bundle.py
python
sha256:b570925a487b44187c0441977d252906b6bf2719253bc53f96412d520cd24d9e
feat: muse#53 item 4 — remote symlog sharing via push/pull/bundle
Sonnet 5
minor
⚠ breaking
4 hours ago
| 1 | """``muse bundle`` — pack and unpack commits for single-file transport. |
| 2 | |
| 3 | A bundle is a self-contained msgpack binary file carrying commits, snapshots, |
| 4 | and objects. It wraps ``muse pack-objects`` / ``unpack-objects`` with friendlier |
| 5 | names and the key value-add of auto-updating local branch refs after ``unbundle``. |
| 6 | |
| 7 | Use bundles to transfer a repository slice between machines without a network |
| 8 | connection — copy the file over SSH, USB, or email. |
| 9 | |
| 10 | Bundle format: binary msgpack encoding of the ``MPack`` TypedDict. |
| 11 | Objects are stored as raw bytes (no base64 overhead). |
| 12 | |
| 13 | Subcommands:: |
| 14 | |
| 15 | muse bundle create <file> [<ref>...] [--have <id>...] [--json] |
| 16 | muse bundle unbundle <file> [--no-update-refs] [--json] |
| 17 | muse bundle verify <file> [-q] [--json] |
| 18 | muse bundle list-heads <file> [--json] |
| 19 | |
| 20 | Security model:: |
| 21 | |
| 22 | Symlinks inside ``.muse/refs/heads/`` are silently skipped during branch |
| 23 | enumeration. A crafted symlink could otherwise expose external path content. |
| 24 | |
| 25 | Ref files are read at most 65 bytes; oversized files are treated as invalid |
| 26 | refs rather than read entirely into memory. |
| 27 | |
| 28 | Bundle file parsing is limited by ``MAX_PACK_MSGPACK_BYTES``; both the file |
| 29 | size and the msgpack payload are checked before any deserialization occurs. |
| 30 | |
| 31 | JSON output schemas:: |
| 32 | |
| 33 | bundle create --json: |
| 34 | {"file": str, "commits": int, "blobs": int, |
| 35 | "size_bytes": int, "branches": [str, ...]} |
| 36 | |
| 37 | bundle unbundle --json: |
| 38 | {"commits_written": int, "snapshots_written": int, |
| 39 | "blobs_written": int, "blobs_skipped": int, |
| 40 | "refs_updated": [str, ...], "verified": bool} |
| 41 | |
| 42 | bundle verify --json: |
| 43 | {"blobs_checked": int, "snapshots_checked": int, |
| 44 | "all_ok": bool, "failures": [str, ...]} |
| 45 | |
| 46 | bundle list-heads --json: |
| 47 | {"<branch>": "<commit_id>", ...} |
| 48 | |
| 49 | Exit codes:: |
| 50 | |
| 51 | 0 — success |
| 52 | 1 — bundle not found, integrity failure, bad arguments |
| 53 | 3 — I/O error |
| 54 | """ |
| 55 | |
| 56 | import argparse |
| 57 | import json |
| 58 | import logging |
| 59 | import os |
| 60 | import pathlib |
| 61 | import sys |
| 62 | import tempfile |
| 63 | from typing import TypedDict, TypeGuard |
| 64 | |
| 65 | import msgpack |
| 66 | |
| 67 | from muse.core.types import blob_id, short_id |
| 68 | from muse.core.paths import heads_dir as _heads_dir |
| 69 | from muse.core.envelope import EnvelopeJson, make_envelope |
| 70 | from muse.core.timing import start_timer |
| 71 | from muse.core.errors import ExitCode |
| 72 | from muse.core.object_store import has_object, write_object |
| 73 | from muse.core.mpack import MPack, apply_mpack, build_mpack |
| 74 | from muse.core.symlog import list_symlog_symbols, merge_remote_symlog_entries, read_symlog, symlog_entries_to_wire |
| 75 | from muse.core.repo import require_repo |
| 76 | from muse.core.io import ( |
| 77 | MAX_PACK_MSGPACK_BYTES, |
| 78 | safe_unpackb, |
| 79 | ) |
| 80 | from muse.core.types import ( |
| 81 | BranchHeads, |
| 82 | MsgpackValue, |
| 83 | ) |
| 84 | from muse.core.refs import ( |
| 85 | get_head_commit_id, |
| 86 | read_current_branch, |
| 87 | write_branch_ref, |
| 88 | ) |
| 89 | from muse.core.commits import ( |
| 90 | CommitDict, |
| 91 | CommitRecord, |
| 92 | read_commit, |
| 93 | resolve_commit_ref, |
| 94 | write_commit, |
| 95 | ) |
| 96 | from muse.core.snapshots import ( |
| 97 | SnapshotDict, |
| 98 | SnapshotRecord, |
| 99 | write_snapshot, |
| 100 | ) |
| 101 | from muse.core.validation import sanitize_display, validate_branch_name |
| 102 | |
| 103 | logger = logging.getLogger(__name__) |
| 104 | |
| 105 | # Maximum bytes read from a branch ref file. A valid canonical commit ID is |
| 106 | # "sha256:" (7 chars) + 64 hex chars + optional newline = 72 bytes. Anything |
| 107 | # larger is treated as corrupt. |
| 108 | _MAX_REF_BYTES = 72 |
| 109 | |
| 110 | # Maximum number of distinct failures collected during ``muse bundle verify``. |
| 111 | # Caps memory and output size when a crafted bundle has many corrupted objects. |
| 112 | _MAX_VERIFY_FAILURES = 100 |
| 113 | |
| 114 | # --------------------------------------------------------------------------- |
| 115 | # JSON wire-format TypedDicts |
| 116 | # --------------------------------------------------------------------------- |
| 117 | |
| 118 | class _BundleCreateJson(EnvelopeJson): |
| 119 | """JSON output for ``muse bundle create --json``.""" |
| 120 | |
| 121 | file: str |
| 122 | commits: int |
| 123 | objects: int |
| 124 | size_bytes: int |
| 125 | branches: list[str] |
| 126 | |
| 127 | class _BundleUnbundleJson(EnvelopeJson): |
| 128 | """JSON output for ``muse bundle unbundle --json``.""" |
| 129 | |
| 130 | commits_written: int |
| 131 | snapshots_written: int |
| 132 | blobs_written: int |
| 133 | blobs_skipped: int |
| 134 | refs_updated: list[str] |
| 135 | verified: bool |
| 136 | #: Number of new symbol-journal entries merged (muse#53 item 4). |
| 137 | symlog_entries_merged: int |
| 138 | |
| 139 | class _BundleVerifyJson(EnvelopeJson): |
| 140 | """JSON output for ``muse bundle verify --json``.""" |
| 141 | |
| 142 | blobs_checked: int |
| 143 | snapshots_checked: int |
| 144 | all_ok: bool |
| 145 | failures: list[str] |
| 146 | |
| 147 | class _BundleInspectCommit(TypedDict): |
| 148 | """Per-commit entry in ``muse bundle inspect --json`` output.""" |
| 149 | |
| 150 | commit_id: str |
| 151 | message: str |
| 152 | committed_at: str |
| 153 | agent_id: str |
| 154 | branches: list[str] |
| 155 | |
| 156 | class _BundleInspectJson(EnvelopeJson): |
| 157 | """JSON output for ``muse bundle inspect --json``.""" |
| 158 | |
| 159 | total_commits: int |
| 160 | total_objects: int |
| 161 | branches: dict[str, str] # branch_name → commit_id (sha256:<64hex>) |
| 162 | commits: list[_BundleInspectCommit] |
| 163 | |
| 164 | class _BundleDiffCommit(TypedDict): |
| 165 | """Per-commit entry in ``muse bundle diff --json`` output.""" |
| 166 | |
| 167 | commit_id: str |
| 168 | message: str |
| 169 | committed_at: str |
| 170 | |
| 171 | class _BundleDiffJson(EnvelopeJson): |
| 172 | """JSON output for ``muse bundle diff --json``.""" |
| 173 | |
| 174 | new_commits: int |
| 175 | known_commits: int |
| 176 | refs_to_advance: list[str] # branch names that would move forward |
| 177 | commits: list[_BundleDiffCommit] # new commits not yet in local repo |
| 178 | |
| 179 | class _BundleListHeadsJson(EnvelopeJson): |
| 180 | """JSON output for ``muse bundle list-heads --json``.""" |
| 181 | |
| 182 | heads: dict[str, str] # branch_name → commit_id |
| 183 | |
| 184 | # --------------------------------------------------------------------------- |
| 185 | # TypeGuards for wire-boundary narrowing |
| 186 | # --------------------------------------------------------------------------- |
| 187 | |
| 188 | def _is_commit_dict(v: MsgpackValue) -> TypeGuard[CommitDict]: |
| 189 | """TypeGuard for narrowing MsgpackValue → CommitDict at the wire boundary.""" |
| 190 | return isinstance(v, dict) |
| 191 | |
| 192 | def _is_snapshot_dict(v: MsgpackValue) -> TypeGuard[SnapshotDict]: |
| 193 | """TypeGuard for narrowing MsgpackValue → SnapshotDict at the wire boundary.""" |
| 194 | return isinstance(v, dict) |
| 195 | |
| 196 | # --------------------------------------------------------------------------- |
| 197 | # Internal helpers |
| 198 | # --------------------------------------------------------------------------- |
| 199 | |
| 200 | def _resolve_refs( |
| 201 | root: pathlib.Path, |
| 202 | branch: str, |
| 203 | refs: list[str], |
| 204 | ) -> list[str]: |
| 205 | """Resolve a list of ref strings to commit IDs. Expands ``HEAD``.""" |
| 206 | ids: list[str] = [] |
| 207 | for ref in refs: |
| 208 | if ref.upper() == "HEAD": |
| 209 | cid = get_head_commit_id(root, branch) |
| 210 | if cid: |
| 211 | ids.append(cid) |
| 212 | else: |
| 213 | rec = resolve_commit_ref(root, branch, ref) |
| 214 | if rec: |
| 215 | ids.append(rec.commit_id) |
| 216 | else: |
| 217 | print(f"❌ Ref '{sanitize_display(ref)}' not found.", file=sys.stderr) |
| 218 | raise SystemExit(ExitCode.USER_ERROR) |
| 219 | return ids |
| 220 | |
| 221 | def _load_bundle(file_path: pathlib.Path) -> MPack: |
| 222 | """Read a msgpack bundle file and return a :class:`~muse.core.mpack.MPack`. |
| 223 | |
| 224 | Security: |
| 225 | The file size is checked before reading. The raw bytes are then |
| 226 | passed through :func:`~muse.core.store.safe_unpackb` which enforces |
| 227 | an upper bound on the deserialized payload as well. This double-check |
| 228 | prevents OOM from crafted size-bomb bundles. |
| 229 | |
| 230 | Exceptions are narrowed to :class:`(OSError, ValueError, |
| 231 | msgpack.UnpackException)` — the precise types raised by the I/O and |
| 232 | parse path — so genuine programming errors are not silently swallowed. |
| 233 | """ |
| 234 | try: |
| 235 | size = file_path.stat().st_size |
| 236 | if size > MAX_PACK_MSGPACK_BYTES: |
| 237 | raise OSError( |
| 238 | f"Bundle file is {size:,} bytes — exceeds the " |
| 239 | f"{MAX_PACK_MSGPACK_BYTES // (1024 * 1024)} MiB safety cap." |
| 240 | ) |
| 241 | raw_bytes = file_path.read_bytes() |
| 242 | parsed = safe_unpackb( |
| 243 | raw_bytes, |
| 244 | context="bundle file", |
| 245 | max_bytes=MAX_PACK_MSGPACK_BYTES, |
| 246 | allow_binary=True, |
| 247 | ) |
| 248 | except FileNotFoundError: |
| 249 | print( |
| 250 | f"❌ Bundle file not found: {sanitize_display(str(file_path))}", |
| 251 | file=sys.stderr, |
| 252 | ) |
| 253 | raise SystemExit(ExitCode.USER_ERROR) |
| 254 | except (OSError, ValueError, msgpack.UnpackException) as exc: |
| 255 | print(f"❌ Bundle is not valid msgpack: {sanitize_display(str(exc))}", file=sys.stderr) |
| 256 | raise SystemExit(ExitCode.USER_ERROR) |
| 257 | |
| 258 | if not isinstance(parsed, dict): |
| 259 | print("❌ Bundle has unexpected structure.", file=sys.stderr) |
| 260 | raise SystemExit(ExitCode.USER_ERROR) |
| 261 | |
| 262 | from muse.core.mpack import BlobPayload |
| 263 | |
| 264 | bundle: MPack = {} |
| 265 | raw_commits = parsed.get("commits") |
| 266 | if isinstance(raw_commits, list): |
| 267 | bundle["commits"] = [r for r in raw_commits if _is_commit_dict(r)] |
| 268 | raw_snapshots = parsed.get("snapshots") |
| 269 | if isinstance(raw_snapshots, list): |
| 270 | bundle["snapshots"] = [r for r in raw_snapshots if _is_snapshot_dict(r)] |
| 271 | if "blobs" in parsed and isinstance(parsed["blobs"], list): |
| 272 | blobs: list[BlobPayload] = [] |
| 273 | for item in parsed["blobs"]: |
| 274 | if isinstance(item, dict): |
| 275 | oid = item.get("object_id") |
| 276 | content = item.get("content") |
| 277 | if isinstance(oid, str) and isinstance(content, (bytes, bytearray)): |
| 278 | blobs.append(BlobPayload(object_id=oid, content=bytes(content))) |
| 279 | bundle["blobs"] = blobs |
| 280 | if "branch_heads" in parsed and isinstance(parsed["branch_heads"], dict): |
| 281 | bundle["branch_heads"] = { |
| 282 | k: v |
| 283 | for k, v in parsed["branch_heads"].items() |
| 284 | if isinstance(k, str) and isinstance(v, str) |
| 285 | } |
| 286 | if "symlogs" in parsed and isinstance(parsed["symlogs"], dict): |
| 287 | symlogs: dict[str, list[dict]] = {} |
| 288 | for addr, entries in parsed["symlogs"].items(): |
| 289 | if not isinstance(addr, str) or not isinstance(entries, list): |
| 290 | continue |
| 291 | symlogs[addr] = [e for e in entries if isinstance(e, dict)] |
| 292 | bundle["symlogs"] = symlogs |
| 293 | return bundle |
| 294 | |
| 295 | def _iter_branches(root: pathlib.Path) -> list[tuple[str, str]]: |
| 296 | """Return ``[(branch_name, commit_id)]`` for all branch ref files. |
| 297 | |
| 298 | Security: |
| 299 | Symlinks inside ``.muse/refs/heads/`` are silently skipped. |
| 300 | Ref files are capped at ``_MAX_REF_BYTES`` (65 bytes) before |
| 301 | decoding — oversized content is decoded but will fail hex |
| 302 | validation downstream. |
| 303 | """ |
| 304 | heads_dir = _heads_dir(root) |
| 305 | if not heads_dir.exists(): |
| 306 | return [] |
| 307 | result: list[tuple[str, str]] = [] |
| 308 | for ref_file in sorted(heads_dir.rglob("*")): |
| 309 | if ref_file.is_symlink(): |
| 310 | logger.warning("⚠️ bundle: skipping symlink ref: %s", ref_file) |
| 311 | continue |
| 312 | if not ref_file.is_file(): |
| 313 | continue |
| 314 | raw_bytes = ref_file.read_bytes()[:_MAX_REF_BYTES] |
| 315 | cid = raw_bytes.decode("utf-8", errors="replace").strip() |
| 316 | if cid: |
| 317 | branch_name = str(ref_file.relative_to(heads_dir).as_posix()) |
| 318 | result.append((branch_name, cid)) |
| 319 | return result |
| 320 | |
| 321 | def _reachable_from(root: pathlib.Path, tip_ids: list[str]) -> set[str]: |
| 322 | """Return all commit IDs reachable from *tip_ids*.""" |
| 323 | from muse.core.graph import ancestor_ids |
| 324 | return ancestor_ids(root, tip_ids) |
| 325 | |
| 326 | # --------------------------------------------------------------------------- |
| 327 | # Registration |
| 328 | # --------------------------------------------------------------------------- |
| 329 | |
| 330 | def register(subparsers: "argparse._SubParsersAction[argparse.ArgumentParser]") -> None: |
| 331 | """Register the bundle subcommand.""" |
| 332 | parser = subparsers.add_parser( |
| 333 | "bundle", |
| 334 | help="Pack and unpack commits into a single portable bundle file.", |
| 335 | description=__doc__, |
| 336 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 337 | ) |
| 338 | subs = parser.add_subparsers(dest="subcommand", metavar="SUBCOMMAND") |
| 339 | subs.required = True |
| 340 | |
| 341 | # create |
| 342 | create_p = subs.add_parser( |
| 343 | "create", |
| 344 | help="Create a bundle file containing commits reachable from <refs>.", |
| 345 | description=( |
| 346 | "Pack commits, snapshots, and objects reachable from <refs> into a\n" |
| 347 | "single self-contained msgpack binary file. Transfer it over SSH,\n" |
| 348 | "USB, or email and apply it with ``muse bundle unbundle``.\n\n" |
| 349 | "Use --have to prune commits the receiver already has, producing a\n" |
| 350 | "smaller incremental bundle.\n\n" |
| 351 | "Agent quickstart\n" |
| 352 | "----------------\n" |
| 353 | " muse bundle create repo.bundle --json\n" |
| 354 | " muse bundle create out.bundle feat/audio --json\n" |
| 355 | " muse bundle create out.bundle HEAD --have <old-sha> --json\n\n" |
| 356 | "JSON output schema\n" |
| 357 | "------------------\n" |
| 358 | ' {"file": "<path>", "commits": <int>, "blobs": <int>,\n' |
| 359 | ' "size_bytes": <int>, "branches": ["<branch>", ...]}\n\n' |
| 360 | "Exit codes\n" |
| 361 | "----------\n" |
| 362 | " 0 — bundle created successfully\n" |
| 363 | " 1 — unknown ref, no commits to bundle, or bad --have ref\n" |
| 364 | " 2 — not inside a Muse repository\n" |
| 365 | ), |
| 366 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 367 | ) |
| 368 | create_p.add_argument("file", help="Output bundle file path.") |
| 369 | create_p.add_argument( |
| 370 | "refs", nargs="*", default=None, |
| 371 | help="Refs to include (default: HEAD).", |
| 372 | ) |
| 373 | create_p.add_argument( |
| 374 | "--have", "-H", nargs="*", default=None, dest="have", |
| 375 | help="Commits the receiver already has (exclude from bundle).", |
| 376 | ) |
| 377 | create_p.add_argument( |
| 378 | "--json", "-j", action="store_true", dest="json_out", |
| 379 | help="Emit a machine-readable JSON summary.", |
| 380 | ) |
| 381 | create_p.set_defaults(func=run_create) |
| 382 | |
| 383 | # list-heads |
| 384 | list_heads_p = subs.add_parser( |
| 385 | "list-heads", |
| 386 | help="List the branch heads recorded in a bundle file.", |
| 387 | description=( |
| 388 | "Print the branch → commit_id map embedded in a bundle file.\n" |
| 389 | "Does NOT require a Muse repository — runs against any bundle\n" |
| 390 | "file in isolation.\n\n" |
| 391 | "Agent quickstart\n" |
| 392 | "----------------\n" |
| 393 | " muse bundle list-heads repo.bundle --json\n" |
| 394 | " muse bundle list-heads repo.bundle --json | jq 'keys'\n\n" |
| 395 | "JSON output schema\n" |
| 396 | "------------------\n" |
| 397 | ' {"<branch>": "sha256:<64hex>", ...}\n\n' |
| 398 | "Exit codes\n" |
| 399 | "----------\n" |
| 400 | " 0 — always (empty object when no heads are recorded)\n" |
| 401 | " 1 — bundle file not found or not valid msgpack\n" |
| 402 | ), |
| 403 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 404 | ) |
| 405 | list_heads_p.add_argument("file", help="Bundle file to inspect.") |
| 406 | list_heads_p.add_argument( |
| 407 | "--json", "-j", action="store_true", dest="json_out", |
| 408 | help="Emit a machine-readable JSON map of branch → commit_id.", |
| 409 | ) |
| 410 | list_heads_p.set_defaults(func=run_list_heads) |
| 411 | |
| 412 | # unbundle |
| 413 | unbundle_p = subs.add_parser( |
| 414 | "unbundle", |
| 415 | help="Apply a bundle to the local store and optionally advance branch refs.", |
| 416 | description=( |
| 417 | "Write all commits, snapshots, and objects from a bundle file into\n" |
| 418 | "the local repository, then advance local branch refs to match the\n" |
| 419 | "bundle's recorded heads. All writes are idempotent — objects\n" |
| 420 | "already present are counted as 'skipped', not errors.\n\n" |
| 421 | "Use --no-update-refs to import objects without moving any branches\n" |
| 422 | "(useful for inspection or partial imports).\n\n" |
| 423 | "Agent quickstart\n" |
| 424 | "----------------\n" |
| 425 | " muse bundle unbundle repo.bundle --json\n" |
| 426 | " muse bundle unbundle repo.bundle --no-update-refs --json\n\n" |
| 427 | "JSON output schema\n" |
| 428 | "------------------\n" |
| 429 | ' {"commits_written": <int>, "snapshots_written": <int>,\n' |
| 430 | ' "blobs_written": <int>, "blobs_skipped": <int>,\n' |
| 431 | ' "refs_updated": ["<branch>", ...]}\n\n' |
| 432 | "Exit codes\n" |
| 433 | "----------\n" |
| 434 | " 0 — bundle applied (even if all objects were already present)\n" |
| 435 | " 1 — bundle file not found or not valid msgpack\n" |
| 436 | " 2 — not inside a Muse repository\n" |
| 437 | ), |
| 438 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 439 | ) |
| 440 | unbundle_p.add_argument("file", help="Bundle file to apply.") |
| 441 | unbundle_p.add_argument( |
| 442 | "--no-update-refs", action="store_false", dest="update_refs", |
| 443 | help="Do not update local branch refs from the bundle's branch_heads.", |
| 444 | ) |
| 445 | unbundle_p.add_argument( |
| 446 | "--verify", action="store_true", dest="verify_first", |
| 447 | help=( |
| 448 | "Verify bundle integrity before applying. Exits 1 without writing " |
| 449 | "anything if the bundle is corrupt." |
| 450 | ), |
| 451 | ) |
| 452 | unbundle_p.add_argument( |
| 453 | "--json", "-j", action="store_true", dest="json_out", |
| 454 | help="Emit a machine-readable JSON summary.", |
| 455 | ) |
| 456 | unbundle_p.set_defaults(func=run_unbundle, update_refs=True, verify_first=False) |
| 457 | |
| 458 | # inspect |
| 459 | inspect_p = subs.add_parser( |
| 460 | "inspect", |
| 461 | help="Show the commit log and branch state recorded in a bundle file.", |
| 462 | description=( |
| 463 | "Read and display commits, branch heads, and agent provenance from\n" |
| 464 | "a bundle file without applying it. No Muse repository required —\n" |
| 465 | "safe to run before deciding whether to unbundle.\n\n" |
| 466 | "Agent quickstart\n" |
| 467 | "----------------\n" |
| 468 | " muse bundle inspect repo.bundle --json\n" |
| 469 | " # → total_commits, branches map, per-commit message + agent_id\n\n" |
| 470 | "Typical agent workflow\n" |
| 471 | "----------------------\n" |
| 472 | " muse bundle inspect bundle.muse --json # what's in it?\n" |
| 473 | " muse bundle verify bundle.muse -q # is it clean?\n" |
| 474 | " muse bundle diff bundle.muse --json # what's new?\n" |
| 475 | " muse bundle unbundle bundle.muse --verify # apply safely\n\n" |
| 476 | "JSON output schema\n" |
| 477 | "------------------\n" |
| 478 | ' {"total_commits": int, "total_objects": int,\n' |
| 479 | ' "branches": {"<name>": "sha256:<64hex>"},\n' |
| 480 | ' "commits": [{"commit_id": str, "message": str,\n' |
| 481 | ' "committed_at": str, "agent_id": str,\n' |
| 482 | ' "branches": [str]}, ...]}\n\n' |
| 483 | "Exit codes\n" |
| 484 | "----------\n" |
| 485 | " 0 — bundle read successfully (even if empty)\n" |
| 486 | " 1 — bundle file not found or not valid msgpack\n" |
| 487 | ), |
| 488 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 489 | ) |
| 490 | inspect_p.add_argument("file", help="Bundle file to inspect.") |
| 491 | inspect_p.add_argument( |
| 492 | "--json", "-j", action="store_true", dest="json_out", |
| 493 | help="Emit a machine-readable JSON summary.", |
| 494 | ) |
| 495 | inspect_p.set_defaults(func=run_inspect) |
| 496 | |
| 497 | # diff |
| 498 | diff_p = subs.add_parser( |
| 499 | "diff", |
| 500 | help="Show which commits in a bundle are not yet in the local repository.", |
| 501 | description=( |
| 502 | "Compare a bundle's commits against the local repository and report\n" |
| 503 | "which ones are new (not yet present). Agents use this to answer\n" |
| 504 | "\"what would this bundle add?\" before deciding to apply it.\n\n" |
| 505 | "Requires a Muse repository (unlike inspect, verify, list-heads).\n\n" |
| 506 | "Agent quickstart\n" |
| 507 | "----------------\n" |
| 508 | " muse bundle diff repo.bundle --json\n" |
| 509 | " # → new_commits, known_commits, refs_to_advance, commit list\n\n" |
| 510 | "JSON output schema\n" |
| 511 | "------------------\n" |
| 512 | ' {"new_commits": int, "known_commits": int,\n' |
| 513 | ' "refs_to_advance": ["<branch>", ...],\n' |
| 514 | ' "commits": [{"commit_id": str, "message": str,\n' |
| 515 | ' "committed_at": str}, ...]}\n\n' |
| 516 | "Exit codes\n" |
| 517 | "----------\n" |
| 518 | " 0 — diff computed (even if new_commits == 0)\n" |
| 519 | " 1 — bundle file not found or not valid msgpack\n" |
| 520 | " 2 — not inside a Muse repository\n" |
| 521 | ), |
| 522 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 523 | ) |
| 524 | diff_p.add_argument("file", help="Bundle file to diff against the local repo.") |
| 525 | diff_p.add_argument( |
| 526 | "--json", "-j", action="store_true", dest="json_out", |
| 527 | help="Emit a machine-readable JSON report.", |
| 528 | ) |
| 529 | diff_p.set_defaults(func=run_diff) |
| 530 | |
| 531 | # verify |
| 532 | verify_p = subs.add_parser( |
| 533 | "verify", |
| 534 | help="Verify the integrity of a bundle file.", |
| 535 | description=( |
| 536 | "Check every object's SHA-256 against its declared object_id\n" |
| 537 | "(detects corruption), and confirm every snapshot's objects are\n" |
| 538 | "present in the bundle (detects truncation). Does NOT require a\n" |
| 539 | "Muse repository — runs against any bundle file in isolation.\n\n" |
| 540 | "Use --quiet for scripting: no output, just the exit code.\n\n" |
| 541 | "Agent quickstart\n" |
| 542 | "----------------\n" |
| 543 | " muse bundle verify repo.bundle --json\n" |
| 544 | " muse bundle verify repo.bundle -q && muse bundle unbundle repo.bundle\n\n" |
| 545 | "JSON output schema\n" |
| 546 | "------------------\n" |
| 547 | ' {"blobs_checked": <int>, "snapshots_checked": <int>,\n' |
| 548 | ' "all_ok": true|false, "failures": ["<description>", ...]}\n\n' |
| 549 | "Exit codes\n" |
| 550 | "----------\n" |
| 551 | " 0 — bundle is clean\n" |
| 552 | " 1 — integrity failures, or bundle file not found / not valid\n" |
| 553 | ), |
| 554 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 555 | ) |
| 556 | verify_p.add_argument("file", help="Bundle file to verify.") |
| 557 | verify_p.add_argument( |
| 558 | "--quiet", "-q", action="store_true", dest="quiet", |
| 559 | help="No output — exit 0 if clean, 1 on failure.", |
| 560 | ) |
| 561 | verify_p.add_argument( |
| 562 | "--json", "-j", action="store_true", dest="json_out", |
| 563 | help="Emit a machine-readable JSON report.", |
| 564 | ) |
| 565 | verify_p.set_defaults(func=run_verify) |
| 566 | |
| 567 | # --------------------------------------------------------------------------- |
| 568 | # Subcommand handlers |
| 569 | # --------------------------------------------------------------------------- |
| 570 | |
| 571 | def run_create(args: argparse.Namespace) -> None: |
| 572 | """Pack commits reachable from <refs> into a portable bundle file. |
| 573 | |
| 574 | A bundle is a self-contained msgpack binary carrying commits, snapshots, |
| 575 | and objects. Copy the file over SSH, USB, or email, then apply it with |
| 576 | ``muse bundle unbundle``. Use ``--have`` to produce an incremental bundle |
| 577 | by pruning commits the receiver already has. The output file is written |
| 578 | atomically — a kill signal never leaves a partial bundle at the target path. |
| 579 | |
| 580 | Agent quickstart |
| 581 | ---------------- |
| 582 | :: |
| 583 | |
| 584 | muse bundle create repo.bundle --json |
| 585 | muse bundle create out.bundle feat/audio --json |
| 586 | muse bundle create out.bundle HEAD --have <old-sha> --json |
| 587 | |
| 588 | JSON fields |
| 589 | ----------- |
| 590 | file Absolute path to the bundle file written. |
| 591 | commits Number of commits packed. |
| 592 | objects Number of content objects packed. |
| 593 | size_bytes Final file size in bytes. |
| 594 | branches List of branch names whose tip is included in the bundle. |
| 595 | |
| 596 | Exit codes |
| 597 | ---------- |
| 598 | 0 Bundle created successfully. |
| 599 | 1 Unknown ref, no commits to bundle, or bad ``--have`` ref. |
| 600 | 2 Not inside a Muse repository. |
| 601 | """ |
| 602 | elapsed = start_timer() |
| 603 | file: str = args.file |
| 604 | refs: list[str] | None = args.refs |
| 605 | have: list[str] | None = args.have |
| 606 | json_out: bool = args.json_out |
| 607 | |
| 608 | root = require_repo() |
| 609 | branch = read_current_branch(root) |
| 610 | |
| 611 | want_refs: list[str] = refs or ["HEAD"] |
| 612 | commit_ids = _resolve_refs(root, branch, want_refs) |
| 613 | |
| 614 | if not commit_ids: |
| 615 | print("❌ No commits to bundle.", file=sys.stderr) |
| 616 | raise SystemExit(ExitCode.USER_ERROR) |
| 617 | |
| 618 | have_ids: list[str] = have or [] |
| 619 | bundle = build_mpack(root, commit_ids, have=have_ids) |
| 620 | |
| 621 | # Pre-compute the reachable set once (O(commits)) so the branch-head |
| 622 | # filter below is O(branches) rather than O(branches × commits). |
| 623 | reachable: set[str] = _reachable_from(root, commit_ids) |
| 624 | heads: BranchHeads = {} |
| 625 | for br_name, cid in _iter_branches(root): |
| 626 | if cid in commit_ids or cid in reachable: |
| 627 | heads[br_name] = cid |
| 628 | if heads: |
| 629 | bundle["branch_heads"] = heads |
| 630 | |
| 631 | # Include every local symbol-journal entry in the bundle (muse#53 item |
| 632 | # 4) -- bundles are meant to be a comprehensive, self-contained offline |
| 633 | # transfer, so unlike the incremental --symlog push flag, no opt-in is |
| 634 | # needed here. Not a declared MPack TypedDict field, same as |
| 635 | # branch_heads above -- bundle-only extra data on top of the regular |
| 636 | # push/fetch mpack shape. |
| 637 | symlog_entries_by_symbol: dict[str, list[dict]] = {} |
| 638 | for addr in list_symlog_symbols(root): |
| 639 | local_entries = read_symlog(root, addr, limit=100_000, follow=False) |
| 640 | if local_entries: |
| 641 | symlog_entries_by_symbol[addr] = symlog_entries_to_wire(list(reversed(local_entries))) |
| 642 | if symlog_entries_by_symbol: |
| 643 | bundle["symlogs"] = symlog_entries_by_symbol |
| 644 | |
| 645 | out_path = pathlib.Path(file) |
| 646 | packed = msgpack.packb(bundle, use_bin_type=True) |
| 647 | |
| 648 | # Atomic write: temp file in same directory → os.replace(). |
| 649 | # Ensures a kill signal never leaves a partial bundle at out_path. |
| 650 | tmp_dir = out_path.parent if out_path.parent != pathlib.Path("") else pathlib.Path(".") |
| 651 | tmp_fd, tmp_str = tempfile.mkstemp(dir=tmp_dir, prefix=".bundle-tmp-") |
| 652 | tmp_path_obj = pathlib.Path(tmp_str) |
| 653 | try: |
| 654 | with os.fdopen(tmp_fd, "wb") as fh: |
| 655 | fh.write(packed) |
| 656 | os.replace(tmp_path_obj, out_path) |
| 657 | except Exception: |
| 658 | tmp_path_obj.unlink(missing_ok=True) |
| 659 | raise |
| 660 | |
| 661 | n_commits = len(bundle.get("commits", [])) |
| 662 | n_blobs = len(bundle.get("blobs", [])) |
| 663 | size_bytes = out_path.stat().st_size |
| 664 | |
| 665 | if json_out: |
| 666 | print(json.dumps(_BundleCreateJson( |
| 667 | **make_envelope(elapsed), |
| 668 | file=str(out_path), |
| 669 | commits=n_commits, |
| 670 | blobs=n_blobs, |
| 671 | size_bytes=size_bytes, |
| 672 | branches=sorted(heads.keys()), |
| 673 | ))) |
| 674 | else: |
| 675 | size_kb = size_bytes / 1024 |
| 676 | print( |
| 677 | f"✅ Bundle: {sanitize_display(str(out_path))} " |
| 678 | f"({n_commits} commits, {n_blobs} blobs, {size_kb:.1f} KiB)" |
| 679 | ) |
| 680 | |
| 681 | def _verify_bundle_integrity(bundle: "MPack") -> list[str]: |
| 682 | """Run integrity checks on *bundle* and return a list of failure strings. |
| 683 | |
| 684 | Returns an empty list when the bundle is clean. This is the shared core |
| 685 | used by both ``run_verify`` and the ``--verify`` flag on ``run_unbundle``. |
| 686 | """ |
| 687 | failures: list[str] = [] |
| 688 | |
| 689 | bundle_obj_ids: set[str] = set() |
| 690 | for obj in bundle.get("blobs", []): |
| 691 | obj_id = obj.get("object_id", "") |
| 692 | raw = obj.get("content", b"") |
| 693 | if not obj_id: |
| 694 | failures.append("blobs list: entry has missing object_id") |
| 695 | continue |
| 696 | # NOTE: raw may be b"" for empty files — that is valid. |
| 697 | actual = blob_id(raw) |
| 698 | if actual != obj_id: |
| 699 | failures.append(f"blob {obj_id}: hash mismatch (corruption)") |
| 700 | else: |
| 701 | bundle_obj_ids.add(obj_id) |
| 702 | |
| 703 | for snap_dict in bundle.get("snapshots", []): |
| 704 | snap_id = snap_dict.get("snapshot_id", "") |
| 705 | # Bundles use delta_upsert since feat(pack): delta-encode snapshots (1a6566e77). |
| 706 | # Only new objects introduced by this delta need to be present in the bundle; |
| 707 | # objects from parent snapshots live on the receiver's side. |
| 708 | delta_upsert: dict[str, str] = snap_dict.get("delta_upsert") or {} |
| 709 | for rel_path, obj_id in delta_upsert.items(): |
| 710 | if obj_id not in bundle_obj_ids: |
| 711 | failures.append( |
| 712 | f"snapshot {snap_id}: " |
| 713 | f"missing object {obj_id} for {rel_path}" |
| 714 | ) |
| 715 | return failures |
| 716 | |
| 717 | def run_inspect(args: argparse.Namespace) -> None: |
| 718 | """Show the commit log and branch state recorded in a bundle file. |
| 719 | |
| 720 | Reads all commits, branch heads, and agent provenance from a bundle file |
| 721 | without writing anything to the local repository. Does NOT require a Muse |
| 722 | repository — safe to run against any bundle file in isolation. Commits are |
| 723 | sorted newest-first by ``committed_at``. |
| 724 | |
| 725 | Agent quickstart |
| 726 | ---------------- |
| 727 | :: |
| 728 | |
| 729 | muse bundle inspect repo.bundle --json |
| 730 | muse bundle inspect repo.bundle --json | jq '.commits[].message' |
| 731 | |
| 732 | JSON fields |
| 733 | ----------- |
| 734 | total_commits Number of commits in the bundle. |
| 735 | total_objects Number of content objects in the bundle. |
| 736 | branches Map of branch name → ``sha256:<64hex>`` commit ID. |
| 737 | commits List of commit objects, each with: ``commit_id``, |
| 738 | ``message`` (first line, max 72 chars), ``committed_at`` |
| 739 | (ISO-8601), ``agent_id``, and ``branches`` (list of branch |
| 740 | names whose tip is this commit). |
| 741 | |
| 742 | Exit codes |
| 743 | ---------- |
| 744 | 0 Bundle read successfully (``total_commits=0`` for an empty bundle). |
| 745 | 1 Bundle file not found or not valid msgpack. |
| 746 | """ |
| 747 | elapsed = start_timer() |
| 748 | file: str = args.file |
| 749 | json_out: bool = args.json_out |
| 750 | |
| 751 | bundle = _load_bundle(pathlib.Path(file)) |
| 752 | |
| 753 | # Branch heads map: branch_name → commit_id (sha256: prefixed). |
| 754 | raw_heads: "BranchHeads" = bundle.get("branch_heads") or {} |
| 755 | # Invert for annotation: commit_id → [branch_name, ...] |
| 756 | cid_to_branches: dict[str, list[str]] = {} |
| 757 | for br, cid in raw_heads.items(): |
| 758 | cid_to_branches.setdefault(cid, []).append(br) |
| 759 | |
| 760 | # Build commit entries from the bundle's commits list. |
| 761 | commit_entries: list["_BundleInspectCommit"] = [] |
| 762 | for c in bundle.get("commits", []): |
| 763 | cid = c.get("commit_id", "") |
| 764 | message_raw = c.get("message", "") |
| 765 | message = sanitize_display(message_raw.splitlines()[0][:72]) if message_raw else "" |
| 766 | committed_at = c.get("committed_at", "") |
| 767 | agent_id = sanitize_display(c.get("agent_id", "") or "") |
| 768 | branches = sorted(sanitize_display(b) for b in cid_to_branches.get(cid, [])) |
| 769 | commit_entries.append(_BundleInspectCommit( |
| 770 | commit_id=cid, |
| 771 | message=message, |
| 772 | committed_at=committed_at, |
| 773 | agent_id=agent_id, |
| 774 | branches=branches, |
| 775 | )) |
| 776 | |
| 777 | # Sort newest-first by committed_at (ISO-8601 sorts lexicographically). |
| 778 | commit_entries.sort(key=lambda e: e["committed_at"], reverse=True) |
| 779 | |
| 780 | total_commits = len(commit_entries) |
| 781 | total_blobs = len(bundle.get("blobs", [])) |
| 782 | branches_out = { |
| 783 | sanitize_display(k): sanitize_display(v) |
| 784 | for k, v in raw_heads.items() |
| 785 | } |
| 786 | |
| 787 | if json_out: |
| 788 | print(json.dumps(_BundleInspectJson( |
| 789 | **make_envelope(elapsed), |
| 790 | total_commits=total_commits, |
| 791 | total_objects=total_blobs, |
| 792 | branches=branches_out, |
| 793 | commits=commit_entries, |
| 794 | ))) |
| 795 | else: |
| 796 | print(f"Commits: {total_commits}") |
| 797 | print(f"Blobs: {total_blobs}") |
| 798 | if branches_out: |
| 799 | print(f"Branches: {', '.join(sorted(branches_out))}") |
| 800 | else: |
| 801 | print("Branches: (none)") |
| 802 | print() |
| 803 | for entry in commit_entries: |
| 804 | short = short_id(entry["commit_id"]) |
| 805 | ts = entry["committed_at"][:19] if entry["committed_at"] else "unknown" |
| 806 | agent = f" [{entry['agent_id']}]" if entry["agent_id"] else "" |
| 807 | br_str = f" ({', '.join(entry['branches'])})" if entry["branches"] else "" |
| 808 | print(f"{short} {ts}{agent}{br_str}") |
| 809 | print(f" {entry['message']}") |
| 810 | |
| 811 | def run_diff(args: argparse.Namespace) -> None: |
| 812 | """Show which commits in a bundle are not yet in the local repository. |
| 813 | |
| 814 | Compares each bundle commit against the local store and reports which are |
| 815 | new. Also reports which branch refs would advance if the bundle were |
| 816 | applied. ``new_commits=0`` means the repo is already up-to-date. Requires |
| 817 | a Muse repository (unlike ``inspect``, ``verify``, ``list-heads``). |
| 818 | |
| 819 | Agent quickstart |
| 820 | ---------------- |
| 821 | :: |
| 822 | |
| 823 | muse bundle diff repo.bundle --json |
| 824 | muse bundle diff repo.bundle --json | jq '.new_commits' |
| 825 | |
| 826 | JSON fields |
| 827 | ----------- |
| 828 | new_commits Number of commits in the bundle not yet in the local store. |
| 829 | known_commits Number of commits already present locally. |
| 830 | refs_to_advance Branch names whose local tip differs from the bundle tip. |
| 831 | commits List of new commit objects, each with: ``commit_id``, |
| 832 | ``message`` (first line, max 72 chars), ``committed_at`` |
| 833 | (ISO-8601). Sorted newest-first. |
| 834 | |
| 835 | Exit codes |
| 836 | ---------- |
| 837 | 0 Diff computed (even when ``new_commits == 0``). |
| 838 | 1 Bundle file not found or not valid msgpack. |
| 839 | 2 Not inside a Muse repository. |
| 840 | """ |
| 841 | elapsed = start_timer() |
| 842 | file: str = args.file |
| 843 | json_out: bool = args.json_out |
| 844 | |
| 845 | root = require_repo() |
| 846 | bundle = _load_bundle(pathlib.Path(file)) |
| 847 | |
| 848 | new_entries: list[_BundleDiffCommit] = [] |
| 849 | known_count = 0 |
| 850 | |
| 851 | for c in bundle.get("commits", []): |
| 852 | cid = c.get("commit_id", "") |
| 853 | if not cid: |
| 854 | continue |
| 855 | local_rec = read_commit(root, cid) |
| 856 | if local_rec is None: |
| 857 | message_raw = c.get("message", "") |
| 858 | message = sanitize_display(message_raw.splitlines()[0][:72]) if message_raw else "" |
| 859 | new_entries.append(_BundleDiffCommit( |
| 860 | commit_id=cid, |
| 861 | message=message, |
| 862 | committed_at=c.get("committed_at", ""), |
| 863 | )) |
| 864 | else: |
| 865 | known_count += 1 |
| 866 | |
| 867 | # Sort new commits newest-first. |
| 868 | new_entries.sort(key=lambda e: e["committed_at"], reverse=True) |
| 869 | |
| 870 | # Determine which branch refs would advance. |
| 871 | branch_heads: "BranchHeads" = bundle.get("branch_heads") or {} |
| 872 | refs_to_advance: list[str] = [] |
| 873 | for br, bundle_cid in sorted(branch_heads.items()): |
| 874 | local_cid = get_head_commit_id(root, br) |
| 875 | if local_cid != bundle_cid: |
| 876 | refs_to_advance.append(sanitize_display(br)) |
| 877 | |
| 878 | if json_out: |
| 879 | print(json.dumps(_BundleDiffJson( |
| 880 | **make_envelope(elapsed), |
| 881 | new_commits=len(new_entries), |
| 882 | known_commits=known_count, |
| 883 | refs_to_advance=refs_to_advance, |
| 884 | commits=new_entries, |
| 885 | ))) |
| 886 | else: |
| 887 | if not new_entries: |
| 888 | print(f"Already up-to-date. 0 new commits.") |
| 889 | else: |
| 890 | print(f"{len(new_entries)} new commit(s), {known_count} already known.") |
| 891 | if refs_to_advance: |
| 892 | print(f"Refs to advance: {', '.join(refs_to_advance)}") |
| 893 | for entry in new_entries: |
| 894 | short = short_id(entry["commit_id"]) |
| 895 | ts = entry["committed_at"][:19] if entry["committed_at"] else "unknown" |
| 896 | print(f" {short} {ts} {entry['message']}") |
| 897 | |
| 898 | def run_unbundle(args: argparse.Namespace) -> None: |
| 899 | """Apply a bundle to the local store and optionally advance branch refs. |
| 900 | |
| 901 | Writes all commits, snapshots, and objects from a bundle into the local |
| 902 | repository, then advances local branch refs to match the bundle's recorded |
| 903 | heads. All writes are idempotent — objects already present are counted as |
| 904 | skipped, not errors. Use ``--verify`` to abort before any write if the |
| 905 | bundle fails an integrity check. |
| 906 | |
| 907 | Agent quickstart |
| 908 | ---------------- |
| 909 | :: |
| 910 | |
| 911 | muse bundle unbundle repo.bundle --verify --json |
| 912 | muse bundle unbundle repo.bundle --no-update-refs --json |
| 913 | |
| 914 | JSON fields |
| 915 | ----------- |
| 916 | commits_written Number of commit records written to the local store. |
| 917 | snapshots_written Number of snapshot records written. |
| 918 | blobs_written Number of blobs written. |
| 919 | blobs_skipped Number of blobs already present (skipped). |
| 920 | refs_updated Branch names whose local tip was advanced. |
| 921 | verified ``true`` if ``--verify`` was set and integrity passed. |
| 922 | |
| 923 | Exit codes |
| 924 | ---------- |
| 925 | 0 Bundle applied (all objects skipped = already up-to-date, also 0). |
| 926 | 1 Bundle not found, not valid msgpack, or ``--verify`` integrity failure. |
| 927 | 2 Not inside a Muse repository. |
| 928 | """ |
| 929 | elapsed = start_timer() |
| 930 | file: str = args.file |
| 931 | update_refs: bool = args.update_refs |
| 932 | verify_first: bool = args.verify_first |
| 933 | json_out: bool = args.json_out |
| 934 | |
| 935 | root = require_repo() |
| 936 | bundle = _load_bundle(pathlib.Path(file)) |
| 937 | |
| 938 | if verify_first: |
| 939 | failures = _verify_bundle_integrity(bundle) |
| 940 | if failures: |
| 941 | if json_out: |
| 942 | print(json.dumps({ |
| 943 | "error": "integrity_failure", |
| 944 | "failures": [sanitize_display(f) for f in failures[:10]], |
| 945 | })) |
| 946 | else: |
| 947 | print( |
| 948 | f"❌ Bundle integrity check failed ({len(failures)} failure(s)) — " |
| 949 | "no changes written.", |
| 950 | file=sys.stderr, |
| 951 | ) |
| 952 | for f in failures[:10]: |
| 953 | print(f" {sanitize_display(f)}", file=sys.stderr) |
| 954 | raise SystemExit(ExitCode.USER_ERROR) |
| 955 | |
| 956 | result = apply_mpack(root, bundle) |
| 957 | |
| 958 | updated: list[str] = [] |
| 959 | if update_refs: |
| 960 | import re as _re |
| 961 | branch_heads: BranchHeads = bundle.get("branch_heads") or {} |
| 962 | for br, cid in branch_heads.items(): |
| 963 | try: |
| 964 | validate_branch_name(br) |
| 965 | except ValueError: |
| 966 | logger.warning("⚠️ bundle: skipping invalid branch name %r", br) |
| 967 | continue |
| 968 | if not _re.fullmatch(r"sha256:[0-9a-f]{64}", cid): |
| 969 | logger.warning("⚠️ bundle: skipping invalid commit ID for %r", br) |
| 970 | continue |
| 971 | write_branch_ref(root, br, cid) |
| 972 | updated.append(br) |
| 973 | |
| 974 | # Merge any symbol-journal entries carried in the bundle (muse#53 item |
| 975 | # 4), directly into the local journals -- best-effort, never fails the |
| 976 | # unbundle. See merge_remote_symlog_entries()'s docstring for why no |
| 977 | # remote-tracking namespace split is needed here (unlike reflog). |
| 978 | symlog_merged_total = 0 |
| 979 | try: |
| 980 | for addr, wire_entries in (bundle.get("symlogs") or {}).items(): |
| 981 | symlog_merged_total += merge_remote_symlog_entries(root, addr, wire_entries) |
| 982 | except Exception as _symlog_exc: |
| 983 | logger.warning("⚠️ bundle: symlog merge failed (non-fatal): %s", _symlog_exc) |
| 984 | |
| 985 | if json_out: |
| 986 | print(json.dumps(_BundleUnbundleJson( |
| 987 | **make_envelope(elapsed), |
| 988 | commits_written=result["commits_written"], |
| 989 | snapshots_written=result["snapshots_written"], |
| 990 | blobs_written=result["blobs_written"], |
| 991 | blobs_skipped=result["blobs_skipped"], |
| 992 | refs_updated=sorted(updated), |
| 993 | verified=verify_first, |
| 994 | symlog_entries_merged=symlog_merged_total, |
| 995 | ))) |
| 996 | else: |
| 997 | print( |
| 998 | f"Unpacked {result['commits_written']} commit(s), " |
| 999 | f"{result['snapshots_written']} snapshot(s), " |
| 1000 | f"{result['blobs_written']} blob(s) " |
| 1001 | f"({result['blobs_skipped']} skipped)" |
| 1002 | ) |
| 1003 | if updated: |
| 1004 | print( |
| 1005 | f"Updated refs: {', '.join(sanitize_display(b) for b in updated)}" |
| 1006 | ) |
| 1007 | if symlog_merged_total: |
| 1008 | print(f"Merged {symlog_merged_total} symlog entr" |
| 1009 | f"{'y' if symlog_merged_total == 1 else 'ies'}.") |
| 1010 | print("✅ Bundle applied.") |
| 1011 | |
| 1012 | def run_verify(args: argparse.Namespace) -> None: |
| 1013 | """Verify the integrity of a bundle file. |
| 1014 | |
| 1015 | Checks every object's SHA-256 against its declared ``object_id`` (hash |
| 1016 | mismatch means corruption) and confirms every snapshot's objects are present |
| 1017 | in the bundle (missing object means truncation). Does NOT require a Muse |
| 1018 | repository. At most 100 distinct failures are collected; beyond that a |
| 1019 | trailing "... and N more" entry is appended. |
| 1020 | |
| 1021 | Agent quickstart |
| 1022 | ---------------- |
| 1023 | :: |
| 1024 | |
| 1025 | muse bundle verify repo.bundle --json |
| 1026 | muse bundle verify repo.bundle -q && muse bundle unbundle repo.bundle |
| 1027 | |
| 1028 | JSON fields |
| 1029 | ----------- |
| 1030 | blobs_checked Number of blobs whose hash was verified. |
| 1031 | snapshots_checked Number of snapshots whose blob references were checked. |
| 1032 | all_ok ``true`` when no integrity failures were found. |
| 1033 | failures List of human-readable failure descriptions (capped at 100). |
| 1034 | |
| 1035 | Exit codes |
| 1036 | ---------- |
| 1037 | 0 Bundle is clean. |
| 1038 | 1 Integrity failures found, or bundle file not found / not valid msgpack. |
| 1039 | """ |
| 1040 | elapsed = start_timer() |
| 1041 | file: str = args.file |
| 1042 | quiet: bool = args.quiet |
| 1043 | json_out: bool = args.json_out |
| 1044 | |
| 1045 | bundle = _load_bundle(pathlib.Path(file)) |
| 1046 | |
| 1047 | failures: list[str] = [] |
| 1048 | blobs_checked = 0 |
| 1049 | snapshots_checked = 0 |
| 1050 | _total_failures = 0 # tracks raw count before the cap kicks in |
| 1051 | |
| 1052 | def _add_failure(msg: str) -> None: |
| 1053 | nonlocal _total_failures |
| 1054 | _total_failures += 1 |
| 1055 | if len(failures) < _MAX_VERIFY_FAILURES: |
| 1056 | failures.append(msg) |
| 1057 | |
| 1058 | # Build set of blob IDs in the bundle and verify each hash. |
| 1059 | # Blob IDs use canonical "sha256:<64hex>" format; compare accordingly. |
| 1060 | bundle_obj_ids: set[str] = set() |
| 1061 | for obj in bundle.get("blobs", []): |
| 1062 | obj_id = obj["object_id"] |
| 1063 | raw = obj["content"] |
| 1064 | if not obj_id: |
| 1065 | _add_failure("blobs list: entry has missing object_id") |
| 1066 | continue |
| 1067 | # NOTE: raw may be b"" for empty files — that is valid. |
| 1068 | actual = blob_id(raw) |
| 1069 | if actual != obj_id: |
| 1070 | _add_failure(f"blob {obj_id}: hash mismatch (corruption)") |
| 1071 | else: |
| 1072 | bundle_obj_ids.add(obj_id) |
| 1073 | blobs_checked += 1 |
| 1074 | |
| 1075 | # Check snapshots reference objects present in the bundle. |
| 1076 | for snap_dict in bundle.get("snapshots", []): |
| 1077 | snap_id = snap_dict.get("snapshot_id", "") |
| 1078 | delta_upsert: dict[str, str] = snap_dict.get("delta_upsert") or {} |
| 1079 | snapshots_checked += 1 |
| 1080 | for rel_path, obj_id in delta_upsert.items(): |
| 1081 | if obj_id not in bundle_obj_ids: |
| 1082 | _add_failure( |
| 1083 | f"snapshot {snap_id}: " |
| 1084 | f"missing object {obj_id} for {rel_path}" |
| 1085 | ) |
| 1086 | |
| 1087 | overflow = _total_failures - len(failures) |
| 1088 | if overflow > 0: |
| 1089 | failures.append(f"... and {overflow} more failure(s) (cap: {_MAX_VERIFY_FAILURES})") |
| 1090 | |
| 1091 | all_ok = _total_failures == 0 |
| 1092 | |
| 1093 | if quiet: |
| 1094 | raise SystemExit(0 if all_ok else ExitCode.USER_ERROR) |
| 1095 | |
| 1096 | if json_out: |
| 1097 | print(json.dumps(_BundleVerifyJson( |
| 1098 | **make_envelope(elapsed), |
| 1099 | blobs_checked=blobs_checked, |
| 1100 | snapshots_checked=snapshots_checked, |
| 1101 | all_ok=all_ok, |
| 1102 | failures=[sanitize_display(f) for f in failures], |
| 1103 | ))) |
| 1104 | else: |
| 1105 | print(f"Blobs checked: {blobs_checked}") |
| 1106 | print(f"Snapshots checked: {snapshots_checked}") |
| 1107 | if all_ok: |
| 1108 | print("✅ Bundle is clean.") |
| 1109 | else: |
| 1110 | print(f"❌ {_total_failures} failure(s):") |
| 1111 | for f in failures: |
| 1112 | print(f" {sanitize_display(str(f))}") |
| 1113 | |
| 1114 | raise SystemExit(0 if all_ok else ExitCode.USER_ERROR) |
| 1115 | |
| 1116 | def run_list_heads(args: argparse.Namespace) -> None: |
| 1117 | """List the branch heads recorded in a bundle file. |
| 1118 | |
| 1119 | Reads the ``branch_heads`` map embedded in the bundle and prints it. Does |
| 1120 | NOT require a Muse repository — runs against any bundle file in isolation. |
| 1121 | Both branch names and commit IDs are sanitized before output to prevent |
| 1122 | ANSI injection from crafted bundles. |
| 1123 | |
| 1124 | Agent quickstart |
| 1125 | ---------------- |
| 1126 | :: |
| 1127 | |
| 1128 | muse bundle list-heads repo.bundle --json |
| 1129 | muse bundle list-heads repo.bundle --json | jq 'keys' |
| 1130 | |
| 1131 | JSON fields |
| 1132 | ----------- |
| 1133 | <branch> Each key is a branch name; value is ``sha256:<64hex>`` commit ID. |
| 1134 | Top level is a flat map — no wrapper key. |
| 1135 | |
| 1136 | Exit codes |
| 1137 | ---------- |
| 1138 | 0 Always (empty object when no heads are recorded in the bundle). |
| 1139 | 1 Bundle file not found or not valid msgpack. |
| 1140 | """ |
| 1141 | elapsed = start_timer() |
| 1142 | file: str = args.file |
| 1143 | json_out: bool = args.json_out |
| 1144 | |
| 1145 | bundle = _load_bundle(pathlib.Path(file)) |
| 1146 | heads: BranchHeads = bundle.get("branch_heads") or {} |
| 1147 | |
| 1148 | if json_out: |
| 1149 | safe_heads = { |
| 1150 | sanitize_display(k): sanitize_display(v) |
| 1151 | for k, v in heads.items() |
| 1152 | } |
| 1153 | print(json.dumps({ |
| 1154 | **make_envelope(elapsed), |
| 1155 | **_BundleListHeadsJson(heads=safe_heads), |
| 1156 | })) |
| 1157 | else: |
| 1158 | if not heads: |
| 1159 | print("No branch heads in bundle.") |
| 1160 | return |
| 1161 | for branch, cid in sorted(heads.items()): |
| 1162 | print(f"{sanitize_display(cid[:len('sha256:') + 12])} {sanitize_display(branch)}") |
File History
1 commit
sha256:b570925a487b44187c0441977d252906b6bf2719253bc53f96412d520cd24d9e
feat: muse#53 item 4 — remote symlog sharing via push/pull/bundle
Sonnet 5
minor
⚠
4 hours ago