appv0.py

import os
import streamlit as st
from langchain_community.document_loaders import PDFPlumberLoader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain.vectorstores import FAISS
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.prompts import ChatPromptTemplate
from langchain.chains import LLMChain
from langchain.llms import CTransformers

# === Configuration ===
pdfs_directory = 'pdfs'
vectorstores_directory = 'vectorstores'
os.makedirs(pdfs_directory, exist_ok=True)
os.makedirs(vectorstores_directory, exist_ok=True)

PREDEFINED_BOOKS = [f for f in os.listdir(pdfs_directory) if f.endswith(".pdf")]

TEMPLATE = """
You are a helpful assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question. 
If you don't know the answer, say "I don't know". Limit your answer to three concise sentences.

Question: {question}
Context: {context}
Answer:
"""

# === Embeddings and LLM (CPU-friendly) ===
embedding_model = HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2')

llm = CTransformers(
    model='TheBloke/Mistral-7B-Instruct-v0.1-GGUF',
    model_file='mistral-7b-instruct-v0.1.Q4_K_M.gguf',
    model_type='mistral',
    config={'max_new_tokens': 512, 'temperature': 0.5}
)

# === Functions ===
def upload_pdf(file):
    save_path = os.path.join(pdfs_directory, file.name)
    with open(save_path, "wb") as f:
        f.write(file.getbuffer())
    return file.name

def load_pdf(file_path):
    loader = PDFPlumberLoader(file_path)
    return loader.load()

def split_text(documents):
    splitter = RecursiveCharacterTextSplitter(
        chunk_size=1000,
        chunk_overlap=200,
        add_start_index=True
    )
    return splitter.split_documents(documents)

def get_vectorstore_path(book_filename):
    base_name = os.path.splitext(book_filename)[0]
    return os.path.join(vectorstores_directory, base_name)

def load_or_create_vectorstore(book_filename, documents=None):
    vs_path = get_vectorstore_path(book_filename)
    
    if os.path.exists(os.path.join(vs_path, "index.faiss")):
        return FAISS.load_local(vs_path, embedding_model, allow_dangerous_deserialization=True)

    if documents is None:
        raise ValueError("Documents must be provided when creating a new vectorstore.")

    os.makedirs(vs_path, exist_ok=True)
    chunks = split_text(documents)
    vector_store = FAISS.from_documents(chunks, embedding_model)
    vector_store.save_local(vs_path)
    return vector_store

def retrieve_docs(vector_store, query):
    return vector_store.similarity_search(query)

def answer_question(question, documents):
    context = "\n\n".join(doc.page_content for doc in documents)
    prompt = ChatPromptTemplate.from_template(TEMPLATE)
    chain = LLMChain(llm=llm, prompt=prompt)
    return chain.run({"question": question, "context": context})

# === UI ===
st.set_page_config(page_title="📄 PDF Q&A (Cached FAISS)", layout="centered")
st.title("📚 Chat with PDF - Cached Vector Stores")

with st.sidebar:
    st.header("Select or Upload a Book")
    selected_book = st.selectbox("Choose a PDF", PREDEFINED_BOOKS + ["Upload new book"])

    if selected_book == "Upload new book":
        uploaded_file = st.file_uploader("Upload PDF", type="pdf")
        if uploaded_file:
            filename = upload_pdf(uploaded_file)
            st.success(f"Uploaded: {filename}")
            selected_book = filename

# === Load or Create Vector Store ===
if selected_book and selected_book != "Upload new book":
    file_path = os.path.join(pdfs_directory, selected_book)
    vectorstore_path = get_vectorstore_path(selected_book)

    try:
        if os.path.exists(os.path.join(vectorstore_path, "index.faiss")):
            st.info("✅ Using cached vector store.")
            vector_store = load_or_create_vectorstore(selected_book)
        else:
            st.warning("⏳ Creating new vector store (first-time load)...")
            documents = load_pdf(file_path)
            vector_store = load_or_create_vectorstore(selected_book, documents)
            st.success("✅ Vector store created and cached.")

        question = st.chat_input("Ask a question about the book...")
        if question:
            st.chat_message("user").write(question)
            related_docs = retrieve_docs(vector_store, question)
            answer = answer_question(question, related_docs)
            st.chat_message("assistant").write(answer)
    except Exception as e:
        st.error(f"❌ Error loading PDF or vector store: {e}")