+            headers = {"api-key": api_key, "Content-Type": "application/json"}
+            try:
+                body = {"prompt": query, "size": conf().get("image_create_size", "1024x1024"), "quality": conf().get("dalle3_image_quality", "standard")}
+                response = requests.post(url, headers=headers, json=body)
+                response.raise_for_status()  # 检查请求是否成功
+                data = response.json()
+
+                # 检查响应中是否包含图像 URL
+                if 'data' in data and len(data['data']) > 0 and 'url' in data['data'][0]:
+                    image_url = data['data'][0]['url']
+                    return True, image_url
+                else:
+                    error_message = "响应中没有图像 URL"
+                    logger.error(error_message)
+                    return False, "图片生成失败"
+
+            except requests.exceptions.RequestException as e:
+                # 捕获所有请求相关的异常
+                try:
+                    error_detail = response.json().get('error', {}).get('message', str(e))
+                except ValueError:
+                    error_detail = str(e)
+                error_message = f"{error_detail}"
+                logger.error(error_message)
+                return False, error_message
+
+            except Exception as e:
+                # 捕获所有其他异常
+                error_message = f"生成图像时发生错误: {e}"
+                logger.error(error_message)
+                return False, "图片生成失败"
+        else:
+            return False, "图片生成失败，未配置text_to_image参数"

models/chatgpt/chat_gpt_session.py

@@ -0,0 +1,104 @@
+from models.session_manager import Session
+from common.log import logger
+from common import const
+
+"""
+    e.g.  [
+        {"role": "system", "content": "You are a helpful assistant."},
+        {"role": "user", "content": "Who won the world series in 2020?"},
+        {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."},
+        {"role": "user", "content": "Where was it played?"}
+    ]
+"""
+
+
+class ChatGPTSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="gpt-3.5-turbo"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        self.reset()
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "user":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens, len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+
+# refer to https://github.com/openai/openai-cookbook/blob/main/examples/How_to_count_tokens_with_tiktoken.ipynb
+def num_tokens_from_messages(messages, model):
+    """Returns the number of tokens used by a list of messages."""
+
+    if model in ["wenxin", "xunfei"] or model.startswith(const.GEMINI):
+        return num_tokens_by_character(messages)
+
+    import tiktoken
+
+    if model in ["gpt-3.5-turbo-0301", "gpt-35-turbo", "gpt-3.5-turbo-1106", "moonshot", const.LINKAI_35]:
+        return num_tokens_from_messages(messages, model="gpt-3.5-turbo")
+    elif model in ["gpt-4-0314", "gpt-4-0613", "gpt-4-32k", "gpt-4-32k-0613", "gpt-3.5-turbo-0613",
+                   "gpt-3.5-turbo-16k", "gpt-3.5-turbo-16k-0613", "gpt-35-turbo-16k", "gpt-4-turbo-preview",
+                   "gpt-4-1106-preview", const.GPT4_TURBO_PREVIEW, const.GPT4_VISION_PREVIEW, const.GPT4_TURBO_01_25,
+                   const.GPT_4o, const.GPT_4O_0806, const.GPT_4o_MINI, const.LINKAI_4o, const.LINKAI_4_TURBO, const.GPT_5, const.GPT_5_MINI, const.GPT_5_NANO]:
+        return num_tokens_from_messages(messages, model="gpt-4")
+    elif model.startswith("claude-3"):
+        return num_tokens_from_messages(messages, model="gpt-3.5-turbo")
+    try:
+        encoding = tiktoken.encoding_for_model(model)
+    except KeyError:
+        logger.debug("Warning: model not found. Using cl100k_base encoding.")
+        encoding = tiktoken.get_encoding("cl100k_base")
+    if model == "gpt-3.5-turbo":
+        tokens_per_message = 4  # every message follows <|start|>{role/name}\n{content}<|end|>\n
+        tokens_per_name = -1  # if there's a name, the role is omitted
+    elif model == "gpt-4":
+        tokens_per_message = 3
+        tokens_per_name = 1
+    else:
+        logger.debug(f"num_tokens_from_messages() is not implemented for model {model}. Returning num tokens assuming gpt-3.5-turbo.")
+        return num_tokens_from_messages(messages, model="gpt-3.5-turbo")
+    num_tokens = 0
+    for message in messages:
+        num_tokens += tokens_per_message
+        for key, value in message.items():
+            num_tokens += len(encoding.encode(value))
+            if key == "name":
+                num_tokens += tokens_per_name
+    num_tokens += 3  # every reply is primed with <|start|>assistant<|message|>
+    return num_tokens
+
+
+def num_tokens_by_character(messages):
+    """Returns the number of tokens used by a list of messages."""
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["content"])
+    return tokens

models/claudeapi/claude_api_bot.py

@@ -0,0 +1,485 @@
+# encoding:utf-8
+
+import json
+import time
+
+import requests
+
+from models.baidu.baidu_wenxin_session import BaiduWenxinSession
+from models.bot import Bot
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common import const
+from common.log import logger
+from config import conf
+
+# Optional OpenAI image support
+try:
+    from models.openai.open_ai_image import OpenAIImage
+    _openai_image_available = True
+except Exception as e:
+    logger.warning(f"OpenAI image support not available: {e}")
+    _openai_image_available = False
+    OpenAIImage = object  # Fallback to object
+
+user_session = dict()
+
+
+# OpenAI对话模型API (可用)
+class ClaudeAPIBot(Bot, OpenAIImage):
+    def __init__(self):
+        super().__init__()
+        self.api_key = conf().get("claude_api_key")
+        self.api_base = conf().get("open_ai_api_base") or "https://api.anthropic.com/v1"
+        self.proxy = conf().get("proxy", None)
+        self.sessions = SessionManager(BaiduWenxinSession, model=conf().get("model") or "text-davinci-003")
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context and context.type:
+            if context.type == ContextType.TEXT:
+                logger.info("[CLAUDE_API] query={}".format(query))
+                session_id = context["session_id"]
+                reply = None
+                if query == "#清除记忆":
+                    self.sessions.clear_session(session_id)
+                    reply = Reply(ReplyType.INFO, "记忆已清除")
+                elif query == "#清除所有":
+                    self.sessions.clear_all_session()
+                    reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+                else:
+                    session = self.sessions.session_query(query, session_id)
+                    result = self.reply_text(session)
+                    logger.info(result)
+                    total_tokens, completion_tokens, reply_content = (
+                        result["total_tokens"],
+                        result["completion_tokens"],
+                        result["content"],
+                    )
+                    logger.debug(
+                        "[CLAUDE_API] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(str(session), session_id, reply_content, completion_tokens)
+                    )
+
+                    if total_tokens == 0:
+                        reply = Reply(ReplyType.ERROR, reply_content)
+                    else:
+                        self.sessions.session_reply(reply_content, session_id, total_tokens)
+                        reply = Reply(ReplyType.TEXT, reply_content)
+                return reply
+            elif context.type == ContextType.IMAGE_CREATE:
+                ok, retstring = self.create_img(query, 0)
+                reply = None
+                if ok:
+                    reply = Reply(ReplyType.IMAGE_URL, retstring)
+                else:
+                    reply = Reply(ReplyType.ERROR, retstring)
+                return reply
+
+    def reply_text(self, session: BaiduWenxinSession, retry_count=0, tools=None):
+        try:
+            actual_model = self._model_mapping(conf().get("model"))
+
+            # Prepare headers
+            headers = {
+                "x-api-key": self.api_key,
+                "anthropic-version": "2023-06-01",
+                "content-type": "application/json"
+            }
+
+            # Extract system prompt if present and prepare Claude-compatible messages
+            system_prompt = conf().get("character_desc", "")
+            claude_messages = []
+
+            for msg in session.messages:
+                if msg.get("role") == "system":
+                    system_prompt = msg["content"]
+                else:
+                    claude_messages.append(msg)
+
+            # Prepare request data
+            data = {
+                "model": actual_model,
+                "messages": claude_messages,
+                "max_tokens": self._get_max_tokens(actual_model)
+            }
+
+            if system_prompt:
+                data["system"] = system_prompt
+
+            if tools:
+                data["tools"] = tools
+
+            # Make HTTP request
+            proxies = {"http": self.proxy, "https": self.proxy} if self.proxy else None
+            response = requests.post(
+                f"{self.api_base}/messages",
+                headers=headers,
+                json=data,
+                proxies=proxies
+            )
+
+            if response.status_code != 200:
+                raise Exception(f"API request failed: {response.status_code} - {response.text}")
+
+            claude_response = response.json()
+            # Handle response content and tool calls
+            res_content = ""
+            tool_calls = []
+
+            content_blocks = claude_response.get("content", [])
+            for block in content_blocks:
+                if block.get("type") == "text":
+                    res_content += block.get("text", "")
+                elif block.get("type") == "tool_use":
+                    tool_calls.append({
+                        "id": block.get("id", ""),
+                        "name": block.get("name", ""),
+                        "arguments": block.get("input", {})
+                    })
+
+            res_content = res_content.strip().replace("<|endoftext|>", "")
+            usage = claude_response.get("usage", {})
+            total_tokens = usage.get("input_tokens", 0) + usage.get("output_tokens", 0)
+            completion_tokens = usage.get("output_tokens", 0)
+
+            logger.info("[CLAUDE_API] reply={}".format(res_content))
+            if tool_calls:
+                logger.info("[CLAUDE_API] tool_calls={}".format(tool_calls))
+
+            result = {
+                "total_tokens": total_tokens,
+                "completion_tokens": completion_tokens,
+                "content": res_content,
+            }
+
+            if tool_calls:
+                result["tool_calls"] = tool_calls
+
+            return result
+        except Exception as e:
+            need_retry = retry_count < 2
+            result = {"total_tokens": 0, "completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+
+            # Handle different types of errors
+            error_str = str(e).lower()
+            if "rate" in error_str or "limit" in error_str:
+                logger.warn("[CLAUDE_API] RateLimitError: {}".format(e))
+                result["content"] = "提问太快啦，请休息一下再问我吧"
+                if need_retry:
+                    time.sleep(20)
+            elif "timeout" in error_str:
+                logger.warn("[CLAUDE_API] Timeout: {}".format(e))
+                result["content"] = "我没有收到你的消息"
+                if need_retry:
+                    time.sleep(5)
+            elif "connection" in error_str or "network" in error_str:
+                logger.warn("[CLAUDE_API] APIConnectionError: {}".format(e))
+                need_retry = False
+                result["content"] = "我连接不到你的网络"
+            else:
+                logger.warn("[CLAUDE_API] Exception: {}".format(e))
+                need_retry = False
+                self.sessions.clear_session(session.session_id)
+
+            if need_retry:
+                logger.warn("[CLAUDE_API] 第{}次重试".format(retry_count + 1))
+                return self.reply_text(session, retry_count + 1, tools)
+            else:
+                return result
+
+    def _model_mapping(self, model) -> str:
+        if model == "claude-3-opus":
+            return const.CLAUDE_3_OPUS
+        elif model == "claude-3-sonnet":
+            return const.CLAUDE_3_SONNET
+        elif model == "claude-3-haiku":
+            return const.CLAUDE_3_HAIKU
+        elif model == "claude-3.5-sonnet":
+            return const.CLAUDE_35_SONNET
+        return model
+
+    def _get_max_tokens(self, model: str) -> int:
+        """
+        Get max_tokens for the model.
+        Reference from pi-mono:
+        - Claude 3.5/3.7: 8192
+        - Claude 3 Opus: 4096
+        - Default: 8192
+        """
+        if model and (model.startswith("claude-3-5") or model.startswith("claude-3-7")):
+            return 8192
+        elif model and model.startswith("claude-3") and "opus" in model:
+            return 4096
+        elif model and (model.startswith("claude-sonnet-4") or model.startswith("claude-opus-4")):
+            return 64000
+        return 8192
+
+    def call_with_tools(self, messages, tools=None, stream=False, **kwargs):
+        """
+        Call Claude API with tool support for agent integration
+
+        Args:
+            messages: List of messages
+            tools: List of tool definitions
+            stream: Whether to use streaming
+            **kwargs: Additional parameters
+            
+        Returns:
+            Formatted response compatible with OpenAI format or generator for streaming
+        """
+        actual_model = self._model_mapping(conf().get("model"))
+
+        # Extract system prompt from messages if present
+        system_prompt = kwargs.get("system", conf().get("character_desc", ""))
+        claude_messages = []
+
+        for msg in messages:
+            if msg.get("role") == "system":
+                system_prompt = msg["content"]
+            else:
+                claude_messages.append(msg)
+
+        request_params = {
+            "model": actual_model,
+            "max_tokens": kwargs.get("max_tokens", self._get_max_tokens(actual_model)),
+            "messages": claude_messages,
+            "stream": stream
+        }
+
+        if system_prompt:
+            request_params["system"] = system_prompt
+
+        if tools:
+            request_params["tools"] = tools
+
+        try:
+            if stream:
+                return self._handle_stream_response(request_params)
+            else:
+                return self._handle_sync_response(request_params)
+        except Exception as e:
+            logger.error(f"Claude API call error: {e}")
+            if stream:
+                # Return error generator for stream
+                def error_generator():
+                    yield {
+                        "error": True,
+                        "message": str(e),
+                        "status_code": 500
+                    }
+
+                return error_generator()
+            else:
+                # Return error response for sync
+                return {
+                    "error": True,
+                    "message": str(e),
+                    "status_code": 500
+                }
+
+    def _handle_sync_response(self, request_params):
+        """Handle synchronous Claude API response"""
+        # Prepare headers
+        headers = {
+            "x-api-key": self.api_key,
+            "anthropic-version": "2023-06-01",
+            "content-type": "application/json"
+        }
+
+        # Make HTTP request
+        proxies = {"http": self.proxy, "https": self.proxy} if self.proxy else None
+        response = requests.post(
+            f"{self.api_base}/messages",
+            headers=headers,
+            json=request_params,
+            proxies=proxies
+        )
+
+        if response.status_code != 200:
+            raise Exception(f"API request failed: {response.status_code} - {response.text}")
+
+        claude_response = response.json()
+
+        # Extract content blocks
+        text_content = ""
+        tool_calls = []
+
+        content_blocks = claude_response.get("content", [])
+        for block in content_blocks:
+            if block.get("type") == "text":
+                text_content += block.get("text", "")
+            elif block.get("type") == "tool_use":
+                tool_calls.append({
+                    "id": block.get("id", ""),
+                    "type": "function",
+                    "function": {
+                        "name": block.get("name", ""),
+                        "arguments": json.dumps(block.get("input", {}))
+                    }
+                })
+
+        # Build message in OpenAI format
+        message = {
+            "role": "assistant",
+            "content": text_content
+        }
+        if tool_calls:
+            message["tool_calls"] = tool_calls
+
+        # Format response to match OpenAI structure
+        usage = claude_response.get("usage", {})
+        formatted_response = {
+            "id": claude_response.get("id", ""),
+            "object": "chat.completion",
+            "created": int(time.time()),
+            "model": claude_response.get("model", request_params["model"]),
+            "choices": [
+                {
+                    "index": 0,
+                    "message": message,
+                    "finish_reason": claude_response.get("stop_reason", "stop")
+                }
+            ],
+            "usage": {
+                "prompt_tokens": usage.get("input_tokens", 0),
+                "completion_tokens": usage.get("output_tokens", 0),
+                "total_tokens": usage.get("input_tokens", 0) + usage.get("output_tokens", 0)
+            }
+        }
+
+        return formatted_response
+
+    def _handle_stream_response(self, request_params):
+        """Handle streaming Claude API response using HTTP requests"""
+        # Prepare headers
+        headers = {
+            "x-api-key": self.api_key,
+            "anthropic-version": "2023-06-01",
+            "content-type": "application/json"
+        }
+
+        # Add stream parameter
+        request_params["stream"] = True
+
+        # Track tool use state
+        tool_uses_map = {}  # {index: {id, name, input}}
+        current_tool_use_index = -1
+
+        try:
+            # Make streaming HTTP request
+            proxies = {"http": self.proxy, "https": self.proxy} if self.proxy else None
+            response = requests.post(
+                f"{self.api_base}/messages",
+                headers=headers,
+                json=request_params,
+                proxies=proxies,
+                stream=True
+            )
+
+            if response.status_code != 200:
+                error_text = response.text
+                try:
+                    error_data = json.loads(error_text)
+                    error_msg = error_data.get("error", {}).get("message", error_text)
+                except:
+                    error_msg = error_text or "Unknown error"
+
+                yield {
+                    "error": True,
+                    "status_code": response.status_code,
+                    "message": error_msg
+                }
+                return
+
+            # Process streaming response
+            for line in response.iter_lines():
+                if line:
+                    line = line.decode('utf-8')
+                    if line.startswith('data: '):
+                        line = line[6:]  # Remove 'data: ' prefix
+                        if line == '[DONE]':
+                            break
+                        try:
+                            event = json.loads(line)
+                            event_type = event.get("type")
+
+                            if event_type == "content_block_start":
+                                # New content block
+                                block = event.get("content_block", {})
+                                if block.get("type") == "tool_use":
+                                    current_tool_use_index = event.get("index", 0)
+                                    tool_uses_map[current_tool_use_index] = {
+                                        "id": block.get("id", ""),
+                                        "name": block.get("name", ""),
+                                        "input": ""
+                                    }
+
+                            elif event_type == "content_block_delta":
+                                delta = event.get("delta", {})
+                                delta_type = delta.get("type")
+
+                                if delta_type == "text_delta":
+                                    # Text content
+                                    content = delta.get("text", "")
+                                    yield {
+                                        "id": event.get("id", ""),
+                                        "object": "chat.completion.chunk",
+                                        "created": int(time.time()),
+                                        "model": request_params["model"],
+                                        "choices": [{
+                                            "index": 0,
+                                            "delta": {"content": content},
+                                            "finish_reason": None
+                                        }]
+                                    }
+
+                                elif delta_type == "input_json_delta":
+                                    # Tool input accumulation
+                                    if current_tool_use_index >= 0:
+                                        tool_uses_map[current_tool_use_index]["input"] += delta.get("partial_json", "")
+
+                            elif event_type == "message_delta":
+                                # Message complete - yield tool calls if any
+                                if tool_uses_map:
+                                    for idx in sorted(tool_uses_map.keys()):
+                                        tool_data = tool_uses_map[idx]
+                                        yield {
+                                            "id": event.get("id", ""),
+                                            "object": "chat.completion.chunk",
+                                            "created": int(time.time()),
+                                            "model": request_params["model"],
+                                            "choices": [{
+                                                "index": 0,
+                                                "delta": {
+                                                    "tool_calls": [{
+                                                        "index": idx,
+                                                        "id": tool_data["id"],
+                                                        "type": "function",
+                                                        "function": {
+                                                            "name": tool_data["name"],
+                                                            "arguments": tool_data["input"]
+                                                        }
+                                                    }]
+                                                },
+                                                "finish_reason": None
+                                            }]
+                                        }
+
+                        except json.JSONDecodeError:
+                            continue
+
+        except requests.RequestException as e:
+            logger.error(f"Claude streaming request error: {e}")
+            yield {
+                "error": True,
+                "message": f"Connection error: {str(e)}",
+                "status_code": 0
+            }
+        except Exception as e:
+            logger.error(f"Claude streaming error: {e}")
+            yield {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }

models/dashscope/dashscope_bot.py

@@ -0,0 +1,117 @@
+# encoding:utf-8
+
+from models.bot import Bot
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf, load_config
+from .dashscope_session import DashscopeSession
+import os
+import dashscope
+from http import HTTPStatus
+
+
+
+dashscope_models = {
+    "qwen-turbo": dashscope.Generation.Models.qwen_turbo,
+    "qwen-plus": dashscope.Generation.Models.qwen_plus,
+    "qwen-max": dashscope.Generation.Models.qwen_max,
+    "qwen-bailian-v1": dashscope.Generation.Models.bailian_v1
+}
+# ZhipuAI对话模型API
+class DashscopeBot(Bot):
+    def __init__(self):
+        super().__init__()
+        self.sessions = SessionManager(DashscopeSession, model=conf().get("model") or "qwen-plus")
+        self.model_name = conf().get("model") or "qwen-plus"
+        self.api_key = conf().get("dashscope_api_key")
+        os.environ["DASHSCOPE_API_KEY"] = self.api_key
+        self.client = dashscope.Generation
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context.type == ContextType.TEXT:
+            logger.info("[DASHSCOPE] query={}".format(query))
+
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[DASHSCOPE] session query={}".format(session.messages))
+
+            reply_content = self.reply_text(session)
+            logger.debug(
+                "[DASHSCOPE] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[DASHSCOPE] reply {} used 0 tokens.".format(reply_content))
+            return reply
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: DashscopeSession, retry_count=0) -> dict:
+        """
+        call openai's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            dashscope.api_key = self.api_key
+            response = self.client.call(
+                dashscope_models[self.model_name],
+                messages=session.messages,
+                result_format="message"
+            )
+            if response.status_code == HTTPStatus.OK:
+                content = response.output.choices[0]["message"]["content"]
+                return {
+                    "total_tokens": response.usage["total_tokens"],
+                    "completion_tokens": response.usage["output_tokens"],
+                    "content": content,
+                }
+            else:
+                logger.error('Request id: %s, Status code: %s, error code: %s, error message: %s' % (
+                    response.request_id, response.status_code,
+                    response.code, response.message
+                ))
+                result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+                need_retry = retry_count < 2
+                result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+                if need_retry:
+                    return self.reply_text(session, retry_count + 1)
+                else:
+                    return result
+        except Exception as e:
+            logger.exception(e)
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if need_retry:
+                return self.reply_text(session, retry_count + 1)
+            else:
+                return result

models/dashscope/dashscope_session.py

@@ -0,0 +1,51 @@
+from models.session_manager import Session
+from common.log import logger
+
+
+class DashscopeSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="qwen-turbo"):
+        super().__init__(session_id)
+        self.reset()
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "user":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens,
+                                                                                       len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages)
+
+
+def num_tokens_from_messages(messages):
+    # 只是大概，具体计算规则：https://help.aliyun.com/zh/dashscope/developer-reference/token-api?spm=a2c4g.11186623.0.0.4d8b12b0BkP3K9
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["content"])
+    return tokens

models/gemini/google_gemini_bot.py

@@ -0,0 +1,704 @@
+"""
+Google gemini bot
+
+@author zhayujie
+@Date 2023/12/15
+"""
+# encoding:utf-8
+
+import json
+import time
+import requests
+from models.bot import Bot
+import google.generativeai as genai
+from models.session_manager import SessionManager
+from bridge.context import ContextType, Context
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf
+from models.chatgpt.chat_gpt_session import ChatGPTSession
+from models.baidu.baidu_wenxin_session import BaiduWenxinSession
+from google.generativeai.types import HarmCategory, HarmBlockThreshold
+
+
+# OpenAI对话模型API (可用)
+class GoogleGeminiBot(Bot):
+
+    def __init__(self):
+        super().__init__()
+        self.api_key = conf().get("gemini_api_key")
+        # 复用chatGPT的token计算方式
+        self.sessions = SessionManager(ChatGPTSession, model=conf().get("model") or "gpt-3.5-turbo")
+        self.model = conf().get("model") or "gemini-pro"
+        if self.model == "gemini":
+            self.model = "gemini-pro"
+        
+        # 支持自定义API base地址，复用open_ai_api_base配置
+        self.api_base = conf().get("open_ai_api_base", "").strip()
+        if self.api_base:
+            # 移除末尾的斜杠
+            self.api_base = self.api_base.rstrip('/')
+            # 如果配置的是OpenAI的地址，则使用默认的Gemini地址
+            if "api.openai.com" in self.api_base or not self.api_base:
+                self.api_base = "https://generativelanguage.googleapis.com"
+            logger.info(f"[Gemini] Using custom API base: {self.api_base}")
+        else:
+            self.api_base = "https://generativelanguage.googleapis.com"
+
+    def reply(self, query, context: Context = None) -> Reply:
+        try:
+            if context.type != ContextType.TEXT:
+                logger.warn(f"[Gemini] Unsupported message type, type={context.type}")
+                return Reply(ReplyType.TEXT, None)
+            logger.info(f"[Gemini] query={query}")
+            session_id = context["session_id"]
+            session = self.sessions.session_query(query, session_id)
+            gemini_messages = self._convert_to_gemini_messages(self.filter_messages(session.messages))
+            logger.debug(f"[Gemini] messages={gemini_messages}")
+            genai.configure(api_key=self.api_key)
+            model = genai.GenerativeModel(self.model)
+            
+            # 添加安全设置
+            safety_settings = {
+                HarmCategory.HARM_CATEGORY_HATE_SPEECH: HarmBlockThreshold.BLOCK_NONE,
+                HarmCategory.HARM_CATEGORY_HARASSMENT: HarmBlockThreshold.BLOCK_NONE,
+                HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: HarmBlockThreshold.BLOCK_NONE,
+                HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.BLOCK_NONE,
+            }
+            
+            # 生成回复，包含安全设置
+            response = model.generate_content(
+                gemini_messages,
+                safety_settings=safety_settings
+            )
+            if response.candidates and response.candidates[0].content:
+                reply_text = response.candidates[0].content.parts[0].text
+                logger.info(f"[Gemini] reply={reply_text}")
+                self.sessions.session_reply(reply_text, session_id)
+                return Reply(ReplyType.TEXT, reply_text)
+            else:
+                # 没有有效响应内容，可能内容被屏蔽，输出安全评分
+                logger.warning("[Gemini] No valid response generated. Checking safety ratings.")
+                if hasattr(response, 'candidates') and response.candidates:
+                    for rating in response.candidates[0].safety_ratings:
+                        logger.warning(f"Safety rating: {rating.category} - {rating.probability}")
+                error_message = "No valid response generated due to safety constraints."
+                self.sessions.session_reply(error_message, session_id)
+                return Reply(ReplyType.ERROR, error_message)
+                    
+        except Exception as e:
+            logger.error(f"[Gemini] Error generating response: {str(e)}", exc_info=True)
+            error_message = "Failed to invoke [Gemini] api!"
+            self.sessions.session_reply(error_message, session_id)
+            return Reply(ReplyType.ERROR, error_message)
+            
+    def _convert_to_gemini_messages(self, messages: list):
+        res = []
+        for msg in messages:
+            if msg.get("role") == "user":
+                role = "user"
+            elif msg.get("role") == "assistant":
+                role = "model"
+            elif msg.get("role") == "system":
+                role = "user"
+            else:
+                continue
+            res.append({
+                "role": role,
+                "parts": [{"text": msg.get("content")}]
+            })
+        return res
+
+    @staticmethod
+    def filter_messages(messages: list):
+        res = []
+        turn = "user"
+        if not messages:
+            return res
+        for i in range(len(messages) - 1, -1, -1):
+            message = messages[i]
+            role = message.get("role")
+            if role == "system":
+                res.insert(0, message)
+                continue
+            if role != turn:
+                continue
+            res.insert(0, message)
+            if turn == "user":
+                turn = "assistant"
+            elif turn == "assistant":
+                turn = "user"
+        return res
+
+    def call_with_tools(self, messages, tools=None, stream=False, **kwargs):
+        """
+        Call Gemini API with tool support using REST API (following official docs)
+        
+        Args:
+            messages: List of messages (OpenAI format)
+            tools: List of tool definitions (OpenAI/Claude format)
+            stream: Whether to use streaming
+            **kwargs: Additional parameters (system, max_tokens, temperature, etc.)
+            
+        Returns:
+            Formatted response compatible with OpenAI format or generator for streaming
+        """
+        try:
+            model_name = kwargs.get("model", self.model or "gemini-1.5-flash")
+            
+            # Build REST API payload
+            payload = {"contents": []}
+            
+            # Extract and set system instruction
+            system_prompt = kwargs.get("system", "")
+            if not system_prompt:
+                for msg in messages:
+                    if msg.get("role") == "system":
+                        system_prompt = msg["content"]
+                        break
+            
+            if system_prompt:
+                payload["system_instruction"] = {
+                    "parts": [{"text": system_prompt}]
+                }
+            
+            # Convert messages to Gemini format
+            for msg in messages:
+                role = msg.get("role")
+                content = msg.get("content", "")
+                
+                if role == "system":
+                    continue
+                
+                # Convert role
+                gemini_role = "user" if role in ["user", "tool"] else "model"
+                
+                # Handle different content formats
+                parts = []
+                
+                if isinstance(content, str):
+                    # Simple text content
+                    parts.append({"text": content})
+                    
+                elif isinstance(content, list):
+                    # List of content blocks (Claude format)
+                    for block in content:
+                        if not isinstance(block, dict):
+                            if isinstance(block, str):
+                                parts.append({"text": block})
+                            continue
+                        
+                        block_type = block.get("type")
+                        
+                        if block_type == "text":
+                            # Text block
+                            parts.append({"text": block.get("text", "")})
+                            
+                        elif block_type == "tool_result":
+                            # Convert Claude tool_result to Gemini functionResponse
+                            tool_use_id = block.get("tool_use_id")
+                            tool_content = block.get("content", "")
+                            
+                            # Try to parse tool content as JSON
+                            try:
+                                if isinstance(tool_content, str):
+                                    tool_result_data = json.loads(tool_content)
+                                else:
+                                    tool_result_data = tool_content
+                            except:
+                                tool_result_data = {"result": tool_content}
+                            
+                            # Find the tool name from previous messages
+                            # Look for the corresponding tool_call in model's message
+                            tool_name = None
+                            for prev_msg in reversed(messages):
+                                if prev_msg.get("role") == "assistant":
+                                    prev_content = prev_msg.get("content", [])
+                                    if isinstance(prev_content, list):
+                                        for prev_block in prev_content:
+                                            if isinstance(prev_block, dict) and prev_block.get("type") == "tool_use":
+                                                if prev_block.get("id") == tool_use_id:
+                                                    tool_name = prev_block.get("name")
+                                                    break
+                                    if tool_name:
+                                        break
+                            
+                            # Gemini functionResponse format
+                            parts.append({
+                                "functionResponse": {
+                                    "name": tool_name or "unknown",
+                                    "response": tool_result_data
+                                }
+                            })
+                            
+                        elif "text" in block:
+                            # Generic text field
+                            parts.append({"text": block["text"]})
+                
+                if parts:
+                    payload["contents"].append({
+                        "role": gemini_role,
+                        "parts": parts
+                    })
+            
+            # Generation config
+            gen_config = {}
+            if kwargs.get("temperature") is not None:
+                gen_config["temperature"] = kwargs["temperature"]
+
+            if gen_config:
+                payload["generationConfig"] = gen_config
+            
+            # Convert tools to Gemini format (REST API style)
+            if tools:
+                gemini_tools = self._convert_tools_to_gemini_rest_format(tools)
+                if gemini_tools:
+                    payload["tools"] = gemini_tools
+                    logger.debug(f"[Gemini] Added {len(tools)} tools to request")
+            
+            # Make REST API call
+            base_url = f"{self.api_base}/v1beta"
+            endpoint = f"{base_url}/models/{model_name}:generateContent"
+            if stream:
+                endpoint = f"{base_url}/models/{model_name}:streamGenerateContent?alt=sse"
+            
+            headers = {
+                "x-goog-api-key": self.api_key,
+                "Content-Type": "application/json"
+            }
+            
+            logger.debug(f"[Gemini] REST API call: {endpoint}")
+            
+            response = requests.post(
+                endpoint,
+                headers=headers,
+                json=payload,
+                stream=stream,
+                timeout=60
+            )
+            
+            # Check HTTP status for stream mode (for non-stream, it's checked in handler)
+            if stream and response.status_code != 200:
+                error_text = response.text
+                logger.error(f"[Gemini] API error ({response.status_code}): {error_text}")
+                def error_generator():
+                    yield {
+                        "error": True,
+                        "message": f"Gemini API error: {error_text}",
+                        "status_code": response.status_code
+                    }
+                return error_generator()
+            
+            if stream:
+                return self._handle_gemini_rest_stream_response(response, model_name)
+            else:
+                return self._handle_gemini_rest_sync_response(response, model_name)
+                
+        except Exception as e:
+            logger.error(f"[Gemini] call_with_tools error: {e}", exc_info=True)
+            error_msg = str(e)  # Capture error message before creating generator
+            if stream:
+                def error_generator():
+                    yield {
+                        "error": True,
+                        "message": error_msg,
+                        "status_code": 500
+                    }
+                return error_generator()
+            else:
+                return {
+                    "error": True,
+                    "message": str(e),
+                    "status_code": 500
+                }
+    
+    def _convert_tools_to_gemini_rest_format(self, tools_list):
+        """
+        Convert tools to Gemini REST API format
+        
+        Handles both OpenAI and Claude/Agent formats.
+        Returns: [{"functionDeclarations": [...]}]
+        """
+        function_declarations = []
+        
+        for tool in tools_list:
+            # Extract name, description, and parameters based on format
+            if tool.get("type") == "function":
+                # OpenAI format: {"type": "function", "function": {...}}
+                func = tool.get("function", {})
+                name = func.get("name")
+                description = func.get("description", "")
+                parameters = func.get("parameters", {})
+            else:
+                # Claude/Agent format: {"name": "...", "description": "...", "input_schema": {...}}
+                name = tool.get("name")
+                description = tool.get("description", "")
+                parameters = tool.get("input_schema", {})
+            
+            if not name:
+                logger.warning(f"[Gemini] Skipping tool without name: {tool}")
+                continue
+            
+            logger.debug(f"[Gemini] Converting tool: {name}")
+            
+            function_declarations.append({
+                "name": name,
+                "description": description,
+                "parameters": parameters
+            })
+        
+        # All functionDeclarations must be in a single tools object (per Gemini REST API spec)
+        return [{
+            "functionDeclarations": function_declarations
+        }] if function_declarations else []
+    
+    def _handle_gemini_rest_sync_response(self, response, model_name):
+        """Handle Gemini REST API sync response and convert to OpenAI format"""
+        try:
+            if response.status_code != 200:
+                error_text = response.text
+                logger.error(f"[Gemini] API error ({response.status_code}): {error_text}")
+                return {
+                    "error": True,
+                    "message": f"Gemini API error: {error_text}",
+                    "status_code": response.status_code
+                }
+            
+            data = response.json()
+            logger.debug(f"[Gemini] Response data: {json.dumps(data, ensure_ascii=False)[:500]}")
+            
+            # Extract from Gemini response format
+            candidates = data.get("candidates", [])
+            if not candidates:
+                logger.warning("[Gemini] No candidates in response")
+                return {
+                    "error": True,
+                    "message": "No candidates in response",
+                    "status_code": 500
+                }
+            
+            candidate = candidates[0]
+            content = candidate.get("content", {})
+            parts = content.get("parts", [])
+            
+            logger.debug(f"[Gemini] Candidate parts count: {len(parts)}")
+            
+            # Extract text and function calls
+            text_content = ""
+            tool_calls = []
+            
+            for part in parts:
+                # Check for text
+                if "text" in part:
+                    text_content += part["text"]
+                    logger.debug(f"[Gemini] Text part: {part['text'][:100]}...")
+                
+                # Check for functionCall (per REST API docs)
+                if "functionCall" in part:
+                    fc = part["functionCall"]
+                    logger.info(f"[Gemini] Function call detected: {fc.get('name')}")
+                    
+                    tool_calls.append({
+                        "id": f"call_{int(time.time() * 1000000)}",
+                        "type": "function",
+                        "function": {
+                            "name": fc.get("name"),
+                            "arguments": json.dumps(fc.get("args", {}))
+                        }
+                    })
+            
+            logger.info(f"[Gemini] Response: text={len(text_content)} chars, tool_calls={len(tool_calls)}")
+            
+            # Build OpenAI format response
+            message_dict = {
+                "role": "assistant",
+                "content": text_content or None
+            }
+            if tool_calls:
+                message_dict["tool_calls"] = tool_calls
+            
+            return {
+                "id": f"chatcmpl-{time.time()}",
+                "object": "chat.completion",
+                "created": int(time.time()),
+                "model": model_name,
+                "choices": [{
+                    "index": 0,
+                    "message": message_dict,
+                    "finish_reason": "tool_calls" if tool_calls else "stop"
+                }],
+                "usage": data.get("usageMetadata", {})
+            }
+            
+        except Exception as e:
+            logger.error(f"[Gemini] sync response error: {e}", exc_info=True)
+            return {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }
+    
+    def _handle_gemini_rest_stream_response(self, response, model_name):
+        """Handle Gemini REST API stream response"""
+        try:
+            all_tool_calls = []
+            has_sent_tool_calls = False
+            has_content = False  # Track if any content was sent
+            chunk_count = 0
+            last_finish_reason = None
+            last_safety_ratings = None
+            
+            for line in response.iter_lines():
+                if not line:
+                    continue
+                
+                line = line.decode('utf-8')
+                
+                # Skip SSE prefixes
+                if line.startswith('data: '):
+                    line = line[6:]
+                
+                if not line or line == '[DONE]':
+                    continue
+                
+                try:
+                    chunk_data = json.loads(line)
+                    chunk_count += 1
+                    logger.debug(f"[Gemini] Stream chunk: {json.dumps(chunk_data, ensure_ascii=False)[:200]}")
+                    
+                    candidates = chunk_data.get("candidates", [])
+                    if not candidates:
+                        logger.debug("[Gemini] No candidates in chunk")
+                        continue
+                    
+                    candidate = candidates[0]
+                    
+                    # 记录 finish_reason 和 safety_ratings
+                    if "finishReason" in candidate:
+                        last_finish_reason = candidate["finishReason"]
+                    if "safetyRatings" in candidate:
+                        last_safety_ratings = candidate["safetyRatings"]
+                    
+                    content = candidate.get("content", {})
+                    parts = content.get("parts", [])
+                    
+                    if not parts:
+                        logger.debug("[Gemini] No parts in candidate content")
+                    
+                    # Stream text content
+                    for part in parts:
+                        if "text" in part and part["text"]:
+                            has_content = True
+                            logger.debug(f"[Gemini] Streaming text: {part['text'][:50]}...")
+                            yield {
+                                "id": f"chatcmpl-{time.time()}",
+                                "object": "chat.completion.chunk",
+                                "created": int(time.time()),
+                                "model": model_name,
+                                "choices": [{
+                                    "index": 0,
+                                    "delta": {"content": part["text"]},
+                                    "finish_reason": None
+                                }]
+                            }
+                        
+                        # Collect function calls
+                        if "functionCall" in part:
+                            fc = part["functionCall"]
+                            logger.debug(f"[Gemini] Function call detected: {fc.get('name')}")
+                            all_tool_calls.append({
+                                "index": len(all_tool_calls),  # Add index to differentiate multiple tool calls
+                                "id": f"call_{int(time.time() * 1000000)}_{len(all_tool_calls)}",
+                                "type": "function",
+                                "function": {
+                                    "name": fc.get("name"),
+                                    "arguments": json.dumps(fc.get("args", {}))
+                                }
+                            })
+                    
+                except json.JSONDecodeError as je:
+                    logger.debug(f"[Gemini] JSON decode error: {je}")
+                    continue
+            
+            # Send tool calls if any were collected
+            if all_tool_calls and not has_sent_tool_calls:
+                logger.debug(f"[Gemini] Stream detected {len(all_tool_calls)} tool calls")
+                yield {
+                    "id": f"chatcmpl-{time.time()}",
+                    "object": "chat.completion.chunk",
+                    "created": int(time.time()),
+                    "model": model_name,
+                    "choices": [{
+                        "index": 0,
+                        "delta": {"tool_calls": all_tool_calls},
+                        "finish_reason": None
+                    }]
+                }
+                has_sent_tool_calls = True
+            
+            # Log summary (only if there's something interesting)
+            if not has_content and not all_tool_calls:
+                logger.debug(f"[Gemini] Stream complete: has_content={has_content}, tool_calls={len(all_tool_calls)}")
+            elif all_tool_calls:
+                logger.debug(f"[Gemini] Stream complete: {len(all_tool_calls)} tool calls")
+            else:
+                logger.debug(f"[Gemini] Stream complete: text response")
+            
+            # 如果返回空响应，记录详细警告
+            if not has_content and not all_tool_calls:
+                logger.warning(f"[Gemini] ⚠️  Empty response detected!")
+            
+            # Final chunk
+            yield {
+                "id": f"chatcmpl-{time.time()}",
+                "object": "chat.completion.chunk",
+                "created": int(time.time()),
+                "model": model_name,
+                "choices": [{
+                    "index": 0,
+                    "delta": {},
+                    "finish_reason": "tool_calls" if all_tool_calls else "stop"
+                }]
+            }
+                    
+        except Exception as e:
+            logger.error(f"[Gemini] stream response error: {e}", exc_info=True)
+            error_msg = str(e)
+            yield {
+                "error": True,
+                "message": error_msg,
+                "status_code": 500
+            }
+    
+    def _convert_tools_to_gemini_format(self, openai_tools):
+        """Convert OpenAI tool format to Gemini function declarations"""
+        import google.generativeai as genai
+        
+        gemini_functions = []
+        for tool in openai_tools:
+            if tool.get("type") == "function":
+                func = tool.get("function", {})
+                gemini_functions.append(
+                    genai.protos.FunctionDeclaration(
+                        name=func.get("name"),
+                        description=func.get("description", ""),
+                        parameters=func.get("parameters", {})
+                    )
+                )
+        
+        if gemini_functions:
+            return [genai.protos.Tool(function_declarations=gemini_functions)]
+        return None
+    
+    def _handle_gemini_sync_response(self, model, messages, request_params, model_name):
+        """Handle synchronous Gemini API response"""
+        import json
+        
+        response = model.generate_content(messages, **request_params)
+        
+        # Extract text content and function calls
+        text_content = ""
+        tool_calls = []
+        
+        if response.candidates and response.candidates[0].content:
+            for part in response.candidates[0].content.parts:
+                if hasattr(part, 'text') and part.text:
+                    text_content += part.text
+                elif hasattr(part, 'function_call') and part.function_call:
+                    # Convert Gemini function call to OpenAI format
+                    func_call = part.function_call
+                    tool_calls.append({
+                        "id": f"call_{hash(func_call.name)}",
+                        "type": "function",
+                        "function": {
+                            "name": func_call.name,
+                            "arguments": json.dumps(dict(func_call.args))
+                        }
+                    })
+        
+        # Build message in OpenAI format
+        message = {
+            "role": "assistant",
+            "content": text_content
+        }
+        if tool_calls:
+            message["tool_calls"] = tool_calls
+        
+        # Format response to match OpenAI structure
+        formatted_response = {
+            "id": f"gemini_{int(time.time())}",
+            "object": "chat.completion",
+            "created": int(time.time()),
+            "model": model_name,
+            "choices": [
+                {
+                    "index": 0,
+                    "message": message,
+                    "finish_reason": "stop" if not tool_calls else "tool_calls"
+                }
+            ],
+            "usage": {
+                "prompt_tokens": 0,  # Gemini doesn't provide token counts in the same way
+                "completion_tokens": 0,
+                "total_tokens": 0
+            }
+        }
+        
+        logger.info(f"[Gemini] call_with_tools reply, model={model_name}")
+        return formatted_response
+    
+    def _handle_gemini_stream_response(self, model, messages, request_params, model_name):
+        """Handle streaming Gemini API response"""
+        import json
+        
+        try:
+            response_stream = model.generate_content(messages, stream=True, **request_params)
+            
+            for chunk in response_stream:
+                if chunk.candidates and chunk.candidates[0].content:
+                    for part in chunk.candidates[0].content.parts:
+                        if hasattr(part, 'text') and part.text:
+                            # Text content
+                            yield {
+                                "id": f"gemini_{int(time.time())}",
+                                "object": "chat.completion.chunk",
+                                "created": int(time.time()),
+                                "model": model_name,
+                                "choices": [{
+                                    "index": 0,
+                                    "delta": {"content": part.text},
+                                    "finish_reason": None
+                                }]
+                            }
+                        elif hasattr(part, 'function_call') and part.function_call:
+                            # Function call
+                            func_call = part.function_call
+                            yield {
+                                "id": f"gemini_{int(time.time())}",
+                                "object": "chat.completion.chunk",
+                                "created": int(time.time()),
+                                "model": model_name,
+                                "choices": [{
+                                    "index": 0,
+                                    "delta": {
+                                        "tool_calls": [{
+                                            "index": 0,
+                                            "id": f"call_{hash(func_call.name)}",
+                                            "type": "function",
+                                            "function": {
+                                                "name": func_call.name,
+                                                "arguments": json.dumps(dict(func_call.args))
+                                            }
+                                        }]
+                                    },
+                                    "finish_reason": None
+                                }]
+                            }
+                            
+        except Exception as e:
+            logger.error(f"[Gemini] stream response error: {e}")
+            yield {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }

models/linkai/link_ai_bot.py

@@ -0,0 +1,646 @@
+# access LinkAI knowledge base platform
+# docs: https://link-ai.tech/platform/link-app/wechat
+
+import re
+import time
+import requests
+import json
+import config
+from models.bot import Bot
+from models.openai_compatible_bot import OpenAICompatibleBot
+from models.chatgpt.chat_gpt_session import ChatGPTSession
+from models.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
+from common import memory, utils
+import base64
+import os
+
+class LinkAIBot(Bot, OpenAICompatibleBot):
+    # authentication failed
+    AUTH_FAILED_CODE = 401
+    NO_QUOTA_CODE = 406
+
+    def __init__(self):
+        super().__init__()
+        self.sessions = LinkAISessionManager(LinkAISession, model=conf().get("model") or "gpt-3.5-turbo")
+        self.args = {}
+    
+    def get_api_config(self):
+        """Get API configuration for OpenAI-compatible base class"""
+        return {
+            'api_key': conf().get("open_ai_api_key"),  # LinkAI uses OpenAI-compatible key
+            'api_base': conf().get("open_ai_api_base", "https://api.link-ai.tech/v1"),
+            'model': conf().get("model", "gpt-3.5-turbo"),
+            'default_temperature': conf().get("temperature", 0.9),
+            'default_top_p': conf().get("top_p", 1.0),
+            'default_frequency_penalty': conf().get("frequency_penalty", 0.0),
+            'default_presence_penalty': conf().get("presence_penalty", 0.0),
+        }
+
+    def reply(self, query, context: Context = None) -> Reply:
+        if context.type == ContextType.TEXT:
+            return self._chat(query, context)
+        elif context.type == ContextType.IMAGE_CREATE:
+            if not conf().get("text_to_image"):
+                logger.warn("[LinkAI] text_to_image is not enabled, ignore the IMAGE_CREATE request")
+                return Reply(ReplyType.TEXT, "")
+            ok, res = self.create_img(query, 0)
+            if ok:
+                reply = Reply(ReplyType.IMAGE_URL, res)
+            else:
+                reply = Reply(ReplyType.ERROR, res)
+            return reply
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def _chat(self, query, context, retry_count=0) -> Reply:
+        """
+        发起对话请求
+        :param query: 请求提示词
+        :param context: 对话上下文
+        :param retry_count: 当前递归重试次数
+        :return: 回复
+        """
+        if retry_count > 2:
+            # exit from retry 2 times
+            logger.warn("[LINKAI] failed after maximum number of retry times")
+            return Reply(ReplyType.TEXT, "请再问我一次吧")
+
+        try:
+            # load config
+            if context.get("generate_breaked_by"):
+                logger.info(f"[LINKAI] won't set appcode because a plugin ({context['generate_breaked_by']}) affected the context")
+                app_code = None
+            else:
+                plugin_app_code = self._find_group_mapping_code(context)
+                app_code = context.kwargs.get("app_code") or plugin_app_code or conf().get("linkai_app_code")
+            linkai_api_key = conf().get("linkai_api_key")
+
+            session_id = context["session_id"]
+            session_message = self.sessions.session_msg_query(query, session_id)
+            logger.debug(f"[LinkAI] session={session_message}, session_id={session_id}")
+
+            # image process
+            img_cache = memory.USER_IMAGE_CACHE.get(session_id)
+            if img_cache:
+                messages = self._process_image_msg(app_code=app_code, session_id=session_id, query=query, img_cache=img_cache)
+                if messages:
+                    session_message = messages
+
+            model = conf().get("model")
+            # remove system message
+            if session_message[0].get("role") == "system":
+                if app_code or model == "wenxin":
+                    session_message.pop(0)
+            body = {
+                "app_code": app_code,
+                "messages": session_message,
+                "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
+                "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
+                "session_id": session_id,
+                "sender_id": session_id,
+                "channel_type": conf().get("channel_type", "wx")
+            }
+            try:
+                from linkai import LinkAIClient
+                client_id = LinkAIClient.fetch_client_id()
+                if client_id:
+                    body["client_id"] = client_id
+                    # start: client info deliver
+                    if context.kwargs.get("msg"):
+                        body["session_id"] = context.kwargs.get("msg").from_user_id
+                        if context.kwargs.get("msg").is_group:
+                            body["is_group"] = True
+                            body["group_name"] = context.kwargs.get("msg").from_user_nickname
+                            body["sender_name"] = context.kwargs.get("msg").actual_user_nickname
+                        else:
+                            if body.get("channel_type") in ["wechatcom_app"]:
+                                body["sender_name"] = context.kwargs.get("msg").from_user_id
+                            else:
+                                body["sender_name"] = context.kwargs.get("msg").from_user_nickname
+
+            except Exception as e:
+                pass
+            file_id = context.kwargs.get("file_id")
+            if file_id:
+                body["file_id"] = file_id
+            logger.info(f"[LINKAI] query={query}, app_code={app_code}, model={body.get('model')}, file_id={file_id}")
+            headers = {"Authorization": "Bearer " + linkai_api_key}
+
+            # do http request
+            base_url = conf().get("linkai_api_base", "https://api.link-ai.tech")
+            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"]
+                res_code = response.get('code')
+                logger.info(f"[LINKAI] reply={reply_content}, total_tokens={total_tokens}, res_code={res_code}")
+                if res_code == 429:
+                    logger.warn(f"[LINKAI] 用户访问超出限流配置，sender_id={body.get('sender_id')}")
+                else:
+                    self.sessions.session_reply(reply_content, session_id, total_tokens, query=query)
+                agent_suffix = self._fetch_agent_suffix(response)
+                if agent_suffix:
+                    reply_content += agent_suffix
+                if not agent_suffix:
+                    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()
+                    reply_content = response["choices"][0].get("text_content")
+                if reply_content:
+                    reply_content = self._process_url(reply_content)
+                return Reply(ReplyType.TEXT, 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._chat(query, context, retry_count + 1)
+
+                error_reply = "提问太快啦，请休息一下再问我吧"
+                if res.status_code == 409:
+                    error_reply = "这个问题我还没有学会，请问我其它问题吧"
+                return Reply(ReplyType.TEXT, error_reply)
+
+        except Exception as e:
+            logger.exception(e)
+            # retry
+            time.sleep(2)
+            logger.warn(f"[LINKAI] do retry, times={retry_count}")
+            return self._chat(query, context, retry_count + 1)
+
+    def _process_image_msg(self, app_code: str, session_id: str, query:str, img_cache: dict):
+        try:
+            enable_image_input = False
+            app_info = self._fetch_app_info(app_code)
+            if not app_info:
+                logger.debug(f"[LinkAI] not found app, can't process images, app_code={app_code}")
+                return None
+            plugins = app_info.get("data").get("plugins")
+            for plugin in plugins:
+                if plugin.get("input_type") and "IMAGE" in plugin.get("input_type"):
+                    enable_image_input = True
+            if not enable_image_input:
+                return
+            msg = img_cache.get("msg")
+            path = img_cache.get("path")
+            msg.prepare()
+            logger.info(f"[LinkAI] query with images, path={path}")
+            messages = self._build_vision_msg(query, path)
+            memory.USER_IMAGE_CACHE[session_id] = None
+            return messages
+        except Exception as e:
+            logger.exception(e)
+
+    def _find_group_mapping_code(self, context):
+        try:
+            if context.kwargs.get("isgroup"):
+                group_name = context.kwargs.get("msg").from_user_nickname
+                if config.plugin_config and config.plugin_config.get("linkai"):
+                    linkai_config = config.plugin_config.get("linkai")
+                    group_mapping = linkai_config.get("group_app_map")
+                    if group_mapping and group_name:
+                        return group_mapping.get(group_name)
+        except Exception as e:
+            logger.exception(e)
+            return None
+
+    def _build_vision_msg(self, query: str, path: str):
+        try:
+            suffix = utils.get_path_suffix(path)
+            with open(path, "rb") as file:
+                base64_str = base64.b64encode(file.read()).decode('utf-8')
+                messages = [{
+                    "role": "user",
+                    "content": [
+                        {
+                            "type": "text",
+                            "text": query
+                        },
+                        {
+                            "type": "image_url",
+                            "image_url": {
+                                "url": f"data:image/{suffix};base64,{base64_str}"
+                            }
+                        }
+                    ]
+                }]
+                return messages
+        except Exception as e:
+            logger.exception(e)
+
+    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.tech")
+            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)
+
+    def _fetch_app_info(self, app_code: str):
+        headers = {"Authorization": "Bearer " + conf().get("linkai_api_key")}
+        # do http request
+        base_url = conf().get("linkai_api_base", "https://api.link-ai.tech")
+        params = {"app_code": app_code}
+        res = requests.get(url=base_url + "/v1/app/info", params=params, headers=headers, timeout=(5, 10))
+        if res.status_code == 200:
+            return res.json()
+        else:
+            logger.warning(f"[LinkAI] find app info exception, res={res}")
+
+    def create_img(self, query, retry_count=0, api_key=None):
+        try:
+            logger.info("[LinkImage] image_query={}".format(query))
+            headers = {
+                "Content-Type": "application/json",
+                "Authorization": f"Bearer {conf().get('linkai_api_key')}"
+            }
+            data = {
+                "prompt": query,
+                "n": 1,
+                "model": conf().get("text_to_image") or "dall-e-2",
+                "response_format": "url",
+                "img_proxy": conf().get("image_proxy")
+            }
+            url = conf().get("linkai_api_base", "https://api.link-ai.tech") + "/v1/images/generations"
+            res = requests.post(url, headers=headers, json=data, timeout=(5, 90))
+            t2 = time.time()
+            image_url = res.json()["data"][0]["url"]
+            logger.info("[OPEN_AI] image_url={}".format(image_url))
+            return True, image_url
+
+        except Exception as e:
+            logger.error(format(e))
+            return False, "画图出现问题，请休息一下再问我吧"
+
+
+    def _fetch_knowledge_search_suffix(self, response) -> str:
+        try:
+            if response.get("knowledge_base"):
+                search_hit = response.get("knowledge_base").get("search_hit")
+                first_similarity = response.get("knowledge_base").get("first_similarity")
+                logger.info(f"[LINKAI] knowledge base, search_hit={search_hit}, first_similarity={first_similarity}")
+                plugin_config = pconf("linkai")
+                if plugin_config and plugin_config.get("knowledge_base") and plugin_config.get("knowledge_base").get("search_miss_text_enabled"):
+                    search_miss_similarity = plugin_config.get("knowledge_base").get("search_miss_similarity")
+                    search_miss_text = plugin_config.get("knowledge_base").get("search_miss_suffix")
+                    if not search_hit:
+                        return search_miss_text
+                    if search_miss_similarity and float(search_miss_similarity) > first_similarity:
+                        return search_miss_text
+        except Exception as e:
+            logger.exception(e)
+
+
+    def _fetch_agent_suffix(self, response):
+        try:
+            plugin_list = []
+            logger.debug(f"[LinkAgent] res={response}")
+            if response.get("agent") and response.get("agent").get("chain") and response.get("agent").get("need_show_plugin"):
+                chain = response.get("agent").get("chain")
+                suffix = "\n\n- - - - - - - - - - - -"
+                i = 0
+                for turn in chain:
+                    plugin_name = turn.get('plugin_name')
+                    suffix += "\n"
+                    need_show_thought = response.get("agent").get("need_show_thought")
+                    if turn.get("thought") and plugin_name and need_show_thought:
+                        suffix += f"{turn.get('thought')}\n"
+                    if plugin_name:
+                        plugin_list.append(turn.get('plugin_name'))
+                        if turn.get('plugin_icon'):
+                            suffix += f"{turn.get('plugin_icon')} "
+                        suffix += f"{turn.get('plugin_name')}"
+                        if turn.get('plugin_input'):
+                            suffix += f"：{turn.get('plugin_input')}"
+                    if i < len(chain) - 1:
+                        suffix += "\n"
+                    i += 1
+                logger.info(f"[LinkAgent] use plugins: {plugin_list}")
+                return suffix
+        except Exception as e:
+            logger.exception(e)
+
+    def _process_url(self, text):
+        try:
+            url_pattern = re.compile(r'\[(.*?)\]\((http[s]?://.*?)\)')
+            def replace_markdown_url(match):
+                return f"{match.group(2)}"
+            return url_pattern.sub(replace_markdown_url, text)
+        except Exception as e:
+            logger.error(e)
+
+    def _send_image(self, channel, context, image_urls):
+        if not image_urls:
+            return
+        max_send_num = conf().get("max_media_send_count")
+        send_interval = conf().get("media_send_interval")
+        file_type = (".pdf", ".doc", ".docx", ".csv", ".xls", ".xlsx", ".txt", ".rtf", ".ppt", ".pptx")
+        try:
+            i = 0
+            for url in image_urls:
+                if max_send_num and i >= max_send_num:
+                    continue
+                i += 1
+                if url.endswith(".mp4"):
+                    reply_type = ReplyType.VIDEO_URL
+                elif url.endswith(file_type):
+                    reply_type = ReplyType.FILE
+                    url = _download_file(url)
+                    if not url:
+                        continue
+                else:
+                    reply_type = ReplyType.IMAGE_URL
+                reply = Reply(reply_type, url)
+                channel.send(reply, context)
+                if send_interval:
+                    time.sleep(send_interval)
+        except Exception as e:
+            logger.error(e)
+
+
+def _download_file(url: str):
+    try:
+        file_path = "tmp"
+        if not os.path.exists(file_path):
+            os.makedirs(file_path)
+        file_name = url.split("/")[-1]  # 获取文件名
+        file_path = os.path.join(file_path, file_name)
+        response = requests.get(url)
+        with open(file_path, "wb") as f:
+            f.write(response.content)
+        return file_path
+    except Exception as e:
+        logger.warn(e)
+
+
+class LinkAISessionManager(SessionManager):
+    def session_msg_query(self, query, session_id):
+        session = self.build_session(session_id)
+        messages = session.messages + [{"role": "user", "content": query}]
+        return messages
+
+    def session_reply(self, reply, session_id, total_tokens=None, query=None):
+        session = self.build_session(session_id)
+        if query:
+            session.add_query(query)
+        session.add_reply(reply)
+        try:
+            max_tokens = conf().get("conversation_max_tokens", 8000)
+            tokens_cnt = session.discard_exceeding(max_tokens, total_tokens)
+            logger.debug(f"[LinkAI] chat history, before tokens={total_tokens}, now tokens={tokens_cnt}")
+        except Exception as e:
+            logger.warning("Exception when counting tokens precisely for session: {}".format(str(e)))
+        return session
+
+
+class LinkAISession(ChatGPTSession):
+    def calc_tokens(self):
+        if not self.messages:
+            return 0
+        return len(str(self.messages))
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        cur_tokens = self.calc_tokens()
+        if cur_tokens > max_tokens:
+            for i in range(0, len(self.messages)):
+                if i > 0 and self.messages[i].get("role") == "assistant" and self.messages[i - 1].get("role") == "user":
+                    self.messages.pop(i)
+                    self.messages.pop(i - 1)
+                    return self.calc_tokens()
+        return cur_tokens
+
+
+# Add call_with_tools method to LinkAIBot class
+def _linkai_call_with_tools(self, messages, tools=None, stream=False, **kwargs):
+    """
+    Call LinkAI API with tool support for agent integration
+    LinkAI is fully compatible with OpenAI's tool calling format
+    
+    Args:
+        messages: List of messages
+        tools: List of tool definitions (OpenAI format)
+        stream: Whether to use streaming
+        **kwargs: Additional parameters (max_tokens, temperature, etc.)
+        
+    Returns:
+        Formatted response in OpenAI format or generator for streaming
+    """
+    try:
+        # Convert messages from Claude format to OpenAI format
+        # This is important because Agent uses Claude format internally
+        messages = self._convert_messages_to_openai_format(messages)
+        
+        # Convert tools from Claude format to OpenAI format
+        if tools:
+            tools = self._convert_tools_to_openai_format(tools)
+        
+        # Handle system prompt (OpenAI uses system message, Claude uses separate parameter)
+        system_prompt = kwargs.get('system')
+        if system_prompt:
+            # Add system message at the beginning if not already present
+            if not messages or messages[0].get('role') != 'system':
+                messages = [{"role": "system", "content": system_prompt}] + messages
+            else:
+                # Replace existing system message
+                messages[0] = {"role": "system", "content": system_prompt}
+        
+        logger.debug(f"[LinkAI] messages: {len(messages)}, tools: {len(tools) if tools else 0}, stream: {stream}")
+        
+        # Build request parameters (LinkAI uses OpenAI-compatible format)
+        body = {
+            "messages": messages,
+            "model": kwargs.get("model", conf().get("model") or "gpt-3.5-turbo"),
+            "temperature": kwargs.get("temperature", conf().get("temperature", 0.9)),
+            "top_p": kwargs.get("top_p", conf().get("top_p", 1)),
+            "frequency_penalty": kwargs.get("frequency_penalty", conf().get("frequency_penalty", 0.0)),
+            "presence_penalty": kwargs.get("presence_penalty", conf().get("presence_penalty", 0.0)),
+            "stream": stream
+        }
+
+        if tools:
+            body["tools"] = tools
+            body["tool_choice"] = kwargs.get("tool_choice", "auto")
+        
+        # Prepare headers
+        headers = {"Authorization": "Bearer " + conf().get("linkai_api_key")}
+        base_url = conf().get("linkai_api_base", "https://api.link-ai.tech")
+        
+        if stream:
+            return self._handle_linkai_stream_response(base_url, headers, body)
+        else:
+            return self._handle_linkai_sync_response(base_url, headers, body)
+            
+    except Exception as e:
+        logger.error(f"[LinkAI] call_with_tools error: {e}")
+        if stream:
+            def error_generator():
+                yield {
+                    "error": True,
+                    "message": str(e),
+                    "status_code": 500
+                }
+            return error_generator()
+        else:
+            return {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }
+
+def _handle_linkai_sync_response(self, base_url, headers, body):
+    """Handle synchronous LinkAI API response"""
+    try:
+        res = requests.post(
+            url=base_url + "/v1/chat/completions",
+            json=body,
+            headers=headers,
+            timeout=conf().get("request_timeout", 180)
+        )
+        
+        if res.status_code == 200:
+            response = res.json()
+            logger.debug(f"[LinkAI] reply: model={response.get('model')}, "
+                        f"tokens={response.get('usage', {}).get('total_tokens', 0)}")
+            
+            # LinkAI response is already in OpenAI-compatible format
+            return response
+        else:
+            error_data = res.json()
+            error_msg = error_data.get("error", {}).get("message", "Unknown error")
+            raise Exception(f"LinkAI API error: {res.status_code} - {error_msg}")
+            
+    except Exception as e:
+        logger.error(f"[LinkAI] sync response error: {e}")
+        raise
+
+def _handle_linkai_stream_response(self, base_url, headers, body):
+    """Handle streaming LinkAI API response"""
+    try:
+        res = requests.post(
+            url=base_url + "/v1/chat/completions",
+            json=body,
+            headers=headers,
+            timeout=conf().get("request_timeout", 180),
+            stream=True
+        )
+        
+        if res.status_code != 200:
+            error_text = res.text
+            try:
+                error_data = json.loads(error_text)
+                error_msg = error_data.get("error", {}).get("message", error_text)
+            except:
+                error_msg = error_text or "Unknown error"
+            
+            yield {
+                "error": True,
+                "status_code": res.status_code,
+                "message": error_msg
+            }
+            return
+        
+        # Process streaming response (OpenAI-compatible SSE format)
+        for line in res.iter_lines():
+            if line:
+                line = line.decode('utf-8')
+                if line.startswith('data: '):
+                    line = line[6:]  # Remove 'data: ' prefix
+                    if line == '[DONE]':
+                        break
+                    try:
+                        chunk = json.loads(line)
+                        yield chunk
+                    except json.JSONDecodeError:
+                        continue
+                        
+    except Exception as e:
+        logger.error(f"[LinkAI] stream response error: {e}")
+        yield {
+            "error": True,
+            "message": str(e),
+            "status_code": 500
+        }
+
+# Attach methods to LinkAIBot class
+LinkAIBot.call_with_tools = _linkai_call_with_tools
+LinkAIBot._handle_linkai_sync_response = _handle_linkai_sync_response
+LinkAIBot._handle_linkai_stream_response = _handle_linkai_stream_response

models/minimax/minimax_bot.py

@@ -0,0 +1,151 @@
+# encoding:utf-8
+
+import time
+
+import openai
+import openai.error
+from models.bot import Bot
+from models.minimax.minimax_session import MinimaxSession
+from models.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, load_config
+from models.chatgpt.chat_gpt_session import ChatGPTSession
+import requests
+from common import const
+
+
+# ZhipuAI对话模型API
+class MinimaxBot(Bot):
+    def __init__(self):
+        super().__init__()
+        self.args = {
+            "model": conf().get("model") or "abab6.5",  # 对话模型的名称
+            "temperature": conf().get("temperature", 0.3),  # 如果设置，值域须为 [0, 1] 我们推荐 0.3，以达到较合适的效果。
+            "top_p": conf().get("top_p", 0.95),  # 使用默认值
+        }
+        self.api_key = conf().get("Minimax_api_key")
+        self.group_id = conf().get("Minimax_group_id")
+        self.base_url = conf().get("Minimax_base_url", f"https://api.minimax.chat/v1/text/chatcompletion_pro?GroupId={self.group_id}")
+        # tokens_to_generate/bot_setting/reply_constraints可自行修改
+        self.request_body = {
+            "model": self.args["model"],
+            "tokens_to_generate": 2048,
+            "reply_constraints": {"sender_type": "BOT", "sender_name": "MM智能助理"},
+            "messages": [],
+            "bot_setting": [
+                {
+                    "bot_name": "MM智能助理",
+                    "content": "MM智能助理是一款由MiniMax自研的，没有调用其他产品的接口的大型语言模型。MiniMax是一家中国科技公司，一直致力于进行大模型相关的研究。",
+                }
+            ],
+        }
+        self.sessions = SessionManager(MinimaxSession, model=const.MiniMax)
+
+    def reply(self, query, context: Context = None) -> Reply:
+        # acquire reply content
+        logger.info("[Minimax_AI] query={}".format(query))
+        if context.type == ContextType.TEXT:
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[Minimax_AI] session query={}".format(session))
+
+            model = context.get("Minimax_model")
+            new_args = self.args.copy()
+            if model:
+                new_args["model"] = model
+            # if context.get('stream'):
+            #     # reply in stream
+            #     return self.reply_text_stream(query, new_query, session_id)
+
+            reply_content = self.reply_text(session, args=new_args)
+            logger.debug(
+                "[Minimax_AI] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[Minimax_AI] reply {} used 0 tokens.".format(reply_content))
+            return reply
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: MinimaxSession, args=None, retry_count=0) -> dict:
+        """
+        call openai's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            headers = {"Content-Type": "application/json", "Authorization": "Bearer " + self.api_key}
+            self.request_body["messages"].extend(session.messages)
+            logger.info("[Minimax_AI] request_body={}".format(self.request_body))
+            # logger.info("[Minimax_AI] reply={}, total_tokens={}".format(response.choices[0]['message']['content'], response["usage"]["total_tokens"]))
+            res = requests.post(self.base_url, headers=headers, json=self.request_body)
+
+            # self.request_body["messages"].extend(response.json()["choices"][0]["messages"])
+            if res.status_code == 200:
+                response = res.json()
+                return {
+                    "total_tokens": response["usage"]["total_tokens"],
+                    "completion_tokens": response["usage"]["total_tokens"],
+                    "content": response["reply"],
+                }
+            else:
+                response = res.json()
+                error = response.get("error")
+                logger.error(f"[Minimax_AI] chat failed, status_code={res.status_code}, " f"msg={error.get('message')}, type={error.get('type')}")
+
+                result = {"completion_tokens": 0, "content": "提问太快啦，请休息一下再问我吧"}
+                need_retry = False
+                if res.status_code >= 500:
+                    # server error, need retry
+                    logger.warn(f"[Minimax_AI] do retry, times={retry_count}")
+                    need_retry = retry_count < 2
+                elif res.status_code == 401:
+                    result["content"] = "授权失败，请检查API Key是否正确"
+                elif res.status_code == 429:
+                    result["content"] = "请求过于频繁，请稍后再试"
+                    need_retry = retry_count < 2
+                else:
+                    need_retry = False
+
+                if need_retry:
+                    time.sleep(3)
+                    return self.reply_text(session, args, retry_count + 1)
+                else:
+                    return result
+        except Exception as e:
+            logger.exception(e)
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if need_retry:
+                return self.reply_text(session, args, retry_count + 1)
+            else:
+                return result

models/minimax/minimax_session.py

@@ -0,0 +1,72 @@
+from models.session_manager import Session
+from common.log import logger
+
+"""
+    e.g.
+    [
+        {"role": "system", "content": "You are a helpful assistant."},
+        {"role": "user", "content": "Who won the world series in 2020?"},
+        {"role": "assistant", "content": "The Los Angeles Dodgers won the World Series in 2020."},
+        {"role": "user", "content": "Where was it played?"}
+    ]
+"""
+
+
+class MinimaxSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="minimax"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        # self.reset()
+
+    def add_query(self, query):
+        user_item = {"sender_type": "USER", "sender_name": self.session_id, "text": query}
+        self.messages.append(user_item)
+
+    def add_reply(self, reply):
+        assistant_item = {"sender_type": "BOT", "sender_name": "MM智能助理", "text": reply}
+        self.messages.append(assistant_item)
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["sender_type"] == "BOT":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["sender_type"] == "USER":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens, len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+
+def num_tokens_from_messages(messages, model):
+    """Returns the number of tokens used by a list of messages."""
+    # 官方token计算规则："对于中文文本来说，1个token通常对应一个汉字；对于英文文本来说，1个token通常对应3至4个字母或1个单词"
+    # 详情请产看文档：https://help.aliyun.com/document_detail/2586397.html
+    # 目前根据字符串长度粗略估计token数，不影响正常使用
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["text"])
+    return tokens

models/modelscope/modelscope_bot.py

@@ -0,0 +1,277 @@
+# encoding:utf-8
+
+import time
+import json
+import openai
+import openai.error
+from models.bot import Bot
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf, load_config
+from .modelscope_session import ModelScopeSession
+import requests
+
+
+# ModelScope对话模型API
+class ModelScopeBot(Bot):
+    def __init__(self):
+        super().__init__()
+        self.sessions = SessionManager(ModelScopeSession, model=conf().get("model") or "Qwen/Qwen2.5-7B-Instruct")
+        model = conf().get("model") or "Qwen/Qwen2.5-7B-Instruct"
+        if model == "modelscope":
+            model = "Qwen/Qwen2.5-7B-Instruct"
+        self.args = {
+            "model": model,  # 对话模型的名称
+            "temperature": conf().get("temperature", 0.3),  # 如果设置，值域须为 [0, 1] 我们推荐 0.3，以达到较合适的效果。
+            "top_p": conf().get("top_p", 1.0),  # 使用默认值
+        }
+        self.api_key = conf().get("modelscope_api_key")
+        self.base_url = conf().get("modelscope_base_url", "https://api-inference.modelscope.cn/v1/chat/completions")
+        """
+        需要获取ModelScope支持API-inference的模型名称列表，请到魔搭社区官网模型中心查看 https://modelscope.cn/models?filter=inference_type&page=1。
+        或者使用命令 curl https://api-inference.modelscope.cn/v1/models 对模型列表和ID进行获取。查看commend/const.py文件也可以获取模型列表。
+        获取ModelScope的免费API Key，请到魔搭社区官网用户中心查看获取方式 https://modelscope.cn/docs/model-service/API-Inference/intro。
+        """
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context.type == ContextType.TEXT:
+            logger.info("[MODELSCOPE_AI] query={}".format(query))
+
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[MODELSCOPE_AI] session query={}".format(session.messages))
+
+            model = context.get("modelscope_model")
+            new_args = self.args.copy()
+            if model:
+                new_args["model"] = model
+
+            if new_args["model"] == "Qwen/QwQ-32B":
+                reply_content = self.reply_text_stream(session, args=new_args)
+            else:
+                reply_content = self.reply_text(session, args=new_args)
+
+            logger.debug(
+                "[MODELSCOPE_AI] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                # 只有当 content 为空且 completion_tokens 为 0 时才标记为错误
+                if len(reply_content["content"]) == 0:
+                    reply = Reply(ReplyType.ERROR, reply_content["content"])
+                else:
+                    reply = Reply(ReplyType.TEXT, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[MODELSCOPE_AI] reply {} used 0 tokens.".format(reply_content))
+            return reply
+        elif context.type == ContextType.IMAGE_CREATE:
+            ok, retstring = self.create_img(query, 0)
+            reply = None
+            if ok:
+                reply = Reply(ReplyType.IMAGE_URL, retstring)
+            else:
+                reply = Reply(ReplyType.ERROR, retstring)
+            return reply
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: ModelScopeSession, args=None, retry_count=0) -> dict:
+        """
+        call openai's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            headers = {
+                "Content-Type": "application/json",
+                "Authorization": "Bearer " + self.api_key
+            }
+            
+            body = args
+            body["messages"] = session.messages
+            res = requests.post(
+                self.base_url,
+                headers=headers,
+                data=json.dumps(body)
+            )
+
+            if res.status_code == 200:
+                response = res.json()
+                return {
+                    "total_tokens": response["usage"]["total_tokens"],
+                    "completion_tokens": response["usage"]["completion_tokens"],
+                    "content": response["choices"][0]["message"]["content"]
+                }
+            else:
+                response = res.json()
+                if "errors" in response:
+                    error = response.get("errors")
+                elif "error" in response:
+                    error = response.get("error")
+                else:
+                    error = "Unknown error"
+                logger.error(f"[MODELSCOPE_AI] chat failed, status_code={res.status_code}, "
+                             f"msg={error.get('message')}, type={error.get('type')}")
+
+                result = {"completion_tokens": 0, "content": "提问太快啦，请休息一下再问我吧"}
+                need_retry = False
+                if res.status_code >= 500:
+                    # server error, need retry
+                    logger.warn(f"[MODELSCOPE_AI] do retry, times={retry_count}")
+                    need_retry = retry_count < 2
+                elif res.status_code == 401:
+                    result["content"] = "授权失败，请检查API Key是否正确"
+                elif res.status_code == 429:
+                    result["content"] = "请求过于频繁，请稍后再试"
+                    need_retry = retry_count < 2
+                else:
+                    need_retry = False
+
+                if need_retry:
+                    time.sleep(3)
+                    return self.reply_text(session, args, retry_count + 1)
+                else:
+                    return result
+        except Exception as e:
+            logger.exception(e)
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if need_retry:
+                return self.reply_text(session, args, retry_count + 1)
+            else:
+                return result
+
+    def reply_text_stream(self, session: ModelScopeSession, args=None, retry_count=0) -> dict:
+        """
+        call ModelScope's ChatCompletion to get the answer with stream response
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            headers = {
+                "Content-Type": "application/json",
+                "Authorization": "Bearer " + self.api_key
+            }
+            
+            body = args
+            body["messages"] = session.messages
+            body["stream"] = True  # 启用流式响应
+
+            res = requests.post(
+                self.base_url,
+                headers=headers,
+                data=json.dumps(body),
+                stream=True
+            )
+            if res.status_code == 200:
+                content = ""
+                for line in res.iter_lines():
+                    if line:
+                        decoded_line = line.decode('utf-8')
+                        if decoded_line.startswith("data: "):
+                            try:
+                                json_data = json.loads(decoded_line[6:])
+                                delta_content = json_data.get("choices", [{}])[0].get("delta", {}).get("content", "")
+                                if delta_content:
+                                    content += delta_content
+                            except json.JSONDecodeError as e:
+                                pass
+                return {
+                    "total_tokens": 1,  # 流式响应通常不返回token使用情况
+                    "completion_tokens": 1,
+                    "content": content
+                }
+            else:
+                response = res.json()
+                if "errors" in response:
+                    error = response.get("errors")
+                elif "error" in response:
+                    error = response.get("error")
+                else:
+                    error = "Unknown error"
+                logger.error(f"[MODELSCOPE_AI] chat failed, status_code={res.status_code}, "
+                             f"msg={error.get('message')}, type={error.get('type')}")
+
+                result = {"completion_tokens": 0, "content": "提问太快啦，请休息一下再问我吧"}
+                need_retry = False
+                if res.status_code >= 500:
+                    # server error, need retry
+                    logger.warn(f"[MODELSCOPE_AI] do retry, times={retry_count}")
+                    need_retry = retry_count < 2
+                elif res.status_code == 401:
+                    result["content"] = "授权失败，请检查API Key是否正确"
+                elif res.status_code == 429:
+                    result["content"] = "请求过于频繁，请稍后再试"
+                    need_retry = retry_count < 2
+                else:
+                    need_retry = False
+
+                if need_retry:
+                    time.sleep(3)
+                    return self.reply_text_stream(session, args, retry_count + 1)
+                else:
+                    return result
+        except Exception as e:
+            logger.exception(e)
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if need_retry:
+                return self.reply_text_stream(session, args, retry_count + 1)
+            else:
+                return result
+    def create_img(self, query, retry_count=0):
+        try:
+            logger.info("[ModelScopeImage] image_query={}".format(query))
+            headers = {
+                "Content-Type": "application/json; charset=utf-8",  # 明确指定编码
+                "Authorization": f"Bearer {self.api_key}"
+            }
+            payload = {
+                "prompt": query,  # required
+                "n": 1,
+                "model": conf().get("text_to_image"),
+            }
+            url = "https://api-inference.modelscope.cn/v1/images/generations"
+            
+            # 手动序列化并保留中文（禁用 ASCII 转义）
+            json_payload = json.dumps(payload, ensure_ascii=False).encode('utf-8')
+            
+            # 使用 data 参数发送原始字符串（requests 会自动处理编码）
+            res = requests.post(url, headers=headers, data=json_payload)
+            
+            response_data = res.json()
+            image_url = response_data['images'][0]['url']
+            logger.info("[ModelScopeImage] image_url={}".format(image_url))
+            return True, image_url
+
+        except Exception as e:
+            logger.error(format(e))
+            return False, "画图出现问题，请休息一下再问我吧"

models/modelscope/modelscope_session.py

@@ -0,0 +1,51 @@
+from models.session_manager import Session
+from common.log import logger
+
+
+class ModelScopeSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="Qwen/Qwen2.5-7B-Instruct"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        self.reset()
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "user":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens,
+                                                                                       len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+
+def num_tokens_from_messages(messages, model):
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["content"])
+    return tokens

models/moonshot/moonshot_bot.py

@@ -0,0 +1,146 @@
+# encoding:utf-8
+
+import time
+
+import openai
+import openai.error
+from models.bot import Bot
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf, load_config
+from .moonshot_session import MoonshotSession
+import requests
+
+
+# ZhipuAI对话模型API
+class MoonshotBot(Bot):
+    def __init__(self):
+        super().__init__()
+        self.sessions = SessionManager(MoonshotSession, model=conf().get("model") or "moonshot-v1-128k")
+        model = conf().get("model") or "moonshot-v1-128k"
+        if model == "moonshot":
+            model = "moonshot-v1-32k"
+        self.args = {
+            "model": model,  # 对话模型的名称
+            "temperature": conf().get("temperature", 0.3),  # 如果设置，值域须为 [0, 1] 我们推荐 0.3，以达到较合适的效果。
+            "top_p": conf().get("top_p", 1.0),  # 使用默认值
+        }
+        self.api_key = conf().get("moonshot_api_key")
+        self.base_url = conf().get("moonshot_base_url", "https://api.moonshot.cn/v1/chat/completions")
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context.type == ContextType.TEXT:
+            logger.info("[MOONSHOT_AI] query={}".format(query))
+
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[MOONSHOT_AI] session query={}".format(session.messages))
+
+            model = context.get("moonshot_model")
+            new_args = self.args.copy()
+            if model:
+                new_args["model"] = model
+            # if context.get('stream'):
+            #     # reply in stream
+            #     return self.reply_text_stream(query, new_query, session_id)
+
+            reply_content = self.reply_text(session, args=new_args)
+            logger.debug(
+                "[MOONSHOT_AI] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[MOONSHOT_AI] reply {} used 0 tokens.".format(reply_content))
+            return reply
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: MoonshotSession, args=None, retry_count=0) -> dict:
+        """
+        call openai's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            headers = {
+                "Content-Type": "application/json",
+                "Authorization": "Bearer " + self.api_key
+            }
+            body = args
+            body["messages"] = session.messages
+            # logger.debug("[MOONSHOT_AI] response={}".format(response))
+            # logger.info("[MOONSHOT_AI] reply={}, total_tokens={}".format(response.choices[0]['message']['content'], response["usage"]["total_tokens"]))
+            res = requests.post(
+                self.base_url,
+                headers=headers,
+                json=body
+            )
+            if res.status_code == 200:
+                response = res.json()
+                return {
+                    "total_tokens": response["usage"]["total_tokens"],
+                    "completion_tokens": response["usage"]["completion_tokens"],
+                    "content": response["choices"][0]["message"]["content"]
+                }
+            else:
+                response = res.json()
+                error = response.get("error")
+                logger.error(f"[MOONSHOT_AI] chat failed, status_code={res.status_code}, "
+                             f"msg={error.get('message')}, type={error.get('type')}")
+
+                result = {"completion_tokens": 0, "content": "提问太快啦，请休息一下再问我吧"}
+                need_retry = False
+                if res.status_code >= 500:
+                    # server error, need retry
+                    logger.warn(f"[MOONSHOT_AI] do retry, times={retry_count}")
+                    need_retry = retry_count < 2
+                elif res.status_code == 401:
+                    result["content"] = "授权失败，请检查API Key是否正确"
+                elif res.status_code == 429:
+                    result["content"] = "请求过于频繁，请稍后再试"
+                    need_retry = retry_count < 2
+                else:
+                    need_retry = False
+
+                if need_retry:
+                    time.sleep(3)
+                    return self.reply_text(session, args, retry_count + 1)
+                else:
+                    return result
+        except Exception as e:
+            logger.exception(e)
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if need_retry:
+                return self.reply_text(session, args, retry_count + 1)
+            else:
+                return result

models/moonshot/moonshot_session.py

@@ -0,0 +1,51 @@
+from models.session_manager import Session
+from common.log import logger
+
+
+class MoonshotSession(Session):
+    def __init__(self, session_id, system_prompt=None, model="moonshot-v1-128k"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        self.reset()
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "user":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens,
+                                                                                       len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+
+def num_tokens_from_messages(messages, model):
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["content"])
+    return tokens

models/openai/open_ai_bot.py

@@ -0,0 +1,230 @@
+# encoding:utf-8
+
+import time
+
+import openai
+import openai.error
+
+from models.bot import Bot
+from models.openai_compatible_bot import OpenAICompatibleBot
+from models.openai.open_ai_image import OpenAIImage
+from models.openai.open_ai_session import OpenAISession
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf
+
+user_session = dict()
+
+
+# OpenAI对话模型API (可用)
+class OpenAIBot(Bot, OpenAIImage, OpenAICompatibleBot):
+    def __init__(self):
+        super().__init__()
+        openai.api_key = conf().get("open_ai_api_key")
+        if conf().get("open_ai_api_base"):
+            openai.api_base = conf().get("open_ai_api_base")
+        proxy = conf().get("proxy")
+        if proxy:
+            openai.proxy = proxy
+
+        self.sessions = SessionManager(OpenAISession, model=conf().get("model") or "text-davinci-003")
+        self.args = {
+            "model": conf().get("model") or "text-davinci-003",  # 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": 1200,  # 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
+            "request_timeout": conf().get("request_timeout", None),  # 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": conf().get("request_timeout", None),  # 重试超时时间，在这个时间内，将会自动重试
+            "stop": ["\n\n\n"],
+        }
+    
+    def get_api_config(self):
+        """Get API configuration for OpenAI-compatible base class"""
+        return {
+            'api_key': conf().get("open_ai_api_key"),
+            'api_base': conf().get("open_ai_api_base"),
+            'model': conf().get("model", "text-davinci-003"),
+            'default_temperature': conf().get("temperature", 0.9),
+            'default_top_p': conf().get("top_p", 1.0),
+            'default_frequency_penalty': conf().get("frequency_penalty", 0.0),
+            'default_presence_penalty': conf().get("presence_penalty", 0.0),
+        }
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context and context.type:
+            if context.type == ContextType.TEXT:
+                logger.info("[OPEN_AI] query={}".format(query))
+                session_id = context["session_id"]
+                reply = None
+                if query == "#清除记忆":
+                    self.sessions.clear_session(session_id)
+                    reply = Reply(ReplyType.INFO, "记忆已清除")
+                elif query == "#清除所有":
+                    self.sessions.clear_all_session()
+                    reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+                else:
+                    session = self.sessions.session_query(query, session_id)
+                    result = self.reply_text(session)
+                    total_tokens, completion_tokens, reply_content = (
+                        result["total_tokens"],
+                        result["completion_tokens"],
+                        result["content"],
+                    )
+                    logger.debug(
+                        "[OPEN_AI] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(str(session), session_id, reply_content, completion_tokens)
+                    )
+
+                    if total_tokens == 0:
+                        reply = Reply(ReplyType.ERROR, reply_content)
+                    else:
+                        self.sessions.session_reply(reply_content, session_id, total_tokens)
+                        reply = Reply(ReplyType.TEXT, reply_content)
+                return reply
+            elif context.type == ContextType.IMAGE_CREATE:
+                ok, retstring = self.create_img(query, 0)
+                reply = None
+                if ok:
+                    reply = Reply(ReplyType.IMAGE_URL, retstring)
+                else:
+                    reply = Reply(ReplyType.ERROR, retstring)
+                return reply
+
+    def reply_text(self, session: OpenAISession, retry_count=0):
+        try:
+            response = openai.Completion.create(prompt=str(session), **self.args)
+            res_content = response.choices[0]["text"].strip().replace("<|endoftext|>", "")
+            total_tokens = response["usage"]["total_tokens"]
+            completion_tokens = response["usage"]["completion_tokens"]
+            logger.info("[OPEN_AI] reply={}".format(res_content))
+            return {
+                "total_tokens": total_tokens,
+                "completion_tokens": completion_tokens,
+                "content": res_content,
+            }
+        except Exception as e:
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if isinstance(e, openai.error.RateLimitError):
+                logger.warn("[OPEN_AI] RateLimitError: {}".format(e))
+                result["content"] = "提问太快啦，请休息一下再问我吧"
+                if need_retry:
+                    time.sleep(20)
+            elif isinstance(e, openai.error.Timeout):
+                logger.warn("[OPEN_AI] Timeout: {}".format(e))
+                result["content"] = "我没有收到你的消息"
+                if need_retry:
+                    time.sleep(5)
+            elif isinstance(e, openai.error.APIConnectionError):
+                logger.warn("[OPEN_AI] APIConnectionError: {}".format(e))
+                need_retry = False
+                result["content"] = "我连接不到你的网络"
+            else:
+                logger.warn("[OPEN_AI] Exception: {}".format(e))
+                need_retry = False
+                self.sessions.clear_session(session.session_id)
+
+            if need_retry:
+                logger.warn("[OPEN_AI] 第{}次重试".format(retry_count + 1))
+                return self.reply_text(session, retry_count + 1)
+            else:
+                return result
+
+    def call_with_tools(self, messages, tools=None, stream=False, **kwargs):
+        """
+        Call OpenAI API with tool support for agent integration
+        Note: This bot uses the old Completion API which doesn't support tools.
+        For tool support, use ChatGPTBot instead.
+        
+        This method converts to ChatCompletion API when tools are provided.
+        
+        Args:
+            messages: List of messages
+            tools: List of tool definitions (OpenAI format)
+            stream: Whether to use streaming
+            **kwargs: Additional parameters
+            
+        Returns:
+            Formatted response in OpenAI format or generator for streaming
+        """
+        try:
+            # The old Completion API doesn't support tools
+            # We need to use ChatCompletion API instead
+            logger.info("[OPEN_AI] Using ChatCompletion API for tool support")
+            
+            # Build request parameters for ChatCompletion
+            request_params = {
+                "model": kwargs.get("model", conf().get("model") or "gpt-3.5-turbo"),
+                "messages": messages,
+                "temperature": kwargs.get("temperature", conf().get("temperature", 0.9)),
+                "top_p": kwargs.get("top_p", 1),
+                "frequency_penalty": kwargs.get("frequency_penalty", conf().get("frequency_penalty", 0.0)),
+                "presence_penalty": kwargs.get("presence_penalty", conf().get("presence_penalty", 0.0)),
+                "stream": stream
+            }
+            
+            # Add max_tokens if specified
+            if kwargs.get("max_tokens"):
+                request_params["max_tokens"] = kwargs["max_tokens"]
+            
+            # Add tools if provided
+            if tools:
+                request_params["tools"] = tools
+                request_params["tool_choice"] = kwargs.get("tool_choice", "auto")
+            
+            # Make API call using ChatCompletion
+            if stream:
+                return self._handle_stream_response(request_params)
+            else:
+                return self._handle_sync_response(request_params)
+                
+        except Exception as e:
+            logger.error(f"[OPEN_AI] call_with_tools error: {e}")
+            if stream:
+                def error_generator():
+                    yield {
+                        "error": True,
+                        "message": str(e),
+                        "status_code": 500
+                    }
+                return error_generator()
+            else:
+                return {
+                    "error": True,
+                    "message": str(e),
+                    "status_code": 500
+                }
+    
+    def _handle_sync_response(self, request_params):
+        """Handle synchronous OpenAI ChatCompletion API response"""
+        try:
+            response = openai.ChatCompletion.create(**request_params)
+            
+            logger.info(f"[OPEN_AI] call_with_tools reply, model={response.get('model')}, "
+                       f"total_tokens={response.get('usage', {}).get('total_tokens', 0)}")
+            
+            return response
+            
+        except Exception as e:
+            logger.error(f"[OPEN_AI] sync response error: {e}")
+            raise
+    
+    def _handle_stream_response(self, request_params):
+        """Handle streaming OpenAI ChatCompletion API response"""
+        try:
+            stream = openai.ChatCompletion.create(**request_params)
+            
+            for chunk in stream:
+                yield chunk
+                
+        except Exception as e:
+            logger.error(f"[OPEN_AI] stream response error: {e}")
+            yield {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }

models/openai/open_ai_image.py

@@ -0,0 +1,43 @@
+import time
+
+import openai
+from models.openai.openai_compat import RateLimitError
+
+from common.log import logger
+from common.token_bucket import TokenBucket
+from config import conf
+
+
+# OPENAI提供的画图接口
+class OpenAIImage(object):
+    def __init__(self):
+        openai.api_key = conf().get("open_ai_api_key")
+        if conf().get("rate_limit_dalle"):
+            self.tb4dalle = TokenBucket(conf().get("rate_limit_dalle", 50))
+
+    def create_img(self, query, retry_count=0, api_key=None, api_base=None):
+        try:
+            if conf().get("rate_limit_dalle") and not self.tb4dalle.get_token():
+                return False, "请求太快了，请休息一下再问我吧"
+            logger.info("[OPEN_AI] image_query={}".format(query))
+            response = openai.Image.create(
+                api_key=api_key,
+                prompt=query,  # 图片描述
+                n=1,  # 每次生成图片的数量
+                model=conf().get("text_to_image") or "dall-e-2",
+                # size=conf().get("image_create_size", "256x256"),  # 图片大小,可选有 256x256, 512x512, 1024x1024
+            )
+            image_url = response["data"][0]["url"]
+            logger.info("[OPEN_AI] image_url={}".format(image_url))
+            return True, image_url
+        except RateLimitError as e:
+            logger.warn(e)
+            if retry_count < 1:
+                time.sleep(5)
+                logger.warn("[OPEN_AI] ImgCreate RateLimit exceed, 第{}次重试".format(retry_count + 1))
+                return self.create_img(query, retry_count + 1)
+            else:
+                return False, "画图出现问题，请休息一下再问我吧"
+        except Exception as e:
+            logger.exception(e)
+            return False, "画图出现问题，请休息一下再问我吧"

models/openai/open_ai_session.py

@@ -0,0 +1,73 @@
+from models.session_manager import Session
+from common.log import logger
+
+
+class OpenAISession(Session):
+    def __init__(self, session_id, system_prompt=None, model="text-davinci-003"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        self.reset()
+
+    def __str__(self):
+        # 构造对话模型的输入
+        """
+        e.g.  Q: xxx
+              A: xxx
+              Q: xxx
+        """
+        prompt = ""
+        for item in self.messages:
+            if item["role"] == "system":
+                prompt += item["content"] + "<|endoftext|>\n\n\n"
+            elif item["role"] == "user":
+                prompt += "Q: " + item["content"] + "\n"
+            elif item["role"] == "assistant":
+                prompt += "\n\nA: " + item["content"] + "<|endoftext|>\n"
+
+        if len(self.messages) > 0 and self.messages[-1]["role"] == "user":
+            prompt += "A: "
+        return prompt
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 1:
+                self.messages.pop(0)
+            elif len(self.messages) == 1 and self.messages[0]["role"] == "assistant":
+                self.messages.pop(0)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = len(str(self))
+                break
+            elif len(self.messages) == 1 and self.messages[0]["role"] == "user":
+                logger.warn("user question exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(conversation)={}".format(max_tokens, cur_tokens, len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = len(str(self))
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_string(str(self), self.model)
+
+
+# refer to https://github.com/openai/openai-cookbook/blob/main/examples/How_to_count_tokens_with_tiktoken.ipynb
+def num_tokens_from_string(string: str, model: str) -> int:
+    """Returns the number of tokens in a text string."""
+    import tiktoken
+
+    encoding = tiktoken.encoding_for_model(model)
+    num_tokens = len(encoding.encode(string, disallowed_special=()))
+    return num_tokens

models/openai/openai_compat.py

@@ -0,0 +1,102 @@
+"""
+OpenAI compatibility layer for different versions.
+
+This module provides a compatibility layer between OpenAI library versions:
+- OpenAI < 1.0 (old API with openai.error module)
+- OpenAI >= 1.0 (new API with direct exception imports)
+"""
+
+try:
+    # Try new OpenAI >= 1.0 API
+    from openai import (
+        OpenAIError,
+        RateLimitError,
+        APIError,
+        APIConnectionError,
+        AuthenticationError,
+        APITimeoutError,
+        BadRequestError,
+    )
+    
+    # Create a mock error module for backward compatibility
+    class ErrorModule:
+        OpenAIError = OpenAIError
+        RateLimitError = RateLimitError
+        APIError = APIError
+        APIConnectionError = APIConnectionError
+        AuthenticationError = AuthenticationError
+        Timeout = APITimeoutError  # Renamed in new version
+        InvalidRequestError = BadRequestError  # Renamed in new version
+    
+    error = ErrorModule()
+    
+    # Also export with new names
+    Timeout = APITimeoutError
+    InvalidRequestError = BadRequestError
+    
+except ImportError:
+    # Fall back to old OpenAI < 1.0 API
+    try:
+        import openai.error as error
+        
+        # Export individual exceptions for direct import
+        OpenAIError = error.OpenAIError
+        RateLimitError = error.RateLimitError
+        APIError = error.APIError
+        APIConnectionError = error.APIConnectionError
+        AuthenticationError = error.AuthenticationError
+        InvalidRequestError = error.InvalidRequestError
+        Timeout = error.Timeout
+        BadRequestError = error.InvalidRequestError  # Alias
+        APITimeoutError = error.Timeout  # Alias
+    except (ImportError, AttributeError):
+        # Neither version works, create dummy classes
+        class OpenAIError(Exception):
+            pass
+        
+        class RateLimitError(OpenAIError):
+            pass
+        
+        class APIError(OpenAIError):
+            pass
+        
+        class APIConnectionError(OpenAIError):
+            pass
+        
+        class AuthenticationError(OpenAIError):
+            pass
+        
+        class InvalidRequestError(OpenAIError):
+            pass
+        
+        class Timeout(OpenAIError):
+            pass
+        
+        BadRequestError = InvalidRequestError
+        APITimeoutError = Timeout
+        
+        # Create error module
+        class ErrorModule:
+            OpenAIError = OpenAIError
+            RateLimitError = RateLimitError
+            APIError = APIError
+            APIConnectionError = APIConnectionError
+            AuthenticationError = AuthenticationError
+            InvalidRequestError = InvalidRequestError
+            Timeout = Timeout
+        
+        error = ErrorModule()
+
+# Export all for easy import
+__all__ = [
+    'error',
+    'OpenAIError',
+    'RateLimitError',
+    'APIError',
+    'APIConnectionError',
+    'AuthenticationError',
+    'InvalidRequestError',
+    'Timeout',
+    'BadRequestError',
+    'APITimeoutError',
+]

models/openai_compatible_bot.py

@@ -0,0 +1,278 @@
+# encoding:utf-8
+
+"""
+OpenAI-Compatible Bot Base Class
+
+Provides a common implementation for bots that are compatible with OpenAI's API format.
+This includes: OpenAI, LinkAI, Azure OpenAI, and many third-party providers.
+"""
+
+import json
+import openai
+from common.log import logger
+
+
+class OpenAICompatibleBot:
+    """
+    Base class for OpenAI-compatible bots.
+    
+    Provides common tool calling implementation that can be inherited by:
+    - ChatGPTBot
+    - LinkAIBot  
+    - OpenAIBot
+    - AzureChatGPTBot
+    - Other OpenAI-compatible providers
+    
+    Subclasses only need to override get_api_config() to provide their specific API settings.
+    """
+    
+    def get_api_config(self):
+        """
+        Get API configuration for this bot.
+        
+        Subclasses should override this to provide their specific config.
+        
+        Returns:
+            dict: {
+                'api_key': str,
+                'api_base': str (optional),
+                'model': str,
+                'default_temperature': float,
+                'default_top_p': float,
+                'default_frequency_penalty': float,
+                'default_presence_penalty': float,
+            }
+        """
+        raise NotImplementedError("Subclasses must implement get_api_config()")
+    
+    def call_with_tools(self, messages, tools=None, stream=False, **kwargs):
+        """
+        Call OpenAI-compatible API with tool support for agent integration
+        
+        This method handles:
+        1. Format conversion (Claude format → OpenAI format)
+        2. System prompt injection
+        3. API calling with proper configuration
+        4. Error handling
+        
+        Args:
+            messages: List of messages (may be in Claude format from agent)
+            tools: List of tool definitions (may be in Claude format from agent)
+            stream: Whether to use streaming
+            **kwargs: Additional parameters (max_tokens, temperature, system, etc.)
+            
+        Returns:
+            Formatted response in OpenAI format or generator for streaming
+        """
+        try:
+            # Get API configuration from subclass
+            api_config = self.get_api_config()
+            
+            # Convert messages from Claude format to OpenAI format
+            messages = self._convert_messages_to_openai_format(messages)
+            
+            # Convert tools from Claude format to OpenAI format
+            if tools:
+                tools = self._convert_tools_to_openai_format(tools)
+            
+            # Handle system prompt (OpenAI uses system message, Claude uses separate parameter)
+            system_prompt = kwargs.get('system')
+            if system_prompt:
+                # Add system message at the beginning if not already present
+                if not messages or messages[0].get('role') != 'system':
+                    messages = [{"role": "system", "content": system_prompt}] + messages
+                else:
+                    # Replace existing system message
+                    messages[0] = {"role": "system", "content": system_prompt}
+            
+            # Build request parameters
+            request_params = {
+                "model": kwargs.get("model", api_config.get('model', 'gpt-3.5-turbo')),
+                "messages": messages,
+                "temperature": kwargs.get("temperature", api_config.get('default_temperature', 0.9)),
+                "top_p": kwargs.get("top_p", api_config.get('default_top_p', 1.0)),
+                "frequency_penalty": kwargs.get("frequency_penalty", api_config.get('default_frequency_penalty', 0.0)),
+                "presence_penalty": kwargs.get("presence_penalty", api_config.get('default_presence_penalty', 0.0)),
+                "stream": stream
+            }
+            
+            # Add max_tokens if specified
+            if kwargs.get("max_tokens"):
+                request_params["max_tokens"] = kwargs["max_tokens"]
+            
+            # Add tools if provided
+            if tools:
+                request_params["tools"] = tools
+                request_params["tool_choice"] = kwargs.get("tool_choice", "auto")
+            
+            # Make API call with proper configuration
+            api_key = api_config.get('api_key')
+            api_base = api_config.get('api_base')
+            
+            if stream:
+                return self._handle_stream_response(request_params, api_key, api_base)
+            else:
+                return self._handle_sync_response(request_params, api_key, api_base)
+                
+        except Exception as e:
+            error_msg = str(e)
+            logger.error(f"[{self.__class__.__name__}] call_with_tools error: {error_msg}")
+            if stream:
+                def error_generator():
+                    yield {
+                        "error": True,
+                        "message": error_msg,
+                        "status_code": 500
+                    }
+                return error_generator()
+            else:
+                return {
+                    "error": True,
+                    "message": error_msg,
+                    "status_code": 500
+                }
+    
+    def _handle_sync_response(self, request_params, api_key, api_base):
+        """Handle synchronous OpenAI API response"""
+        try:
+            # Build kwargs with explicit API configuration
+            kwargs = dict(request_params)
+            if api_key:
+                kwargs["api_key"] = api_key
+            if api_base:
+                kwargs["api_base"] = api_base
+            
+            response = openai.ChatCompletion.create(**kwargs)
+            return response
+            
+        except Exception as e:
+            logger.error(f"[{self.__class__.__name__}] sync response error: {e}")
+            return {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }
+    
+    def _handle_stream_response(self, request_params, api_key, api_base):
+        """Handle streaming OpenAI API response"""
+        try:
+            # Build kwargs with explicit API configuration
+            kwargs = dict(request_params)
+            if api_key:
+                kwargs["api_key"] = api_key
+            if api_base:
+                kwargs["api_base"] = api_base
+            
+            stream = openai.ChatCompletion.create(**kwargs)
+            
+            # Stream chunks to caller
+            for chunk in stream:
+                yield chunk
+                
+        except Exception as e:
+            logger.error(f"[{self.__class__.__name__}] stream response error: {e}")
+            yield {
+                "error": True,
+                "message": str(e),
+                "status_code": 500
+            }
+    
+    def _convert_tools_to_openai_format(self, tools):
+        """
+        Convert tools from Claude format to OpenAI format
+        
+        Claude format: {name, description, input_schema}
+        OpenAI format: {type: "function", function: {name, description, parameters}}
+        """
+        if not tools:
+            return None
+        
+        openai_tools = []
+        for tool in tools:
+            # Check if already in OpenAI format
+            if 'type' in tool and tool['type'] == 'function':
+                openai_tools.append(tool)
+            else:
+                # Convert from Claude format
+                openai_tools.append({
+                    "type": "function",
+                    "function": {
+                        "name": tool.get("name"),
+                        "description": tool.get("description"),
+                        "parameters": tool.get("input_schema", {})
+                    }
+                })
+        
+        return openai_tools
+    
+    def _convert_messages_to_openai_format(self, messages):
+        """
+        Convert messages from Claude format to OpenAI format
+        
+        Claude uses content blocks with types like 'tool_use', 'tool_result'
+        OpenAI uses 'tool_calls' in assistant messages and 'tool' role for results
+        """
+        if not messages:
+            return []
+        
+        openai_messages = []
+        
+        for msg in messages:
+            role = msg.get("role")
+            content = msg.get("content")
+            
+            # Handle string content (already in correct format)
+            if isinstance(content, str):
+                openai_messages.append(msg)
+                continue
+            
+            # Handle list content (Claude format with content blocks)
+            if isinstance(content, list):
+                # Check if this is a tool result message (user role with tool_result blocks)
+                if role == "user" and any(block.get("type") == "tool_result" for block in content):
+                    # Convert each tool_result block to a separate tool message
+                    for block in content:
+                        if block.get("type") == "tool_result":
+                            openai_messages.append({
+                                "role": "tool",
+                                "tool_call_id": block.get("tool_use_id"),
+                                "content": block.get("content", "")
+                            })
+                
+                # Check if this is an assistant message with tool_use blocks
+                elif role == "assistant":
+                    # Separate text content and tool_use blocks
+                    text_parts = []
+                    tool_calls = []
+                    
+                    for block in content:
+                        if block.get("type") == "text":
+                            text_parts.append(block.get("text", ""))
+                        elif block.get("type") == "tool_use":
+                            tool_calls.append({
+                                "id": block.get("id"),
+                                "type": "function",
+                                "function": {
+                                    "name": block.get("name"),
+                                    "arguments": json.dumps(block.get("input", {}))
+                                }
+                            })
+                    
+                    # Build OpenAI format assistant message
+                    openai_msg = {
+                        "role": "assistant",
+                        "content": " ".join(text_parts) if text_parts else None
+                    }
+                    
+                    if tool_calls:
+                        openai_msg["tool_calls"] = tool_calls
+                    
+                    openai_messages.append(openai_msg)
+                else:
+                    # Other list content, keep as is
+                    openai_messages.append(msg)
+            else:
+                # Other formats, keep as is
+                openai_messages.append(msg)
+        
+        return openai_messages

models/session_manager.py

@@ -0,0 +1,91 @@
+from common.expired_dict import ExpiredDict
+from common.log import logger
+from config import conf
+
+
+class Session(object):
+    def __init__(self, session_id, system_prompt=None):
+        self.session_id = session_id
+        self.messages = []
+        if system_prompt is None:
+            self.system_prompt = conf().get("character_desc", "")
+        else:
+            self.system_prompt = system_prompt
+
+    # 重置会话
+    def reset(self):
+        system_item = {"role": "system", "content": self.system_prompt}
+        self.messages = [system_item]
+
+    def set_system_prompt(self, system_prompt):
+        self.system_prompt = system_prompt
+        self.reset()
+
+    def add_query(self, query):
+        user_item = {"role": "user", "content": query}
+        self.messages.append(user_item)
+
+    def add_reply(self, reply):
+        assistant_item = {"role": "assistant", "content": reply}
+        self.messages.append(assistant_item)
+
+    def discard_exceeding(self, max_tokens=None, cur_tokens=None):
+        raise NotImplementedError
+
+    def calc_tokens(self):
+        raise NotImplementedError
+
+
+class SessionManager(object):
+    def __init__(self, sessioncls, **session_args):
+        if conf().get("expires_in_seconds"):
+            sessions = ExpiredDict(conf().get("expires_in_seconds"))
+        else:
+            sessions = dict()
+        self.sessions = sessions
+        self.sessioncls = sessioncls
+        self.session_args = session_args
+
+    def build_session(self, session_id, system_prompt=None):
+        """
+        如果session_id不在sessions中，创建一个新的session并添加到sessions中
+        如果system_prompt不会空，会更新session的system_prompt并重置session
+        """
+        if session_id is None:
+            return self.sessioncls(session_id, system_prompt, **self.session_args)
+
+        if session_id not in self.sessions:
+            self.sessions[session_id] = self.sessioncls(session_id, system_prompt, **self.session_args)
+        elif system_prompt is not None:  # 如果有新的system_prompt，更新并重置session
+            self.sessions[session_id].set_system_prompt(system_prompt)
+        session = self.sessions[session_id]
+        return session
+
+    def session_query(self, query, session_id):
+        session = self.build_session(session_id)
+        session.add_query(query)
+        try:
+            max_tokens = conf().get("conversation_max_tokens", 1000)
+            total_tokens = session.discard_exceeding(max_tokens, None)
+            logger.debug("prompt tokens used={}".format(total_tokens))
+        except Exception as e:
+            logger.warning("Exception when counting tokens precisely for prompt: {}".format(str(e)))
+        return session
+
+    def session_reply(self, reply, session_id, total_tokens=None):
+        session = self.build_session(session_id)
+        session.add_reply(reply)
+        try:
+            max_tokens = conf().get("conversation_max_tokens", 1000)
+            tokens_cnt = session.discard_exceeding(max_tokens, total_tokens)
+            logger.debug("raw total_tokens={}, savesession tokens={}".format(total_tokens, tokens_cnt))
+        except Exception as e:
+            logger.warning("Exception when counting tokens precisely for session: {}".format(str(e)))
+        return session
+
+    def clear_session(self, session_id):
+        if session_id in self.sessions:
+            del self.sessions[session_id]
+
+    def clear_all_session(self):
+        self.sessions.clear()

models/xunfei/xunfei_spark_bot.py

@@ -0,0 +1,269 @@
+# encoding:utf-8
+
+import requests, json
+from models.bot import Bot
+from models.session_manager import SessionManager
+from models.chatgpt.chat_gpt_session import ChatGPTSession
+from bridge.context import ContextType, Context
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf
+from common import const
+import time
+import _thread as thread
+import datetime
+from datetime import datetime
+from wsgiref.handlers import format_date_time
+from urllib.parse import urlencode
+import base64
+import ssl
+import hashlib
+import hmac
+import json
+from time import mktime
+from urllib.parse import urlparse
+import websocket
+import queue
+import threading
+import random
+
+# 消息队列 map
+queue_map = dict()
+
+# 响应队列 map
+reply_map = dict()
+
+
+class XunFeiBot(Bot):
+    def __init__(self):
+        super().__init__()
+        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: "generalv2"
+        # Spark Lite请求地址(spark_url): wss://spark-api.xf-yun.com/v1.1/chat, 对应的domain参数为: "lite"
+        # Spark V2.0请求地址(spark_url): wss://spark-api.xf-yun.com/v2.1/chat, 对应的domain参数为: "generalv2"
+        # Spark Pro 请求地址(spark_url): wss://spark-api.xf-yun.com/v3.1/chat, 对应的domain参数为: "generalv3"
+        # Spark Pro-128K请求地址(spark_url):  wss://spark-api.xf-yun.com/chat/pro-128k, 对应的domain参数为: "pro-128k"
+        # Spark Max 请求地址(spark_url): wss://spark-api.xf-yun.com/v3.5/chat, 对应的domain参数为: "generalv3.5"
+        # Spark4.0 Ultra 请求地址(spark_url): wss://spark-api.xf-yun.com/v4.0/chat, 对应的domain参数为: "4.0Ultra"
+        # 后续模型更新，对应的参数可以参考官网文档获取：https://www.xfyun.cn/doc/spark/Web.html
+        self.domain = conf().get("xunfei_domain", "generalv3.5")
+        self.spark_url = conf().get("xunfei_spark_url", "wss://spark-api.xf-yun.com/v3.5/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
+        self.sessions = SessionManager(ChatGPTSession, model=const.XUNFEI)
+
+    def reply(self, query, context: Context = None) -> Reply:
+        if context.type == ContextType.TEXT:
+            logger.info("[XunFei] query={}".format(query))
+            session_id = context["session_id"]
+            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()
+            depth = 0
+            time.sleep(0.1)
+            t1 = time.time()
+            usage = {}
+            while depth <= 300:
+                try:
+                    data_queue = queue_map.get(request_id)
+                    if not data_queue:
+                        depth += 1
+                        time.sleep(0.1)
+                        continue
+                    data_item = data_queue.get(block=True, timeout=0.1)
+                    if data_item.is_end:
+                        # 请求结束
+                        del queue_map[request_id]
+                        if data_item.reply:
+                            reply_map[request_id] += data_item.reply
+                        usage = data_item.usage
+                        break
+
+                    reply_map[request_id] += data_item.reply
+                    depth += 1
+                except Exception as e:
+                    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"))
+            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))
+            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,
+                                    on_open=on_open)
+        data_queue = queue.Queue(1000)
+        queue_map[session_id] = data_queue
+        ws.appid = self.app_id
+        ws.question = prompt
+        ws.domain = self.domain
+        ws.session_id = session_id
+        ws.temperature = temperature
+        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))
+
+    # 生成url
+    def create_url(self):
+        # 生成RFC1123格式的时间戳
+        now = datetime.now()
+        date = format_date_time(mktime(now.timetuple()))
+
+        # 拼接字符串
+        signature_origin = "host: " + self.host + "\n"
+        signature_origin += "date: " + date + "\n"
+        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'),
+                                 digestmod=hashlib.sha256).digest()
+
+        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')
+
+        # 将请求的鉴权参数组合为字典
+        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
+        return url
+
+    def gen_params(self, appid, domain, question):
+        """
+        通过appid和用户的提问来生成请参数
+        """
+        data = {
+            "header": {
+                "app_id": appid,
+                "uid": "1234"
+            },
+            "parameter": {
+                "chat": {
+                    "domain": domain,
+                    "random_threshold": 0.5,
+                    "max_tokens": 2048,
+                    "auditing": "default"
+                }
+            },
+            "payload": {
+                "message": {
+                    "text": question
+                }
+            }
+        }
+        return data
+
+
+class ReplyItem:
+    def __init__(self, reply, usage=None, is_end=False):
+        self.is_end = is_end
+        self.reply = reply
+        self.usage = usage
+
+
+# 收到websocket错误的处理
+def on_error(ws, error):
+    logger.error(f"[XunFei] error: {str(error)}")
+
+
+# 收到websocket关闭的处理
+def on_close(ws, one, two):
+    data_queue = queue_map.get(ws.session_id)
+    data_queue.put("END")
+
+
+# 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, ))
+
+
+def run(ws, *args):
+    data = json.dumps(
+        gen_params(appid=ws.appid,
+                   domain=ws.domain,
+                   question=ws.question,
+                   temperature=ws.temperature))
+    ws.send(data)
+
+
+# Websocket 操作
+# 收到websocket消息的处理
+def on_message(ws, message):
+    data = json.loads(message)
+    code = data['header']['code']
+    if code != 0:
+        logger.error(f'请求错误: {code}, {data}')
+        ws.close()
+    else:
+        choices = data["payload"]["choices"]
+        status = choices["status"]
+        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}")
+            return
+        reply_item = ReplyItem(content)
+        if status == 2:
+            usage = data["payload"].get("usage")
+            reply_item = ReplyItem(content, usage)
+            reply_item.is_end = True
+            ws.close()
+        data_queue.put(reply_item)
+
+
+def gen_params(appid, domain, question, temperature=0.5):
+    """
+    通过appid和用户的提问来生成请参数
+    """
+    data = {
+        "header": {
+            "app_id": appid,
+            "uid": "1234"
+        },
+        "parameter": {
+            "chat": {
+                "domain": domain,
+                "temperature": temperature,
+                "random_threshold": 0.5,
+                "max_tokens": 2048,
+                "auditing": "default"
+            }
+        },
+        "payload": {
+            "message": {
+                "text": question
+            }
+        }
+    }
+    return data

models/zhipuai/zhipu_ai_image.py

@@ -0,0 +1,29 @@
+from common.log import logger
+from config import conf
+
+
+# ZhipuAI提供的画图接口
+
+class ZhipuAIImage(object):
+    def __init__(self):
+        from zhipuai import ZhipuAI
+        self.client = ZhipuAI(api_key=conf().get("zhipu_ai_api_key"))
+
+    def create_img(self, query, retry_count=0, api_key=None, api_base=None):
+        try:
+            if conf().get("rate_limit_dalle"):
+                return False, "请求太快了，请休息一下再问我吧"
+            logger.info("[ZHIPU_AI] image_query={}".format(query))
+            response = self.client.images.generations(
+                prompt=query,
+                n=1,  # 每次生成图片的数量
+                model=conf().get("text_to_image") or "cogview-3",
+                size=conf().get("image_create_size", "1024x1024"),  # 图片大小,可选有 256x256, 512x512, 1024x1024
+                quality="standard",
+            )
+            image_url = response.data[0].url
+            logger.info("[ZHIPU_AI] image_url={}".format(image_url))
+            return True, image_url
+        except Exception as e:
+            logger.exception(e)
+            return False, "画图出现问题，请休息一下再问我吧"

models/zhipuai/zhipu_ai_session.py

@@ -0,0 +1,53 @@
+from models.session_manager import Session
+from common.log import logger
+
+
+class ZhipuAISession(Session):
+    def __init__(self, session_id, system_prompt=None, model="glm-4"):
+        super().__init__(session_id, system_prompt)
+        self.model = model
+        self.reset()
+        if not system_prompt:
+            logger.warn("[ZhiPu] `character_desc` can not be empty")
+
+    def discard_exceeding(self, max_tokens, cur_tokens=None):
+        precise = True
+        try:
+            cur_tokens = self.calc_tokens()
+        except Exception as e:
+            precise = False
+            if cur_tokens is None:
+                raise e
+            logger.debug("Exception when counting tokens precisely for query: {}".format(e))
+        while cur_tokens > max_tokens:
+            if len(self.messages) > 2:
+                self.messages.pop(1)
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "assistant":
+                self.messages.pop(1)
+                if precise:
+                    cur_tokens = self.calc_tokens()
+                else:
+                    cur_tokens = cur_tokens - max_tokens
+                break
+            elif len(self.messages) == 2 and self.messages[1]["role"] == "user":
+                logger.warn("user message exceed max_tokens. total_tokens={}".format(cur_tokens))
+                break
+            else:
+                logger.debug("max_tokens={}, total_tokens={}, len(messages)={}".format(max_tokens, cur_tokens,
+                                                                                       len(self.messages)))
+                break
+            if precise:
+                cur_tokens = self.calc_tokens()
+            else:
+                cur_tokens = cur_tokens - max_tokens
+        return cur_tokens
+
+    def calc_tokens(self):
+        return num_tokens_from_messages(self.messages, self.model)
+
+
+def num_tokens_from_messages(messages, model):
+    tokens = 0
+    for msg in messages:
+        tokens += len(msg["content"])
+    return tokens

models/zhipuai/zhipuai_bot.py

@@ -0,0 +1,149 @@
+# encoding:utf-8
+
+import time
+
+import openai
+import openai.error
+from models.bot import Bot
+from models.zhipuai.zhipu_ai_session import ZhipuAISession
+from models.zhipuai.zhipu_ai_image import ZhipuAIImage
+from models.session_manager import SessionManager
+from bridge.context import ContextType
+from bridge.reply import Reply, ReplyType
+from common.log import logger
+from config import conf, load_config
+from zhipuai import ZhipuAI
+
+
+# ZhipuAI对话模型API
+class ZHIPUAIBot(Bot, ZhipuAIImage):
+    def __init__(self):
+        super().__init__()
+        self.sessions = SessionManager(ZhipuAISession, model=conf().get("model") or "ZHIPU_AI")
+        self.args = {
+            "model": conf().get("model") or "glm-4",  # 对话模型的名称
+            "temperature": conf().get("temperature", 0.9),  # 值在(0,1)之间(智谱AI 的温度不能取 0 或者 1)
+            "top_p": conf().get("top_p", 0.7),  # 值在(0,1)之间(智谱AI 的 top_p 不能取 0 或者 1)
+        }
+        self.client = ZhipuAI(api_key=conf().get("zhipu_ai_api_key"))
+
+    def reply(self, query, context=None):
+        # acquire reply content
+        if context.type == ContextType.TEXT:
+            logger.info("[ZHIPU_AI] query={}".format(query))
+
+            session_id = context["session_id"]
+            reply = None
+            clear_memory_commands = conf().get("clear_memory_commands", ["#清除记忆"])
+            if query in clear_memory_commands:
+                self.sessions.clear_session(session_id)
+                reply = Reply(ReplyType.INFO, "记忆已清除")
+            elif query == "#清除所有":
+                self.sessions.clear_all_session()
+                reply = Reply(ReplyType.INFO, "所有人记忆已清除")
+            elif query == "#更新配置":
+                load_config()
+                reply = Reply(ReplyType.INFO, "配置已更新")
+            if reply:
+                return reply
+            session = self.sessions.session_query(query, session_id)
+            logger.debug("[ZHIPU_AI] session query={}".format(session.messages))
+
+            api_key = context.get("openai_api_key") or openai.api_key
+            model = context.get("gpt_model")
+            new_args = None
+            if model:
+                new_args = self.args.copy()
+                new_args["model"] = model
+            # if context.get('stream'):
+            #     # reply in stream
+            #     return self.reply_text_stream(query, new_query, session_id)
+
+            reply_content = self.reply_text(session, api_key, args=new_args)
+            logger.debug(
+                "[ZHIPU_AI] new_query={}, session_id={}, reply_cont={}, completion_tokens={}".format(
+                    session.messages,
+                    session_id,
+                    reply_content["content"],
+                    reply_content["completion_tokens"],
+                )
+            )
+            if reply_content["completion_tokens"] == 0 and len(reply_content["content"]) > 0:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+            elif reply_content["completion_tokens"] > 0:
+                self.sessions.session_reply(reply_content["content"], session_id, reply_content["total_tokens"])
+                reply = Reply(ReplyType.TEXT, reply_content["content"])
+            else:
+                reply = Reply(ReplyType.ERROR, reply_content["content"])
+                logger.debug("[ZHIPU_AI] reply {} used 0 tokens.".format(reply_content))
+            return reply
+        elif context.type == ContextType.IMAGE_CREATE:
+            ok, retstring = self.create_img(query, 0)
+            reply = None
+            if ok:
+                reply = Reply(ReplyType.IMAGE_URL, retstring)
+            else:
+                reply = Reply(ReplyType.ERROR, retstring)
+            return reply
+
+        else:
+            reply = Reply(ReplyType.ERROR, "Bot不支持处理{}类型的消息".format(context.type))
+            return reply
+
+    def reply_text(self, session: ZhipuAISession, api_key=None, args=None, retry_count=0) -> dict:
+        """
+        call openai's ChatCompletion to get the answer
+        :param session: a conversation session
+        :param session_id: session id
+        :param retry_count: retry count
+        :return: {}
+        """
+        try:
+            # if conf().get("rate_limit_chatgpt") and not self.tb4chatgpt.get_token():
+            #     raise openai.error.RateLimitError("RateLimitError: rate limit exceeded")
+            # if api_key == None, the default openai.api_key will be used
+            if args is None:
+                args = self.args
+            # response = openai.ChatCompletion.create(api_key=api_key, messages=session.messages, **args)
+            response = self.client.chat.completions.create(messages=session.messages, **args)
+            # logger.debug("[ZHIPU_AI] response={}".format(response))
+            # logger.info("[ZHIPU_AI] reply={}, total_tokens={}".format(response.choices[0]['message']['content'], response["usage"]["total_tokens"]))
+
+            return {
+                "total_tokens": response.usage.total_tokens,
+                "completion_tokens": response.usage.completion_tokens,
+                "content": response.choices[0].message.content,
+            }
+        except Exception as e:
+            need_retry = retry_count < 2
+            result = {"completion_tokens": 0, "content": "我现在有点累了，等会再来吧"}
+            if isinstance(e, openai.error.RateLimitError):
+                logger.warn("[ZHIPU_AI] RateLimitError: {}".format(e))
+                result["content"] = "提问太快啦，请休息一下再问我吧"
+                if need_retry:
+                    time.sleep(20)
+            elif isinstance(e, openai.error.Timeout):
+                logger.warn("[ZHIPU_AI] Timeout: {}".format(e))
+                result["content"] = "我没有收到你的消息"
+                if need_retry:
+                    time.sleep(5)
+            elif isinstance(e, openai.error.APIError):
+                logger.warn("[ZHIPU_AI] Bad Gateway: {}".format(e))
+                result["content"] = "请再问我一次"
+                if need_retry:
+                    time.sleep(10)
+            elif isinstance(e, openai.error.APIConnectionError):
+                logger.warn("[ZHIPU_AI] APIConnectionError: {}".format(e))
+                result["content"] = "我连接不到你的网络"
+                if need_retry:
+                    time.sleep(5)
+            else:
+                logger.exception("[ZHIPU_AI] Exception: {}".format(e), e)
+                need_retry = False
+                self.sessions.clear_session(session.session_id)
+
+            if need_retry:
+                logger.warn("[ZHIPU_AI] 第{}次重试".format(retry_count + 1))
+                return self.reply_text(session, api_key, args, retry_count + 1)
+            else:
+                return result