app.py

import gradio as gr
import os
from groq import Groq

from langchain_community.document_loaders import PyPDFLoader
from langchain_community.vectorstores import FAISS
from langchain_community.embeddings import SentenceTransformerEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter


# ---------------- CONFIG ----------------
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
client = Groq(api_key=GROQ_API_KEY)

embeddings = SentenceTransformerEmbeddings(
    model_name="all-MiniLM-L6-v2"
)

vector_store = None
chat_history = []

# ---------------- FUNCTIONS ----------------

def process_pdfs(files):
    global vector_store

    docs = []
    for file in files:
        loader = PyPDFLoader(file.name)
        docs.extend(loader.load())

    splitter = RecursiveCharacterTextSplitter(
        chunk_size=800,
        chunk_overlap=150
    )
    chunks = splitter.split_documents(docs)

    vector_store = FAISS.from_documents(chunks, embeddings)
    return "✅ PDFs processed successfully!"

def ask_question(question):
    global chat_history, vector_store

    try:
        # ---------- Check 1: PDFs processed ----------
        if vector_store is None:
            return "❌ Please upload and process PDF files first."

        if question.strip() == "":
            return "❌ Please enter a valid question."

        # ---------- Retrieve documents ----------
        docs = vector_store.similarity_search(question, k=4)

        if not docs:
            return "⚠️ No relevant information found in the documents."

        context = ""
        sources = []

        for d in docs:
            context += d.page_content + "\n"
            page = d.metadata.get("page", "N/A")
            sources.append(f"Page {page + 1}" if page != "N/A" else "Unknown page")

        # ---------- Prompt ----------
        prompt = f"""
You are a helpful assistant.
Answer ONLY using the context below.

Context:
{context}

Question:
{question}
"""

        # ---------- Groq API ----------
        response = client.chat.completions.create(
    model="llama-3.1-8b-instant",
    messages=[
        {"role": "system", "content": "Answer strictly from the documents."},
        {"role": "user", "content": prompt}
    ]
)

        # ---------- Safe response extraction ----------
        answer = response.choices[0].message.content.strip()

        final_answer = f"{answer}\n\n📄 Sources: {', '.join(set(sources))}"
        chat_history.append((question, final_answer))

        return final_answer

    except Exception as e:
        return f"❌ Error occurred: {str(e)}"

# ---------------- UI ----------------

with gr.Blocks() as demo:
    gr.Markdown("## 📘Kashif's Enhanced RAG-Based Chatbot")

    with gr.Row():
        pdfs = gr.File(file_types=[".pdf"], file_count="multiple")
        upload_btn = gr.Button("Process PDFs")

    status = gr.Textbox()
    upload_btn.click(process_pdfs, pdfs, status)

    question = gr.Textbox(label="Ask a Question")
    answer = gr.Textbox(label="Answer")

    ask_btn = gr.Button("Ask")
    ask_btn.click(ask_question, question, answer)

demo.launch()