优化工具定义；添加 openai 兼容的统一 chat 接口 (#3570)

- 修复：
    - Qwen Agent 的 OutputParser 不再抛出异常，遇到非 COT 文本直接返回
    - CallbackHandler 正确处理工具调用信息

- 重写 tool 定义方式：
    - 添加 regist_tool 简化 tool 定义:
        - 可以指定一个用户友好的名称
        - 自动将函数的 __doc__ 作为 tool.description
	- 支持用 Field 定义参数，不再需要额外定义 ModelSchema
        - 添加 BaseToolOutput 封装 tool	返回结果，以便同时获取原始值、给LLM的字符串值
        - 支持工具热加载（有待测试）

- 增加 openai 兼容的统一 chat 接口，通过 tools/tool_choice/extra_body 不同参数组合支持：
    - Agent 对话
    - 指定工具调用（如知识库RAG）
    - LLM 对话

- 根据后端功能更新 webui

chatchat-server/chatchat/server/agent/agent_factory/qwen_agent.py

@@ -92,7 +92,8 @@ class QwenChatAgentOutputParserCustom(StructuredChatOutputParser):
             s = s[-1]
             return AgentFinish({"output": s}, log=text)
         else:
-            raise OutputParserException(f"Could not parse LLM output: {text}")
+            return AgentFinish({"output": text}, log=text)
+            # raise OutputParserException(f"Could not parse LLM output: {text}")
 
     @property
     def _type(self) -> str:

chatchat-server/chatchat/server/agent/tools_factory/aqa_processor.py

@@ -1,7 +1,7 @@
 import base64
 from chatchat.server.pydantic_v1 import Field
 from chatchat.server.utils import get_tool_config
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
 def save_base64_audio(base64_audio, file_path):
@@ -16,8 +16,8 @@ def aqa_run(model, tokenizer, query):
     return response
 
 
-@regist_tool
+@regist_tool(title="音频问答")
-def aqa_processor(query: str = Field(description="The question of the image in English")):
+def aqa_processor(query: str = Field(description="The question of the audio in English")):
     '''use this tool to get answer for audio question'''
 
     from chatchat.server.agent.container import container
@@ -28,6 +28,8 @@ def aqa_processor(query: str = Field(description="The question of the image in E
             "audio": file_path,
             "text": query,
         }
-        return aqa_run(tokenizer=container.audio_tokenizer, query=query_input, model=container.audio_model)
+        ret = aqa_run(tokenizer=container.audio_tokenizer, query=query_input, model=container.audio_model)
     else:
-        return "No Audio, Please Try Again"
+        ret = "No Audio, Please Try Again"
+    
+    return BaseToolOutput(ret)

chatchat-server/chatchat/server/agent/tools_factory/arxiv.py

@@ -1,12 +1,12 @@
 # LangChain 的 ArxivQueryRun 工具
 from chatchat.server.pydantic_v1 import Field
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
-@regist_tool
+@regist_tool(title="ARXIV论文")
 def arxiv(query: str = Field(description="The search query title")):
     '''A wrapper around Arxiv.org for searching and retrieving scientific articles in various fields.'''
     from langchain.tools.arxiv.tool import ArxivQueryRun
 
     tool = ArxivQueryRun()
-    return tool.run(tool_input=query)
+    return BaseToolOutput(tool.run(tool_input=query))

chatchat-server/chatchat/server/agent/tools_factory/calculate.py

@@ -1,8 +1,8 @@
 from chatchat.server.pydantic_v1 import Field
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
-@regist_tool
+@regist_tool(title="数学计算器")
 def calculate(text: str = Field(description="a math expression")) -> float:
     '''
     Useful to answer questions about simple calculations.
@@ -11,6 +11,8 @@ def calculate(text: str = Field(description="a math expression")) -> float:
     import numexpr
 
     try:
-        return str(numexpr.evaluate(text))
+        ret = str(numexpr.evaluate(text))
     except Exception as e:
-        return f"wrong: {e}"
+        ret = f"wrong: {e}"
+
+    return BaseToolOutput(ret)

chatchat-server/chatchat/server/agent/tools_factory/search_internet.py

@@ -9,7 +9,7 @@ from strsimpy.normalized_levenshtein import NormalizedLevenshtein
 
 from chatchat.server.utils import get_tool_config
 from chatchat.server.pydantic_v1 import Field
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
 def bing_search(text, config):
@@ -95,8 +95,8 @@ def search_engine(query: str,
     return context
 
 
-@regist_tool
+@regist_tool(title="互联网搜索")
 def search_internet(query: str = Field(description="query for Internet search")):
     '''Use this tool to use bing search engine to search the internet and get information.'''
     tool_config = get_tool_config("search_internet")
-    return search_engine(query=query, config=tool_config)
+    return BaseToolOutput(search_engine(query=query, config=tool_config))

chatchat-server/chatchat/server/agent/tools_factory/search_local_knowledgebase.py

@@ -1,8 +1,9 @@
 from urllib.parse import urlencode
 from chatchat.server.utils import get_tool_config
 from chatchat.server.pydantic_v1 import Field
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
-from chatchat.server.knowledge_base.kb_doc_api import search_docs
+from chatchat.server.knowledge_base.kb_api import list_kbs
+from chatchat.server.knowledge_base.kb_doc_api import search_docs, DocumentWithVSId
 from chatchat.configs import KB_INFO
 
 
@@ -11,35 +12,44 @@ KB_info_str = '\n'.join([f"{key}: {value}" for key, value in KB_INFO.items()])
 template_knowledge = template.format(KB_info=KB_info_str, key="samples")
 
 
+class KBToolOutput(BaseToolOutput):
+    def __str__(self) -> str:
+        context = ""
+        docs = self.data
+        source_documents = []
+
+        for inum, doc in enumerate(docs):
+            doc = DocumentWithVSId.parse_obj(doc)
+            filename = doc.metadata.get("source")
+            parameters = urlencode({"knowledge_base_name": self.extras.get("database"), "file_name": filename})
+            url = f"download_doc?" + parameters
+            text = f"""出处 [{inum + 1}] [{filename}]({url}) \n\n{doc.page_content}\n\n"""
+            source_documents.append(text)
+
+        if len(source_documents) == 0:
+            context = "没有找到相关文档,请更换关键词重试"
+        else:
+            for doc in source_documents:
+                context += doc + "\n"
+
+        return context
+
+
 def search_knowledgebase(query: str, database: str, config: dict):
     docs = search_docs(
         query=query,
         knowledge_base_name=database,
         top_k=config["top_k"],
         score_threshold=config["score_threshold"])
-    context = ""
+    return docs
-    source_documents = []
-    for inum, doc in enumerate(docs):
-        filename = doc.metadata.get("source")
-        parameters = urlencode({"knowledge_base_name": database, "file_name": filename})
-        url = f"download_doc?" + parameters
-        text = f"""出处 [{inum + 1}] [{filename}]({url}) \n\n{doc.page_content}\n\n"""
-        source_documents.append(text)
-
-    if len(source_documents) == 0:
-        context = "没有找到相关文档,请更换关键词重试"
-    else:
-        for doc in source_documents:
-            context += doc + "\n"
-
-    return context
 
 
-@regist_tool(description=template_knowledge)
+@regist_tool(description=template_knowledge, title="本地知识库")
 def search_local_knowledgebase(
-    database: str = Field(description="Database for Knowledge Search"),
+    database: str = Field(description="Database for Knowledge Search", choices=list_kbs().data),
     query: str = Field(description="Query for Knowledge Search"),
 ):
     ''''''
     tool_config = get_tool_config("search_local_knowledgebase")
-    return search_knowledgebase(query=query, database=database, config=tool_config)
+    ret = search_knowledgebase(query=query, database=database, config=tool_config)
+    return BaseToolOutput(ret, database=database)

chatchat-server/chatchat/server/agent/tools_factory/search_youtube.py

@@ -1,10 +1,10 @@
 from chatchat.server.pydantic_v1 import Field
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
-@regist_tool
+@regist_tool(title="油管视频")
 def search_youtube(query: str = Field(description="Query for Videos search")):
     '''use this tools_factory to search youtube videos'''
     from langchain_community.tools import YouTubeSearchTool
     tool = YouTubeSearchTool()
-    return tool.run(tool_input=query)
+    return BaseToolOutput(tool.run(tool_input=query))

chatchat-server/chatchat/server/agent/tools_factory/shell.py

@@ -2,11 +2,11 @@
 from langchain.tools.shell import ShellTool
 
 from chatchat.server.pydantic_v1 import Field
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
-@regist_tool
+@regist_tool(title="系统命令")
 def shell(query: str = Field(description="The command to execute")):
     '''Use Shell to execute system shell commands'''
     tool = ShellTool()
-    return tool.run(tool_input=query)
+    return BaseToolOutput(tool.run(tool_input=query))

chatchat-server/chatchat/server/agent/tools_factory/text2image.py

@@ -7,7 +7,7 @@ import uuid
 
 from chatchat.server.pydantic_v1 import Field
 from chatchat.server.utils import get_tool_config
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 import openai
 
 from chatchat.configs.basic_config import MEDIA_PATH
@@ -26,7 +26,7 @@ def get_image_model_config() -> dict:
             return config
 
 
-@regist_tool(return_direct=True)
+@regist_tool(title="文生图", return_direct=True)
 def text2images(
     prompt: str,
     n: int = Field(1, description="需生成图片的数量"),
@@ -56,7 +56,7 @@ def text2images(
         with open(os.path.join(MEDIA_PATH, filename), "wb") as fp:
             fp.write(base64.b64decode(x.b64_json))
         images.append(filename)
-    return json.dumps({"message_type": MsgType.IMAGE, "images": images})
+    return BaseToolOutput({"message_type": MsgType.IMAGE, "images": images}, format="json")
 
 
 if __name__ == "__main__":

chatchat-server/chatchat/server/agent/tools_factory/tools_registry.py

@@ -1,15 +1,22 @@
+import json
 import re
 from typing import Any, Union, Dict, Tuple, Callable, Optional, Type
 
 from langchain.agents import tool
 from langchain_core.tools import BaseTool
 
-from chatchat.server.pydantic_v1 import BaseModel
+from chatchat.server.pydantic_v1 import BaseModel, Extra
+
+
+__all__ = ["regist_tool", "BaseToolOutput"]
 
 
 _TOOLS_REGISTRY = {}
 
 
+# patch BaseTool to support extra fields e.g. a title
+BaseTool.Config.extra = Extra.allow
+
 ################################### TODO: workaround to langchain #15855
 # patch BaseTool to support tool parameters defined using pydantic Field
 
@@ -60,6 +67,7 @@ BaseTool._to_args_and_kwargs = _new_to_args_and_kwargs
 
 def regist_tool(
     *args: Any,
+    title: str = "",
     description: str = "",
     return_direct: bool = False,
     args_schema: Optional[Type[BaseModel]] = None,
@@ -70,9 +78,10 @@ def regist_tool(
     add tool to regstiry automatically
     '''
     def _parse_tool(t: BaseTool):
-        nonlocal description
+        nonlocal description, title
 
         _TOOLS_REGISTRY[t.name] = t
+
         # change default description
         if not description:
             if t.func is not None:
@@ -80,6 +89,10 @@ def regist_tool(
             elif t.coroutine is not None:
                 description = t.coroutine.__doc__ 
         t.description = " ".join(re.split(r"\n+\s*", description))
+        # set a default title for human
+        if not title:
+            title = "".join([x.capitalize() for x in t.name.split("_")])
+        t.title = title
 
     def wrapper(def_func: Callable) -> BaseTool:
         partial_ = tool(*args,
@@ -101,3 +114,30 @@ def regist_tool(
                 )
         _parse_tool(t)
         return t
+
+
+class BaseToolOutput:
+    '''
+    LLM 要求 Tool 的输出为 str，但 Tool 用在别处时希望它正常返回结构化数据。
+    只需要将 Tool 返回值用该类封装，能同时满足两者的需要。
+    基类简单的将返回值字符串化，或指定 format="json" 将其转为 json。
+    用户也可以继承该类定义自己的转换方法。
+    '''
+    def __init__(
+        self,
+        data: Any,
+        format: str="",
+        data_alias: str="",
+        **extras: Any,
+    ) -> None:
+        self.data = data
+        self.format = format
+        self.extras = extras
+        if data_alias:
+            setattr(self, data_alias, property(lambda obj: obj.data))
+    
+    def __str__(self) -> str:
+        if self.format == "json":
+            return json.dumps(self.data, ensure_ascii=False, indent=2)
+        else:
+            return str(self.data)

chatchat-server/chatchat/server/agent/tools_factory/vqa_processor.py

@@ -6,7 +6,7 @@ from io import BytesIO
 from PIL import Image, ImageDraw
 from chatchat.server.pydantic_v1 import Field
 from chatchat.server.utils import get_tool_config
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 import re
 from chatchat.server.agent.container import container
 
@@ -99,7 +99,7 @@ def vqa_run(model, tokenizer, image_base_64, query, history=[], device="cuda", m
     return response
 
 
-@regist_tool
+@regist_tool(title="图片对话")
 def vqa_processor(query: str = Field(description="The question of the image in English")):
     '''use this tool to get answer for image question'''
 
@@ -119,6 +119,8 @@ def vqa_processor(query: str = Field(description="The question of the image in E
         # end_marker = "Grounded Operation:"
         # ans = extract_between_markers(ans, start_marker, end_marker)
 
-        return ans
+        ret = ans
     else:
-        return "No Image, Please Try Again"
+        ret = "No Image, Please Try Again"
+
+    return BaseToolOutput(ret)

chatchat-server/chatchat/server/agent/tools_factory/weather_check.py

@@ -3,11 +3,11 @@
 """
 from chatchat.server.pydantic_v1 import Field
 from chatchat.server.utils import get_tool_config
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 import requests
 
 
-@regist_tool
+@regist_tool(title="天气查询")
 def weather_check(city: str = Field(description="City name,include city and county,like '厦门'")):
     '''Use this tool to check the weather at a specific city'''
 
@@ -21,7 +21,7 @@ def weather_check(city: str = Field(description="City name,include city and coun
             "temperature": data["results"][0]["now"]["temperature"],
             "description": data["results"][0]["now"]["text"],
         }
-        return weather
+        return BaseToolOutput(weather)
     else:
         raise Exception(
             f"Failed to retrieve weather: {response.status_code}")

chatchat-server/chatchat/server/agent/tools_factory/wolfram.py

@@ -2,7 +2,7 @@
 
 from chatchat.server.pydantic_v1 import Field
 from chatchat.server.utils import get_tool_config
-from .tools_registry import regist_tool
+from .tools_registry import regist_tool, BaseToolOutput
 
 
 @regist_tool
@@ -13,4 +13,4 @@ def wolfram(query: str = Field(description="The formula to be calculated")):
 
     wolfram = WolframAlphaAPIWrapper(wolfram_alpha_appid=get_tool_config("wolfram").get("appid"))
     ans = wolfram.run(query)
-    return ans
+    return BaseToolOutput(ans)

chatchat-server/chatchat/server/api_server/api_schemas.py

@@ -1,10 +1,10 @@
 from __future__ import annotations
 
-import re
+import json
+import time
 from typing import Dict, List, Literal, Optional, Union
 
 from fastapi import UploadFile
-from chatchat.server.pydantic_v2 import BaseModel, Field, AnyUrl, root_validator
 from openai.types.chat import (
     ChatCompletionMessageParam,
     ChatCompletionToolChoiceOptionParam,
@@ -12,8 +12,10 @@ from openai.types.chat import (
     completion_create_params,
 )
 
-from chatchat.configs import DEFAULT_LLM_MODEL, TEMPERATURE, LLM_MODEL_CONFIG
+from chatchat.configs import DEFAULT_LLM_MODEL, TEMPERATURE
-
+from chatchat.server.callback_handler.agent_callback_handler import AgentStatus
+from chatchat.server.pydantic_v2 import BaseModel, Field, AnyUrl
+from chatchat.server.utils import MsgType
 
 class OpenAIBaseInput(BaseModel):
     user: Optional[str] = None
@@ -21,7 +23,7 @@ class OpenAIBaseInput(BaseModel):
     # The extra values given here take precedence over values defined on the client or passed to this method.
     extra_headers: Optional[Dict] = None
     extra_query: Optional[Dict] = None
-    extra_body: Optional[Dict] = None
+    extra_json: Optional[Dict] = Field(None, alias="extra_body")
     timeout: Optional[float] = None
 
     class Config:
@@ -44,8 +46,8 @@ class OpenAIChatInput(OpenAIBaseInput):
     stop: Union[Optional[str], List[str]] = None
     stream: Optional[bool] = None
     temperature: Optional[float] = TEMPERATURE
-    tool_choice: Optional[ChatCompletionToolChoiceOptionParam] = None
+    tool_choice: Optional[Union[ChatCompletionToolChoiceOptionParam, str]] = None
-    tools: List[ChatCompletionToolParam] = None
+    tools: List[Union[ChatCompletionToolParam, str]] = None
     top_logprobs: Optional[int] = None
     top_p: Optional[float] = None
 
@@ -98,3 +100,62 @@ class OpenAIAudioSpeechInput(OpenAIBaseInput):
     voice: str
     response_format: Optional[Literal["mp3", "opus", "aac", "flac", "pcm", "wav"]] = None
     speed: Optional[float] = None
+
+
+class OpenAIBaseOutput(BaseModel):
+    id: Optional[str] = None
+    content: Optional[str] = None
+    model: Optional[str] = None
+    object: Literal["chat.completion", "chat.completion.chunk"] = "chat.completion.chunk"
+    role: Literal["assistant"] = "assistant"
+    finish_reason: Optional[str] = None
+    created: int = Field(default_factory=lambda : int(time.time()))
+    tool_calls: List[Dict] = []
+
+    status: Optional[int] = None # AgentStatus
+    message_type: int = MsgType.TEXT
+    message_id: Optional[str] = None # id in database table
+    is_ref: bool = False # wheather show in seperated expander
+
+    class Config:
+        extra = "allow"
+
+    def model_dump(self) -> dict:
+        result = {
+            "id": self.id,
+            "object": self.object,
+            "model": self.model,
+            "created": self.created,
+
+            "status": self.status,
+            "message_type": self.message_type,
+            "message_id": self.message_id,
+            "is_ref": self.is_ref,
+            **(self.model_extra or {}),
+        }
+
+        if self.object == "chat.completion.chunk":
+            result["choices"] = [{
+                "delta": {
+                    "content": self.content,
+                    "tool_calls": self.tool_calls,
+                },
+                "role": self.role,
+            }]
+        elif self.object == "chat.completion":
+            result["choices"] = [{
+                "message": {
+                    "role": self.role,
+                    "content": self.content,
+                    "finish_reason": self.finish_reason,
+                    "tool_calls": self.tool_calls,
+                }
+            }]
+        return result
+
+    def model_dump_json(self):
+        return json.dumps(self.model_dump(), ensure_ascii=False)
+
+
+class OpenAIChatOutput(OpenAIBaseOutput):
+    ...

chatchat-server/chatchat/server/api_server/chat_routes.py

@@ -1,12 +1,17 @@
 from __future__ import annotations
 
-from typing import List
+from typing import List, Dict
 
 from fastapi import APIRouter, Request
+from langchain.prompts.prompt import PromptTemplate
 
+from chatchat.server.api_server.api_schemas import OpenAIChatInput, MsgType, AgentStatus
 from chatchat.server.chat.chat import chat
 from chatchat.server.chat.feedback import chat_feedback
 from chatchat.server.chat.file_chat import file_chat
+from chatchat.server.db.repository import add_message_to_db
+from chatchat.server.utils import get_OpenAIClient, get_tool, get_tool_config, get_prompt_template
+from .openai_routes import openai_request
 
 
 chat_router = APIRouter(prefix="/chat", tags=["ChatChat 对话"])
@@ -21,4 +26,115 @@ chat_router.post("/feedback",
 
 chat_router.post("/file_chat",
             summary="文件对话"
-            )(file_chat)
+            )(file_chat)
+
+
+@chat_router.post("/chat/completions", summary="兼容 openai 的统一 chat 接口")
+async def chat_completions(
+    request: Request,
+    body: OpenAIChatInput,
+) -> Dict:
+    '''
+    请求参数与 openai.chat.completions.create 一致，可以通过 extra_body 传入额外参数
+    tools 和 tool_choice 可以直接传工具名称，会根据项目里包含的 tools 进行转换
+    通过不同的参数组合调用不同的 chat 功能：
+    - tool_choice
+        - extra_body 中包含 tool_input: 直接调用 tool_choice(tool_input)
+        - extra_body 中不包含 tool_input: 通过 agent 调用 tool_choice
+    - tools: agent 对话
+    - 其它：LLM 对话
+    以后还要考虑其它的组合（如文件对话）
+    返回与 openai 兼容的 Dict
+    '''
+    client = get_OpenAIClient(model_name=body.model, is_async=True)
+    extra = {**body.model_extra} or {}
+    for key in list(extra):
+        delattr(body, key)
+
+    # check tools & tool_choice in request body
+    if isinstance(body.tool_choice, str):
+        if t := get_tool(body.tool_choice):
+            body.tool_choice = {"function": {"name": t.name}, "type": "function"}
+    if isinstance(body.tools, list):
+        for i in range(len(body.tools)):
+            if isinstance(body.tools[i], str):
+                if t := get_tool(body.tools[i]):
+                    body.tools[i] = {
+                        "type": "function",
+                        "function": {
+                            "name": t.name,
+                            "description": t.description,
+                            "parameters": t.args,
+                        }
+                    }
+
+    conversation_id = extra.get("conversation_id")
+
+    # chat based on result from one choiced tool
+    if body.tool_choice:
+        tool = get_tool(body.tool_choice["function"]["name"])
+        if not body.tools:
+            body.tools = [{
+                        "type": "function",
+                        "function": {
+                            "name": tool.name,
+                            "description": tool.description,
+                            "parameters": tool.args,
+                        }
+                    }]
+        if tool_input := extra.get("tool_input"):
+            message_id = add_message_to_db(
+                chat_type="tool_call",
+                query=body.messages[-1]["content"],
+                conversation_id=conversation_id
+            ) if conversation_id else None
+
+            tool_result = await tool.ainvoke(tool_input)
+            prompt_template = PromptTemplate.from_template(get_prompt_template("llm_model", "rag"))
+            body.messages[-1]["content"] = prompt_template.format(context=tool_result, question=body.messages[-1]["content"])
+            del body.tools
+            del body.tool_choice
+            extra_json = {
+                "message_id": message_id,
+                "status": None,
+            }
+            header = [{**extra_json,
+                       "content": f"知识库参考资料：\n\n{tool_result}\n\n",
+                       "tool_output":tool_result.data,
+                       "is_ref": True,
+                       }]
+            return await openai_request(client.chat.completions.create, body, extra_json=extra_json, header=header)
+
+    # agent chat with tool calls
+    if body.tools:
+        message_id = add_message_to_db(
+            chat_type="agent_chat",
+            query=body.messages[-1]["content"],
+            conversation_id=conversation_id
+        ) if conversation_id else None
+
+        chat_model_config = {} # TODO: 前端支持配置模型
+        tool_names = [x["function"]["name"] for x in body.tools]
+        tool_config = {name: get_tool_config(name) for name in tool_names}
+        result = await chat(query=body.messages[-1]["content"],
+                        metadata=extra.get("metadata", {}),
+                        conversation_id=extra.get("conversation_id", ""),
+                        message_id=message_id,
+                        history_len=-1,
+                        history=body.messages[:-1],
+                        stream=body.stream,
+                        chat_model_config=extra.get("chat_model_config", chat_model_config),
+                        tool_config=extra.get("tool_config", tool_config),
+                        )
+        return result
+    else: # LLM chat directly
+        message_id = add_message_to_db(
+            chat_type="llm_chat",
+            query=body.messages[-1]["content"],
+            conversation_id=conversation_id
+        ) if conversation_id else None
+        extra_json = {
+            "message_id": message_id,
+            "status": None,
+        }
+        return await openai_request(client.chat.completions.create, body, extra_json=extra_json)

chatchat-server/chatchat/server/api_server/openai_routes.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 import asyncio
 from contextlib import asynccontextmanager
-from typing import Dict, Tuple, AsyncGenerator
+from typing import Dict, Tuple, AsyncGenerator, Iterable
 
 from fastapi import APIRouter, Request
 from openai import AsyncClient
@@ -19,7 +19,7 @@ openai_router = APIRouter(prefix="/v1", tags=["OpenAI 兼容平台整合接口"]
 
 
 @asynccontextmanager
-async def acquire_model_client(model_name: str) -> AsyncGenerator[AsyncClient]:
+async def get_model_client(model_name: str) -> AsyncGenerator[AsyncClient]:
     '''
     对重名模型进行调度，依次选择：空闲的模型 -> 当前访问数最少的模型
     '''
@@ -49,19 +49,47 @@ async def acquire_model_client(model_name: str) -> AsyncGenerator[AsyncClient]:
         semaphore.release()
 
 
-async def openai_request(method, body):
+async def openai_request(method, body, extra_json: Dict={}, header: Iterable=[], tail: Iterable=[]):
     '''
-    helper function to make openai request
+    helper function to make openai request with extra fields
     '''
     async def generator():
-        async for chunk in await method(**params):
+        for x in header:
-            yield {"data": chunk.json()}
+            if isinstance(x, str):
+                x = OpenAIChatOutput(content=x, object="chat.completion.chunk")
+            elif isinstance(x, dict):
+                x = OpenAIChatOutput.model_validate(x)
+            else:
+                raise RuntimeError(f"unsupported value: {header}")
+            for k, v in extra_json.items():
+                setattr(x, k, v)
+            yield x.model_dump_json()
+
+        async for chunk in await method(**params):
+            for k, v in extra_json.items():
+                setattr(chunk, k, v)
+            yield chunk.model_dump_json()
+
+        for x in tail:
+            if isinstance(x, str):
+                x = OpenAIChatOutput(content=x, object="chat.completion.chunk")
+            elif isinstance(x, dict):
+                x = OpenAIChatOutput.model_validate(x)
+            else:
+                raise RuntimeError(f"unsupported value: {tail}")
+            for k, v in extra_json.items():
+                setattr(x, k, v)
+            yield x.model_dump_json()
+
+    params = body.model_dump(exclude_unset=True)
 
-    params = body.dict(exclude_unset=True)
     if hasattr(body, "stream") and body.stream:
         return EventSourceResponse(generator())
     else:
-        return (await method(**params)).dict()
+        result = await method(**params)
+        for k, v in extra_json.items():
+            setattr(result, k, v)
+        return result.model_dump()
 
 
 @openai_router.get("/models")
@@ -74,12 +102,12 @@ async def list_models() -> List:
             client = get_OpenAIClient(name, is_async=True)
             models = await client.models.list()
             if config.get("platform_type") == "xinference":
-                models = models.dict(exclude={"data":..., "object":...})
+                models = models.model_dump(exclude={"data":..., "object":...})
                 for x in models:
                     models[x]["platform_name"] = name
                 return [{**v, "id": k} for k, v in models.items()]
             elif config.get("platform_type") == "oneapi":
-                return [{**x.dict(), "platform_name": name} for x in models.data]
+                return [{**x.model_dump(), "platform_name": name} for x in models.data]
         except Exception:
             logger.error(f"failed request to platform: {name}", exc_info=True)
             return {}
@@ -97,12 +125,8 @@ async def create_chat_completions(
     request: Request,
     body: OpenAIChatInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         result = await openai_request(client.chat.completions.create, body)
-        # result["related_docs"] = ["doc1"]
-        # result["choices"][0]["message"]["related_docs"] = ["doc1"]
-        # print(result)
-        # breakpoint()
         return result
 
 
@@ -111,7 +135,7 @@ async def create_completions(
     request: Request,
     body: OpenAIChatInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.completions.create, body)
 
 
@@ -130,7 +154,7 @@ async def create_image_generations(
     request: Request,
     body: OpenAIImageGenerationsInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.images.generate, body)
 
 
@@ -139,7 +163,7 @@ async def create_image_variations(
     request: Request,
     body: OpenAIImageVariationsInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.images.create_variation, body)
 
 
@@ -148,7 +172,7 @@ async def create_image_edit(
     request: Request,
     body: OpenAIImageEditsInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.images.edit, body)
 
     
@@ -157,7 +181,7 @@ async def create_audio_translations(
     request: Request,
     body: OpenAIAudioTranslationsInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.audio.translations.create, body)
 
 
@@ -166,7 +190,7 @@ async def create_audio_transcriptions(
     request: Request,
     body: OpenAIAudioTranscriptionsInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.audio.transcriptions.create, body)
 
 
@@ -175,7 +199,7 @@ async def create_audio_speech(
     request: Request,
     body: OpenAIAudioSpeechInput,
 ):
-    async with acquire_model_client(body.model) as client:
+    async with get_model_client(body.model) as client:
         return await openai_request(client.audio.speech.create, body)
 
 

chatchat-server/chatchat/server/api_server/tool_routes.py

@@ -5,7 +5,7 @@ from typing import List
 from fastapi import APIRouter, Request, Body
 
 from chatchat.configs import logger
-from chatchat.server.utils import BaseResponse, get_tool
+from chatchat.server.utils import BaseResponse, get_tool, get_tool_config
 
 
 tool_router = APIRouter(prefix="/tools", tags=["Toolkits"])
@@ -14,20 +14,24 @@ tool_router = APIRouter(prefix="/tools", tags=["Toolkits"])
 @tool_router.get("/", response_model=BaseResponse)
 async def list_tools():
     tools = get_tool()
-    data = {t.name: {"name": t.name, "description": t.description, "args": t.args} for t in tools}
+    data = {t.name: {
+                "name": t.name,
+                "title": t.title,
+                "description": t.description,
+                "args": t.args,
+                "config": get_tool_config(t.name),
+            } for t in tools.values()}
     return {"data": data}
 
 
 @tool_router.post("/call", response_model=BaseResponse)
 async def call_tool(
     name: str = Body(examples=["calculate"]),
-    kwargs: dict = Body({}, examples=[{"a":1,"b":2,"operator":"+"}]),
+    tool_input: dict = Body({}, examples=[{"text": "3+5/2"}]),
 ):
-    tools = get_tool()
+    if tool := get_tool(name):
-
-    if tool := tools.get(name):
         try:
-            result = await tool.ainvoke(kwargs)
+            result = await tool.ainvoke(tool_input)
             return {"data": result}
         except Exception:
             msg = f"failed to call tool '{name}'"

chatchat-server/chatchat/server/callback_handler/agent_callback_handler.py

@@ -102,7 +102,13 @@ class AgentExecutorAsyncIteratorCallbackHandler(AsyncIteratorCallbackHandler):
             metadata: Optional[Dict[str, Any]] = None,
             **kwargs: Any,
     ) -> None:
-       pass
+       data = {
+           "run_id": str(run_id),
+           "status": AgentStatus.tool_start,
+           "tool": serialized["name"],
+           "tool_input": input_str,
+       }
+       self.queue.put_nowait(dumps(data))
 
 
     async def on_tool_end(
@@ -116,6 +122,7 @@ class AgentExecutorAsyncIteratorCallbackHandler(AsyncIteratorCallbackHandler):
     ) -> None:
         """Run when tool ends running."""
         data = {
+            "run_id": str(run_id),
             "status": AgentStatus.tool_end,
             "tool_output": output,
         }
@@ -133,8 +140,10 @@ class AgentExecutorAsyncIteratorCallbackHandler(AsyncIteratorCallbackHandler):
     ) -> None:
         """Run when tool errors."""
         data = {
+            "run_id": str(run_id),
             "status": AgentStatus.tool_end,
-            "text": str(error),
+            "tool_output": str(error),
+            "is_error": True,
         }
         # self.done.clear()
         self.queue.put_nowait(dumps(data))

chatchat-server/chatchat/server/chat/chat.py

@@ -1,6 +1,8 @@
 import asyncio
 import json
+import time
 from typing import AsyncIterable, List
+import uuid
 
 from fastapi import Body
 from sse_starlette.sse import EventSourceResponse
@@ -10,19 +12,19 @@ from langchain_core.messages import AIMessage, HumanMessage
 from langchain.chains import LLMChain
 from langchain.prompts.chat import ChatPromptTemplate
 from langchain.prompts import PromptTemplate
+
+from chatchat.configs.model_config import LLM_MODEL_CONFIG
 from chatchat.server.agent.agent_factory.agents_registry import agents_registry
 from chatchat.server.agent.container import container
-
+from chatchat.server.api_server.api_schemas import OpenAIChatOutput
 from chatchat.server.utils import wrap_done, get_ChatOpenAI, get_prompt_template, MsgType, get_tool
 from chatchat.server.chat.utils import History
 from chatchat.server.memory.conversation_db_buffer_memory import ConversationBufferDBMemory
-from chatchat.server.db.repository import add_message_to_db
 from chatchat.server.callback_handler.agent_callback_handler import AgentExecutorAsyncIteratorCallbackHandler, AgentStatus
 
 
 def create_models_from_config(configs, callbacks, stream):
-    if configs is None:
+    configs = configs or LLM_MODEL_CONFIG
-        configs = {}
     models = {}
     prompts = {}
     for model_type, model_configs in configs.items():
@@ -99,6 +101,7 @@ def create_models_chains(history, history_len, prompts, models, tools, callbacks
 async def chat(query: str = Body(..., description="用户输入", examples=["恼羞成怒"]),
                metadata: dict = Body({}, description="附件，可能是图像或者其他功能", examples=[]),
                conversation_id: str = Body("", description="对话框ID"),
+               message_id: str = Body(None, description="数据库消息ID"),
                history_len: int = Body(-1, description="从数据库中取历史消息的数量"),
                history: List[History] = Body(
                    [],
@@ -115,13 +118,9 @@ async def chat(query: str = Body(..., description="用户输入", examples=["恼
                chat_model_config: dict = Body({}, description="LLM 模型配置", examples=[]),
                tool_config: dict = Body({}, description="工具配置", examples=[]),
                ):
-    async def chat_iterator() -> AsyncIterable[str]:
+    '''Agent 对话'''
-        message_id = add_message_to_db(
-            chat_type="llm_chat",
-            query=query,
-            conversation_id=conversation_id
-        ) if conversation_id else None
 
+    async def chat_iterator() -> AsyncIterable[OpenAIChatOutput]:
         callback = AgentExecutorAsyncIteratorCallbackHandler()
         callbacks = [callback]
         models, prompts = create_models_from_config(callbacks=callbacks, configs=chat_model_config,
@@ -145,9 +144,32 @@ async def chat(query: str = Body(..., description="用户输入", examples=["恼
                 }
             ), callback.done))
 
+        last_tool = {}
         async for chunk in callback.aiter():
             data = json.loads(chunk)
-            if data["status"] == AgentStatus.tool_end:
+            data["tool_calls"] = []
+            data["message_type"] = MsgType.TEXT
+
+            if data["status"] == AgentStatus.tool_start:
+                last_tool = {
+                    "index": 0,
+                    "id": data["run_id"],
+                    "type": "function",
+                    "function": {
+                        "name": data["tool"],
+                        "arguments": data["tool_input"],
+                    },
+                    "tool_output": None,
+                    "is_error": False,
+                }
+                data["tool_calls"].append(last_tool)
+            if data["status"] in [AgentStatus.tool_end]:
+                last_tool.update(
+                    tool_output=data["tool_output"],
+                    is_error=data.get("is_error", False)
+                )
+                data["tool_calls"] = [last_tool]
+                last_tool = {}
                 try:
                     tool_output = json.loads(data["tool_output"])
                     if message_type := tool_output.get("message_type"):
@@ -161,10 +183,43 @@ async def chat(query: str = Body(..., description="用户输入", examples=["恼
                         data["message_type"] = message_type
                 except:
                     ...
-            data.setdefault("message_type", MsgType.TEXT)
-            data["message_id"] = message_id
-            yield json.dumps(data, ensure_ascii=False)
 
+            ret = OpenAIChatOutput(
+                id=f"chat{uuid.uuid4()}",
+                object="chat.completion.chunk",
+                content=data.get("text", ""),
+                role="assistant",
+                tool_calls=data["tool_calls"],
+                model=models["llm_model"].model_name,
+                status = data["status"],
+                message_type = data["message_type"],
+                message_id=message_id,
+            )
+            yield ret.model_dump_json()
         await task
 
-    return EventSourceResponse(chat_iterator())
+    if stream:
+        return EventSourceResponse(chat_iterator())
+    else:
+        ret = OpenAIChatOutput(
+            id=f"chat{uuid.uuid4()}",
+            object="chat.completion",
+            content="",
+            role="assistant",
+            finish_reason="stop",
+            tool_calls=[],
+            status = AgentStatus.agent_finish,
+            message_type = MsgType.TEXT,
+            message_id=message_id,
+        )
+
+        async for chunk in chat_iterator():
+            data = json.loads(chunk)
+            if text := data["choices"][0]["delta"]["content"]:
+                ret.content += text
+            if data["status"] == AgentStatus.tool_end:
+                ret.tool_calls += data["choices"][0]["delta"]["tool_calls"]
+            ret.model = data["model"]
+            ret.created = data["created"]
+
+        return ret.model_dump()

chatchat-server/chatchat/server/knowledge_base/kb_service/es_kb_service.py

@@ -96,11 +96,7 @@ class ESKBService(KBService):
         return os.path.join(ESKBService.get_kb_path(knowledge_base_name), "vector_store")
 
     def do_create_kb(self):
-        if os.path.exists(self.doc_path):
+        ...
-            if not os.path.exists(os.path.join(self.kb_path, "vector_store")):
-                os.makedirs(os.path.join(self.kb_path, "vector_store"))
-            else:
-                logger.warning("directory `vector_store` already exists.")
 
     def vs_type(self) -> str:
         return SupportedVSType.ES

chatchat-server/chatchat/server/knowledge_base/kb_service/faiss_kb_service.py

@@ -65,7 +65,6 @@ class FaissKBService(KBService):
         embed_func = get_Embeddings(self.embed_model)
         embeddings = embed_func.embed_query(query)
         with self.load_vector_store().acquire() as vs:
-            embeddings = vs.embeddings.embed_query(query)
             docs = vs.similarity_search_with_score_by_vector(embeddings, k=top_k, score_threshold=score_threshold)
         return docs
 

chatchat-server/chatchat/webui_pages/dialogue/dialogue.py

@@ -1,24 +1,25 @@
 import base64
-
+import uuid
-from chatchat.server.utils import get_tool_config
-import streamlit as st
-from streamlit_antd_components.utils import ParseItems
-
-from chatchat.webui_pages.dialogue.utils import process_files
-# from chatchat.webui_pages.loom_view_client import build_providers_model_plugins_name, find_menu_items_by_index, set_llm_select, \
-#     get_select_model_endpoint
-from chatchat.webui_pages.utils import *
-from streamlit_chatbox import *
-from streamlit_modal import Modal
-from datetime import datetime
 import os
 import re
 import time
+from typing import List, Dict
+
+import streamlit as st
+from streamlit_antd_components.utils import ParseItems
+
+import openai
+from streamlit_chatbox import *
+from streamlit_modal import Modal
+from datetime import datetime
+
 from chatchat.configs import (LLM_MODEL_CONFIG, SUPPORT_AGENT_MODELS, MODEL_PLATFORMS)
 from chatchat.server.callback_handler.agent_callback_handler import AgentStatus
 from chatchat.server.utils import MsgType, get_config_models
-import uuid
+from chatchat.server.utils import get_tool_config
-from typing import List, Dict
+from chatchat.webui_pages.utils import *
+from chatchat.webui_pages.dialogue.utils import process_files
+
 
 img_dir = (Path(__file__).absolute().parent.parent.parent)
 
@@ -121,69 +122,66 @@ def dialogue_page(api: ApiRequest, is_lite: bool = False):
             st.write("\n\n".join(cmds))
 
     with st.sidebar:
-        conv_names = list(st.session_state["conversation_ids"].keys())
+        tab1, tab2 = st.tabs(["对话设置", "模型设置"])
-        index = 0
 
-        if st.session_state.get("cur_conv_name") in conv_names:
+        with tab1:
-            index = conv_names.index(st.session_state.get("cur_conv_name"))
+            use_agent = st.checkbox("启用Agent", True, help="请确保选择的模型具备Agent能力")
-        conversation_name = st.selectbox("当前会话", conv_names, index=index)
+            # 选择工具
-        chat_box.use_chat_name(conversation_name)
+            tools = api.list_tools()
-        conversation_id = st.session_state["conversation_ids"][conversation_name]
+            if use_agent:
+                selected_tools = st.multiselect("选择工具", list(tools), format_func=lambda x: tools[x]["title"])
+            else:
+                selected_tool = st.selectbox("选择工具", list(tools), format_func=lambda x: tools[x]["title"])
+                selected_tools = [selected_tool]
+            selected_tool_configs = {name: tool["config"] for name, tool in tools.items() if name in selected_tools}
 
-        platforms = ["所有"] + [x["platform_name"] for x in MODEL_PLATFORMS]
+            # 当不启用Agent时，手动生成工具参数
-        platform = st.selectbox("选择模型平台", platforms)
+            # TODO: 需要更精细的控制控件
-        llm_models = list(get_config_models(model_type="llm", platform_name=None if platform=="所有" else platform))
+            tool_input = {}
-        llm_model = st.selectbox("选择LLM模型", llm_models)
+            if not use_agent and len(selected_tools) == 1:
+                with st.expander("工具参数", True):
+                    for k, v in tools[selected_tools[0]]["args"].items():
+                        if choices := v.get("choices", v.get("enum")):
+                            tool_input[k] = st.selectbox(v["title"], choices)
+                        else:
+                            if v["type"] == "integer":
+                                tool_input[k] = st.slider(v["title"], value=v.get("default"))
+                            elif v["type"] == "number":
+                                tool_input[k] = st.slider(v["title"], value=v.get("default"), step=0.1)
+                            else:
+                                tool_input[k] = st.text_input(v["title"], v.get("default"))
 
-        #  传入后端的内容
-        chat_model_config = {key: {} for key in LLM_MODEL_CONFIG.keys()}
-        tool_use = True
-        for key in LLM_MODEL_CONFIG:
-            if key == 'llm_model':
-                continue
-            if key == 'action_model':
-                first_key = next(iter(LLM_MODEL_CONFIG[key]))
-                if first_key not in SUPPORT_AGENT_MODELS:
-                    st.warning("不支持Agent的模型，无法执行任何工具调用")
-                    tool_use = False
-                    continue
-            if LLM_MODEL_CONFIG[key]:
-                first_key = next(iter(LLM_MODEL_CONFIG[key]))
-                chat_model_config[key][first_key] = LLM_MODEL_CONFIG[key][first_key]
 
-        # 选择工具
+            uploaded_file = st.file_uploader("上传附件", accept_multiple_files=False)
-        selected_tool_configs = {}
+            files_upload = process_files(files=[uploaded_file]) if uploaded_file else None
-        if tool_use:
+        
-            from chatchat.configs import model_config as model_config_py
+        with tab2:
-            import importlib
+            # 会话
-            importlib.reload(model_config_py)
+            conv_names = list(st.session_state["conversation_ids"].keys())
+            index = 0
+            if st.session_state.get("cur_conv_name") in conv_names:
+                index = conv_names.index(st.session_state.get("cur_conv_name"))
+            conversation_name = st.selectbox("当前会话", conv_names, index=index)
+            chat_box.use_chat_name(conversation_name)
+            conversation_id = st.session_state["conversation_ids"][conversation_name]
 
-            tools = get_tool_config()
+            # 模型
-            with st.expander("工具栏"):
+            platforms = ["所有"] + [x["platform_name"] for x in MODEL_PLATFORMS]
-                for tool in tools:
+            platform = st.selectbox("选择模型平台", platforms)
-                    is_selected = st.checkbox(tool, value=tools[tool]["use"], key=tool)
+            llm_models = list(get_config_models(model_type="llm", platform_name=None if platform=="所有" else platform))
-                    if is_selected:
+            llm_model = st.selectbox("选择LLM模型", llm_models)
-                        selected_tool_configs[tool] = tools[tool]
 
-        if llm_model is not None:
+            #  传入后端的内容
-            chat_model_config['llm_model'][llm_model] = LLM_MODEL_CONFIG['llm_model'].get(llm_model, {})
+            chat_model_config = {key: {} for key in LLM_MODEL_CONFIG.keys()}
+            for key in LLM_MODEL_CONFIG:
+                if LLM_MODEL_CONFIG[key]:
+                    first_key = next(iter(LLM_MODEL_CONFIG[key]))
+                    chat_model_config[key][first_key] = LLM_MODEL_CONFIG[key][first_key]
 
-        uploaded_file = st.file_uploader("上传附件", accept_multiple_files=False)
+            if llm_model is not None:
-        files_upload = process_files(files=[uploaded_file]) if uploaded_file else None
+                chat_model_config['llm_model'][llm_model] = LLM_MODEL_CONFIG['llm_model'].get(llm_model, {})
-        # print(len(files_upload["audios"])) if files_upload else None
 
-        # if dialogue_mode == "文件对话":
-        #     with st.expander("文件对话配置", True):
-        #         files = st.file_uploader("上传知识文件：",
-        #                                  [i for ls in LOADER_DICT.values() for i in ls],
-        #                                  accept_multiple_files=True,
-        #                                  )
-        #         kb_top_k = st.number_input("匹配知识条数：", 1, 20, VECTOR_SEARCH_TOP_K)
-        #         score_threshold = st.slider("知识匹配分数阈值：", 0.0, 2.0, float(SCORE_THRESHOLD), 0.01)
-        #         if st.button("开始上传", disabled=len(files) == 0):
-        #             st.session_state["file_chat_id"] = upload_temp_docs(files, api)
     # Display chat messages from history on app rerun
-
     chat_box.output_messages()
     chat_input_placeholder = "请输入对话内容，换行请使用Shift+Enter。输入/help查看自定义命令 "
 
@@ -228,50 +226,85 @@ def dialogue_page(api: ApiRequest, is_lite: bool = False):
 
             chat_box.ai_say("正在思考...")
             text = ""
-            text_action = ""
+            started = False
-            element_index = 0
 
-            for d in api.chat_chat(query=prompt,
+            client = openai.Client(base_url=f"{api_address()}/chat", api_key="NONE")
-                                   metadata=files_upload,
+            messages = history + [{"role": "user", "content": prompt}]
-                                   history=history,
+            tools = list(selected_tool_configs)
-                                   chat_model_config=chat_model_config,
+            if len(selected_tools) == 1:
-                                   conversation_id=conversation_id,
+                tool_choice = selected_tools[0]
-                                   tool_config=selected_tool_configs,
+            else:
-                                   ):
+                tool_choice = None
-                message_id = d.get("message_id", "")
+            # 如果 tool_input 中有空的字段，设为用户输入
+            for k in tool_input:
+                if tool_input[k] in [None, ""]:
+                    tool_input[k] = prompt
+
+            extra_body = dict(
+                            metadata=files_upload,
+                            chat_model_config=chat_model_config,
+                            conversation_id=conversation_id,
+                            tool_input = tool_input,
+                            )
+            for d in client.chat.completions.create(
+                    messages=messages,
+                    model=llm_model,
+                    stream=True,
+                    tools=tools,
+                    tool_choice=tool_choice,
+                    extra_body=extra_body,
+                ):
+                print("\n\n", d.status, "\n", d, "\n\n")
+                message_id = d.message_id
                 metadata = {
                     "message_id": message_id,
                 }
-                print(d)
+
-                if d["status"] == AgentStatus.error:
+                if d.status == AgentStatus.error:
-                    st.error(d["text"])
+                    st.error(d.choices[0].delta.content)
-                elif d["status"] == AgentStatus.agent_action:
+                elif d.status == AgentStatus.llm_start:
-                    formatted_data = {
+                    if not started:
-                        "Function": d["tool_name"],
+                        started = True
-                        "function_input": d["tool_input"]
-                    }
-                    element_index += 1
-                    formatted_json = json.dumps(formatted_data, indent=2, ensure_ascii=False)
-                    chat_box.insert_msg(
-                        Markdown(title="Function call", in_expander=True, expanded=True, state="running"))
-                    text = """\n```{}\n```\n""".format(formatted_json)
-                    chat_box.update_msg(Markdown(text), element_index=element_index)
-                elif d["status"] == AgentStatus.tool_end:
-                    text += """\n```\nObservation:\n{}\n```\n""".format(d["tool_output"])
-                    chat_box.update_msg(Markdown(text), element_index=element_index, expanded=False, state="complete")
-                elif d["status"] == AgentStatus.llm_new_token:
-                    text += d["text"]
-                    chat_box.update_msg(text, streaming=True, element_index=element_index, metadata=metadata)
-                elif d["status"] == AgentStatus.llm_end:
-                    chat_box.update_msg(text, streaming=False, element_index=element_index, metadata=metadata)
-                elif d["status"] == AgentStatus.agent_finish:
-                    if d["message_type"] == MsgType.IMAGE:
-                        for url in json.loads(d["text"]).get("images", []):
-                            url = f"{api.base_url}/media/{url}"
-                            chat_box.insert_msg(Image(url))
-                        chat_box.update_msg(element_index=element_index, expanded=False, state="complete")
                     else:
-                        chat_box.insert_msg(Markdown(d["text"], expanded=True))
+                        chat_box.insert_msg("正在解读工具输出结果...")
+                    text = d.choices[0].delta.content or ""
+                elif d.status == AgentStatus.llm_new_token:
+                    text += d.choices[0].delta.content or ""
+                    chat_box.update_msg(text.replace("\n", "\n\n"), streaming=True, metadata=metadata)
+                elif d.status == AgentStatus.llm_end:
+                    text += d.choices[0].delta.content or ""
+                    chat_box.update_msg(text.replace("\n", "\n\n"), streaming=False, metadata=metadata)
+                # tool 的输出与 llm 输出重复了
+                # elif d.status == AgentStatus.tool_start:
+                #     formatted_data = {
+                #         "Function": d.choices[0].delta.tool_calls[0].function.name,
+                #         "function_input": d.choices[0].delta.tool_calls[0].function.arguments,
+                #     }
+                #     formatted_json = json.dumps(formatted_data, indent=2, ensure_ascii=False)
+                #     text = """\n```{}\n```\n""".format(formatted_json)
+                #     chat_box.insert_msg( # TODO: insert text directly not shown
+                #         Markdown(text, title="Function call", in_expander=True, expanded=True, state="running"))
+                # elif d.status == AgentStatus.tool_end:
+                #     tool_output = d.choices[0].delta.tool_calls[0].tool_output
+                #     if d.message_type == MsgType.IMAGE:
+                #         for url in json.loads(tool_output).get("images", []):
+                #             url = f"{api.base_url}/media/{url}"
+                #             chat_box.insert_msg(Image(url))
+                #         chat_box.update_msg(expanded=False, state="complete")
+                #     else:
+                #         text += """\n```\nObservation:\n{}\n```\n""".format(tool_output)
+                #         chat_box.update_msg(text, streaming=False, expanded=False, state="complete")
+                elif d.status == AgentStatus.agent_finish:
+                    text = d.choices[0].delta.content or ""
+                    chat_box.update_msg(text.replace("\n", "\n\n"))
+                elif d.status == None: # not agent chat
+                    if getattr(d, "is_ref", False):
+                        chat_box.insert_msg(Markdown(d.choices[0].delta.content or "", in_expander=True, state="complete", title="参考资料"))
+                        chat_box.insert_msg("")
+                    else:
+                        text += d.choices[0].delta.content or ""
+                        chat_box.update_msg(text.replace("\n", "\n\n"), streaming=True, metadata=metadata)
+            chat_box.update_msg(text, streaming=False, metadata=metadata)
 
             if os.path.exists("tmp/image.jpg"):
                 with open("tmp/image.jpg", "rb") as image_file:
@@ -313,8 +346,7 @@ def dialogue_page(api: ApiRequest, is_lite: bool = False):
         st.rerun()
 
     now = datetime.now()
-    with st.sidebar:
+    with tab1:
-
         cols = st.columns(2)
         export_btn = cols[0]
         if cols[1].button(
@@ -333,3 +365,5 @@ def dialogue_page(api: ApiRequest, is_lite: bool = False):
         mime="text/markdown",
         use_container_width=True,
     )
+
+    # st.write(chat_box.history)

chatchat-server/chatchat/webui_pages/utils.py

@@ -50,6 +50,14 @@ class ApiRequest:
                                             timeout=self.timeout)
         return self._client
 
+    def _check_url(self, url: str) -> str:
+        '''
+        新版 httpx 强制要求 url 以 / 结尾，否则会返回 307
+        '''
+        if not url.endswith("/"):
+            url = url + "/"
+        return url
+
     def get(
             self,
             url: str,
@@ -58,6 +66,7 @@ class ApiRequest:
             stream: bool = False,
             **kwargs: Any,
     ) -> Union[httpx.Response, Iterator[httpx.Response], None]:
+        url = self._check_url(url)
         while retry > 0:
             try:
                 if stream:
@@ -79,6 +88,7 @@ class ApiRequest:
             stream: bool = False,
             **kwargs: Any
     ) -> Union[httpx.Response, Iterator[httpx.Response], None]:
+        url = self._check_url(url)
         while retry > 0:
             try:
                 # print(kwargs)
@@ -101,6 +111,7 @@ class ApiRequest:
             stream: bool = False,
             **kwargs: Any
     ) -> Union[httpx.Response, Iterator[httpx.Response], None]:
+        url = self._check_url(url)
         while retry > 0:
             try:
                 if stream:
@@ -638,6 +649,27 @@ class ApiRequest:
         resp = self.post("/chat/feedback", json=data)
         return self._get_response_value(resp)
 
+    def list_tools(self) -> Dict:
+        '''
+        列出所有工具
+        '''
+        resp = self.get("/tools")
+        return self._get_response_value(resp, as_json=True, value_func=lambda r: r.get("data", {}))
+
+    def call_tool(
+            self,
+            name: str,
+            tool_input: Dict = {},
+    ):
+        '''
+        调用工具
+        '''
+        data = {
+            "name": name,
+            "tool_input": tool_input,
+        }
+        resp = self.post("/tools/call", json=data)
+        return self._get_response_value(resp, as_json=True, value_func=lambda r: r.get("data"))
 
 class AsyncApiRequest(ApiRequest):
     def __init__(self, base_url: str = api_address(), timeout: float = HTTPX_DEFAULT_TIMEOUT):