test_cmd_content_grep.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
128 days ago
| 1 | """Tests for ``muse content-grep``. |
| 2 | |
| 3 | Covers: no match exit-1, pattern found, --files-only, --count, --ignore-case, |
| 4 | --format json, binary skip, multi-file, stress: 100 files. |
| 5 | Working-tree mode: --working-tree searches disk, not the committed snapshot. |
| 6 | """ |
| 7 | |
| 8 | from __future__ import annotations |
| 9 | |
| 10 | type _FileStore = dict[str, bytes] |
| 11 | |
| 12 | import datetime |
| 13 | import json |
| 14 | import pathlib |
| 15 | |
| 16 | import pytest |
| 17 | from tests.cli_test_helper import CliRunner |
| 18 | |
| 19 | cli = None # argparse migration — CliRunner ignores this arg |
| 20 | from muse.core.object_store import write_object |
| 21 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 22 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 23 | from muse.core.types import Manifest, blob_id |
| 24 | from muse.core.paths import heads_dir, muse_dir |
| 25 | |
| 26 | runner = CliRunner() |
| 27 | |
| 28 | _REPO_ID = "cgrep-test" |
| 29 | |
| 30 | |
| 31 | # --------------------------------------------------------------------------- |
| 32 | # Helpers |
| 33 | # --------------------------------------------------------------------------- |
| 34 | |
| 35 | |
| 36 | |
| 37 | |
| 38 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 39 | dot_muse = muse_dir(path) |
| 40 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 41 | (dot_muse / d).mkdir(parents=True, exist_ok=True) |
| 42 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 43 | (dot_muse / "repo.json").write_text( |
| 44 | json.dumps({"repo_id": _REPO_ID, "domain": "midi"}), encoding="utf-8" |
| 45 | ) |
| 46 | return path |
| 47 | |
| 48 | |
| 49 | def _env(repo: pathlib.Path) -> Manifest: |
| 50 | return {"MUSE_REPO_ROOT": str(repo)} |
| 51 | |
| 52 | |
| 53 | _counter = 0 |
| 54 | |
| 55 | |
| 56 | def _commit_files(root: pathlib.Path, files: _FileStore) -> str: |
| 57 | global _counter |
| 58 | _counter += 1 |
| 59 | manifest: Manifest = {} |
| 60 | for rel_path, content in files.items(): |
| 61 | obj_id = blob_id(content) |
| 62 | write_object(root, obj_id, content) |
| 63 | manifest[rel_path] = obj_id |
| 64 | snap_id = compute_snapshot_id(manifest) |
| 65 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 66 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 67 | commit_id = compute_commit_id( parent_ids=[], |
| 68 | snapshot_id=snap_id, |
| 69 | message=f"commit {_counter}", |
| 70 | committed_at_iso=committed_at.isoformat(), |
| 71 | ) |
| 72 | write_commit(root, CommitRecord( |
| 73 | commit_id=commit_id, |
| 74 | repo_id="test-repo", |
| 75 | branch="main", |
| 76 | snapshot_id=snap_id, |
| 77 | message=f"commit {_counter}", |
| 78 | committed_at=committed_at, |
| 79 | )) |
| 80 | (heads_dir(root) / "main").write_text(commit_id, encoding="utf-8") |
| 81 | return commit_id |
| 82 | |
| 83 | |
| 84 | # --------------------------------------------------------------------------- |
| 85 | # Unit: help |
| 86 | # --------------------------------------------------------------------------- |
| 87 | |
| 88 | |
| 89 | def test_content_grep_help() -> None: |
| 90 | result = runner.invoke(cli, ["content-grep", "--help"]) |
| 91 | assert result.exit_code == 0 |
| 92 | assert "pattern" in result.output |
| 93 | |
| 94 | |
| 95 | # --------------------------------------------------------------------------- |
| 96 | # Unit: no match → exit 1 |
| 97 | # --------------------------------------------------------------------------- |
| 98 | |
| 99 | |
| 100 | def test_content_grep_no_match(tmp_path: pathlib.Path) -> None: |
| 101 | _init_repo(tmp_path) |
| 102 | _commit_files(tmp_path, {"song.txt": b"chord: Am\ntempo: 120\n"}) |
| 103 | result = runner.invoke(cli, ["content-grep", "ZZZNOMATCH", "--json"], env=_env(tmp_path)) |
| 104 | assert result.exit_code != 0 |
| 105 | # --json must always emit valid JSON even on no-match so agents can parse safely. |
| 106 | data = json.loads(result.output) |
| 107 | assert data["total_matches"] == 0 |
| 108 | assert data["results"] == [] |
| 109 | |
| 110 | |
| 111 | # --------------------------------------------------------------------------- |
| 112 | # Unit: match found → exit 0 |
| 113 | # --------------------------------------------------------------------------- |
| 114 | |
| 115 | |
| 116 | def test_content_grep_match_found(tmp_path: pathlib.Path) -> None: |
| 117 | _init_repo(tmp_path) |
| 118 | _commit_files(tmp_path, {"song.txt": b"chord: Cm7\ntempo: 120\n"}) |
| 119 | result = runner.invoke(cli, ["content-grep", "Cm7"], env=_env(tmp_path)) |
| 120 | assert result.exit_code == 0 |
| 121 | assert "song.txt" in result.output |
| 122 | |
| 123 | |
| 124 | # --------------------------------------------------------------------------- |
| 125 | # Unit: --ignore-case |
| 126 | # --------------------------------------------------------------------------- |
| 127 | |
| 128 | |
| 129 | def test_content_grep_ignore_case(tmp_path: pathlib.Path) -> None: |
| 130 | _init_repo(tmp_path) |
| 131 | _commit_files(tmp_path, {"notes.txt": b"VERSE: intro melody\n"}) |
| 132 | result = runner.invoke( |
| 133 | cli, ["content-grep", "verse", "--ignore-case"], env=_env(tmp_path) |
| 134 | ) |
| 135 | assert result.exit_code == 0 |
| 136 | assert "notes.txt" in result.output |
| 137 | |
| 138 | |
| 139 | def test_content_grep_case_sensitive_no_match(tmp_path: pathlib.Path) -> None: |
| 140 | _init_repo(tmp_path) |
| 141 | _commit_files(tmp_path, {"notes.txt": b"VERSE: intro melody\n"}) |
| 142 | result = runner.invoke( |
| 143 | cli, ["content-grep", "verse"], env=_env(tmp_path) |
| 144 | ) |
| 145 | # Case-sensitive: "verse" ≠ "VERSE" → no match. |
| 146 | assert result.exit_code != 0 |
| 147 | |
| 148 | |
| 149 | # --------------------------------------------------------------------------- |
| 150 | # Unit: --files-only |
| 151 | # --------------------------------------------------------------------------- |
| 152 | |
| 153 | |
| 154 | def test_content_grep_files_only(tmp_path: pathlib.Path) -> None: |
| 155 | _init_repo(tmp_path) |
| 156 | _commit_files(tmp_path, { |
| 157 | "a.txt": b"match here\n", |
| 158 | "b.txt": b"match here too\n", |
| 159 | }) |
| 160 | result = runner.invoke( |
| 161 | cli, ["content-grep", "match", "--files-only"], env=_env(tmp_path) |
| 162 | ) |
| 163 | assert result.exit_code == 0 |
| 164 | lines = [l.strip() for l in result.output.strip().split("\n") if l.strip()] |
| 165 | for line in lines: |
| 166 | assert ":" not in line or line.startswith("a.txt") or line.startswith("b.txt") |
| 167 | |
| 168 | |
| 169 | # --------------------------------------------------------------------------- |
| 170 | # Unit: --count |
| 171 | # --------------------------------------------------------------------------- |
| 172 | |
| 173 | |
| 174 | def test_content_grep_count(tmp_path: pathlib.Path) -> None: |
| 175 | _init_repo(tmp_path) |
| 176 | _commit_files(tmp_path, {"multi.txt": b"hit\nhit\nhit\nmiss\n"}) |
| 177 | result = runner.invoke( |
| 178 | cli, ["content-grep", "hit", "--count"], env=_env(tmp_path) |
| 179 | ) |
| 180 | assert result.exit_code == 0 |
| 181 | assert "3" in result.output |
| 182 | |
| 183 | |
| 184 | # --------------------------------------------------------------------------- |
| 185 | # Unit: --format json |
| 186 | # --------------------------------------------------------------------------- |
| 187 | |
| 188 | |
| 189 | def test_content_grep_json_output(tmp_path: pathlib.Path) -> None: |
| 190 | _init_repo(tmp_path) |
| 191 | _commit_files(tmp_path, {"song.midi.txt": b"note: C4\nnote: D4\n"}) |
| 192 | result = runner.invoke( |
| 193 | cli, ["content-grep", "note", "--json"], env=_env(tmp_path) |
| 194 | ) |
| 195 | assert result.exit_code == 0 |
| 196 | data = json.loads(result.output) |
| 197 | assert isinstance(data, dict) |
| 198 | assert len(data["results"]) >= 1 |
| 199 | assert data["results"][0]["match_count"] >= 2 |
| 200 | |
| 201 | |
| 202 | # --------------------------------------------------------------------------- |
| 203 | # Unit: binary file skipped silently |
| 204 | # --------------------------------------------------------------------------- |
| 205 | |
| 206 | |
| 207 | def test_content_grep_binary_skipped(tmp_path: pathlib.Path) -> None: |
| 208 | _init_repo(tmp_path) |
| 209 | binary_content = b"\x00\x01\x02\x03" * 100 |
| 210 | text_content = b"searchable text here\n" |
| 211 | _commit_files(tmp_path, { |
| 212 | "binary.bin": binary_content, |
| 213 | "text.txt": text_content, |
| 214 | }) |
| 215 | result = runner.invoke( |
| 216 | cli, ["content-grep", "searchable"], env=_env(tmp_path) |
| 217 | ) |
| 218 | assert result.exit_code == 0 |
| 219 | assert "text.txt" in result.output |
| 220 | |
| 221 | |
| 222 | # --------------------------------------------------------------------------- |
| 223 | # Unit: short flags work |
| 224 | # --------------------------------------------------------------------------- |
| 225 | |
| 226 | |
| 227 | def test_content_grep_short_flags(tmp_path: pathlib.Path) -> None: |
| 228 | _init_repo(tmp_path) |
| 229 | _commit_files(tmp_path, {"f.txt": b"hello world\n"}) |
| 230 | result = runner.invoke( |
| 231 | cli, ["content-grep", "hello", "-i", "--json"], env=_env(tmp_path) |
| 232 | ) |
| 233 | assert result.exit_code == 0 |
| 234 | data = json.loads(result.output) |
| 235 | assert len(data["results"]) >= 1 |
| 236 | |
| 237 | |
| 238 | # --------------------------------------------------------------------------- |
| 239 | # Stress: 100 files, pattern matches 50 |
| 240 | # --------------------------------------------------------------------------- |
| 241 | |
| 242 | |
| 243 | def test_content_grep_stress_100_files(tmp_path: pathlib.Path) -> None: |
| 244 | _init_repo(tmp_path) |
| 245 | files: _FileStore = {} |
| 246 | for i in range(100): |
| 247 | content = b"TARGET_LINE\n" if i % 2 == 0 else b"other content\n" |
| 248 | files[f"file_{i:04d}.txt"] = content |
| 249 | _commit_files(tmp_path, files) |
| 250 | result = runner.invoke( |
| 251 | cli, ["content-grep", "TARGET_LINE", "--json"], env=_env(tmp_path) |
| 252 | ) |
| 253 | assert result.exit_code == 0 |
| 254 | data = json.loads(result.output) |
| 255 | assert len(data["results"]) == 50 |
| 256 | |
| 257 | |
| 258 | # --------------------------------------------------------------------------- |
| 259 | # Working-tree mode: --working-tree searches disk, not the committed snapshot |
| 260 | # --------------------------------------------------------------------------- |
| 261 | |
| 262 | |
| 263 | def test_content_grep_working_tree_finds_uncommitted_edit(tmp_path: pathlib.Path) -> None: |
| 264 | """--working-tree finds content written to disk that is not yet committed.""" |
| 265 | _init_repo(tmp_path) |
| 266 | # Commit a file with one pattern. |
| 267 | _commit_files(tmp_path, {"song.txt": b"chord: Am\n"}) |
| 268 | # Write an uncommitted edit with a different pattern. |
| 269 | (tmp_path / "song.txt").write_bytes(b"chord: WORKING_TREE_ONLY\n") |
| 270 | |
| 271 | # Without --working-tree, finds the committed content. |
| 272 | result_committed = runner.invoke( |
| 273 | cli, ["content-grep", "Am"], env=_env(tmp_path) |
| 274 | ) |
| 275 | assert result_committed.exit_code == 0 |
| 276 | |
| 277 | # With --working-tree, finds the disk content. |
| 278 | result_wt = runner.invoke( |
| 279 | cli, ["content-grep", "WORKING_TREE_ONLY", "--working-tree"], |
| 280 | env=_env(tmp_path), |
| 281 | ) |
| 282 | assert result_wt.exit_code == 0 |
| 283 | assert "song.txt" in result_wt.output |
| 284 | |
| 285 | |
| 286 | def test_content_grep_working_tree_no_match(tmp_path: pathlib.Path) -> None: |
| 287 | """--working-tree returns exit 1 when pattern absent; --json still emits valid JSON.""" |
| 288 | _init_repo(tmp_path) |
| 289 | (tmp_path / "notes.txt").write_bytes(b"hello world\n") |
| 290 | result = runner.invoke( |
| 291 | cli, ["content-grep", "ZZZNOMATCH", "--working-tree", "--json"], |
| 292 | env=_env(tmp_path), |
| 293 | ) |
| 294 | assert result.exit_code != 0 |
| 295 | data = json.loads(result.output) |
| 296 | assert data["total_matches"] == 0 |
| 297 | assert data["results"] == [] |
| 298 | |
| 299 | |
| 300 | def test_content_grep_working_tree_skips_muse_dir(tmp_path: pathlib.Path) -> None: |
| 301 | """--working-tree never searches inside the .muse object store.""" |
| 302 | _init_repo(tmp_path) |
| 303 | # Write a matching string inside .muse/ — must NOT be found. |
| 304 | (muse_dir(tmp_path) / "stray.txt").write_bytes(b"SECRET_IN_MUSE\n") |
| 305 | # Write the same string outside .muse/ — must be found. |
| 306 | (tmp_path / "real.txt").write_bytes(b"SECRET_IN_MUSE\n") |
| 307 | |
| 308 | result = runner.invoke( |
| 309 | cli, ["content-grep", "SECRET_IN_MUSE", "--working-tree", "--json"], |
| 310 | env=_env(tmp_path), |
| 311 | ) |
| 312 | assert result.exit_code == 0 |
| 313 | data = json.loads(result.output) |
| 314 | paths = [r["path"] for r in data["results"]] |
| 315 | assert "real.txt" in paths |
| 316 | assert not any(".muse" in p for p in paths) |
| 317 | |
| 318 | |
| 319 | def test_content_grep_working_tree_json_schema(tmp_path: pathlib.Path) -> None: |
| 320 | """--working-tree JSON output has source=working-tree and null commit_id/snapshot_id.""" |
| 321 | _init_repo(tmp_path) |
| 322 | (tmp_path / "f.txt").write_bytes(b"TARGET\n") |
| 323 | result = runner.invoke( |
| 324 | cli, ["content-grep", "TARGET", "--working-tree", "--json"], |
| 325 | env=_env(tmp_path), |
| 326 | ) |
| 327 | assert result.exit_code == 0 |
| 328 | data = json.loads(result.output) |
| 329 | assert data["source"] == "working-tree" |
| 330 | assert data["commit_id"] is None |
| 331 | assert data["snapshot_id"] is None |
| 332 | assert data["results"][0]["object_id"] is None |
| 333 | |
| 334 | |
| 335 | def test_content_grep_working_tree_files_only(tmp_path: pathlib.Path) -> None: |
| 336 | """--working-tree --files-only prints only file paths, no line numbers.""" |
| 337 | _init_repo(tmp_path) |
| 338 | (tmp_path / "a.txt").write_bytes(b"match\n") |
| 339 | (tmp_path / "b.txt").write_bytes(b"match\n") |
| 340 | result = runner.invoke( |
| 341 | cli, ["content-grep", "match", "--working-tree", "--files-only"], |
| 342 | env=_env(tmp_path), |
| 343 | ) |
| 344 | assert result.exit_code == 0 |
| 345 | lines = [l.strip() for l in result.output.strip().splitlines() if l.strip()] |
| 346 | assert all(":" not in l for l in lines) |
| 347 | assert {"a.txt", "b.txt"}.issubset(set(lines)) |
| 348 | |
| 349 | |
| 350 | def test_content_grep_working_tree_and_ref_mutually_exclusive(tmp_path: pathlib.Path) -> None: |
| 351 | """Passing both --working-tree and --ref is a user error (exit non-zero).""" |
| 352 | _init_repo(tmp_path) |
| 353 | _commit_files(tmp_path, {"f.txt": b"content\n"}) |
| 354 | result = runner.invoke( |
| 355 | cli, |
| 356 | ["content-grep", "content", "--working-tree", "--ref", "main"], |
| 357 | env=_env(tmp_path), |
| 358 | ) |
| 359 | assert result.exit_code != 0 |
| 360 | |
| 361 | |
| 362 | def test_content_grep_snapshot_json_has_source_commit(tmp_path: pathlib.Path) -> None: |
| 363 | """Snapshot mode JSON output has source=commit and non-null commit_id/snapshot_id.""" |
| 364 | _init_repo(tmp_path) |
| 365 | _commit_files(tmp_path, {"f.txt": b"TARGET\n"}) |
| 366 | result = runner.invoke( |
| 367 | cli, ["content-grep", "TARGET", "--json"], env=_env(tmp_path) |
| 368 | ) |
| 369 | assert result.exit_code == 0 |
| 370 | data = json.loads(result.output) |
| 371 | assert data["source"] == "commit" |
| 372 | assert data["commit_id"] is not None |
| 373 | assert data["snapshot_id"] is not None |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
128 days ago