read_tree.py
python
| 1 | """muse read-tree — read a snapshot into the muse-work/ directory. |
| 2 | |
| 3 | This is a low-level plumbing command (analogous to ``git read-tree``) that |
| 4 | hydrates ``muse-work/`` from a stored snapshot manifest WITHOUT touching any |
| 5 | branch or HEAD references. It is intentionally destructive by default: any |
| 6 | file in ``muse-work/`` whose path appears in the snapshot is overwritten. |
| 7 | |
| 8 | Use cases |
| 9 | --------- |
| 10 | - Inspect an older snapshot without losing your current branch position. |
| 11 | - Restore a specific state before running ``muse commit``. |
| 12 | - Agent tooling that needs to populate a clean working directory from a |
| 13 | known snapshot (e.g., after ``muse pull`` or before an automated test). |
| 14 | |
| 15 | Flags |
| 16 | ----- |
| 17 | ``<snapshot_id>`` |
| 18 | Positional — the full or abbreviated (≥ 4 chars) snapshot SHA to restore. |
| 19 | |
| 20 | ``--dry-run`` |
| 21 | Print the list of files that *would* be written without touching disk. |
| 22 | Exit 0 on success. |
| 23 | |
| 24 | ``--reset`` |
| 25 | Remove all existing files from ``muse-work/`` before populating. |
| 26 | Without this flag, only the files referenced by the snapshot are |
| 27 | written (files not in the snapshot are left untouched). |
| 28 | |
| 29 | Algorithm |
| 30 | --------- |
| 31 | 1. Resolve *snapshot_id* — accept full 64-char SHA or a ≥4-char prefix via |
| 32 | a DB prefix scan. |
| 33 | 2. Fetch the snapshot manifest: ``{rel_path → object_id}``. |
| 34 | 3. For each entry, read the object from ``.muse/objects/<object_id>``. |
| 35 | 4. If ``--reset``: remove all files currently in ``muse-work/``. |
| 36 | 5. Write each file to ``muse-work/<rel_path>`` (creating parent dirs). |
| 37 | 6. Does NOT update ``.muse/HEAD`` or any branch ref. |
| 38 | """ |
| 39 | from __future__ import annotations |
| 40 | |
| 41 | import asyncio |
| 42 | import logging |
| 43 | import pathlib |
| 44 | |
| 45 | import typer |
| 46 | from sqlalchemy.ext.asyncio import AsyncSession |
| 47 | from sqlalchemy.future import select |
| 48 | |
| 49 | from maestro.muse_cli._repo import require_repo |
| 50 | from maestro.muse_cli.db import open_session |
| 51 | from maestro.muse_cli.errors import ExitCode |
| 52 | from maestro.muse_cli.models import MuseCliSnapshot |
| 53 | from maestro.muse_cli.object_store import read_object |
| 54 | |
| 55 | logger = logging.getLogger(__name__) |
| 56 | |
| 57 | app = typer.Typer() |
| 58 | |
| 59 | # Minimum prefix length for abbreviated snapshot IDs. |
| 60 | _MIN_PREFIX_LEN = 4 |
| 61 | |
| 62 | |
| 63 | # --------------------------------------------------------------------------- |
| 64 | # Result type |
| 65 | # --------------------------------------------------------------------------- |
| 66 | |
| 67 | |
| 68 | class ReadTreeResult: |
| 69 | """Structured result of a ``read-tree`` operation. |
| 70 | |
| 71 | Carries the set of files written (or that would have been written in |
| 72 | dry-run mode) so that tests can assert on content without inspecting |
| 73 | the filesystem. |
| 74 | |
| 75 | Attributes: |
| 76 | snapshot_id: Full 64-char snapshot ID that was resolved. |
| 77 | files_written: Relative paths of files written to muse-work/. |
| 78 | dry_run: True when ``--dry-run`` was requested (no writes made). |
| 79 | reset: True when ``--reset`` cleared muse-work/ first. |
| 80 | """ |
| 81 | |
| 82 | def __init__( |
| 83 | self, |
| 84 | *, |
| 85 | snapshot_id: str, |
| 86 | files_written: list[str], |
| 87 | dry_run: bool, |
| 88 | reset: bool, |
| 89 | ) -> None: |
| 90 | self.snapshot_id = snapshot_id |
| 91 | self.files_written = files_written |
| 92 | self.dry_run = dry_run |
| 93 | self.reset = reset |
| 94 | |
| 95 | def __repr__(self) -> str: |
| 96 | return ( |
| 97 | f"<ReadTreeResult snap={self.snapshot_id[:8]}" |
| 98 | f" files={len(self.files_written)}" |
| 99 | f" dry_run={self.dry_run} reset={self.reset}>" |
| 100 | ) |
| 101 | |
| 102 | |
| 103 | # --------------------------------------------------------------------------- |
| 104 | # Testable async core |
| 105 | # --------------------------------------------------------------------------- |
| 106 | |
| 107 | |
| 108 | async def _resolve_snapshot( |
| 109 | session: AsyncSession, |
| 110 | snapshot_id: str, |
| 111 | ) -> MuseCliSnapshot | None: |
| 112 | """Resolve *snapshot_id* to a :class:`MuseCliSnapshot` row. |
| 113 | |
| 114 | Accepts both full 64-char SHAs and abbreviated prefixes (≥ 4 chars). |
| 115 | When a prefix is provided, a table scan is used to find the first |
| 116 | matching row (acceptable for CLI use; snapshot IDs are content-addressed |
| 117 | so collisions are astronomically unlikely). |
| 118 | |
| 119 | Returns ``None`` when no snapshot matches. |
| 120 | """ |
| 121 | if len(snapshot_id) == 64: |
| 122 | return await session.get(MuseCliSnapshot, snapshot_id) |
| 123 | |
| 124 | # Abbreviated prefix scan. |
| 125 | result = await session.execute( |
| 126 | select(MuseCliSnapshot).where( |
| 127 | MuseCliSnapshot.snapshot_id.startswith(snapshot_id) |
| 128 | ) |
| 129 | ) |
| 130 | return result.scalars().first() |
| 131 | |
| 132 | |
| 133 | async def _read_tree_async( |
| 134 | *, |
| 135 | snapshot_id: str, |
| 136 | root: pathlib.Path, |
| 137 | session: AsyncSession, |
| 138 | dry_run: bool = False, |
| 139 | reset: bool = False, |
| 140 | ) -> ReadTreeResult: |
| 141 | """Core read-tree logic — fully injectable for tests. |
| 142 | |
| 143 | Resolves the snapshot from the DB, then hydrates ``muse-work/`` from the |
| 144 | local object store. Raises ``typer.Exit`` with a clean exit code on any |
| 145 | user-facing error so the Typer callback can surface it without a traceback. |
| 146 | |
| 147 | Args: |
| 148 | snapshot_id: Full or abbreviated snapshot ID to restore. |
| 149 | root: Muse repository root (directory containing ``.muse/``). |
| 150 | session: Open async DB session used for snapshot lookup. |
| 151 | dry_run: When True, report what would be written but do nothing. |
| 152 | reset: When True, clear muse-work/ before populating. |
| 153 | |
| 154 | Returns: |
| 155 | :class:`ReadTreeResult` describing what was (or would have been) done. |
| 156 | |
| 157 | Raises: |
| 158 | typer.Exit: On user-facing errors (unknown snapshot, missing objects). |
| 159 | """ |
| 160 | if len(snapshot_id) < _MIN_PREFIX_LEN: |
| 161 | typer.echo( |
| 162 | f"❌ Snapshot ID too short: '{snapshot_id}' " |
| 163 | f"(need at least {_MIN_PREFIX_LEN} hex chars)." |
| 164 | ) |
| 165 | raise typer.Exit(code=ExitCode.USER_ERROR) |
| 166 | |
| 167 | # ── 1. Resolve snapshot ────────────────────────────────────────────── |
| 168 | snapshot = await _resolve_snapshot(session, snapshot_id.lower()) |
| 169 | if snapshot is None: |
| 170 | typer.echo(f"❌ No snapshot found matching '{snapshot_id[:8]}'.") |
| 171 | raise typer.Exit(code=ExitCode.USER_ERROR) |
| 172 | |
| 173 | manifest: dict[str, str] = dict(snapshot.manifest) |
| 174 | if not manifest: |
| 175 | typer.echo(f"⚠️ Snapshot {snapshot.snapshot_id[:8]} has an empty manifest.") |
| 176 | raise typer.Exit(code=ExitCode.USER_ERROR) |
| 177 | |
| 178 | workdir = root / "muse-work" |
| 179 | sorted_paths = sorted(manifest.keys()) |
| 180 | |
| 181 | # ── 2. Dry-run: list files and exit ───────────────────────────────── |
| 182 | if dry_run: |
| 183 | typer.echo(f"Snapshot {snapshot.snapshot_id[:8]} — {len(manifest)} file(s):") |
| 184 | for rel_path in sorted_paths: |
| 185 | obj_id = manifest[rel_path] |
| 186 | typer.echo(f" {rel_path} ({obj_id[:8]})") |
| 187 | return ReadTreeResult( |
| 188 | snapshot_id=snapshot.snapshot_id, |
| 189 | files_written=sorted_paths, |
| 190 | dry_run=True, |
| 191 | reset=reset, |
| 192 | ) |
| 193 | |
| 194 | # ── 3. Pre-flight: verify all objects are present in the store ─────── |
| 195 | missing: list[str] = [] |
| 196 | for rel_path, object_id in manifest.items(): |
| 197 | content = read_object(root, object_id) |
| 198 | if content is None: |
| 199 | missing.append(rel_path) |
| 200 | |
| 201 | if missing: |
| 202 | typer.echo( |
| 203 | f"❌ {len(missing)} object(s) missing from local store " |
| 204 | f"(snapshot {snapshot.snapshot_id[:8]}):" |
| 205 | ) |
| 206 | for p in sorted(missing): |
| 207 | typer.echo(f" {p} ({manifest[p][:8]})") |
| 208 | typer.echo( |
| 209 | " Objects are written by 'muse commit'. " |
| 210 | "Re-commit this snapshot or fetch it via 'muse pull'." |
| 211 | ) |
| 212 | raise typer.Exit(code=ExitCode.USER_ERROR) |
| 213 | |
| 214 | # ── 4. Optional reset: remove all files from muse-work/ ───────────── |
| 215 | if reset and workdir.exists(): |
| 216 | removed = 0 |
| 217 | for existing_file in sorted(workdir.rglob("*")): |
| 218 | if existing_file.is_file(): |
| 219 | existing_file.unlink() |
| 220 | removed += 1 |
| 221 | logger.info("⚠️ --reset: removed %d file(s) from muse-work/", removed) |
| 222 | # Clean up empty directories left behind. |
| 223 | for d in sorted(workdir.rglob("*"), reverse=True): |
| 224 | if d.is_dir(): |
| 225 | try: |
| 226 | d.rmdir() |
| 227 | except OSError: |
| 228 | pass # Not empty — leave it. |
| 229 | |
| 230 | # ── 5. Write objects to muse-work/ ────────────────────────────────── |
| 231 | workdir.mkdir(parents=True, exist_ok=True) |
| 232 | files_written: list[str] = [] |
| 233 | |
| 234 | for rel_path in sorted_paths: |
| 235 | object_id = manifest[rel_path] |
| 236 | content = read_object(root, object_id) |
| 237 | # content is guaranteed non-None by the pre-flight check above. |
| 238 | assert content is not None |
| 239 | |
| 240 | dest = workdir / rel_path |
| 241 | dest.parent.mkdir(parents=True, exist_ok=True) |
| 242 | dest.write_bytes(content) |
| 243 | files_written.append(rel_path) |
| 244 | logger.debug("✅ Wrote %s (%d bytes)", rel_path, len(content)) |
| 245 | |
| 246 | return ReadTreeResult( |
| 247 | snapshot_id=snapshot.snapshot_id, |
| 248 | files_written=files_written, |
| 249 | dry_run=False, |
| 250 | reset=reset, |
| 251 | ) |
| 252 | |
| 253 | |
| 254 | # --------------------------------------------------------------------------- |
| 255 | # Typer command |
| 256 | # --------------------------------------------------------------------------- |
| 257 | |
| 258 | |
| 259 | @app.callback(invoke_without_command=True) |
| 260 | def read_tree( |
| 261 | ctx: typer.Context, |
| 262 | snapshot_id: str = typer.Argument( |
| 263 | ..., |
| 264 | help=( |
| 265 | "Snapshot ID to restore into muse-work/. " |
| 266 | "Accepts the full 64-char SHA or an abbreviated prefix (≥ 4 chars)." |
| 267 | ), |
| 268 | ), |
| 269 | dry_run: bool = typer.Option( |
| 270 | False, |
| 271 | "--dry-run", |
| 272 | help="Print the file list without writing anything.", |
| 273 | ), |
| 274 | reset: bool = typer.Option( |
| 275 | False, |
| 276 | "--reset", |
| 277 | help="Remove all existing muse-work/ files before populating.", |
| 278 | ), |
| 279 | ) -> None: |
| 280 | """Read a snapshot into muse-work/ without updating HEAD. |
| 281 | |
| 282 | Populates muse-work/ with the exact file set recorded in SNAPSHOT_ID. |
| 283 | Files not referenced by the snapshot are left untouched unless --reset |
| 284 | is specified. HEAD and branch refs are never modified. |
| 285 | """ |
| 286 | root = require_repo() |
| 287 | |
| 288 | async def _run() -> ReadTreeResult: |
| 289 | async with open_session() as session: |
| 290 | return await _read_tree_async( |
| 291 | snapshot_id=snapshot_id, |
| 292 | root=root, |
| 293 | session=session, |
| 294 | dry_run=dry_run, |
| 295 | reset=reset, |
| 296 | ) |
| 297 | |
| 298 | try: |
| 299 | result = asyncio.run(_run()) |
| 300 | except typer.Exit: |
| 301 | raise |
| 302 | except Exception as exc: |
| 303 | typer.echo(f"❌ muse read-tree failed: {exc}") |
| 304 | logger.error("❌ muse read-tree error: %s", exc, exc_info=True) |
| 305 | raise typer.Exit(code=ExitCode.INTERNAL_ERROR) |
| 306 | |
| 307 | if result.dry_run: |
| 308 | return |
| 309 | |
| 310 | action = "reset and populated" if result.reset else "populated" |
| 311 | typer.echo( |
| 312 | f"✅ muse-work/ {action} from snapshot {result.snapshot_id[:8]} " |
| 313 | f"({len(result.files_written)} file(s))." |
| 314 | ) |
| 315 | logger.info( |
| 316 | "✅ read-tree snapshot=%s files=%d reset=%s", |
| 317 | result.snapshot_id[:8], |
| 318 | len(result.files_written), |
| 319 | result.reset, |
| 320 | ) |