test_read_commit_supercharge.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Supercharge tests for ``muse read-commit``. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | - Unit: _short_id helper — prefix preservation, hex length |
| 6 | - Integration: duration_ms + exit_code in JSON; text short-ID format |
| 7 | - Data integrity: sha256: prefix on all ID fields; valid JSON output |
| 8 | - Edge cases: --fields empty/duplicate; HEAD~N beyond depth; unknown branch |
| 9 | - Merge: parent2_commit_id in output |
| 10 | - Performance: single read under threshold |
| 11 | """ |
| 12 | from __future__ import annotations |
| 13 | |
| 14 | import datetime |
| 15 | import json |
| 16 | import pathlib |
| 17 | import re |
| 18 | import time |
| 19 | |
| 20 | import pytest |
| 21 | |
| 22 | from muse.core.errors import ExitCode |
| 23 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 24 | from muse.core.commits import ( |
| 25 | CommitRecord, |
| 26 | write_commit, |
| 27 | ) |
| 28 | from muse.core.snapshots import ( |
| 29 | SnapshotRecord, |
| 30 | write_snapshot, |
| 31 | ) |
| 32 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 33 | from muse.core.types import fake_id, long_id, split_id |
| 34 | from muse.core.paths import heads_dir, muse_dir |
| 35 | |
| 36 | runner = CliRunner() |
| 37 | |
| 38 | _SNAP_ID: str = compute_snapshot_id({}) |
| 39 | _COMMITTED_AT: datetime.datetime = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 40 | |
| 41 | _SHA256_FULL = re.compile(r"^sha256:[0-9a-f]{64}$") |
| 42 | _SHA256_SHORT_19 = re.compile(r"^sha256:[0-9a-f]{12}$") # "sha256:" (7) + 12 hex = 19 chars |
| 43 | |
| 44 | |
| 45 | # --------------------------------------------------------------------------- |
| 46 | # Helpers |
| 47 | # --------------------------------------------------------------------------- |
| 48 | |
| 49 | |
| 50 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 51 | repo = tmp_path / "repo" |
| 52 | dot_muse = muse_dir(repo) |
| 53 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 54 | (dot_muse / sub).mkdir(parents=True) |
| 55 | (dot_muse / "HEAD").write_text("ref: refs/heads/main") |
| 56 | (dot_muse / "repo.json").write_text(json.dumps({"repo_id": "test-repo", "domain": "code"})) |
| 57 | return repo |
| 58 | |
| 59 | |
| 60 | def _commit( |
| 61 | repo: pathlib.Path, |
| 62 | *, |
| 63 | branch: str = "main", |
| 64 | message: str = "test commit", |
| 65 | author: str = "tester", |
| 66 | parent: str | None = None, |
| 67 | parent2: str | None = None, |
| 68 | agent_id: str = "", |
| 69 | model_id: str = "", |
| 70 | snap_id: str | None = None, |
| 71 | committed_at: datetime.datetime | None = None, |
| 72 | ) -> str: |
| 73 | """Write a commit with a real content-addressed ID; return the commit_id.""" |
| 74 | sid = snap_id or _SNAP_ID |
| 75 | ts = committed_at or _COMMITTED_AT |
| 76 | parent_ids: list[str] = [p for p in (parent, parent2) if p] |
| 77 | commit_id = compute_commit_id( |
| 78 | parent_ids=parent_ids, |
| 79 | snapshot_id=sid, |
| 80 | message=message, |
| 81 | committed_at_iso=ts.isoformat(), |
| 82 | author=author, |
| 83 | ) |
| 84 | write_snapshot(repo, SnapshotRecord( |
| 85 | snapshot_id=sid, |
| 86 | manifest={}, |
| 87 | created_at=ts, |
| 88 | )) |
| 89 | rec = CommitRecord( |
| 90 | commit_id=commit_id, |
| 91 | branch=branch, |
| 92 | snapshot_id=sid, |
| 93 | message=message, |
| 94 | committed_at=ts, |
| 95 | author=author, |
| 96 | parent_commit_id=parent, |
| 97 | parent2_commit_id=parent2, |
| 98 | agent_id=agent_id, |
| 99 | model_id=model_id, |
| 100 | ) |
| 101 | write_commit(repo, rec) |
| 102 | return commit_id |
| 103 | |
| 104 | |
| 105 | def _rc(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 106 | from muse.cli.app import main as cli |
| 107 | return runner.invoke( |
| 108 | cli, |
| 109 | ["read-commit", *args], |
| 110 | env={"MUSE_REPO_ROOT": str(repo)}, |
| 111 | ) |
| 112 | |
| 113 | |
| 114 | def _rcj(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 115 | """Like _rc but always passes --json.""" |
| 116 | return _rc(repo, "--json", *args) |
| 117 | |
| 118 | |
| 119 | # --------------------------------------------------------------------------- |
| 120 | # Unit — _short_id |
| 121 | # --------------------------------------------------------------------------- |
| 122 | |
| 123 | |
| 124 | class TestCommitIdFormat: |
| 125 | """Text output must emit the full sha256:<64-hex> commit ID.""" |
| 126 | |
| 127 | def test_commit_id_keeps_sha256_prefix(self) -> None: |
| 128 | cid = long_id("a" * 64) |
| 129 | assert cid.startswith("sha256:") |
| 130 | |
| 131 | def test_commit_id_total_length_is_71(self) -> None: |
| 132 | cid = long_id("c0ffee" * 11) |
| 133 | assert len(cid[:71]) == 71 # sha256: (7) + 64 hex |
| 134 | |
| 135 | def test_full_id_matches_regex(self) -> None: |
| 136 | cid = long_id("abcdef01" * 8) |
| 137 | assert _SHA256_FULL.match(cid) |
| 138 | |
| 139 | |
| 140 | # --------------------------------------------------------------------------- |
| 141 | # Integration — duration_ms and exit_code in JSON output |
| 142 | # --------------------------------------------------------------------------- |
| 143 | |
| 144 | |
| 145 | class TestDurationAndExitCode: |
| 146 | def test_duration_ms_present_on_success(self, tmp_path: pathlib.Path) -> None: |
| 147 | repo = _make_repo(tmp_path) |
| 148 | cid = _commit(repo, message="timing test") |
| 149 | data = json.loads(_rcj(repo, cid).output) |
| 150 | assert "duration_ms" in data, "duration_ms must be present in JSON success output" |
| 151 | |
| 152 | def test_exit_code_zero_on_success(self, tmp_path: pathlib.Path) -> None: |
| 153 | repo = _make_repo(tmp_path) |
| 154 | cid = _commit(repo, message="exit code test") |
| 155 | data = json.loads(_rcj(repo, cid).output) |
| 156 | assert data["exit_code"] == 0 |
| 157 | |
| 158 | def test_duration_ms_is_float(self, tmp_path: pathlib.Path) -> None: |
| 159 | repo = _make_repo(tmp_path) |
| 160 | cid = _commit(repo, message="float timing") |
| 161 | data = json.loads(_rcj(repo, cid).output) |
| 162 | assert isinstance(data["duration_ms"], float) |
| 163 | |
| 164 | def test_duration_ms_non_negative(self, tmp_path: pathlib.Path) -> None: |
| 165 | repo = _make_repo(tmp_path) |
| 166 | cid = _commit(repo, message="positive timing") |
| 167 | data = json.loads(_rcj(repo, cid).output) |
| 168 | assert data["duration_ms"] >= 0.0 |
| 169 | |
| 170 | def test_fields_filter_preserves_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 171 | """duration_ms is command metadata, not a commit field — --fields must not drop it.""" |
| 172 | repo = _make_repo(tmp_path) |
| 173 | cid = _commit(repo, message="fields + duration") |
| 174 | data = json.loads(_rcj(repo, "--fields", "commit_id,message", cid).output) |
| 175 | assert "duration_ms" in data, "--fields must not filter out duration_ms" |
| 176 | |
| 177 | def test_fields_filter_preserves_exit_code(self, tmp_path: pathlib.Path) -> None: |
| 178 | """exit_code is command metadata — --fields must not drop it.""" |
| 179 | repo = _make_repo(tmp_path) |
| 180 | cid = _commit(repo, message="fields + exit_code") |
| 181 | data = json.loads(_rcj(repo, "--fields", "commit_id", cid).output) |
| 182 | assert "exit_code" in data, "--fields must not filter out exit_code" |
| 183 | |
| 184 | def test_duration_ms_3dp_precision(self, tmp_path: pathlib.Path) -> None: |
| 185 | """duration_ms must be rounded to 3 decimal places (millisecond precision).""" |
| 186 | repo = _make_repo(tmp_path) |
| 187 | cid = _commit(repo, message="precision test") |
| 188 | data = json.loads(_rcj(repo, cid).output) |
| 189 | ms = data["duration_ms"] |
| 190 | # round-trips through json.dumps — check at most 3 decimal places |
| 191 | assert round(ms, 3) == ms |
| 192 | |
| 193 | |
| 194 | # --------------------------------------------------------------------------- |
| 195 | # Integration — text format short ID |
| 196 | # --------------------------------------------------------------------------- |
| 197 | |
| 198 | |
| 199 | class TestTextFormatFullId: |
| 200 | """Text format must emit the full sha256:<64-hex> (71 chars) commit ID.""" |
| 201 | |
| 202 | def _full_token(self, line: str) -> str | None: |
| 203 | for tok in line.split(): |
| 204 | if _SHA256_FULL.match(tok): |
| 205 | return tok |
| 206 | return None |
| 207 | |
| 208 | def test_text_full_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 209 | repo = _make_repo(tmp_path) |
| 210 | cid = _commit(repo, message="full id prefix test") |
| 211 | result = _rc(repo, cid) |
| 212 | assert result.exit_code == 0 |
| 213 | line = result.output.strip() |
| 214 | tok = self._full_token(line) |
| 215 | assert tok is not None, f"no sha256:<64-hex> token in text output: {line!r}" |
| 216 | assert tok.startswith("sha256:") |
| 217 | |
| 218 | def test_text_full_id_has_64_hex_chars(self, tmp_path: pathlib.Path) -> None: |
| 219 | repo = _make_repo(tmp_path) |
| 220 | cid = _commit(repo, message="full id hex length test") |
| 221 | result = _rc(repo, cid) |
| 222 | line = result.output.strip() |
| 223 | tok = self._full_token(line) |
| 224 | assert tok is not None, f"no sha256:<64-hex> token in text output: {line!r}" |
| 225 | hex_part = tok[len("sha256:"):] |
| 226 | assert len(hex_part) == 64, f"expected 64 hex chars after prefix, got {len(hex_part)}: {tok!r}" |
| 227 | |
| 228 | def test_text_full_id_total_length_is_71(self, tmp_path: pathlib.Path) -> None: |
| 229 | repo = _make_repo(tmp_path) |
| 230 | cid = _commit(repo, message="full id length test") |
| 231 | result = _rc(repo, cid) |
| 232 | line = result.output.strip() |
| 233 | tok = self._full_token(line) |
| 234 | assert tok is not None |
| 235 | assert len(tok) == 71, f"commit ID must be exactly 71 chars, got {len(tok)}: {tok!r}" |
| 236 | |
| 237 | def test_text_full_id_matches_commit_id(self, tmp_path: pathlib.Path) -> None: |
| 238 | repo = _make_repo(tmp_path) |
| 239 | cid = _commit(repo, message="full id matches test") |
| 240 | result = _rc(repo, cid) |
| 241 | line = result.output.strip() |
| 242 | tok = self._full_token(line) |
| 243 | assert tok is not None |
| 244 | assert tok == cid, f"text output ID {tok!r} does not match commit_id {cid!r}" |
| 245 | |
| 246 | |
| 247 | # --------------------------------------------------------------------------- |
| 248 | # Data integrity |
| 249 | # --------------------------------------------------------------------------- |
| 250 | |
| 251 | |
| 252 | class TestDataIntegrity: |
| 253 | def test_commit_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 254 | repo = _make_repo(tmp_path) |
| 255 | cid = _commit(repo, message="id prefix test") |
| 256 | data = json.loads(_rcj(repo, cid).output) |
| 257 | assert _SHA256_FULL.match(data["commit_id"]), \ |
| 258 | f"commit_id must be sha256:<64hex>, got {data['commit_id']!r}" |
| 259 | |
| 260 | def test_snapshot_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 261 | repo = _make_repo(tmp_path) |
| 262 | cid = _commit(repo, message="snapshot id test") |
| 263 | data = json.loads(_rcj(repo, cid).output) |
| 264 | assert _SHA256_FULL.match(data["snapshot_id"]), \ |
| 265 | f"snapshot_id must be sha256:<64hex>, got {data['snapshot_id']!r}" |
| 266 | |
| 267 | def test_parent_commit_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 268 | repo = _make_repo(tmp_path) |
| 269 | parent = _commit(repo, message="parent") |
| 270 | child = _commit(repo, message="child", parent=parent) |
| 271 | data = json.loads(_rcj(repo, child).output) |
| 272 | assert _SHA256_FULL.match(data["parent_commit_id"]), \ |
| 273 | f"parent_commit_id must be sha256:<64hex>, got {data['parent_commit_id']!r}" |
| 274 | |
| 275 | def test_json_output_is_valid_json(self, tmp_path: pathlib.Path) -> None: |
| 276 | repo = _make_repo(tmp_path) |
| 277 | cid = _commit(repo, message="valid json test") |
| 278 | result = _rcj(repo, cid) |
| 279 | assert result.exit_code == 0 |
| 280 | # Must not raise |
| 281 | data = json.loads(result.output) |
| 282 | assert isinstance(data, dict) |
| 283 | |
| 284 | def test_message_with_special_chars_in_json(self, tmp_path: pathlib.Path) -> None: |
| 285 | """Control chars and quotes in message must not break JSON output.""" |
| 286 | repo = _make_repo(tmp_path) |
| 287 | # tab, backslash, double-quote — all must be escaped in JSON |
| 288 | msg = 'feat: say "hello"\twith backslash \\' |
| 289 | cid = _commit(repo, message=msg) |
| 290 | result = _rcj(repo, cid) |
| 291 | assert result.exit_code == 0 |
| 292 | data = json.loads(result.output) |
| 293 | assert data["message"] == msg |
| 294 | |
| 295 | def test_message_with_unicode_in_json(self, tmp_path: pathlib.Path) -> None: |
| 296 | repo = _make_repo(tmp_path) |
| 297 | msg = "feat: 音楽 🎵 café naïve" |
| 298 | cid = _commit(repo, message=msg) |
| 299 | result = _rcj(repo, cid) |
| 300 | assert result.exit_code == 0 |
| 301 | data = json.loads(result.output) |
| 302 | assert data["message"] == msg |
| 303 | |
| 304 | |
| 305 | # --------------------------------------------------------------------------- |
| 306 | # Edge cases — --fields |
| 307 | # --------------------------------------------------------------------------- |
| 308 | |
| 309 | |
| 310 | class TestFieldsEdgeCases: |
| 311 | def test_fields_empty_string_errors(self, tmp_path: pathlib.Path) -> None: |
| 312 | """--fields '' with no real field names should error (empty requested set).""" |
| 313 | repo = _make_repo(tmp_path) |
| 314 | cid = _commit(repo, message="empty fields test") |
| 315 | result = _rc(repo, "--fields", "", cid) |
| 316 | # Empty --fields is ambiguous — should either error or return only metadata. |
| 317 | # At minimum the output must be valid JSON. |
| 318 | assert result.exit_code == 0 or result.exit_code == ExitCode.USER_ERROR |
| 319 | |
| 320 | def test_fields_duplicate_deduplicated(self, tmp_path: pathlib.Path) -> None: |
| 321 | """Duplicate field names in --fields must not crash and produce one key.""" |
| 322 | repo = _make_repo(tmp_path) |
| 323 | cid = _commit(repo, message="duplicate fields test") |
| 324 | result = _rcj(repo, "--fields", "commit_id,commit_id,message", cid) |
| 325 | assert result.exit_code == 0 |
| 326 | data = json.loads(result.output) |
| 327 | # Only one commit_id key, one message key |
| 328 | assert "commit_id" in data |
| 329 | assert "message" in data |
| 330 | |
| 331 | def test_fields_whitespace_only_errors(self, tmp_path: pathlib.Path) -> None: |
| 332 | """--fields ' , ' (only whitespace/commas) should error.""" |
| 333 | repo = _make_repo(tmp_path) |
| 334 | cid = _commit(repo, message="whitespace fields test") |
| 335 | result = _rc(repo, "--fields", " , ", cid) |
| 336 | # Parts after strip are empty — should error |
| 337 | assert result.exit_code == 0 or result.exit_code == ExitCode.USER_ERROR |
| 338 | |
| 339 | |
| 340 | # --------------------------------------------------------------------------- |
| 341 | # Edge cases — symbolic refs |
| 342 | # --------------------------------------------------------------------------- |
| 343 | |
| 344 | |
| 345 | class TestSymbolicRefEdgeCases: |
| 346 | def test_head_tilde_exceeds_chain_depth_errors(self, tmp_path: pathlib.Path) -> None: |
| 347 | """HEAD~99 on a 1-commit repo must exit with USER_ERROR, not crash.""" |
| 348 | repo = _make_repo(tmp_path) |
| 349 | cid = _commit(repo, branch="main", message="only commit") |
| 350 | (heads_dir(repo) / "main").write_text(cid) |
| 351 | result = _rc(repo, "HEAD~99") |
| 352 | assert result.exit_code == ExitCode.USER_ERROR |
| 353 | assert "Traceback" not in result.output |
| 354 | |
| 355 | def test_unknown_branch_name_errors(self, tmp_path: pathlib.Path) -> None: |
| 356 | """A branch name that doesn't exist must exit USER_ERROR cleanly.""" |
| 357 | repo = _make_repo(tmp_path) |
| 358 | _commit(repo, message="root") |
| 359 | result = _rc(repo, "nonexistent-branch-xyz") |
| 360 | assert result.exit_code == ExitCode.USER_ERROR |
| 361 | assert "Traceback" not in result.output |
| 362 | |
| 363 | |
| 364 | # --------------------------------------------------------------------------- |
| 365 | # Merge commit |
| 366 | # --------------------------------------------------------------------------- |
| 367 | |
| 368 | |
| 369 | class TestMergeCommit: |
| 370 | def test_parent2_commit_id_in_json_output(self, tmp_path: pathlib.Path) -> None: |
| 371 | """Merge commits must expose parent2_commit_id in JSON output.""" |
| 372 | repo = _make_repo(tmp_path) |
| 373 | p1 = _commit(repo, message="parent one") |
| 374 | p2 = _commit(repo, message="parent two", committed_at=datetime.datetime(2026, 1, 2, tzinfo=datetime.timezone.utc)) |
| 375 | merge = _commit(repo, message="merge commit", parent=p1, parent2=p2, |
| 376 | committed_at=datetime.datetime(2026, 1, 3, tzinfo=datetime.timezone.utc)) |
| 377 | data = json.loads(_rcj(repo, merge).output) |
| 378 | assert data["parent_commit_id"] == p1 |
| 379 | assert data["parent2_commit_id"] == p2 |
| 380 | |
| 381 | def test_parent2_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 382 | repo = _make_repo(tmp_path) |
| 383 | p1 = _commit(repo, message="p1") |
| 384 | p2 = _commit(repo, message="p2", committed_at=datetime.datetime(2026, 1, 2, tzinfo=datetime.timezone.utc)) |
| 385 | merge = _commit(repo, message="merge", parent=p1, parent2=p2, |
| 386 | committed_at=datetime.datetime(2026, 1, 3, tzinfo=datetime.timezone.utc)) |
| 387 | data = json.loads(_rcj(repo, merge).output) |
| 388 | assert _SHA256_FULL.match(data["parent2_commit_id"]), \ |
| 389 | f"parent2_commit_id must be sha256:<64hex>, got {data['parent2_commit_id']!r}" |
| 390 | |
| 391 | |
| 392 | # --------------------------------------------------------------------------- |
| 393 | # Performance |
| 394 | # --------------------------------------------------------------------------- |
| 395 | |
| 396 | |
| 397 | class TestPerformance: |
| 398 | def test_single_read_under_500ms(self, tmp_path: pathlib.Path) -> None: |
| 399 | """A single read-commit invocation must complete in under 500ms.""" |
| 400 | repo = _make_repo(tmp_path) |
| 401 | cid = _commit(repo, message="perf test") |
| 402 | t0 = time.monotonic() |
| 403 | result = _rc(repo, cid) |
| 404 | duration_ms = (time.monotonic() - t0) * 1000 |
| 405 | assert result.exit_code == 0 |
| 406 | assert duration_ms < 500, f"read-commit took {duration_ms:.1f}ms — over 500ms threshold" |
| 407 | |
| 408 | def test_duration_ms_in_output_plausible(self, tmp_path: pathlib.Path) -> None: |
| 409 | """duration_ms in the JSON output must be less than 500ms for a warm read.""" |
| 410 | repo = _make_repo(tmp_path) |
| 411 | cid = _commit(repo, message="plausible timing") |
| 412 | data = json.loads(_rcj(repo, cid).output) |
| 413 | assert data["duration_ms"] < 500, \ |
| 414 | f"duration_ms={data['duration_ms']} — suspiciously slow or not measuring correctly" |
| 415 | |
| 416 | |
| 417 | class TestRegisterFlags: |
| 418 | def _parse(self, *args: str) -> "argparse.Namespace": |
| 419 | import argparse |
| 420 | from muse.cli.commands.read_commit import register |
| 421 | p = argparse.ArgumentParser() |
| 422 | subs = p.add_subparsers() |
| 423 | register(subs) |
| 424 | return p.parse_args(["read-commit", fake_id("a"), *args]) |
| 425 | |
| 426 | def test_json_short_flag(self) -> None: |
| 427 | args = self._parse("-j") |
| 428 | assert args.json_out is True |
| 429 | |
| 430 | def test_json_long_flag(self) -> None: |
| 431 | args = self._parse("--json") |
| 432 | assert args.json_out is True |
| 433 | |
| 434 | def test_default_no_json(self) -> None: |
| 435 | args = self._parse() |
| 436 | assert args.json_out is False |