app.py

import os
import base64
import gc
import uuid
from pathlib import Path
from llama_index.core import Settings
from llama_index.llms.ollama import Ollama
from llama_index.embeddings.fastembed import FastEmbedEmbedding
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, get_response_synthesizer
from llama_index.core import DocumentSummaryIndex
from llama_index.core.node_parser import SentenceSplitter
import streamlit as st

# Initialize session state
if "id" not in st.session_state:
    st.session_state.id = uuid.uuid4()
    st.session_state.file_cache = {}
    st.session_state.processing = False
    st.session_state.messages = []

session_id = st.session_state.id

embed_model = FastEmbedEmbedding(model_name="BAAI/bge-large-en-v1.5")
llama3 = Ollama(model="llama3", request_timeout=60.0)

Settings.llm = llama3
Settings.embed_model = embed_model

# Ensure the documents directory exists
os.makedirs("./documents", exist_ok=True)


# Load existing documents from the directory
def load_existing_documents():
    for file_name in os.listdir("./documents"):
        file_path = os.path.join("./documents", file_name)
        if os.path.isfile(file_path) and file_name.endswith(".pdf"):
            st.session_state.file_cache[file_name] = file_path


def reset_app():
    st.session_state.file_cache = {}
    st.session_state.messages = []
    gc.collect()


def display_pdf(file_path):
    with open(file_path, "rb") as f:
        base64_pdf = base64.b64encode(f.read()).decode("utf-8")
    pdf_display = f"""<iframe src="data:application/pdf;base64,{base64_pdf}" width="400" height="100%" type="application/pdf" style="height:100vh; width:100%"></iframe>"""
    st.markdown(pdf_display, unsafe_allow_html=True)


def process_and_summarize_selected_doc(file_path):
    file_name = Path(file_path).stem
    docs = SimpleDirectoryReader(input_files=[file_path]).load_data()[:1]
    docs[0].doc_id = file_name

    splitter = SentenceSplitter(chunk_size=4096)

    response_synthesizer = get_response_synthesizer(
        response_mode="tree_summarize", use_async=True
    )

    doc_summary_index = DocumentSummaryIndex.from_documents(
        docs,
        transformations=[splitter],
        response_synthesizer=response_synthesizer,
        show_progress=True,
        # streaming=True,
    )

    return doc_summary_index, doc_summary_index.get_document_summary(file_name)


def get_query_engine(doc_summary_index):
    return doc_summary_index.as_query_engine(
        response_mode="tree_summarize", use_async=True, streaming=True
    )


def reset_chat():
    st.session_state.messages = []
    gc.collect()


# Load existing documents when the app loads
load_existing_documents()

with st.sidebar:
    st.header("Upload your documents!")
    uploaded_file = st.file_uploader("Choose your `.pdf` file", type="pdf")

    if uploaded_file:
        try:
            file_path = os.path.join("./documents", uploaded_file.name)
            with open(file_path, "wb") as f:
                f.write(uploaded_file.getvalue())

            file_key = uploaded_file.name

            if file_key not in st.session_state.get("file_cache", {}):
                st.session_state.file_cache[file_key] = file_path

            st.success("File uploaded and saved successfully!")
            display_pdf(file_path)

        except Exception as e:
            st.error(f"An error occurred: {e}")
            st.stop()

col1, col2 = st.columns([3, 1])

with col1:
    st.header("Documents Summarizer! 🚀")
    if st.session_state.get("file_cache"):
        with st.expander("Select Document"):
            selected_file_key = st.radio("", list(st.session_state.file_cache.keys()))
            selected_file_path = st.session_state.file_cache[selected_file_key]
    else:
        st.write("No documents uploaded yet. Please upload a document to get started.")

    if st.button("Summarize"):
        if "selected_file_path" in locals():
            st.session_state.processing = True
            with st.spinner('Processing...'):
                doc_summary_index, summary = process_and_summarize_selected_doc(selected_file_path)
            st.session_state.processing = False
            st.markdown("### Summary")
            st.write(summary)

            # Initialize the query engine
            query_engine = get_query_engine(doc_summary_index)
            st.session_state.query_engine = query_engine
        else:
            st.error("Please select a document to summarize.")

with col2:
    if st.button("Clear ↺"):
        reset_app()
        st.experimental_rerun()

if "query_engine" in st.session_state:
    st.header("Chat for follow up question! 💬")
    if "messages" not in st.session_state:
        reset_chat()

    # Display chat messages from history on app rerun
    for message in st.session_state.messages:
        with st.chat_message(message["role"]):
            st.markdown(message["content"])

    # Accept user input
    if prompt := st.chat_input("Ask something about the document ..."):
        # Add user message to chat history
        st.session_state.messages.append({"role": "user", "content": prompt})
        # Display user message in chat message container
        with st.chat_message("user"):
            st.markdown(prompt)

        # Display assistant response in chat message container
        with st.chat_message("assistant"):
            message_placeholder = st.empty()
            full_response = ""

            # Simulate stream of response with milliseconds delay
            streaming_response = st.session_state.query_engine.query(prompt)

            for chunk in streaming_response.response_gen:
                full_response += chunk
                message_placeholder.markdown(full_response + "▌")

            message_placeholder.markdown(full_response)

        # Add assistant response to chat history
        st.session_state.messages.append({"role": "assistant", "content": full_response})