verify_project.py

from pathlib import Path
import py_compile
import json

ROOT = Path(__file__).resolve().parent

required_files = [
    "README.md",
    "backend/main.py",
    "backend/requirements.txt",
    "frontend/package.json",
    "frontend/src/App.jsx",
    "frontend/src/main.jsx",
    "notebooks/01_eda_and_embeddings.ipynb",
    "notebooks/02_search_engine.ipynb",
    "notebooks/03_pdf_paper_analyzer.ipynb",
    "notebooks/04_recommendation_system.ipynb",
]

print("Checking required files...")
missing = []
for file in required_files:
    path = ROOT / file
    if not path.exists():
        missing.append(file)
    else:
        print("OK:", file)

if missing:
    raise SystemExit(f"Missing files: {missing}")

print("\nChecking Python syntax...")
python_files = list((ROOT / "backend").rglob("*.py"))
for path in python_files:
    # Let py_compile manage its own cache path to avoid Windows temp-file locking issues.
    py_compile.compile(str(path), doraise=True)
    print("OK:", path.relative_to(ROOT))

print("\nChecking notebooks...")
for path in (ROOT / "notebooks").glob("*.ipynb"):
    with open(path, "r", encoding="utf-8") as f:
        nb = json.load(f)
    cells = nb.get("cells", [])
    if not cells:
        raise SystemExit(f"Notebook has no cells: {path}")
    print("OK:", path.name, "-", len(cells), "cells")

print("\nProject validation passed.")