修改流程，改成无agent流程，新增地区提取工具

src/agent/tool_divisions.py

@@ -9,7 +9,7 @@ divisions = [
         },
         {
         "name": "海东市",
-        "counties": ["乐都区", "平安区", "民和回族土族自治县", "互助土族自治县", "化隆回族自治县", "循化撒拉族自治县"]
+        "counties": ["乐都区", "平安区", "民和回族土族自治县", "互助土族自治县", "化隆县", "循化撒拉族自治县"]
         },
         {
         "name": "海北藏族自治州",

src/config/prompts.py

@@ -306,4 +306,41 @@ A: 你可以在 nginx 官网上下载 nginx。
 {histories}
 '''
 新问题：
+"""
+
+
+# 结合历史问答对话，提取问题中的位置信息
+PROMPT_LOCATION_EXTEND = """
+做为一个行政区提取助手，你的任务是结合历史对话，提取出问题所涉及到的行政区。例如：
+历史记录: 
+'''
+'''
+原问题: 西宁市各区县谁的年平均降雨量大。
+提取到的行政区: [西宁市]
+----------------
+历史记录:
+'''
+Q: 对话背景。
+A: 当前对话是关于化隆县的介绍。
+'''
+原问题: 其年最高温是多少
+提取到的行政区: [化隆县]
+----------------
+历史记录: 
+'''
+'''
+原问题: 请问 nginx 如何下载？
+提取到的行政区: []
+----------------
+历史记录: 
+'''
+'''
+原问题: 大通县和湟源县谁的降雨量高？
+提取到的行政区:[大通县,湟源县]
+----------------
+'''
+{histories}
+'''
+原问题: {query}
+提取到的行政区:
 """
\ No newline at end of file

src/controller/web.py

@@ -8,22 +8,15 @@ from src.pgdb.chat.c_db import UPostgresDB
 import uvicorn
 import json
 from src.pgdb.chat.crud import CRUD
-from langchain.agents import AgentExecutor
-from langchain_core.prompts.chat import ChatPromptTemplate,HumanMessagePromptTemplate,SystemMessagePromptTemplate,MessagesPlaceholder
-from langchain.chains import LLMChain
-import langchain_core
-from typing import List,Union
+
 from src.pgdb.knowledge.similarity import VectorStore_FAISS
-from src.server.get_similarity import QAExt
-from src.server.agent import create_chart_agent
+
 from src.pgdb.knowledge.k_db import PostgresDB
 from src.pgdb.knowledge.txt_doc_table import TxtDoc
-from src.agent.tool_divisions import AdministrativeDivision
-from src.agent.rag_agent import RAGQuery
-from langchain_core.prompts import PromptTemplate
+
 from langchain_openai import ChatOpenAI
 from langchain_core.documents import Document
-import re
+from src.server.rag_query import RagQuery
 from src.controller.request import (
     PhoneLoginRequest,
     ChatRequest,
@@ -44,10 +37,7 @@ from src.config.consts import (
     VEC_DB_USER,
     VEC_DB_DBNAME,
     SIMILARITY_SHOW_NUMBER,
-    prompt_enhancement_history_template,
-    prompt1
 )
-from src.config.prompts import PROMPT_AGENT_SYS_VARS,PROMPT_AGENT_SYS,PROMPT_AGENT_CHAT_HUMAN,PROMPT_AGENT_CHAT_HUMAN_VARS,PROMPT_AGENT_EXTEND_SYS
 
 
 app = FastAPI()
@@ -82,54 +72,10 @@ base_llm = ChatOpenAI(
     verbose=True
 )
 
-ext = QAExt(base_llm)
-
-llm_chain = LLMChain(llm=base_llm, prompt=PromptTemplate(input_variables=["history","context", "question"], template=prompt1), llm_kwargs= {"temperature": 0})
-tools = [RAGQuery(vecstore_faiss,ext,PromptTemplate(input_variables=["history","context", "question"], template=prompt_enhancement_history_template),_db=TxtDoc(k_db),_llm_chain=llm_chain)]
-# input_variables=['agent_scratchpad', 'input', 'tool_names', 'tools','chart_tool']
-input_variables=[]
-input_variables.extend(PROMPT_AGENT_CHAT_HUMAN_VARS)
-input_variables.extend(PROMPT_AGENT_SYS_VARS)
-input_types={'chat_history': List[Union[langchain_core.messages.ai.AIMessage, langchain_core.messages.human.HumanMessage, langchain_core.messages.chat.ChatMessage, langchain_core.messages.system.SystemMessage, langchain_core.messages.function.FunctionMessage, langchain_core.messages.tool.ToolMessage]]}
-messages=[
-#   SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=['tool_names', 'tools'], template='Respond to the human as helpfully and accurately as possible. You have access to the following tools:\n\n{tools}\n\nUse a json blob to specify a tool by providing an action key (tool name) and an action_input key (tool input).\n\nValid "action" values: "Final Answer" or {tool_names}\n\nProvide only ONE action per $JSON_BLOB, as shown:\n\n```\n{{\n  "action": $TOOL_NAME,\n  "action_input": $INPUT\n}}\n```\n\nFollow this format:\n\nQuestion: input question to answer\nThought: consider previous and subsequent steps\nAction:\n```\n$JSON_BLOB\n```\nObservation: action result\n... (repeat Thought/Action/Observation N times)\nThought: I know what to respond\nAction:\n```\n{{\n  "action": "Final Answer",\n  "action_input": "Final response to human"\n}}\n\nBegin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:```$JSON_BLOB```then Observation')),
-  SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=PROMPT_AGENT_SYS_VARS, template=PROMPT_AGENT_SYS)),
-  MessagesPlaceholder(variable_name='chat_history', optional=True),
-#   HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['agent_scratchpad', 'input'], template='{input}\n\n{agent_scratchpad}\n (reminder to respond in a JSON blob no matter what)'))
-  HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=PROMPT_AGENT_CHAT_HUMAN_VARS, template=PROMPT_AGENT_CHAT_HUMAN))
-]
-
-prompt = ChatPromptTemplate(
-    input_variables=input_variables,
-    input_types=input_types,
-    # metadata=metadata,
-    messages=messages
-)
+rag_query = RagQuery(base_llm=base_llm,_faiss_db=vecstore_faiss,_db=TxtDoc(k_db))
 
 
-# agent = create_structured_chat_agent(llm=base_llm, tools=tools, prompt=prompt)
-agent = create_chart_agent(base_llm, tools, prompt, chart_tool="chart")
-agent_executor = AgentExecutor(agent=agent, tools=tools,verbose=True,handle_parsing_errors=True,return_intermediate_steps=True)
-
-administrative_messages=[
-#   SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=['tool_names', 'tools'], template='Respond to the human as helpfully and accurately as possible. You have access to the following tools:\n\n{tools}\n\nUse a json blob to specify a tool by providing an action key (tool name) and an action_input key (tool input).\n\nValid "action" values: "Final Answer" or {tool_names}\n\nProvide only ONE action per $JSON_BLOB, as shown:\n\n```\n{{\n  "action": $TOOL_NAME,\n  "action_input": $INPUT\n}}\n```\n\nFollow this format:\n\nQuestion: input question to answer\nThought: consider previous and subsequent steps\nAction:\n```\n$JSON_BLOB\n```\nObservation: action result\n... (repeat Thought/Action/Observation N times)\nThought: I know what to respond\nAction:\n```\n{{\n  "action": "Final Answer",\n  "action_input": "Final response to human"\n}}\n\nBegin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:```$JSON_BLOB```then Observation')),
-  SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=PROMPT_AGENT_SYS_VARS, template=PROMPT_AGENT_EXTEND_SYS)),
-  MessagesPlaceholder(variable_name='chat_history', optional=True),
-#   HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['agent_scratchpad', 'input'], template='{input}\n\n{agent_scratchpad}\n (reminder to respond in a JSON blob no matter what)'))
-  HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=PROMPT_AGENT_CHAT_HUMAN_VARS, template=PROMPT_AGENT_CHAT_HUMAN))
-]
-administrative_prompt = ChatPromptTemplate(
-    input_variables=input_variables,
-    input_types=input_types,
-    # metadata=metadata,
-    messages=administrative_messages
-)
-AdministrativeTools =[AdministrativeDivision()]
-administrative_agent = create_chart_agent(base_llm, AdministrativeTools, administrative_prompt, chart_tool="chart")
-administrative_agent_executor = AgentExecutor(agent=administrative_agent, tools=AdministrativeTools,verbose=True,handle_parsing_errors=True,return_intermediate_steps=True)
 
-# my_chat = QA(PromptTemplate(input_variables=["history","context", "question"], template=prompt_enhancement_history_template), base_llm,
-#              {"temperature": 0.9}, ['history','context', 'question'], _db=c_db, _faiss_db=vecstore_faiss,rerank=True)
 
 @app.post('/api/login')
 def login(phone_request: PhoneLoginRequest):
@@ -155,7 +101,7 @@ def login(phone_request: PhoneLoginRequest):
 
 
 @app.get('/api/sessions/chat/')
-def get_sessions(timestamp: int = Query(None, alias="_"),token: str = Header(None)):
+def get_sessions(token: str = Header(None)):
     if not token:
         return {
             'code': 404,
@@ -224,28 +170,14 @@ def question(chat_request: ChatRequest, token: str = Header(None)):
     history = []
     if session_id !="":
         history = crud.get_last_history(str(session_id))
-    # answer = my_chat.chat(question)
-    # result = ext.extend_query(question, history)
-    # matches = re.findall(r'"([^"]+)"', result.content)
-    # print(matches)
-    # if len(matches)>3:
-    #     matches = matches[:3]
-    # print(matches)
-    # prompt = ""
-    # for h in history:
-    #     prompt += "问：{}\n答：{}\n\n".format(h[0], h[1])
-    res_a = administrative_agent_executor.invoke({"input": question, "histories": history})
-    new_question = res_a['output']
-
-    res = agent_executor.invoke({"input": new_question, "histories": history})
-    answer = res["output"]
-
-    # answer, docs = my_chat.chat_with_history_with_ext(question,ext=matches,history=prompt, with_similarity=True)
-    docs_json = []
-    for step in res["intermediate_steps"]:
-        if "rag_query" == step[0].tool:
-            j = json.loads(step[1]["参考文档"], strict=False)
-            docs_json.extend(j)
+    prompt = ""
+    for h in history:
+        prompt += "Q: {}\nA:{}\n".format(h[0], h[1])
+    res = rag_query.query(question=question,history=prompt)
+    answer = res["answer"]
+    docs = res["docs"]
+    docs_json = json.loads(docs, strict=False)
+
     print(len(docs_json))
     doc_hash = []
     for d in docs_json:
@@ -287,26 +219,14 @@ def re_generate(chat_request: ReGenerateRequest, token: str = Header(None)):
     crud = CRUD(_db=c_db)
     last_turn_id = crud.get_last_turn_num(str(session_id))
     history = crud.get_last_history_before_turn_id(str(session_id),last_turn_id)
-    # result = ext.extend_query(question, history)
-    # matches = re.findall(r'"([^"]+)"', result.content)
-    # if len(matches)>3:
-    #     matches = matches[:3]
-    # print(matches)
-    # prompt = ""
-    # for h in history:
-    #     prompt += "问：{}\n答：{}\n\n".format(h[0], h[1])
-    # answer, docs = my_chat.chat_with_history_with_ext(question,ext=matches, history=prompt, with_similarity=True)
-    # docs_json = []
-    res_a = administrative_agent_executor.invoke({"input": question, "histories": history})
-    new_question = res_a['output']
-
-    res = agent_executor.invoke({"input": new_question, "histories": history})
-    answer = res["output"]
-    docs_json = []
-    for step in res["intermediate_steps"]:
-        if "rag_query" == step[0].tool:
-            j = json.loads(step[1]["参考文档"], strict=False)
-            docs_json.extend(j)
+    prompt = ""
+    for h in history:
+        prompt += "Q: {}\nA:{}\n".format(h[0], h[1])
+    res = rag_query.query(question=question, history=prompt)
+    answer = res["answer"]
+    docs = res["docs"]
+    docs_json = json.loads(docs, strict=False)
+
 
     doc_hash = []
     for d in docs_json:

src/server/extend.py

+from langchain_core.prompts import PromptTemplate
+from src.config.prompts import PROMPT_LOCATION_EXTEND
+
+class LocationExt:
+    llm = None
+
+    def __init__(self, llm) -> None:
+        self.llm = llm
+        prompt = PromptTemplate.from_template(PROMPT_LOCATION_EXTEND)
+        # parser = ListOutputParser()
+        self.query_extend = prompt | llm
+
+    def extend_query(self, question, messages=None):
+        """
+            question: str
+            messages: list of tuple (str,str)
+                eg:
+                [
+                    ("Q1","A1"),
+                    ("Q2","A2"),
+                    ...
+                ]
+
+        """
+        if not messages:
+            messages = []
+        history = ""
+        for msg in messages:
+            history += f"Q: {msg[0]}\nA: {msg[1]}\n"
+        return self.query_extend.invoke(input={"histories": history, "query": question})
+
+    def extend_query_str(self, question, history):
+        return self.query_extend.invoke(input={"histories": history, "query": question})
\ No newline at end of file

src/server/rag_query.py

+import re
+from src.server.get_similarity import GetSimilarityWithExt
+import time
+from src.server.rerank import BgeRerank
+from langchain_core.documents import Document
+from langchain_core.prompts import PromptTemplate
+from src.server.get_similarity import QAExt
+from src.server.extend import LocationExt
+import json
+from src.agent.tool_divisions import complete_administrative_division,divisions
+from langchain.chains import LLMChain
+from src.config.consts import (
+    RERANK_MODEL_PATH,
+    prompt1
+)
+
+
+class RagQuery():
+    def __init__(self,base_llm,_faiss_db,_db):
+        self.qa_ext = QAExt(base_llm)
+        self.location_ext = LocationExt(base_llm)
+        self.rerank_model = BgeRerank(RERANK_MODEL_PATH)
+        self.faiss_db = _faiss_db
+        self.db = _db
+        self.llm_chain = LLMChain(llm=base_llm, prompt=PromptTemplate(input_variables=["history","context", "question"], template=prompt1), llm_kwargs= {"temperature": 0})
+
+
+    def get_similarity_with_ext_origin(self, _ext,_location):
+        return GetSimilarityWithExt(_question=_ext, _faiss_db=self.faiss_db,_location=_location)
+
+    def query(self, question: str, history: str) :
+        location_result = self.location_ext.extend_query_str(question=question, history=history)
+        index = location_result.content.find("提取到的行政区:")
+        if index == -1:
+            location_str = location_result.content
+        else:
+            location_str = location_result.content[index + len("提取到的行政区:"):]
+
+        pattern = r'\[([^\]]+)\]'
+        match = re.search(pattern, location_str)
+        cities = []
+        if match:
+            cities = match.group(1).split(', ')
+        cities_ext = []
+        for m in cities:
+            city_ext = complete_administrative_division(m, divisions)
+            cities_ext.append(city_ext)
+        location = []
+        prompt = ""
+        for city in cities_ext:
+            if city is not None and "县(区)" in city:
+                if isinstance(city["县(区)"], str):
+                    location.append(city["县(区)"])
+                    prompt += city["县(区)"] + "位于" + city["省"] + city["市"] + "，"
+                if isinstance(city["县(区)"], list):
+                    location.extend(city["县(区)"])
+                    prompt += city["省"] + city["市"] + "管辖"
+                    for x in city["县(区)"]:
+                        prompt += x + "，"
+        new_question = prompt + question
+        print(new_question)
+        split_docs_list = []
+        for l in location:
+            start = time.time()
+            answer = self.db.find_like_doc(l)
+            end = time.time()
+            print('find_like_doc time: %s Seconds' % (end - start))
+            print(len(answer) if answer else 0)
+            split_docs = []
+            for a in answer if answer else []:
+                d = Document(page_content=a[0], metadata=json.loads(a[1]))
+                split_docs.append(d)
+            print(len(split_docs))
+            # if len(split_docs) > 5:
+            #     split_docs = split_docs[:5]
+            split_docs_list.append(split_docs)
+
+        start = time.time()
+        result = self.qa_ext.extend_query_with_str(new_question, history)
+        end = time.time()
+        print('extend_query_with_str time: %s Seconds' % (end - start))
+        print(result)
+        matches = re.findall(r'"([^"]+)"', result.content)
+
+        print(matches)
+        similarity = self.get_similarity_with_ext_origin(matches,_location=location)
+        cur_similarity = similarity.get_rerank_with_doc(self.rerank_model,split_docs_list)
+
+        cur_answer = self.llm_chain.run(context=cur_similarity, question=new_question, history=history)
+        return {"answer":cur_answer,"docs": cur_similarity}
\ No newline at end of file

test/extend_test.py

+from src.server.extend import LocationExt
+from langchain_openai import ChatOpenAI
+import re
+from src.agent.tool_divisions import complete_administrative_division,divisions
+
+
+
+base_llm = ChatOpenAI(
+    openai_api_key='xxxxxxxxxxxxx',
+    openai_api_base='http://192.168.10.14:8000/v1',
+    model_name='Qwen2-7B',
+    verbose=True
+)
+
+ext = LocationExt(base_llm)
+message = [
+
+    ]
+question = "大通县和化隆县最大降雨量是多少"
+result =  ext.extend_query(question=question,messages=message)
+print(result.content)
+index = result.content.find("提取到的行政区:")
+if index==-1:
+    location = result.content
+else:
+    location = result.content[index + len("提取到的行政区:"):]
+
+# 获取特定子字符串之后的字符串
+print(location)
+pattern = r'\[([^\]]+)\]'
+match = re.search(pattern, location)
+cities = []
+if match:
+    # 匹配到的内容是括号内的全部内容，包括逗号和空格
+    # 再次使用split分割这个字符串
+    cities = match.group(1).split(', ')
+print(cities)
+location_exts = []
+for m in cities:
+    print(m)
+    location_ext = complete_administrative_division(m,divisions)
+    print(location_ext)
+    location_exts.append(location_ext)
+prompt1 = ""
+locations = []
+for l in location_exts:
+    if l is not None and "县(区)" in l:
+        if isinstance(l["县(区)"], str):
+            locations.append(l["县(区)"])
+            prompt1 += l["县(区)"] + "位于" + l["省"]+ l["市"]+"，"
+        if isinstance(l["县(区)"], list):
+            locations.extend(l["县(区)"])
+            prompt1 += l["省"] + l["市"] + "拥有"
+            for x in l["县(区)"]:
+                prompt1+= x+"，"
+
+print(locations)
+question = prompt1+question
+print(question)
\ No newline at end of file

test/rag_query_test.py

+from langchain_openai import ChatOpenAI
+from src.server.rag_query import RagQuery
+from src.pgdb.knowledge.txt_doc_table import TxtDoc
+from src.pgdb.knowledge.k_db import PostgresDB
+from src.pgdb.knowledge.similarity import VectorStore_FAISS
+import json
+from src.config.consts import (
+    EMBEEDING_MODEL_PATH,
+    FAISS_STORE_PATH,
+    INDEX_NAME,
+    VEC_DB_HOST,
+    VEC_DB_PASSWORD,
+    VEC_DB_PORT,
+    VEC_DB_USER,
+    VEC_DB_DBNAME,
+    SIMILARITY_SHOW_NUMBER,
+)
+
+base_llm = ChatOpenAI(
+    openai_api_key='xxxxxxxxxxxxx',
+    openai_api_base='http://192.168.10.14:8000/v1',
+    model_name='Qwen2-7B',
+    verbose=True
+)
+
+vecstore_faiss = VectorStore_FAISS(
+        embedding_model_name=EMBEEDING_MODEL_PATH,
+        store_path=FAISS_STORE_PATH,
+        index_name=INDEX_NAME,
+        info={"port": VEC_DB_PORT, "host": VEC_DB_HOST, "dbname": VEC_DB_DBNAME, "username": VEC_DB_USER,
+              "password": VEC_DB_PASSWORD},
+        show_number=SIMILARITY_SHOW_NUMBER,
+        reset=False)
+
+k_db = PostgresDB(host=VEC_DB_HOST, database=VEC_DB_DBNAME, user=VEC_DB_USER, password=VEC_DB_PASSWORD, port=VEC_DB_PORT)
+k_db.connect()
+db = TxtDoc(k_db)
+
+rag_query = RagQuery(base_llm=base_llm,_faiss_db=vecstore_faiss,_db=db)
+question = "大通县和化隆县最大降雨量是多少"
+history = ""
+result = rag_query.query(question=question,history=history)
+print(result)
+j = json.loads(result["docs"], strict=False)
+print(type(j))
+print(j)
\ No newline at end of file