app.py

import flask
from flask import Flask, request, jsonify
#import streamlit as st
from langchain.chat_models import ChatOpenAI
from langchain.document_loaders import WebBaseLoader
from langchain.chains.summarize import load_summarize_chain
from bs4 import BeautifulSoup
import os
from dotenv import load_dotenv
load_dotenv()
from langchain import HuggingFaceHub
from huggingface_hub import InferenceClient
import requests
import sys

#OPENAI_API_KEY = os.environ.get('OPENAI_API_KEY')

hf_token = os.environ.get('HUGGINGFACEHUB_API_TOKEN')
#starchat_repo_id = os.environ.get('starchat_repo_id')
repo_id=os.environ.get('repo_id')
#port = os.getenv('port')

llm = HuggingFaceHub(repo_id=repo_id,  #for Llama2
                     #repo_id=starchat_repo_id,     #for StarChat                  
                     huggingfacehub_api_token=hf_token,
                     model_kwargs={#"min_length":512,  #for StarChat
                                   "min_length":1024,  #for Llama2
                                   "max_new_tokens":3072, "do_sample":True,  #for StarChat
                                   #"max_new_tokens":5632, "do_sample":True,  #for Llama2                                 
                                   "temperature":0.1,
                                   "top_k":50,
                                   "top_p":0.95, "eos_token_id":49155}) 
#llm = ChatOpenAI(temperature=0, model_name="gpt-3.5-turbo-16k")
chain = load_summarize_chain(llm, chain_type="stuff")

app = Flask(__name__)
@app.route('/', methods=['POST'])
def home_api():
    data = request.get_json()
    user_query = data['user_question']
    print(user_query)
    return {"Message":"Flask Home API Deploy Success on HF"}  

@app.route('/api/chat', methods=['POST'])
def chat():
#async def chat():  #Failed - Flask API 不支持async???
    data = request.get_json()
    user_query = data['user_question']
    if user_query !="" and not user_query.strip().isspace() and not user_query == "" and not user_query.strip() == "" and not user_query.isspace():
    try:
        loader = WebBaseLoader(user_query)
        with st.spinner("AI Thinking...Please wait a while to Cheers!"):
            docs = loader.load()
            result=chain.run(docs)
            print(user_query)
            print("AI Summarization: "+result)
            #st.write("AI Summarization:")
            #st.write(result)
            return jsonify({'response': result})
    except Exception as e:
        #st.write("Wrong URL or URL not parsable.") 
        err_msg="Wrong URL or URL not parsable."
        return jsonify({'response': err_msg})
        
    #initial_response = llm_chain.run(user_query)
    #return jsonify({'response': initial_response})
    #找到问题了：jsonify在Huggingface不支持；在Github然后部署到Render是可以的！---NO No No, it's supported
    #return {'response': initial_response}
    #return jsonify({'response': initial_response})  #tried and OKed!
    

#url=st.text_input("Enter webiste URL to summarize (format: https://www.usinoip.com):")

#loader = WebBaseLoader("https://www.usinoip.com/")

#if url !="" and not url.strip().isspace() and not url == "" and not url.strip() == "" and not url.isspace():
#    try:
#        loader = WebBaseLoader(url)
#        with st.spinner("AI Thinking...Please wait a while to Cheers!"):
#            docs = loader.load()
#            result=chain.run(docs)
#            print(url)
#            print("AI Summarization: "+result)
#            st.write("AI Summarization:")
#            st.write(result)
#    except Exception as e:
#        st.write("Wrong URL or URL not parsable.")    
#**************************************************************#

#try:
#    loader = WebBaseLoader(url)
#    with st.spinner("AI Thinking...Please wait a while to Cheers!"):
#        docs = loader.load()
#        result=chain.run(docs)
#        print(result)
#        st.write("AI Summarization:")
#        st.write(result)
#except Exception as e:
#    st.write("Wrong URL")