interim.py

import streamlit as st
import os
from openai import OpenAI
import tempfile
from langchain.chains import ConversationalRetrievalChain
from langchain_openai import ChatOpenAI, OpenAIEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.vectorstores import Chroma
from langchain_community.document_loaders import (
    PyPDFLoader, 
    TextLoader, 
    CSVLoader
)
from datetime import datetime
import pytz

from langchain.chains import ConversationalRetrievalChain
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_openai import ChatOpenAI, OpenAIEmbeddings
from langchain_community.vectorstores import Chroma
from langchain_community.document_loaders import PyPDFLoader, TextLoader, CSVLoader
import os
import tempfile
from datetime import datetime
import pytz


class DocumentRAG:
    def __init__(self):
        self.document_store = None
        self.qa_chain = None
        self.document_summary = ""
        self.chat_history = []
        self.last_processed_time = None
        self.api_key = os.getenv("OPENAI_API_KEY")  # Fetch the API key from environment variable
        self.init_time = datetime.now(pytz.UTC)

        if not self.api_key:
            raise ValueError("API Key not found. Make sure to set the 'OPENAI_API_KEY' environment variable.")

        # Persistent directory for Chroma to avoid tenant-related errors
        self.chroma_persist_dir = "./chroma_storage"
        os.makedirs(self.chroma_persist_dir, exist_ok=True)

    def process_documents(self, uploaded_files):
        """Process uploaded files by saving them temporarily and extracting content."""
        if not self.api_key:
            return "Please set the OpenAI API key in the environment variables."
        if not uploaded_files:
            return "Please upload documents first."

        try:
            documents = []
            for uploaded_file in uploaded_files:
                # Save uploaded file to a temporary location
                temp_file_path = tempfile.NamedTemporaryFile(delete=False, suffix=os.path.splitext(uploaded_file.name)[1]).name
                with open(temp_file_path, "wb") as temp_file:
                    temp_file.write(uploaded_file.read())

                # Determine the loader based on the file type
                if temp_file_path.endswith('.pdf'):
                    loader = PyPDFLoader(temp_file_path)
                elif temp_file_path.endswith('.txt'):
                    loader = TextLoader(temp_file_path)
                elif temp_file_path.endswith('.csv'):
                    loader = CSVLoader(temp_file_path)
                else:
                    return f"Unsupported file type: {uploaded_file.name}"

                # Load the documents
                try:
                    documents.extend(loader.load())
                except Exception as e:
                    return f"Error loading {uploaded_file.name}: {str(e)}"

            if not documents:
                return "No valid documents were processed. Please check your files."

            # Split text for better processing
            text_splitter = RecursiveCharacterTextSplitter(
                chunk_size=1000,
                chunk_overlap=200,
                length_function=len
            )
            documents = text_splitter.split_documents(documents)

            # Combine text for summary
            combined_text = " ".join([doc.page_content for doc in documents])
            self.document_summary = self.generate_summary(combined_text)

            # Create embeddings and initialize retrieval chain
            embeddings = OpenAIEmbeddings(api_key=self.api_key)
            self.document_store = Chroma.from_documents(
                documents,
                embeddings,
                persist_directory=self.chroma_persist_dir  # Persistent directory for Chroma
            )

            self.qa_chain = ConversationalRetrievalChain.from_llm(
                ChatOpenAI(temperature=0, model_name='gpt-4', api_key=self.api_key),
                self.document_store.as_retriever(search_kwargs={'k': 6}),
                return_source_documents=True,
                verbose=False
            )

            self.last_processed_time = datetime.now(pytz.UTC)
            return "Documents processed successfully!"
        except Exception as e:
            return f"Error processing documents: {str(e)}"

    def generate_summary(self, text):
        """Generate a summary of the provided text."""
        if not self.api_key:
            return "API Key not set. Please set it in the environment variables."
        try:
            client = OpenAI(api_key=self.api_key)
            response = client.chat.completions.create(
                model="gpt-4",
                messages=[
                    {"role": "system", "content": "Summarize the document content concisely and provide 3-5 key points for discussion."},
                    {"role": "user", "content": text[:4000]}
                ],
                temperature=0.3
            )
            return response.choices[0].message.content
        except Exception as e:
            return f"Error generating summary: {str(e)}"

    def handle_query(self, question, history):
        """Handle user queries."""
        if not self.qa_chain:
            return history + [("System", "Please process the documents first.")]
        try:
            preface = """
            Instruction: Respond in English. Be professional and concise, keeping the response under 300 words.
            If you cannot provide an answer, say: "I am not sure about this question. Please try asking something else."
            """
            query = f"{preface}\nQuery: {question}"

            result = self.qa_chain({
                "question": query,
                "chat_history": [(q, a) for q, a in history]
            })

            if "answer" not in result:
                return history + [("System", "Sorry, an error occurred.")]

            history.append((question, result["answer"]))
            return history
        except Exception as e:
            return history + [("System", f"Error: {str(e)}")]

# Initialize RAG system in session state
if "rag_system" not in st.session_state:
    st.session_state.rag_system = DocumentRAG()

# Streamlit UI
st.title("Document Analyzer and Podcast Generator")

# Fetch the API key status
if "OPENAI_API_KEY" not in os.environ or not os.getenv("OPENAI_API_KEY"):
    st.error("The 'OPENAI_API_KEY' environment variable is not set. Please configure it in your hosting environment.")

# File upload
st.subheader("Step 1: Upload Documents")
uploaded_files = st.file_uploader("Upload files (PDF, TXT, CSV)", accept_multiple_files=True)

if st.button("Process Documents"):
    if uploaded_files:
        # Process the uploaded files
        result = st.session_state.rag_system.process_documents(uploaded_files)
        if "successfully" in result:
            st.success(result)
        else:
            st.error(result)
    else:
        st.warning("No files uploaded.")

# Document Q&A
st.subheader("Step 2: Ask Questions")
if st.session_state.rag_system.qa_chain:
    history = []
    user_question = st.text_input("Ask a question:")
    if st.button("Submit Question"):
        # Handle the user query
        history = st.session_state.rag_system.handle_query(user_question, history)
        for question, answer in history:
            st.chat_message("user").write(question)
            st.chat_message("assistant").write(answer)
else:
    st.info("Please process documents before asking questions.")