app.py

import os
import gradio as gr

from langchain_community.document_loaders import PyPDFLoader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_groq import ChatGroq
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser


# ───────────────────────── CONFIG ─────────────────────────
EMBED_MODEL = "sentence-transformers/all-MiniLM-L6-v2"
GROQ_MODEL = "llama-3.1-8b-instant"
TOP_K = 3

os.environ["GROQ_API_KEY"] = os.getenv("GROQ_API_KEY")


# ───────────────────────── INIT MODELS ─────────────────────────
embeddings = HuggingFaceEmbeddings(
    model_name=EMBED_MODEL,
    model_kwargs={"device": "cpu"},
    encode_kwargs={"normalize_embeddings": True}
)


def create_llm():
    return ChatGroq(
        model=GROQ_MODEL,
        temperature=0.2,
        max_tokens=1024,
        groq_api_key=os.environ["GROQ_API_KEY"]
    )


RAG_PROMPT = ChatPromptTemplate.from_template("""
You are a helpful assistant.
Answer ONLY using the context below.
If not found, say you don't have enough information.
Context:
{context}
Question: {question}
Answer:
""")


def format_docs(docs):
    return "\n\n".join(d.page_content for d in docs)


# ───────────────────────── GLOBAL STATE ─────────────────────────
vectorstore = None
rag_chain = None


# ───────────────────────── PROCESS PDF ─────────────────────────
def process_pdf(file):
    global vectorstore, rag_chain

    if file is None:
        return "Upload a PDF first."

    path = file.name

    # Load
    loader = PyPDFLoader(path)
    docs = loader.load()

    # Split
    splitter = RecursiveCharacterTextSplitter(
        chunk_size=500,
        chunk_overlap=50
    )
    chunks = splitter.split_documents(docs)

    # Vector store
    if vectorstore is None:
        vectorstore = FAISS.from_documents(chunks, embeddings)
    else:
        vectorstore.add_documents(chunks)

    retriever = vectorstore.as_retriever(search_kwargs={"k": TOP_K})

    llm = create_llm()

    rag_chain = (
        {
            "context": retriever | format_docs,
            "question": RunnablePassthrough()
        }
        | RAG_PROMPT
        | llm
        | StrOutputParser()
    )

    return f"✅ PDF processed successfully!\nChunks: {len(chunks)}"


# ───────────────────────── CHAT FUNCTION ─────────────────────────
def chat(message, history):

    if rag_chain is None:
        history.append({"role": "user", "content": message})
        history.append({"role": "assistant", "content": "Please upload a PDF first."})
        return "", history

    response = rag_chain.invoke(message)

    history.append({"role": "user", "content": message})
    history.append({"role": "assistant", "content": response})

    return "", history


# ───────────────────────── UI ─────────────────────────
with gr.Blocks(title="RAG Chatbot") as demo:

    gr.Markdown("## 📄 PDF RAG Chatbot (Groq + FAISS + LangChain)")

    with gr.Row():
        file = gr.File(label="Upload PDF")
        upload_btn = gr.Button("Process PDF")

    status = gr.Textbox(label="Status")

    chatbot = gr.Chatbot()
    msg = gr.Textbox(label="Ask a question")

    upload_btn.click(process_pdf, inputs=file, outputs=status)
    msg.submit(chat, inputs=[msg, chatbot], outputs=[msg, chatbot])

demo.launch()