修改了单元测试的问题生成代码

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "webapp"]
+	path = webapp
+	url = https://git.97id.com/ly/webapp.git

backend/.env.example

@@ -1,7 +1,13 @@
+JIEBA_DATA=./nltk_data
+NLTK_DATA=./nltk_data
+SQLITE_DATABASE_URL=sqlite:///./source.db
+DATA_SOURCE_CACHE=./restapi
+
 # The Llama Cloud API key.
 # LLAMA_CLOUD_API_KEY=
 SQL_DATABASE_URL=mysql+pymysql://zjinfo1:Dy2Bcr53Hm5xRkba@110.42.234.166:3306/zjinfo1
 #SQL_DATABASE_URL=mysql+pymysql://zjinfo2:GSKcziSdBixDXwcd@110.42.234.166:3306/zjinfo2
+SQLITE_DATABASE_URL=sqlite:///./source.db
 
 DASHSCOPE_API_KEY=sk-02c8540e86d84b7ca0e6f4f51bac6e60
 # The provider for the AI models to use.
@@ -49,6 +55,7 @@ VECTOR_STORE_COLLECTION=default
 # Specify this if you are using a local vector database.
 # Otherwise, use VECTOR_STORE__HOST and VECTOR_STORE__PORT config above
 VECTOR_STORE_PATH=./storage_vector
+BM_RETRIEVER_PATH =./storage_bm
 
 
 
@@ -78,3 +85,4 @@ SYSTEM_PROMPT="You are a weather forecast agent. You help users to get the weath
 - You can install any pip package (if it exists) by running a cell with pip install.
 "
 
+PROJECT_TITLE = "您好，我是博微工程理解小助手，您可以问我有关[线路工程]工程数据的相关问题！"

backend/.env.xinference

@@ -1,8 +1,29 @@
+JIEBA_DATA=./nltk_data
+NLTK_DATA=./nltk_data
+SQLITE_DATABASE_URL=sqlite:///./source.db
+DATA_SOURCE_CACHE=./restapi
+
 # The Llama Cloud API key.
 # LLAMA_CLOUD_API_KEY=
 SQL_DATABASE_URL=mysql+pymysql://zjinfo1:Dy2Bcr53Hm5xRkba@110.42.234.166:3306/zjinfo1
 #SQL_DATABASE_URL=mysql+pymysql://zjinfo2:GSKcziSdBixDXwcd@110.42.234.166:3306/zjinfo2
+SQLITE_DATABASE_URL=sqlite:///./source.db
 
+# The number of similar embeddings to return when retrieving documents.
+TOP_K=10
+#--------------------------
+# 是否启用混合检索
+HYBRID_ENABLED = true
+# 混合检索阈值
+HYBRID_ALPHA = 0.6
+#--------------------------
+# 是否启用检索重排功能
+RERANK_ENABLED=true
+# Rerank model
+RERANK_MODEL=bge-reranker-v2-m3
+RERANK_BASE_URL=http://10.1.16.39:9995
+RERANK_TOP_N=5
+RERANK_THRESHOLD=0.3
 #----------   Xinference    ----------------
 # The provider for the AI models to use.
 MODEL_PROVIDER=xinference
@@ -19,9 +40,7 @@ EMBEDDING_MODEL=bge-m3
 EMBEDDING_BASE_URL=http://10.1.16.39:9995
 # Dimension of the embedding model to use.
 EMBEDDING_DIM=1024
-# Rerank model
+
-RERANK_MODEL=bge-reranker-v2-m3
-RERANK_BASE_URL=http://10.1.16.39:9995
 ##----------   OpenAI    ----------------
 ## The provider for the AI models to use.
 #MODEL_PROVIDER=openai
@@ -46,17 +65,10 @@ RERANK_BASE_URL=http://10.1.16.39:9995
 ## Name of the embedding model to use.
 #EMBEDDING_MODEL=text-embedding-v2
 
-#--------------------------
-# 是否启用检索重排功能
-ENABLE_RERANK=true
-
 
 # The questions to help users get started (multi-line).
 CONVERSATION_STARTERS=本工程指什么？\n总算表有哪些费用？\n项目划分哪些内容构成？\n其他费用表有哪些内容？
 
-# The number of similar embeddings to return when retrieving documents.
-TOP_K=5
-
 # The time in milliseconds to wait for the stream to return a response.
 STREAM_TIMEOUT=60000
 
@@ -75,7 +87,7 @@ VECTOR_STORE_COLLECTION=default
 # Specify this if you are using a local vector database.
 # Otherwise, use VECTOR_STORE__HOST and VECTOR_STORE__PORT config above
 VECTOR_STORE_PATH=./storage_vector
-
+BM_RETRIEVER_PATH =./storage_bm
 
 
 PHOENIX_API_KEY=123456
@@ -104,3 +116,4 @@ SYSTEM_PROMPT="You are a weather forecast agent. You help users to get the weath
 - You can install any pip package (if it exists) by running a cell with pip install.
 "
 
+PROJECT_TITLE = "您好，我是博微工程理解小助手，您可以问我有关[线路工程]工程数据的相关问题！"

backend/app/api/routers/app.py

@@ -0,0 +1,490 @@
+import asyncio
+import json
+import logging
+import time
+from typing import Dict, List, Any, Optional, AsyncGenerator
+from collections import deque
+
+from aiostream import stream
+from fastapi import APIRouter, Request
+from fastapi.responses import StreamingResponse
+from llama_index.core import BaseCallbackHandler
+from llama_index.core.base.llms.types import ChatMessage
+from llama_index.core.callbacks import CBEventType
+from llama_index.core.chat_engine.types import StreamingAgentChatResponse
+from llama_index.core.tools import ToolOutput
+from pydantic import BaseModel
+from app.api.routers.request.base import userMng, conversations,message,parameter,feedback
+from app.api.routers.request.baseConfig import *
+from app.api.routers.request.models import ChatRequestData,ChatFileUploadRequest
+from app.engine import get_chat_engine
+import uuid
+
+logger = logging.getLogger("uvicorn")
+
+api_router = r = APIRouter()
+v1_router = v = APIRouter()
+
+class ChatCallbackEvent(BaseModel):
+    event_type: ChatEventType
+    payload: Optional[Dict[str, Any]] = None
+
+    def get_common_param(self)-> dict:
+        return {
+            'event': self.event_type.name,
+            'conversation_id':self.payload.get("conversation_id"),
+            'message_id': self.payload.get("message_id"), 
+            'created_at':  int(time.time()),
+            'task_id': self.payload.get("task_id")
+        }
+
+    def get_WorkflowStart_param(self) -> dict:
+        params = self.get_common_param()
+        params.update({
+            'workflow_run_id':self.payload.get('workflow_run_id'),
+            'data':{
+                "id": self.payload.get('workflow_run_id'),
+                "workflow_id": self.payload.get('workflow_id'),
+                "sequence_number": 1709,
+                "inputs": {
+                    "sys.query": self.payload.get('query'),
+                    "sys.files": [],
+                    "sys.conversation_id": self.payload.get('conversation_id'),
+                    "sys.user_id": self.payload.get('use_id')
+                },
+                "created_at": int(time.time())
+            }
+        })
+        return params
+
+    def get_WorkflowFinished_param(self) -> dict:
+        params = self.get_common_param()
+        params.update({
+            'workflow_run_id':self.payload.get('workflow_run_id'),
+            'data':{
+                "id": self.payload.get('workflow_run_id'),
+                "workflow_id": self.payload.get('workflow_id'),
+                "sequence_number": 1709,
+                "status": "succeeded",
+                "outputs": {
+                    "answer": self.payload.get('response')
+                },
+                "error": '',
+                "elapsed_time": 36.03764106379822,
+                "total_tokens": 11707,
+                "total_steps": 10,
+                "created_by": {
+                "id": str(uuid.uuid4()),
+                "user": self.payload.get('use_id')
+                },
+                "created_at": int(time.time()),
+                "finished_at": int(time.time()),
+                "files": []
+            }
+        })
+        return params
+    
+    def get_NodeStart_param(self) -> dict:
+        params = self.get_common_param()
+        params.update({
+            'workflow_run_id':self.payload.get('workflow_run_id'),
+            'data':{
+                "id": self.payload.get('nodeid'),
+                "node_id": self.payload.get('nodeid'),
+                "node_type": "http-request",
+                "title": self.payload.get('title'),
+                "index": self.payload.get('index'),
+                "predecessor_node_id": self.payload.get('predecessor_node_id'),
+                "inputs": '',
+                "created_at": 1724398751,
+                "extras": {}
+            }
+        })
+        return params     
+
+    def get_NodeFinished_param(self) -> dict:
+        params = self.get_common_param()
+        params.update({
+            'workflow_run_id':self.payload.get('workflow_run_id'),
+            'data':{
+                "id": self.payload.get('nodeid'),
+                "node_id": self.payload.get('nodeid'),
+                "node_type": "http-request",
+                "title": self.payload.get('title'),
+                "index": self.payload.get('index'),
+                "predecessor_node_id": self.payload.get('predecessor_node_id'),
+                "inputs": '',
+                "process_data": '',
+                "outputs": '',
+                "status": "succeeded",
+                "error": '',
+                "elapsed_time": 0.10402441816404462,
+                "execution_metadata": '',
+                "created_at": 1724398751,
+                "finished_at": 1724398751,
+                "files": []
+            }
+        })
+        return params  
+
+    def get_Message_param(self) -> dict:
+        params = self.get_common_param()
+        params.update({
+            'id':self.payload.get('message_id'),
+            'answer':self.payload.get('answer')
+        })
+        return params 
+    
+    def get_MessageEnd_param(self) -> dict:
+        params = self.get_common_param()
+        params.update({
+            'id':self.payload.get('message_id'),
+            'metadata':self.payload.get('metadata')
+        })
+        return params 
+
+    def to_response(self)-> dict|None:
+        try:
+            match self.event_type:
+                case "workflow_started":
+                    return self.get_WorkflowStart_param()
+                case "workflow_finished":
+                    return self.get_WorkflowFinished_param()
+                case "node_started":
+                    return self.get_NodeStart_param()
+                case 'node_finished':
+                    return self.get_NodeFinished_param()
+                case 'message':
+                    return self.get_Message_param()
+                case 'message_end':
+                    return self.get_MessageEnd_param()
+                case _:
+                    return None
+        except Exception as e:
+            logger.error(f"转换回应时间时发生错误，原因: {e}")
+            return None
+
+class ChatEventCallbackHandler(BaseCallbackHandler):
+    _aqueue: asyncio.Queue
+    is_done: bool = False
+
+    def __init__(self,**params):
+        """Initialize the base callback handler."""
+        ignored_events = [
+            # CBEventType.CHUNKING,
+            # CBEventType.NODE_PARSING,
+            # CBEventType.EMBEDDING,
+            # CBEventType.LLM,
+            # CBEventType.TEMPLATING,
+        ]
+        super().__init__(ignored_events, ignored_events)
+        self._aqueue = asyncio.Queue()
+        self._response:str = ''
+        self._params:Dict[str,Any] = params
+        self._nodeStack:deque = deque()
+
+        #添加工作流开始事件     
+        data:ChatRequestData = self._params['data']
+        args:Dict[str,Any] = self._params['ids']
+        args.update(
+            {
+                'use_id': data.user,
+                'query': data.query,
+                'conversation_id': data.conversation_id  
+            }
+        )
+        wf_event = ChatCallbackEvent(event_type = ChatEventType.WORKFLOW_START,payload = args)
+        if wf_event.to_response() is not None:
+            self._aqueue.put_nowait(wf_event)
+
+    def on_event_start(
+            self,
+            event_type: CBEventType,
+            payload: Optional[Dict[str, Any]] = None,
+            event_id: str = "",
+            **kwargs: Any,
+    ) -> str:
+        logger.info("event_start:{} type:{} payload:{}\n".format(event_id, event_type, payload))
+
+        self._nodeStack.append(event_id)      
+        nindex = self._nodeStack.count() - 1
+        args:Dict[str,Any] = self._params['ids']
+        args.update(
+            {
+                'nodeid':event_id,
+                'title':event_type.name,
+                'index':nindex + 1,
+                'predecessor_node_id': self._nodeStack[nindex - 1] if nindex > 0 else ''
+            }
+        )
+        nd_event = ChatCallbackEvent(event_type = ChatEventType.NODE_START,payload = args)
+        if nd_event.to_response() is not None:
+            self._aqueue.put_nowait(nd_event)
+
+
+    def on_event_end(
+            self,
+            event_type: CBEventType,
+            payload: Optional[Dict[str, Any]] = None,
+            event_id: str = "",
+            **kwargs: Any,
+    ) -> None:
+        logger.info("event_end:{} type:{} payload:{}\n".format(event_id, event_type, payload))
+        
+        #self.response = payload.get("response","")
+        args:Dict[str,Any] = self._params['ids']
+        nodeID = self._nodeStack[-1]
+        if nodeID == event_id:
+            nindex = self._nodeStack.count() - 1
+            args.update(
+                {
+                    'nodeid':event_id,
+                    'title':event_type.name,
+                    'index':nindex + 1,
+                    'predecessor_node_id':self._nodeStack[nindex - 1] if nindex > 0 else ''
+                }
+            )
+            nd_event = ChatCallbackEvent(event_type = ChatEventType.NODE_FINISHED,payload = args)
+            if nd_event.to_response() is not None:
+                self._aqueue.put_nowait(nd_event)
+            self._nodeStack.pop()
+
+
+    def start_trace(self, trace_id: Optional[str] = None) -> None:
+        """No-op."""
+        logger.info("trace_start:{}\n".format(trace_id))
+
+    def end_trace(
+            self,
+            trace_id: Optional[str] = None,
+            trace_map: Optional[Dict[str, List[str]]] = None,
+    ) -> None:
+        """No-op."""
+        logger.info("trace_end:{} trace_map:{}\n".format(trace_id, trace_map))
+        data:ChatRequestData = self._params['data']    
+        args:Dict[str,Any] = self._params['ids']
+        args.update(
+            {
+                'response':self._response,
+                'conversation_id': data.conversation_id
+            }
+        )
+        wf_event = ChatCallbackEvent(event_type = ChatEventType.WORKFLOW_FINISHED,payload = args)
+        if wf_event.to_response() is not None:
+            self._aqueue.put_nowait(wf_event)
+
+        
+        args:Dict[str,Any] = self._params['ids']
+        msgEnt_event = ChatCallbackEvent(event_type = ChatEventType.MESSAGE_END,payload = args)
+        if msgEnt_event.to_response() is not None:
+            self._aqueue.put_nowait(msgEnt_event)                 
+
+    async def async_event_gen(self) -> AsyncGenerator[ChatCallbackEvent, None]:
+        while not self._aqueue.empty() or not self.is_done:
+            try:
+                yield await asyncio.wait_for(self._aqueue.get(), timeout=0.1)
+            except asyncio.TimeoutError:
+                pass
+
+class IDManager:
+    def createID(self):
+        return {
+            "message_id" : str(uuid.uuid4()),
+            'task_id':str(uuid.uuid4()),
+            'workflow_run_id': str(uuid.uuid4()),
+            "workflow_id": str(uuid.uuid4())
+        }
+
+class ChatStreamResponse(StreamingResponse):
+    TEXT_PREFIX = "data: "
+    DATA_PREFIX = "data: "
+    ids:Dict[str,Any] = {}
+    data:ChatRequestData = None
+
+    @classmethod
+    def convert_Message(cls, token: str):
+        params = cls.ids
+        params.update({
+            'answer':token,
+            'conversation_id':cls.data.conversation_id
+        })
+        event = ChatCallbackEvent(event_type = ChatEventType.MESSAGE,payload = params)
+        data_str = json.dumps(event.to_response())
+        return f"{cls.DATA_PREFIX}{data_str}\n\n"
+
+    @classmethod
+    def convert_Event(cls, data: dict):
+        data_str = json.dumps(data)
+        return f"{cls.DATA_PREFIX}{data_str}\n\n"
+
+    def __init__(
+            self,
+            request: Request,
+            event_handler: ChatEventCallbackHandler,
+            response: StreamingAgentChatResponse,
+            data: ChatRequestData,
+            ids:Dict[str,Any]
+    ):
+        ChatStreamResponse.ids = ids
+        ChatStreamResponse.data = data
+        content = ChatStreamResponse.content_generator(
+            request, event_handler, response, data
+        )
+        super().__init__(content=content)
+
+    @classmethod
+    async def content_generator(
+            cls,
+            request: Request,
+            event_handler: ChatEventCallbackHandler,
+            response: StreamingAgentChatResponse,
+            data: ChatRequestData
+    ):
+        
+        # Yield the text response
+        async def _chat_response_generator():
+            final_response = ""
+            async for token in response.async_response_gen():
+                final_response += token
+                yield ChatStreamResponse.convert_Message(token)
+
+            # 存储消息历史
+            message().add(user_id=data.user,conversation_id=data.conversation_id,query=data.query,answer=final_response)
+
+            # the text_generator is the leading stream, once it's finished, also finish the event stream
+            event_handler.is_done = True
+       
+        # Yield the events from the event handler
+        async def _event_generator():
+            async for event in event_handler.async_event_gen():
+                event_response = event.to_response()
+                if event_response is not None:
+                    yield ChatStreamResponse.convert_Event(event_response)
+
+        combine = stream.merge(_chat_response_generator(), _event_generator())
+        is_stream_started = False
+        async with combine.stream() as streamer:
+            async for output in streamer:
+                if not is_stream_started:
+                    is_stream_started = True
+
+                yield output
+
+                if await request.is_disconnected():
+                    break
+
+@v.post("/chat-messages")
+async def post_conversations(request: Request, data: ChatRequestData):
+    userMng.findNoExistCreate(data.user)
+    data.conversation_id = data.conversation_id if data.conversation_id else str(uuid.uuid4())
+
+    conversaObj = conversations()
+    conversationinfo = conversaObj.get(data.conversation_id)
+    if conversationinfo is None:
+        conversationinfo = conversaObj.add(data.conversation_id, data.user, "新建会话")
+
+    # 生成聊天参数
+    last_message_content = ChatMessage.from_str(data.query)
+    filters = None
+    params = data.inputs or {}
+
+    # 获取聊天引擎对象
+    chat_engine = get_chat_engine(filters=filters, params=params)
+
+    # 启动聊天事件监听
+    ids = IDManager().createID()
+    event_handler = ChatEventCallbackHandler(ids = ids,data = data)
+    chat_engine.callback_manager.handlers.append(event_handler)  # type: ignore
+
+    # 执行异步聊天
+    response = await chat_engine.astream_chat(data.query)
+
+    # 返回异步消息回应
+    return ChatStreamResponse(request, event_handler, response, data,ids)
+
+@v.get("/messages")
+async def query_messages(user:str, conversation_id:str):
+    #conversation_id = default_conversation_id if conversation_id is None else conversation_id
+    datas = []
+    records = message().gets(user,conversation_id)
+    if records is None:
+        return {
+            "limit": 20,
+            "has_more": False,
+            "data": []
+        }
+
+    for record in records:
+        res = record.dict()
+        feeds = feedback().query(res['id'])  
+        res["message_files"] = []
+        res["feedback"] = {'rating':feeds['rating'] } if feeds != None else ''
+        res["retriever_resources"] = []
+        res["created_at"] = 1723444905
+        res["agent_thoughts"] = []
+        res["status"] =  "normal"
+        res["error"] = ''
+        datas.append(res)
+
+    return {
+        "limit": 20,
+        "has_more": False,
+        "data": datas
+    }  
+    
+@v.post("/conversations/{itemid}/name")
+async def post_conversations(request: Request,itemid:str,params:Dict[str,Any]):
+    consaObj = conversations()
+    consaObj.rename(itemid,'知识问答')
+    cond = {
+        'id':itemid,
+        'user_id':params['user']
+    }
+    results = consaObj.query(**cond)
+    if len(results) > 0:
+        res = results[0]
+        return {
+            "id": res['id'],
+            "name": res['name'],
+            "inputs": res['inputs'],
+            "status": res['status'],
+            "introduction": res['introduction'],
+            "created_at": res['created_at'],
+            #"工程位置"
+        }
+    return 'null'
+
+@v.get("/conversations")
+async def query_conversations(user:str, first_id:str = None, limit:str = None, pinned:str = None):
+    user_id = '' if user is None else user
+    userMng.findNoExistCreate(user_id)
+
+    return {
+        "limit": 20,
+        "has_more": False,
+        "data": conversations().gets(user_id)
+    }    
+
+@v.get("/parameters")
+async def query_parameters(user:str):
+    params = parameter().get(user)
+    if len(params) == 0:
+        params = BaseConfig().ParamterCfg() 
+    return params
+
+@v.post("/messages/{message_id}/feedbacks")
+async def post_feedbacks(request: Request,message_id:str,params:Dict[str,Any]):
+    if params['rating'] =='null':
+        feedback().delete(message_id)
+    else:
+        condition = {'id':message_id}
+        results = message().query(**condition)
+        if len(results) > 0:
+            result = results[0]
+            feedback().add(message_id=message_id,query=result['query'],
+                        answer=result['answer'],rating=params['rating'])
+
+@r.post("")
+def upload_file(request: ChatFileUploadRequest) -> List[str]:
+    pass
+

backend/app/api/routers/chat.py

@@ -48,6 +48,8 @@ async def chat(
 ):
     try:
         last_message_content = data.get_last_message_content()
+        # 由于基于历史消息的提示词没有调整好，所以暂时屏蔽历史消息
+        data.messages.clear()
         messages = data.get_history_messages()
 
         doc_ids = data.get_chat_document_ids()
@@ -59,8 +61,6 @@ async def chat(
         event_handler = EventCallbackHandler()
         chat_engine.callback_manager.handlers.append(event_handler)  # type: ignore
 
-        # 由于基于历史消息的提示词没有调整好，所以暂时屏蔽历史消息
-        messages = None
         response = await chat_engine.astream_chat(last_message_content, messages)
         process_response_nodes(response.source_nodes, background_tasks)
 

backend/app/api/routers/request/base.py

@@ -0,0 +1,155 @@
+from datetime import datetime
+import uuid
+from app.api.routers.request.baseConfig import BaseConfig
+from app.api.routers.request.dbOrm import DBManager
+
+dbManage = DBManager()
+
+class conversations:
+    def __init__(self) -> None:
+        self._tableName = 'conversations'
+        dbManage.createTable(self._tableName)
+
+    def gets(self,user_id:str):
+        records = dbManage.query(self._tableName,user_id = user_id)
+        datas = []
+        for record in records:
+            datas.append(record)
+
+        return datas
+
+    def get(self, id:str):
+        records = dbManage.query(self._tableName, id=id)
+        if len(records) >0:
+            return records[0]
+        return None
+
+    def add(self,id:str, user_id:str, name:str):
+        template = BaseConfig().ConversationCfg()
+        template['id'] = id
+        template['user_id'] = user_id
+        template['name'] = name
+        template['created_at'] = 1724399038
+        dbManage.addRecord(self._tableName,template)
+        
+    def delete(self,id:str):
+        dbManage.delete(self._tableName,id=id)
+    
+    def rename(self,id:str,name:str):
+        data = {'name':name}
+        dbManage.update(self._tableName,data,id=id)
+
+    def query(self,**condition):
+        results = []
+        records = dbManage.query(self._tableName,**condition)
+        for record in records:
+            results.append(record.dict())
+        return results
+
+class user:
+    def __init__(self) -> None:
+        self._tableName = 'user'
+        dbManage.createTable(self._tableName)
+
+    def gets(self):
+       return dbManage.query(self._tableName)
+
+    def get(self,id:str):
+        return dbManage.query(self._tableName,id = id)
+
+    def add(self,id:str):
+        info = {
+            'id':id,
+            'createtime': datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+        }
+        dbManage.addRecord(self._tableName,info)
+
+    def delete(self,id:str):
+        dbManage.delete(self._tableName,id = id)
+
+class userMng:
+    userObj = user()
+    @classmethod
+    def findNoExistCreate(cls,user_id:str):
+        userInfo = cls.userObj.get(user_id)
+        if len(userInfo) == 0:
+            cls.userObj.add(user_id)
+    
+    def remove(cls,user_id:str):
+        cls.userObj.delete(user_id)
+
+class parameter:
+    def __init__(self) -> None:
+        self._tableName = 'parameters'
+        dbManage.createTable(self._tableName)
+
+    def get(self,user_id:str):
+        records = dbManage.query(self._tableName,user_id = user_id)
+        data = {}
+        for record in records:
+            key = record['name']
+            value = record['value']
+            data[key] = value
+        return data
+
+    def set(self,user_id:str):
+        dbManage.addRecord(self._tableName,{})
+
+    def delete(self,user_id:str):
+        dbManage.delete(self._tableName,user_id = user_id)
+
+class message:
+    def __init__(self) -> None:
+        self._tableName = 'messages'
+        dbManage.createTable(self._tableName)
+
+    def gets(self,user_id:str,conversation_id:str):
+        records = dbManage.query(self._tableName,user_id = user_id,conversation_id = conversation_id)
+        datas = []
+        for record in records:
+            datas.append(record)
+        return datas
+
+    def add(self,user_id:str,conversation_id:str,query:str,answer:str):
+        template = BaseConfig.MessageCfg()
+        template['id'] = str(uuid.uuid4())
+        template['user_id'] = user_id
+        template['conversation_id'] = conversation_id
+        template['query'] = query
+        template['answer'] = answer
+        dbManage.addRecord(self._tableName,template)
+
+    def delete(self,user_id:str):
+        dbManage.delete(self._tableName,user_id = user_id)
+
+    def query(self,**condition):
+        results = []
+        records = dbManage.query(self._tableName,**condition)
+        for record in records:
+            results.append(record.dict())
+        return results        
+
+class feedback:
+    def __init__(self) -> None:
+        self._tableName = 'feedbacks'
+        dbManage.createTable(self._tableName)
+
+    def add(self,message_id:str,query:str,answer:str,rating:str):
+        record = {
+            'message_id': message_id,
+            'query': query,  
+            'answer': answer,  
+            'rating': rating,  
+        }
+        dbManage.addRecord(self._tableName,record)
+ 
+    def delete(self,message_id:str):
+        cond = {'message_id':message_id}
+        dbManage.delete(self._tableName,**cond)
+    
+    def query(self,message_id:str):
+        cond = {'message_id':message_id}
+        records = dbManage.query(self._tableName,**cond)
+        if len(records) > 0:
+            return records[0].dict()    
+        return None

backend/app/api/routers/request/baseConfig.py

@@ -0,0 +1,80 @@
+from pydantic import BaseModel
+import os
+from enum import Enum
+
+class BaseConfig(BaseModel):
+	projectInfo:str = os.getenv("PROJECT_TITLE","您好，我是博微工程理解小助手，您可以问我有关[线路工程]工程数据的相关问题！")
+
+	def ParamterCfg(self):
+		questions = os.getenv("CONVERSATION_STARTERS", "dev")
+		return{
+			"opening_statement": self.projectInfo,
+			"suggested_questions": questions.split('\n'),
+			"suggested_questions_after_answer": {
+				"enabled": False
+			},
+			"speech_to_text": {
+				"enabled": False
+			},
+			"text_to_speech": {
+				"enabled": False,
+				"language": "",
+				"voice": ""
+			},
+			"retriever_resource": {
+				"enabled": True
+			},
+			"annotation_reply": {
+				"enabled": False
+			},
+			"more_like_this": {
+				"enabled": False
+			},
+			"user_input_form": [],
+			"sensitive_word_avoidance": {
+				"enabled": False
+			},
+			"file_upload": {
+				"image": {
+					"enabled": False,
+					"number_limits": 3,
+					"transfer_methods": [
+						"remote_url"
+					]
+				}
+			},
+			"system_parameters": {
+				"image_file_size_limit": "10"
+			}
+		}
+	
+	def ConversationCfg(self):
+		return{
+			"id": "",
+			'user_id':'',
+			"name": "",
+			"inputs": {},
+			"status": "normal",
+			"introduction": self.projectInfo,
+			"created_at":''
+		}
+
+	@classmethod 
+	def MessageCfg(cls):
+		return {
+		    "id": "",
+			'user_id':'',
+            "conversation_id": "",
+            "inputs": {},
+            "query": "",
+            "answer": ""
+		}
+
+
+class ChatEventType(str, Enum):
+    WORKFLOW_START = "workflow_started"
+    WORKFLOW_FINISHED = "workflow_finished"
+    NODE_START = "node_started"
+    NODE_FINISHED = "node_finished"
+    MESSAGE = "message"
+    MESSAGE_END = "message_end"

backend/app/api/routers/request/dbOrm.py

@@ -0,0 +1,220 @@
+import os
+from typing import Dict, List, Any
+
+from pydantic import BaseModel
+from sqlalchemy import create_engine, Column, String, Integer, JSON,Float
+from sqlalchemy.engine.reflection import Inspector
+from sqlalchemy.orm import sessionmaker, declarative_base
+
+Base = declarative_base()
+
+#orm类
+class ConversationOrm(Base):
+    __tablename__ = "conversations"
+
+    id = Column(String, primary_key=True)
+    user_id = Column(String)
+    name = Column(String)
+    inputs = Column(JSON)
+    status = Column(String)
+    introduction = Column(String)
+    created_at = Column(Integer)
+
+    def update(self,data:Dict[str,Any]):
+        if 'name' in data:
+            self.name = data['name']
+
+class UserOrm(Base):
+    __tablename__ = "user"
+
+    id = Column(String, primary_key=True)
+    createtime = Column(String)
+
+class ParametersOrm(Base):
+    __tablename__ = "parameters"
+
+    user_id = Column(String,primary_key=True)
+    name = Column(String)
+    value = Column(JSON)
+
+class MessagesOrm(Base):
+    __tablename__ = "messages"
+
+    id = Column(String,primary_key=True)
+    user_id = Column(String)
+    conversation_id = Column(String)
+    inputs = Column(JSON)
+    query = Column(String)
+    answer = Column(String)
+
+class FeedBackOrm(Base):
+    __tablename__ = "feedbacks"
+
+    message_id = Column(String,primary_key=True)
+    query = Column(String)
+    answer = Column(String)
+    rating = Column(String)
+
+#数据结构
+class ConversationModel(BaseModel):
+    id: str
+    name: str
+    inputs: Dict[str, Any]
+    status: str
+    introduction: str
+    created_at: int
+
+    class Config:
+        from_attributes=True
+
+    @classmethod
+    def orm(cls):
+        return ConversationOrm
+
+class UserModel(BaseModel):
+    id: str
+    createtime: str
+
+    class Config:
+        from_attributes=True
+
+    @classmethod
+    def orm(cls):
+        return UserOrm
+
+class ParametersModel(BaseModel):
+    user_id : str
+    name : str
+    value : Dict[str, Any]
+
+    class Config:
+        from_attributes=True
+
+    @classmethod
+    def orm(cls):
+        return ParametersOrm
+
+class MessagesModel(BaseModel):
+    id :str
+    conversation_id :str
+    inputs : Dict[str, Any]
+    query : str
+    answer : str
+
+    class Config:
+        from_attributes=True
+
+    @classmethod
+    def orm(cls):
+        return MessagesOrm
+
+class FeedBackModel(BaseModel):
+    message_id :str
+    query :str
+    answer :str
+    rating :str
+
+    class Config:
+        from_attributes=True
+
+    @classmethod
+    def orm(cls):
+        return FeedBackOrm
+
+class DBManager:
+    def __init__(self) -> None:
+        DATABASE_URL = os.getenv("SQLITE_DATABASE_URL")
+        self._engine = create_engine(DATABASE_URL)
+        self.SessionLocal = sessionmaker(autocommit=False, autoflush=False, bind=self._engine)
+    
+    def createTable(self,tableName:str):
+        if self._engine is None:
+            return
+        if not self.exist(tableName):
+            Base.metadata.tables[tableName].create(self._engine)
+
+    def addRecord(self,tableName:str,record:Dict[str,Any]):
+        ormCls = self._get_orm(tableName)
+        if ormCls is None:
+            return 
+        session = self.SessionLocal()
+        data = ormCls(**record)
+        session.add(data)
+        session.commit()
+
+    def addRecords(self,tableName:str,records:List[Dict[str,Any]]):
+        ormCls = self._get_orm(tableName)
+        if ormCls is None:
+            return 
+        datas = []
+        session = self.SessionLocal()
+        for record in records:
+            datas.append(ormCls(**record))
+        session.add(datas)
+        session.commit()
+ 
+    def delete(self,tableName:str,**filter):
+        session = self.SessionLocal()
+        ormCls = self._get_orm(tableName)
+        if ormCls is None:
+            return 
+        records = session.query(ormCls).filter_by(**filter).all()
+        if records is not None:
+            session.delete(records)
+            session.commit()
+    
+    def update(self,tableName:str,data:Dict[str,Any],**filter):
+        if not self.exist(tableName):
+            return
+        session = self.SessionLocal()
+        ormCls = self._get_orm(tableName)
+        if ormCls is None:
+            return 
+        if len(filter) > 0:
+            records = session.query(ormCls).filter_by(**filter).all()
+        else:
+            records = session.query(ormCls).all()
+        for record in records:
+            if record is not None:
+                record.update(data)
+                session.commit()    
+
+    def query(self,tableName:str,**filter):
+        session = self.SessionLocal()
+        ormCls = self._get_orm(tableName)
+        if ormCls is None:
+            return 
+        modelCls = self._get_model(ormCls)
+        if modelCls is None:
+            return 
+        
+        if filter is not None:
+            records = session.query(ormCls).filter_by(**filter).all()
+        else:
+            records = session.query(ormCls).all() 
+
+        datas  = []
+        for record in records:
+            datas.append(modelCls.from_orm(record))
+        return datas
+
+    def exist(self,tableName:str)->bool:
+        if self._engine is None:
+            return
+        inspector = Inspector.from_engine(self._engine)
+        return inspector.has_table(tableName)
+
+    def _get_orm(self,tableName:str):
+        subClss = Base.__subclasses__()
+        for sunCls in subClss:
+            if sunCls.__tablename__ == tableName:
+                return sunCls
+        return None
+
+    def _get_model(self,orm:Any):
+        subClss = BaseModel.__subclasses__()
+        for sunCls in subClss:
+            if 'orm' in sunCls.__dict__ and sunCls.orm() == orm:
+                return sunCls
+        return None
+

backend/app/api/routers/request/models.py

@@ -0,0 +1,17 @@
+
+from typing import Dict, Any
+from pydantic import BaseModel
+from typing import Optional
+
+class ChatRequestData(BaseModel):
+    inputs: Dict[str,Any]
+    query: str
+    user: str
+    response_mode: str
+    files: Any
+    conversation_id: str = None
+
+class ChatFileUploadRequest(BaseModel):
+    base64: str
+
+

backend/app/engine/__init__.py

@@ -1,81 +1,63 @@
 import os
 
-from llama_index.core import SQLDatabase, SummaryIndex, VectorStoreIndex
+from llama_index.core.agent import AgentRunner, ReActChatFormatter
-from llama_index.core.indices.struct_store import SQLTableRetrieverQueryEngine
-from llama_index.core.objects import SQLTableNodeMapping, ObjectIndex
 from llama_index.core.settings import Settings
-from llama_index.core.agent import AgentRunner, StructuredPlannerAgent, FunctionCallingAgentWorker
 from llama_index.core.tools.query_engine import QueryEngineTool
-from sqlalchemy import create_engine, Engine
 
-from app.engine.loaders.db import makeDescriptionByEngine
+from app.engine.engine import create_query_engine, create_summary_query_engine
-from app.engine.tools import ToolFactory
 from app.engine.index import get_index
-from app.settings import get_node_postprocessors
+#from app.engine.loaders.db import makeDescriptionByEngine
+from app.engine.tools import ToolFactory
 
-sql_database = None
-sql_obj_index = None
 
 def get_chat_engine(filters=None, params=None):
     system_prompt = os.getenv("SYSTEM_PROMPT")
     top_k = int(os.getenv("TOP_K", "3"))
+    use_reranker = os.getenv("RERANK_ENABLED")
     tools = []
 
-    global sql_obj_index
-    global sql_database
-    if sql_obj_index is None:
-        sqlengine = create_engine(os.getenv("SQL_DATABASE_URL", ""))
-        sql_database = SQLDatabase(sqlengine)
-        table_schema_objs = makeDescriptionByEngine(sql_database)
-        table_node_mapping = SQLTableNodeMapping(sql_database)
-
-        sql_obj_index = ObjectIndex.from_objects(
-            table_schema_objs,
-            table_node_mapping,
-            index_cls=VectorStoreIndex,
-        )
-
     # 创建SQL查询工具
-    sql_query_engine = SQLTableRetrieverQueryEngine(sql_database,
+#    sql_query_engine = create_summary_query_engine(index)
-                                                    sql_obj_index.as_retriever(similarity_top_k=top_k),
+    # sql_query_tool = QueryEngineTool.from_defaults(query_engine=sql_query_engine,
-                                                    verbose=True,)
+    #                                                name="zjdata_query_tool",
-    sql_query_tool = QueryEngineTool.from_defaults(query_engine=sql_query_engine,
+    #                                                description="来源于一个由博微公司电力造价软件编制的造价工程文件。该文件以多张表格的形式存储存储了整个工程的全部数据内容。适用于以详细的自然语言查询表格数据方式查询造价工程各项具体属性、费用的数值。请先使用“zj_query_tool”无法解决才使用本工具"
-                                                   name="zjdata_query_tool",
+    #                                                )
-                                                   description="来源于一个由博微公司电力造价软件编制的造价工程文件。该文件以多张表格的形式存储存储了整个工程的全部数据内容。适用于以详细的自然语言查询表格数据方式查询造价工程各项具体属性、费用的数值。请先使用“zj_query_tool”无法解决才使用本工具")
+    #tools.append(sql_query_tool)
 
     # Add query tool if index exists
     index = get_index()
     if index is not None:
-        summary_index = SummaryIndex(index.vector_store.get_nodes(node_ids=None))
+        summary_query_engine = create_summary_query_engine(index,top_k,use_reranker,filters)
-        summary_query_engine = summary_index.as_query_engine()
         summary_query_tool = QueryEngineTool.from_defaults( query_engine=summary_query_engine, name="summary_query_tool",
                                                             description="适用于任何需要进行全面总结、概括的要求。",
-                                                            #description="适用于任何需要对所有内容进行全面总结的请求。有关电力造价领域更具体部分的问题，请使用zj_query_engine_tool",
                                                             )
-
+        query_engine = create_query_engine(index,top_k,use_reranker,filters,response_mode = "COMPACT")
-        # 创建向量检索查询工具
-        postprocess = get_node_postprocessors()
-        query_engine = index.as_query_engine(
-            similarity_top_k=top_k, filters=filters,
-            node_postprocessors=postprocess,
-        )
         query_engine_tool = QueryEngineTool.from_defaults(query_engine=query_engine, name="zj_query_tool",
-                                                          description="由博微公司编制的关于电力造价知识、电力造价编制软件知识和造价工程文件结构的知识库。适用于查询电力领域、电力造价领域、博微、博微电力、博微造价等业务等内容。如果本知识库没有直接答案但有解决思路的可以返回解决办法后建议使用“zjdata_query_tool”工具。如果你不知道答案，就说你不知道，不要编造答案。",
+                                                          description="由博微公司编制的关于电力造价知识、电力造价编制软件知识和造价工程文件结构的知识库。适用于查询电力领域、电力造价领域、博微、博微电力、博微造价等业务等内容。如果本知识库没有直接答案但有解决思路的可以返回解决办法后建议使用“zjdata_query_tool”工具。",
+                                                          )
+        
+        query_engine = create_query_engine(index,top_k,use_reranker,filters,response_mode = "TREE_SUMMARIZE")
+        query_engine_tool_1 = QueryEngineTool.from_defaults(query_engine=query_engine, name="zj_query_tool_1",
+                                                          description="由博微公司编制的关于电力造价知识、电力造价编制软件知识和造价工程文件结构的知识库。适用于查询电力领域、电力造价领域、博微、博微电力、博微造价等业务等内容。如果本知识库没有直接答案但有解决思路的可以返回解决办法后，且在询问工程中单位的具体数值，例如用量，费率，合计，金额等的时候建议使用“zj_query_tool_1”工具。",
                                                           )
 
         tools.append(summary_query_tool)
         tools.append(query_engine_tool)
-        #tools.append(sql_query_tool)
+        tools.append(query_engine_tool_1)
 
     # Add additional tools
     tools += ToolFactory.from_env()
 
-    return AgentRunner.from_llm(
+    prefix_messages = ("""您的设计旨在帮助完成各种任务，从回答问题到提供其他类型分析的摘要。\n\n##工具\n\n你可以访问各种工具。你有责任按照你认为合适的顺序使用这些工具来完成当前的任务。\n这可能需要将任务分解为子任务，并使用不同的工具来完成每个子任务。\n\n你可以访问以下工具：\n{tool_desc}\n\n\n##输出格式\n\n请用与问题相同的语言回答，并使用以下格式：\n\n   \nThought: 用户当前的语言是：(user's language)。我需要使用工具来帮助我回答问题。\nAction: 如果使用工具,则为工具名称(one of {tool_names})。\nAction Input: 输入给工具的内容，使用JSON格式表示kwargs（例如{{\"input\": \"hello world\", \"num_beams\": 5}}）\n   \n\n请始终以Thought开始。\n\n请始终以Thought开始。\n\n请始终以Thought开始。\n\n请始终以Thought开始。\n\n切勿用Markdown代码标记包围你的响应。如果需要，可以在响应中使用代码标记。\n\n请为Action Input使用有效的JSON格式。不要这样做{{\'input\': \'hello world\', \'num_beams\': 5}}。\n\n如果使用此格式，用户将以下面的格式进行回应：\n\n   \nObservation: 工具响应\n   \n\n你应该继续重复上述格式，直到你有足够的信息来回答问题而无需使用更多工具。此时，你必须使用以下两种格式之一进行回答：\n\n   \nThought: 我可以不用任何工具来回答。我将使用用户的语言来回答。\nAnswer: [你的答案（与用户问题相同的语言）]\n   \n\n   \nThought: 我无法使用提供的工具回答问题。\nAnswer: [你的答案（与用户问题相同的语言）]\n   \n\n##如果从工具中得到的回应是Empty Response，那么只需要回答“我不知道”，不需要额外回答别的内容。## 当前对话\n\n以下是当前对话，由人类和助手的消息交替组成。\n""")
+    react_chat_formatter = ReActChatFormatter.from_defaults(prefix_messages)
+    agentrunner = AgentRunner.from_llm(
         llm=Settings.llm,
         tools=tools,
+        react_chat_formatter=react_chat_formatter,
         system_prompt=system_prompt,
         verbose=True,
     )
+    return agentrunner
     # create the function calling worker for reasoning
     # worker = FunctionCallingAgentWorker.from_tools(
     #     tools, verbose=True

backend/app/engine/engine.py

@@ -0,0 +1,109 @@
+import os
+
+from llama_index.core import SummaryIndex, SQLDatabase, VectorStoreIndex
+from llama_index.core.indices.struct_store import SQLTableRetrieverQueryEngine
+from llama_index.core.objects import SQLTableNodeMapping, ObjectIndex, SQLTableSchema
+from llama_index.core.query_engine import RetrieverQueryEngine
+from llama_index.core.response_synthesizers import ResponseMode
+from llama_index.readers.database import DatabaseReader
+from sqlalchemy import create_engine
+
+from app.engine.prompt import text_qa_template, refine_template, summary_template, simple_template
+from app.engine.retriever.HybridRetriever import HybridRetriever
+from app.settings import get_node_postprocessors
+
+def makeDescriptionByEngine(sql_database:SQLDatabase):
+    reader = DatabaseReader(sql_database)
+
+    table_names = sql_database.get_usable_table_names()
+    table_schema_objs = []
+    for table_name in table_names:
+        columns = sql_database.get_table_columns(table_name)
+        if len(columns) > 150:
+            continue
+        stats_txt = ""
+
+        if table_name == 'gongchengshuxing':
+            stats_txt = '该表中有以下属性:'
+            documents = reader.load_data(query='select name from gongchengshuxing')
+            for index in range(len(documents) if len(documents) < 30 else 30):
+                if index == 0:
+                    continue
+                elif index > 1:
+                    stats_txt += ','
+                stats_txt += documents[index].text.split(':')[1]
+
+        tbSchema = (SQLTableSchema(table_name=table_name, context_str=stats_txt))
+        table_schema_objs.append(tbSchema)
+
+    return table_schema_objs
+
+def get_Retriever(index,**kwargs):
+    strEnableHybrid = os.getenv("HYBRID_ENABLED",'False')
+    bEnableHybrid = True if strEnableHybrid is not None and strEnableHybrid.title() == 'True' else False
+    if bEnableHybrid:
+        alpha = float(os.getenv("HYBRID_ALPHA", "0.5"))
+        retriever = HybridRetriever(index,alpha = alpha,**kwargs)
+    else:
+        retriever = index.as_retriever(**kwargs)
+    return retriever
+
+
+sql_database = None
+sql_obj_index = None
+
+# Create a summary query engine
+def create_summary_query_engine(top_k=3, use_reranker=False, filters=None):
+    global sql_obj_index
+    global sql_database
+    if sql_obj_index is None or sql_database is None:
+        sqlengine = create_engine(os.getenv("SQL_DATABASE_URL", ""))
+        sql_database = SQLDatabase(sqlengine)
+        table_schema_objs = makeDescriptionByEngine(sql_database)
+        table_node_mapping = SQLTableNodeMapping(sql_database)
+
+        sql_obj_index = ObjectIndex.from_objects(
+            table_schema_objs,
+            table_node_mapping,
+            index_cls=VectorStoreIndex,
+        )
+
+    # 创建SQL查询工具
+    sql_query_engine = SQLTableRetrieverQueryEngine(sql_database,
+                                                    sql_obj_index.as_retriever(similarity_top_k=top_k),
+                                                    verbose=True,
+                                                    )
+    return sql_query_engine
+
+# Create a summary query engine
+def create_summary_query_engine(index, top_k=3, use_reranker=False, filters=None):
+    summary_index = SummaryIndex(index.vector_store.get_nodes(node_ids=None))
+    summary_query_engine = summary_index.as_query_engine(
+    response_mode=ResponseMode.TREE_SUMMARIZE,
+    use_async=True,
+    streaming=True,
+    )
+    return summary_query_engine
+
+# Create a query engine
+def create_query_engine(index, top_k=3, use_reranker=False, filters=None, response_mode=None):
+    # 创建向量检索查询工具
+    postprocess = None
+    if use_reranker:
+        postprocess = get_node_postprocessors()
+
+    query_engine = RetrieverQueryEngine.from_args(
+        get_Retriever(index,
+                      similarity_top_k=top_k,
+                      filters=filters),
+        text_qa_template=text_qa_template,
+        refine_template=refine_template,
+        summary_template = summary_template,
+        simple_template = simple_template,
+        node_postprocessors=postprocess,
+        use_async=True,
+        streaming=True,
+        ResponseMode = response_mode
+    )
+
+    return query_engine

backend/app/engine/generate.py

@@ -8,6 +8,7 @@ import os
 from app.engine.loaders import get_documents
 from app.engine.vectordb import get_vector_store
 from app.settings import init_settings
+from app.engine.retriever.CHBM25Retriever import CHBM25Retriever
 from llama_index.core.ingestion import IngestionPipeline
 from llama_index.core.node_parser import SentenceSplitter
 from llama_index.core.settings import Settings
@@ -58,6 +59,13 @@ def persist_storage(docstore, vector_store):
     storage_context.persist(STORAGE_DIR)
 
 
+def persist_BMRetriever(vector_store):
+    STORAGE_DIR = os.getenv("BM_RETRIEVER_PATH", "storage_bm")
+    top_k = int(os.getenv("TOP_K", "3"))
+    bmRetriver = CHBM25Retriever.from_defaults(similarity_top_k=top_k,nodes=vector_store.get_nodes([]))
+    bmRetriver.persist(STORAGE_DIR)
+
+
 def generate_datasource():
     init_settings()
     logger.info("Generate index for the provided data")
@@ -75,6 +83,7 @@ def generate_datasource():
 
     # Build the index and persist storage
     persist_storage(docstore, vector_store)
+    persist_BMRetriever(vector_store)
 
     logger.info("Finished generating the index")
 

backend/app/engine/loaders/__init__.py

@@ -1,5 +1,4 @@
 import logging
-
 import yaml
 from app.engine.loaders.db import DBLoaderConfig, get_db_documents
 from app.engine.loaders.file import FileLoaderConfig, get_file_documents
@@ -9,7 +8,7 @@ logger = logging.getLogger(__name__)
 
 
 def load_configs():
-    with open("config/loaders.yaml") as f:
+    with open("config/loaders.yaml",encoding='UTF-8') as f:
         configs = yaml.safe_load(f)
     return configs
 
@@ -17,24 +16,26 @@ def load_configs():
 def get_documents():
     documents = []
     config = load_configs()
+
     if config is None or len(config.items()) == 0:
-        return  documents
+        return documents
 
     for loader_type, loader_config in config.items():
-        logger.info(
+        if loader_config.get('enable', True):  # 检查 enable 字段
-            f"Loading documents from loader: {loader_type}, config: {loader_config}"
+            logger.info(
-        )
+                f"Loading documents from loader: {loader_type}, config: {loader_config}"
+            )
 
-        loader_config = loader_config or []
+            loader_config = loader_config or []
-        match loader_type:
+            match loader_type:
-            case "file":
+                case "file":
-                document = get_file_documents(FileLoaderConfig(**loader_config))
+                    document = get_file_documents(FileLoaderConfig(**loader_config))
-            case "web":
+                case "web":
-                document = get_web_documents(WebLoaderConfig(**loader_config))
+                    document = get_web_documents(WebLoaderConfig(**loader_config))
-            case "db":
+                case "db":
-                document = get_db_documents(configs=[DBLoaderConfig(**cfg) for cfg in loader_config])
+                    document = get_db_documents(configs=[DBLoaderConfig(**cfg) for cfg in loader_config])
-            case _:
+                case _:
-                raise ValueError(f"Invalid loader type: {loader_type}")
+                    raise ValueError(f"Invalid loader type: {loader_type}")
-        documents.extend(document)
+            documents.extend(document)
 
-    return documents
+    return documents

backend/app/engine/loaders/db.py

@@ -1,24 +1,15 @@
-import os
 import logging
-from typing import List
 from typing import Any, List, Optional
 
-from llama_index.core.readers.base import BaseReader
-from llama_index.core.schema import Document
-from llama_index.core.utilities.sql_wrapper import SQLDatabase
-from sqlalchemy import text
-from sqlalchemy.engine import Engine
 from llama_index.core import SQLDatabase, Document
-from llama_index.core.objects import SQLTableSchema, SQLTableNodeMapping
-from llama_index.core.readers.base import BaseReader
 from llama_index.readers.database import DatabaseReader
-from pydantic import BaseModel, validator
+from pydantic import BaseModel
-from llama_index.core.indices.vector_store import VectorStoreIndex
+from sqlalchemy import create_engine, text
-from sqlalchemy import create_engine
+from sqlalchemy.engine import Engine
 
 logger = logging.getLogger(__name__)
 
-class CustomDatabaseReader(BaseReader):
+class CustomDatabaseReader(DatabaseReader):
     """Simple Database reader.
 
     Concatenates each row into Document used by LlamaIndex.
@@ -91,19 +82,20 @@ class CustomDatabaseReader(BaseReader):
         Returns:
             List[Document]: A list of Document objects.
         """
-        dco_str = ""
+        dco_str = ""  
+
         with self.sql_database.engine.connect() as connection:
             if query is None:
                 raise ValueError("A query parameter is necessary to filter the data")
             else:
                 result = connection.execute(text(query))
 
-            dco_str = ", ".join(
+            dco_str += ", ".join(
                 [f"{entry}" for entry in result.keys()]
-            )
+            ) + "\n"
 
             for item in result.fetchall():
-                # fetch each item
+                # Fetch each item
                 record_str = ", ".join(
                     [f"{entry}" for col, entry in zip(result.keys(), item)]
                 )
@@ -117,71 +109,36 @@ class CustomDatabaseReader(BaseReader):
 
 class DBLoaderConfig(BaseModel):
     uri: str
-    queries: List[str]
+    queries: List[dict]  
 
-def makeDescriptionByEngine(sql_database:SQLDatabase):
+def get_db_documents(configs: List[DBLoaderConfig]) -> List[Document]:
-    reader = DatabaseReader(sql_database)
-
-    table_names = sql_database.get_usable_table_names()
-    table_schema_objs = []
-    for table_name in table_names:
-        columns = sql_database.get_table_columns(table_name)
-        if len(columns) > 150:
-            continue
-        stats_txt = ""
-
-        if table_name == 'gongchengshuxing':
-            stats_txt = '该表中有以下属性:'
-            documents = reader.load_data(query='select name from gongchengshuxing')
-            for index in range(len(documents) if len(documents) < 30 else 30):
-                if index == 0:
-                    continue
-                elif index > 1:
-                    stats_txt += ','
-                stats_txt += documents[index].text.split(':')[1]
-
-        tbSchema = (SQLTableSchema(table_name=table_name, context_str=stats_txt))
-        table_schema_objs.append(tbSchema)
-
-    return table_schema_objs
-
-def get_db_documents(configs: list[DBLoaderConfig]):
     docs = []
 
-    if len(configs) == 0 or configs[0].uri == "":
+    if not configs or not configs[0].uri:
         logger.warning(
             f"Failed to load database, error message: uri is empty. Return as empty document list."
         )
         return docs
 
     metadata = {
-        #'file_name':'',
+        'file_type': 'application/booway.document.zj',
-        'file_type':'application/booway.document.zj',
-        #'file_path':'',
-        #'file_size':'',
-        #'creation_date':'',
-        #'last_modified_date':'',
     }
 
-    #from llama_index.readers.database import DatabaseReader
     for entry in configs:
         engine = create_engine(entry.uri)
         sql_database = SQLDatabase(engine)
 
-        table_schema_objs = makeDescriptionByEngine(sql_database)
-        table_node_mapping = SQLTableNodeMapping(sql_database)
-
-        nodes = table_node_mapping.to_nodes(table_schema_objs)
-        for node in nodes:
-            node.metadata.update(metadata)
-
-        docs.extend(nodes)
-
-        queries = entry.queries or []
         loader = CustomDatabaseReader(sql_database)
-        for query in queries:
+        for query_dict in entry.queries:
+            query = query_dict.get("sql", "")
+            explanation = query_dict.get("explanation", "")
             logger.info(f"Loading data from database with query: {query}")
             documents = loader.load_data(query=query)
 
-            docs.extend(documents)
+            # 添加解释到元数据中
-    return docs
+            for doc in documents:
+                doc.metadata["explanation"] = explanation
+                doc.metadata.update(metadata)  # 更新或添加额外的元数据
+                docs.append(doc)
+
+    return docs

backend/app/engine/prompt.py

@@ -0,0 +1,95 @@
+from llama_index.core import PromptTemplate
+
+text_qa_template_str = (
+    "# 角色\n"
+    "你是一名博微造价工程数据查询助手，专精于电力工程文件中的信息。"
+    "你的职责是提供有关电力造价、造价编制软件、文件结构及相关数据的精准、客观的回答，"
+    "如同直接从文件中提取的内容。\n"
+    "知识库中已经导入一个工程的全部数据，请你站在当前工程的角度回答用户关于工程文件的问题。\n"
+    "例如：询问“此工程”指当前导入的工程。询问“此工程名称”指当前导入的工程的工程名称。\n"
+    
+    "## 技能\n"
+    "### 技能 1: 数据查询与提供\n"
+    "- 准确回答所有关于电力工程造价的相关问题。\n"
+    "- 提供具体数据，如成本估算、材料清单、劳动力需求等。\n"
+    "- 确保提供的信息严格基于工程文档中的记录。\n"
+    
+    "### 技能 2: 技术性解释\n"
+    "- 解释造价工程中的技术术语和概念。\n"
+    "- 为复杂的工程细节提供清晰易懂的说明。\n"
+    
+    "## 约束\n"
+    "- 仅回答与电力工程造价文件相关的具体问题。\n"
+    "- 不进行任何超出文件内容的猜测或假设。\n"
+    "- 所有回答均基于文件内容，采用客观和技术性的语言。\n"
+    "- 请基于这些信息回答问题。如果无法找到相关信息，请不要额外发散回答，不要回答多余的信息，只需要回答“我不知道这个问题的答案”。\n"
+    "以下为上下文信息\n"
+    "---------------------\n"
+    "{context_str}\n"
+    "---------------------\n"
+    "请根据上下文信息而非先前知识回答我的问题或回复我的指令。前面的上下文信息可能有用，也可能没用，你需要从我给出的上下文信息中选出与我的问题最相关的那些，来为你的回答提供依据。回答一定要忠于原文，简洁但不丢信息，不要胡乱编造。如果无法找到相关信息，请不要额外发散回答，不要回答多余的信息，只需要回答“我不知道这个问题的答案”。我的问题或指令是什么语种，你就用什么语种回复。\n"
+    "如果是表结构或者是数据库的相关内容，只用于推导问题，不需要告诉用户数据库或表结构等物理信息。\n"
+
+    "问题：{query_str}\n"
+    "你的回复： "
+)
+
+
+text_qa_template = PromptTemplate(text_qa_template_str)
+
+refine_template_str = (
+    "这是原本的问题： {query_str}\n"
+    "我们已经提供了回答: {existing_answer}\n"
+    "现在我们有机会改进这个回答 "
+    "使用以下更多上下文（仅当有助于改进回答时使用）\n"
+    "你需要仔细的判断新的上下文的信息与原本问题必须一个字都不差，如果有一点差别，那就不能改变我现有的回答。\n"
+    "在判断回答是否正确的时候，你应该仔细对比新的上下文中包含的信息是否与原本的问题一字不差，如果一字不差，才能当作新的正确回答。\n"
+    "如果新的上下文对回答没有影响，或者原来的回答已经正确，不要在上次回答的后边再加上多余的补充信息，直接返回原本的回答。\n"
+    "判断一下如果原回答正确，且在新的上下文仍然包含正确的回答，请将新的回答与原回答一起返回。\n"
+    "------------\n"
+    "{context_msg}\n"
+    "------------\n"
+    "如果回答中已经包含有正确答案，不要返回多余的解释等信息，只返回正确答案\n"
+    "如果是表结构或者是数据库的相关内容，仅用于推导问题，不需要告诉用户数据库或表结构等物理信息。\n"
+    "改进的回答: "
+)
+
+refine_template = PromptTemplate(refine_template_str)
+
+summary_template_str = (
+    "# 角色\n"
+    "你是一名博微造价工程数据查询助手，专精于电力工程文件中的信息。"
+    "你的职责是提供有关电力造价、造价编制软件、文件结构及相关数据的精准、客观的回答，"
+    "如同直接从文件中提取的内容。\n"
+    
+    "## 技能\n"
+    "### 技能 1: 数据查询与提供\n"
+    "- 准确回答所有关于电力工程造价的相关问题。\n"
+    "- 提供具体数据，如成本估算、材料清单、劳动力需求等。\n"
+    "- 确保提供的信息严格基于工程文档中的记录。\n"
+    
+    "### 技能 2: 技术性解释\n"
+    "- 解释造价工程中的技术术语和概念。\n"
+    "- 为复杂的工程细节提供清晰易懂的说明。\n"
+    
+    "## 约束\n"
+    "- 仅回答与电力工程造价文件相关的具体问题。\n"
+    "- 不进行任何超出文件内容的猜测或假设。\n"
+    "- 所有回答均基于文件内容，采用客观和技术性的语言。\n"
+    "- 请基于这些信息回答问题。如果无法找到相关信息，请不要额外发散回答，不要回答多余的信息，只需要回答“我不知道这个问题的答案”。\n"
+    "来自多个来源的上下文信息如下。\n"
+    "---------------------\n"
+    "{context_str}\n"
+    "---------------------\n"
+    "鉴于来自多个来源的信息而非先验知识， "
+    "回答查询。\n"
+    "如果是表结构或者是数据库的相关内容，只用于推导问题，不需要告诉用户数据库或表结构等物理信息。\n"
+    "Query: {query_str}\n"
+    "Answer: "
+)
+summary_template = PromptTemplate(summary_template_str)
+
+simple_template_str = (
+    "{query_str}"
+)
+simple_template = PromptTemplate(simple_template_str)

backend/app/engine/retriever/CHBM25Retriever.py

@@ -0,0 +1,133 @@
+import json
+import logging
+import os
+
+from typing import Any, Callable, Dict, List, Optional, cast
+
+from llama_index.core.base.base_retriever import BaseRetriever
+from llama_index.core.callbacks.base import CallbackManager
+from llama_index.core.constants import DEFAULT_SIMILARITY_TOP_K
+from llama_index.core.indices.vector_store.base import VectorStoreIndex
+from llama_index.core.schema import BaseNode, IndexNode, NodeWithScore, QueryBundle
+from llama_index.core.storage.docstore.types import BaseDocumentStore
+from llama_index.core.vector_stores.utils import (
+    node_to_metadata_dict,
+    metadata_dict_to_node,
+)
+
+import bm25s
+from app.engine.retriever.CHTokener import chTokenize
+
+CHDEFAULT_PERSIST_ARGS = {"similarity_top_k": "similarity_top_k", "_verbose": "verbose"}
+
+CHDEFAULT_PERSIST_FILENAME = "retriever.json"
+
+class CHBM25Retriever(BaseRetriever):
+    def __init__(
+        self,
+        nodes: Optional[List[BaseNode]] = None,
+        existing_bm25: Optional[bm25s.BM25] = None,
+        similarity_top_k: int = DEFAULT_SIMILARITY_TOP_K,
+        callback_manager: Optional[CallbackManager] = None,
+        objects: Optional[List[IndexNode]] = None,
+        object_map: Optional[dict] = None,
+        verbose: bool = False,
+    ) -> None:
+        self.similarity_top_k = similarity_top_k
+        if existing_bm25 is not None:
+            self.bm25 = existing_bm25
+            self.corpus = existing_bm25.corpus
+        else:
+            from nltk.corpus import stopwords
+            if nodes is None:
+                raise ValueError("Please pass nodes or an existing BM25 object.")
+
+            self.corpus = [node_to_metadata_dict(node) for node in nodes]
+
+            corpus_tokens = chTokenize(
+                [node.get_content() for node in nodes],
+                show_progress=verbose,
+            )
+            self.bm25 = bm25s.BM25()
+            self.bm25.index(corpus_tokens, show_progress=verbose)
+        super().__init__(
+            callback_manager=callback_manager,
+            object_map=object_map,
+            objects=objects,
+            verbose=verbose,
+        )
+
+    @classmethod
+    def from_defaults(
+        cls,
+        index: Optional[VectorStoreIndex] = None,
+        nodes: Optional[List[BaseNode]] = None,
+        docstore: Optional[BaseDocumentStore] = None,
+        similarity_top_k: int = DEFAULT_SIMILARITY_TOP_K,
+        verbose: bool = False,
+    ) -> "CHBM25Retriever":
+        if sum(bool(val) for val in [index, nodes, docstore]) != 1:
+            raise ValueError("Please pass exactly one of index, nodes, or docstore.")
+
+        if index is not None:
+            docstore = index.docstore
+
+        if docstore is not None:
+            nodes = cast(List[BaseNode], list(docstore.docs.values()))
+
+        assert (
+            nodes is not None
+        ), "Please pass exactly one of index, nodes, or docstore."
+
+        return cls(
+            nodes=nodes,
+            similarity_top_k=similarity_top_k,
+            verbose=verbose,
+        )
+
+    def get_persist_args(self) -> Dict[str, Any]:
+        """Get Persist Args Dict to Save."""
+        return {
+            CHDEFAULT_PERSIST_ARGS[key]: getattr(self, key)
+            for key in CHDEFAULT_PERSIST_ARGS
+            if hasattr(self, key)
+        }
+
+    def persist(self, path: str, **kwargs: Any) -> None:
+        """Persist the retriever to a directory."""
+        self.bm25.save(path, corpus=self.corpus, **kwargs)
+        with open(os.path.join(path, CHDEFAULT_PERSIST_FILENAME), "w") as f:
+            json.dump(self.get_persist_args(), f, indent=2)
+
+    @classmethod
+    def from_persist_dir(cls, path: str, **kwargs: Any) -> "CHBM25Retriever":
+        """Load the retriever from a directory."""
+        bm25 = bm25s.BM25.load(path, load_corpus=True, **kwargs)
+        with open(os.path.join(path, CHDEFAULT_PERSIST_FILENAME)) as f:
+            retriever_data = json.load(f)
+        return cls(existing_bm25=bm25, **retriever_data)
+
+    def _retrieve(self, query_bundle: QueryBundle) -> List[NodeWithScore]:
+        query = query_bundle.query_str
+        tokenized_query = chTokenize(
+            query,show_progress=self._verbose
+        )
+        indexes, scores = self.bm25.retrieve(
+            tokenized_query, k=self.similarity_top_k, show_progress=self._verbose
+        )
+
+        # batched, but only one query
+        indexes = indexes[0]
+        scores = scores[0]
+
+        nodes: List[NodeWithScore] = []
+        for idx, score in zip(indexes, scores):
+            # idx can be an int or a dict of the node
+            if isinstance(idx, dict):
+                node = metadata_dict_to_node(idx)
+            else:
+                node_dict = self.corpus[int(idx)]
+                node = metadata_dict_to_node(node_dict)
+            nodes.append(NodeWithScore(node=node, score=float(score)))
+
+        return nodes

backend/app/engine/retriever/CHTokener.py

@@ -0,0 +1,50 @@
+import os
+from typing import Any, Dict, List, Union, Callable, NamedTuple
+from bm25s.tokenization import *
+
+try:
+    from tqdm.auto import tqdm
+except ImportError:
+
+    def tqdm(iterable, *args, **kwargs):
+        return iterable
+
+import jieba
+jiebapath = os.environ.get("JIEBA_DATA", "")
+jieba.set_dictionary(os.path.join(jiebapath, 'dict.txt')) #设置字典
+jieba.initialize() #初始化jeiba
+
+def chinese_tokenizer(text: str) -> List[str]:
+    from nltk.corpus import stopwords
+    tokens = jieba.lcut(text)
+    return [token for token in tokens if token not in stopwords.words('chinese')]
+
+def chTokenize(
+    texts,
+    show_progress: bool = True,
+    leave: bool = False,
+) -> Union[List[List[str]], Tokenized]:
+    if isinstance(texts, str):
+        texts = [texts]
+
+    corpus_ids = []
+    token_to_index = {}
+
+    for text in tqdm(
+        texts, desc="Split strings", leave=leave, disable=not show_progress
+    ):
+        
+        splitted = chinese_tokenizer(text)
+        doc_ids = []
+
+        for token in splitted:
+            if token not in token_to_index:
+                token_to_index[token] = len(token_to_index)
+
+            token_id = token_to_index[token]
+            doc_ids.append(token_id)
+
+        corpus_ids.append(doc_ids)
+
+    return Tokenized(ids=corpus_ids, vocab=token_to_index)
+

backend/app/engine/retriever/HybridRetriever.py

@@ -0,0 +1,67 @@
+import os
+from typing import Optional, Any, Dict, List
+
+from llama_index.core.base.base_retriever import BaseRetriever
+from llama_index.core.schema import NodeWithScore, QueryBundle
+
+from app.engine.retriever.CHBM25Retriever import CHBM25Retriever
+
+
+class HybridRetriever(BaseRetriever):
+    def __init__(
+            self,
+            vector_index,
+            similarity_top_k: int = 2,
+            out_top_k: Optional[int] = None,
+            alpha: float = 0.5,
+            filters = None,
+            **kwargs: Any,
+    ) -> None:
+        super().__init__(**kwargs)
+        self._vector_index = vector_index
+        self._embed_model = vector_index._embed_model
+        self._out_top_k = out_top_k or similarity_top_k
+        self._vecRetriever = vector_index.as_retriever(
+            similarity_top_k=similarity_top_k,filters = filters
+        )
+
+        STORAGE_DIR = os.getenv("BM_RETRIEVER_PATH", "storage_bm")
+        if os.path.exists(STORAGE_DIR) and len(os.listdir(STORAGE_DIR)) > 0:
+            self._bm25Retriever = CHBM25Retriever.from_persist_dir(STORAGE_DIR)
+        else:
+            bmRetriver = CHBM25Retriever.from_defaults(similarity_top_k=similarity_top_k,nodes=self._vector_index.vector_store.get_nodes(None))
+            bmRetriver.persist(STORAGE_DIR)
+        self._alpha = alpha
+
+
+
+    def _retrieve(self, query_bundle: QueryBundle) -> List[NodeWithScore]:
+        vecNodes:List[NodeWithScore] = self._vecRetriever.retrieve(query_bundle.query_str)
+        bmNodes:List[NodeWithScore] = self._bm25Retriever.retrieve(query_bundle.query_str)
+
+        bmDic:Dict[str,NodeWithScore] = {}
+        for node in bmNodes:
+            bmDic[node.node_id] = node
+
+        result_tups = []
+        for i in range(len(vecNodes)):
+            node = vecNodes[i]
+            bmScore = 0.0
+            if node.node_id in bmDic:
+                bmScore = bmDic[node.node_id].score
+                bmDic.pop(node.node_id)
+            else:
+                bmScore = 0.0
+            full_similarity = (self._alpha * node.score) + (
+                    (1 - self._alpha) * bmScore
+            )
+            result_tups.append((full_similarity, node))
+
+        for _,node in bmDic.items():
+            full_similarity = (1 - self._alpha) * node.score
+            result_tups.append((full_similarity, node))
+
+        result_tups = sorted(result_tups, key=lambda x: x[0], reverse=True)
+        for full_score, node in result_tups:
+            node.score = full_score
+        return [n for _, n in result_tups][:self._out_top_k]

backend/app/engine/tools/__init__.py

@@ -1,10 +1,9 @@
-import os
-import yaml
-import json
 import importlib
-from cachetools import cached, LRUCache
+import os
-from llama_index.core.tools.tool_spec.base import BaseToolSpec
+
+import yaml
 from llama_index.core.tools.function_tool import FunctionTool
+from llama_index.core.tools.tool_spec.base import BaseToolSpec
 
 
 class ToolType:
@@ -46,7 +45,7 @@ class ToolFactory:
     def from_env() -> list[FunctionTool]:
         tools = []
         if os.path.exists("config/tools.yaml"):
-            with open("config/tools.yaml", "r") as f:
+            with open("config/tools.yaml", "r", encoding='UTF-8') as f:
                 tool_configs = yaml.safe_load(f)
                 if tool_configs != None and len(tool_configs.items()) != 0:
                     for tool_type, config_entries in tool_configs.items():

backend/app/settings.py

@@ -3,18 +3,23 @@ from typing import Dict
 
 from llama_index.core.constants import DEFAULT_TEMPERATURE
 from llama_index.core.settings import Settings
+from app.xinference.base import XinferenceEmbedding, XinferenceRerank
 from llama_index.llms.xinference import Xinference
 from llama_index.llms.xinference.base import DEFAULT_XINFERENCE_TEMP
 
-from app.xinference.base import XinferenceEmbedding, XinferenceRerank
-
 
 def get_node_postprocessors():
+    rerank_enabled = os.getenv("RERANK_ENABLED").title()
+    if rerank_enabled is None or rerank_enabled == 'False':
+        return []
+
     rerank_model = os.getenv("RERANK_MODEL")
     rerank_url = os.getenv("RERANK_BASE_URL")
+    rerank_top_n = os.getenv("RERANK_TOP_N")
+    rerank_threshold = os.getenv("RERANK_THRESHOLD")
     postprocess = None
-    if rerank_model is None:
+    if rerank_model is not None:
-        postprocess = [XinferenceRerank(rerank_model, rerank_url)]
+        postprocess = [XinferenceRerank(rerank_model, rerank_url, top_n=rerank_top_n, threshold=rerank_threshold)]
     return postprocess
 
 def init_settings():
@@ -79,7 +84,7 @@ def init_xinference():
     embed_model_name = os.getenv("EMBEDDING_MODEL")
     dimensions = os.getenv("EMBEDDING_DIM")
     dimensions = int(dimensions) if dimensions is not None else None
-    Settings.embed_model = XinferenceEmbedding(embed_model_name, embedding_base_url)
+    Settings.embed_model = XinferenceEmbedding(embed_model_name, embedding_base_url, dimensions=dimensions)
 
 def init_openai():
     from llama_index.core.constants import DEFAULT_TEMPERATURE
@@ -226,4 +231,4 @@ def init_mistral():
     #
     # Settings.llm = MistralAI(model=os.getenv("MODEL"))
     # Settings.embed_model = MistralAIEmbedding(model_name=os.getenv("EMBEDDING_MODEL"))
-    pass
+    pass

backend/app/xinference/base.py

@@ -5,10 +5,11 @@ from enum import Enum
 from http import HTTPStatus
 from typing import Any, Dict, List, Optional, Union, Tuple
 
-from llama_index.core.base.embeddings.base import BaseEmbedding, Embedding
+from llama_index.core.base.embeddings.base import BaseEmbedding, Embedding, dispatcher
 from llama_index.core.bridge.pydantic import PrivateAttr
 from llama_index.core.callbacks import CBEventType, EventPayload
 from llama_index.core.embeddings.multi_modal_base import MultiModalEmbedding
+from llama_index.core.instrumentation.events.rerank import ReRankStartEvent, ReRankEndEvent
 from llama_index.core.postprocessor.types import BaseNodePostprocessor
 from llama_index.core.schema import ImageType, NodeWithScore, QueryBundle
 from pydantic import Field
@@ -51,7 +52,7 @@ class XinferenceEmbedding(BaseEmbedding):
             # num_workers: Optional[int] = None,
             **kwargs: Any,
     ) -> None:
-        generator, model_description = self.load_model(
+        generator, model_description, embed_batch_size, dimensions = self.load_model(
                 model_uid, endpoint
             )
         self._generator = generator
@@ -106,8 +107,11 @@ class XinferenceEmbedding(BaseEmbedding):
             )
 
         model = model_description["model_name"]
+        replica = model_description['replica']
+        dimensions = model_description['dimensions']
+        max_tokens = model_description['max_tokens']
 
-        return generator, model_description
+        return generator, model_description, replica, dimensions
 
     @classmethod
     def class_name(cls) -> str:
@@ -153,21 +157,27 @@ class XinferenceRerank(BaseNodePostprocessor):
     _generator: Any = PrivateAttr()
     _model_uid: str = Field(description="The Xinference model to use.")
     _endpoint: str = Field(description="The Xinference endpoint URL to use.")
-    #model: str = Field(description="Dashscope rerank model name.")
+    model: str = Field(description="Dashscope rerank model name.")
     top_n: int = Field(description="Top N nodes to return.")
+    threshold: float = Field(description="threshold nodes to return.")
 
     def __init__(
             self,
             model_uid: str,
             endpoint: str,
-            top_n: int = 3,
+            top_n: int = None,
+            threshold: float = None,
             return_documents: bool = False
     ):
+        _model_uid = model_uid
+        _endpoint = endpoint
+        _op_n = top_n
+        threshold = threshold
         generator, model_description = self.load_model(
             model_uid, endpoint
         )
         self._generator = generator
-        super().__init__(top_n=top_n, model=model_uid, return_documents=return_documents)
+        super().__init__(top_n=top_n, model=model_uid, model_uid=model_uid, threshold = threshold, return_documents=return_documents)
 
     @classmethod
     def class_name(cls) -> str:
@@ -183,6 +193,15 @@ class XinferenceRerank(BaseNodePostprocessor):
         if len(nodes) == 0:
             return []
 
+        dispatcher.event(
+            ReRankStartEvent(
+                nodes = nodes,
+                top_n = self.top_n,
+                query = query_bundle,
+                model_name = self.model
+            )
+        )
+
         with self.callback_manager.event(
                 CBEventType.RERANKING,
                 payload={
@@ -199,12 +218,22 @@ class XinferenceRerank(BaseNodePostprocessor):
                 new_node_with_score = NodeWithScore(
                     node=nodes[result['index']].node, score=result['relevance_score']
                 )
-                print(new_node_with_score.node.get_content)
+                if self.threshold is not None:
-                print('\n')
+                    if new_node_with_score.score >=self.threshold:
-                print(new_node_with_score.score)
+                        new_nodes.append(new_node_with_score)
-                new_nodes.append(new_node_with_score)
+
+                if self.top_n is not None:
+                    if len(new_nodes) > self.top_n:
+                        for index in new_nodes[self.top_n:-1]:
+                            new_nodes.remove(index)
+
             event.on_end(payload={EventPayload.NODES: new_nodes})
 
+        dispatcher.event(
+            ReRankEndEvent(
+                nodes= new_nodes
+            )
+        )
         return new_nodes
 
     def load_model(self, model_uid: str, endpoint: str) -> Tuple[Any, int, dict]:

backend/config/loaders.yaml

@@ -1,4 +1,5 @@
 file:
+  enable: true  # 添加 enable 字段
   # use_llama_parse: Use LlamaParse if `true`. Needs a `LLAMA_CLOUD_API_KEY` from https://cloud.llamaindex.ai set as environment variable
   use_llama_parse: false
 
@@ -7,14 +8,41 @@ db:
   # uri: The URI for the database. E.g.: mysql+pymysql://user:password@localhost:3306/db or postgresql+psycopg2://user:password@localhost:5432/db
   # query: The query to fetch data from the database. E.g.: SELECT * FROM table
   - uri: mysql+pymysql://zjinfo1:Dy2Bcr53Hm5xRkba@110.42.234.166:3306/zjinfo1
-  #- uri: mysql+pymysql://zjinfo:Y6EAjEEdSYmskA8B@110.42.234.166:3306/zjinfo
+    enable: true  # 添加 enable 字段
-#  - uri: mysql+pymysql://zjinfo2:GSKcziSdBixDXwcd@110.42.234.166:3306/zjinfo2
     queries:
-      - select * from ProjectProperties limit 30;
+      - sql: select * from ProjectProperties;
-      - select Name, Code, Amount, Amount_Total from TotalCalculateTable
+        explanation: "工程属性表数据，层级关系包含在博微电力造价工程文件格式_ProjectProperties.json文件中。"
-      - select SerialNumber, Name, Quantity, Rate, Sum_Price from ProjectDivision where Level = 1  limit 30;
-      - select Name, Code, Rate, Amount from OtherFee
 
+      - sql: select Id, ParentId, Level, Name, Code, Amount, Amount_Total from TotalCalculateTable;
+        explanation: "总算表数据，层级关系包含在博微电力造价工程文件格式_TotalCalculateTable.json文件中。"
+
+      - sql: select Id, ParentId, Level, SerialNumber, Name, Quantity, Rate, Sum_Price from ProjectDivision where ProfessionalType = '线路';
+        explanation: "专业类型为线路的项目划分表数据，层级关系包含在博微电力造价工程文件格式_ProjectDivision.json文件中。"
+      - sql: select Id, ParentId, Level, SerialNumber, Name, Quantity, Rate, Sum_Price from ProjectDivision where ProfessionalType = '余物清理';
+        explanation: "专业类型为余物清理的项目划分表数据，层级关系包含在博微电力造价工程文件格式_ProjectDivision.json文件中。"
+      - sql: select Id, ParentId, Level, SerialNumber, Name, Quantity, Rate, Sum_Price from ProjectDivision where ProfessionalType = '拆除线路';
+        explanation: "专业类型为拆除线路的项目划分表数据，层级关系包含在博微电力造价工程文件格式_ProjectDivision.json文件中。"
+
+      - sql: select Id, ParentId, Level, Name, Code, Rate, Amount from OtherFee;
+        explanation: "其他费用表数据，层级关系包含在博微电力造价工程文件格式_OtherFee.json文件中"
+
+      - sql: select Name, Code, Calculation_Formula, Rate, from FeeCollectionTable where FeeCollection_Table_Name = '线路取费表'
+        explanation: "取费表名称为线路取费表的取费表数据，层级关系包含在博微电力造价工程文件格式_FeeCollectionTable.json文件中"
+      - sql: select Name, Code, Calculation_Formula, Rate, from FeeCollectionTable where FeeCollection_Table_Name = '线路取费表(调试工程)aa'
+        explanation: "取费表名称为线路取费表的取费表数据，层级关系包含在博微电力造价工程文件格式_FeeCollectionTable.json文件中"
+      - sql: select Name, Code, Calculation_Formula, Rate, from FeeCollectionTable where FeeCollection_Table_Name = '大型土石方取费表'
+        explanation: "取费表名称为线路取费表的取费表数据，层级关系包含在博微电力造价工程文件格式_FeeCollectionTable.json文件中"
+      - sql: select Name, Code, Calculation_Formula, Rate, from FeeCollectionTable where FeeCollection_Table_Name = '线路取费表（余物清理）'
+        explanation: "取费表名称为线路取费表的取费表数据，层级关系包含在博微电力造价工程文件格式_FeeCollectionTable.json文件中"
+      - sql: select Name, Code, Calculation_Formula, Rate, from FeeCollectionTable where FeeCollection_Table_Name = '线路取费表（余物清理）(1)'
+        explanation: "取费表名称为线路取费表的取费表数据，层级关系包含在博微电力造价工程文件格式_FeeCollectionTable.json文件中"
+      - sql: select Name, Code, Calculation_Formula, Rate, from FeeCollectionTable where FeeCollection_Table_Name = '线路取费表（拆除）'
+        explanation: "取费表名称为线路取费表的取费表数据，层级关系包含在博微电力造价工程文件格式_FeeCollectionTable.json文件中"
+      
+      - sql: select Name, Code, Calculation_Formula, Rate, from ProjectQuantities where Professional_Type = '线路'
+        explanation: "专业类型为线路的工程量表数据，层级关系包含在博微电力造价工程文件格式_ProjectQuantities.json文件中"
+      - sql: select Name, Code, Calculation_Formula, Rate, from ProjectQuantities where Professional_Type = '余物清理'
+        explanation: "专业类型为余物清理的工程量表数据，层级关系包含在博微电力造价工程文件格式_ProjectQuantities.json文件中"
 #web:
 #  driver_arguments:
 #    # The arguments to pass to the webdriver. E.g.: add --headless to run in headless mode

backend/data/projectstruct/博微电力造价工程文件格式_FeeCollectionTable.json

@@ -4,56 +4,66 @@
             "name": "FeeCollectionTable",
             "alias": "",
             "comment": "取费表是取费设置中各取费表明细。查询示例: SELECT Rate FROM FeeCollectionTable WHERE Name = 'findname'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "FeeCollectionTableName",
-                    "alias": "取费名，取费名称，取费表名称",
+                    "alias": "取费表名称，取费名称，取费名",
-                    "comment": "取费表名称"
+                    "comment": "取费表名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Name",
-                    "alias": "费用名，名称，项目名",
+                    "alias": "名称，费用名，项目名",
-                    "comment": "费用名称，项目名称"
+                    "comment": "费用名称，项目名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "SerialNumber",
-                    "alias": "费用序号，序号，序列号",
+                    "alias": "序号，序列号，费用序号",
-                    "comment": "费用表序号"
+                    "comment": "费用表序号",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Code",
-                    "alias": "代码，代号，编号",
+                    "alias": "编号，代号，代码",
-                    "comment": "费用代码"
+                    "comment": "费用代码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "CalculationFormula",
-                    "alias": "表达式，公式，计算式",
+                    "alias": "公式，表达式，计算式",
-                    "comment": "取费基数"
+                    "comment": "取费基数",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Rate",
                     "alias": "费用利率，费率",
-                    "comment": "取费费率"
+                    "comment": "取费费率",
+                    "type": "REAL"
                 },
                 {
                     "name": "Remarks",
-                    "alias": "备注，说明",
+                    "alias": "说明，备注",
-                    "comment": "费用项备注说明"
+                    "comment": "费用项备注说明",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Major",
                     "alias": "专业",
-                    "comment": "取费表专业"
+                    "comment": "取费表专业",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Type",
                     "alias": "类型，取费类型",
-                    "comment": "取费表类型"
+                    "comment": "取费表类型",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Path",
-                    "alias": "费用全路径，路径",
+                    "alias": "路径，费用全路径",
-                    "comment": "费用项层级全路径"
+                    "comment": "费用项层级全路径",
+                    "type": "VARCHAR"
                 }
             ]
         }

backend/data/projectstruct/博微电力造价工程文件格式_OtherFee.json

@@ -4,71 +4,84 @@
             "name": "OtherFee",
             "alias": "",
             "comment": "其他费用表被称为“工程费用中其他费用明细”。其他费用是指为完成工程项目建设所必需的，但不属于建筑工程费、安装工程费、设备购置费、基本预备费的其他相关费用。包括建设场地征用及清理费、项目建设管理费、项目建设技术服务费、生产准备费、大件运输措施费、专业爆破服务费等。查询示例: SELECT Rate FROM OtherFee WHERE Name = 'findname'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "Id",
-                    "alias": "id，项目id，费用id",
+                    "alias": "项目id，id，费用id",
-                    "comment": "费用项目id"
+                    "comment": "费用项目id",
+                    "type": "INT"
                 },
                 {
                     "name": "ParentId",
-                    "alias": "父id，父级id",
+                    "alias": "父级id，父id",
-                    "comment": "费用项目父级id"
+                    "comment": "费用项目父级id",
+                    "type": "INT"
                 },
                 {
                     "name": "Level",
-                    "alias": "层级，层编号，层号",
+                    "alias": "层号，层级，层编号",
-                    "comment": "层级编号，从1开始"
+                    "comment": "层级编号，从1开始",
+                    "type": "INT"
                 },
                 {
                     "name": "Name",
-                    "alias": "费用名，名称，项目名",
+                    "alias": "名称，费用名，项目名",
-                    "comment": "费用名称，项目名称"
+                    "comment": "费用名称，项目名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "SerialNumber",
                     "alias": "序号，序列号",
-                    "comment": "费用表序号"
+                    "comment": "费用表序号",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Code",
-                    "alias": "代码，代号，编号",
+                    "alias": "编号，代号，代码",
-                    "comment": "费用代码"
+                    "comment": "费用代码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "CalculationFormula",
-                    "alias": "表达式，公式，计算式",
+                    "alias": "公式，表达式，计算式",
-                    "comment": "取费基数"
+                    "comment": "取费基数",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Rate",
                     "alias": "费用利率，费率",
-                    "comment": "取费费率"
+                    "comment": "取费费率",
+                    "type": "REAL"
                 },
                 {
                     "name": "Amount",
                     "alias": "金额，价格",
-                    "comment": "金额、合计、费用，\n单位为元"
+                    "comment": "金额、合计、费用，\n单位为元",
+                    "type": "REAL"
                 },
                 {
                     "name": "Remarks",
-                    "alias": "备注，说明",
+                    "alias": "说明，备注",
-                    "comment": "费用项备注说明"
+                    "comment": "费用项备注说明",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Compilation_Basis",
                     "alias": "编制依据，编制来源",
-                    "comment": "费用项编制依据"
+                    "comment": "费用项编制依据",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "WBS_Code",
                     "alias": "WBS编号，WBS编码",
-                    "comment": "费用项WBS编码"
+                    "comment": "费用项WBS编码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Path",
-                    "alias": "费用全路径，路径",
+                    "alias": "路径，费用全路径",
-                    "comment": "费用项层级全路径"
+                    "comment": "费用项层级全路径",
+                    "type": "VARCHAR"
                 }
             ]
         }

backend/data/projectstruct/博微电力造价工程文件格式_ProjectDivision.json

@@ -4,121 +4,144 @@
             "name": "ProjectDivision",
             "alias": "",
             "comment": "项目划分表是用于存储工程项目划分树状数据。内部包含安装工程项目划分，建筑工程项目划分，线路项目划分，工程分部分项。查询示例: SELECT Sum_Price FROM ProjectDivision WHERE Name = 'findname'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "Id",
-                    "alias": "id，项目id，费用id",
+                    "alias": "项目id，id，费用id",
-                    "comment": "项目划分id"
+                    "comment": "项目划分id",
+                    "type": "INT"
                 },
                 {
                     "name": "ParentId",
-                    "alias": "父id，父级id",
+                    "alias": "父级id，父id",
-                    "comment": "项目划分父级id"
+                    "comment": "项目划分父级id",
+                    "type": "INT"
                 },
                 {
                     "name": "Level",
-                    "alias": "层级，层编号，层号",
+                    "alias": "层号，层级，层编号",
-                    "comment": "层级编号，从1开始"
+                    "comment": "层级编号，从1开始",
+                    "type": "INT"
                 },
                 {
                     "name": "Quantity",
                     "alias": "个数，数量，数目",
-                    "comment": "项目划分数量"
+                    "comment": "项目划分数量",
+                    "type": "REAL"
                 },
                 {
                     "name": "SerialNumber",
-                    "alias": "项目序号，序号，序列号",
+                    "alias": "序号，序列号，项目序号",
-                    "comment": "项目划分序号"
+                    "comment": "项目划分序号",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Name",
-                    "alias": "项目名，名称",
+                    "alias": "名称，项目名",
-                    "comment": "项目名称"
+                    "comment": "项目名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Encoding",
-                    "alias": "译码，编码",
+                    "alias": "编码，译码",
-                    "comment": "项目划分编码"
+                    "comment": "项目划分编码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Sum_Price",
                     "alias": "合计，合价",
-                    "comment": "项目划分合价，分部分项费用"
+                    "comment": "项目划分合价，分部分项费用",
+                    "type": "REAL"
                 },
                 {
                     "name": "FeeCollectionTableName",
                     "alias": "取费表",
-                    "comment": "项目划分的取费表，此项目划分选用的取费表"
+                    "comment": "项目划分的取费表，此项目划分选用的取费表",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Remarks",
-                    "alias": "备注，说明",
+                    "alias": "说明，备注",
-                    "comment": "备注"
+                    "comment": "备注",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "WBS_Code",
                     "alias": "WBS编号，WBS编码",
-                    "comment": "WBS编码"
+                    "comment": "WBS编码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Manual_Adjustment_Coefficient",
                     "alias": "人工调差系数",
-                    "comment": "此项目划分下人工调差系数"
+                    "comment": "此项目划分下人工调差系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "Material_Adjustment_Coefficient",
                     "alias": "材料调差系数",
-                    "comment": "此项目划分下材料调差系数"
+                    "comment": "此项目划分下材料调差系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "Mechanical_Adjustment_Coefficient",
                     "alias": "机械调差系数",
-                    "comment": "此项目划分下机械调差系数"
+                    "comment": "此项目划分下机械调差系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "Demolition_Manual_Adjustment_Coefficient",
                     "alias": "拆除人工调差系数",
-                    "comment": "此项目划分下拆除人工调差系数"
+                    "comment": "此项目划分下拆除人工调差系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "Demolition_Material_Adjustment_Coefficient",
                     "alias": "拆除材料调差系数",
-                    "comment": "此项目划分下拆除材料调差系数"
+                    "comment": "此项目划分下拆除材料调差系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "Demolition_Mechanical_Adjustment_Coefficient",
                     "alias": "拆除机械调差系数",
-                    "comment": "此项目划分下拆除机械调差系数"
+                    "comment": "此项目划分下拆除机械调差系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "ProfessionalType",
                     "alias": "专业类型",
-                    "comment": "专业类型，字段值有变电安装、变电建筑、线路等。变电安装等于安装工程，变电建筑等于建筑工程，线路等于安装工程。"
+                    "comment": "专业类型，字段值有变电安装、变电建筑、线路等。变电安装等于安装工程，变电建筑等于建筑工程，线路等于安装工程。",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Unit",
                     "alias": "单位",
-                    "comment": "项目划分单位"
+                    "comment": "项目划分单位",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "CalculationFormula",
-                    "alias": "表达式，公式，计算式",
+                    "alias": "公式，表达式，计算式",
-                    "comment": "项目划分计算式"
+                    "comment": "项目划分计算式",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Rate",
                     "alias": "费用利率，费率",
-                    "comment": "项目划分费率"
+                    "comment": "项目划分费率",
+                    "type": "REAL"
                 },
                 {
                     "name": "Code",
-                    "alias": "代码，代号，编号",
+                    "alias": "编号，代号，代码",
-                    "comment": "项目划分代码"
+                    "comment": "项目划分代码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Path",
                     "alias": "路径，项目全路径",
-                    "comment": "项目划分层级全路径"
+                    "comment": "项目划分层级全路径",
+                    "type": "VARCHAR"
                 }
             ]
         }

backend/data/projectstruct/博微电力造价工程文件格式_ProjectDivisions_CostPreview.json

@@ -4,196 +4,234 @@
             "name": "ProjectDivisions_CostPreview",
             "alias": "",
             "comment": "项目划分_费用预览表也被称为“项目划分费用预览”、“项目划分取费费用”。其中包含项目划分合价、直接费、间接费、利润、税金、主材费等。查询示例: SELECT Total FROM ProjectDivisions_CostPreview WHERE Id = '15'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "Id",
                     "alias": "id，项目id",
-                    "comment": "项目划分id"
+                    "comment": "项目划分id",
+                    "type": "INT"
                 },
                 {
                     "name": "ParentId",
-                    "alias": "父id，父级id",
+                    "alias": "父级id，父id",
-                    "comment": "项目划分父级id"
+                    "comment": "项目划分父级id",
+                    "type": "INT"
                 },
                 {
                     "name": "Level",
-                    "alias": "层级，层编号，层号",
+                    "alias": "层号，层级，层编号",
-                    "comment": "层级编号，从1开始"
+                    "comment": "层级编号，从1开始",
+                    "type": "INT"
                 },
                 {
                     "name": "ProfessionalType",
                     "alias": "专业类型",
-                    "comment": "专业类型，字段值有变电安装、变电建筑、线路等。变电安装等于安装工程，变电建筑等于建筑工程，线路等于安装工程。"
+                    "comment": "专业类型，字段值有变电安装、变电建筑、线路等。变电安装等于安装工程，变电建筑等于建筑工程，线路等于安装工程。",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "FeeCollectionTableName",
                     "alias": "取费表",
-                    "comment": "项目划分的取费表，此项目划分选用的取费表"
+                    "comment": "项目划分的取费表，此项目划分选用的取费表",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Direct_Cost",
                     "alias": "直接费",
-                    "comment": "直接费是指施工过程中直接耗用于建筑、安装工程产品的各项费用的总和。包括直接工程费和措施费。"
+                    "comment": "直接费是指施工过程中直接耗用于建筑、安装工程产品的各项费用的总和。包括直接工程费和措施费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Direct_Project_Cost",
                     "alias": "直接工程费",
-                    "comment": "直接工程费是指按照正常的施工条件，在施工过程中耗费的构成工程实体的各项费用。包括人工费、材料费和施工机械使用费。"
+                    "comment": "直接工程费是指按照正常的施工条件，在施工过程中耗费的构成工程实体的各项费用。包括人工费、材料费和施工机械使用费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Quota_Direct_Cost",
                     "alias": "定额直接费",
-                    "comment": "定额直接费，包含人工费、材料费中已进入定额基价的消耗性材料费和施工机械使用费。"
+                    "comment": "定额直接费，包含人工费、材料费中已进入定额基价的消耗性材料费和施工机械使用费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Labor_Cost",
                     "alias": "人工费",
-                    "comment": "人工费是指支付给直接从事建筑安装工程施工作业的生产人员的各项费用。包括基本工资、工资性补贴、辅助工资、职工福利费、生产人员劳动保护费。"
+                    "comment": "人工费是指支付给直接从事建筑安装工程施工作业的生产人员的各项费用。包括基本工资、工资性补贴、辅助工资、职工福利费、生产人员劳动保护费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Material_Cost",
                     "alias": "材料费",
-                    "comment": "材料费是指施工过程中一次性消耗材料及摊销材料的费用。指已进入定额基价的消耗性材料费。"
+                    "comment": "材料费是指施工过程中一次性消耗材料及摊销材料的费用。指已进入定额基价的消耗性材料费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Construction_Machinery_Cost",
                     "alias": "施工机械使用费",
-                    "comment": "施工机械使用费是指施工机械作业所发生的机械使用费以及机械的现场安拆费和场外运费。包括折旧费、检修费、维护费、安装及拆卸费、场外运费、操作人员人工费、燃料动力费、其他费等。"
+                    "comment": "施工机械使用费是指施工机械作业所发生的机械使用费以及机械的现场安拆费和场外运费。包括折旧费、检修费、维护费、安装及拆卸费、场外运费、操作人员人工费、燃料动力费、其他费等。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Installation_Material_Cost",
                     "alias": "装置性材料费",
-                    "comment": "装置性材料费是指建设工程中构成工艺系统实体的工艺性材料，也称主要材料费。装置性材料通常在概算或预算定额中未计价，也称未计价材料，也称主材。"
+                    "comment": "装置性材料费是指建设工程中构成工艺系统实体的工艺性材料，也称主要材料费。装置性材料通常在概算或预算定额中未计价，也称未计价材料，也称主材。",
+                    "type": "REAL"
                 },
                 {
                     "name": "A_Supply_Installation_Material_Cost",
                     "alias": "甲供装置性材料费",
-                    "comment": "供货方为甲供的装置性材料费。"
+                    "comment": "供货方为甲供的装置性材料费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "B_Supply_Installation_Material_Cost",
                     "alias": "乙供装置性材料费",
-                    "comment": "供货方为乙供的装置性材料费。"
+                    "comment": "供货方为乙供的装置性材料费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Measure_Cost",
                     "alias": "措施费",
-                    "comment": "措施费是指为完成工程项目施工而进行施工准备、克服自然条件的不利影响和辅助施工所发生的不构成工程实体的各项费用。包括冬雨季施工增加费、夜间施工增加费、施工工具用具使用费、特殊地区施工增加费、临时设施费、施工机构迁移费、安全文明施工费。"
+                    "comment": "措施费是指为完成工程项目施工而进行施工准备、克服自然条件的不利影响和辅助施工所发生的不构成工程实体的各项费用。包括冬雨季施工增加费、夜间施工增加费、施工工具用具使用费、特殊地区施工增加费、临时设施费、施工机构迁移费、安全文明施工费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "WinterRainySeasons_Additional_Construction_Cost",
                     "alias": "冬雨季施工增加费",
-                    "comment": "冬雨季施工增加费是指按照合理的工期要求，建筑、安装工程必须在冬季、雨季期间连续施工而需要增加的费用。"
+                    "comment": "冬雨季施工增加费是指按照合理的工期要求，建筑、安装工程必须在冬季、雨季期间连续施工而需要增加的费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Night_Additional_Construction_Cost",
                     "alias": "夜间施工增加费",
-                    "comment": "夜间施工增加费是指按照规程要求，工程必须在夜间连续施工所发生的夜班补助、夜间施工降效、夜间施工照明设备摊销及照明用电等费用。"
+                    "comment": "夜间施工增加费是指按照规程要求，工程必须在夜间连续施工所发生的夜班补助、夜间施工降效、夜间施工照明设备摊销及照明用电等费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Construction_Tool_Usage_Cost",
                     "alias": "施工工具用具使用费",
-                    "comment": "施工工具用具使用费是指施工企业的生产、检验、试验部门使用的不属于固定资产的工具用具和仪器仪表的购置、摊销和维护费用。"
+                    "comment": "施工工具用具使用费是指施工企业的生产、检验、试验部门使用的不属于固定资产的工具用具和仪器仪表的购置、摊销和维护费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Special_Areas_Additional_Construction_Cost",
                     "alias": "特殊地区施工增加费",
-                    "comment": "特殊地区施工增加费是指在高海拔、酷热、严寒等地区施工:因特殊自然条件影响而需额外增加的施工费用。"
+                    "comment": "特殊地区施工增加费是指在高海拔、酷热、严寒等地区施工:因特殊自然条件影响而需额外增加的施工费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Temporary_Facility_Cost",
                     "alias": "临时设施费",
-                    "comment": "临时设施费是指施工企业为满足现场正常生产、生活需要在现场必须搭设的生产、生活用临时建筑物、构筑物和其他临时设施所发生的费用，以及维修、拆除、折旧及摊销费，或临时设施的租赁费等。"
+                    "comment": "临时设施费是指施工企业为满足现场正常生产、生活需要在现场必须搭设的生产、生活用临时建筑物、构筑物和其他临时设施所发生的费用，以及维修、拆除、折旧及摊销费，或临时设施的租赁费等。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Construction_Organization_Relocation_Cost",
                     "alias": "施工机构迁移费",
-                    "comment": "施工机构迁移费是指施工企业派遣施工队伍到所承建工程现场所发生的搬迁费用。包括职工调遣差旅费和调遣期间的工资，以及办公设备、工器具、家具、材料用品和施工机械等的搬迁费用。"
+                    "comment": "施工机构迁移费是指施工企业派遣施工队伍到所承建工程现场所发生的搬迁费用。包括职工调遣差旅费和调遣期间的工资，以及办公设备、工器具、家具、材料用品和施工机械等的搬迁费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Safe_Civilized_Construction_Cost",
                     "alias": "安全文明施工费",
-                    "comment": "安全文明施工费，包括安全生产费、文明施工费、环境保护费。"
+                    "comment": "安全文明施工费，包括安全生产费、文明施工费、环境保护费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Indirect_Cost",
                     "alias": "间接费",
-                    "comment": "间接费是指建筑安装工程的施工过程中，为全工程项目服务而不直接消耗在特定产品对象上的费用。包括规费、企业管理费和施工企业配合调试费。"
+                    "comment": "间接费是指建筑安装工程的施工过程中，为全工程项目服务而不直接消耗在特定产品对象上的费用。包括规费、企业管理费和施工企业配合调试费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Regulatory_Cost",
                     "alias": "规费",
-                    "comment": "规费是指按照国家行政主管部门或省级政府和省级有关权力部门规定必须缴纳并计入建筑安装工程造价的费用。包括社会保险费和住房公积金。"
+                    "comment": "规费是指按照国家行政主管部门或省级政府和省级有关权力部门规定必须缴纳并计入建筑安装工程造价的费用。包括社会保险费和住房公积金。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Social_Insurance_Premiums",
                     "alias": "社会保险费",
-                    "comment": "社会保险费包括养老保险费、失业保险费、医疗保险费、生育保险费和工伤保险费。"
+                    "comment": "社会保险费包括养老保险费、失业保险费、医疗保险费、生育保险费和工伤保险费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Housing_Provident_Fund",
                     "alias": "住房公积金",
-                    "comment": "住房公积金是指企业按照规定标准为职工缴纳的住房公积金。"
+                    "comment": "住房公积金是指企业按照规定标准为职工缴纳的住房公积金。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Enterprise_Management_Cost",
                     "alias": "企业管理费",
-                    "comment": "企业管理费是指建筑安装施工企业为组织施工生产和经营管理所发生的费用。"
+                    "comment": "企业管理费是指建筑安装施工企业为组织施工生产和经营管理所发生的费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Construction_Enterprise_Cooperation_Debugging_Cost",
                     "alias": "施工企业配合调试费",
-                    "comment": "施工企业配合调试费是指在工程整套启动试运阶段，施工企业安装专业配合调试所发生的费用。"
+                    "comment": "施工企业配合调试费是指在工程整套启动试运阶段，施工企业安装专业配合调试所发生的费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Profit",
                     "alias": "利润",
-                    "comment": "利润是指施工企业完成所承包工程获得的盈利。"
+                    "comment": "利润是指施工企业完成所承包工程获得的盈利。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Taxes",
                     "alias": "税金",
-                    "comment": "税金是指按照国家税法规定应计入建筑安装工程造价内的销项税额。"
+                    "comment": "税金是指按照国家税法规定应计入建筑安装工程造价内的销项税额。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Equipment_Cost",
                     "alias": "设备费",
-                    "comment": "设备购置费是指为项目建设而购置或自制各种设备，并将设备运至施工现场指定位置所支出的费用。包括设备费和设备运杂费。"
+                    "comment": "设备购置费是指为项目建设而购置或自制各种设备，并将设备运至施工现场指定位置所支出的费用。包括设备费和设备运杂费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "B_Supply_Equipment_Excluding_Tax_Price",
                     "alias": "乙供设备不含税价",
-                    "comment": "设备费中，供货方为乙供设备，不含税价"
+                    "comment": "设备费中，供货方为乙供设备，不含税价",
+                    "type": "REAL"
                 },
                 {
                     "name": "A_Supply_Equipment_Tax_Price",
                     "alias": "甲供设备含税价",
-                    "comment": "设备费中，供货方为甲供设备，含税价"
+                    "comment": "设备费中，供货方为甲供设备，含税价",
+                    "type": "REAL"
                 },
                 {
                     "name": "Installation_Cost",
                     "alias": "安装费",
-                    "comment": "安装费包含定额直接费、措施费、间接费、利润、税金和一笔性费用。"
+                    "comment": "安装费包含定额直接费、措施费、间接费、利润、税金和一笔性费用。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Main_Material_Cost",
                     "alias": "主材费",
-                    "comment": "主材费指装置性材料费"
+                    "comment": "主材费指装置性材料费",
+                    "type": "REAL"
                 },
                 {
                     "name": "Total",
-                    "alias": "总体费用，总计，总价，总的费用",
+                    "alias": "总价，总计，总体费用，总的费用",
-                    "comment": "总计包含安装费、主材费、设备费。"
+                    "comment": "总计包含安装费、主材费、设备费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Sum",
                     "alias": "合计，合价",
-                    "comment": "项目划分合价，分部分项费用，项目划分费用。合计包含安装费和主材费。"
+                    "comment": "项目划分合价，分部分项费用，项目划分费用。合计包含安装费和主材费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Path",
                     "alias": "路径，项目划分全路径",
-                    "comment": "项目划分层级全路径"
+                    "comment": "项目划分层级全路径",
+                    "type": "VARCHAR"
                 }
             ]
         }

backend/data/projectstruct/博微电力造价工程文件格式_ProjectProperties.json

@@ -4,26 +4,30 @@
             "name": "ProjectProperties",
             "alias": "",
             "comment": "工程属性表是用于存储整个工程的重要属性，访问该表都是为了通过属性名查找属性值。通常属性值有工程信息、工程属性、技经参数，表中包含工程总投资、工程总费用，工程主要费用，工程技经参数等。查询示例: SELECT Value FROM ProjectProperties WHERE Name = 'findname'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "Name\n",
-                    "alias": "属性名，属性名称，属性",
+                    "alias": "名称、属性、属性名称、字段、字段名称、变量、参数，属性名",
-                    "comment": "属性的唯一标识"
+                    "comment": "属性的唯一标识",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Value",
-                    "alias": "属性值",
+                    "alias": "值、变量值、参数值、数值，属性值",
-                    "comment": "属性对应的实际值"
+                    "comment": "属性对应的实际值",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Type",
-                    "alias": "类型，属性类型",
+                    "alias": "类型、变量类型、数值类型，属性类型",
-                    "comment": "属性变量的类型"
+                    "comment": "属性变量的类型",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Unit",
                     "alias": "单位",
-                    "comment": "单位"
+                    "comment": "单位",
+                    "type": "VARCHAR"
                 }
             ]
         }

backend/data/projectstruct/博微电力造价工程文件格式_ProjectQuantities.json

@@ -4,276 +4,330 @@
             "name": "ProjectQuantities",
             "alias": "",
             "comment": "工程量表是项目划分下工程量，包含定额、主材、设备、一笔性费用。查询示例: SELECT BudgetPrice FROM ProjectQuantities WHERE Name = 'findname'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "Id",
                     "alias": "id",
-                    "comment": "消耗量id，工程量id"
+                    "comment": "消耗量id，工程量id",
+                    "type": "INT"
                 },
                 {
                     "name": "ParentId",
-                    "alias": "父id，父级id",
+                    "alias": "父级id，父id",
-                    "comment": "父级id"
+                    "comment": "父级id",
+                    "type": "INT"
                 },
                 {
                     "name": "ProjectDivisionId",
                     "alias": "项目划分id，项目id",
-                    "comment": "父级项目划分id"
+                    "comment": "父级项目划分id",
+                    "type": "INT"
                 },
                 {
                     "name": "Quantity",
                     "alias": "个数，数量，数目",
-                    "comment": "数量，消耗量数量，工程量数量，主材数量，定额数量，设备数量，项目划分单位"
+                    "comment": "数量，消耗量数量，工程量数量，主材数量，定额数量，设备数量，项目划分单位",
+                    "type": "REAL"
                 },
                 {
                     "name": "FeatureSegment",
                     "alias": "特征段",
-                    "comment": "线路特征段"
+                    "comment": "线路特征段",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "ParentQuantity",
                     "alias": "父级个数，父级数量",
-                    "comment": "父级id的数量"
+                    "comment": "父级id的数量",
+                    "type": "REAL"
                 },
                 {
                     "name": "Name",
                     "alias": "名称",
-                    "comment": "项目名称，工程量名称，消耗量名称，主材名称，定额名称，设备名称，材料名称"
+                    "comment": "项目名称，工程量名称，消耗量名称，主材名称，定额名称，设备名称，材料名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Encoding",
-                    "alias": "译码，编码",
+                    "alias": "编码，译码",
-                    "comment": "编码，定额编码，主材编码，设备编码"
+                    "comment": "编码，定额编码，主材编码，设备编码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "SpecificationModel",
                     "alias": "规格型号",
-                    "comment": "规格型号，主材规格型号，设备规格型号"
+                    "comment": "规格型号，主材规格型号，设备规格型号",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Unit",
                     "alias": "单位",
-                    "comment": "单位，主材单位，定额单位，设备单位，项目划分单位"
+                    "comment": "单位，主材单位，定额单位，设备单位，项目划分单位",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "BasePrice",
                     "alias": "基价",
-                    "comment": "定额基价"
+                    "comment": "定额基价",
+                    "type": "REAL"
                 },
                 {
                     "name": "LaborCost",
                     "alias": "人工费",
-                    "comment": "定额人工费"
+                    "comment": "定额人工费",
+                    "type": "REAL"
                 },
                 {
                     "name": "MaterialCost",
                     "alias": "材料费",
-                    "comment": "定额材料费"
+                    "comment": "定额材料费",
+                    "type": "REAL"
                 },
                 {
                     "name": "MachineryCost",
                     "alias": "机械费",
-                    "comment": "定额机械费"
+                    "comment": "定额机械费",
+                    "type": "REAL"
                 },
                 {
                     "name": "QuotaCoefficient",
                     "alias": "定额系数",
-                    "comment": "定额系数"
+                    "comment": "定额系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "LaborCoefficient",
                     "alias": "人工系数",
-                    "comment": "定额人工系数"
+                    "comment": "定额人工系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "MaterialCoefficient",
                     "alias": "材料系数",
-                    "comment": "定额材料系数"
+                    "comment": "定额材料系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "MechanicalCoefficient",
                     "alias": "机械系数",
-                    "comment": "定额机械系数"
+                    "comment": "定额机械系数",
+                    "type": "REAL"
                 },
                 {
                     "name": "ExpenseType",
                     "alias": "费用类型",
-                    "comment": "费用类型，取值为取费、不取费"
+                    "comment": "费用类型，取值为取费、不取费",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "BudgetPrice",
                     "alias": "预算价",
-                    "comment": "预算价"
+                    "comment": "预算价",
+                    "type": "REAL"
                 },
                 {
                     "name": "MarketPrice",
                     "alias": "市场价",
-                    "comment": "间接费是指建筑安装工程的施工过程中，为全工程项目服务而不直接消耗在特定产品对象上的费用。包括规费、企业管理费和施工企业配合调试费。"
+                    "comment": "间接费是指建筑安装工程的施工过程中，为全工程项目服务而不直接消耗在特定产品对象上的费用。包括规费、企业管理费和施工企业配合调试费。",
+                    "type": "REAL"
                 },
                 {
                     "name": "Supplier",
                     "alias": "供货方",
-                    "comment": "供货方，设备供货方，主材供货方，取值为甲供、乙供"
+                    "comment": "供货方，设备供货方，主材供货方，取值为甲供、乙供",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Type",
                     "alias": "类型",
-                    "comment": "工程量类型，取值定额、主材、设备、一笔性费用"
+                    "comment": "工程量类型，取值定额、主材、设备、一笔性费用",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "QuotaRange",
                     "alias": "定额范围",
-                    "comment": "定额范围，取值概算、预算"
+                    "comment": "定额范围，取值概算、预算",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "A_Supply_Material_Cost_Excluding_Tax",
                     "alias": "甲供材料费不含税",
-                    "comment": "甲供材料费不含税"
+                    "comment": "甲供材料费不含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "A_Supply_Material_Cost_Including_Tax",
                     "alias": "甲供材料费含税",
-                    "comment": "甲供材料费含税"
+                    "comment": "甲供材料费含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "B_Supply_Material_Cost_Excluding_Tax",
                     "alias": "乙供材料费不含税",
-                    "comment": "乙供材料费不含税"
+                    "comment": "乙供材料费不含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "B_Supply_Material_Cost_Including_Tax",
                     "alias": "乙供材料费含税",
-                    "comment": "乙供材料费含税"
+                    "comment": "乙供材料费含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "ScaffoldCalculation",
                     "alias": "脚手架计取",
-                    "comment": "脚手架计取，取值计取、不计取"
+                    "comment": "脚手架计取，取值计取、不计取",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Remarks",
-                    "alias": "备注，说明",
+                    "alias": "说明，备注",
-                    "comment": "备注，说明"
+                    "comment": "备注，说明",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "FeeCollectionTableName",
                     "alias": "取费表",
-                    "comment": "项目划分的取费表，工程量的取费表"
+                    "comment": "项目划分的取费表，工程量的取费表",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Quota_Section_Name",
                     "alias": "定额章节名称",
-                    "comment": "定额章节名称"
+                    "comment": "定额章节名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "ProfessionalType",
                     "alias": "专业类型",
-                    "comment": "专业类型，字段值有变电安装、变电建筑、线路等。变电安装等于安装工程，变电建筑等于建筑工程，线路等于安装工程。"
+                    "comment": "专业类型，字段值有变电安装、变电建筑、线路等。变电安装等于安装工程，变电建筑等于建筑工程，线路等于安装工程。",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "split",
                     "alias": "拆分",
-                    "comment": "是否为拆分材料，取值1为拆分，取值0为不拆分"
+                    "comment": "是否为拆分材料，取值1为拆分，取值0为不拆分",
+                    "type": "INT"
                 },
                 {
                     "name": "Loss",
                     "alias": "损耗",
-                    "comment": "损耗率，主材损耗率"
+                    "comment": "损耗率，主材损耗率",
+                    "type": "REAL"
                 },
                 {
                     "name": "SingleWeight",
                     "alias": "单重",
-                    "comment": "单重，主材单重"
+                    "comment": "单重，主材单重",
+                    "type": "REAL"
                 },
                 {
                     "name": "LineWeight",
                     "alias": "线重",
-                    "comment": "线重，主材线重"
+                    "comment": "线重，主材线重",
+                    "type": "REAL"
                 },
                 {
                     "name": "SupervisedMaterials",
                     "alias": "监造物料",
-                    "comment": "监造物料，取值1为监造物料，取值0为非监造物料"
+                    "comment": "监造物料，取值1为监造物料，取值0为非监造物料",
+                    "type": "INT"
                 },
                 {
                     "name": "EquipmentMaterials",
                     "alias": "设备性材料",
-                    "comment": "设备性材料，取值1为设备性材料，取值0为主材"
+                    "comment": "设备性材料，取值1为设备性材料，取值0为主材",
+                    "type": "INT"
                 },
                 {
                     "name": "GrossWeight",
                     "alias": "毛重",
-                    "comment": "毛重，主材毛重"
+                    "comment": "毛重，主材毛重",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "TransportationType",
                     "alias": "运输类型",
-                    "comment": "运输类型，主材运输类型"
+                    "comment": "运输类型，主材运输类型",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "TransportationMiscellaneous",
                     "alias": "运杂费率",
-                    "comment": "运杂费率，设备运杂费率"
+                    "comment": "运杂费率，设备运杂费率",
+                    "type": "REAL"
                 },
                 {
                     "name": "EquipmentType",
                     "alias": "设备类型",
-                    "comment": "设备类型，取值为主要设备、普通设备"
+                    "comment": "设备类型，取值为主要设备、普通设备",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "UnitPrice",
                     "alias": "单价",
-                    "comment": "单价"
+                    "comment": "单价",
+                    "type": "REAL"
                 },
                 {
                     "name": "Market_Price_Excluding_Tax",
                     "alias": "市场价不含税",
-                    "comment": "市场价不含税"
+                    "comment": "市场价不含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "Market_Price_Including_Tax",
                     "alias": "市场价含税",
-                    "comment": "市场价含税，设备含税价"
+                    "comment": "市场价含税，设备含税价",
+                    "type": "REAL"
                 },
                 {
                     "name": "Budget_Price_Excluding_Tax",
                     "alias": "预算价不含税",
-                    "comment": "预算价不含税"
+                    "comment": "预算价不含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "Budget_Price_Including_Tax",
                     "alias": "预算价含税",
-                    "comment": "预算价含税"
+                    "comment": "预算价含税",
+                    "type": "REAL"
                 },
                 {
                     "name": "Unit_Price_Excluding_Tax",
                     "alias": "单价不含税",
-                    "comment": "单价不含税，设备不含税价"
+                    "comment": "单价不含税，设备不含税价",
+                    "type": "REAL"
                 },
                 {
                     "name": "GroupPrice",
                     "alias": "分组合价",
-                    "comment": "分组合价"
+                    "comment": "分组合价",
+                    "type": "REAL"
                 },
                 {
                     "name": "Pump_Truck_Pouring",
                     "alias": "泵车浇制",
-                    "comment": "泵车浇制，取值1为泵车浇制，取值0为非泵车浇制"
+                    "comment": "泵车浇制，取值1为泵车浇制，取值0为非泵车浇制",
+                    "type": "INT"
                 },
                 {
                     "name": "On_Site_Preparation",
                     "alias": "现场制备",
-                    "comment": "现场制备，取值1为现场制备，取值0为非现场制备"
+                    "comment": "现场制备，取值1为现场制备，取值0为非现场制备",
+                    "type": "INT"
                 },
                 {
                     "name": "Clear_Water_Concrete",
                     "alias": "清水混凝土",
-                    "comment": "清水混凝土，取值1为清水混凝土，取值0为非清水混凝土"
+                    "comment": "清水混凝土，取值1为清水混凝土，取值0为非清水混凝土",
+                    "type": "INT"
                 },
                 {
                     "name": "Debugging_Fee_Calculation",
                     "alias": "调试费计取",
-                    "comment": "调试费计取，取值计取、不计取"
+                    "comment": "调试费计取，取值计取、不计取",
+                    "type": "VARCHAR"
                 }
             ]
         }

backend/data/projectstruct/博微电力造价工程文件格式_TotalCalculateTable.json

@@ -2,83 +2,98 @@
     "Table": [
         {
             "name": "TotalCalculateTable",
-            "alias": "总算表",
+            "alias": "",
             "comment": "总算表也被称为“工程总费用”、“工程费用”。其中包含本地工程、辅助设施工程、编制基准期价差、设备购置费、其他费用、基本预备费、特殊费用、工程静态投资、动态费用、价差预备费、建设期贷款利息、工程动态投资、可抵扣增值税额。查询示例: SELECT Amount FROM TotalCalculateTable WHERE Name = 'findname'。",
-            "fileds": [
+            "fields": [
                 {
                     "name": "Id",
-                    "alias": "id，项目id，费用id",
+                    "alias": "项目id，id，费用id",
-                    "comment": "费用项目id"
+                    "comment": "费用项目id",
+                    "type": "INT"
                 },
                 {
                     "name": "ParentId",
-                    "alias": "父id，父级id",
+                    "alias": "父级id，父id",
-                    "comment": "费用项目父级id"
+                    "comment": "费用项目父级id",
+                    "type": "INT"
                 },
                 {
                     "name": "Level",
-                    "alias": "层级，层编号，层号",
+                    "alias": "层号，层级，层编号",
-                    "comment": "层级编号，从1开始"
+                    "comment": "层级编号，从1开始",
+                    "type": "INT"
                 },
                 {
                     "name": "Name",
-                    "alias": "费用名，名称，项目名",
+                    "alias": "名称，费用名，项目名",
-                    "comment": "费用名称，项目名称"
+                    "comment": "费用名称，项目名称",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "SerialNumber",
                     "alias": "序号",
-                    "comment": "工程费用序号"
+                    "comment": "工程费用序号",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Code",
-                    "alias": "代码，代号，编号",
+                    "alias": "编号，代号，代码",
-                    "comment": "费用代码"
+                    "comment": "费用代码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Rate",
                     "alias": "费用利率，费率",
-                    "comment": "费率"
+                    "comment": "费率",
+                    "type": "REAL"
                 },
                 {
                     "name": "Amount",
                     "alias": "金额，价格",
-                    "comment": "合计费"
+                    "comment": "合计费",
+                    "type": "REAL"
                 },
                 {
                     "name": "WBS_Code",
                     "alias": "WBS编号，WBS编码",
-                    "comment": "费用编码"
+                    "comment": "费用编码",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Path",
-                    "alias": "费用全路径，路径",
+                    "alias": "路径，费用全路径",
-                    "comment": "费用名称全路径"
+                    "comment": "费用名称全路径",
+                    "type": "VARCHAR"
                 },
                 {
                     "name": "Amount_InstallationCost",
-                    "alias": "安装价格，安装金额，金额_安装费",
+                    "alias": "安装金额，金额_安装费，安装价格",
-                    "comment": "安装费金额"
+                    "comment": "安装费金额",
+                    "type": "REAL"
                 },
                 {
                     "name": "Amount_EquipmentCost",
-                    "alias": "设备价格，金额_设备费，设备金额",
+                    "alias": "金额_设备费，设备金额，设备价格",
-                    "comment": "设备费金额"
+                    "comment": "设备费金额",
+                    "type": "REAL"
                 },
                 {
                     "name": "Amount_OtherCost",
-                    "alias": "其他费用价格，其他费用金额，金额_其他费",
+                    "alias": "其他费用金额，金额_其他费，其他费用价格",
-                    "comment": "其他费金额"
+                    "comment": "其他费金额",
+                    "type": "REAL"
                 },
                 {
                     "name": "Amount_Total",
                     "alias": "总的金额，金额_占总计，总体金额",
-                    "comment": "合计费占总计"
+                    "comment": "合计费占总计",
+                    "type": "REAL"
                 },
                 {
                     "name": "Amount_UnitInvestment",
-                    "alias": "合计投资金额，金额_单位投资",
+                    "alias": "金额_单位投资，合计投资金额",
-                    "comment": "合计费单位投资"
+                    "comment": "合计费单位投资",
+                    "type": "REAL"
                 }
             ]
         }

backend/main.py

@@ -1,6 +1,5 @@
-from dotenv import load_dotenv
-from llama_index.core.node_parser import SentenceSplitter
 
+from dotenv import load_dotenv
 load_dotenv()
 
 import logging
@@ -11,16 +10,20 @@ from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import RedirectResponse
 from app.api.routers.chat import chat_router
 from app.api.routers.upload import file_upload_router
+from app.api.routers.app import v1_router
 from app.settings import init_settings
 from app.observability import init_observability
 from fastapi.staticfiles import StaticFiles
 from phoenix.trace import using_project
 
+
 logger = logging.getLogger("uvicorn")
 
+
 usPrj = using_project(os.getenv("PHOENIX_PROJECT_NAME"))
 usPrj.__enter__()
 
+
 init_settings()
 init_observability()
 
@@ -52,12 +55,12 @@ mount_static_files("data_output", "/api/files/output")
 app.include_router(chat_router, prefix="/api/chat")
 app.include_router(file_upload_router, prefix="/api/chat/upload")
 
-# Redirect to documentation page when accessing base URL
+app.include_router(v1_router, prefix="/v1")
+
 @app.get("/")
 async def redirect_to_docs():
     return RedirectResponse(url="/docs")
 
-SentenceSplitter
 if __name__ == "__main__":
     app_host = os.getenv("APP_HOST", "0.0.0.0")
     app_port = int(os.getenv("APP_PORT", "8000"))
@@ -65,4 +68,3 @@ if __name__ == "__main__":
     reload = False
     uvicorn.run(app="main:app", host=app_host, port=app_port, reload=reload)
 
-    #usPrj.__exit__()

backend/pyproject.toml

@@ -17,6 +17,8 @@ aiostream = "^0.6.2"
 llama-index = "0.10.63"
 cachetools = "^5.3.3"
 protobuf = "4.25.4"
+nltk = "^3.9.1"
+jieba = "^0.42.1"
 
 #arize-phoenix = "^4.12.0"
 openinference-instrumentation-llama-index="2.2.3"
@@ -32,6 +34,8 @@ llama-index-vector-stores-qdrant = "^0.2.14"
 chroma="^0.2.0"
 llama-index-vector-stores-chroma = "^0.1.10"
 llama-index-readers-json = "^0.1.5"
+llama-index-retrievers-bm25 = "^0.2.2"
+llama-index-experimental = "^0.2.0"
 
 duckduckgo_search = "^6.2.6"
 
@@ -59,6 +63,12 @@ version = "^0.8"
 version = "0.0.7"
 
 
+
+[[tool.poetry.source]]
+name = "mirrors"
+url = "https://pypi.tuna.tsinghua.edu.cn/simple/"
+priority = "default"
+
 [build-system]
 requires = [ "poetry-core" ]
 build-backend = "poetry.core.masonry.api"

backend/test1/ParamTuner.py

@@ -0,0 +1,138 @@
+import nest_asyncio
+nest_asyncio.apply()
+from llama_index.core import SimpleDirectoryReader
+from llama_index.core.node_parser import SentenceSplitter
+from llama_index.core import VectorStoreIndex
+from llama_index.core.evaluation import (
+    FaithfulnessEvaluator, 
+    DatasetGenerator, 
+    CorrectnessEvaluator, 
+    SemanticSimilarityEvaluator,
+)
+from llama_index.experimental.param_tuner import ParamTuner
+from llama_index.experimental.param_tuner.base import RunResult
+from llama_index.llms.openai import OpenAI
+
+import asyncio
+
+# 初始化环境
+from app.observability import init_observability
+from app.settings import init_settings
+from dotenv import load_dotenv
+
+load_dotenv()
+init_settings()
+init_observability()
+
+# 读取文档
+documents = SimpleDirectoryReader("D:/LLM_model/text2sql/zjdataai-app-test/backend/data-test").load_data()
+
+# 参数字典
+param_dict = {
+    "chunk_size": [512, 1024],
+    "top_k": [1, 5],
+    "temperature": [0.1, 1.0]
+}
+
+# 辅助函数
+def _build_index(chunk_size, documents):
+    # 构建索引
+    splitter = SentenceSplitter(chunk_size=chunk_size)
+    vector_index = VectorStoreIndex.from_documents(
+        documents, transformations=[splitter],
+    )
+    return vector_index
+
+# 评估函数
+def evaluate_query_engine(query_engine, questions):
+    loop = asyncio.get_event_loop()
+    correct, total = loop.run_until_complete(_evaluate_query_engine_async(query_engine, questions))
+    return correct, total
+
+async def _evaluate_query_engine_async(query_engine, questions):
+    c = [query_engine.aquery(q) for q in questions]
+    gathering_future = asyncio.gather(*c)
+    results = await gathering_future
+
+    total_correct = 0
+    for r in results:
+        eval_result = (
+            1 if FaithfulnessEvaluator().evaluate_response(response=r).passing else 0
+        )
+        total_correct += eval_result
+
+    return total_correct, len(results)
+
+
+
+# 生成问题
+question_generator = DatasetGenerator.from_documents(documents)
+eval_questions = question_generator.generate_questions_from_nodes(1)  # 假设生成10个问题
+
+# 打印生成的问题
+for i, q in enumerate(eval_questions, start=1):
+    print(f"问题 {i}: {q}")
+
+# 目标函数
+def objective_function(params_dict, documents, questions):
+    chunk_size = params_dict["chunk_size"]
+    top_k = params_dict["top_k"]
+    temperature = params_dict["temperature"]
+
+    # 构建索引
+    vector_index = _build_index(chunk_size, documents)
+    
+    # 查询引擎
+    query_engine = vector_index.as_query_engine(
+        similarity_top_k=top_k, temperature=temperature
+    )
+
+    # 评估查询引擎
+    correct, total = 0, len(questions)
+    question_answers = []  # 添加列表来收集问题和答案
+    
+    for question in questions:
+        response = query_engine.query(question)
+        if response is not None:
+            question_answers.append((question, response.response))
+            eval_result = FaithfulnessEvaluator().evaluate_response(response=response, query_str=question)
+            if eval_result.passing:
+                correct += 1
+
+    # 计算分数
+    score = correct / total if total > 0 else 0
+    return RunResult(score=score, params=params_dict, question_answers=question_answers)
+
+# 创建 ParamTuner 实例
+param_tuner = ParamTuner(
+    param_fn=lambda params_dict: objective_function(params_dict, documents, eval_questions),
+    param_dict=param_dict,
+    show_progress=True,
+)
+
+# 调用 tune 方法
+results = param_tuner.tune()
+best_result = results.best_run_result
+best_top_k = best_result.params["top_k"]
+best_chunk_size = best_result.params["chunk_size"]
+best_temperature = best_result.params["temperature"]
+print(f"得分: {best_result.score}")
+print(f"Top-k: {best_top_k}")
+print(f"文本块大小: {best_chunk_size}")
+print(f"温度: {best_temperature}")
+
+# 使用最佳参数再次运行查询引擎，并打印问题与答案
+best_vector_index = _build_index(best_chunk_size, documents)
+best_query_engine = best_vector_index.as_query_engine(
+    similarity_top_k=best_top_k, temperature=best_temperature
+)
+
+best_question_answers = []
+for question in eval_questions:
+    response = best_query_engine.query(question)
+    if response is not None:
+        best_question_answers.append((question, response.response))
+
+# 打印最佳参数下的问题与答案
+for i, (question, answer) in enumerate(best_question_answers, start=1):
+    print(f"最佳参数 - 问题 {i}: {question}\n答案: {answer}\n")

backend/test1/evaluator.py

@@ -0,0 +1,81 @@
+from app.observability import init_observability
+from app.settings import init_settings
+from dotenv import load_dotenv
+
+import nest_asyncio
+nest_asyncio.apply()
+
+load_dotenv()
+
+
+from llama_index.core.node_parser import SentenceSplitter
+from llama_index.core import (
+    VectorStoreIndex,
+    SimpleDirectoryReader,
+    Response,
+)
+from llama_index.core.evaluation import (
+    FaithfulnessEvaluator, 
+    DatasetGenerator, 
+    CorrectnessEvaluator, 
+    SemanticSimilarityEvaluator,)
+
+
+
+init_settings()
+init_observability()
+
+faith_evaluator_qwen = FaithfulnessEvaluator() #诚实度评测
+corr_evaluator_qwen = CorrectnessEvaluator() #准确率评测
+Seman_evaluator_qwen = SemanticSimilarityEvaluator()#嵌入相似度评估
+
+documents = SimpleDirectoryReader("D:/LLM_model/text2sql/zjdataai-app-test/backend/data-test").load_data()
+
+splitter = SentenceSplitter(chunk_size=512)
+
+
+vector_index = VectorStoreIndex.from_documents(
+    documents, transformations=[splitter],
+)
+
+
+# # 运行评估
+# query_engine = vector_index.as_query_engine()
+# response_vector = query_engine.query("工程监理费的金额是多少？")
+# eval_result = evaluator_qwen.evaluate_response(response=response_vector)
+
+# print(response_vector)
+# print(eval_result)
+
+
+question_generator = DatasetGenerator.from_documents(documents)
+eval_questions = question_generator.generate_questions_from_nodes(5)
+print(eval_questions)
+
+import asyncio
+
+async def evaluate_query_engine_async(query_engine, questions):
+    c = [query_engine.aquery(q) for q in questions]
+    gathering_future = asyncio.gather(*c)
+    results = await gathering_future
+    #print(results)
+    
+    total_correct = 0
+    for r in results:
+        eval_result = (
+            1 if faith_evaluator_qwen.evaluate_response(response=r).passing else 0
+        )
+        total_correct += eval_result
+
+    return total_correct, len(results)
+
+def evaluate_query_engine(query_engine, questions):
+    loop = asyncio.get_event_loop()
+    correct, total = loop.run_until_complete(evaluate_query_engine_async(query_engine, questions))
+    return correct, total
+
+# 使用 evaluate_query_engine 函数
+vector_query_engine = vector_index.as_query_engine()
+correct, total = evaluate_query_engine(vector_query_engine, eval_questions[:5])
+
+print(f"score: {correct}/{total}")

backend/unit_test/corr_test.py

@@ -0,0 +1,121 @@
+from dotenv import load_dotenv
+load_dotenv()
+
+from llama_index.core.evaluation import CorrectnessEvaluator
+from app.engine import get_chat_engine  
+from app.engine.index import get_index
+from app.observability import init_observability
+from app.settings import init_settings
+
+init_settings()
+init_observability()
+
+index = get_index()
+
+
+import os
+import json
+import asyncio
+import nest_asyncio
+nest_asyncio.apply()
+from llama_index.core.prompts import (
+    ChatMessage,
+    ChatPromptTemplate,
+    MessageRole
+)
+
+DEFAULT_SYSTEM_TEMPLATE = """
+您是一个问答聊天机器人的专业评估系统。
+
+您将获得以下信息：
+
+- 用户查询，
+- 生成的回答，
+
+也可能提供一个参考答案作为评估的依据。
+
+您的任务是判断生成回答的相关性和正确性。
+输出一个代表全面评估的单一分数。
+您必须在一行中仅返回该分数。
+不要以其他任何格式返回答案。
+在单独的一行提供给定分数的理由。
+
+请遵循以下评分指南：
+
+- 您的分数必须在1到5之间，其中1是最差，5是最好的。
+-如果生成的回答与用户查询不相关，您应该给出1分。
+-如果生成的回答相关但包含错误，您应该给出2到3分之间的分数。
+-如果生成的回答相关且完全正确，您应该给出4到5分之间的分数。
+示例响应：
+4.0
+生成的回答与参考答案的指标完全相同，但不够精炼。
+
+"""
+
+DEFAULT_USER_TEMPLATE = """
+## User Query
+{query}
+
+## Reference Answer
+{reference_answer}
+
+## Generated Answer
+{generated_answer}
+"""
+
+DEFAULT_EVAL_TEMPLATE = ChatPromptTemplate(
+    message_templates=[
+        ChatMessage(role=MessageRole.SYSTEM, content=DEFAULT_SYSTEM_TEMPLATE),
+        ChatMessage(role=MessageRole.USER, content=DEFAULT_USER_TEMPLATE),
+    ]
+)
+
+
+# 初始化聊天引擎和评估器
+chat_engine = get_chat_engine()
+corr_evaluator_qwen = CorrectnessEvaluator()
+
+# 加载本地问题回答文件
+script_dir = os.path.dirname(os.path.abspath(__file__))
+file_path = os.path.join(script_dir, 'questions_and_answers.json')
+output_file_path = file_path.replace('.json', '_test.json')
+
+with open(file_path, 'r', encoding='utf-8') as f:
+    data = json.load(f)
+
+# 异步函数用于评估查询
+async def evaluate_query(question, answer, index, output_file):
+    response = await chat_engine.astream_chat(question)
+
+    # 检查sources是否为空
+    if response.sources:
+        content_str = str(response.sources[0])
+    else:
+        content_str = "<无回答>"
+
+    result = corr_evaluator_qwen.evaluate(
+        query=question,
+        response=content_str, 
+        reference=answer,
+    )
+    
+    result_dict = {
+        "编号": index,
+        "问题": question,
+        "答案": answer,
+        "回答": result.response,
+        "得分(1~5)": result.score,
+        "评价": result.feedback
+    }
+    
+    with open(output_file, 'a', encoding='utf-8') as f:
+        f.write(json.dumps(result_dict, ensure_ascii=False, indent=4))
+        f.write(',\n')
+
+# 主异步函数
+async def main():
+    for index, item in enumerate(data, start=1):
+        await evaluate_query(item['question'], item['answer'], index, output_file_path)
+
+# 运行主协程
+asyncio.run(main())

backend/unit_test/prompts.py

@@ -0,0 +1,55 @@
+Attribute_Prompt = (
+    "你是一个电力造价工程相关的项目经理，现在给你一些上下文信息，"
+    "你需要根据现有的上下文信息，来生成{num_questions_per_chunk}个电力造价工程相关的问题和对应的回答，"
+	"现在需要你针对数据中属性一列进行提问和回答。"
+    "问题和回答的示例应该是这种类型的，示例：'工程总投资（万元）,工程总投资（万元）是77469835.590045万元','尖峰及施工基面土石方量,尖峰及施工基面土石方量是8377.6','截止阀的编码,截止阀的编码是F01010203',"
+    "你生成的回答必须严格按照示例中的格式（'问题, 回答'），不允许有丝毫的变动。问题和回答应该在一个单引号内。"
+    "这种类似的问题和答案，生成的问题和答案必须一一对应，要符合文件里的内容，不要生成一些无关的问题，不要生成一些重复的问题，"
+    "不要生成一些过于简单的问题，不要生成一些过于复杂的问题。"
+)
+
+
+Amount_Prompt = (
+    "你是一个电力造价工程相关的项目经理，现在给你一些上下文信息，"
+    "你需要根据现有的上下文信息，来生成{num_questions_per_chunk}个电力造价工程相关的问题和对应的回答，"
+	"现在需要你针对上下文信息中的金额或者合价进行提问和回答。" 	
+    "问题和回答的示例应该是这种类型的，示例：'项目建设技术服务费的金额,项目建设技术服务费的金额是16855957065.4302','项目后评价费的费率,项目后评价费的费率是0.5','架空输电线路本体工程的金额,架空输电线路本体工程的金额是55105688268.5176','工程静态投资的金额,工程静态投资的金额是715035853336.391'"
+    "你生成的回答必须严格按照示例中的格式（'问题, 回答'），不允许有丝毫的变动。问题和回答应该在一个单引号内。"
+    "这种类似的问题和答案，生成的问题和答案必须一一对应，要符合文件里的内容，不要生成一些无关的问题，不要生成一些重复的问题，"
+    "不要生成一些过于简单的问题，不要生成一些过于复杂的问题。"
+)
+
+
+
+Units_Prompt = (
+    "你是一个电力造价工程相关的项目经理，现在给你一些上下文信息，"
+    "你需要根据现有的上下文信息，来生成{num_questions_per_chunk}个电力造价工程相关的问题和对应的回答，"
+	"现在需要你针对上下文信息来进行单位转化问题提问和回答。"  
+    "问题和回答的示例应该是这种类型的，示例：'工程总投资（万元）结果用元表示,工程总投资（万元）是774698355900.45元','本体工程（元）结果用万元表示,本体工程（元）是5490494.261046万元'"
+    "你生成的回答必须严格按照示例中的格式（'问题, 回答'），不允许有丝毫的变动。问题和回答应该在一个单引号内。"
+    "这种类似的问题和答案，生成的问题和答案必须一一对应，要符合文件里的内容，不要生成一些无关的问题，不要生成一些重复的问题，"
+    "不要生成一些过于简单的问题，不要生成一些过于复杂的问题。"
+)
+
+Name_Prompt = (
+    "你是一个电力造价工程相关的项目经理，现在给你一些上下文信息，"
+    "你需要根据现有的上下文信息，来生成{num_questions_per_chunk}个电力造价工程相关的问题和对应的回答，"
+	"现在需要你针对上下文信息中的重名问题进行提问和回答。"    
+    "问题和回答的示例应该是这种类型的，示例：'专业类型为线路的杆塔工程项目划分的合价,专业类型为线路的杆塔工程项目划分的合价是220969744.905856','专业类型为线路清理的杆塔工程项目划分的合价,电缆工程的合价是0'"
+    "你生成的回答必须严格按照示例中的格式（'问题, 回答'），不允许有丝毫的变动。问题和回答应该在一个单引号内。"
+    "这种类似的问题和答案，生成的问题和答案必须一一对应，要符合文件里的内容，不要生成一些无关的问题，不要生成一些重复的问题，"
+    "不要生成一些过于简单的问题，不要生成一些过于复杂的问题。"
+)
+
+
+All_Amount_Prompt = (
+    "你是一个电力造价工程相关的项目经理，现在给你一些上下文信息，"
+    "你需要根据现有的上下文信息，来生成{num_questions_per_chunk}个电力造价工程相关的问题和对应的回答，"
+	"现在需要你针对上下文信息中的总体金额进行提问和回答。" 	
+    "问题和回答的示例应该是这种类型的，示例：'架空输电线路本体工程的总体金额,架空输电线路本体工程的总体金额是7.706703','工程静态投资的总体金额,工程静态投资的总体金额是100'"
+    "你生成的回答必须严格按照示例中的格式（'问题, 回答'），不允许有丝毫的变动。问题和回答应该在一个单引号内。"
+    "这种类似的问题和答案，生成的问题和答案必须一一对应，要符合文件里的内容，不要生成一些无关的问题，不要生成一些重复的问题，"
+    "不要生成一些过于简单的问题，不要生成一些过于复杂的问题。"
+)
+
+

backend/unit_test/question.py

@@ -0,0 +1,144 @@
+
+from dotenv import load_dotenv
+load_dotenv()
+
+import json
+import sys
+
+
+from app.observability import init_observability
+from app.settings import init_settings
+
+import nest_asyncio
+nest_asyncio.apply()
+
+from llama_index.core.node_parser import SentenceSplitter
+from llama_index.core import SimpleDirectoryReader
+from llama_index.core.evaluation import DatasetGenerator
+
+import prompts  
+
+init_settings()
+init_observability()
+
+# 读取所有文档（即所有表格）
+documents = SimpleDirectoryReader("D:/LLM_model/text2sql/zjdataai-app-test/backend/data-test").load_data()
+
+# 定义表格名称和索引的对应关系
+table_names = {
+    "工程信息表": 0,
+    "其他费用表": 1,
+    "取费表": 2,
+    "项目划分表": 3,
+    "项目划分_费用预览表": 4,
+    "总算表": 5,
+    "工程量表": 6
+}
+
+# 定义中文提示词和Python代码中提示词名称的映射
+prompt_mapping = {
+    "普通属性": "Attribute_Prompt",
+    "金额查询": "Amount_Prompt",
+    "单位换算": "Units_Prompt",
+    "重名项目划分": "Name_Prompt",
+    "总体金额查询": "All_Amount_Prompt"
+}
+
+# 定义表格与其对应的查询类别
+table_prompt_mapping = {
+    "工程信息表": ["普通属性", "单位换算"],
+    "其他费用表": ["金额查询", "单位换算"],
+    "取费表": ["金额查询"],
+    "总算表": ["金额查询", "总体金额查询"],
+    "工程量表": ["普通属性", "重名项目划分"]
+}
+
+# 根据表格名称选择特定的表格
+def select_document(documents, table_name):
+    if table_name not in table_names:
+        raise ValueError(f"未找到名为 '{table_name}' 的表格")
+    index = table_names[table_name]
+    return [documents[index]]  # 返回一个包含所选表格的列表
+
+# 选择提示词
+def select_prompt(prompt_category):
+    prompt_name = prompt_mapping.get(prompt_category)
+    if not prompt_name:
+        raise ValueError(f"未找到名为 '{prompt_category}' 的提示词")
+    try:
+        return getattr(prompts, prompt_name)
+    except AttributeError:
+        raise ValueError(f"未找到提示词 '{prompt_name}' 对应的函数")
+
+# 生成问题和答案
+def generate_questions_from_document(document, quest_prompt, num_questions):
+    question_generator = DatasetGenerator.from_documents(
+        documents=document,
+        question_gen_query=quest_prompt,
+        num_questions_per_chunk=num_questions
+    )
+
+    eval_questions = question_generator.generate_questions_from_nodes(num_questions)
+    print(eval_questions)
+
+    qa_pairs = []
+    for qa in eval_questions:
+        if ',' in qa:
+            question, answer = qa.split(",", 1)
+            qa_pairs.append({
+                "question": question.strip(),
+                "answer": answer.strip()
+            })
+        else:
+            print(f"无法处理的问题和答案: {qa}")
+
+    return qa_pairs
+
+# 主函数，控制生成多个表格的问题和使用多个提示词，并将结果合并到一个文件中
+def main(documents, table_names_input, prompt_categories_input, num_questions_per_prompt):
+    if table_names_input == "all":
+        selected_tables = list(table_prompt_mapping.keys())
+    else:
+        selected_tables = table_names_input.strip('[]').split(',')
+
+    all_results = {}
+
+    for table_name in selected_tables:
+        table_name = table_name.strip()  # 去掉前后空格
+        document = select_document(documents, table_name)
+
+        if prompt_categories_input == "all":
+            selected_prompts = table_prompt_mapping[table_name]
+        else:
+            selected_prompts = prompt_categories_input.strip('[]').split(',')
+            selected_prompts = [p.strip() for p in selected_prompts]  # 去掉前后空格
+
+        for prompt_category in selected_prompts:
+            if prompt_category not in table_prompt_mapping[table_name]:
+                print(f"跳过表格 '{table_name}' 的提示词 '{prompt_category}'，因为该表中不包含该类别的信息")
+                continue
+
+            quest_prompt = select_prompt(prompt_category).format(num_questions_per_chunk=num_questions_per_prompt)
+            qa_pairs = generate_questions_from_document(document, quest_prompt, num_questions_per_prompt)
+
+            label = f"test:{table_name}_{prompt_category}"
+            all_results[label] = qa_pairs
+
+    # 自动生成输出文件名
+    output_file = "combined_test.json"
+
+    with open(output_file, "w", encoding="utf-8") as f:
+        json.dump(all_results, f, ensure_ascii=False, indent=4)
+
+    print(f"All questions and answers have been saved to '{output_file}'")
+
+# 获取命令行参数
+if __name__ == "__main__":
+    if len(sys.argv) != 4:
+        print("Usage: python script.py <table_names_input> <prompt_categories_input> <num_questions_per_prompt>")
+    else:
+        table_names_input = sys.argv[1]
+        prompt_categories_input = sys.argv[2]
+        num_questions_per_prompt = int(sys.argv[3])
+
+        main(documents, table_names_input, prompt_categories_input, num_questions_per_prompt)

phoenixserver/phoenixserver.py

@@ -1,9 +1,10 @@
 import os
+from dotenv import load_dotenv
+load_dotenv()
+
 import phoenix as px
 
 
-os.environ['PHOENIX_HOST'] = "0.0.0.0"
-
 session = px.launch_app(use_temp_dir=False)
 
 import msvcrt