test_cmd_commit.py
python
sha256:4d09a52c06fbc389006963ad1e5ca6ee48c3cb72799f1a322561035b263db67d
merge conflict resolve
Human
patch
3 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, harmony. |
| 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.refs import ( |
| 27 | get_head_commit_id, |
| 28 | read_current_branch, |
| 29 | ) |
| 30 | from muse.core.commits import read_commit |
| 31 | from muse.core.snapshots import read_snapshot |
| 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 | _invoke(repo, ["code", "add", "."]) |
| 52 | return _invoke(repo, ["commit", *extra]) |
| 53 | |
| 54 | |
| 55 | def _init_repo(repo: pathlib.Path) -> InvokeResult: |
| 56 | repo.mkdir(parents=True, exist_ok=True) |
| 57 | return _invoke(repo, ["init"]) |
| 58 | |
| 59 | |
| 60 | # ────────────────────────────────────────────────────────────────────────────── |
| 61 | # Fixtures |
| 62 | # ────────────────────────────────────────────────────────────────────────────── |
| 63 | |
| 64 | |
| 65 | @pytest.fixture() |
| 66 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 67 | """Initialised repo with one tracked file ready to commit.""" |
| 68 | _init_repo(tmp_path) |
| 69 | (tmp_path / "a.py").write_text("x = 1\n") |
| 70 | return tmp_path |
| 71 | |
| 72 | |
| 73 | # ────────────────────────────────────────────────────────────────────────────── |
| 74 | # Unit — parser flags |
| 75 | # ────────────────────────────────────────────────────────────────────────────── |
| 76 | |
| 77 | |
| 78 | class TestRegisterFlags: |
| 79 | """All expected CLI flags are registered on the commit subcommand.""" |
| 80 | |
| 81 | def _parse(self, *args: str) -> argparse.Namespace: |
| 82 | from muse.cli.commands.commit import register |
| 83 | |
| 84 | p = argparse.ArgumentParser() |
| 85 | sub = p.add_subparsers() |
| 86 | register(sub) |
| 87 | return p.parse_args(["commit", *args]) |
| 88 | |
| 89 | def test_message_flag(self) -> None: |
| 90 | ns = self._parse("-m", "hello") |
| 91 | assert ns.message == "hello" |
| 92 | |
| 93 | def test_allow_empty_flag(self) -> None: |
| 94 | ns = self._parse("-m", "x", "--allow-empty") |
| 95 | assert ns.allow_empty is True |
| 96 | |
| 97 | def test_dry_run_short_flag(self) -> None: |
| 98 | ns = self._parse("-m", "x", "-n") |
| 99 | assert ns.dry_run is True |
| 100 | |
| 101 | def test_dry_run_long_flag(self) -> None: |
| 102 | ns = self._parse("-m", "x", "--dry-run") |
| 103 | assert ns.dry_run is True |
| 104 | |
| 105 | def test_json_flag(self) -> None: |
| 106 | ns = self._parse("-m", "x", "--json") |
| 107 | assert ns.json_out is True |
| 108 | |
| 109 | def test_j_shorthand(self) -> None: |
| 110 | ns = self._parse("-m", "x", "-j") |
| 111 | assert ns.json_out is True |
| 112 | |
| 113 | def test_default_json_out_is_false(self) -> None: |
| 114 | ns = self._parse("-m", "x") |
| 115 | assert ns.json_out is False |
| 116 | |
| 117 | def test_agent_id_flag(self) -> None: |
| 118 | ns = self._parse("-m", "x", "--agent-id", "bot-1") |
| 119 | assert ns.agent_id == "bot-1" |
| 120 | |
| 121 | def test_model_id_flag(self) -> None: |
| 122 | ns = self._parse("-m", "x", "--model-id", "claude-4") |
| 123 | assert ns.model_id == "claude-4" |
| 124 | |
| 125 | def test_toolchain_id_flag(self) -> None: |
| 126 | ns = self._parse("-m", "x", "--toolchain-id", "cursor-v1") |
| 127 | assert ns.toolchain_id == "cursor-v1" |
| 128 | |
| 129 | def test_section_flag(self) -> None: |
| 130 | ns = self._parse("-m", "x", "--section", "chorus") |
| 131 | assert ns.section == "chorus" |
| 132 | |
| 133 | def test_track_flag(self) -> None: |
| 134 | ns = self._parse("-m", "x", "--track", "bass") |
| 135 | assert ns.track == "bass" |
| 136 | |
| 137 | def test_emotion_flag(self) -> None: |
| 138 | ns = self._parse("-m", "x", "--emotion", "joyful") |
| 139 | assert ns.emotion == "joyful" |
| 140 | |
| 141 | def test_author_flag(self) -> None: |
| 142 | ns = self._parse("-m", "x", "--author", "alice") |
| 143 | assert ns.author == "alice" |
| 144 | |
| 145 | def test_sign_flag(self) -> None: |
| 146 | ns = self._parse("-m", "x", "--sign") |
| 147 | assert ns.sign is True |
| 148 | |
| 149 | |
| 150 | # ────────────────────────────────────────────────────────────────────────────── |
| 151 | # Unit — _MAX_FIELD_LEN constant |
| 152 | # ────────────────────────────────────────────────────────────────────────────── |
| 153 | |
| 154 | |
| 155 | class TestMaxFieldLen: |
| 156 | def test_constant_exists_and_is_256(self) -> None: |
| 157 | from muse.cli.commands.commit import _MAX_FIELD_LEN |
| 158 | |
| 159 | assert _MAX_FIELD_LEN == 256 |
| 160 | |
| 161 | def test_no_separate_max_author_constant(self) -> None: |
| 162 | import muse.cli.commands.commit as m |
| 163 | |
| 164 | assert not hasattr(m, "_MAX_AUTHOR"), "_MAX_AUTHOR should be replaced by _MAX_FIELD_LEN" |
| 165 | assert not hasattr(m, "_MAX_PROV"), "_MAX_PROV should be replaced by _MAX_FIELD_LEN" |
| 166 | |
| 167 | |
| 168 | # ────────────────────────────────────────────────────────────────────────────── |
| 169 | # Unit — dead-code removal |
| 170 | # ────────────────────────────────────────────────────────────────────────────── |
| 171 | |
| 172 | |
| 173 | class TestDeadCodeRemoved: |
| 174 | def test_read_branch_removed(self) -> None: |
| 175 | import muse.cli.commands.commit as m |
| 176 | |
| 177 | assert not hasattr(m, "_read_branch"), ( |
| 178 | "_read_branch was a dead wrapper; it should have been deleted" |
| 179 | ) |
| 180 | |
| 181 | def test_read_parent_id_removed(self) -> None: |
| 182 | import muse.cli.commands.commit as m |
| 183 | |
| 184 | assert not hasattr(m, "_read_parent_id"), ( |
| 185 | "_read_parent_id was a dead wrapper; it should have been deleted" |
| 186 | ) |
| 187 | |
| 188 | |
| 189 | # ────────────────────────────────────────────────────────────────────────────── |
| 190 | # Unit — inline imports removed |
| 191 | # ────────────────────────────────────────────────────────────────────────────── |
| 192 | |
| 193 | |
| 194 | class TestNoInlineImports: |
| 195 | def test_sign_commit_record_is_module_level_import(self) -> None: |
| 196 | import inspect |
| 197 | |
| 198 | import muse.cli.commands.commit as m |
| 199 | |
| 200 | src = inspect.getsource(m.run) |
| 201 | assert "from muse.core.provenance import sign_commit_record" not in src, ( |
| 202 | "sign_commit_record import must be at module level, not inside run()" |
| 203 | ) |
| 204 | |
| 205 | def test_no_inline_store_imports(self) -> None: |
| 206 | import inspect |
| 207 | |
| 208 | import muse.cli.commands.commit as m |
| 209 | |
| 210 | src = inspect.getsource(m.run) |
| 211 | assert "from muse.core.store import" not in src, ( |
| 212 | "store imports inside run() should be at module level" |
| 213 | ) |
| 214 | |
| 215 | |
| 216 | # ────────────────────────────────────────────────────────────────────────────── |
| 217 | # Integration — basic commit lifecycle |
| 218 | # ────────────────────────────────────────────────────────────────────────────── |
| 219 | |
| 220 | |
| 221 | class TestBasicCommit: |
| 222 | def test_first_commit_succeeds(self, repo: pathlib.Path) -> None: |
| 223 | result = _commit(repo, "-m", "init") |
| 224 | assert result.exit_code == 0 |
| 225 | assert "init" in result.output |
| 226 | |
| 227 | def test_commit_creates_commit_record(self, repo: pathlib.Path) -> None: |
| 228 | _commit(repo, "-m", "first") |
| 229 | branch = read_current_branch(repo) |
| 230 | cid = get_head_commit_id(repo, branch) |
| 231 | assert cid is not None |
| 232 | rec = read_commit(repo, cid) |
| 233 | assert rec is not None |
| 234 | assert rec.message == "first" |
| 235 | |
| 236 | def test_commit_creates_snapshot(self, repo: pathlib.Path) -> None: |
| 237 | _commit(repo, "-m", "snap") |
| 238 | branch = read_current_branch(repo) |
| 239 | cid = get_head_commit_id(repo, branch) |
| 240 | assert cid is not None |
| 241 | rec = read_commit(repo, cid) |
| 242 | assert rec is not None |
| 243 | snap = read_snapshot(repo, rec.snapshot_id) |
| 244 | assert snap is not None |
| 245 | assert len(snap.manifest) >= 1 |
| 246 | |
| 247 | def test_commit_advances_branch_ref(self, repo: pathlib.Path) -> None: |
| 248 | _commit(repo, "-m", "first") |
| 249 | cid1 = get_head_commit_id(repo, "main") |
| 250 | (repo / "b.py").write_text("y = 2\n") |
| 251 | _commit(repo, "-m", "second") |
| 252 | cid2 = get_head_commit_id(repo, "main") |
| 253 | assert cid1 != cid2 |
| 254 | |
| 255 | def test_second_commit_has_parent(self, repo: pathlib.Path) -> None: |
| 256 | _commit(repo, "-m", "first") |
| 257 | cid1 = get_head_commit_id(repo, "main") |
| 258 | (repo / "b.py").write_text("y = 2\n") |
| 259 | _commit(repo, "-m", "second") |
| 260 | cid2 = get_head_commit_id(repo, "main") |
| 261 | assert cid2 is not None |
| 262 | rec2 = read_commit(repo, cid2) |
| 263 | assert rec2 is not None |
| 264 | assert rec2.parent_commit_id == cid1 |
| 265 | |
| 266 | def test_nothing_to_commit_exits_0(self, repo: pathlib.Path) -> None: |
| 267 | _commit(repo, "-m", "first") |
| 268 | result = _commit(repo, "-m", "second") |
| 269 | assert result.exit_code == 0 |
| 270 | assert "Nothing to commit" in result.output |
| 271 | |
| 272 | def test_metadata_section_stored(self, repo: pathlib.Path) -> None: |
| 273 | _commit(repo, "-m", "chorus", "--section", "chorus") |
| 274 | branch = read_current_branch(repo) |
| 275 | cid = get_head_commit_id(repo, branch) |
| 276 | assert cid is not None |
| 277 | rec = read_commit(repo, cid) |
| 278 | assert rec is not None |
| 279 | assert rec.metadata.get("section") == "chorus" |
| 280 | |
| 281 | def test_metadata_track_stored(self, repo: pathlib.Path) -> None: |
| 282 | _commit(repo, "-m", "bass", "--track", "bass") |
| 283 | branch = read_current_branch(repo) |
| 284 | cid = get_head_commit_id(repo, branch) |
| 285 | assert cid is not None |
| 286 | rec = read_commit(repo, cid) |
| 287 | assert rec is not None |
| 288 | assert rec.metadata.get("track") == "bass" |
| 289 | |
| 290 | def test_metadata_emotion_stored(self, repo: pathlib.Path) -> None: |
| 291 | _commit(repo, "-m", "joy", "--emotion", "joyful") |
| 292 | branch = read_current_branch(repo) |
| 293 | cid = get_head_commit_id(repo, branch) |
| 294 | assert cid is not None |
| 295 | rec = read_commit(repo, cid) |
| 296 | assert rec is not None |
| 297 | assert rec.metadata.get("emotion") == "joyful" |
| 298 | |
| 299 | |
| 300 | # ────────────────────────────────────────────────────────────────────────────── |
| 301 | # Integration — allow-empty |
| 302 | # ────────────────────────────────────────────────────────────────────────────── |
| 303 | |
| 304 | |
| 305 | class TestAllowEmpty: |
| 306 | def test_allow_empty_creates_commit(self, repo: pathlib.Path) -> None: |
| 307 | result = _commit(repo, "-m", "empty", "--allow-empty") |
| 308 | assert result.exit_code == 0 |
| 309 | |
| 310 | def test_allow_empty_without_message_warns( |
| 311 | self, repo: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 312 | ) -> None: |
| 313 | import logging |
| 314 | |
| 315 | with caplog.at_level(logging.WARNING, logger="muse.cli.commands.commit"): |
| 316 | _commit(repo, "--allow-empty") |
| 317 | assert any( |
| 318 | "empty message" in r.message or "--allow-empty" in r.message |
| 319 | for r in caplog.records |
| 320 | ) |
| 321 | |
| 322 | def test_allow_empty_without_message_exits_0(self, repo: pathlib.Path) -> None: |
| 323 | result = _commit(repo, "--allow-empty") |
| 324 | assert result.exit_code == 0 |
| 325 | |
| 326 | def test_allow_empty_json_message_is_empty_string(self, repo: pathlib.Path) -> None: |
| 327 | result = _commit(repo, "--allow-empty", "--json") |
| 328 | data = json.loads(result.output) |
| 329 | assert data["message"] == "" |
| 330 | |
| 331 | |
| 332 | # ────────────────────────────────────────────────────────────────────────────── |
| 333 | # Integration — validation errors |
| 334 | # ────────────────────────────────────────────────────────────────────────────── |
| 335 | |
| 336 | |
| 337 | class TestValidation: |
| 338 | def test_missing_message_exits_1(self, repo: pathlib.Path) -> None: |
| 339 | result = _commit(repo) |
| 340 | assert result.exit_code == 1 |
| 341 | |
| 342 | def test_missing_message_prints_hint(self, repo: pathlib.Path) -> None: |
| 343 | result = _commit(repo) |
| 344 | assert "-m" in result.stderr or "message" in result.stderr.lower() |
| 345 | |
| 346 | def test_unknown_flag_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 347 | result = _commit(repo, "-m", "x", "--no-such-flag") |
| 348 | assert result.exit_code != 0 |
| 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 | muse_dir(bare).mkdir() |
| 357 | (head_path(bare)).write_text("ref: refs/heads/main\n") |
| 358 | (muse_dir(bare) / "refs").mkdir() |
| 359 | (heads_dir(bare)).mkdir() |
| 360 | (repo_json_path(bare)).write_text( |
| 361 | f'{{"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 {"added", "modified", "deleted", "total"} <= set(fc.keys()) |
| 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 sha256:X...] message" |
| 567 | # The sha256: prefix is canonical — check for it directly. |
| 568 | assert re.search(r"sha256:[0-9a-f]+", result.output), ( |
| 569 | f"No sha256:-prefixed commit ID found in: {result.output!r}" |
| 570 | ) |
| 571 | |
| 572 | def test_text_shows_message(self, repo: pathlib.Path) -> None: |
| 573 | result = _commit(repo, "-m", "verse melody") |
| 574 | assert "verse melody" in result.output |
| 575 | |
| 576 | def test_text_shows_files_changed(self, repo: pathlib.Path) -> None: |
| 577 | result = _commit(repo, "-m", "x") |
| 578 | assert "file" in result.output |
| 579 | |
| 580 | def test_text_nothing_to_commit_message(self, repo: pathlib.Path) -> None: |
| 581 | _commit(repo, "-m", "first") |
| 582 | result = _commit(repo, "-m", "second") |
| 583 | assert "Nothing to commit" in result.output |
| 584 | |
| 585 | |
| 586 | # ────────────────────────────────────────────────────────────────────────────── |
| 587 | # Security — ANSI injection prevention |
| 588 | # ────────────────────────────────────────────────────────────────────────────── |
| 589 | |
| 590 | |
| 591 | class TestSecurityAnsi: |
| 592 | """Text output must never emit raw ANSI escape sequences from user input.""" |
| 593 | |
| 594 | def _has_ansi(self, s: str) -> bool: |
| 595 | return "\x1b[" in s or "\x1b]" in s |
| 596 | |
| 597 | def test_ansi_in_message_stripped_from_text_output(self, repo: pathlib.Path) -> None: |
| 598 | msg = "hello \x1b[31mred\x1b[0m world" |
| 599 | result = _commit(repo, "-m", msg) |
| 600 | assert not self._has_ansi(result.output), "ANSI in message leaked to text output" |
| 601 | |
| 602 | def test_ansi_in_message_flag_sanitized(self, repo: pathlib.Path) -> None: |
| 603 | result = _commit(repo, "-m", "\x1b[31mmalicious\x1b[0m message") |
| 604 | assert not self._has_ansi(result.output) |
| 605 | |
| 606 | def test_ansi_in_author_sanitized(self, repo: pathlib.Path) -> None: |
| 607 | result = _commit(repo, "-m", "x", "--author", "\x1b[1mmalicious\x1b[0m") |
| 608 | assert not self._has_ansi(result.output) |
| 609 | |
| 610 | |
| 611 | # ────────────────────────────────────────────────────────────────────────────── |
| 612 | # Security — author / provenance field caps |
| 613 | # ────────────────────────────────────────────────────────────────────────────── |
| 614 | |
| 615 | |
| 616 | class TestSecurityProvenance: |
| 617 | def test_author_capped_at_256_chars(self, repo: pathlib.Path) -> None: |
| 618 | long_author = "a" * 500 |
| 619 | _commit(repo, "-m", "x", "--author", long_author) |
| 620 | branch = read_current_branch(repo) |
| 621 | cid = get_head_commit_id(repo, branch) |
| 622 | assert cid is not None |
| 623 | rec = read_commit(repo, cid) |
| 624 | assert rec is not None |
| 625 | assert len(rec.author) <= 256 |
| 626 | |
| 627 | def test_agent_id_capped_at_256_chars(self, repo: pathlib.Path) -> None: |
| 628 | long_id = "b" * 500 |
| 629 | _commit(repo, "-m", "x", "--agent-id", long_id) |
| 630 | branch = read_current_branch(repo) |
| 631 | cid = get_head_commit_id(repo, branch) |
| 632 | assert cid is not None |
| 633 | rec = read_commit(repo, cid) |
| 634 | assert rec is not None |
| 635 | assert len(rec.agent_id) <= 256 |
| 636 | |
| 637 | def test_author_control_chars_stripped(self, repo: pathlib.Path) -> None: |
| 638 | _commit(repo, "-m", "x", "--author", "alice\x00\x01\x02") |
| 639 | branch = read_current_branch(repo) |
| 640 | cid = get_head_commit_id(repo, branch) |
| 641 | assert cid is not None |
| 642 | rec = read_commit(repo, cid) |
| 643 | assert rec is not None |
| 644 | assert "\x00" not in rec.author |
| 645 | assert "\x01" not in rec.author |
| 646 | |
| 647 | def test_author_override_emits_warning( |
| 648 | self, repo: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 649 | ) -> None: |
| 650 | import logging |
| 651 | |
| 652 | with caplog.at_level(logging.WARNING, logger="muse.cli.commands.commit"): |
| 653 | _commit(repo, "-m", "x", "--author", "malicious-impersonator") |
| 654 | assert any( |
| 655 | "impersonation" in r.message or "--author" in r.message |
| 656 | for r in caplog.records |
| 657 | ) |
| 658 | |
| 659 | def test_agent_id_from_flag_overrides_env( |
| 660 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 661 | ) -> None: |
| 662 | monkeypatch.setenv("MUSE_AGENT_ID", "env-agent") |
| 663 | result = _invoke(repo, ["commit", "-m", "x", "--agent-id", "flag-agent", "--json"]) |
| 664 | data = json.loads(result.output) |
| 665 | assert data["agent_id"] == "flag-agent" |
| 666 | |
| 667 | |
| 668 | # ────────────────────────────────────────────────────────────────────────────── |
| 669 | # Integration — merge-parent recording |
| 670 | # ────────────────────────────────────────────────────────────────────────────── |
| 671 | |
| 672 | |
| 673 | class TestMergeParent: |
| 674 | """When a merge commit is created, parent2_commit_id must be set.""" |
| 675 | |
| 676 | def test_merge_commit_has_two_parents(self, repo: pathlib.Path) -> None: |
| 677 | _commit(repo, "-m", "base") |
| 678 | _invoke(repo, ["branch", "feat"]) |
| 679 | _invoke(repo, ["checkout", "feat"]) |
| 680 | (repo / "feat.py").write_text("f = 1\n") |
| 681 | _commit(repo, "-m", "feat commit") |
| 682 | _invoke(repo, ["checkout", "main"]) |
| 683 | (repo / "main_only.py").write_text("m = 1\n") |
| 684 | _commit(repo, "-m", "main commit") |
| 685 | _invoke(repo, ["merge", "feat"]) |
| 686 | cid = get_head_commit_id(repo, "main") |
| 687 | assert cid is not None |
| 688 | rec = read_commit(repo, cid) |
| 689 | assert rec is not None |
| 690 | assert rec.parent2_commit_id is not None |
| 691 | |
| 692 | def test_regular_commit_parent2_is_none(self, repo: pathlib.Path) -> None: |
| 693 | _commit(repo, "-m", "first") |
| 694 | (repo / "b.py").write_text("b=1\n") |
| 695 | _commit(repo, "-m", "second") |
| 696 | branch = read_current_branch(repo) |
| 697 | cid = get_head_commit_id(repo, branch) |
| 698 | assert cid is not None |
| 699 | rec = read_commit(repo, cid) |
| 700 | assert rec is not None |
| 701 | assert rec.parent2_commit_id is None |
| 702 | |
| 703 | |
| 704 | # ────────────────────────────────────────────────────────────────────────────── |
| 705 | # Integration — SemVer bump inference |
| 706 | # ────────────────────────────────────────────────────────────────────────────── |
| 707 | |
| 708 | |
| 709 | class TestSemVerBump: |
| 710 | def test_first_commit_sem_ver_bump_valid(self, repo: pathlib.Path) -> None: |
| 711 | _commit(repo, "-m", "init") |
| 712 | cid = get_head_commit_id(repo, "main") |
| 713 | assert cid is not None |
| 714 | rec = read_commit(repo, cid) |
| 715 | assert rec is not None |
| 716 | assert rec.sem_ver_bump in ("none", "patch", "minor", "major") |
| 717 | |
| 718 | def test_json_sem_ver_bump_is_valid_value(self, repo: pathlib.Path) -> None: |
| 719 | result = _commit(repo, "-m", "x", "--json") |
| 720 | data = json.loads(result.output) |
| 721 | assert data["sem_ver_bump"] in ("none", "patch", "minor", "major") |
| 722 | |
| 723 | def test_breaking_changes_list_in_record(self, repo: pathlib.Path) -> None: |
| 724 | _commit(repo, "-m", "first") |
| 725 | branch = read_current_branch(repo) |
| 726 | cid = get_head_commit_id(repo, branch) |
| 727 | assert cid is not None |
| 728 | rec = read_commit(repo, cid) |
| 729 | assert rec is not None |
| 730 | assert isinstance(rec.breaking_changes, list) |
| 731 | |
| 732 | |
| 733 | # ────────────────────────────────────────────────────────────────────────────── |
| 734 | # Integration — reflog |
| 735 | # ────────────────────────────────────────────────────────────────────────────── |
| 736 | |
| 737 | |
| 738 | class TestReflog: |
| 739 | def test_commit_appends_reflog_entry(self, repo: pathlib.Path) -> None: |
| 740 | from muse.core.reflog import read_reflog |
| 741 | |
| 742 | _commit(repo, "-m", "logged") |
| 743 | entries = read_reflog(repo, "main") |
| 744 | assert len(entries) >= 1 |
| 745 | assert any( |
| 746 | "logged" in e.operation or "commit" in e.operation for e in entries |
| 747 | ) |
| 748 | |
| 749 | def test_reflog_contains_commit_id(self, repo: pathlib.Path) -> None: |
| 750 | from muse.core.reflog import read_reflog |
| 751 | |
| 752 | _commit(repo, "-m", "ref-entry") |
| 753 | cid = get_head_commit_id(repo, "main") |
| 754 | entries = read_reflog(repo, "main") |
| 755 | assert any(e.new_id == cid for e in entries) |
| 756 | |
| 757 | |
| 758 | # ────────────────────────────────────────────────────────────────────────────── |
| 759 | # Integration — stage cleared after commit |
| 760 | # ────────────────────────────────────────────────────────────────────────────── |
| 761 | |
| 762 | |
| 763 | class TestStageClearAfterCommit: |
| 764 | def test_stage_is_cleared(self, repo: pathlib.Path) -> None: |
| 765 | _invoke(repo, ["code", "add", "."]) |
| 766 | _commit(repo, "-m", "staged") |
| 767 | stage_path = muse_dir(repo) / "stage.json" |
| 768 | if stage_path.exists(): |
| 769 | data = json.loads(stage_path.read_text()) |
| 770 | assert data == {} or data.get("files") == {} |
| 771 | |
| 772 | |
| 773 | # ────────────────────────────────────────────────────────────────────────────── |
| 774 | # End-to-end — provenance env vars |
| 775 | # ────────────────────────────────────────────────────────────────────────────── |
| 776 | |
| 777 | |
| 778 | class TestProvenanceEnvVars: |
| 779 | def test_model_id_from_env( |
| 780 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 781 | ) -> None: |
| 782 | monkeypatch.setenv("MUSE_MODEL_ID", "gpt-5") |
| 783 | _invoke(repo, ["commit", "-m", "x"]) |
| 784 | branch = read_current_branch(repo) |
| 785 | cid = get_head_commit_id(repo, branch) |
| 786 | assert cid is not None |
| 787 | rec = read_commit(repo, cid) |
| 788 | assert rec is not None |
| 789 | assert rec.model_id == "gpt-5" |
| 790 | |
| 791 | def test_toolchain_id_from_env( |
| 792 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 793 | ) -> None: |
| 794 | monkeypatch.setenv("MUSE_TOOLCHAIN_ID", "cursor-v42") |
| 795 | _invoke(repo, ["commit", "-m", "x"]) |
| 796 | branch = read_current_branch(repo) |
| 797 | cid = get_head_commit_id(repo, branch) |
| 798 | assert cid is not None |
| 799 | rec = read_commit(repo, cid) |
| 800 | assert rec is not None |
| 801 | assert rec.toolchain_id == "cursor-v42" |
| 802 | |
| 803 | def test_prompt_hash_bare_hex_gets_prefixed( |
| 804 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 805 | ) -> None: |
| 806 | bare = "a" * 64 |
| 807 | monkeypatch.setenv("MUSE_PROMPT_HASH", bare) |
| 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 == f"sha256:{bare}" |
| 815 | |
| 816 | def test_prompt_hash_already_prefixed_unchanged( |
| 817 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 818 | ) -> None: |
| 819 | prefixed = f"sha256:{'b' * 64}" |
| 820 | monkeypatch.setenv("MUSE_PROMPT_HASH", prefixed) |
| 821 | _invoke(repo, ["commit", "-m", "x"]) |
| 822 | branch = read_current_branch(repo) |
| 823 | cid = get_head_commit_id(repo, branch) |
| 824 | assert cid is not None |
| 825 | rec = read_commit(repo, cid) |
| 826 | assert rec is not None |
| 827 | assert rec.prompt_hash == prefixed |
| 828 | |
| 829 | def test_prompt_hash_invalid_not_stored( |
| 830 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 831 | ) -> None: |
| 832 | monkeypatch.setenv("MUSE_PROMPT_HASH", "abc123") |
| 833 | _invoke(repo, ["commit", "-m", "x"]) |
| 834 | branch = read_current_branch(repo) |
| 835 | cid = get_head_commit_id(repo, branch) |
| 836 | assert cid is not None |
| 837 | rec = read_commit(repo, cid) |
| 838 | assert rec is not None |
| 839 | assert rec.prompt_hash == "" |
| 840 | |
| 841 | def test_flag_overrides_env_for_model_id( |
| 842 | self, repo: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 843 | ) -> None: |
| 844 | monkeypatch.setenv("MUSE_MODEL_ID", "env-model") |
| 845 | _invoke(repo, ["commit", "-m", "x", "--model-id", "flag-model"]) |
| 846 | branch = read_current_branch(repo) |
| 847 | cid = get_head_commit_id(repo, branch) |
| 848 | assert cid is not None |
| 849 | rec = read_commit(repo, cid) |
| 850 | assert rec is not None |
| 851 | assert rec.model_id == "flag-model" |
| 852 | |
| 853 | |
| 854 | # ────────────────────────────────────────────────────────────────────────────── |
| 855 | # Integration — parent manifest not double-read |
| 856 | # ────────────────────────────────────────────────────────────────────────────── |
| 857 | |
| 858 | |
| 859 | class TestParentManifestSingleRead: |
| 860 | """ |
| 861 | The parent snapshot must be loaded only once per commit, not twice. |
| 862 | We verify via call counts on read_snapshot. |
| 863 | """ |
| 864 | |
| 865 | def test_parent_snapshot_read_at_most_once(self, repo: pathlib.Path) -> None: |
| 866 | _commit(repo, "-m", "first") |
| 867 | (repo / "b.py").write_text("b=1\n") |
| 868 | call_count: list[int] = [0] |
| 869 | original_read_snapshot = read_snapshot |
| 870 | |
| 871 | from muse.core.snapshots import SnapshotRecord |
| 872 | |
| 873 | def counting_read_snapshot( |
| 874 | root: pathlib.Path, sid: str |
| 875 | ) -> SnapshotRecord | None: |
| 876 | call_count[0] += 1 |
| 877 | return original_read_snapshot(root, sid) |
| 878 | |
| 879 | with patch( |
| 880 | "muse.cli.commands.commit.read_snapshot", |
| 881 | side_effect=counting_read_snapshot, |
| 882 | ): |
| 883 | _commit(repo, "-m", "second") |
| 884 | |
| 885 | # Should be ≤1 (one read of the parent snapshot). |
| 886 | # Previously the bug caused 2 reads: one for structured_delta, one for file counts. |
| 887 | assert call_count[0] <= 1, ( |
| 888 | f"read_snapshot called {call_count[0]} times; expected ≤1 (parent double-read bug)" |
| 889 | ) |
| 890 | |
| 891 | |
| 892 | # ────────────────────────────────────────────────────────────────────────────── |
| 893 | # Stress — sequential commits |
| 894 | # ────────────────────────────────────────────────────────────────────────────── |
| 895 | |
| 896 | |
| 897 | @pytest.mark.slow |
| 898 | class TestStressSequential: |
| 899 | def test_100_commits_all_succeed(self, repo: pathlib.Path) -> None: |
| 900 | for i in range(100): |
| 901 | (repo / f"f{i:04d}.py").write_text(f"x = {i}\n") |
| 902 | result = _commit(repo, "-m", f"commit {i}") |
| 903 | assert result.exit_code == 0, f"Commit {i} failed: {result.output}" |
| 904 | |
| 905 | def test_100_commits_branch_advances(self, repo: pathlib.Path) -> None: |
| 906 | seen_ids: set[str] = set() |
| 907 | for i in range(100): |
| 908 | (repo / f"g{i:04d}.py").write_text(f"y = {i}\n") |
| 909 | _commit(repo, "-m", f"c{i}") |
| 910 | cid = get_head_commit_id(repo, "main") |
| 911 | assert cid not in seen_ids, f"Duplicate commit ID at commit {i}" |
| 912 | if cid: |
| 913 | seen_ids.add(cid) |
| 914 | assert len(seen_ids) == 100 |
| 915 | |
| 916 | |
| 917 | @pytest.mark.slow |
| 918 | class TestStressLargeManifest: |
| 919 | def test_500_file_commit_succeeds(self, repo: pathlib.Path) -> None: |
| 920 | for i in range(500): |
| 921 | (repo / f"h{i:04d}.py").write_text(f"z = {i}\n") |
| 922 | t0 = time.perf_counter() |
| 923 | result = _commit(repo, "-m", "big") |
| 924 | elapsed = (time.perf_counter() - t0) * 1000 |
| 925 | assert result.exit_code == 0 |
| 926 | assert elapsed < 3000, f"Commit too slow: {elapsed:.0f}ms" |
| 927 | |
| 928 | def test_500_file_single_change_commit(self, repo: pathlib.Path) -> None: |
| 929 | for i in range(500): |
| 930 | (repo / f"k{i:04d}.py").write_text(f"a = {i}\n") |
| 931 | _commit(repo, "-m", "base") |
| 932 | (repo / "k0000.py").write_text("a = 999\n") |
| 933 | t0 = time.perf_counter() |
| 934 | result = _commit(repo, "-m", "one change") |
| 935 | elapsed = (time.perf_counter() - t0) * 1000 |
| 936 | assert result.exit_code == 0 |
| 937 | assert elapsed < 2000, f"Single-file commit too slow: {elapsed:.0f}ms" |
| 938 | |
| 939 | |
| 940 | # ────────────────────────────────────────────────────────────────────────────── |
| 941 | # Stress — concurrent commits to different repos |
| 942 | # ────────────────────────────────────────────────────────────────────────────── |
| 943 | |
| 944 | |
| 945 | @pytest.mark.slow |
| 946 | class TestStressConcurrent: |
| 947 | def test_concurrent_commits_to_separate_repos(self, tmp_path: pathlib.Path) -> None: |
| 948 | """16 threads each commit to their own isolated repo — no interference.""" |
| 949 | errors: list[str] = [] |
| 950 | |
| 951 | def do_commit(idx: int) -> None: |
| 952 | repo_dir = tmp_path / f"repo_{idx}" |
| 953 | repo_dir.mkdir() |
| 954 | subprocess.run( |
| 955 | ["muse", "init"], cwd=str(repo_dir), capture_output=True |
| 956 | ) |
| 957 | (repo_dir / "x.py").write_text(f"x = {idx}\n") |
| 958 | r = subprocess.run( |
| 959 | ["muse", "commit", "-m", f"c{idx}", "--json"], |
| 960 | cwd=str(repo_dir), |
| 961 | capture_output=True, |
| 962 | text=True, |
| 963 | ) |
| 964 | if r.returncode != 0: |
| 965 | errors.append(f"repo_{idx}: {r.stderr}") |
| 966 | return |
| 967 | data = json.loads(r.stdout) |
| 968 | if "commit_id" not in data: |
| 969 | errors.append(f"repo_{idx}: no commit_id in output") |
| 970 | |
| 971 | threads = [threading.Thread(target=do_commit, args=(i,)) for i in range(16)] |
| 972 | for t in threads: |
| 973 | t.start() |
| 974 | for t in threads: |
| 975 | t.join() |
| 976 | |
| 977 | |
| 978 | # --------------------------------------------------------------------------- |
| 979 | # commit must NOT touch the working tree (Git-compatible behaviour) |
| 980 | # |
| 981 | # muse commit writes to the object store and advances the branch ref. |
| 982 | # It must never call apply_manifest — that belongs only in checkout/merge/pull. |
| 983 | # Unstaged changes on disk survive a commit unchanged. |
| 984 | # See tests/test_commit_workdir_preservation.py for the full regression suite. |
| 985 | # --------------------------------------------------------------------------- |
| 986 | |
| 987 | |
| 988 | # ────────────────────────────────────────────────────────────────────────────── |
| 989 | # Bug: commit refuses when only staged deletions remain (empty snapshot) |
| 990 | # ────────────────────────────────────────────────────────────────────────────── |
| 991 | |
| 992 | |
| 993 | class TestCommitAllDeletions: |
| 994 | """muse commit must succeed when the only staged changes are deletions. |
| 995 | |
| 996 | The previous bug: plugin.snapshot() returns an empty manifest when all |
| 997 | on-disk files are gone, and the guard ``if not manifest and not allow_empty`` |
| 998 | fired — refusing the commit with "nothing tracked". But staged deletions |
| 999 | ARE meaningful changes; the snapshot is intentionally empty. |
| 1000 | """ |
| 1001 | |
| 1002 | def _committed_repo(self, tmp_path: pathlib.Path) -> pathlib.Path: |
| 1003 | """Init repo, add files, make a first commit. Returns the repo path.""" |
| 1004 | _init_repo(tmp_path) |
| 1005 | (tmp_path / "a.txt").write_text("alpha\n") |
| 1006 | (tmp_path / "b.txt").write_text("beta\n") |
| 1007 | _invoke(tmp_path, ["code", "add", "."]) |
| 1008 | _commit(tmp_path, "-m", "initial") |
| 1009 | return tmp_path |
| 1010 | |
| 1011 | def test_commit_after_rm_all_succeeds(self, tmp_path: pathlib.Path) -> None: |
| 1012 | """muse commit must exit 0 after muse rm removes all tracked files.""" |
| 1013 | repo = self._committed_repo(tmp_path) |
| 1014 | _invoke(repo, ["rm", "a.txt"]) |
| 1015 | _invoke(repo, ["rm", "b.txt"]) |
| 1016 | result = _commit(repo, "-m", "remove everything") |
| 1017 | assert result.exit_code == 0, result.output |
| 1018 | |
| 1019 | def test_commit_after_rm_all_creates_second_commit(self, tmp_path: pathlib.Path) -> None: |
| 1020 | repo = self._committed_repo(tmp_path) |
| 1021 | _invoke(repo, ["rm", "a.txt"]) |
| 1022 | _invoke(repo, ["rm", "b.txt"]) |
| 1023 | _commit(repo, "-m", "remove everything") |
| 1024 | branch = read_current_branch(repo) |
| 1025 | commit_id = get_head_commit_id(repo, branch) |
| 1026 | assert commit_id is not None |
| 1027 | commit = read_commit(repo, commit_id) |
| 1028 | assert commit is not None |
| 1029 | assert commit.message == "remove everything" |
| 1030 | |
| 1031 | def test_commit_after_rm_all_snapshot_is_empty(self, tmp_path: pathlib.Path) -> None: |
| 1032 | """The snapshot produced by an all-deletions commit must be empty.""" |
| 1033 | repo = self._committed_repo(tmp_path) |
| 1034 | _invoke(repo, ["rm", "a.txt"]) |
| 1035 | _invoke(repo, ["rm", "b.txt"]) |
| 1036 | _commit(repo, "-m", "remove everything") |
| 1037 | branch = read_current_branch(repo) |
| 1038 | commit_id = get_head_commit_id(repo, branch) |
| 1039 | commit = read_commit(repo, commit_id) |
| 1040 | snap = read_snapshot(repo, commit.snapshot_id) |
| 1041 | assert snap is not None |
| 1042 | # muse init creates .museattributes and .museignore — these are tracked |
| 1043 | # alongside user files and remain in the snapshot after user files are removed. |
| 1044 | assert "a.txt" not in snap.manifest |
| 1045 | assert "b.txt" not in snap.manifest |
| 1046 | |
| 1047 | def test_commit_after_rm_one_file_leaves_one_in_snapshot( |
| 1048 | self, tmp_path: pathlib.Path |
| 1049 | ) -> None: |
| 1050 | """Removing one of two files produces a one-entry snapshot.""" |
| 1051 | repo = self._committed_repo(tmp_path) |
| 1052 | _invoke(repo, ["rm", "a.txt"]) |
| 1053 | _commit(repo, "-m", "remove a.txt") |
| 1054 | branch = read_current_branch(repo) |
| 1055 | commit_id = get_head_commit_id(repo, branch) |
| 1056 | commit = read_commit(repo, commit_id) |
| 1057 | snap = read_snapshot(repo, commit.snapshot_id) |
| 1058 | assert snap is not None |
| 1059 | assert "a.txt" not in snap.manifest |
| 1060 | assert "b.txt" in snap.manifest |
| 1061 | |
| 1062 | def test_json_output_on_all_deletions_commit(self, tmp_path: pathlib.Path) -> None: |
| 1063 | """--json output must be valid and show exit_code 0 for an all-deletions commit.""" |
| 1064 | repo = self._committed_repo(tmp_path) |
| 1065 | _invoke(repo, ["rm", "a.txt"]) |
| 1066 | _invoke(repo, ["rm", "b.txt"]) |
| 1067 | result = _commit(repo, "-m", "rm all", "--json") |
| 1068 | assert result.exit_code == 0, result.output |
| 1069 | data = json.loads(result.output) |
| 1070 | assert data.get("exit_code", data.get("code", 0)) == 0 or "commit_id" in data |
| 1071 | |
| 1072 | def test_status_clean_after_all_deletions_commit(self, tmp_path: pathlib.Path) -> None: |
| 1073 | """After committing all deletions, muse status must show clean=True.""" |
| 1074 | repo = self._committed_repo(tmp_path) |
| 1075 | _invoke(repo, ["rm", "a.txt"]) |
| 1076 | _invoke(repo, ["rm", "b.txt"]) |
| 1077 | result = _commit(repo, "-m", "remove everything") |
| 1078 | assert result.exit_code == 0, result.output |
| 1079 | status = _invoke(repo, ["status", "--json"]) |
| 1080 | data = json.loads(status.output) |
| 1081 | assert data["clean"] is True |
| 1082 | assert data["staged"]["deleted"] == [] |
| 1083 | |
| 1084 | def test_recursive_rm_then_commit_succeeds(self, tmp_path: pathlib.Path) -> None: |
| 1085 | """muse rm -r <dir> then commit must succeed even if all files were in that dir.""" |
| 1086 | _init_repo(tmp_path) |
| 1087 | (tmp_path / "src").mkdir() |
| 1088 | (tmp_path / "src" / "main.py").write_text("main()\n") |
| 1089 | (tmp_path / "src" / "utils.py").write_text("pass\n") |
| 1090 | _invoke(tmp_path, ["code", "add", "."]) |
| 1091 | _commit(tmp_path, "-m", "initial") |
| 1092 | _invoke(tmp_path, ["rm", "-r", "src"]) |
| 1093 | result = _commit(tmp_path, "-m", "remove src/") |
| 1094 | assert result.exit_code == 0, result.output |
| 1095 | |
| 1096 | def test_dry_run_with_all_deletions_staged(self, tmp_path: pathlib.Path) -> None: |
| 1097 | """--dry-run must exit 0 (changes pending) when deletions are staged.""" |
| 1098 | repo = self._committed_repo(tmp_path) |
| 1099 | _invoke(repo, ["rm", "a.txt"]) |
| 1100 | _invoke(repo, ["rm", "b.txt"]) |
| 1101 | result = _commit(repo, "-m", "rm all", "--dry-run") |
| 1102 | assert result.exit_code == 0, result.output |
| 1103 | |
| 1104 | def test_cached_rm_file_stays_on_disk_after_commit(self, tmp_path: pathlib.Path) -> None: |
| 1105 | """muse rm --cached keeps the file on disk; after commit it is untracked.""" |
| 1106 | repo = self._committed_repo(tmp_path) |
| 1107 | # Stage deletion of a.txt but keep it on disk; delete b.txt from disk too. |
| 1108 | _invoke(repo, ["rm", "--cached", "a.txt"]) |
| 1109 | _invoke(repo, ["rm", "b.txt"]) |
| 1110 | result = _commit(repo, "-m", "untrack a.txt, delete b.txt") |
| 1111 | assert result.exit_code == 0, result.output |
| 1112 | # a.txt must still exist on disk (it was --cached) |
| 1113 | assert (repo / "a.txt").exists(), "a.txt should remain on disk after --cached rm" |
| 1114 | # b.txt was deleted from disk by muse rm |
| 1115 | assert not (repo / "b.txt").exists() |
| 1116 | # a.txt is now untracked |
| 1117 | status = json.loads(_invoke(repo, ["status", "--json"]).output) |
| 1118 | assert "a.txt" in status["untracked"] |
| 1119 | |
| 1120 | def test_all_cached_rm_then_commit_leaves_files_on_disk( |
| 1121 | self, tmp_path: pathlib.Path |
| 1122 | ) -> None: |
| 1123 | """All files removed with --cached must survive on disk after commit.""" |
| 1124 | repo = self._committed_repo(tmp_path) |
| 1125 | _invoke(repo, ["rm", "--cached", "a.txt"]) |
| 1126 | _invoke(repo, ["rm", "--cached", "b.txt"]) |
| 1127 | result = _commit(repo, "-m", "untrack everything") |
| 1128 | assert result.exit_code == 0, result.output |
| 1129 | assert (repo / "a.txt").exists(), "a.txt must stay on disk" |
| 1130 | assert (repo / "b.txt").exists(), "b.txt must stay on disk" |
| 1131 | status = json.loads(_invoke(repo, ["status", "--json"]).output) |
| 1132 | # Untracked files make the repo dirty (mirrors git behaviour) |
| 1133 | assert status["clean"] is False |
| 1134 | assert "a.txt" in status["untracked"] |
| 1135 | assert "b.txt" in status["untracked"] |
| 1136 | |
| 1137 | |
| 1138 | # --------------------------------------------------------------------------- |
| 1139 | # Flag registration tests |
| 1140 | # --------------------------------------------------------------------------- |
| 1141 | |
| 1142 | import argparse as _argparse |
| 1143 | from muse.cli.commands.commit import register as _register_commit |
| 1144 | from muse.core.paths import head_path, heads_dir, muse_dir, repo_json_path |
| 1145 | |
| 1146 | |
| 1147 | def _parse_commit(*args: str) -> _argparse.Namespace: |
| 1148 | root_p = _argparse.ArgumentParser() |
| 1149 | subs = root_p.add_subparsers(dest="cmd") |
| 1150 | _register_commit(subs) |
| 1151 | return root_p.parse_args(["commit", *args]) |
| 1152 | |
| 1153 | |
| 1154 | class TestRegisterFlags: |
| 1155 | def test_default_json_out_is_false(self) -> None: |
| 1156 | ns = _parse_commit("-m", "msg") |
| 1157 | assert ns.json_out is False |
| 1158 | |
| 1159 | def test_json_flag_sets_json_out(self) -> None: |
| 1160 | ns = _parse_commit("-m", "msg", "--json") |
| 1161 | assert ns.json_out is True |
| 1162 | |
| 1163 | def test_j_shorthand_sets_json_out(self) -> None: |
| 1164 | ns = _parse_commit("-m", "msg", "-j") |
| 1165 | assert ns.json_out is True |
| 1166 | |
| 1167 | def test_m_shorthand_for_message(self) -> None: |
| 1168 | ns = _parse_commit("-m", "hello") |
| 1169 | assert ns.message == "hello" |
| 1170 | |
| 1171 | |
| 1172 | # --------------------------------------------------------------------------- |
| 1173 | # Genesis commit — structured_delta must be populated (TDD) |
| 1174 | # --------------------------------------------------------------------------- |
| 1175 | |
| 1176 | |
| 1177 | class TestGenesisStructuredDelta: |
| 1178 | """The very first commit (no parent) must produce a structured_delta with |
| 1179 | insert ops for every tracked symbol, so indexers can record the birth op |
| 1180 | as ``add`` rather than ``modify``. |
| 1181 | |
| 1182 | Prior to the fix, ``structured_delta`` was ``None`` for genesis commits |
| 1183 | because the diff path was guarded by ``if parent_id is not None``. |
| 1184 | """ |
| 1185 | |
| 1186 | def test_genesis_commit_has_structured_delta(self, repo: pathlib.Path) -> None: |
| 1187 | """structured_delta must not be None on the first commit.""" |
| 1188 | _invoke(repo, ["code", "add", "."]) |
| 1189 | result = _commit(repo, "-m", "init: genesis", "--json") |
| 1190 | assert result.exit_code == 0, result.output |
| 1191 | data = json.loads(result.output) |
| 1192 | cid = data.get("commit_id") |
| 1193 | assert cid is not None |
| 1194 | rec = read_commit(repo, cid) |
| 1195 | assert rec is not None |
| 1196 | assert rec.structured_delta is not None, ( |
| 1197 | "Genesis commit must carry a structured_delta so indexers can " |
| 1198 | "record symbol births as op=add" |
| 1199 | ) |
| 1200 | |
| 1201 | def test_genesis_structured_delta_has_insert_ops(self, repo: pathlib.Path) -> None: |
| 1202 | """All symbols in a genesis commit must have op=insert (not replace).""" |
| 1203 | _invoke(repo, ["code", "add", "."]) |
| 1204 | result = _commit(repo, "-m", "init", "--json") |
| 1205 | data = json.loads(result.output) |
| 1206 | cid = data["commit_id"] |
| 1207 | rec = read_commit(repo, cid) |
| 1208 | assert rec is not None |
| 1209 | delta = rec.structured_delta |
| 1210 | assert delta is not None |
| 1211 | ops = delta.get("ops", []) |
| 1212 | assert ops, "Genesis delta must have at least one op" |
| 1213 | op_types = {op.get("op") for op in ops} |
| 1214 | assert "insert" in op_types, ( |
| 1215 | f"Expected insert ops in genesis delta; got: {op_types}" |
| 1216 | ) |
| 1217 | assert "replace" not in op_types, ( |
| 1218 | f"Genesis delta must not contain replace ops; got: {op_types}" |
| 1219 | ) |
| 1220 | |
| 1221 | def test_genesis_structured_delta_content_ids_present( |
| 1222 | self, repo: pathlib.Path |
| 1223 | ) -> None: |
| 1224 | """Each insert op in the genesis delta must carry a new_content_id.""" |
| 1225 | _invoke(repo, ["code", "add", "."]) |
| 1226 | result = _commit(repo, "-m", "init", "--json") |
| 1227 | data = json.loads(result.output) |
| 1228 | cid = data["commit_id"] |
| 1229 | rec = read_commit(repo, cid) |
| 1230 | assert rec is not None |
| 1231 | delta = rec.structured_delta |
| 1232 | assert delta is not None |
| 1233 | ops = delta.get("ops", []) |
| 1234 | for op in ops: |
| 1235 | if op.get("op") == "insert": |
| 1236 | assert op.get("new_content_id") or op.get("content_id"), ( |
| 1237 | f"Insert op missing content id: {op}" |
| 1238 | ) |
File History
1 commit
sha256:4d09a52c06fbc389006963ad1e5ca6ee48c3cb72799f1a322561035b263db67d
merge conflict resolve
Human
patch
3 days ago