backfill_content_ids_from_snapshots.py
python
sha256:fc04e4cae9e1774d6a21b65c45daeed0e6787eb581d13aa1b03bfe9384a34226
Merge branch 'fix/two-column-scroll-layout' into dev
Human
8 days ago
| 1 | #!/usr/bin/env python3 |
| 2 | """One-time backfill: fill missing content_id values on file-level symbol |
| 3 | history entries by reading the canonical source — the snapshot manifest. |
| 4 | |
| 5 | Background |
| 6 | ---------- |
| 7 | When the indexer processes a file-level move/rename op (structured_delta op |
| 8 | type 'patch' with from_address), it records no new_content_id at the symbol |
| 9 | level. The snapshot stored at that commit, however, has the full |
| 10 | {path: content_id} manifest. This script reads those manifests and fills |
| 11 | the gaps. |
| 12 | |
| 13 | Only file-level addresses (no '::' separator) are handled — they map directly |
| 14 | to a manifest path. Symbol-level addresses are not yet backfillable this way. |
| 15 | |
| 16 | Usage |
| 17 | ----- |
| 18 | # Dry-run: count rows that would be changed (no writes) |
| 19 | docker exec musehub-blue python3 /app/deploy/backfill_content_ids_from_snapshots.py --dry-run |
| 20 | |
| 21 | # Run for all repos |
| 22 | docker exec musehub-blue python3 /app/deploy/backfill_content_ids_from_snapshots.py |
| 23 | |
| 24 | # Run for a single repo |
| 25 | docker exec musehub-blue python3 /app/deploy/backfill_content_ids_from_snapshots.py --repo-id <repo_id> |
| 26 | |
| 27 | # Quiet (no progress output) |
| 28 | docker exec musehub-blue python3 /app/deploy/backfill_content_ids_from_snapshots.py -q |
| 29 | """ |
| 30 | from __future__ import annotations |
| 31 | |
| 32 | import argparse |
| 33 | import asyncio |
| 34 | import sys |
| 35 | import time |
| 36 | |
| 37 | import sqlalchemy as sa |
| 38 | from sqlalchemy.ext.asyncio import AsyncSession, create_async_engine |
| 39 | from sqlalchemy.orm import sessionmaker |
| 40 | |
| 41 | sys.path.insert(0, "/app") |
| 42 | sys.path.insert(0, "/tmp/devpkgs") |
| 43 | |
| 44 | from musehub.db.database import get_database_url |
| 45 | from musehub.services.musehub_symbol_indexer import backfill_content_ids_from_snapshots |
| 46 | |
| 47 | |
| 48 | async def run(dry_run: bool, quiet: bool, repo_id: str | None) -> int: |
| 49 | engine = create_async_engine(get_database_url(), echo=False) |
| 50 | Session = sessionmaker(engine, class_=AsyncSession, expire_on_commit=False) |
| 51 | |
| 52 | async with Session() as session: |
| 53 | if not quiet: |
| 54 | scope = f"repo {repo_id}" if repo_id else "all repos" |
| 55 | mode = "[DRY RUN] " if dry_run else "" |
| 56 | print(f"{mode}Scanning {scope} for file-level entries with missing content_id …") |
| 57 | |
| 58 | t0 = time.monotonic() |
| 59 | count = await backfill_content_ids_from_snapshots(session, repo_id=repo_id, dry_run=dry_run) |
| 60 | |
| 61 | if not dry_run: |
| 62 | await session.commit() |
| 63 | |
| 64 | elapsed = time.monotonic() - t0 |
| 65 | |
| 66 | if dry_run: |
| 67 | print(f"Would update {count} entr{'y' if count == 1 else 'ies'} ({elapsed:.1f}s)") |
| 68 | else: |
| 69 | print(f"Updated {count} entr{'y' if count == 1 else 'ies'} ({elapsed:.1f}s)") |
| 70 | |
| 71 | await engine.dispose() |
| 72 | return count |
| 73 | |
| 74 | |
| 75 | def main() -> None: |
| 76 | p = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter) |
| 77 | p.add_argument("--dry-run", action="store_true", help="Count rows without writing") |
| 78 | p.add_argument("--repo-id", metavar="REPO_ID", help="Limit to a single repo") |
| 79 | p.add_argument("-q", "--quiet", action="store_true", help="Suppress progress output") |
| 80 | args = p.parse_args() |
| 81 | |
| 82 | count = asyncio.run(run(dry_run=args.dry_run, quiet=args.quiet, repo_id=args.repo_id)) |
| 83 | sys.exit(0 if count >= 0 else 1) |
| 84 | |
| 85 | |
| 86 | if __name__ == "__main__": |
| 87 | main() |
File History
13 commits
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