_tools/restore_upstream_audio.py

"""Restore exact-name audio assets from upstream when the catalog mirror returns 404.
Provenance is explicit; equivalence with the missing mirror bytes is not asserted.
"""
import os,json,hashlib,requests,tempfile
from pathlib import Path
from huggingface_hub import HfApi
R=Path('/home/user/game-archive/reports');B='smodusermc/offline-games';api=HfApi(token=os.environ['HF_TOKEN']);report=[]
plans=[(6,'games/30dolar/assets/'+n+'.wav','https://thirtydollar.website/assets/'+n+'.wav') for n in ['you_deadass_built_like_a','white_people_be_like']]
plans +=[(101,'games/cookie-clicker/snd/'+n+'.mp3','https://orteil.dashnet.org/cookieclicker/snd/'+n+'.mp3') for n in ['tickOff','clickOn2','clickOff2','squeak1','squeak2','squeak3','squeak4']]
manifests={i:json.loads((R/f'game-{i}.json').read_text()) for i in [6,101]}
with tempfile.TemporaryDirectory(prefix='upstream-audio-') as tmp:
 for i,path,url in plans:
  response=requests.get(url,timeout=25);response.raise_for_status();body=response.content
  assert len(body)<5*1024**2 and not body.lstrip().lower().startswith((b'<html',b'<!doctype'))
  local=Path(tmp)/'audio';local.write_bytes(body);key='site/'+path;sha=hashlib.sha256(body).hexdigest()
  api.batch_bucket_files(B,add=[(local,key)]);back=Path(tmp)/'back';api.download_bucket_files(B,[(key,back)],raise_on_missing_files=True)
  assert hashlib.sha256(back.read_bytes()).hexdigest()==sha
  f={'path':key,'url':url,'bytes':len(body),'sha256':sha,'content_type':'audio/wav' if path.endswith('.wav') else 'audio/mpeg','provenance':{'original_url':'https://garbsoftball.com/'+path,'original_status':404,'replacement':'same filename on upstream project/creator asset host','byte_equivalence_to_missing_original':'not established'}}
  records={x['path']:x for x in manifests[i]['files']};records[key]=f;manifests[i]['files']=list(records.values());report.append({'game_id':i,**f});local.unlink();back.unlink()
for i,m in manifests.items():(R/f'game-{i}.json').write_text(json.dumps(m,indent=2))
(R/'upstream-audio-restorations.json').write_text(json.dumps(report,indent=2))
api.batch_bucket_files(B,add=[(R/'upstream-audio-restorations.json','_verification/upstream-audio-restorations.json')]+[(R/f'game-{i}.json',f'_reports/games/{i:04d}.json') for i in manifests])
print('Restored and downloaded-back SHA256 verified',len(report),'upstream audio files')