test_cmd_commit.py
python
sha256:4d09a52c06fbc389006963ad1e5ca6ee48c3cb72799f1a322561035b263db67d
merge conflict resolve
Human
patch
41 days ago
| 1 | """Tests for ``muse commit``. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | Unit — parser flags, pure-logic helpers, sanitization. |
| 6 | Integration — actual repo operations: commits, snapshots, reflog, rerere. |
| 7 | End-to-end — CLI invocations, text and JSON output paths. |
| 8 | Security — ANSI injection, author impersonation, provenance field caps. |
| 9 | Stress — 100 sequential commits, large manifests, concurrent writes. |
| 10 | """ |
| 11 | |
| 12 | from __future__ import annotations |
| 13 | |
| 14 | import argparse |
| 15 | import json |
| 16 | import os |
| 17 | import pathlib |
| 18 | import subprocess |
| 19 | import threading |
| 20 | import time |
| 21 | from unittest.mock import patch |
| 22 | |
| 23 | import pytest |
| 24 | |
| 25 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 26 | from muse.core.store import ( |
| 27 | get_head_commit_id, |
| 28 | read_commit, |
| 29 | read_current_branch, |
| 30 | read_snapshot, |
| 31 | ) |
| 32 | |
| 33 | runner = CliRunner() |
| 34 | |
| 35 | # ────────────────────────────────────────────────────────────────────────────── |
| 36 | # Helpers |
| 37 | # ────────────────────────────────────────────────────────────────────────────── |
| 38 | |
| 39 | |
| 40 | def _invoke(repo: pathlib.Path, args: list[str]) -> InvokeResult: |
| 41 | """Run a muse command in *repo* and return the result.""" |
| 42 | saved = os.getcwd() |
| 43 | try: |
| 44 | os.chdir(repo) |
| 45 | return runner.invoke(None, args) |
| 46 | finally: |
| 47 | os.chdir(saved) |
| 48 | |
| 49 | |
| 50 | def _commit(repo: pathlib.Path, *extra: str) -> InvokeResult: |
| 51 | return _invoke(repo, ["commit", *extra]) |
| 52 | |
| 53 | |
| 54 | def _init_repo(repo: pathlib.Path) -> InvokeResult: |
| 55 | repo.mkdir(parents=True, exist_ok=True) |
| 56 | return _invoke(repo, ["init"]) |
| 57 | |
| 58 | |
| 59 | # ────────────────────────────────────────────────────────────────────────────── |
| 60 | # Fixtures |
| 61 | # ────────────────────────────────────────────────────────────────────────────── |
| 62 | |
| 63 | |
| 64 | @pytest.fixture() |
| 65 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 66 | """Initialised repo with one tracked file ready to commit.""" |
| 67 | _init_repo(tmp_path) |
| 68 | (tmp_path / "a.py").write_text("x = 1\n") |
| 69 | return tmp_path |
| 70 | |
| 71 | |
| 72 | # ────────────────────────────────────────────────────────────────────────────── |
| 73 | # Unit — parser flags |
| 74 | # ────────────────────────────────────────────────────────────────────────────── |
| 75 | |
| 76 | |
| 77 | class TestRegisterFlags: |
| 78 | """All expected CLI flags are registered on the commit subcommand.""" |
| 79 | |
| 80 | def _parse(self, *args: str) -> argparse.Namespace: |
| 81 | from muse.cli.commands.commit import register |
| 82 | |
| 83 | p = argparse.ArgumentParser() |
| 84 | sub = p.add_subparsers() |
| 85 | register(sub) |
| 86 | return p.parse_args(["commit", *args]) |
| 87 | |
| 88 | def test_message_flag(self) -> None: |
| 89 | ns = self._parse("-m", "hello") |
| 90 | assert ns.message == "hello" |
| 91 | |
| 92 | def test_allow_empty_flag(self) -> None: |
| 93 | ns = self._parse("-m", "x", "--allow-empty") |
| 94 | assert ns.allow_empty is True |
| 95 | |
| 96 | def test_dry_run_short_flag(self) -> None: |
| 97 | ns = self._parse("-m", "x", "-n") |
| 98 | assert ns.dry_run is True |
| 99 | |
| 100 | def test_dry_run_long_flag(self) -> None: |
| 101 | ns = self._parse("-m", "x", "--dry-run") |
| 102 | assert ns.dry_run is True |
| 103 | |
| 104 | def test_json_flag(self) -> None: |
| 105 | ns = self._parse("-m", "x", "--json") |
| 106 | assert ns.fmt == "json" |
| 107 | |
| 108 | def test_format_text_default(self) -> None: |
| 109 | ns = self._parse("-m", "x") |
| 110 | assert ns.fmt == "text" |
| 111 | |
| 112 | def test_agent_id_flag(self) -> None: |
| 113 | ns = self._parse("-m", "x", "--agent-id", "bot-1") |
| 114 | assert ns.agent_id == "bot-1" |
| 115 | |
| 116 | def test_model_id_flag(self) -> None: |
| 117 | ns = self._parse("-m", "x", "--model-id", "claude-4") |
| 118 | assert ns.model_id == "claude-4" |
| 119 | |
| 120 | def test_toolchain_id_flag(self) -> None: |
| 121 | ns = self._parse("-m", "x", "--toolchain-id", "cursor-v1") |
| 122 | assert ns.toolchain_id == "cursor-v1" |
| 123 | |
| 124 | def test_section_flag(self) -> None: |
| 125 | ns = self._parse("-m", "x", "--section", "chorus") |
| 126 | assert ns.section == "chorus" |
| 127 | |
| 128 | def test_track_flag(self) -> None: |
| 129 | ns = self._parse("-m", "x", "--track", "bass") |
| 130 | assert ns.track == "bass" |
| 131 | |
| 132 | def test_emotion_flag(self) -> None: |
| 133 | ns = self._parse("-m", "x", "--emotion", "joyful") |
| 134 | assert ns.emotion == "joyful" |
| 135 | |
| 136 | def test_author_flag(self) -> None: |
| 137 | ns = self._parse("-m", "x", "--author", "alice") |
| 138 | assert ns.author == "alice" |
| 139 | |
| 140 | def test_sign_flag(self) -> None: |
| 141 | ns = self._parse("-m", "x", "--sign") |
| 142 | assert ns.sign is True |
| 143 | |
| 144 | |
| 145 | # ────────────────────────────────────────────────────────────────────────────── |
| 146 | # Unit — _MAX_FIELD_LEN constant |
| 147 | # ────────────────────────────────────────────────────────────────────────────── |
| 148 | |
| 149 | |
| 150 | class TestMaxFieldLen: |
| 151 | def test_constant_exists_and_is_256(self) -> None: |
| 152 | from muse.cli.commands.commit import _MAX_FIELD_LEN |
| 153 | |
| 154 | assert _MAX_FIELD_LEN == 256 |
| 155 | |
| 156 | def test_no_separate_max_author_constant(self) -> None: |
| 157 | import muse.cli.commands.commit as m |
| 158 | |
| 159 | assert not hasattr(m, "_MAX_AUTHOR"), "_MAX_AUTHOR should be replaced by _MAX_FIELD_LEN" |
| 160 | assert not hasattr(m, "_MAX_PROV"), "_MAX_PROV should be replaced by _MAX_FIELD_LEN" |
| 161 | |
| 162 | |
| 163 | # ────────────────────────────────────────────────────────────────────────────── |
| 164 | # Unit — dead-code removal |
| 165 | # ────────────────────────────────────────────────────────────────────────────── |
| 166 | |
| 167 | |
| 168 | class TestDeadCodeRemoved: |
| 169 | def test_read_branch_removed(self) -> None: |
| 170 | import muse.cli.commands.commit as m |
| 171 | |
| 172 | assert not hasattr(m, "_read_branch"), ( |
| 173 | "_read_branch was a dead wrapper; it should have been deleted" |
| 174 | ) |
| 175 | |
| 176 | def test_read_parent_id_removed(self) -> None: |
| 177 | import muse.cli.commands.commit as m |
| 178 | |
| 179 | assert not hasattr(m, "_read_parent_id"), ( |
| 180 | "_read_parent_id was a dead wrapper; it should have been deleted" |
| 181 | ) |
| 182 | |
| 183 | |
| 184 | # ────────────────────────────────────────────────────────────────────────────── |
| 185 | # Unit — inline imports removed |
| 186 | # ────────────────────────────────────────────────────────────────────────────── |
| 187 | |
| 188 | |
| 189 | class TestNoInlineImports: |
| 190 | def test_sign_commit_record_is_module_level_import(self) -> None: |
| 191 | import inspect |
| 192 | |
| 193 | import muse.cli.commands.commit as m |
| 194 | |
| 195 | src = inspect.getsource(m.run) |
| 196 | assert "from muse.core.provenance import sign_commit_record" not in src, ( |
| 197 | "sign_commit_record import must be at module level, not inside run()" |
| 198 | ) |
| 199 | |
| 200 | def test_no_inline_store_imports(self) -> None: |
| 201 | import inspect |
| 202 | |
| 203 | import muse.cli.commands.commit as m |
| 204 | |
| 205 | src = inspect.getsource(m.run) |
| 206 | assert "from muse.core.store import" not in src, ( |
| 207 | "store imports inside run() should be at module level" |
| 208 | ) |
| 209 | |
| 210 | |
| 211 | # ────────────────────────────────────────────────────────────────────────────── |
| 212 | # Integration — basic commit lifecycle |
| 213 | # ────────────────────────────────────────────────────────────────────────────── |
| 214 | |
| 215 | |
| 216 | class TestBasicCommit: |
| 217 | def test_first_commit_succeeds(self, repo: pathlib.Path) -> None: |
| 218 | result = _commit(repo, "-m", "init") |
| 219 | assert result.exit_code == 0 |
| 220 | assert "init" in result.output |
| 221 | |
| 222 | def test_commit_creates_commit_record(self, repo: pathlib.Path) -> None: |
| 223 | _commit(repo, "-m", "first") |
| 224 | branch = read_current_branch(repo) |
| 225 | cid = get_head_commit_id(repo, branch) |
| 226 | assert cid is not None |
| 227 | rec = read_commit(repo, cid) |
| 228 | assert rec is not None |
| 229 | assert rec.message == "first" |
| 230 | |
| 231 | def test_commit_creates_snapshot(self, repo: pathlib.Path) -> None: |
| 232 | _commit(repo, "-m", "snap") |
| 233 | branch = read_current_branch(repo) |
| 234 | cid = get_head_commit_id(repo, branch) |
| 235 | assert cid is not None |
| 236 | rec = read_commit(repo, cid) |
| 237 | assert rec is not None |
| 238 | snap = read_snapshot(repo, rec.snapshot_id) |
| 239 | assert snap is not None |
| 240 | assert len(snap.manifest) >= 1 |
| 241 | |
| 242 | def test_commit_advances_branch_ref(self, repo: pathlib.Path) -> None: |
| 243 | _commit(repo, "-m", "first") |
| 244 | cid1 = get_head_commit_id(repo, "main") |
| 245 | (repo / "b.py").write_text("y = 2\n") |
| 246 | _commit(repo, "-m", "second") |
| 247 | cid2 = get_head_commit_id(repo, "main") |
| 248 | assert cid1 != cid2 |
| 249 | |
| 250 | def test_second_commit_has_parent(self, repo: pathlib.Path) -> None: |
| 251 | _commit(repo, "-m", "first") |
| 252 | cid1 = get_head_commit_id(repo, "main") |
| 253 | (repo / "b.py").write_text("y = 2\n") |
| 254 | _commit(repo, "-m", "second") |
| 255 | cid2 = get_head_commit_id(repo, "main") |
| 256 | assert cid2 is not None |
| 257 | rec2 = read_commit(repo, cid2) |
| 258 | assert rec2 is not None |
| 259 | assert rec2.parent_commit_id == cid1 |
| 260 | |
| 261 | def test_nothing_to_commit_exits_0(self, repo: pathlib.Path) -> None: |
| 262 | _commit(repo, "-m", "first") |
| 263 | result = _commit(repo, "-m", "second") |
| 264 | assert result.exit_code == 0 |
| 265 | assert "Nothing to commit" in result.output |
| 266 | |
| 267 | def test_metadata_section_stored(self, repo: pathlib.Path) -> None: |
| 268 | _commit(repo, "-m", "chorus", "--section", "chorus") |
| 269 | branch = read_current_branch(repo) |
| 270 | cid = get_head_commit_id(repo, branch) |
| 271 | assert cid is not None |
| 272 | rec = read_commit(repo, cid) |
| 273 | assert rec is not None |
| 274 | assert rec.metadata.get("section") == "chorus" |
| 275 | |
| 276 | def test_metadata_track_stored(self, repo: pathlib.Path) -> None: |
| 277 | _commit(repo, "-m", "bass", "--track", "bass") |
| 278 | branch = read_current_branch(repo) |
| 279 | cid = get_head_commit_id(repo, branch) |
| 280 | assert cid is not None |
| 281 | rec = read_commit(repo, cid) |
| 282 | assert rec is not None |
| 283 | assert rec.metadata.get("track") == "bass" |
| 284 | |
| 285 | def test_metadata_emotion_stored(self, repo: pathlib.Path) -> None: |
| 286 | _commit(repo, "-m", "joy", "--emotion", "joyful") |
| 287 | branch = read_current_branch(repo) |
| 288 | cid = get_head_commit_id(repo, branch) |
| 289 | assert cid is not None |
| 290 | rec = read_commit(repo, cid) |
| 291 | assert rec is not None |
| 292 | assert rec.metadata.get("emotion") == "joyful" |
| 293 | |
| 294 | |
| 295 | # ────────────────────────────────────────────────────────────────────────────── |
| 296 | # Integration — allow-empty |
| 297 | # ────────────────────────────────────────────────────────────────────────────── |
| 298 | |
| 299 | |
| 300 | class TestAllowEmpty: |
| 301 | def test_allow_empty_creates_commit(self, repo: pathlib.Path) -> None: |
| 302 | result = _commit(repo, "-m", "empty", "--allow-empty") |
| 303 | assert result.exit_code == 0 |
| 304 | |
| 305 | def test_allow_empty_without_message_warns( |
| 306 | self, repo: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 307 | ) -> None: |
| 308 | import logging |
| 309 | |
| 310 | with caplog.at_level(logging.WARNING, logger="muse.cli.commands.commit"): |
| 311 | _commit(repo, "--allow-empty") |
| 312 | assert any( |
| 313 | "empty message" in r.message or "--allow-empty" in r.message |
| 314 | for r in caplog.records |
| 315 | ) |
| 316 | |
| 317 | def test_allow_empty_without_message_exits_0(self, repo: pathlib.Path) -> None: |
| 318 | result = _commit(repo, "--allow-empty") |
| 319 | assert result.exit_code == 0 |
| 320 | |
| 321 | def test_allow_empty_json_message_is_empty_string(self, repo: pathlib.Path) -> None: |
| 322 | result = _commit(repo, "--allow-empty", "--json") |
| 323 | data = json.loads(result.output) |
| 324 | assert data["message"] == "" |
| 325 | |
| 326 | |
| 327 | # ────────────────────────────────────────────────────────────────────────────── |
| 328 | # Integration — validation errors |
| 329 | # ────────────────────────────────────────────────────────────────────────────── |
| 330 | |
| 331 | |
| 332 | class TestValidation: |
| 333 | def test_missing_message_exits_1(self, repo: pathlib.Path) -> None: |
| 334 | result = _commit(repo) |
| 335 | assert result.exit_code == 1 |
| 336 | |
| 337 | def test_missing_message_prints_hint(self, repo: pathlib.Path) -> None: |
| 338 | result = _commit(repo) |
| 339 | assert "-m" in result.output or "message" in result.output.lower() |
| 340 | |
| 341 | def test_unknown_format_exits_1(self, repo: pathlib.Path) -> None: |
| 342 | result = _commit(repo, "-m", "x", "--format", "xml") |
| 343 | assert result.exit_code == 1 |
| 344 | |
| 345 | def test_unknown_format_sanitized_in_error(self, repo: pathlib.Path) -> None: |
| 346 | evil = "\x1b[31mred\x1b[0m" |
| 347 | result = _commit(repo, "-m", "x", "--format", evil) |
| 348 | assert "\x1b" not in result.output |
| 349 | |
| 350 | def test_empty_tree_without_allow_empty_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 351 | # Create a bare .muse structure with no tracked files at all (pre-init state). |
| 352 | # This is the only scenario where the "empty tree" guard fires, because |
| 353 | # muse init always writes .museattributes and .museignore as tracked files. |
| 354 | bare = tmp_path / "bare" |
| 355 | bare.mkdir() |
| 356 | (bare / ".muse").mkdir() |
| 357 | (bare / ".muse" / "HEAD").write_text("ref: refs/heads/main\n") |
| 358 | (bare / ".muse" / "refs").mkdir() |
| 359 | (bare / ".muse" / "refs" / "heads").mkdir() |
| 360 | (bare / ".muse" / "repo.json").write_text( |
| 361 | '{"repo_id": "' + "a" * 36 + '", "schema_version": 1, "domain": "code"}' |
| 362 | ) |
| 363 | result = _invoke(bare, ["commit", "-m", "empty tree"]) |
| 364 | # Either exits 1 (empty tree guard) or 0 (domain plugin tracks no files). |
| 365 | # The point is that it must not crash. |
| 366 | assert result.exit_code in (0, 1) |
| 367 | |
| 368 | |
| 369 | # ────────────────────────────────────────────────────────────────────────────── |
| 370 | # End-to-end — JSON output schema |
| 371 | # ────────────────────────────────────────────────────────────────────────────── |
| 372 | |
| 373 | |
| 374 | class TestJsonSchema: |
| 375 | """All keys agents depend on must be present in every JSON response.""" |
| 376 | |
| 377 | REQUIRED_KEYS = { |
| 378 | "commit_id", |
| 379 | "branch", |
| 380 | "snapshot_id", |
| 381 | "message", |
| 382 | "parent_commit_id", |
| 383 | "parent2_commit_id", |
| 384 | "committed_at", |
| 385 | "author", |
| 386 | "agent_id", |
| 387 | "sem_ver_bump", |
| 388 | "breaking_changes", |
| 389 | "files_changed", |
| 390 | "dry_run", |
| 391 | } |
| 392 | |
| 393 | def test_first_commit_json_keys(self, repo: pathlib.Path) -> None: |
| 394 | result = _commit(repo, "-m", "first", "--json") |
| 395 | assert result.exit_code == 0 |
| 396 | data = json.loads(result.output) |
| 397 | missing = self.REQUIRED_KEYS - set(data) |
| 398 | assert not missing, f"Missing keys: {missing}" |
| 399 | |
| 400 | def test_parent_commit_id_null_on_first_commit(self, repo: pathlib.Path) -> None: |
| 401 | result = _commit(repo, "-m", "first", "--json") |
| 402 | data = json.loads(result.output) |
| 403 | assert data["parent_commit_id"] is None |
| 404 | |
| 405 | def test_parent_commit_id_populated_on_second_commit(self, repo: pathlib.Path) -> None: |
| 406 | _commit(repo, "-m", "first") |
| 407 | cid1 = get_head_commit_id(repo, "main") |
| 408 | (repo / "b.py").write_text("y=2\n") |
| 409 | result = _commit(repo, "-m", "second", "--json") |
| 410 | data = json.loads(result.output) |
| 411 | assert data["parent_commit_id"] == cid1 |
| 412 | |
| 413 | def test_parent2_commit_id_null_on_regular_commit(self, repo: pathlib.Path) -> None: |
| 414 | result = _commit(repo, "-m", "first", "--json") |
| 415 | data = json.loads(result.output) |
| 416 | assert data["parent2_commit_id"] is None |
| 417 | |
| 418 | def test_breaking_changes_is_list(self, repo: pathlib.Path) -> None: |
| 419 | result = _commit(repo, "-m", "first", "--json") |
| 420 | data = json.loads(result.output) |
| 421 | assert isinstance(data["breaking_changes"], list) |
| 422 | |
| 423 | def test_sem_ver_bump_is_string(self, repo: pathlib.Path) -> None: |
| 424 | result = _commit(repo, "-m", "first", "--json") |
| 425 | data = json.loads(result.output) |
| 426 | assert isinstance(data["sem_ver_bump"], str) |
| 427 | |
| 428 | def test_agent_id_default_empty_string(self, repo: pathlib.Path) -> None: |
| 429 | result = _commit(repo, "-m", "first", "--json") |
| 430 | data = json.loads(result.output) |
| 431 | assert data["agent_id"] == "" |
| 432 | |
| 433 | def test_agent_id_from_flag(self, repo: pathlib.Path) -> None: |
| 434 | result = _commit(repo, "-m", "x", "--agent-id", "bot-42", "--json") |
| 435 | data = json.loads(result.output) |
| 436 | assert data["agent_id"] == "bot-42" |
| 437 | |
| 438 | def test_agent_id_from_env(self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None: |
| 439 | monkeypatch.setenv("MUSE_AGENT_ID", "env-bot") |
| 440 | result = _invoke(repo, ["commit", "-m", "x", "--json"]) |
| 441 | data = json.loads(result.output) |
| 442 | assert data["agent_id"] == "env-bot" |
| 443 | |
| 444 | def test_dry_run_false_on_real_commit(self, repo: pathlib.Path) -> None: |
| 445 | result = _commit(repo, "-m", "x", "--json") |
| 446 | data = json.loads(result.output) |
| 447 | assert data["dry_run"] is False |
| 448 | |
| 449 | def test_files_changed_structure(self, repo: pathlib.Path) -> None: |
| 450 | result = _commit(repo, "-m", "x", "--json") |
| 451 | data = json.loads(result.output) |
| 452 | fc = data["files_changed"] |
| 453 | assert isinstance(fc, dict) |
| 454 | assert set(fc.keys()) == {"added", "modified", "deleted"} |
| 455 | |
| 456 | def test_files_added_counted(self, repo: pathlib.Path) -> None: |
| 457 | result = _commit(repo, "-m", "x", "--json") |
| 458 | data = json.loads(result.output) |
| 459 | assert data["files_changed"]["added"] >= 1 |
| 460 | |
| 461 | def test_files_modified_counted(self, repo: pathlib.Path) -> None: |
| 462 | _commit(repo, "-m", "first") |
| 463 | (repo / "a.py").write_text("x = 99\n") |
| 464 | result = _commit(repo, "-m", "mod", "--json") |
| 465 | data = json.loads(result.output) |
| 466 | assert data["files_changed"]["modified"] == 1 |
| 467 | assert data["files_changed"]["added"] == 0 |
| 468 | |
| 469 | def test_files_deleted_counted(self, repo: pathlib.Path) -> None: |
| 470 | (repo / "del.py").write_text("z = 3\n") |
| 471 | _commit(repo, "-m", "add del.py") |
| 472 | (repo / "del.py").unlink() |
| 473 | result = _commit(repo, "-m", "remove", "--json") |
| 474 | data = json.loads(result.output) |
| 475 | assert data["files_changed"]["deleted"] == 1 |
| 476 | |
| 477 | def test_committed_at_is_utc_iso(self, repo: pathlib.Path) -> None: |
| 478 | import datetime |
| 479 | |
| 480 | result = _commit(repo, "-m", "x", "--json") |
| 481 | data = json.loads(result.output) |
| 482 | dt = datetime.datetime.fromisoformat(data["committed_at"]) |
| 483 | assert dt.tzinfo is not None |
| 484 | |
| 485 | |
| 486 | # ────────────────────────────────────────────────────────────────────────────── |
| 487 | # End-to-end — dry-run |
| 488 | # ────────────────────────────────────────────────────────────────────────────── |
| 489 | |
| 490 | |
| 491 | class TestDryRun: |
| 492 | def test_dry_run_no_commit_written(self, repo: pathlib.Path) -> None: |
| 493 | result = _commit(repo, "-m", "dr", "--dry-run") |
| 494 | assert result.exit_code == 0 |
| 495 | assert get_head_commit_id(repo, "main") is None |
| 496 | |
| 497 | def test_dry_run_json_schema(self, repo: pathlib.Path) -> None: |
| 498 | result = _commit(repo, "-m", "dr", "--dry-run", "--json") |
| 499 | assert result.exit_code == 0 |
| 500 | data = json.loads(result.output) |
| 501 | assert data["dry_run"] is True |
| 502 | assert data["clean"] is False |
| 503 | assert "commit_id" in data |
| 504 | assert "files_changed" in data |
| 505 | |
| 506 | def test_dry_run_snapshot_id_stable(self, repo: pathlib.Path) -> None: |
| 507 | """Same tree content → same snapshot_id on repeated dry-runs.""" |
| 508 | r1 = _commit(repo, "-m", "dr", "--dry-run", "--json") |
| 509 | r2 = _commit(repo, "-m", "dr", "--dry-run", "--json") |
| 510 | d1 = json.loads(r1.output) |
| 511 | d2 = json.loads(r2.output) |
| 512 | assert d1["snapshot_id"] == d2["snapshot_id"] |
| 513 | |
| 514 | def test_dry_run_clean_tree_exits_1(self, repo: pathlib.Path) -> None: |
| 515 | _commit(repo, "-m", "first") |
| 516 | result = _commit(repo, "-m", "no changes", "--dry-run") |
| 517 | assert result.exit_code == 1 |
| 518 | |
| 519 | def test_dry_run_clean_tree_json_clean_flag(self, repo: pathlib.Path) -> None: |
| 520 | _commit(repo, "-m", "first") |
| 521 | result = _commit(repo, "-m", "no changes", "--dry-run", "--json") |
| 522 | data = json.loads(result.output) |
| 523 | assert data["clean"] is True |
| 524 | |
| 525 | def test_dry_run_text_output_prefix(self, repo: pathlib.Path) -> None: |
| 526 | result = _commit(repo, "-m", "preview", "--dry-run") |
| 527 | assert "dry-run" in result.output |
| 528 | |
| 529 | def test_dry_run_text_output_nothing_written_note(self, repo: pathlib.Path) -> None: |
| 530 | result = _commit(repo, "-m", "preview", "--dry-run") |
| 531 | assert "nothing written" in result.output |
| 532 | |
| 533 | def test_dry_run_shows_sem_ver_in_json(self, repo: pathlib.Path) -> None: |
| 534 | result = _commit(repo, "-m", "dr", "--dry-run", "--json") |
| 535 | data = json.loads(result.output) |
| 536 | assert "sem_ver_bump" in data |
| 537 | |
| 538 | def test_dry_run_does_not_advance_branch(self, repo: pathlib.Path) -> None: |
| 539 | _commit(repo, "-m", "first") |
| 540 | cid_before = get_head_commit_id(repo, "main") |
| 541 | (repo / "b.py").write_text("z=9\n") |
| 542 | _commit(repo, "-m", "second", "--dry-run") |
| 543 | cid_after = get_head_commit_id(repo, "main") |
| 544 | assert cid_before == cid_after |
| 545 | |
| 546 | def test_dry_run_parent_commit_id_in_json(self, repo: pathlib.Path) -> None: |
| 547 | _commit(repo, "-m", "first") |
| 548 | cid1 = get_head_commit_id(repo, "main") |
| 549 | (repo / "b.py").write_text("z=9\n") |
| 550 | result = _commit(repo, "-m", "second", "--dry-run", "--json") |
| 551 | data = json.loads(result.output) |
| 552 | assert data["parent_commit_id"] == cid1 |
| 553 | |
| 554 | |
| 555 | # ────────────────────────────────────────────────────────────────────────────── |
| 556 | # End-to-end — text output |
| 557 | # ────────────────────────────────────────────────────────────────────────────── |
| 558 | |
| 559 | |
| 560 | class TestTextOutput: |
| 561 | def test_text_shows_branch_and_short_id(self, repo: pathlib.Path) -> None: |
| 562 | import re |
| 563 | |
| 564 | result = _commit(repo, "-m", "hello") |
| 565 | assert "main" in result.output |
| 566 | # Output format: "[main <8-hex-char>] message" |
| 567 | # Extract all 8-char hex tokens (stripping punctuation like ']'). |
| 568 | tokens = [re.sub(r"[^0-9a-f]", "", w) for w in result.output.split()] |
| 569 | assert any(len(t) == 8 for t in tokens), ( |
| 570 | f"No 8-char hex commit ID found in: {result.output!r}" |
| 571 | ) |
| 572 | |
| 573 | def test_text_shows_message(self, repo: pathlib.Path) -> None: |
| 574 | result = _commit(repo, "-m", "verse melody") |
| 575 | assert "verse melody" in result.output |
| 576 | |
| 577 | def test_text_shows_files_changed(self, repo: pathlib.Path) -> None: |
| 578 | result = _commit(repo, "-m", "x") |
| 579 | assert "file" in result.output |
| 580 | |
| 581 | def test_text_nothing_to_commit_message(self, repo: pathlib.Path) -> None: |
| 582 | _commit(repo, "-m", "first") |
| 583 | result = _commit(repo, "-m", "second") |
| 584 | assert "Nothing to commit" in result.output |
| 585 | |
| 586 | |
| 587 | # ────────────────────────────────────────────────────────────────────────────── |
| 588 | # Security — ANSI injection prevention |
| 589 | # ────────────────────────────────────────────────────────────────────────────── |
| 590 | |
| 591 | |
| 592 | class TestSecurityAnsi: |
| 593 | """Text output must never emit raw ANSI escape sequences from user input.""" |
| 594 | |
| 595 | def _has_ansi(self, s: str) -> bool: |
| 596 | return "\x1b[" in s or "\x1b]" in s |
| 597 | |
| 598 | def test_ansi_in_message_stripped_from_text_output(self, repo: pathlib.Path) -> None: |
| 599 | msg = "hello \x1b[31mred\x1b[0m world" |
| 600 | result = _commit(repo, "-m", msg) |
| 601 | assert not self._has_ansi(result.output), "ANSI in message leaked to text output" |
| 602 | |
| 603 | def test_ansi_in_format_flag_sanitized(self, repo: pathlib.Path) -> None: |
| 604 | result = _commit(repo, "-m", "x", "--format", "\x1b[31mxml\x1b[0m") |
| 605 | assert not self._has_ansi(result.output) |
| 606 | |
| 607 | def test_ansi_in_author_sanitized(self, repo: pathlib.Path) -> None: |
| 608 | result = _commit(repo, "-m", "x", "--author", "\x1b[1mevil\x1b[0m") |
| 609 | assert not self._has_ansi(result.output) |
| 610 | |
| 611 | |
| 612 | # ────────────────────────────────────────────────────────────────────────────── |
| 613 | # Security — author / provenance field caps |
| 614 | # ────────────────────────────────────────────────────────────────────────────── |
| 615 | |
| 616 | |
| 617 | class TestSecurityProvenance: |
| 618 | def test_author_capped_at_256_chars(self, repo: pathlib.Path) -> None: |
| 619 | long_author = "a" * 500 |
| 620 | _commit(repo, "-m", "x", "--author", long_author) |
| 621 | branch = read_current_branch(repo) |
| 622 | cid = get_head_commit_id(repo, branch) |
| 623 | assert cid is not None |
| 624 | rec = read_commit(repo, cid) |
| 625 | assert rec is not None |
| 626 | assert len(rec.author) <= 256 |
| 627 | |
| 628 | def test_agent_id_capped_at_256_chars(self, repo: pathlib.Path) -> None: |
| 629 | long_id = "b" * 500 |
| 630 | _commit(repo, "-m", "x", "--agent-id", long_id) |
| 631 | branch = read_current_branch(repo) |
| 632 | cid = get_head_commit_id(repo, branch) |
| 633 | assert cid is not None |
| 634 | rec = read_commit(repo, cid) |
| 635 | assert rec is not None |
| 636 | assert len(rec.agent_id) <= 256 |
| 637 | |
| 638 | def test_author_control_chars_stripped(self, repo: pathlib.Path) -> None: |
| 639 | _commit(repo, "-m", "x", "--author", "alice\x00\x01\x02") |
| 640 | branch = read_current_branch(repo) |
| 641 | cid = get_head_commit_id(repo, branch) |
| 642 | assert cid is not None |
| 643 | rec = read_commit(repo, cid) |
| 644 | assert rec is not None |
| 645 | assert "\x00" not in rec.author |
| 646 | assert "\x01" not in rec.author |
| 647 | |
| 648 | def test_author_override_emits_warning( |
| 649 | self, repo: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 650 | ) -> None: |
| 651 | import logging |
| 652 | |
| 653 | with caplog.at_level(logging.WARNING, logger="muse.cli.commands.commit"): |
| 654 | _commit(repo, "-m", "x", "--author", "evil-impersonator") |
| 655 | assert any( |
| 656 | "impersonation" in r.message or "--author" in r.message |
| 657 | for r in caplog.records |
| 658 | ) |
| 659 | |
| 660 | def test_agent_id_from_flag_overrides_env( |
| 661 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 662 | ) -> None: |
| 663 | monkeypatch.setenv("MUSE_AGENT_ID", "env-agent") |
| 664 | result = _invoke(repo, ["commit", "-m", "x", "--agent-id", "flag-agent", "--json"]) |
| 665 | data = json.loads(result.output) |
| 666 | assert data["agent_id"] == "flag-agent" |
| 667 | |
| 668 | |
| 669 | # ────────────────────────────────────────────────────────────────────────────── |
| 670 | # Integration — merge-parent recording |
| 671 | # ────────────────────────────────────────────────────────────────────────────── |
| 672 | |
| 673 | |
| 674 | class TestMergeParent: |
| 675 | """When a merge commit is created, parent2_commit_id must be set.""" |
| 676 | |
| 677 | def test_merge_commit_has_two_parents(self, repo: pathlib.Path) -> None: |
| 678 | _commit(repo, "-m", "base") |
| 679 | _invoke(repo, ["branch", "feat"]) |
| 680 | _invoke(repo, ["checkout", "feat"]) |
| 681 | (repo / "feat.py").write_text("f = 1\n") |
| 682 | _commit(repo, "-m", "feat commit") |
| 683 | _invoke(repo, ["checkout", "main"]) |
| 684 | (repo / "main_only.py").write_text("m = 1\n") |
| 685 | _commit(repo, "-m", "main commit") |
| 686 | _invoke(repo, ["merge", "feat"]) |
| 687 | cid = get_head_commit_id(repo, "main") |
| 688 | assert cid is not None |
| 689 | rec = read_commit(repo, cid) |
| 690 | assert rec is not None |
| 691 | assert rec.parent2_commit_id is not None |
| 692 | |
| 693 | def test_regular_commit_parent2_is_none(self, repo: pathlib.Path) -> None: |
| 694 | _commit(repo, "-m", "first") |
| 695 | (repo / "b.py").write_text("b=1\n") |
| 696 | _commit(repo, "-m", "second") |
| 697 | branch = read_current_branch(repo) |
| 698 | cid = get_head_commit_id(repo, branch) |
| 699 | assert cid is not None |
| 700 | rec = read_commit(repo, cid) |
| 701 | assert rec is not None |
| 702 | assert rec.parent2_commit_id is None |
| 703 | |
| 704 | |
| 705 | # ────────────────────────────────────────────────────────────────────────────── |
| 706 | # Integration — SemVer bump inference |
| 707 | # ────────────────────────────────────────────────────────────────────────────── |
| 708 | |
| 709 | |
| 710 | class TestSemVerBump: |
| 711 | def test_first_commit_sem_ver_bump_valid(self, repo: pathlib.Path) -> None: |
| 712 | _commit(repo, "-m", "init") |
| 713 | cid = get_head_commit_id(repo, "main") |
| 714 | assert cid is not None |
| 715 | rec = read_commit(repo, cid) |
| 716 | assert rec is not None |
| 717 | assert rec.sem_ver_bump in ("none", "patch", "minor", "major") |
| 718 | |
| 719 | def test_json_sem_ver_bump_is_valid_value(self, repo: pathlib.Path) -> None: |
| 720 | result = _commit(repo, "-m", "x", "--json") |
| 721 | data = json.loads(result.output) |
| 722 | assert data["sem_ver_bump"] in ("none", "patch", "minor", "major") |
| 723 | |
| 724 | def test_breaking_changes_list_in_record(self, repo: pathlib.Path) -> None: |
| 725 | _commit(repo, "-m", "first") |
| 726 | branch = read_current_branch(repo) |
| 727 | cid = get_head_commit_id(repo, branch) |
| 728 | assert cid is not None |
| 729 | rec = read_commit(repo, cid) |
| 730 | assert rec is not None |
| 731 | assert isinstance(rec.breaking_changes, list) |
| 732 | |
| 733 | |
| 734 | # ────────────────────────────────────────────────────────────────────────────── |
| 735 | # Integration — reflog |
| 736 | # ────────────────────────────────────────────────────────────────────────────── |
| 737 | |
| 738 | |
| 739 | class TestReflog: |
| 740 | def test_commit_appends_reflog_entry(self, repo: pathlib.Path) -> None: |
| 741 | from muse.core.reflog import read_reflog |
| 742 | |
| 743 | _commit(repo, "-m", "logged") |
| 744 | entries = read_reflog(repo, "main") |
| 745 | assert len(entries) >= 1 |
| 746 | assert any( |
| 747 | "logged" in e.operation or "commit" in e.operation for e in entries |
| 748 | ) |
| 749 | |
| 750 | def test_reflog_contains_commit_id(self, repo: pathlib.Path) -> None: |
| 751 | from muse.core.reflog import read_reflog |
| 752 | |
| 753 | _commit(repo, "-m", "ref-entry") |
| 754 | cid = get_head_commit_id(repo, "main") |
| 755 | entries = read_reflog(repo, "main") |
| 756 | assert any(e.new_id == cid for e in entries) |
| 757 | |
| 758 | |
| 759 | # ────────────────────────────────────────────────────────────────────────────── |
| 760 | # Integration — stage cleared after commit |
| 761 | # ────────────────────────────────────────────────────────────────────────────── |
| 762 | |
| 763 | |
| 764 | class TestStageClearAfterCommit: |
| 765 | def test_stage_is_cleared(self, repo: pathlib.Path) -> None: |
| 766 | _invoke(repo, ["code", "add", "."]) |
| 767 | _commit(repo, "-m", "staged") |
| 768 | stage_path = repo / ".muse" / "stage.json" |
| 769 | if stage_path.exists(): |
| 770 | data = json.loads(stage_path.read_text()) |
| 771 | assert data == {} or data.get("files") == {} |
| 772 | |
| 773 | |
| 774 | # ────────────────────────────────────────────────────────────────────────────── |
| 775 | # End-to-end — provenance env vars |
| 776 | # ────────────────────────────────────────────────────────────────────────────── |
| 777 | |
| 778 | |
| 779 | class TestProvenanceEnvVars: |
| 780 | def test_model_id_from_env( |
| 781 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 782 | ) -> None: |
| 783 | monkeypatch.setenv("MUSE_MODEL_ID", "gpt-5") |
| 784 | _invoke(repo, ["commit", "-m", "x"]) |
| 785 | branch = read_current_branch(repo) |
| 786 | cid = get_head_commit_id(repo, branch) |
| 787 | assert cid is not None |
| 788 | rec = read_commit(repo, cid) |
| 789 | assert rec is not None |
| 790 | assert rec.model_id == "gpt-5" |
| 791 | |
| 792 | def test_toolchain_id_from_env( |
| 793 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 794 | ) -> None: |
| 795 | monkeypatch.setenv("MUSE_TOOLCHAIN_ID", "cursor-v42") |
| 796 | _invoke(repo, ["commit", "-m", "x"]) |
| 797 | branch = read_current_branch(repo) |
| 798 | cid = get_head_commit_id(repo, branch) |
| 799 | assert cid is not None |
| 800 | rec = read_commit(repo, cid) |
| 801 | assert rec is not None |
| 802 | assert rec.toolchain_id == "cursor-v42" |
| 803 | |
| 804 | def test_prompt_hash_from_env( |
| 805 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 806 | ) -> None: |
| 807 | monkeypatch.setenv("MUSE_PROMPT_HASH", "abc123") |
| 808 | _invoke(repo, ["commit", "-m", "x"]) |
| 809 | branch = read_current_branch(repo) |
| 810 | cid = get_head_commit_id(repo, branch) |
| 811 | assert cid is not None |
| 812 | rec = read_commit(repo, cid) |
| 813 | assert rec is not None |
| 814 | assert rec.prompt_hash == "abc123" |
| 815 | |
| 816 | def test_flag_overrides_env_for_model_id( |
| 817 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 818 | ) -> None: |
| 819 | monkeypatch.setenv("MUSE_MODEL_ID", "env-model") |
| 820 | _invoke(repo, ["commit", "-m", "x", "--model-id", "flag-model"]) |
| 821 | branch = read_current_branch(repo) |
| 822 | cid = get_head_commit_id(repo, branch) |
| 823 | assert cid is not None |
| 824 | rec = read_commit(repo, cid) |
| 825 | assert rec is not None |
| 826 | assert rec.model_id == "flag-model" |
| 827 | |
| 828 | |
| 829 | # ────────────────────────────────────────────────────────────────────────────── |
| 830 | # Integration — parent manifest not double-read |
| 831 | # ────────────────────────────────────────────────────────────────────────────── |
| 832 | |
| 833 | |
| 834 | class TestParentManifestSingleRead: |
| 835 | """ |
| 836 | The parent snapshot must be loaded only once per commit, not twice. |
| 837 | We verify via call counts on read_snapshot. |
| 838 | """ |
| 839 | |
| 840 | def test_parent_snapshot_read_at_most_once(self, repo: pathlib.Path) -> None: |
| 841 | _commit(repo, "-m", "first") |
| 842 | (repo / "b.py").write_text("b=1\n") |
| 843 | call_count: list[int] = [0] |
| 844 | original_read_snapshot = read_snapshot |
| 845 | |
| 846 | from muse.core.store import SnapshotRecord |
| 847 | |
| 848 | def counting_read_snapshot( |
| 849 | root: pathlib.Path, sid: str |
| 850 | ) -> SnapshotRecord | None: |
| 851 | call_count[0] += 1 |
| 852 | return original_read_snapshot(root, sid) |
| 853 | |
| 854 | with patch( |
| 855 | "muse.cli.commands.commit.read_snapshot", |
| 856 | side_effect=counting_read_snapshot, |
| 857 | ): |
| 858 | _commit(repo, "-m", "second") |
| 859 | |
| 860 | # Should be ≤1 (one read of the parent snapshot). |
| 861 | # Previously the bug caused 2 reads: one for structured_delta, one for file counts. |
| 862 | assert call_count[0] <= 1, ( |
| 863 | f"read_snapshot called {call_count[0]} times; expected ≤1 (parent double-read bug)" |
| 864 | ) |
| 865 | |
| 866 | |
| 867 | # ────────────────────────────────────────────────────────────────────────────── |
| 868 | # Stress — sequential commits |
| 869 | # ────────────────────────────────────────────────────────────────────────────── |
| 870 | |
| 871 | |
| 872 | @pytest.mark.slow |
| 873 | class TestStressSequential: |
| 874 | def test_100_commits_all_succeed(self, repo: pathlib.Path) -> None: |
| 875 | for i in range(100): |
| 876 | (repo / f"f{i:04d}.py").write_text(f"x = {i}\n") |
| 877 | result = _commit(repo, "-m", f"commit {i}") |
| 878 | assert result.exit_code == 0, f"Commit {i} failed: {result.output}" |
| 879 | |
| 880 | def test_100_commits_branch_advances(self, repo: pathlib.Path) -> None: |
| 881 | seen_ids: set[str] = set() |
| 882 | for i in range(100): |
| 883 | (repo / f"g{i:04d}.py").write_text(f"y = {i}\n") |
| 884 | _commit(repo, "-m", f"c{i}") |
| 885 | cid = get_head_commit_id(repo, "main") |
| 886 | assert cid not in seen_ids, f"Duplicate commit ID at commit {i}" |
| 887 | if cid: |
| 888 | seen_ids.add(cid) |
| 889 | assert len(seen_ids) == 100 |
| 890 | |
| 891 | |
| 892 | @pytest.mark.slow |
| 893 | class TestStressLargeManifest: |
| 894 | def test_500_file_commit_succeeds(self, repo: pathlib.Path) -> None: |
| 895 | for i in range(500): |
| 896 | (repo / f"h{i:04d}.py").write_text(f"z = {i}\n") |
| 897 | t0 = time.perf_counter() |
| 898 | result = _commit(repo, "-m", "big") |
| 899 | elapsed = (time.perf_counter() - t0) * 1000 |
| 900 | assert result.exit_code == 0 |
| 901 | assert elapsed < 3000, f"Commit too slow: {elapsed:.0f}ms" |
| 902 | |
| 903 | def test_500_file_single_change_commit(self, repo: pathlib.Path) -> None: |
| 904 | for i in range(500): |
| 905 | (repo / f"k{i:04d}.py").write_text(f"a = {i}\n") |
| 906 | _commit(repo, "-m", "base") |
| 907 | (repo / "k0000.py").write_text("a = 999\n") |
| 908 | t0 = time.perf_counter() |
| 909 | result = _commit(repo, "-m", "one change") |
| 910 | elapsed = (time.perf_counter() - t0) * 1000 |
| 911 | assert result.exit_code == 0 |
| 912 | assert elapsed < 2000, f"Single-file commit too slow: {elapsed:.0f}ms" |
| 913 | |
| 914 | |
| 915 | # ────────────────────────────────────────────────────────────────────────────── |
| 916 | # Stress — concurrent commits to different repos |
| 917 | # ────────────────────────────────────────────────────────────────────────────── |
| 918 | |
| 919 | |
| 920 | @pytest.mark.slow |
| 921 | class TestStressConcurrent: |
| 922 | def test_concurrent_commits_to_separate_repos(self, tmp_path: pathlib.Path) -> None: |
| 923 | """16 threads each commit to their own isolated repo — no interference.""" |
| 924 | errors: list[str] = [] |
| 925 | |
| 926 | def do_commit(idx: int) -> None: |
| 927 | repo_dir = tmp_path / f"repo_{idx}" |
| 928 | repo_dir.mkdir() |
| 929 | subprocess.run( |
| 930 | ["muse", "init"], cwd=str(repo_dir), capture_output=True |
| 931 | ) |
| 932 | (repo_dir / "x.py").write_text(f"x = {idx}\n") |
| 933 | r = subprocess.run( |
| 934 | ["muse", "commit", "-m", f"c{idx}", "--json"], |
| 935 | cwd=str(repo_dir), |
| 936 | capture_output=True, |
| 937 | text=True, |
| 938 | ) |
| 939 | if r.returncode != 0: |
| 940 | errors.append(f"repo_{idx}: {r.stderr}") |
| 941 | return |
| 942 | data = json.loads(r.stdout) |
| 943 | if "commit_id" not in data: |
| 944 | errors.append(f"repo_{idx}: no commit_id in output") |
| 945 | |
| 946 | threads = [threading.Thread(target=do_commit, args=(i,)) for i in range(16)] |
| 947 | for t in threads: |
| 948 | t.start() |
| 949 | for t in threads: |
| 950 | t.join() |
| 951 | |
| 952 | assert not errors, "Concurrent commit errors:\n" + "\n".join(errors) |
File History
1 commit
sha256:4d09a52c06fbc389006963ad1e5ca6ee48c3cb72799f1a322561035b263db67d
merge conflict resolve
Human
patch
41 days ago