test_cmd_describe_hardening.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Hardening test suite for ``muse describe``. |
| 2 | |
| 3 | Coverage: |
| 4 | - Unit: describe_commit core — no tags, exact, distance, long, match_pattern, |
| 5 | first_parent, abbrev, exact_match, _MAX_WALK budget, multi-tag tie-break |
| 6 | - Security: ANSI injection in tag names sanitized in text output, |
| 7 | raw in JSON; --ref ANSI passthrough in error message |
| 8 | - Error routing: all user errors routed to stderr |
| 9 | - JSON schema: _DescribeJson shape, all fields present, repo_id + branch |
| 10 | - New flags: --match, --exact-match, --first-parent, --abbrev, --json |
| 11 | - Integration: tag walk across merge commits, --first-parent vs full walk |
| 12 | - E2E: help output, combined flags |
| 13 | - Stress: 5 000-commit chain, 200-tag repo, concurrent reads |
| 14 | """ |
| 15 | |
| 16 | from __future__ import annotations |
| 17 | |
| 18 | import datetime |
| 19 | import json |
| 20 | import pathlib |
| 21 | import threading |
| 22 | from typing import TypedDict |
| 23 | from unittest.mock import patch |
| 24 | |
| 25 | import pytest |
| 26 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 27 | |
| 28 | from muse.core.describe import describe_commit, _MAX_WALK |
| 29 | from muse.core.object_store import write_object |
| 30 | from muse.core.ids import hash_commit, hash_snapshot |
| 31 | from muse.core.commits import ( |
| 32 | CommitRecord, |
| 33 | write_commit, |
| 34 | ) |
| 35 | from muse.core.snapshots import ( |
| 36 | SnapshotRecord, |
| 37 | write_snapshot, |
| 38 | ) |
| 39 | from muse.core.tags import ( |
| 40 | TagRecord, |
| 41 | write_tag, |
| 42 | ) |
| 43 | from muse.core.types import Manifest, blob_id, fake_id, content_hash |
| 44 | from muse.core.paths import muse_dir, ref_path |
| 45 | |
| 46 | runner = CliRunner() |
| 47 | _REPO_ID = content_hash({"name": "describe-hard-test"}) |
| 48 | |
| 49 | |
| 50 | # --------------------------------------------------------------------------- |
| 51 | # Helpers |
| 52 | # --------------------------------------------------------------------------- |
| 53 | |
| 54 | |
| 55 | |
| 56 | |
| 57 | def _init_repo(path: pathlib.Path, *, domain: str = "midi") -> pathlib.Path: |
| 58 | dot_muse = muse_dir(path) |
| 59 | for sub in ("commits", "snapshots", "objects", "refs/heads", "tags"): |
| 60 | (dot_muse / sub).mkdir(parents=True, exist_ok=True) |
| 61 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 62 | (dot_muse / "repo.json").write_text( |
| 63 | json.dumps({"repo_id": _REPO_ID, "domain": domain}), |
| 64 | encoding="utf-8", |
| 65 | ) |
| 66 | return path |
| 67 | |
| 68 | |
| 69 | def _make_commit( |
| 70 | root: pathlib.Path, |
| 71 | parent_id: str | None = None, |
| 72 | parent2_id: str | None = None, |
| 73 | content: bytes = b"data", |
| 74 | branch: str = "main", |
| 75 | ) -> str: |
| 76 | obj_id = blob_id(content) |
| 77 | write_object(root, obj_id, content) |
| 78 | manifest = {f"f_{obj_id[len('sha256:'):len('sha256:') + 8]}.txt": obj_id} |
| 79 | snap_id = hash_snapshot(manifest) |
| 80 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 81 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 82 | parent_ids = [pid for pid in (parent_id, parent2_id) if pid is not None] |
| 83 | commit_id = hash_commit( |
| 84 | parent_ids=parent_ids, |
| 85 | snapshot_id=snap_id, |
| 86 | message="msg", |
| 87 | committed_at_iso=committed_at.isoformat(), |
| 88 | ) |
| 89 | rec = CommitRecord( |
| 90 | commit_id=commit_id, |
| 91 | branch=branch, |
| 92 | snapshot_id=snap_id, |
| 93 | message="msg", |
| 94 | committed_at=committed_at, |
| 95 | parent_commit_id=parent_id, |
| 96 | parent2_commit_id=parent2_id, |
| 97 | ) |
| 98 | write_commit(root, rec) |
| 99 | (ref_path(root, branch)).write_text( |
| 100 | commit_id, encoding="utf-8" |
| 101 | ) |
| 102 | return commit_id |
| 103 | |
| 104 | |
| 105 | def _make_tag(root: pathlib.Path, tag: str, commit_id: str) -> None: |
| 106 | write_tag( |
| 107 | root, |
| 108 | TagRecord( |
| 109 | tag_id=content_hash({"tag": tag, "commit_id": commit_id}), |
| 110 | repo_id=_REPO_ID, |
| 111 | tag=tag, |
| 112 | commit_id=commit_id, |
| 113 | created_at=datetime.datetime.now(datetime.timezone.utc), |
| 114 | ), |
| 115 | ) |
| 116 | |
| 117 | |
| 118 | def _env(repo: pathlib.Path) -> Manifest: |
| 119 | return {"MUSE_REPO_ROOT": str(repo)} |
| 120 | |
| 121 | |
| 122 | def _invoke(args: list[str], env: Manifest) -> InvokeResult: |
| 123 | return runner.invoke(None, args, env=env) |
| 124 | |
| 125 | |
| 126 | class _DescribeOut(TypedDict): |
| 127 | commit_id: str |
| 128 | tag: str | None |
| 129 | distance: int |
| 130 | short_sha: str |
| 131 | name: str |
| 132 | exact: bool |
| 133 | repo_id: str |
| 134 | branch: str |
| 135 | duration_ms: float |
| 136 | exit_code: int |
| 137 | |
| 138 | |
| 139 | def _parse_json(result: InvokeResult) -> _DescribeOut: |
| 140 | raw = json.loads(result.output.strip()) |
| 141 | return _DescribeOut( |
| 142 | commit_id=raw["commit_id"], |
| 143 | tag=raw["tag"], |
| 144 | distance=raw["distance"], |
| 145 | short_sha=raw["short_sha"], |
| 146 | name=raw["name"], |
| 147 | exact=raw["exact"], |
| 148 | repo_id=raw["repo_id"], |
| 149 | branch=raw["branch"], |
| 150 | duration_ms=raw["duration_ms"], |
| 151 | exit_code=raw["exit_code"], |
| 152 | ) |
| 153 | |
| 154 | |
| 155 | # --------------------------------------------------------------------------- |
| 156 | # Unit: describe_commit core |
| 157 | # --------------------------------------------------------------------------- |
| 158 | |
| 159 | |
| 160 | def test_core_no_tags_returns_shortblob_id(tmp_path: pathlib.Path) -> None: |
| 161 | _init_repo(tmp_path) |
| 162 | cid = _make_commit(tmp_path, content=b"a") |
| 163 | r = describe_commit(tmp_path, _REPO_ID, cid) |
| 164 | assert r["tag"] is None |
| 165 | assert r["short_sha"] == cid[:len("sha256:") + 12] |
| 166 | assert r["name"] == r["short_sha"] |
| 167 | assert r["exact"] is False |
| 168 | |
| 169 | |
| 170 | def test_core_exact_tag(tmp_path: pathlib.Path) -> None: |
| 171 | _init_repo(tmp_path) |
| 172 | cid = _make_commit(tmp_path, content=b"b") |
| 173 | _make_tag(tmp_path, "v1.0.0", cid) |
| 174 | r = describe_commit(tmp_path, _REPO_ID, cid) |
| 175 | assert r["tag"] == "v1.0.0" |
| 176 | assert r["distance"] == 0 |
| 177 | assert r["exact"] is True |
| 178 | assert r["name"] == "v1.0.0" |
| 179 | |
| 180 | |
| 181 | def test_core_distance_one(tmp_path: pathlib.Path) -> None: |
| 182 | _init_repo(tmp_path) |
| 183 | c1 = _make_commit(tmp_path, content=b"c1") |
| 184 | _make_tag(tmp_path, "v0.9", c1) |
| 185 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"c2") |
| 186 | r = describe_commit(tmp_path, _REPO_ID, c2) |
| 187 | assert r["tag"] == "v0.9" |
| 188 | assert r["distance"] == 1 |
| 189 | assert r["exact"] is False |
| 190 | assert r["name"] == "v0.9~1" |
| 191 | |
| 192 | |
| 193 | def test_core_long_format_on_tag(tmp_path: pathlib.Path) -> None: |
| 194 | _init_repo(tmp_path) |
| 195 | cid = _make_commit(tmp_path, content=b"long") |
| 196 | _make_tag(tmp_path, "v2.0.0", cid) |
| 197 | r = describe_commit(tmp_path, _REPO_ID, cid, long_format=True) |
| 198 | assert r["name"].startswith("v2.0.0-0-sha256:") |
| 199 | |
| 200 | |
| 201 | def test_core_long_format_with_distance(tmp_path: pathlib.Path) -> None: |
| 202 | _init_repo(tmp_path) |
| 203 | c1 = _make_commit(tmp_path, content=b"root") |
| 204 | _make_tag(tmp_path, "v1.0", c1) |
| 205 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"next") |
| 206 | r = describe_commit(tmp_path, _REPO_ID, c2, long_format=True) |
| 207 | assert "-1-sha256:" in r["name"] |
| 208 | |
| 209 | |
| 210 | def test_core_abbrev_controls_sha_length(tmp_path: pathlib.Path) -> None: |
| 211 | _init_repo(tmp_path) |
| 212 | cid = _make_commit(tmp_path, content=b"abbrev") |
| 213 | r = describe_commit(tmp_path, _REPO_ID, cid, abbrev=8) |
| 214 | assert r["short_sha"] == cid[:len("sha256:") + 8] |
| 215 | assert r["short_sha"].startswith("sha256:") |
| 216 | assert len(r["short_sha"]) == len("sha256:") + 8 |
| 217 | |
| 218 | |
| 219 | def test_core_match_pattern_filters_tags(tmp_path: pathlib.Path) -> None: |
| 220 | _init_repo(tmp_path) |
| 221 | cid = _make_commit(tmp_path, content=b"match") |
| 222 | _make_tag(tmp_path, "release-1", cid) |
| 223 | _make_tag(tmp_path, "v1.0.0", cid) |
| 224 | # Only semver tags — "release-1" excluded. |
| 225 | r = describe_commit(tmp_path, _REPO_ID, cid, match_pattern="v*") |
| 226 | assert r["tag"] == "v1.0.0" |
| 227 | |
| 228 | |
| 229 | def test_core_match_pattern_no_match_returnsblob_id(tmp_path: pathlib.Path) -> None: |
| 230 | _init_repo(tmp_path) |
| 231 | cid = _make_commit(tmp_path, content=b"nomatch") |
| 232 | _make_tag(tmp_path, "nightly-123", cid) |
| 233 | r = describe_commit(tmp_path, _REPO_ID, cid, match_pattern="v*") |
| 234 | assert r["tag"] is None |
| 235 | assert r["name"] == cid[:len("sha256:") + 12] |
| 236 | |
| 237 | |
| 238 | def test_core_exact_match_on_tag(tmp_path: pathlib.Path) -> None: |
| 239 | _init_repo(tmp_path) |
| 240 | cid = _make_commit(tmp_path, content=b"exact") |
| 241 | _make_tag(tmp_path, "v3.0", cid) |
| 242 | r = describe_commit(tmp_path, _REPO_ID, cid, exact_match=True) |
| 243 | assert r["tag"] == "v3.0" |
| 244 | assert r["exact"] is True |
| 245 | |
| 246 | |
| 247 | def test_core_exact_match_off_tag_returnsblob_id(tmp_path: pathlib.Path) -> None: |
| 248 | _init_repo(tmp_path) |
| 249 | c1 = _make_commit(tmp_path, content=b"root") |
| 250 | _make_tag(tmp_path, "v3.0", c1) |
| 251 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"after") |
| 252 | r = describe_commit(tmp_path, _REPO_ID, c2, exact_match=True) |
| 253 | assert r["tag"] is None |
| 254 | assert r["name"] == c2[:len("sha256:") + 12] |
| 255 | |
| 256 | |
| 257 | def test_core_first_parent_skips_merge_branch(tmp_path: pathlib.Path) -> None: |
| 258 | """With --first-parent the tag on a merged branch is invisible.""" |
| 259 | _init_repo(tmp_path) |
| 260 | # main: c1 → c3 (merge of feat) |
| 261 | # feat: c1 → c2 (tag here) |
| 262 | c1 = _make_commit(tmp_path, content=b"root") |
| 263 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"feat", branch="feat") |
| 264 | _make_tag(tmp_path, "feat-tag", c2) |
| 265 | c3 = _make_commit( |
| 266 | tmp_path, parent_id=c1, parent2_id=c2, content=b"merge", branch="main" |
| 267 | ) |
| 268 | # Without first_parent: feat-tag is reachable via second parent. |
| 269 | r_full = describe_commit(tmp_path, _REPO_ID, c3) |
| 270 | # With first_parent: only first parent chain; feat-tag not reachable. |
| 271 | r_fp = describe_commit(tmp_path, _REPO_ID, c3, first_parent=True) |
| 272 | assert r_fp["tag"] is None |
| 273 | # Full walk should find the tag (via c2). |
| 274 | assert r_full["tag"] == "feat-tag" |
| 275 | |
| 276 | |
| 277 | def test_core_multi_tag_same_commit_lex_greatest(tmp_path: pathlib.Path) -> None: |
| 278 | """When multiple tags point at the same commit, greatest lex name wins.""" |
| 279 | _init_repo(tmp_path) |
| 280 | cid = _make_commit(tmp_path, content=b"multi") |
| 281 | _make_tag(tmp_path, "v1.0.0", cid) |
| 282 | _make_tag(tmp_path, "v2.0.0", cid) |
| 283 | _make_tag(tmp_path, "v1.5.0", cid) |
| 284 | r = describe_commit(tmp_path, _REPO_ID, cid) |
| 285 | assert r["tag"] == "v2.0.0" |
| 286 | |
| 287 | |
| 288 | def test_core_max_walk_budget(tmp_path: pathlib.Path) -> None: |
| 289 | """Walk stops at _MAX_WALK without crashing; returns short-SHA fallback.""" |
| 290 | _init_repo(tmp_path) |
| 291 | # Inject a fake read_commit that always returns a parent so BFS never |
| 292 | # finds a commit-store miss — we just want to trigger the budget guard. |
| 293 | cid = _make_commit(tmp_path, content=b"budget") |
| 294 | _make_tag(tmp_path, "very-far", cid) |
| 295 | |
| 296 | call_count = 0 |
| 297 | |
| 298 | import muse.core.describe as _describe_mod |
| 299 | from muse.core.commits import read_commit as _orig_read_commit, CommitRecord as _CR |
| 300 | import datetime as _dt |
| 301 | |
| 302 | orig = _orig_read_commit |
| 303 | |
| 304 | def _fake_read(root: pathlib.Path, cid: str) -> _CR | None: |
| 305 | nonlocal call_count |
| 306 | call_count += 1 |
| 307 | if call_count > _MAX_WALK + 5: |
| 308 | return None |
| 309 | fake_parent = fake_id(cid) |
| 310 | return _CR( |
| 311 | commit_id=cid, |
| 312 | branch="main", |
| 313 | snapshot_id="snap", |
| 314 | message="x", |
| 315 | committed_at=_dt.datetime.now(_dt.timezone.utc), |
| 316 | parent_commit_id=fake_parent, |
| 317 | ) |
| 318 | |
| 319 | with patch.object(_describe_mod, "read_commit", side_effect=_fake_read): |
| 320 | # Start from a commit far from any tag — BFS will exhaust budget. |
| 321 | far_commit = blob_id(b"far") |
| 322 | r = describe_commit(tmp_path, _REPO_ID, far_commit) |
| 323 | |
| 324 | # Budget exhausted → tag not found → name is short SHA. |
| 325 | assert r["tag"] is None |
| 326 | |
| 327 | |
| 328 | # --------------------------------------------------------------------------- |
| 329 | # Security: ANSI injection in tag names |
| 330 | # --------------------------------------------------------------------------- |
| 331 | |
| 332 | |
| 333 | def test_ansi_in_tag_name_stripped_in_text_output(tmp_path: pathlib.Path) -> None: |
| 334 | _init_repo(tmp_path) |
| 335 | cid = _make_commit(tmp_path, content=b"ansi") |
| 336 | malicious_tag = "v1.0\x1b[31mRED\x1b[0m" |
| 337 | _make_tag(tmp_path, malicious_tag, cid) |
| 338 | result = _invoke(["describe"], _env(tmp_path)) |
| 339 | assert result.exit_code == 0 |
| 340 | assert "\x1b[31m" not in result.output |
| 341 | |
| 342 | |
| 343 | def test_ansi_in_tag_name_preserved_in_json(tmp_path: pathlib.Path) -> None: |
| 344 | """JSON output must not sanitize so callers see the raw value.""" |
| 345 | _init_repo(tmp_path) |
| 346 | cid = _make_commit(tmp_path, content=b"ansi-json") |
| 347 | malicious_tag = "v1.0\x1b[31mRED\x1b[0m" |
| 348 | _make_tag(tmp_path, malicious_tag, cid) |
| 349 | result = _invoke(["describe", "--json"], _env(tmp_path)) |
| 350 | assert result.exit_code == 0 |
| 351 | data = _parse_json(result) |
| 352 | assert data["tag"] == malicious_tag |
| 353 | |
| 354 | |
| 355 | # --------------------------------------------------------------------------- |
| 356 | # Error routing: all user errors go to stderr |
| 357 | # --------------------------------------------------------------------------- |
| 358 | |
| 359 | |
| 360 | def test_no_commits_error_on_stderr(tmp_path: pathlib.Path) -> None: |
| 361 | _init_repo(tmp_path) |
| 362 | result = _invoke(["describe"], _env(tmp_path)) |
| 363 | assert result.exit_code != 0 |
| 364 | assert result.stderr != "" or "commits" in result.output.lower() |
| 365 | |
| 366 | |
| 367 | def test_ref_not_found_error_on_stderr(tmp_path: pathlib.Path) -> None: |
| 368 | _init_repo(tmp_path) |
| 369 | _make_commit(tmp_path, content=b"x") |
| 370 | result = _invoke(["describe", "--ref", "nonexistent"], _env(tmp_path)) |
| 371 | assert result.exit_code != 0 |
| 372 | |
| 373 | |
| 374 | def test_require_tag_no_tags_error(tmp_path: pathlib.Path) -> None: |
| 375 | _init_repo(tmp_path) |
| 376 | _make_commit(tmp_path, content=b"no-tag") |
| 377 | result = _invoke(["describe", "--require-tag"], _env(tmp_path)) |
| 378 | assert result.exit_code != 0 |
| 379 | |
| 380 | |
| 381 | def test_exact_match_not_on_tag_error(tmp_path: pathlib.Path) -> None: |
| 382 | _init_repo(tmp_path) |
| 383 | c1 = _make_commit(tmp_path, content=b"c1") |
| 384 | _make_tag(tmp_path, "v1", c1) |
| 385 | _make_commit(tmp_path, parent_id=c1, content=b"c2") |
| 386 | result = _invoke(["describe", "--exact-match"], _env(tmp_path)) |
| 387 | assert result.exit_code != 0 |
| 388 | |
| 389 | |
| 390 | def test_abbrev_too_small_error(tmp_path: pathlib.Path) -> None: |
| 391 | _init_repo(tmp_path) |
| 392 | _make_commit(tmp_path, content=b"ab") |
| 393 | result = _invoke(["describe", "--abbrev", "2"], _env(tmp_path)) |
| 394 | assert result.exit_code != 0 |
| 395 | |
| 396 | |
| 397 | def test_abbrev_too_large_error(tmp_path: pathlib.Path) -> None: |
| 398 | _init_repo(tmp_path) |
| 399 | _make_commit(tmp_path, content=b"ab") |
| 400 | result = _invoke(["describe", "--abbrev", "65"], _env(tmp_path)) |
| 401 | assert result.exit_code != 0 |
| 402 | |
| 403 | |
| 404 | # --------------------------------------------------------------------------- |
| 405 | # JSON schema: _DescribeJson |
| 406 | # --------------------------------------------------------------------------- |
| 407 | |
| 408 | |
| 409 | def test_json_schema_all_fields(tmp_path: pathlib.Path) -> None: |
| 410 | _init_repo(tmp_path) |
| 411 | cid = _make_commit(tmp_path, content=b"schema") |
| 412 | _make_tag(tmp_path, "v1.0.0", cid) |
| 413 | result = _invoke(["describe", "--json"], _env(tmp_path)) |
| 414 | assert result.exit_code == 0 |
| 415 | data = _parse_json(result) |
| 416 | assert data["tag"] == "v1.0.0" |
| 417 | assert data["distance"] == 0 |
| 418 | assert data["exact"] is True |
| 419 | assert data["repo_id"] == _REPO_ID |
| 420 | assert data["branch"] == "main" |
| 421 | assert data["commit_id"] == cid |
| 422 | assert data["short_sha"] == cid[:len("sha256:") + 12] |
| 423 | |
| 424 | |
| 425 | def test_json_schema_no_tag(tmp_path: pathlib.Path) -> None: |
| 426 | _init_repo(tmp_path) |
| 427 | cid = _make_commit(tmp_path, content=b"no-tag-json") |
| 428 | result = _invoke(["describe", "--json"], _env(tmp_path)) |
| 429 | assert result.exit_code == 0 |
| 430 | data = _parse_json(result) |
| 431 | assert data["tag"] is None |
| 432 | assert data["name"] == cid[:len("sha256:") + 12] |
| 433 | assert data["exact"] is False |
| 434 | |
| 435 | |
| 436 | def test_json_schema_with_distance(tmp_path: pathlib.Path) -> None: |
| 437 | _init_repo(tmp_path) |
| 438 | c1 = _make_commit(tmp_path, content=b"root") |
| 439 | _make_tag(tmp_path, "v0.1", c1) |
| 440 | _make_commit(tmp_path, parent_id=c1, content=b"next") |
| 441 | result = _invoke(["describe", "--json"], _env(tmp_path)) |
| 442 | assert result.exit_code == 0 |
| 443 | data = _parse_json(result) |
| 444 | assert data["tag"] == "v0.1" |
| 445 | assert data["distance"] == 1 |
| 446 | assert data["exact"] is False |
| 447 | |
| 448 | |
| 449 | def test_json_abbrev_reflected(tmp_path: pathlib.Path) -> None: |
| 450 | _init_repo(tmp_path) |
| 451 | cid = _make_commit(tmp_path, content=b"abbrev-json") |
| 452 | result = _invoke(["describe", "--abbrev", "8", "--json"], _env(tmp_path)) |
| 453 | assert result.exit_code == 0 |
| 454 | data = _parse_json(result) |
| 455 | assert data["short_sha"].startswith("sha256:") |
| 456 | assert len(data["short_sha"]) == len("sha256:") + 8 |
| 457 | |
| 458 | |
| 459 | # --------------------------------------------------------------------------- |
| 460 | # New flags: --match, --exact-match, --first-parent, --abbrev |
| 461 | # --------------------------------------------------------------------------- |
| 462 | |
| 463 | |
| 464 | def test_flag_match_filters_tags(tmp_path: pathlib.Path) -> None: |
| 465 | _init_repo(tmp_path) |
| 466 | cid = _make_commit(tmp_path, content=b"match-flag") |
| 467 | _make_tag(tmp_path, "nightly-1", cid) |
| 468 | _make_tag(tmp_path, "v1.0.0", cid) |
| 469 | result = _invoke(["describe", "--match", "v*", "--json"], _env(tmp_path)) |
| 470 | assert result.exit_code == 0 |
| 471 | data = _parse_json(result) |
| 472 | assert data["tag"] == "v1.0.0" |
| 473 | |
| 474 | |
| 475 | def test_flag_match_no_matching_tag(tmp_path: pathlib.Path) -> None: |
| 476 | _init_repo(tmp_path) |
| 477 | cid = _make_commit(tmp_path, content=b"match-none") |
| 478 | _make_tag(tmp_path, "nightly-1", cid) |
| 479 | result = _invoke(["describe", "--match", "v*", "--json"], _env(tmp_path)) |
| 480 | assert result.exit_code == 0 |
| 481 | data = _parse_json(result) |
| 482 | assert data["tag"] is None |
| 483 | |
| 484 | |
| 485 | def test_flag_exact_match_on_tag(tmp_path: pathlib.Path) -> None: |
| 486 | _init_repo(tmp_path) |
| 487 | cid = _make_commit(tmp_path, content=b"exact-flag") |
| 488 | _make_tag(tmp_path, "v1.0", cid) |
| 489 | result = _invoke(["describe", "--exact-match", "--json"], _env(tmp_path)) |
| 490 | assert result.exit_code == 0 |
| 491 | data = _parse_json(result) |
| 492 | assert data["exact"] is True |
| 493 | |
| 494 | |
| 495 | def test_flag_exact_match_off_tag_fails(tmp_path: pathlib.Path) -> None: |
| 496 | _init_repo(tmp_path) |
| 497 | c1 = _make_commit(tmp_path, content=b"em-root") |
| 498 | _make_tag(tmp_path, "v1.0", c1) |
| 499 | _make_commit(tmp_path, parent_id=c1, content=b"em-next") |
| 500 | result = _invoke(["describe", "--exact-match"], _env(tmp_path)) |
| 501 | assert result.exit_code != 0 |
| 502 | |
| 503 | |
| 504 | def test_flag_first_parent(tmp_path: pathlib.Path) -> None: |
| 505 | _init_repo(tmp_path) |
| 506 | c1 = _make_commit(tmp_path, content=b"fp-root") |
| 507 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"feat-side", branch="feat") |
| 508 | _make_tag(tmp_path, "side-tag", c2) |
| 509 | c3 = _make_commit( |
| 510 | tmp_path, parent_id=c1, parent2_id=c2, content=b"fp-merge", branch="main" |
| 511 | ) |
| 512 | # --first-parent should not see side-tag. |
| 513 | result = _invoke(["describe", "--first-parent", "--json"], _env(tmp_path)) |
| 514 | assert result.exit_code == 0 |
| 515 | data = _parse_json(result) |
| 516 | assert data["tag"] is None # side-tag not reachable via first-parent |
| 517 | |
| 518 | |
| 519 | def test_flag_abbrev(tmp_path: pathlib.Path) -> None: |
| 520 | _init_repo(tmp_path) |
| 521 | _make_commit(tmp_path, content=b"abbrev-flag") |
| 522 | result = _invoke(["describe", "--abbrev", "16", "--json"], _env(tmp_path)) |
| 523 | assert result.exit_code == 0 |
| 524 | data = _parse_json(result) |
| 525 | assert len(data["short_sha"]) == len("sha256:") + 16 |
| 526 | |
| 527 | |
| 528 | # --------------------------------------------------------------------------- |
| 529 | # Integration |
| 530 | # --------------------------------------------------------------------------- |
| 531 | |
| 532 | |
| 533 | def test_integration_ref_to_branch_tip(tmp_path: pathlib.Path) -> None: |
| 534 | _init_repo(tmp_path) |
| 535 | c1 = _make_commit(tmp_path, content=b"ref-root") |
| 536 | _make_tag(tmp_path, "v10.0", c1) |
| 537 | _make_commit(tmp_path, parent_id=c1, content=b"ref-next") |
| 538 | # Describe the HEAD (which is 1 hop past the tag). |
| 539 | result = _invoke(["describe", "--json"], _env(tmp_path)) |
| 540 | assert result.exit_code == 0 |
| 541 | data = _parse_json(result) |
| 542 | assert data["distance"] == 1 |
| 543 | assert data["tag"] == "v10.0" |
| 544 | |
| 545 | |
| 546 | def test_integration_long_and_match_combined(tmp_path: pathlib.Path) -> None: |
| 547 | _init_repo(tmp_path) |
| 548 | cid = _make_commit(tmp_path, content=b"combo") |
| 549 | _make_tag(tmp_path, "v5.0.0", cid) |
| 550 | result = _invoke( |
| 551 | ["describe", "--long", "--match", "v*", "--json"], _env(tmp_path) |
| 552 | ) |
| 553 | assert result.exit_code == 0 |
| 554 | data = _parse_json(result) |
| 555 | assert data["name"].startswith("v5.0.0-0-sha256:") |
| 556 | |
| 557 | |
| 558 | def test_integration_require_tag_passes_when_tag_exists( |
| 559 | tmp_path: pathlib.Path, |
| 560 | ) -> None: |
| 561 | _init_repo(tmp_path) |
| 562 | cid = _make_commit(tmp_path, content=b"req-tag") |
| 563 | _make_tag(tmp_path, "v7.0", cid) |
| 564 | result = _invoke(["describe", "--require-tag", "--json"], _env(tmp_path)) |
| 565 | assert result.exit_code == 0 |
| 566 | |
| 567 | |
| 568 | def test_integration_text_output_sanitized(tmp_path: pathlib.Path) -> None: |
| 569 | _init_repo(tmp_path) |
| 570 | cid = _make_commit(tmp_path, content=b"text-sanitize") |
| 571 | _make_tag(tmp_path, "v1.0\x1b[1mBOLD\x1b[0m", cid) |
| 572 | result = _invoke(["describe"], _env(tmp_path)) |
| 573 | assert result.exit_code == 0 |
| 574 | assert "\x1b[1m" not in result.output |
| 575 | |
| 576 | |
| 577 | # --------------------------------------------------------------------------- |
| 578 | # E2E: help output |
| 579 | # --------------------------------------------------------------------------- |
| 580 | |
| 581 | |
| 582 | def test_help_contains_new_flags() -> None: |
| 583 | result = _invoke(["describe", "--help"], {}) |
| 584 | assert result.exit_code == 0 |
| 585 | for flag in ("--match", "--exact-match", "--first-parent", "--abbrev", "--json"): |
| 586 | assert flag in result.output, f"Missing flag in help: {flag}" |
| 587 | |
| 588 | |
| 589 | def test_help_mentions_json_schema() -> None: |
| 590 | result = _invoke(["describe", "--help"], {}) |
| 591 | assert "json" in result.output.lower() |
| 592 | |
| 593 | |
| 594 | # --------------------------------------------------------------------------- |
| 595 | # Stress: deep ancestry + many tags + concurrent reads |
| 596 | # --------------------------------------------------------------------------- |
| 597 | |
| 598 | |
| 599 | def test_stress_5000_commit_chain(tmp_path: pathlib.Path) -> None: |
| 600 | _init_repo(tmp_path) |
| 601 | prev: str | None = None |
| 602 | root_cid = "" |
| 603 | for i in range(5_000): |
| 604 | cid = _make_commit(tmp_path, parent_id=prev, content=f"s{i}".encode()) |
| 605 | if i == 0: |
| 606 | root_cid = cid |
| 607 | prev = cid |
| 608 | |
| 609 | _make_tag(tmp_path, "v-deep", root_cid) |
| 610 | assert prev is not None |
| 611 | r = describe_commit(tmp_path, _REPO_ID, prev) |
| 612 | assert r["tag"] == "v-deep" |
| 613 | assert r["distance"] == 4_999 |
| 614 | |
| 615 | |
| 616 | def test_stress_200_tags_repo(tmp_path: pathlib.Path) -> None: |
| 617 | """Many tags — describe still picks the nearest one efficiently.""" |
| 618 | _init_repo(tmp_path) |
| 619 | commits: list[str] = [] |
| 620 | prev: str | None = None |
| 621 | for i in range(200): |
| 622 | cid = _make_commit(tmp_path, parent_id=prev, content=f"t{i}".encode()) |
| 623 | commits.append(cid) |
| 624 | # Tag every 10th commit. |
| 625 | if i % 10 == 0: |
| 626 | _make_tag(tmp_path, f"v{i}.0", cid) |
| 627 | prev = cid |
| 628 | |
| 629 | # HEAD is commits[-1], nearest tag is v190.0 (at commits[190]). |
| 630 | r = describe_commit(tmp_path, _REPO_ID, commits[-1]) |
| 631 | assert r["tag"] == "v190.0" |
| 632 | assert r["distance"] == 9 |
| 633 | |
| 634 | |
| 635 | def test_stress_concurrent_describe(tmp_path: pathlib.Path) -> None: |
| 636 | """Concurrent --json calls must all return consistent, valid JSON.""" |
| 637 | _init_repo(tmp_path) |
| 638 | c1 = _make_commit(tmp_path, content=b"conc-root") |
| 639 | _make_tag(tmp_path, "v-conc", c1) |
| 640 | _make_commit(tmp_path, parent_id=c1, content=b"conc-next") |
| 641 | |
| 642 | invoke_lock = threading.Lock() |
| 643 | errors: list[str] = [] |
| 644 | |
| 645 | def _worker() -> None: |
| 646 | with invoke_lock: |
| 647 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 648 | try: |
| 649 | assert r.exit_code == 0 |
| 650 | data = _parse_json(r) |
| 651 | assert data["tag"] == "v-conc" |
| 652 | assert data["distance"] == 1 |
| 653 | except Exception as exc: |
| 654 | errors.append(str(exc)) |
| 655 | |
| 656 | threads = [threading.Thread(target=_worker) for _ in range(8)] |
| 657 | for t in threads: |
| 658 | t.start() |
| 659 | for t in threads: |
| 660 | t.join() |
| 661 | |
| 662 | assert errors == [], f"Concurrent failures: {errors}" |
| 663 | |
| 664 | |
| 665 | # --------------------------------------------------------------------------- |
| 666 | # JSON schema: duration_ms + exit_code always present |
| 667 | # --------------------------------------------------------------------------- |
| 668 | |
| 669 | |
| 670 | class TestJsonSchemaComplete: |
| 671 | """Every --json path includes duration_ms and exit_code.""" |
| 672 | |
| 673 | def test_elapsed_present_on_tag(self, tmp_path: pathlib.Path) -> None: |
| 674 | _init_repo(tmp_path) |
| 675 | cid = _make_commit(tmp_path, content=b"sc-tag") |
| 676 | _make_tag(tmp_path, "v1.0.0", cid) |
| 677 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 678 | assert r.exit_code == 0 |
| 679 | raw = json.loads(r.output) |
| 680 | assert "duration_ms" in raw |
| 681 | assert "exit_code" in raw |
| 682 | |
| 683 | def test_elapsed_present_no_tag(self, tmp_path: pathlib.Path) -> None: |
| 684 | _init_repo(tmp_path) |
| 685 | _make_commit(tmp_path, content=b"sc-notag") |
| 686 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 687 | assert r.exit_code == 0 |
| 688 | raw = json.loads(r.output) |
| 689 | assert "duration_ms" in raw |
| 690 | assert "exit_code" in raw |
| 691 | |
| 692 | def test_elapsed_present_with_distance(self, tmp_path: pathlib.Path) -> None: |
| 693 | _init_repo(tmp_path) |
| 694 | c1 = _make_commit(tmp_path, content=b"sc-dist-root") |
| 695 | _make_tag(tmp_path, "v0.1", c1) |
| 696 | _make_commit(tmp_path, parent_id=c1, content=b"sc-dist-next") |
| 697 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 698 | assert r.exit_code == 0 |
| 699 | raw = json.loads(r.output) |
| 700 | assert "duration_ms" in raw |
| 701 | assert raw["exit_code"] == 0 |
| 702 | |
| 703 | def test_elapsed_present_long_format(self, tmp_path: pathlib.Path) -> None: |
| 704 | _init_repo(tmp_path) |
| 705 | cid = _make_commit(tmp_path, content=b"sc-long") |
| 706 | _make_tag(tmp_path, "v2.0.0", cid) |
| 707 | r = _invoke(["describe", "--long", "--json"], _env(tmp_path)) |
| 708 | assert r.exit_code == 0 |
| 709 | raw = json.loads(r.output) |
| 710 | assert "duration_ms" in raw |
| 711 | assert raw["exit_code"] == 0 |
| 712 | |
| 713 | def test_elapsed_present_with_match(self, tmp_path: pathlib.Path) -> None: |
| 714 | _init_repo(tmp_path) |
| 715 | cid = _make_commit(tmp_path, content=b"sc-match") |
| 716 | _make_tag(tmp_path, "v3.0.0", cid) |
| 717 | r = _invoke(["describe", "--match", "v*", "--json"], _env(tmp_path)) |
| 718 | assert r.exit_code == 0 |
| 719 | raw = json.loads(r.output) |
| 720 | assert "duration_ms" in raw |
| 721 | |
| 722 | def test_elapsed_present_with_abbrev(self, tmp_path: pathlib.Path) -> None: |
| 723 | _init_repo(tmp_path) |
| 724 | _make_commit(tmp_path, content=b"sc-abbrev") |
| 725 | r = _invoke(["describe", "--abbrev", "8", "--json"], _env(tmp_path)) |
| 726 | assert r.exit_code == 0 |
| 727 | raw = json.loads(r.output) |
| 728 | assert "duration_ms" in raw |
| 729 | assert raw["exit_code"] == 0 |
| 730 | |
| 731 | def test_all_eight_base_fields_present(self, tmp_path: pathlib.Path) -> None: |
| 732 | _init_repo(tmp_path) |
| 733 | cid = _make_commit(tmp_path, content=b"sc-all") |
| 734 | _make_tag(tmp_path, "v9.0.0", cid) |
| 735 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 736 | assert r.exit_code == 0 |
| 737 | raw = json.loads(r.output) |
| 738 | for field in ("commit_id", "tag", "distance", "short_sha", "name", |
| 739 | "exact", "repo_id", "branch", "duration_ms", "exit_code"): |
| 740 | assert field in raw, f"Missing field: {field}" |
| 741 | |
| 742 | def test_exit_code_field_is_zero(self, tmp_path: pathlib.Path) -> None: |
| 743 | _init_repo(tmp_path) |
| 744 | cid = _make_commit(tmp_path, content=b"sc-exit") |
| 745 | _make_tag(tmp_path, "v10.0.0", cid) |
| 746 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 747 | assert r.exit_code == 0 |
| 748 | raw = json.loads(r.output) |
| 749 | assert raw["exit_code"] == 0 |
| 750 | |
| 751 | |
| 752 | # --------------------------------------------------------------------------- |
| 753 | # duration_ms: type and magnitude checks |
| 754 | # --------------------------------------------------------------------------- |
| 755 | |
| 756 | |
| 757 | class TestElapsedSeconds: |
| 758 | """duration_ms is a non-negative float in a reasonable range.""" |
| 759 | |
| 760 | def test_elapsed_is_float(self, tmp_path: pathlib.Path) -> None: |
| 761 | _init_repo(tmp_path) |
| 762 | _make_commit(tmp_path, content=b"el-float") |
| 763 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 764 | raw = json.loads(r.output) |
| 765 | assert isinstance(raw["duration_ms"], float) |
| 766 | |
| 767 | def test_elapsed_non_negative(self, tmp_path: pathlib.Path) -> None: |
| 768 | _init_repo(tmp_path) |
| 769 | _make_commit(tmp_path, content=b"el-nonneg") |
| 770 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 771 | raw = json.loads(r.output) |
| 772 | assert raw["duration_ms"] >= 0.0 |
| 773 | |
| 774 | def test_elapsed_under_ten_seconds(self, tmp_path: pathlib.Path) -> None: |
| 775 | _init_repo(tmp_path) |
| 776 | _make_commit(tmp_path, content=b"el-under") |
| 777 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 778 | raw = json.loads(r.output) |
| 779 | assert raw["duration_ms"] < 10.0 |
| 780 | |
| 781 | def test_elapsed_with_tag(self, tmp_path: pathlib.Path) -> None: |
| 782 | _init_repo(tmp_path) |
| 783 | cid = _make_commit(tmp_path, content=b"el-tag") |
| 784 | _make_tag(tmp_path, "v1.2.3", cid) |
| 785 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 786 | raw = json.loads(r.output) |
| 787 | assert raw["duration_ms"] >= 0.0 |
| 788 | |
| 789 | def test_elapsed_with_require_tag(self, tmp_path: pathlib.Path) -> None: |
| 790 | _init_repo(tmp_path) |
| 791 | cid = _make_commit(tmp_path, content=b"el-req") |
| 792 | _make_tag(tmp_path, "v1.0", cid) |
| 793 | r = _invoke(["describe", "--require-tag", "--json"], _env(tmp_path)) |
| 794 | assert r.exit_code == 0 |
| 795 | raw = json.loads(r.output) |
| 796 | assert "duration_ms" in raw |
| 797 | |
| 798 | def test_elapsed_six_decimal_places(self, tmp_path: pathlib.Path) -> None: |
| 799 | _init_repo(tmp_path) |
| 800 | _make_commit(tmp_path, content=b"el-prec") |
| 801 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 802 | raw = json.loads(r.output) |
| 803 | # round(..., 6) produces at most 6 decimal places — str check |
| 804 | s = str(raw["duration_ms"]) |
| 805 | dec = s.split(".")[-1] if "." in s else "" |
| 806 | assert len(dec) <= 6 |
| 807 | |
| 808 | |
| 809 | # --------------------------------------------------------------------------- |
| 810 | # exit_code field |
| 811 | # --------------------------------------------------------------------------- |
| 812 | |
| 813 | |
| 814 | class TestExitCode: |
| 815 | """exit_code field mirrors process exit code; always 0 on success.""" |
| 816 | |
| 817 | def test_exit_code_zero_no_tag(self, tmp_path: pathlib.Path) -> None: |
| 818 | _init_repo(tmp_path) |
| 819 | _make_commit(tmp_path, content=b"ec-notag") |
| 820 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 821 | assert r.exit_code == 0 |
| 822 | assert json.loads(r.output)["exit_code"] == 0 |
| 823 | |
| 824 | def test_exit_code_zero_on_tag(self, tmp_path: pathlib.Path) -> None: |
| 825 | _init_repo(tmp_path) |
| 826 | cid = _make_commit(tmp_path, content=b"ec-tag") |
| 827 | _make_tag(tmp_path, "v1.0.0", cid) |
| 828 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 829 | assert r.exit_code == 0 |
| 830 | assert json.loads(r.output)["exit_code"] == 0 |
| 831 | |
| 832 | def test_exit_code_zero_with_distance(self, tmp_path: pathlib.Path) -> None: |
| 833 | _init_repo(tmp_path) |
| 834 | c1 = _make_commit(tmp_path, content=b"ec-dist-root") |
| 835 | _make_tag(tmp_path, "v0.5", c1) |
| 836 | _make_commit(tmp_path, parent_id=c1, content=b"ec-dist-next") |
| 837 | r = _invoke(["describe", "--json"], _env(tmp_path)) |
| 838 | assert r.exit_code == 0 |
| 839 | assert json.loads(r.output)["exit_code"] == 0 |
| 840 | |
| 841 | def test_exit_code_zero_long_format(self, tmp_path: pathlib.Path) -> None: |
| 842 | _init_repo(tmp_path) |
| 843 | cid = _make_commit(tmp_path, content=b"ec-long") |
| 844 | _make_tag(tmp_path, "v2.0.0", cid) |
| 845 | r = _invoke(["describe", "--long", "--json"], _env(tmp_path)) |
| 846 | assert r.exit_code == 0 |
| 847 | assert json.loads(r.output)["exit_code"] == 0 |
| 848 | |
| 849 | def test_exit_code_zero_with_abbrev(self, tmp_path: pathlib.Path) -> None: |
| 850 | _init_repo(tmp_path) |
| 851 | _make_commit(tmp_path, content=b"ec-abbrev") |
| 852 | r = _invoke(["describe", "--abbrev", "10", "--json"], _env(tmp_path)) |
| 853 | assert r.exit_code == 0 |
| 854 | assert json.loads(r.output)["exit_code"] == 0 |