#!/usr/bin/env python3
from __future__ import annotations
import hashlib,json,subprocess,tarfile
from pathlib import Path
R=Path(__file__).resolve().parent;S=R/'source';E=R/'source_extract';V='2606.02073v1'
def sha(p):return hashlib.sha256(p.read_bytes()).hexdigest()
def main():
 S.mkdir(parents=True,exist_ok=True);E.mkdir(parents=True,exist_ok=True);pdf=S/f'{V}.pdf';arc=S/f'{V}.tar'
 subprocess.run(['curl','-fsSL','--retry','3',f'https://arxiv.org/pdf/{V}','-o',str(pdf)],check=True);subprocess.run(['curl','-fsSL','--retry','3',f'https://arxiv.org/e-print/{V}','-o',str(arc)],check=True)
 with tarfile.open(arc,'r:*') as h:
  for m in h.getmembers():
   d=(E/m.name).resolve()
   if E.resolve() not in d.parents and d!=E.resolve():raise ValueError(m.name)
  h.extractall(E,filter='data')
 print(json.dumps({'paper_id':'nbU2LNYdZN','arxiv_id':V,'pdf_sha256':sha(pdf),'source_sha256':sha(arc)},indent=2))
if __name__=='__main__':main()