ui/chat_handler.py

import os
from utils.central_logging import setup_logging,get_logger
import textwrap
from langchain_openai import OpenAI
from langchain_chroma import Chroma
#from langchain_community.document_loaders import SeleniumURLLoader
from dotenv import load_dotenv
import os
import openai


from langchain_openai import ChatOpenAI
from langchain_core.runnables import RunnableLambda
import chromadb

import gradio as gr
import time
import asyncio
import nest_asyncio
import threading
import re
from openai import OpenAI
#import streamlit as st

from whisper_singleton import get_embedding,save_file,transcribe_content
from extract_text import pdf_to_documents,store_data
from prompt import get_prompt,get_system_prompt


load_dotenv("./.env")

setup_logging()
logger = get_logger("chat")


_embedding = None
_retriever = None
_vectore_store = None

openai_api_key  = os.getenv("OPENAI_API_KEY")

if openai_api_key:
    logger.info("Open ai api key has been set")
else:
    logger.error("No open ai api key has been found")




try:
    llm_openai = ChatOpenAI(model='gpt-3.5-turbo',temperature=0)
    client = OpenAI()
    logger.info("Clients has been initialized")
except Exception  as e:
    logger.exception(f"An exception occured: {e}")



def handle_upload(file_path):
    global _embedding
    global _retriever
    _embedding = get_embedding()
    text_content = ""
    status_message = ""
    file_name = "./transcribe.txt"
    try:
        if file_path.lower().endswith(".pdf"):
            
            collection_name = "pdffiles"
            pdf_docs,_vectore_store =  pdf_to_documents(file_path,"transcribe_db",collection_name,_embedding)
            text_content = "\n\n".join([doc.page_content for doc in pdf_docs])
            status_message = "📄 PDF file uploaded — extraction implemented."
            logger.info(status_message)
            #save_file(file_name,text_content)
        elif file_path.lower().endswith(".mp3") or file_path.lower().endswith('.mp4'):
            print(f"path:{file_path}")
            if file_path.lower().endswith(".mp3"):
                collection_name = "audios"
                status_message = "🎧 MP3 uploaded — transcription implemented."
                logger.info(status_message)
            else:
                collection_name = "videos"
                status_message = "🎬 MP4 uploaded — video transcription implemented."
                logger.info(status_message)
            
            text_content = transcribe_content(file_path)
            _vectore_store = store_data(text_content,"transcribe_db",collection_name,_embedding)
            #save_file(file_name,text_content)
        else:
            status_message = "Invalid file format"
    except Exception as e:
        status_message = f"❌ Error processing file: {e}"
        logger.exception(status_message)
    _retriever = _vectore_store.as_retriever()
    return status_message,text_content



def stream_response(user_input,history):
    
    history = history or []

    history.append({"role": "user", "content": user_input})
    history.append({"role": "assistant", "content": ""})
    
    context = ""
    if _retriever is not None:
        docs = _retriever.invoke(user_input)
        context = "\n\n".join([d.page_content for d in docs])

    formatted_history = "\n".join(
        f"{m['role'].capitalize()}: {m['content']}"
        for m in history
    )

    

    system_prompt = get_system_prompt().format(
        history=formatted_history,
        context=context,
        user_message=user_input
    )

    messages = [
        {"role": "system", "content": system_prompt},
        {"role": "user", "content": user_input},
    ]

    partial_reply = ""

    stream = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=messages,
        stream=True,
        temperature = 0
    )

    for event in stream:
        delta = event.choices[0].delta
        if delta and delta.content:
            token = delta.content
            partial_reply += token
            history[-1]["content"] = partial_reply 
            yield history, history, "" 

    history[-1]["content"] = partial_reply
    yield history, history, ""