106 lines
3.8 KiB
Python
Executable File
106 lines
3.8 KiB
Python
Executable File
import json
|
|
import chatbot_handler as chat
|
|
# import bale_qabot
|
|
import os
|
|
import numpy as np
|
|
import torch
|
|
import faiss
|
|
from typing import List, Tuple
|
|
from sentence_transformers import SentenceTransformer
|
|
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
|
from sklearn.feature_extraction.text import TfidfVectorizer
|
|
from sklearn.metrics.pairwise import cosine_similarity
|
|
import datetime
|
|
import re
|
|
import random
|
|
from fastapi.middleware.cors import CORSMiddleware
|
|
from embedder_sbert_qavanin_285k import PersianVectorAnalyzer
|
|
#from normalizer import cleaning
|
|
from fastapi import FastAPI ,Header
|
|
from pydantic import BaseModel
|
|
# LLM Libs
|
|
from openai import OpenAI
|
|
from langchain_openai import ChatOpenAI # pip install -U langchain_openai
|
|
import requests
|
|
from FlagEmbedding import FlagReranker # deldar-reranker-v2
|
|
import aiofiles
|
|
|
|
chatbot = FastAPI()
|
|
origins = ["*"]
|
|
|
|
chatbot.add_middleware(
|
|
CORSMiddleware,
|
|
allow_origins=origins,
|
|
allow_credentials=True,
|
|
allow_methods=["*"],
|
|
allow_headers=["*"],
|
|
)
|
|
|
|
print('#'*19)
|
|
print('-Chatbot is Ready-')
|
|
print('#'*19)
|
|
|
|
# تعریف مدل داده‌ها برای درخواست‌های API
|
|
class Query(BaseModel):
|
|
query: str
|
|
# مسیر API برای اجرا کردن run_chatbot
|
|
@chatbot.get("/")
|
|
async def simple():
|
|
return "ai rag caht qanon OK"
|
|
|
|
@chatbot.get("/ping")
|
|
async def ping():
|
|
return "ai rag caht qanon OK"
|
|
|
|
|
|
@chatbot.post("/run_chat")
|
|
async def run_chat(query: Query):
|
|
print('generate answer ...')
|
|
chat_id = await chat.create_chat_id()
|
|
answer = await chat.run_chatbot(query.query, chat_id)
|
|
await chat.credit_refresh()
|
|
|
|
return {"answer": answer}
|
|
|
|
# uvicorn src.app:app --reload
|
|
|
|
if __name__ == "__main__":
|
|
|
|
# query = 'در قانون حمایت از خانواده و جوانی جمعیت چه خدماتی در نظر گرفته شده است؟'
|
|
while True:
|
|
query = input('enter your qustion:')
|
|
if query == '':
|
|
print('لطفا متن سوال را وارد نمائید')
|
|
continue
|
|
start = (datetime.datetime.now())
|
|
# result = test_dataset()
|
|
result = chat.single_query(query)
|
|
end_retrive = datetime.datetime.now()
|
|
print('-'*40)
|
|
print(f'retrive duration: {(end_retrive - start).total_seconds()}')
|
|
|
|
prompt = f'برای پرسش "{query}" از میان مواد قانونی "{result}" .پاسخ مناسب و دقیق را استخراج کن. درصورتی که مطلبی مرتبط با پرسش در متن پیدا نشد، فقط پاسخ بده: "متاسفانه در منابع، پاسخی پیدا نشد!"'
|
|
llm_answer = chat.llm_request(prompt)
|
|
|
|
print('-'*40)
|
|
print(f'llm duration: {(datetime.datetime.now() - end_retrive).total_seconds()}')
|
|
|
|
refrences = ''
|
|
recognized_refrences = chat.find_refrences(llm_answer)
|
|
llm_answer = chat.replace_refrences(llm_answer, recognized_refrences)
|
|
|
|
with open('./llm-answer/result.txt', mode='a+', encoding='utf-8') as file:
|
|
result_message = f'متن پرامپت: {query.strip()}\n\nپاسخ: {llm_answer} \n----------------------------------------------------------\n'
|
|
file.write(result_message)
|
|
|
|
with open('./llm-answer/passages.txt', mode='a+', encoding='utf-8') as file:
|
|
result_message = f'متن پرامپت: {query.strip()}\n\مواد مشابه: {result} \n----------------------------------------------------------\n'
|
|
file.write(result_message)
|
|
|
|
|
|
|
|
print('----------------------------------------------------------')
|
|
print(f'full duration: {(datetime.datetime.now() - start).total_seconds()}')
|
|
print('----------------------------------------------------------')
|
|
print('~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~')
|
|
|