test_cmd_shortlog_hardening.py
python
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
| 1 | """Hardening test suite for ``muse shortlog``. |
| 2 | |
| 3 | Coverage: |
| 4 | - Unit: _branch_names (symlink guard), _group_key (all four modes), |
| 5 | _build_groups (email flag, dedup), _parse_date (valid + invalid) |
| 6 | - Security: ANSI in author/message sanitized in text, raw in JSON; |
| 7 | symlink inside refs/heads is skipped |
| 8 | - Error routing: all user errors go to stderr |
| 9 | - JSON schema: _ShortlogJson shape (repo_id, branch, groups), all fields |
| 10 | - New flags: --group-by (agent, model, branch), --summary, --no-merges, |
| 11 | --since, --until, combined filters |
| 12 | - --json: empty, single group, multi-group, provenance fields |
| 13 | - Integration: --all branches with dedup, --limit early-exit, date range |
| 14 | - E2E: help output, combined flags |
| 15 | - Stress: 500 commits × 5 authors, 50-branch repo, concurrent reads |
| 16 | """ |
| 17 | |
| 18 | from __future__ import annotations |
| 19 | from collections.abc import Mapping |
| 20 | |
| 21 | import datetime |
| 22 | import json |
| 23 | import os |
| 24 | import pathlib |
| 25 | from typing import TypedDict |
| 26 | from unittest.mock import patch |
| 27 | |
| 28 | import pytest |
| 29 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 30 | |
| 31 | from muse.cli.commands.shortlog import _branch_names, _build_groups, _group_key, _parse_date |
| 32 | from muse.core.object_store import write_object |
| 33 | from muse.core.ids import hash_commit, hash_snapshot |
| 34 | from muse.core.commits import ( |
| 35 | CommitRecord, |
| 36 | write_commit, |
| 37 | ) |
| 38 | from muse.core.snapshots import ( |
| 39 | SnapshotRecord, |
| 40 | write_snapshot, |
| 41 | ) |
| 42 | from muse.core.types import Manifest, blob_id |
| 43 | from muse.core.paths import heads_dir, muse_dir |
| 44 | |
| 45 | runner = CliRunner() |
| 46 | _REPO_ID = "shortlog-hard-test" |
| 47 | |
| 48 | # Tracks the latest commit_id per (str(root), branch) so _make_commit |
| 49 | # can auto-chain without callers needing to pass parent_id explicitly. |
| 50 | _branch_heads_map: Manifest = {} |
| 51 | |
| 52 | |
| 53 | # --------------------------------------------------------------------------- |
| 54 | # Helpers |
| 55 | # --------------------------------------------------------------------------- |
| 56 | |
| 57 | |
| 58 | |
| 59 | |
| 60 | def _init_repo(path: pathlib.Path, *, domain: str = "code") -> pathlib.Path: |
| 61 | muse = muse_dir(path) |
| 62 | for sub in ("commits", "snapshots", "objects", "refs/heads"): |
| 63 | (muse / sub).mkdir(parents=True, exist_ok=True) |
| 64 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 65 | (muse / "repo.json").write_text( |
| 66 | json.dumps({"repo_id": _REPO_ID, "domain": domain}), |
| 67 | encoding="utf-8", |
| 68 | ) |
| 69 | return path |
| 70 | |
| 71 | |
| 72 | _commit_counter = 0 |
| 73 | |
| 74 | |
| 75 | def _make_commit( |
| 76 | root: pathlib.Path, |
| 77 | *, |
| 78 | author: str = "Alice", |
| 79 | agent_id: str | None = None, |
| 80 | model_id: str | None = None, |
| 81 | branch: str = "main", |
| 82 | parent_id: str | None = None, |
| 83 | parent2_id: str | None = None, |
| 84 | committed_at: datetime.datetime | None = None, |
| 85 | ) -> str: |
| 86 | """Create and store a commit, auto-chaining to the previous on the same branch.""" |
| 87 | global _commit_counter |
| 88 | _commit_counter += 1 |
| 89 | content = f"c{_commit_counter}".encode() |
| 90 | obj_id = blob_id(content) |
| 91 | write_object(root, obj_id, content) |
| 92 | manifest = {f"f{_commit_counter}.txt": obj_id} |
| 93 | snap_id = hash_snapshot(manifest) |
| 94 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 95 | ts = committed_at or datetime.datetime.now(datetime.timezone.utc) |
| 96 | |
| 97 | # Auto-chain: if caller didn't provide parent_id, use the last known head. |
| 98 | effective_parent = parent_id |
| 99 | if effective_parent is None: |
| 100 | effective_parent = _branch_heads_map.get(f"{root}:{branch}") |
| 101 | |
| 102 | pids = [pid for pid in (effective_parent, parent2_id) if pid is not None] |
| 103 | commit_id = hash_commit( |
| 104 | parent_ids=pids, |
| 105 | snapshot_id=snap_id, |
| 106 | message=f"msg {_commit_counter}", |
| 107 | committed_at_iso=ts.isoformat(), |
| 108 | author=author, |
| 109 | ) |
| 110 | rec = CommitRecord( |
| 111 | commit_id=commit_id, |
| 112 | branch=branch, |
| 113 | snapshot_id=snap_id, |
| 114 | message=f"msg {_commit_counter}", |
| 115 | committed_at=ts, |
| 116 | parent_commit_id=effective_parent, |
| 117 | parent2_commit_id=parent2_id, |
| 118 | author=author, |
| 119 | agent_id=agent_id or "", |
| 120 | model_id=model_id or "", |
| 121 | ) |
| 122 | write_commit(root, rec) |
| 123 | ref_dir = heads_dir(root) |
| 124 | ref_file = ref_dir / branch |
| 125 | ref_file.parent.mkdir(parents=True, exist_ok=True) |
| 126 | ref_file.write_text(commit_id, encoding="utf-8") |
| 127 | _branch_heads_map[f"{root}:{branch}"] = commit_id |
| 128 | return commit_id |
| 129 | |
| 130 | |
| 131 | def _env(repo: pathlib.Path) -> Manifest: |
| 132 | return {"MUSE_REPO_ROOT": str(repo)} |
| 133 | |
| 134 | |
| 135 | def _invoke(args: list[str], env: Manifest) -> InvokeResult: |
| 136 | return runner.invoke(None, args, env=env) |
| 137 | |
| 138 | |
| 139 | class _GroupOut(TypedDict): |
| 140 | key: str |
| 141 | count: int |
| 142 | commits: list[Mapping[str, str | None]] |
| 143 | |
| 144 | |
| 145 | class _ShortlogOut(TypedDict): |
| 146 | repo_id: str |
| 147 | branch: str |
| 148 | groups: list[_GroupOut] |
| 149 | |
| 150 | |
| 151 | def _parse_json(result: InvokeResult) -> _ShortlogOut: |
| 152 | raw = json.loads(result.output.strip()) |
| 153 | groups: list[_GroupOut] = [ |
| 154 | _GroupOut( |
| 155 | key=g["key"], |
| 156 | count=g["count"], |
| 157 | commits=g["commits"], |
| 158 | ) |
| 159 | for g in raw["groups"] |
| 160 | ] |
| 161 | return _ShortlogOut( |
| 162 | repo_id=raw["repo_id"], |
| 163 | branch=raw["branch"], |
| 164 | groups=groups, |
| 165 | ) |
| 166 | |
| 167 | |
| 168 | # --------------------------------------------------------------------------- |
| 169 | # Unit: _branch_names — symlink guard |
| 170 | # --------------------------------------------------------------------------- |
| 171 | |
| 172 | |
| 173 | def test_branch_names_returns_normal_branches(tmp_path: pathlib.Path) -> None: |
| 174 | _init_repo(tmp_path) |
| 175 | _make_commit(tmp_path, branch="main") |
| 176 | _make_commit(tmp_path, branch="dev") |
| 177 | names = _branch_names(tmp_path) |
| 178 | assert "main" in names |
| 179 | assert "dev" in names |
| 180 | |
| 181 | |
| 182 | def test_branch_names_skips_symlinks(tmp_path: pathlib.Path) -> None: |
| 183 | _init_repo(tmp_path) |
| 184 | _make_commit(tmp_path, branch="main") |
| 185 | h_dir = heads_dir(tmp_path) |
| 186 | malicious = h_dir / "malicious-branch" |
| 187 | try: |
| 188 | malicious.symlink_to(tmp_path / "some_other_file") |
| 189 | except OSError: |
| 190 | pytest.skip("filesystem does not support symlinks") |
| 191 | names = _branch_names(tmp_path) |
| 192 | assert "malicious-branch" not in names |
| 193 | assert "main" in names |
| 194 | |
| 195 | |
| 196 | def test_branch_names_missing_heads_dir(tmp_path: pathlib.Path) -> None: |
| 197 | _init_repo(tmp_path) |
| 198 | import shutil |
| 199 | shutil.rmtree(heads_dir(tmp_path)) |
| 200 | assert _branch_names(tmp_path) == [] |
| 201 | |
| 202 | |
| 203 | # --------------------------------------------------------------------------- |
| 204 | # Unit: _group_key |
| 205 | # --------------------------------------------------------------------------- |
| 206 | |
| 207 | |
| 208 | def _make_rec( |
| 209 | *, |
| 210 | author: str = "", |
| 211 | agent_id: str = "", |
| 212 | model_id: str = "", |
| 213 | branch: str = "main", |
| 214 | ) -> CommitRecord: |
| 215 | return CommitRecord( |
| 216 | commit_id="aaa", |
| 217 | branch=branch, |
| 218 | snapshot_id="snap", |
| 219 | message="x", |
| 220 | committed_at=datetime.datetime.now(datetime.timezone.utc), |
| 221 | author=author, |
| 222 | agent_id=agent_id, |
| 223 | model_id=model_id, |
| 224 | ) |
| 225 | |
| 226 | |
| 227 | def test_group_key_author_with_author() -> None: |
| 228 | rec = _make_rec(author="Alice") |
| 229 | assert _group_key(rec, "author") == "Alice" |
| 230 | |
| 231 | |
| 232 | def test_group_key_author_fallback_to_agent() -> None: |
| 233 | rec = _make_rec(agent_id="bot-1") |
| 234 | assert _group_key(rec, "author") == "bot-1 (agent)" |
| 235 | |
| 236 | |
| 237 | def test_group_key_author_unknown() -> None: |
| 238 | rec = _make_rec() |
| 239 | assert _group_key(rec, "author") == "(unknown)" |
| 240 | |
| 241 | |
| 242 | def test_group_key_agent() -> None: |
| 243 | rec = _make_rec(agent_id="gpt-agent") |
| 244 | assert _group_key(rec, "agent") == "gpt-agent" |
| 245 | |
| 246 | |
| 247 | def test_group_key_agent_no_agent() -> None: |
| 248 | rec = _make_rec() |
| 249 | assert _group_key(rec, "agent") == "(no agent)" |
| 250 | |
| 251 | |
| 252 | def test_group_key_model() -> None: |
| 253 | rec = _make_rec(model_id="gpt-4o") |
| 254 | assert _group_key(rec, "model") == "gpt-4o" |
| 255 | |
| 256 | |
| 257 | def test_group_key_model_no_model() -> None: |
| 258 | rec = _make_rec() |
| 259 | assert _group_key(rec, "model") == "(no model)" |
| 260 | |
| 261 | |
| 262 | def test_group_key_branch() -> None: |
| 263 | rec = _make_rec(branch="feat/my-thing") |
| 264 | assert _group_key(rec, "branch") == "feat/my-thing" |
| 265 | |
| 266 | |
| 267 | # --------------------------------------------------------------------------- |
| 268 | # Unit: _parse_date |
| 269 | # --------------------------------------------------------------------------- |
| 270 | |
| 271 | |
| 272 | def test_parse_date_valid() -> None: |
| 273 | dt = _parse_date("2025-03-15", "--since") |
| 274 | assert dt.year == 2025 |
| 275 | assert dt.month == 3 |
| 276 | assert dt.day == 15 |
| 277 | assert dt.tzinfo == datetime.timezone.utc |
| 278 | |
| 279 | |
| 280 | def test_parse_date_invalid_exits() -> None: |
| 281 | with pytest.raises(ValueError): |
| 282 | _parse_date("not-a-date", "--since") |
| 283 | |
| 284 | |
| 285 | def test_parse_date_wrong_format_exits() -> None: |
| 286 | with pytest.raises(ValueError): |
| 287 | _parse_date("15/03/2025", "--since") |
| 288 | |
| 289 | |
| 290 | # --------------------------------------------------------------------------- |
| 291 | # Security: ANSI injection |
| 292 | # --------------------------------------------------------------------------- |
| 293 | |
| 294 | |
| 295 | def test_ansi_in_author_name_stripped_text(tmp_path: pathlib.Path) -> None: |
| 296 | _init_repo(tmp_path) |
| 297 | _make_commit(tmp_path, author="Evil\x1b[31mRED\x1b[0m") |
| 298 | result = _invoke(["shortlog"], _env(tmp_path)) |
| 299 | assert result.exit_code == 0 |
| 300 | assert "\x1b[31m" not in result.output |
| 301 | |
| 302 | |
| 303 | def test_ansi_in_author_name_raw_in_json(tmp_path: pathlib.Path) -> None: |
| 304 | _init_repo(tmp_path) |
| 305 | _make_commit(tmp_path, author="Evil\x1b[31mRED\x1b[0m") |
| 306 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 307 | assert result.exit_code == 0 |
| 308 | data = _parse_json(result) |
| 309 | assert data["groups"][0]["key"] == "Evil\x1b[31mRED\x1b[0m" |
| 310 | |
| 311 | |
| 312 | def test_ansi_in_message_stripped_text(tmp_path: pathlib.Path) -> None: |
| 313 | _init_repo(tmp_path) |
| 314 | commit_id = _make_commit(tmp_path) |
| 315 | # Directly overwrite message in stored commit to contain ANSI. |
| 316 | from muse.core.commits import read_commit |
| 317 | original = read_commit(tmp_path, commit_id) |
| 318 | assert original is not None |
| 319 | from muse.core.ids import hash_commit |
| 320 | from muse.core.commits import write_commit |
| 321 | malicious_msg = "fix: \x1b[1mBOLD\x1b[0m thing" |
| 322 | parent_ids = [original.parent_commit_id] if original.parent_commit_id else [] |
| 323 | new_cid = hash_commit( |
| 324 | parent_ids=parent_ids, |
| 325 | snapshot_id=original.snapshot_id, |
| 326 | message=malicious_msg, |
| 327 | committed_at_iso=original.committed_at.isoformat(), |
| 328 | author=original.author, |
| 329 | ) |
| 330 | patched = CommitRecord( |
| 331 | commit_id=new_cid, |
| 332 | branch=original.branch, |
| 333 | snapshot_id=original.snapshot_id, |
| 334 | message=malicious_msg, |
| 335 | committed_at=original.committed_at, |
| 336 | author=original.author, |
| 337 | ) |
| 338 | write_commit(tmp_path, patched) |
| 339 | (heads_dir(tmp_path) / "main").write_text(new_cid) |
| 340 | result = _invoke(["shortlog"], _env(tmp_path)) |
| 341 | assert "\x1b[1m" not in result.output |
| 342 | |
| 343 | |
| 344 | # --------------------------------------------------------------------------- |
| 345 | # Error routing: all user errors go to stderr |
| 346 | # --------------------------------------------------------------------------- |
| 347 | |
| 348 | |
| 349 | def test_since_invalid_format_stderr(tmp_path: pathlib.Path) -> None: |
| 350 | _init_repo(tmp_path) |
| 351 | _make_commit(tmp_path) |
| 352 | result = _invoke(["shortlog", "--since", "01-01-2025"], _env(tmp_path)) |
| 353 | assert result.exit_code != 0 |
| 354 | |
| 355 | |
| 356 | def test_until_invalid_format_stderr(tmp_path: pathlib.Path) -> None: |
| 357 | _init_repo(tmp_path) |
| 358 | _make_commit(tmp_path) |
| 359 | result = _invoke(["shortlog", "--until", "not-a-date"], _env(tmp_path)) |
| 360 | assert result.exit_code != 0 |
| 361 | |
| 362 | |
| 363 | # --------------------------------------------------------------------------- |
| 364 | # JSON schema: _ShortlogJson |
| 365 | # --------------------------------------------------------------------------- |
| 366 | |
| 367 | |
| 368 | def test_json_schema_empty_repo(tmp_path: pathlib.Path) -> None: |
| 369 | _init_repo(tmp_path) |
| 370 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 371 | assert result.exit_code == 0 |
| 372 | data = _parse_json(result) |
| 373 | assert data["repo_id"] == _REPO_ID |
| 374 | assert data["branch"] == "main" |
| 375 | assert data["groups"] == [] |
| 376 | |
| 377 | |
| 378 | def test_json_schema_all_fields_present(tmp_path: pathlib.Path) -> None: |
| 379 | _init_repo(tmp_path) |
| 380 | _make_commit(tmp_path, author="Alice", agent_id="bot-1", model_id="gpt-4o") |
| 381 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 382 | assert result.exit_code == 0 |
| 383 | data = _parse_json(result) |
| 384 | assert data["repo_id"] == _REPO_ID |
| 385 | assert data["branch"] == "main" |
| 386 | grp = data["groups"][0] |
| 387 | assert grp["key"] == "Alice" |
| 388 | assert grp["count"] == 1 |
| 389 | commit_entry = grp["commits"][0] |
| 390 | assert "commit_id" in commit_entry |
| 391 | assert "message" in commit_entry |
| 392 | assert "committed_at" in commit_entry |
| 393 | assert "author" in commit_entry |
| 394 | assert "agent_id" in commit_entry |
| 395 | assert "model_id" in commit_entry |
| 396 | |
| 397 | |
| 398 | def test_json_schema_repo_id_and_branch_in_output(tmp_path: pathlib.Path) -> None: |
| 399 | _init_repo(tmp_path) |
| 400 | _make_commit(tmp_path, branch="main") |
| 401 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 402 | assert result.exit_code == 0 |
| 403 | data = _parse_json(result) |
| 404 | assert data["repo_id"] == _REPO_ID |
| 405 | assert data["branch"] == "main" |
| 406 | |
| 407 | |
| 408 | def test_json_schema_all_branches_label(tmp_path: pathlib.Path) -> None: |
| 409 | _init_repo(tmp_path) |
| 410 | _make_commit(tmp_path, branch="main") |
| 411 | result = _invoke(["shortlog", "--all", "--json"], _env(tmp_path)) |
| 412 | assert result.exit_code == 0 |
| 413 | data = _parse_json(result) |
| 414 | assert data["branch"] == "__all__" |
| 415 | |
| 416 | |
| 417 | def test_json_agent_id_and_model_id_present(tmp_path: pathlib.Path) -> None: |
| 418 | _init_repo(tmp_path) |
| 419 | _make_commit(tmp_path, agent_id="agent-007", model_id="claude-3") |
| 420 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 421 | assert result.exit_code == 0 |
| 422 | data = _parse_json(result) |
| 423 | entry = data["groups"][0]["commits"][0] |
| 424 | assert entry["agent_id"] == "agent-007" |
| 425 | assert entry["model_id"] == "claude-3" |
| 426 | |
| 427 | |
| 428 | # --------------------------------------------------------------------------- |
| 429 | # New flag: --group-by |
| 430 | # --------------------------------------------------------------------------- |
| 431 | |
| 432 | |
| 433 | def test_group_by_agent(tmp_path: pathlib.Path) -> None: |
| 434 | _init_repo(tmp_path) |
| 435 | _make_commit(tmp_path, author="Alice", agent_id="bot-1") |
| 436 | _make_commit(tmp_path, author="Bob", agent_id="bot-2") |
| 437 | _make_commit(tmp_path, author="Alice", agent_id="bot-1") |
| 438 | result = _invoke(["shortlog", "--group-by", "agent", "--json"], _env(tmp_path)) |
| 439 | assert result.exit_code == 0 |
| 440 | data = _parse_json(result) |
| 441 | keys = {g["key"] for g in data["groups"]} |
| 442 | assert "bot-1" in keys |
| 443 | assert "bot-2" in keys |
| 444 | |
| 445 | |
| 446 | def test_group_by_model(tmp_path: pathlib.Path) -> None: |
| 447 | _init_repo(tmp_path) |
| 448 | _make_commit(tmp_path, model_id="gpt-4o") |
| 449 | _make_commit(tmp_path, model_id="claude-3") |
| 450 | _make_commit(tmp_path, model_id="gpt-4o") |
| 451 | result = _invoke(["shortlog", "--group-by", "model", "--json"], _env(tmp_path)) |
| 452 | assert result.exit_code == 0 |
| 453 | data = _parse_json(result) |
| 454 | keys = {g["key"] for g in data["groups"]} |
| 455 | assert "gpt-4o" in keys |
| 456 | assert "claude-3" in keys |
| 457 | gpt_count = next(g["count"] for g in data["groups"] if g["key"] == "gpt-4o") |
| 458 | assert gpt_count == 2 |
| 459 | |
| 460 | |
| 461 | def test_group_by_branch(tmp_path: pathlib.Path) -> None: |
| 462 | _init_repo(tmp_path) |
| 463 | _make_commit(tmp_path, branch="main") |
| 464 | _make_commit(tmp_path, branch="dev") |
| 465 | _make_commit(tmp_path, branch="main") |
| 466 | result = _invoke( |
| 467 | ["shortlog", "--all", "--group-by", "branch", "--json"], _env(tmp_path) |
| 468 | ) |
| 469 | assert result.exit_code == 0 |
| 470 | data = _parse_json(result) |
| 471 | keys = {g["key"] for g in data["groups"]} |
| 472 | assert "main" in keys |
| 473 | assert "dev" in keys |
| 474 | |
| 475 | |
| 476 | def test_group_by_invalid_choice(tmp_path: pathlib.Path) -> None: |
| 477 | _init_repo(tmp_path) |
| 478 | result = _invoke(["shortlog", "--group-by", "badfield"], _env(tmp_path)) |
| 479 | assert result.exit_code != 0 |
| 480 | |
| 481 | |
| 482 | # --------------------------------------------------------------------------- |
| 483 | # New flag: --summary |
| 484 | # --------------------------------------------------------------------------- |
| 485 | |
| 486 | |
| 487 | def test_summary_suppresses_messages(tmp_path: pathlib.Path) -> None: |
| 488 | _init_repo(tmp_path) |
| 489 | _make_commit(tmp_path, author="Alice") |
| 490 | _make_commit(tmp_path, author="Alice") |
| 491 | result = _invoke(["shortlog", "--summary"], _env(tmp_path)) |
| 492 | assert result.exit_code == 0 |
| 493 | # Author line should still appear. |
| 494 | assert "Alice" in result.output |
| 495 | # Individual commit messages should not appear (they start with spaces). |
| 496 | assert "msg" not in result.output |
| 497 | |
| 498 | |
| 499 | def test_summary_with_json_still_includes_commits(tmp_path: pathlib.Path) -> None: |
| 500 | """--summary only suppresses messages in text mode; JSON always includes them.""" |
| 501 | _init_repo(tmp_path) |
| 502 | _make_commit(tmp_path, author="Alice") |
| 503 | result = _invoke(["shortlog", "--summary", "--json"], _env(tmp_path)) |
| 504 | assert result.exit_code == 0 |
| 505 | data = _parse_json(result) |
| 506 | assert len(data["groups"][0]["commits"]) >= 1 |
| 507 | |
| 508 | |
| 509 | # --------------------------------------------------------------------------- |
| 510 | # New flag: --no-merges |
| 511 | # --------------------------------------------------------------------------- |
| 512 | |
| 513 | |
| 514 | def test_no_merges_excludes_merge_commits(tmp_path: pathlib.Path) -> None: |
| 515 | """get_commits_for_branch follows first-parent only. |
| 516 | |
| 517 | Chain: c1 → c2 → c3(merge, parent2=c1) → c4 |
| 518 | First-parent walk from c4 returns [c4, c3, c2, c1]. |
| 519 | With --no-merges, c3 is excluded → 3 commits remain. |
| 520 | """ |
| 521 | _init_repo(tmp_path) |
| 522 | c1 = _make_commit(tmp_path, author="Alice") |
| 523 | c2 = _make_commit(tmp_path, author="Bob") # chains to c1 |
| 524 | # Merge commit: auto-chains first-parent to c2; parent2 points to c1. |
| 525 | _make_commit(tmp_path, author="Alice", parent2_id=c1) # chains to c2 |
| 526 | _make_commit(tmp_path, author="Bob") # chains to merge |
| 527 | result = _invoke(["shortlog", "--no-merges", "--json"], _env(tmp_path)) |
| 528 | assert result.exit_code == 0 |
| 529 | data = _parse_json(result) |
| 530 | total = sum(g["count"] for g in data["groups"]) |
| 531 | assert total == 3 # c1, c2, c4 — c3 (merge) excluded |
| 532 | |
| 533 | |
| 534 | def test_no_merges_with_all_non_merges(tmp_path: pathlib.Path) -> None: |
| 535 | _init_repo(tmp_path) |
| 536 | for _ in range(5): |
| 537 | _make_commit(tmp_path, author="Alice") |
| 538 | result = _invoke(["shortlog", "--no-merges", "--json"], _env(tmp_path)) |
| 539 | assert result.exit_code == 0 |
| 540 | data = _parse_json(result) |
| 541 | assert sum(g["count"] for g in data["groups"]) == 5 |
| 542 | |
| 543 | |
| 544 | # --------------------------------------------------------------------------- |
| 545 | # New flags: --since / --until |
| 546 | # --------------------------------------------------------------------------- |
| 547 | |
| 548 | |
| 549 | def test_since_filters_old_commits(tmp_path: pathlib.Path) -> None: |
| 550 | _init_repo(tmp_path) |
| 551 | old = datetime.datetime(2020, 1, 1, tzinfo=datetime.timezone.utc) |
| 552 | new = datetime.datetime(2025, 6, 1, tzinfo=datetime.timezone.utc) |
| 553 | _make_commit(tmp_path, author="Old", committed_at=old) |
| 554 | _make_commit(tmp_path, author="New", committed_at=new) |
| 555 | result = _invoke(["shortlog", "--since", "2025-01-01", "--json"], _env(tmp_path)) |
| 556 | assert result.exit_code == 0 |
| 557 | data = _parse_json(result) |
| 558 | keys = {g["key"] for g in data["groups"]} |
| 559 | assert "New" in keys |
| 560 | assert "Old" not in keys |
| 561 | |
| 562 | |
| 563 | def test_until_filters_future_commits(tmp_path: pathlib.Path) -> None: |
| 564 | _init_repo(tmp_path) |
| 565 | old = datetime.datetime(2020, 1, 1, tzinfo=datetime.timezone.utc) |
| 566 | new = datetime.datetime(2025, 6, 1, tzinfo=datetime.timezone.utc) |
| 567 | _make_commit(tmp_path, author="Old", committed_at=old) |
| 568 | _make_commit(tmp_path, author="New", committed_at=new) |
| 569 | result = _invoke(["shortlog", "--until", "2022-12-31", "--json"], _env(tmp_path)) |
| 570 | assert result.exit_code == 0 |
| 571 | data = _parse_json(result) |
| 572 | keys = {g["key"] for g in data["groups"]} |
| 573 | assert "Old" in keys |
| 574 | assert "New" not in keys |
| 575 | |
| 576 | |
| 577 | def test_since_and_until_window(tmp_path: pathlib.Path) -> None: |
| 578 | _init_repo(tmp_path) |
| 579 | dates = [ |
| 580 | datetime.datetime(2024, 1, 1, tzinfo=datetime.timezone.utc), |
| 581 | datetime.datetime(2025, 3, 15, tzinfo=datetime.timezone.utc), |
| 582 | datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc), |
| 583 | ] |
| 584 | authors = ["Before", "Inside", "After"] |
| 585 | for a, d in zip(authors, dates): |
| 586 | _make_commit(tmp_path, author=a, committed_at=d) |
| 587 | result = _invoke( |
| 588 | ["shortlog", "--since", "2025-01-01", "--until", "2025-12-31", "--json"], |
| 589 | _env(tmp_path), |
| 590 | ) |
| 591 | assert result.exit_code == 0 |
| 592 | data = _parse_json(result) |
| 593 | keys = {g["key"] for g in data["groups"]} |
| 594 | assert "Inside" in keys |
| 595 | assert "Before" not in keys |
| 596 | assert "After" not in keys |
| 597 | |
| 598 | |
| 599 | def test_since_no_results_returns_empty_json(tmp_path: pathlib.Path) -> None: |
| 600 | _init_repo(tmp_path) |
| 601 | _make_commit( |
| 602 | tmp_path, |
| 603 | author="Old", |
| 604 | committed_at=datetime.datetime(2020, 1, 1, tzinfo=datetime.timezone.utc), |
| 605 | ) |
| 606 | result = _invoke(["shortlog", "--since", "2030-01-01", "--json"], _env(tmp_path)) |
| 607 | assert result.exit_code == 0 |
| 608 | data = _parse_json(result) |
| 609 | assert data["groups"] == [] |
| 610 | |
| 611 | |
| 612 | # --------------------------------------------------------------------------- |
| 613 | # Integration |
| 614 | # --------------------------------------------------------------------------- |
| 615 | |
| 616 | |
| 617 | def test_integration_all_branches_dedup(tmp_path: pathlib.Path) -> None: |
| 618 | """A commit reachable from two branches should count once.""" |
| 619 | _init_repo(tmp_path) |
| 620 | shared = _make_commit(tmp_path, author="Alice", branch="main") |
| 621 | # Create dev branch pointing at same commit (by writing the ref file). |
| 622 | dev_ref = heads_dir(tmp_path) / "dev" |
| 623 | dev_ref.write_text(shared, encoding="utf-8") |
| 624 | result = _invoke(["shortlog", "--all", "--json"], _env(tmp_path)) |
| 625 | assert result.exit_code == 0 |
| 626 | data = _parse_json(result) |
| 627 | total = sum(g["count"] for g in data["groups"]) |
| 628 | assert total == 1 # deduplicated |
| 629 | |
| 630 | |
| 631 | def test_integration_limit_early_exit(tmp_path: pathlib.Path) -> None: |
| 632 | _init_repo(tmp_path) |
| 633 | for i in range(50): |
| 634 | _make_commit(tmp_path, author=f"Author{i % 5}") |
| 635 | result = _invoke(["shortlog", "--limit", "10", "--json"], _env(tmp_path)) |
| 636 | assert result.exit_code == 0 |
| 637 | data = _parse_json(result) |
| 638 | total = sum(g["count"] for g in data["groups"]) |
| 639 | assert total <= 10 |
| 640 | |
| 641 | |
| 642 | def test_integration_numbered_combined_with_since(tmp_path: pathlib.Path) -> None: |
| 643 | _init_repo(tmp_path) |
| 644 | old = datetime.datetime(2020, 1, 1, tzinfo=datetime.timezone.utc) |
| 645 | new = datetime.datetime(2025, 6, 1, tzinfo=datetime.timezone.utc) |
| 646 | for _ in range(3): |
| 647 | _make_commit(tmp_path, author="Prolific", committed_at=new) |
| 648 | _make_commit(tmp_path, author="Old", committed_at=old) |
| 649 | result = _invoke( |
| 650 | ["shortlog", "--since", "2025-01-01", "--numbered", "--json"], |
| 651 | _env(tmp_path), |
| 652 | ) |
| 653 | assert result.exit_code == 0 |
| 654 | data = _parse_json(result) |
| 655 | assert data["groups"][0]["key"] == "Prolific" |
| 656 | assert "Old" not in {g["key"] for g in data["groups"]} |
| 657 | |
| 658 | |
| 659 | # --------------------------------------------------------------------------- |
| 660 | # E2E: help output |
| 661 | # --------------------------------------------------------------------------- |
| 662 | |
| 663 | |
| 664 | def test_help_shows_new_flags() -> None: |
| 665 | result = _invoke(["shortlog", "--help"], {}) |
| 666 | assert result.exit_code == 0 |
| 667 | for flag in ("--group-by", "--summary", "--no-merges", "--since", "--until", "--json"): |
| 668 | assert flag in result.output, f"Missing flag: {flag}" |
| 669 | |
| 670 | |
| 671 | def test_help_mentions_group_by_choices() -> None: |
| 672 | result = _invoke(["shortlog", "--help"], {}) |
| 673 | for choice in ("author", "agent", "model", "branch"): |
| 674 | assert choice in result.output |
| 675 | |
| 676 | |
| 677 | # --------------------------------------------------------------------------- |
| 678 | # Stress: 500 commits × 5 authors |
| 679 | # --------------------------------------------------------------------------- |
| 680 | |
| 681 | |
| 682 | def test_stress_500_commits(tmp_path: pathlib.Path) -> None: |
| 683 | _init_repo(tmp_path) |
| 684 | authors = ["Amy", "Ben", "Cleo", "Dan", "Eva"] |
| 685 | for i in range(500): |
| 686 | _make_commit(tmp_path, author=authors[i % 5]) |
| 687 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 688 | assert result.exit_code == 0 |
| 689 | data = _parse_json(result) |
| 690 | total = sum(g["count"] for g in data["groups"]) |
| 691 | assert total == 500 |
| 692 | assert len(data["groups"]) == 5 |
| 693 | |
| 694 | |
| 695 | def test_stress_500_commits_numbered(tmp_path: pathlib.Path) -> None: |
| 696 | _init_repo(tmp_path) |
| 697 | # Give Alice 300, Bob 200. |
| 698 | for _ in range(300): |
| 699 | _make_commit(tmp_path, author="Alice") |
| 700 | for _ in range(200): |
| 701 | _make_commit(tmp_path, author="Bob") |
| 702 | result = _invoke(["shortlog", "--numbered", "--json"], _env(tmp_path)) |
| 703 | assert result.exit_code == 0 |
| 704 | data = _parse_json(result) |
| 705 | assert data["groups"][0]["key"] == "Alice" |
| 706 | assert data["groups"][0]["count"] == 300 |
| 707 | |
| 708 | |
| 709 | # --------------------------------------------------------------------------- |
| 710 | # JSON schema — duration_ms + exit_code + truncated on every output path |
| 711 | # --------------------------------------------------------------------------- |
| 712 | |
| 713 | |
| 714 | class TestJsonSchema: |
| 715 | """Every --json response must carry duration_ms, exit_code, and truncated.""" |
| 716 | |
| 717 | def _assert_schema(self, d: Mapping[str, object], *, exit_code: int = 0) -> None: |
| 718 | assert "duration_ms" in d, f"duration_ms missing: {d}" |
| 719 | assert isinstance(d["duration_ms"], (int, float)) |
| 720 | assert d["duration_ms"] >= 0 |
| 721 | assert "exit_code" in d, f"exit_code missing: {d}" |
| 722 | assert d["exit_code"] == exit_code |
| 723 | assert "truncated" in d, f"truncated missing: {d}" |
| 724 | |
| 725 | def test_normal_output_has_schema(self, tmp_path: pathlib.Path) -> None: |
| 726 | _init_repo(tmp_path) |
| 727 | _make_commit(tmp_path, author="Alice") |
| 728 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 729 | assert result.exit_code == 0 |
| 730 | self._assert_schema(json.loads(result.output)) |
| 731 | |
| 732 | def test_empty_repo_json_has_schema(self, tmp_path: pathlib.Path) -> None: |
| 733 | _init_repo(tmp_path) |
| 734 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 735 | assert result.exit_code == 0 |
| 736 | self._assert_schema(json.loads(result.output)) |
| 737 | |
| 738 | def test_all_branches_json_has_schema(self, tmp_path: pathlib.Path) -> None: |
| 739 | _init_repo(tmp_path) |
| 740 | _make_commit(tmp_path, branch="main") |
| 741 | result = _invoke(["shortlog", "--all", "--json"], _env(tmp_path)) |
| 742 | assert result.exit_code == 0 |
| 743 | self._assert_schema(json.loads(result.output)) |
| 744 | |
| 745 | def test_numbered_json_has_schema(self, tmp_path: pathlib.Path) -> None: |
| 746 | _init_repo(tmp_path) |
| 747 | _make_commit(tmp_path, author="Alice") |
| 748 | _make_commit(tmp_path, author="Bob") |
| 749 | result = _invoke(["shortlog", "--numbered", "--json"], _env(tmp_path)) |
| 750 | assert result.exit_code == 0 |
| 751 | self._assert_schema(json.loads(result.output)) |
| 752 | |
| 753 | def test_since_filtered_empty_has_schema(self, tmp_path: pathlib.Path) -> None: |
| 754 | """_emit_empty path (after filtering) must also carry the schema.""" |
| 755 | _init_repo(tmp_path) |
| 756 | _make_commit( |
| 757 | tmp_path, author="Old", |
| 758 | committed_at=datetime.datetime(2020, 1, 1, tzinfo=datetime.timezone.utc), |
| 759 | ) |
| 760 | result = _invoke(["shortlog", "--since", "2030-01-01", "--json"], _env(tmp_path)) |
| 761 | assert result.exit_code == 0 |
| 762 | self._assert_schema(json.loads(result.output)) |
| 763 | |
| 764 | def test_exit_code_zero_on_success(self, tmp_path: pathlib.Path) -> None: |
| 765 | _init_repo(tmp_path) |
| 766 | _make_commit(tmp_path) |
| 767 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 768 | d = json.loads(result.output) |
| 769 | assert d["exit_code"] == 0 |
| 770 | |
| 771 | |
| 772 | # --------------------------------------------------------------------------- |
| 773 | # truncated flag — set when --limit caps the result |
| 774 | # --------------------------------------------------------------------------- |
| 775 | |
| 776 | |
| 777 | class TestTruncated: |
| 778 | """truncated:true when --limit hit; false otherwise.""" |
| 779 | |
| 780 | def test_truncated_true_when_limit_hit(self, tmp_path: pathlib.Path) -> None: |
| 781 | _init_repo(tmp_path) |
| 782 | for _ in range(10): |
| 783 | _make_commit(tmp_path, author="Alice") |
| 784 | result = _invoke(["shortlog", "--limit", "3", "--json"], _env(tmp_path)) |
| 785 | assert result.exit_code == 0 |
| 786 | d = json.loads(result.output) |
| 787 | assert d["truncated"] is True |
| 788 | |
| 789 | def test_truncated_false_when_under_limit(self, tmp_path: pathlib.Path) -> None: |
| 790 | _init_repo(tmp_path) |
| 791 | for _ in range(5): |
| 792 | _make_commit(tmp_path, author="Alice") |
| 793 | result = _invoke(["shortlog", "--limit", "10", "--json"], _env(tmp_path)) |
| 794 | assert result.exit_code == 0 |
| 795 | d = json.loads(result.output) |
| 796 | assert d["truncated"] is False |
| 797 | |
| 798 | def test_truncated_false_when_no_limit(self, tmp_path: pathlib.Path) -> None: |
| 799 | _init_repo(tmp_path) |
| 800 | for _ in range(5): |
| 801 | _make_commit(tmp_path, author="Alice") |
| 802 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 803 | assert result.exit_code == 0 |
| 804 | d = json.loads(result.output) |
| 805 | assert d["truncated"] is False |
| 806 | |
| 807 | def test_truncated_false_on_empty_repo(self, tmp_path: pathlib.Path) -> None: |
| 808 | _init_repo(tmp_path) |
| 809 | result = _invoke(["shortlog", "--json"], _env(tmp_path)) |
| 810 | assert result.exit_code == 0 |
| 811 | d = json.loads(result.output) |
| 812 | assert d["truncated"] is False |
| 813 | |
| 814 | |
| 815 | # --------------------------------------------------------------------------- |
| 816 | # Error JSON — date parse errors emit structured JSON to stdout with --json |
| 817 | # --------------------------------------------------------------------------- |
| 818 | |
| 819 | |
| 820 | class TestErrorJson: |
| 821 | """--since / --until bad dates must emit JSON to stdout when --json is set.""" |
| 822 | |
| 823 | def _assert_error(self, result: InvokeResult) -> Mapping[str, object]: |
| 824 | assert result.exit_code != 0 |
| 825 | d = json.loads(result.output) |
| 826 | assert "error" in d |
| 827 | assert "duration_ms" in d |
| 828 | assert "exit_code" in d |
| 829 | assert d["exit_code"] != 0 |
| 830 | return d |
| 831 | |
| 832 | def test_since_bad_date_json_error(self, tmp_path: pathlib.Path) -> None: |
| 833 | _init_repo(tmp_path) |
| 834 | _make_commit(tmp_path) |
| 835 | result = _invoke(["shortlog", "--json", "--since", "not-a-date"], _env(tmp_path)) |
| 836 | self._assert_error(result) |
| 837 | |
| 838 | def test_until_bad_date_json_error(self, tmp_path: pathlib.Path) -> None: |
| 839 | _init_repo(tmp_path) |
| 840 | _make_commit(tmp_path) |
| 841 | result = _invoke(["shortlog", "--json", "--until", "01/01/2025"], _env(tmp_path)) |
| 842 | self._assert_error(result) |
| 843 | |
| 844 | def test_date_error_has_message(self, tmp_path: pathlib.Path) -> None: |
| 845 | _init_repo(tmp_path) |
| 846 | result = _invoke(["shortlog", "--json", "--since", "garbage"], _env(tmp_path)) |
| 847 | d = self._assert_error(result) |
| 848 | assert isinstance(d["message"], str) and len(d["message"]) > 0 |
| 849 | |
| 850 | |
| 851 | # --------------------------------------------------------------------------- |
| 852 | # _parse_date refactor — now raises ValueError, not SystemExit |
| 853 | # --------------------------------------------------------------------------- |
| 854 | |
| 855 | |
| 856 | class TestParseDateRefactor: |
| 857 | """_parse_date is a pure parser; it raises ValueError, not SystemExit.""" |
| 858 | |
| 859 | def test_invalid_date_raises_value_error(self) -> None: |
| 860 | with pytest.raises(ValueError): |
| 861 | _parse_date("not-a-date", "--since") |
| 862 | |
| 863 | def test_wrong_format_raises_value_error(self) -> None: |
| 864 | with pytest.raises(ValueError): |
| 865 | _parse_date("15/03/2025", "--since") |
File History
4 commits
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
29 days ago