app.py

from fastapi import FastAPI, UploadFile, File, Form
from pydantic import BaseModel
import openai
import faiss
import numpy as np
import os
from dotenv import load_dotenv
from fastapi.middleware.cors import CORSMiddleware
from pypdf import PdfReader

load_dotenv()
openai.api_key = os.getenv("OPENAI_API_KEY")

app = FastAPI()

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Store documents and vectors per notebook
notebooks = {}

class Query(BaseModel):
    question: str
    notebook_id: str

@app.get("/")
def read_root():
    return {
        "message": "✅ NotebookLM OpenAI Backend is running!",
        "endpoints": {
            "/upload-pdf": "POST a PDF file with notebook_id",
            "/ask": "POST question + notebook_id to get answer"
        }
    }

@app.post("/ask")
def ask(query: Query):
    nb = notebooks.get(query.notebook_id)
    if not nb:
        return {"answer": "Notebook not found."}

    question_embedding = openai.Embedding.create(
        input=[query.question],
        model="text-embedding-ada-002"
    )["data"][0]["embedding"]

    if len(nb["texts"]) == 0:
        return {"answer": "No documents indexed in this notebook."}

    D, I = nb["index"].search(np.array([question_embedding]).astype("float32"), k=3)
    context = "\n\n".join([f"[{i+1}] {nb['texts'][i]}" for i in I[0]])
    citation_refs = [nb['citations'][i] for i in I[0]]

    response = openai.ChatCompletion.create(
        model="gpt-4",
        messages=[
            {"role": "system", "content": "You are an AI assistant that answers based on uploaded documents. Cite sources using [1], [2], etc."},
            {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {query.question}"}
        ],
        temperature=0.3
    )
    return {"answer": response.choices[0].message.content.strip(), "citations": citation_refs}

@app.post("/upload-pdf")
def upload_pdf(notebook_id: str = Form(...), file: UploadFile = File(...)):
    if notebook_id not in notebooks:
        notebooks[notebook_id] = {
            "index": faiss.IndexFlatL2(1536),
            "texts": [],
            "citations": []
        }

    nb = notebooks[notebook_id]
    reader = PdfReader(file.file)
    for i, page in enumerate(reader.pages):
        content = page.extract_text()
        if content:
            embedding = openai.Embedding.create(
                input=[content],
                model="text-embedding-ada-002"
            )["data"][0]["embedding"]
            nb["index"].add(np.array([embedding]).astype("float32"))
            nb["texts"].append(content)
            nb["citations"].append(f"{file.filename}, page {i+1}")

    return {"status": f"{file.filename} uploaded and parsed"}