app.py

from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.document_loaders import PyMuPDFLoader
from langchain.chains import RetrievalQA
from langchain.llms.base import LLM

from groq import Groq
from typing import List, Optional
import os
import gradio as gr



class GroqLLM(LLM):
    model: str = "llama3-8b-8192"
    api_key: str = "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"
    temperature: float = 0.0

    def _call(self, prompt: str, stop: Optional[List[str]] = None) -> str:
        client = Groq(api_key=self.api_key)
        messages = [
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt}
        ]
        response = client.chat.completions.create(
            model=self.model,
            messages=messages,
            temperature=self.temperature,
        )
        return response.choices[0].message.content

    @property
    def _llm_type(self) -> str:
        return "groq-llm"


def process_pdf(pdf_path):
    loader = PyMuPDFLoader(pdf_path)
    documents = loader.load()

    splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
    chunks = splitter.split_documents(documents)

    embedding = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
    vectorstore = FAISS.from_documents(chunks, embedding)

    retriever = vectorstore.as_retriever()
    llm = GroqLLM(api_key="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")

    qa = RetrievalQA.from_chain_type(
        llm=llm,
        retriever=retriever,
        return_source_documents=True
    )

    return qa



qa_chain = None

def upload_file(file):
    global qa_chain
    qa_chain = process_pdf(file.name)
    return "PDF processed! You can now ask questions."

def ask_question(query):
    if qa_chain is None:
        return "Please upload a PDF first."
    result = qa_chain({"query": query})
    return result["result"]


with gr.Blocks() as demo:
    gr.Markdown("# 🧠 PDF Q&A with GROQ + LangChain")
    with gr.Row():
        uploader = gr.File(label="Upload your PDF")
        status = gr.Textbox(label="Status")

    uploader.change(fn=upload_file, inputs=uploader, outputs=status)

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

    question.submit(fn=ask_question, inputs=question, outputs=answer)

demo.launch(share=True)