musehub_wire_fetch.py
python
sha256:f99af7b1a7f36c4d537d1c630d4b71fc39222b1255f82e930929e2fc89015e11
fix: relax browse_repo perf budget to 500ms — 200ms was too…
Sonnet 4.6
99 days ago
| 1 | """Fetch path — wire_fetch_presign, wire_fetch_mpack, wire_fetch, process_mpack_gc_job.""" |
| 2 | |
| 3 | import asyncio |
| 4 | import hashlib |
| 5 | import logging |
| 6 | import msgpack as _msgpack |
| 7 | import time as _time_module |
| 8 | from datetime import datetime, timezone |
| 9 | from typing import TypedDict |
| 10 | |
| 11 | from sqlalchemy import func, select, text as _sa_text |
| 12 | from sqlalchemy.dialects.postgresql import insert as _pg_insert |
| 13 | from sqlalchemy.ext.asyncio import AsyncSession |
| 14 | |
| 15 | from musehub.db.musehub_repo_models import ( |
| 16 | MusehubBranch, |
| 17 | MusehubCommit, |
| 18 | MusehubCommitGraph, |
| 19 | MusehubMPackIndex, |
| 20 | MusehubObject, |
| 21 | MusehubObjectRef, |
| 22 | MusehubRepo, |
| 23 | MusehubSnapshot, |
| 24 | ) |
| 25 | from musehub.models.wire import WireFetchRequest |
| 26 | from muse.core.types import blob_id |
| 27 | from musehub.storage import get_backend |
| 28 | |
| 29 | from musehub.services.musehub_wire_shared import ( |
| 30 | FetchMPackResult, |
| 31 | FetchNotIndexedError, |
| 32 | FetchPresignResult, |
| 33 | MPackValidationError, |
| 34 | _reconstruct_manifest, |
| 35 | _snap_row_to_wire, |
| 36 | _to_wire_commit, |
| 37 | _utc_now, |
| 38 | logger, |
| 39 | ) |
| 40 | |
| 41 | |
| 42 | type _CommitDeltaMap = dict[str, MusehubCommitGraph | MusehubCommit] |
| 43 | |
| 44 | |
| 45 | async def _walk_commit_delta( |
| 46 | session: AsyncSession, |
| 47 | want: list[str] | set[str], |
| 48 | have: list[str] | set[str], |
| 49 | ) -> _CommitDeltaMap: |
| 50 | _wcd_t0 = _time_module.perf_counter() |
| 51 | _want_list = list(want) |
| 52 | _have_list = list(have) |
| 53 | logger.info("[_walk_commit_delta] START want=%d have=%d want_ids=%s", |
| 54 | len(_want_list), len(_have_list), |
| 55 | [cid[:16] for cid in _want_list[:5]]) |
| 56 | |
| 57 | have_set: frozenset[str] = frozenset(_have_list) |
| 58 | starts = [cid for cid in _want_list if cid not in have_set] |
| 59 | if not starts: |
| 60 | logger.info("[_walk_commit_delta] SKIP — all want in have, 0ms") |
| 61 | return {} |
| 62 | |
| 63 | if True: # fast path always active — commit graph is global (no repo_id) |
| 64 | from sqlalchemy import func as _func |
| 65 | |
| 66 | want_gen_q = await session.execute( |
| 67 | select(_func.max(MusehubCommitGraph.generation)) |
| 68 | .where(MusehubCommitGraph.commit_id.in_(starts)) |
| 69 | ) |
| 70 | _want_gen_raw = want_gen_q.scalar() |
| 71 | max_want_gen: int = _want_gen_raw or 0 |
| 72 | |
| 73 | _missing_from_graph: list[str] = [] |
| 74 | _found_in_graph: list[tuple[str, int]] = [] |
| 75 | for _scid in starts[:10]: |
| 76 | _sg = await session.execute( |
| 77 | select(MusehubCommitGraph.generation) |
| 78 | .where(MusehubCommitGraph.commit_id == _scid) |
| 79 | ) |
| 80 | _sg_val = _sg.scalar_one_or_none() |
| 81 | if _sg_val is None: |
| 82 | _missing_from_graph.append(_scid[:16]) |
| 83 | else: |
| 84 | _found_in_graph.append((_scid[:16], _sg_val)) |
| 85 | logger.info( |
| 86 | "[_walk_commit_delta] want_gen_raw=%s max_want_gen=%d " |
| 87 | "starts_in_graph=%s starts_missing_from_graph=%s", |
| 88 | _want_gen_raw, max_want_gen, _found_in_graph, _missing_from_graph, |
| 89 | ) |
| 90 | |
| 91 | min_have_gen: int = -1 |
| 92 | if have_set: |
| 93 | have_gen_q = await session.execute( |
| 94 | select(_func.max(MusehubCommitGraph.generation)) |
| 95 | .where(MusehubCommitGraph.commit_id.in_(list(have_set))) |
| 96 | ) |
| 97 | min_have_gen = have_gen_q.scalar() or -1 |
| 98 | |
| 99 | range_q = await session.execute( |
| 100 | select( |
| 101 | MusehubCommitGraph.commit_id, |
| 102 | MusehubCommitGraph.parent_ids, |
| 103 | MusehubCommitGraph.snapshot_id, |
| 104 | ) |
| 105 | .where(MusehubCommitGraph.generation > min_have_gen) |
| 106 | .where(MusehubCommitGraph.generation <= max_want_gen) |
| 107 | ) |
| 108 | graph_map: dict[str, tuple[list[str], str | None]] = { |
| 109 | cid: (pids or [], sid) for cid, pids, sid in range_q |
| 110 | } |
| 111 | logger.info( |
| 112 | "[_walk_commit_delta] range_scan gen=(%d,%d] returned %d rows " |
| 113 | "starts_in_map=%s", |
| 114 | min_have_gen, max_want_gen, len(graph_map), |
| 115 | [cid[:16] for cid in starts if cid in graph_map], |
| 116 | ) |
| 117 | |
| 118 | visited_mem: set[str] = set(have_set) |
| 119 | frontier_mem = [cid for cid in starts if cid not in visited_mem] |
| 120 | reachable_cids: set[str] = set() |
| 121 | while frontier_mem: |
| 122 | next_mem: list[str] = [] |
| 123 | for cid in frontier_mem: |
| 124 | if cid in visited_mem: |
| 125 | continue |
| 126 | visited_mem.add(cid) |
| 127 | reachable_cids.add(cid) |
| 128 | pids_for_cid, _ = graph_map.get(cid, ([], None)) |
| 129 | for p in pids_for_cid: |
| 130 | if p not in visited_mem and p not in have_set: |
| 131 | next_mem.append(p) |
| 132 | frontier_mem = next_mem |
| 133 | |
| 134 | from types import SimpleNamespace as _SN |
| 135 | needed_graph: dict[str, _SN] = {} |
| 136 | for cid in reachable_cids: |
| 137 | pids_ns, sid_ns = graph_map.get(cid, ([], None)) |
| 138 | needed_graph[cid] = _SN(commit_id=cid, snapshot_id=sid_ns, parent_ids=pids_ns) |
| 139 | |
| 140 | _wcd_elapsed = (_time_module.perf_counter() - _wcd_t0) * 1000 |
| 141 | logger.info( |
| 142 | "[_walk_commit_delta] DONE (graph) commits=%d elapsed=%.1fms (%.3fms/commit) " |
| 143 | "gen_range=(%d,%d] graph_rows=%d reachable=%d", |
| 144 | len(needed_graph), _wcd_elapsed, _wcd_elapsed / max(len(needed_graph), 1), |
| 145 | min_have_gen, max_want_gen, len(graph_map), len(reachable_cids), |
| 146 | ) |
| 147 | return needed_graph # type: ignore[return-value] |
| 148 | |
| 149 | # Legacy fallback |
| 150 | from musehub.graph.walk import walk_dag_async |
| 151 | |
| 152 | _row_cache: dict[str, MusehubCommit] = {} |
| 153 | _db_calls = 0 |
| 154 | |
| 155 | async def _adj(cid: str) -> list[str]: |
| 156 | nonlocal _db_calls |
| 157 | _db_calls += 1 |
| 158 | row = await session.get(MusehubCommit, cid) |
| 159 | if row is not None: |
| 160 | _row_cache[cid] = row |
| 161 | return row.parent_ids or [] if row else [] |
| 162 | |
| 163 | needed_legacy: dict[str, MusehubCommit] = {} |
| 164 | async for cid in walk_dag_async(starts, _adj, exclude=have_set): |
| 165 | if cid in _row_cache: |
| 166 | needed_legacy[cid] = _row_cache[cid] |
| 167 | |
| 168 | _wcd_elapsed = (_time_module.perf_counter() - _wcd_t0) * 1000 |
| 169 | logger.info( |
| 170 | "[_walk_commit_delta] DONE (legacy) commits=%d db_calls=%d elapsed=%.1fms (%.2fms/commit)", |
| 171 | len(needed_legacy), _db_calls, _wcd_elapsed, |
| 172 | _wcd_elapsed / max(len(needed_legacy), 1), |
| 173 | ) |
| 174 | return needed_legacy |
| 175 | |
| 176 | |
| 177 | async def wire_fetch_presign( |
| 178 | session: AsyncSession, |
| 179 | repo_id: str, |
| 180 | req: WireFetchRequest, |
| 181 | ttl_seconds: int = 3600, |
| 182 | ) -> FetchPresignResult: |
| 183 | import asyncio |
| 184 | from datetime import timedelta |
| 185 | |
| 186 | _empty: FetchPresignResult = { |
| 187 | "presign": False, |
| 188 | "blob_urls": {}, |
| 189 | "commits": [], |
| 190 | "snapshots": [], |
| 191 | "branch_heads": {}, |
| 192 | "repo_id": repo_id, |
| 193 | "domain": "", |
| 194 | "default_branch": "main", |
| 195 | "expires_at": None, |
| 196 | "commit_count": 0, |
| 197 | "blob_count": 0, |
| 198 | } |
| 199 | |
| 200 | if not req.want: |
| 201 | return _empty |
| 202 | |
| 203 | repo_row = await session.get(MusehubRepo, repo_id) |
| 204 | if repo_row is None: |
| 205 | return _empty |
| 206 | _domain: str = repo_row.domain_id or "" |
| 207 | _default_branch: str = repo_row.default_branch if repo_row.default_branch else "main" |
| 208 | _empty["domain"] = _domain |
| 209 | _empty["default_branch"] = _default_branch |
| 210 | _empty["repo_id"] = repo_id |
| 211 | |
| 212 | have_set = set(req.have) |
| 213 | needed_rows = await _walk_commit_delta(session, req.want, have_set) |
| 214 | |
| 215 | if not needed_rows: |
| 216 | return {**_empty, "domain": _domain, "default_branch": _default_branch} |
| 217 | |
| 218 | _presign_commit_rows: dict[str, MusehubCommit] = {} |
| 219 | _presign_any = next(iter(needed_rows.values())) |
| 220 | if not isinstance(_presign_any, MusehubCommit): |
| 221 | _PRESIGN_WIRE_BATCH = 2000 |
| 222 | _presign_cids = list(needed_rows.keys()) |
| 223 | for _pi in range(0, len(_presign_cids), _PRESIGN_WIRE_BATCH): |
| 224 | _pq = await session.execute( |
| 225 | select(MusehubCommit).where(MusehubCommit.commit_id.in_(_presign_cids[_pi : _pi + _PRESIGN_WIRE_BATCH])) |
| 226 | ) |
| 227 | for _pr in _pq.scalars(): |
| 228 | _presign_commit_rows[_pr.commit_id] = _pr |
| 229 | else: |
| 230 | _presign_commit_rows = needed_rows # type: ignore[assignment] |
| 231 | |
| 232 | snap_ids = [r.snapshot_id for r in needed_rows.values() if r.snapshot_id] |
| 233 | all_oids: set[str] = set() |
| 234 | if snap_ids: |
| 235 | snaps_q = await session.execute( |
| 236 | select(MusehubSnapshot).where(MusehubSnapshot.snapshot_id.in_(snap_ids)) |
| 237 | ) |
| 238 | for snap in snaps_q.scalars().all(): |
| 239 | manifest = ( |
| 240 | _msgpack.unpackb(snap.manifest_blob, raw=False) |
| 241 | if snap.manifest_blob |
| 242 | else await _reconstruct_manifest(session, snap.snapshot_id) |
| 243 | ) |
| 244 | all_oids.update(v for v in manifest.values() if v) |
| 245 | |
| 246 | have_snap_ids: list[str] = [] |
| 247 | if have_set: |
| 248 | have_commits_q = await session.execute( |
| 249 | select(MusehubCommit).where(MusehubCommit.commit_id.in_(have_set)) |
| 250 | ) |
| 251 | have_snap_ids = [r.snapshot_id for r in have_commits_q.scalars().all() if r.snapshot_id] |
| 252 | have_oids: set[str] = set() |
| 253 | if have_snap_ids: |
| 254 | have_snaps_q = await session.execute( |
| 255 | select(MusehubSnapshot).where(MusehubSnapshot.snapshot_id.in_(have_snap_ids)) |
| 256 | ) |
| 257 | for snap in have_snaps_q.scalars().all(): |
| 258 | manifest = ( |
| 259 | _msgpack.unpackb(snap.manifest_blob, raw=False) |
| 260 | if snap.manifest_blob |
| 261 | else await _reconstruct_manifest(session, snap.snapshot_id) |
| 262 | ) |
| 263 | have_oids.update(v for v in manifest.values() if v) |
| 264 | |
| 265 | new_oids = all_oids - have_oids |
| 266 | n_objects = len(new_oids) |
| 267 | n_commits = len(needed_rows) |
| 268 | |
| 269 | total_size = 0 |
| 270 | if new_oids: |
| 271 | size_q = await session.execute( |
| 272 | select(func.coalesce(func.sum(MusehubObject.size_bytes), 0)).where( |
| 273 | MusehubObject.object_id.in_(list(new_oids)) |
| 274 | ) |
| 275 | ) |
| 276 | total_size = int(size_q.scalar() or 0) |
| 277 | |
| 278 | wire_commits = [_to_wire_commit(row).model_dump() for row in _presign_commit_rows.values()] |
| 279 | |
| 280 | snap_rows_q = await session.execute( |
| 281 | select(MusehubSnapshot).where(MusehubSnapshot.snapshot_id.in_(snap_ids)) |
| 282 | ) |
| 283 | wire_snaps = [_snap_row_to_wire(snap) for snap in snap_rows_q.scalars().all()] |
| 284 | |
| 285 | branch_rows_q = await session.execute( |
| 286 | select(MusehubBranch).where(MusehubBranch.repo_id == repo_id) |
| 287 | ) |
| 288 | branch_heads = { |
| 289 | b.name: b.head_commit_id |
| 290 | for b in branch_rows_q.scalars().all() |
| 291 | if b.head_commit_id |
| 292 | } |
| 293 | |
| 294 | backend = get_backend() |
| 295 | |
| 296 | sem = asyncio.Semaphore(50) |
| 297 | |
| 298 | logger.info( |
| 299 | "fetch/presign: generating %d presigned GET URLs repo=%s/%s", |
| 300 | len(new_oids), repo_row.owner, repo_row.slug, |
| 301 | ) |
| 302 | |
| 303 | async def _presign_one(oid: str) -> tuple[str, str]: |
| 304 | async with sem: |
| 305 | url = await backend.presign_get(oid, ttl_seconds) |
| 306 | logger.debug("fetch/presign: presigned oid=%s", oid) |
| 307 | return oid, url |
| 308 | |
| 309 | pairs = await asyncio.gather(*(_presign_one(oid) for oid in new_oids)) |
| 310 | blob_urls = {oid: url for oid, url in pairs} |
| 311 | expires_at = (_utc_now() + timedelta(seconds=ttl_seconds)).isoformat() |
| 312 | |
| 313 | return { |
| 314 | "presign": True, |
| 315 | "blob_urls": blob_urls, |
| 316 | "commits": wire_commits, |
| 317 | "snapshots": wire_snaps, |
| 318 | "branch_heads": branch_heads, |
| 319 | "repo_id": repo_id, |
| 320 | "domain": _domain, |
| 321 | "default_branch": _default_branch, |
| 322 | "expires_at": expires_at, |
| 323 | "commit_count": n_commits, |
| 324 | "blob_count": n_objects, |
| 325 | } |
| 326 | |
| 327 | async def wire_fetch_mpack( |
| 328 | session: AsyncSession, |
| 329 | repo_id: str, |
| 330 | want: list[str], |
| 331 | have: list[str], |
| 332 | ttl_seconds: int = 3600, |
| 333 | ) -> FetchMPackResult: |
| 334 | import msgpack as _msgpack_local |
| 335 | |
| 336 | _t0 = _time_module.perf_counter() |
| 337 | def _ms() -> float: |
| 338 | return (_time_module.perf_counter() - _t0) * 1000 |
| 339 | |
| 340 | logger.info("[wire_fetch_mpack] START repo_id=%s want=%d have=%d want_ids=%s", |
| 341 | repo_id, len(want), len(have), [cid[:16] for cid in want[:5]]) |
| 342 | |
| 343 | _up_to_date: FetchMPackResult = { |
| 344 | "mpack_url": None, |
| 345 | "mpack_id": None, |
| 346 | "commit_count": 0, |
| 347 | "blob_count": 0, |
| 348 | } |
| 349 | |
| 350 | if not want: |
| 351 | logger.info("[wire_fetch_mpack] SKIP — want is empty") |
| 352 | return _up_to_date |
| 353 | |
| 354 | have_set = set(have) |
| 355 | logger.info("[wire_fetch_mpack] step=1 DAG walk starting t=%.1fms", _ms()) |
| 356 | needed_rows = await _walk_commit_delta(session, want, have_set) |
| 357 | logger.info("[wire_fetch_mpack] step=1 DAG walk done commits=%d t=%.1fms", len(needed_rows), _ms()) |
| 358 | |
| 359 | if not needed_rows: |
| 360 | logger.info("[wire_fetch_mpack] SKIP — client already up-to-date (needed_rows empty)") |
| 361 | return _up_to_date |
| 362 | |
| 363 | commit_rows: dict[str, MusehubCommit] = {} |
| 364 | _any = next(iter(needed_rows.values())) |
| 365 | _is_proxy = not isinstance(_any, MusehubCommit) |
| 366 | logger.info("[wire_fetch_mpack] step=1b needed_rows=%d is_proxy=%s t=%.1fms", |
| 367 | len(needed_rows), _is_proxy, _ms()) |
| 368 | if _is_proxy: |
| 369 | _cids = list(needed_rows.keys()) |
| 370 | _q = await session.execute( |
| 371 | select(MusehubCommit).where( |
| 372 | _sa_text("commit_id = ANY(:ids)").bindparams(ids=_cids) |
| 373 | ) |
| 374 | ) |
| 375 | for _row in _q.scalars(): |
| 376 | commit_rows[_row.commit_id] = _row |
| 377 | _missing_from_db = set(_cids) - set(commit_rows.keys()) |
| 378 | logger.info( |
| 379 | "[wire_fetch_mpack] step=1b bulk fetch done commits_in_db=%d missing_from_db=%d " |
| 380 | "missing_ids=%s t=%.1fms", |
| 381 | len(commit_rows), len(_missing_from_db), |
| 382 | [cid[:16] for cid in list(_missing_from_db)[:5]], _ms(), |
| 383 | ) |
| 384 | else: |
| 385 | commit_rows = needed_rows # type: ignore[assignment] |
| 386 | logger.info("[wire_fetch_mpack] step=1b using MusehubCommit rows directly commits=%d t=%.1fms", |
| 387 | len(commit_rows), _ms()) |
| 388 | |
| 389 | _proxy_snap_ids_raw = [r.snapshot_id for r in needed_rows.values()] |
| 390 | snap_ids = [sid for sid in _proxy_snap_ids_raw if sid] |
| 391 | _proxy_snap_none_count = sum(1 for s in _proxy_snap_ids_raw if not s) |
| 392 | _commit_row_snap_ids = [r.snapshot_id for r in commit_rows.values() if r.snapshot_id] |
| 393 | logger.info( |
| 394 | "[wire_fetch_mpack] step=2 snap_ids_from_graph=%d snap_ids_none_in_graph=%d " |
| 395 | "snap_ids_from_commit_rows=%d t=%.1fms", |
| 396 | len(snap_ids), _proxy_snap_none_count, len(_commit_row_snap_ids), _ms(), |
| 397 | ) |
| 398 | |
| 399 | snap_map: dict[str, dict] = {} |
| 400 | if snap_ids: |
| 401 | snaps_q = await session.execute( |
| 402 | select(MusehubSnapshot).where( |
| 403 | _sa_text("snapshot_id = ANY(:ids)").bindparams(ids=snap_ids) |
| 404 | ) |
| 405 | ) |
| 406 | for snap in snaps_q.scalars().all(): |
| 407 | snap_map[snap.snapshot_id] = _snap_row_to_wire(snap) |
| 408 | logger.info("[wire_fetch_mpack] step=2 snap_map loaded=%d t=%.1fms", len(snap_map), _ms()) |
| 409 | |
| 410 | all_oids: set[str] = set() |
| 411 | _needed_cids = list(needed_rows.keys()) |
| 412 | logger.warning("[GRAPH-DEBUG] wire_fetch_mpack: needed_rows=%d needed_cids_sample=%s", |
| 413 | len(_needed_cids), [c[:16] for c in _needed_cids[:3]]) |
| 414 | _debug_graph_q = await session.execute( |
| 415 | select(MusehubCommitGraph.commit_id, MusehubCommitGraph.generation, MusehubCommitGraph.snapshot_id) |
| 416 | .where(MusehubCommitGraph.commit_id.in_(_needed_cids)) |
| 417 | .order_by(MusehubCommitGraph.generation.desc()) |
| 418 | .limit(5) |
| 419 | ) |
| 420 | _debug_graph_rows = _debug_graph_q.all() |
| 421 | logger.warning("[GRAPH-DEBUG] wire_fetch_mpack: CommitGraph has %d rows for needed_cids (top 5 by gen): %s", |
| 422 | len(_debug_graph_rows), |
| 423 | [(r[1], r[0][:16]) for r in _debug_graph_rows]) |
| 424 | want_tip_snap_q = await session.execute( |
| 425 | select(MusehubCommitGraph.snapshot_id) |
| 426 | .where(MusehubCommitGraph.commit_id.in_(_needed_cids)) |
| 427 | .order_by(MusehubCommitGraph.generation.desc()) |
| 428 | .limit(1) |
| 429 | ) |
| 430 | want_tip_snap_id = want_tip_snap_q.scalar_one_or_none() |
| 431 | logger.warning("[GRAPH-DEBUG] wire_fetch_mpack: want_tip_snap_id=%s", |
| 432 | want_tip_snap_id[:20] if want_tip_snap_id else "NONE") |
| 433 | logger.info("[wire_fetch_mpack] step=2 want_tip_snap_id=%s (from CommitGraph) t=%.1fms", |
| 434 | want_tip_snap_id[:16] if want_tip_snap_id else None, _ms()) |
| 435 | if want_tip_snap_id: |
| 436 | wt_blob_q = await session.execute( |
| 437 | select(MusehubSnapshot.manifest_blob) |
| 438 | .where(MusehubSnapshot.snapshot_id == want_tip_snap_id) |
| 439 | ) |
| 440 | wt_blob = wt_blob_q.scalar_one_or_none() |
| 441 | if wt_blob: |
| 442 | all_oids.update(v for v in _msgpack_local.unpackb(wt_blob, raw=False).values() if v) |
| 443 | logger.warning("[GRAPH-DEBUG] wire_fetch_mpack: want_tip manifest all_oids=%d wt_blob_present=%s", |
| 444 | len(all_oids), wt_blob is not None) |
| 445 | logger.info("[wire_fetch_mpack] step=2 want_tip manifest all_oids=%d wt_blob_present=%s t=%.1fms", |
| 446 | len(all_oids), wt_blob is not None, _ms()) |
| 447 | else: |
| 448 | logger.warning( |
| 449 | "[wire_fetch_mpack] step=2 WARN want_tip_snap_id=None — CommitGraph missing tip " |
| 450 | "needed_cids=%s commit_rows_snap_ids=%s", |
| 451 | [cid[:16] for cid in list(needed_rows.keys())[:5]], |
| 452 | [sid[:16] for sid in _commit_row_snap_ids[:5]], |
| 453 | ) |
| 454 | |
| 455 | have_oids: set[str] = set() |
| 456 | if have_set: |
| 457 | ht_snap_q = await session.execute( |
| 458 | select(MusehubCommitGraph.snapshot_id) |
| 459 | .where(MusehubCommitGraph.commit_id.in_(list(have_set))) |
| 460 | .order_by(MusehubCommitGraph.generation.desc()) |
| 461 | .limit(1) |
| 462 | ) |
| 463 | have_tip_snap_id = ht_snap_q.scalar_one_or_none() |
| 464 | if have_tip_snap_id: |
| 465 | ht_blob_q = await session.execute( |
| 466 | select(MusehubSnapshot.manifest_blob) |
| 467 | .where(MusehubSnapshot.snapshot_id == have_tip_snap_id) |
| 468 | ) |
| 469 | ht_blob = ht_blob_q.scalar_one_or_none() |
| 470 | if ht_blob: |
| 471 | have_oids.update(v for v in _msgpack_local.unpackb(ht_blob, raw=False).values() if v) |
| 472 | |
| 473 | new_oids = all_oids - have_oids |
| 474 | logger.info( |
| 475 | "[wire_fetch_mpack] step=2 done snap_map=%d all_oids=%d have_oids=%d new_oids=%d t=%.1fms", |
| 476 | len(snap_map), len(all_oids), len(have_oids), len(new_oids), _ms(), |
| 477 | ) |
| 478 | |
| 479 | if new_oids: |
| 480 | indexed_q = await session.execute( |
| 481 | select(MusehubMPackIndex.entity_id) |
| 482 | .where(MusehubMPackIndex.entity_id.in_(list(new_oids))) |
| 483 | .where(MusehubMPackIndex.entity_type == "object") |
| 484 | ) |
| 485 | indexed_oids = {row[0] for row in indexed_q} |
| 486 | missing = new_oids - indexed_oids |
| 487 | if missing: |
| 488 | logger.warning( |
| 489 | "[wire_fetch_mpack] step=3 NOT INDEXED %d/%d objects — raising FetchNotIndexedError t=%.1fms", |
| 490 | len(missing), len(new_oids), _ms(), |
| 491 | ) |
| 492 | raise FetchNotIndexedError(len(missing)) |
| 493 | logger.info("[wire_fetch_mpack] step=3 index coverage OK oids=%d t=%.1fms", len(new_oids), _ms()) |
| 494 | |
| 495 | backend = get_backend() |
| 496 | |
| 497 | cache_hits: dict[str, bytes] = {} |
| 498 | if new_oids: |
| 499 | _CACHE_CHUNK = 10000 |
| 500 | _new_oid_list = list(new_oids) |
| 501 | for _ci in range(0, len(_new_oid_list), _CACHE_CHUNK): |
| 502 | _chunk = _new_oid_list[_ci : _ci + _CACHE_CHUNK] |
| 503 | _cache_q = await session.execute( |
| 504 | select(MusehubObject.object_id, MusehubObject.content_cache) |
| 505 | .where(MusehubObject.object_id.in_(_chunk)) |
| 506 | .where(MusehubObject.content_cache.isnot(None)) |
| 507 | ) |
| 508 | for _oid, _cached in _cache_q: |
| 509 | if _cached: |
| 510 | cache_hits[_oid] = bytes(_cached) |
| 511 | |
| 512 | cache_miss_oids = [oid for oid in new_oids if oid not in cache_hits] |
| 513 | |
| 514 | oid_to_mpack: dict[str, str] = {} |
| 515 | if cache_miss_oids: |
| 516 | _MIDX_CHUNK = 10000 |
| 517 | for _ci in range(0, len(cache_miss_oids), _MIDX_CHUNK): |
| 518 | _chunk = cache_miss_oids[_ci : _ci + _MIDX_CHUNK] |
| 519 | _midx_q = await session.execute( |
| 520 | select(MusehubMPackIndex.entity_id, MusehubMPackIndex.mpack_id) |
| 521 | .where(MusehubMPackIndex.entity_id.in_(_chunk)) |
| 522 | .where(MusehubMPackIndex.entity_type == "object") |
| 523 | ) |
| 524 | for _oid, _mid in _midx_q: |
| 525 | oid_to_mpack[_oid] = _mid |
| 526 | |
| 527 | mpack_to_oids: dict[str, list[str]] = {} |
| 528 | no_mpack_oids: list[str] = [] |
| 529 | for oid in cache_miss_oids: |
| 530 | mid = oid_to_mpack.get(oid) |
| 531 | if mid: |
| 532 | mpack_to_oids.setdefault(mid, []).append(oid) |
| 533 | else: |
| 534 | no_mpack_oids.append(oid) |
| 535 | |
| 536 | mpack_hits: dict[str, bytes] = {} |
| 537 | mpack_miss_oids: list[str] = [] |
| 538 | _sem_mpack = asyncio.Semaphore(8) |
| 539 | |
| 540 | async def _extract_from_mpack(mpack_id: str, oids: list[str]) -> None: |
| 541 | async with _sem_mpack: |
| 542 | raw = await backend.get_mpack(mpack_id) |
| 543 | if raw is None: |
| 544 | mpack_miss_oids.extend(oids) |
| 545 | return |
| 546 | import zstandard as _zstd_phase1 |
| 547 | _dctx_phase1 = _zstd_phase1.ZstdDecompressor() |
| 548 | try: |
| 549 | if raw[:4] == b"MUSE": |
| 550 | from muse.core.mpack import parse_wire_mpack as _parse_wire_fetch |
| 551 | payload = _parse_wire_fetch(raw) |
| 552 | else: |
| 553 | payload = _msgpack_local.unpackb(raw, raw=False) |
| 554 | except Exception as _parse_err: |
| 555 | logger.warning( |
| 556 | "[_extract_from_mpack] failed to parse mpack=%s: %s", |
| 557 | mpack_id[:20], _parse_err, |
| 558 | ) |
| 559 | mpack_miss_oids.extend(oids) |
| 560 | return |
| 561 | obj_index: dict[str, bytes] = {} |
| 562 | for o in payload.get("blobs", []): |
| 563 | oid_entry = o.get("object_id", "") |
| 564 | content = o.get("content") or b"" |
| 565 | if not isinstance(content, bytes): |
| 566 | content = bytes(content) |
| 567 | _ZSTD_MAGIC = b"\x28\xb5\x2f\xfd" |
| 568 | if (o.get("encoding") == "zstd" or content[:4] == _ZSTD_MAGIC) and content: |
| 569 | try: |
| 570 | content = _dctx_phase1.decompress(content) |
| 571 | except Exception as _decomp_err: |
| 572 | logger.warning( |
| 573 | "[_extract_from_mpack] zstd decompress failed oid=%s: %s", |
| 574 | oid_entry[:20], _decomp_err, |
| 575 | ) |
| 576 | continue |
| 577 | obj_index[oid_entry] = content |
| 578 | for oid in oids: |
| 579 | content = obj_index.get(oid) |
| 580 | if content is not None: |
| 581 | mpack_hits[oid] = content |
| 582 | else: |
| 583 | mpack_miss_oids.append(oid) |
| 584 | |
| 585 | if mpack_to_oids: |
| 586 | await asyncio.gather( |
| 587 | *(_extract_from_mpack(mid, oids) for mid, oids in mpack_to_oids.items()) |
| 588 | ) |
| 589 | |
| 590 | legacy_hits: dict[str, bytes] = {} |
| 591 | _fallback_oids = no_mpack_oids + mpack_miss_oids |
| 592 | if _fallback_oids: |
| 593 | _sem_legacy = asyncio.Semaphore(50) |
| 594 | |
| 595 | async def _get_legacy(oid: str) -> None: |
| 596 | async with _sem_legacy: |
| 597 | data = await backend.get(oid) |
| 598 | if data: |
| 599 | legacy_hits[oid] = data |
| 600 | |
| 601 | await asyncio.gather(*(_get_legacy(oid) for oid in _fallback_oids)) |
| 602 | |
| 603 | _all_blob_bytes: dict[str, bytes] = {**legacy_hits, **mpack_hits, **cache_hits} |
| 604 | blob_pairs = [(oid, _all_blob_bytes[oid]) for oid in new_oids if oid in _all_blob_bytes] |
| 605 | logger.info( |
| 606 | "[wire_fetch_mpack] step=4 fetched %d blobs (cache=%d mpack=%d legacy=%d) t=%.1fms", |
| 607 | len(blob_pairs), len(cache_hits), len(mpack_hits), len(legacy_hits), _ms(), |
| 608 | ) |
| 609 | |
| 610 | wire_commits = [_to_wire_commit(row).model_dump() for row in commit_rows.values()] |
| 611 | wire_snaps = [snap_map[sid] for sid in snap_ids if sid in snap_map] |
| 612 | wire_blobs = [ |
| 613 | {"object_id": oid, "content": data} |
| 614 | for oid, data in blob_pairs |
| 615 | if data |
| 616 | ] |
| 617 | logger.info( |
| 618 | "[wire_fetch_mpack] step=5 assembly: wire_commits=%d wire_snaps=%d wire_blobs=%d " |
| 619 | "snap_ids_total=%d snap_ids_in_map=%d commit_rows=%d t=%.1fms", |
| 620 | len(wire_commits), len(wire_snaps), len(wire_blobs), |
| 621 | len(snap_ids), sum(1 for sid in snap_ids if sid in snap_map), |
| 622 | len(commit_rows), _ms(), |
| 623 | ) |
| 624 | |
| 625 | from muse.core.mpack import build_wire_mpack as _build_wire_mpack |
| 626 | _head_commit_id = want[0] if want else "" |
| 627 | mpack_bytes = _build_wire_mpack( |
| 628 | { |
| 629 | "commits": wire_commits, |
| 630 | "snapshots": wire_snaps, |
| 631 | "blobs": wire_blobs, |
| 632 | "tags": [], |
| 633 | }, |
| 634 | meta={"repo_id": repo_id, "head_commit_id": _head_commit_id}, |
| 635 | ) |
| 636 | mpack_id = blob_id(mpack_bytes) |
| 637 | |
| 638 | n_commits = len(wire_commits) |
| 639 | n_blobs = len(wire_blobs) |
| 640 | logger.info( |
| 641 | "[wire_fetch_mpack] step=5 assembled commits=%d snapshots=%d blobs=%d bytes=%d t=%.1fms", |
| 642 | n_commits, len(wire_snaps), n_blobs, len(mpack_bytes), _ms(), |
| 643 | ) |
| 644 | |
| 645 | await backend.put_mpack(mpack_id, mpack_bytes) |
| 646 | mpack_url = await backend.presign_mpack_get(mpack_id, ttl_seconds) |
| 647 | logger.info( |
| 648 | "[wire_fetch_mpack] step=6 mpack_id=%s mpack_url=%s t=%.1fms", |
| 649 | mpack_id[:20], mpack_url[:80] if mpack_url else None, _ms(), |
| 650 | ) |
| 651 | logger.info("[wire_fetch_mpack] RETURN commits=%d blobs=%d TOTAL=%.1fms", n_commits, n_blobs, _ms()) |
| 652 | |
| 653 | async def _cleanup() -> None: |
| 654 | await asyncio.sleep(ttl_seconds) |
| 655 | try: |
| 656 | await backend.delete(mpack_id) |
| 657 | except Exception: |
| 658 | pass |
| 659 | |
| 660 | asyncio.ensure_future(_cleanup()) |
| 661 | |
| 662 | return { |
| 663 | "mpack_url": mpack_url, |
| 664 | "mpack_id": mpack_id, |
| 665 | "commit_count": n_commits, |
| 666 | "blob_count": n_blobs, |
| 667 | } |
| 668 | |
| 669 | async def _check_missing_objects( |
| 670 | session: AsyncSession, |
| 671 | needs_check: set[str], |
| 672 | ) -> set[str]: |
| 673 | if not needs_check: |
| 674 | return set() |
| 675 | from musehub.db.musehub_repo_models import MusehubObject |
| 676 | registered: set[str] = set( |
| 677 | (await session.execute( |
| 678 | select(MusehubObject.object_id).where( |
| 679 | MusehubObject.object_id.in_(list(needs_check)), |
| 680 | MusehubObject.deleted_at.is_(None), |
| 681 | ) |
| 682 | )).scalars().all() |
| 683 | ) |
| 684 | return needs_check - registered |
| 685 | |
| 686 | |
| 687 | class MPackGCResult(TypedDict): |
| 688 | skipped: bool |
| 689 | packs_before: int |
| 690 | packs_after: int |
| 691 | consolidated_key: str |
| 692 | |
| 693 | |
| 694 | async def process_mpack_gc_job(session: AsyncSession, repo_id: str) -> MPackGCResult: |
| 695 | import msgpack as _mp |
| 696 | |
| 697 | _skipped: MPackGCResult = { |
| 698 | "skipped": True, |
| 699 | "packs_before": 0, |
| 700 | "packs_after": 0, |
| 701 | "consolidated_key": "", |
| 702 | } |
| 703 | |
| 704 | repo_oids_q = await session.execute( |
| 705 | select(MusehubObjectRef.object_id) |
| 706 | .where(MusehubObjectRef.repo_id == repo_id) |
| 707 | ) |
| 708 | repo_oid_set = {row[0] for row in repo_oids_q} |
| 709 | mpack_q = await session.execute( |
| 710 | select(MusehubMPackIndex.mpack_id) |
| 711 | .where(MusehubMPackIndex.entity_id.in_(list(repo_oid_set))) |
| 712 | .where(MusehubMPackIndex.entity_type == "object") |
| 713 | .distinct() |
| 714 | ) |
| 715 | mpack_ids = [row[0] for row in mpack_q] |
| 716 | packs_before = len(mpack_ids) |
| 717 | |
| 718 | if packs_before <= 1: |
| 719 | _skipped["packs_before"] = packs_before |
| 720 | if mpack_ids: |
| 721 | _skipped["consolidated_key"] = mpack_ids[0] |
| 722 | return _skipped |
| 723 | |
| 724 | import musehub.storage.backends as _backends_mod |
| 725 | backend = _backends_mod.get_backend() |
| 726 | |
| 727 | merged_objects: dict[str, bytes] = {} |
| 728 | |
| 729 | async def _download(pid: str) -> None: |
| 730 | raw = await backend.get_mpack(pid) |
| 731 | if not raw: |
| 732 | logger.warning("[mpack_gc] mpack not found in storage: %s", pid) |
| 733 | return |
| 734 | if raw[:4] == b"MUSE": |
| 735 | from muse.core.mpack import parse_wire_mpack as _parse_gc |
| 736 | _parsed = _parse_gc(raw) |
| 737 | else: |
| 738 | _parsed = _mp.unpackb(raw, raw=False) |
| 739 | for obj in _parsed.get("blobs", []): |
| 740 | oid = obj.get("object_id", "") |
| 741 | content = obj.get("content", b"") |
| 742 | if oid and oid not in merged_objects: |
| 743 | merged_objects[oid] = content |
| 744 | |
| 745 | await asyncio.gather(*(_download(pid) for pid in mpack_ids)) |
| 746 | |
| 747 | from muse.core.mpack import build_wire_mpack as _build_gc_mpack |
| 748 | consolidated_bytes = _build_gc_mpack({ |
| 749 | "commits": [], |
| 750 | "snapshots": [], |
| 751 | "blobs": [ |
| 752 | {"object_id": oid, "content": merged_objects[oid]} |
| 753 | for oid in sorted(merged_objects) |
| 754 | ], |
| 755 | "tags": [], |
| 756 | }) |
| 757 | consolidated_key = "sha256:" + hashlib.sha256(consolidated_bytes).hexdigest() |
| 758 | |
| 759 | await backend.put_mpack(consolidated_key, consolidated_bytes) |
| 760 | |
| 761 | old_mpack_ids = [p for p in mpack_ids if p != consolidated_key] |
| 762 | if old_mpack_ids: |
| 763 | from sqlalchemy import delete as sa_delete |
| 764 | await session.execute( |
| 765 | sa_delete(MusehubMPackIndex) |
| 766 | .where(MusehubMPackIndex.mpack_id.in_(old_mpack_ids)) |
| 767 | .where(MusehubMPackIndex.entity_type == "object") |
| 768 | ) |
| 769 | _gc_now = datetime.now(timezone.utc) |
| 770 | new_rows = [ |
| 771 | { |
| 772 | "entity_id": oid, |
| 773 | "mpack_id": consolidated_key, |
| 774 | "entity_type": "object", |
| 775 | "created_at": _gc_now, |
| 776 | } |
| 777 | for oid in merged_objects |
| 778 | ] |
| 779 | if new_rows: |
| 780 | _GC_MIDX_CHUNK = 5000 |
| 781 | for _gmi in range(0, len(new_rows), _GC_MIDX_CHUNK): |
| 782 | await session.execute( |
| 783 | _pg_insert(MusehubMPackIndex) |
| 784 | .values(new_rows[_gmi : _gmi + _GC_MIDX_CHUNK]) |
| 785 | .on_conflict_do_nothing(index_elements=["entity_id", "mpack_id"]) |
| 786 | ) |
| 787 | |
| 788 | logger.info( |
| 789 | "[mpack_gc] repo=%s consolidated %d mpacks → 1 (objects=%d key=%s)", |
| 790 | repo_id, packs_before, len(merged_objects), consolidated_key, |
| 791 | ) |
| 792 | |
| 793 | return { |
| 794 | "skipped": False, |
| 795 | "packs_before": packs_before, |
| 796 | "packs_after": 1, |
| 797 | "consolidated_key": consolidated_key, |
| 798 | } |
| 799 | |
| 800 | |
| 801 | class FetchResult(TypedDict): |
| 802 | mpack_id: str |
| 803 | mpack_url: str | None |
| 804 | commit_count: int |
| 805 | blob_count: int |
| 806 | |
| 807 | |
| 808 | class FetchCommitNotFound(Exception): |
| 809 | """A want commit_id does not exist in musehub_commits.""" |
| 810 | |
| 811 | |
| 812 | class FetchNotReady(Exception): |
| 813 | """Needed objects are absent from musehub_mpack_index — client must retry.""" |
| 814 | |
| 815 | |
| 816 | async def wire_fetch( |
| 817 | session: AsyncSession, |
| 818 | repo_id: str, |
| 819 | want: list[str], |
| 820 | have: list[str], |
| 821 | ttl_seconds: int = 3600, |
| 822 | ) -> FetchResult: |
| 823 | import msgpack as _mp |
| 824 | |
| 825 | _empty: FetchResult = {"mpack_id": "", "mpack_url": None, "commit_count": 0, "blob_count": 0} |
| 826 | |
| 827 | for entry in want: |
| 828 | if not (isinstance(entry, str) and entry.startswith("sha256:")): |
| 829 | raise MPackValidationError(f"want entry is not a sha256: id: {entry!r}") |
| 830 | |
| 831 | for entry in have: |
| 832 | if not (isinstance(entry, str) and entry.startswith("sha256:")): |
| 833 | raise MPackValidationError(f"have entry is not a sha256: id: {entry!r}") |
| 834 | |
| 835 | if want: |
| 836 | existing_q = await session.execute( |
| 837 | select(MusehubCommit.commit_id).where(MusehubCommit.commit_id.in_(want)) |
| 838 | ) |
| 839 | found = {row[0] for row in existing_q} |
| 840 | missing_want = [cid for cid in want if cid not in found] |
| 841 | if missing_want: |
| 842 | raise FetchCommitNotFound(missing_want[0]) |
| 843 | |
| 844 | have_set = set(have) |
| 845 | needed = await _walk_commit_delta(session, want, have_set) |
| 846 | if not needed: |
| 847 | return _empty |
| 848 | |
| 849 | cids = list(needed.keys()) |
| 850 | commit_rows: dict[str, MusehubCommit] = {} |
| 851 | for i in range(0, len(cids), 2000): |
| 852 | q = await session.execute( |
| 853 | select(MusehubCommit).where(MusehubCommit.commit_id.in_(cids[i:i + 2000])) |
| 854 | ) |
| 855 | for row in q.scalars(): |
| 856 | commit_rows[row.commit_id] = row |
| 857 | |
| 858 | want_snap_ids = {r.snapshot_id for r in needed.values() if r.snapshot_id} |
| 859 | have_snap_ids: set[str] = set() |
| 860 | if have_set: |
| 861 | have_commits_q = await session.execute( |
| 862 | select(MusehubCommit.snapshot_id).where(MusehubCommit.commit_id.in_(list(have_set))) |
| 863 | ) |
| 864 | have_snap_ids = {row[0] for row in have_commits_q if row[0]} |
| 865 | |
| 866 | new_snap_ids = want_snap_ids - have_snap_ids |
| 867 | snap_map: dict[str, dict] = {} |
| 868 | new_oids: set[str] = set() |
| 869 | if new_snap_ids: |
| 870 | snaps_q = await session.execute( |
| 871 | select(MusehubSnapshot).where(MusehubSnapshot.snapshot_id.in_(list(new_snap_ids))) |
| 872 | ) |
| 873 | for snap in snaps_q.scalars(): |
| 874 | manifest = ( |
| 875 | _mp.unpackb(snap.manifest_blob, raw=False) |
| 876 | if snap.manifest_blob |
| 877 | else await _reconstruct_manifest(session, snap.snapshot_id) |
| 878 | ) |
| 879 | new_oids.update(v for v in manifest.values() if v) |
| 880 | snap_map[snap.snapshot_id] = _snap_row_to_wire(snap) |
| 881 | |
| 882 | if have_snap_ids: |
| 883 | have_snaps_q = await session.execute( |
| 884 | select(MusehubSnapshot).where(MusehubSnapshot.snapshot_id.in_(list(have_snap_ids))) |
| 885 | ) |
| 886 | for snap in have_snaps_q.scalars(): |
| 887 | m = ( |
| 888 | _mp.unpackb(snap.manifest_blob, raw=False) |
| 889 | if snap.manifest_blob |
| 890 | else await _reconstruct_manifest(session, snap.snapshot_id) |
| 891 | ) |
| 892 | new_oids -= {v for v in m.values() if v} |
| 893 | |
| 894 | if new_oids: |
| 895 | idx_q = await session.execute( |
| 896 | select(MusehubMPackIndex.entity_id).where( |
| 897 | MusehubMPackIndex.entity_id.in_(list(new_oids)), |
| 898 | MusehubMPackIndex.entity_type == "object", |
| 899 | ) |
| 900 | ) |
| 901 | indexed = {row[0] for row in idx_q} |
| 902 | unindexed = new_oids - indexed |
| 903 | if unindexed: |
| 904 | raise FetchNotReady(f"{len(unindexed)} object(s) not yet in mpack_index") |
| 905 | |
| 906 | backend = get_backend() |
| 907 | objects: list[dict] = [] |
| 908 | if new_oids: |
| 909 | obj_q = await session.execute( |
| 910 | select(MusehubObject).where(MusehubObject.object_id.in_(list(new_oids))) |
| 911 | ) |
| 912 | for obj_row in obj_q.scalars(): |
| 913 | if obj_row.content_cache is not None: |
| 914 | content = obj_row.content_cache |
| 915 | else: |
| 916 | content = await backend.get(obj_row.object_id) or b"" |
| 917 | objects.append({"object_id": obj_row.object_id, "content": content}) |
| 918 | |
| 919 | wire_commits = [_to_wire_commit(r).model_dump() for r in commit_rows.values()] |
| 920 | from muse.core.mpack import build_wire_mpack as _build_fetch_mpack |
| 921 | wire_bytes = _build_fetch_mpack({ |
| 922 | "commits": wire_commits, |
| 923 | "snapshots": list(snap_map.values()), |
| 924 | "blobs": objects, |
| 925 | "tags": [], |
| 926 | }) |
| 927 | mpack_id = blob_id(wire_bytes) |
| 928 | |
| 929 | await backend.put_mpack(mpack_id, wire_bytes) |
| 930 | mpack_url = await backend.presign_mpack_get(mpack_id, ttl_seconds) |
| 931 | |
| 932 | return { |
| 933 | "mpack_id": mpack_id, |
| 934 | "mpack_url": mpack_url, |
| 935 | "commit_count": len(commit_rows), |
| 936 | "blob_count": len(objects), |
| 937 | } |
File History
1 commit
sha256:f99af7b1a7f36c4d537d1c630d4b71fc39222b1255f82e930929e2fc89015e11
fix: relax browse_repo perf budget to 500ms — 200ms was too…
Sonnet 4.6
99 days ago