gabriel / musehub public
count_compressed.py python
68 lines 2.0 KB
Raw
sha256:34035d72cef530c1ab9d6a6f53be18d803dde705fc3157617d70352a96d0747b Merge branch 'fix/wire-push-external-parent-manifest' into dev Human 8 days ago
1 """Count how many objects in R2 are still zlib-compressed.
2
3 Prints a progress line every 100 objects and a final summary.
4 Run: docker exec musehub-blue python3 /app/deploy/count_compressed.py
5 """
6 import sys
7
8 import sqlalchemy as sa
9 from sqlalchemy import create_engine
10 from sqlalchemy.orm import Session
11
12 from musehub.config import settings
13 from musehub.db.musehub_repo_models import MusehubObject
14 from musehub.storage import get_backend
15
16 ZLIB_MAGIC = (b"\x78\x01", b"\x78\x9c", b"\x78\xda", b"\x78\x5e")
17
18
19 def check_header(backend, oid: str) -> bool:
20 """Return True if the object starts with a zlib magic header."""
21 client = backend._get_client()
22 key = backend._key(oid)
23 try:
24 resp = client.get_object(Bucket=backend._bucket, Key=key, Range="bytes=0-1")
25 header = resp["Body"].read(2)
26 return header in ZLIB_MAGIC
27 except Exception as e:
28 print(f" ERROR {oid[:20]}: {e}", flush=True)
29 return False
30
31
32 def main() -> None:
33 # Sync engine — no asyncio, no threads, no surprises.
34 sync_url = settings.database_url.replace("+asyncpg", "").replace("+aiosqlite", "")
35 engine = create_engine(sync_url)
36 backend = get_backend()
37
38 with Session(engine) as session:
39 rows = session.execute(
40 sa.select(MusehubObject.object_id)
41 .where(
42 MusehubObject.storage_uri.like("s3://%"),
43 MusehubObject.deleted_at.is_(None),
44 )
45 .order_by(MusehubObject.object_id)
46 ).scalars().all()
47
48 total = len(rows)
49 print(f"Total objects: {total}", flush=True)
50
51 compressed = 0
52 plain = 0
53
54 for i, oid in enumerate(rows, 1):
55 if check_header(backend, oid):
56 compressed += 1
57 else:
58 plain += 1
59
60 if i % 100 == 0 or i == total:
61 print(f" [{i}/{total}] plain={plain} compressed={compressed}", flush=True)
62
63 print(f"\nDone. plain={plain} compressed={compressed}", flush=True)
64 sys.exit(1 if compressed else 0)
65
66
67 if __name__ == "__main__":
68 main()
File History 14 commits
sha256:34035d72cef530c1ab9d6a6f53be18d803dde705fc3157617d70352a96d0747b Merge branch 'fix/wire-push-external-parent-manifest' into dev Human 8 days ago
sha256:fc04e4cae9e1774d6a21b65c45daeed0e6787eb581d13aa1b03bfe9384a34226 Merge branch 'fix/two-column-scroll-layout' into dev Human 8 days ago
sha256:408916fc5973ba59c6e4eebaa80ebdcc801c0a63205651e25009d11548f79454 chore: bump version to 0.2.0.dev2 — nightly.2, matching muse Sonnet 4.6 patch 11 days ago
sha256:d035733f21ccff27735fddebfbbe0ed24565a32a22db8de5885402262671ecd2 chore: bump version to 0.2.0rc15 for musehub#113 fix release Sonnet 4.6 patch 14 days ago
sha256:0032d6cfa33bc3c8367436ad768e7dd0e339b4332153160247da8266cb5fa352 Merge branch 'task/version-tags-phase3-server' into dev Human 17 days ago
sha256:4669620efda9ff41c55bdefd1f7bfe1c239d468428744c84ead9957e5a003a53 merge: rescue snapshot-recovery hardening (c00aa21d) into d… Opus 4.8 minor 29 days ago
sha256:a59da49c4611b970fc4b6ae48678ce4943261c213a07ddbd73ce9201df869b4a fix: remove false-positive proposal_comments index drop fro… Sonnet 4.6 patch 33 days ago
sha256:0a240d6dbff234f07d98a28a4a9a68db702f3f9ff9260196f24219bdb1c0b6f3 feat: render markdown mists as HTML with heading anchor links Sonnet 4.6 patch 34 days ago
sha256:24a7d47486ebc4ebd1832830580e177ec6f877b48dced8c000e198cdec4ce9d6 Merge 'task/bump-version-rc12' into 'dev' — proposal: Bump … Human 35 days ago
sha256:b9ff931d147e0114a1f17060f415b89ed551c170a91ff226c70437aa5c85f9ee Merge 'task/bump-version-rc12' into 'dev' — proposal: Bump … Human 35 days ago
sha256:d1122d21e73471879b460037b22c0b50fded7c423444a176f248428f75dac39c Merge 'task/fix-issue-pagination-cursor' into 'dev' — propo… Human 35 days ago
sha256:01e18975e73d2b3cd5b6db7929c895bef9aa6e0d4391dc5b2adfc548b41318dd Merge 'feat/adding-debug-logs-to-staging' into 'dev' — prop… Human 35 days ago
sha256:6b1949fc2797ca4c1936a637a4cbfec828ef56cf52398a2e74ca3c4f494e728f fix: use wire_bytes not mpack_bytes_raw in compute_object_b… Sonnet 4.6 patch 48 days ago
sha256:b99f2455dc346966d040133f5203297e6e3ef5803a93728a2c30568d0a0f7583 rename: delta_add → delta_upsert across wire format, models… Sonnet 4.6 patch 50 days ago