test_coord_pull_null_records.py
python
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
| 1 | """ |
| 2 | Tests for the bug: hub response with null/non-list records or null/non-integer |
| 3 | cursor causes unhandled TypeError that escapes run_pull as a raw traceback. |
| 4 | |
| 5 | Root cause (coord_bus.py::pull_from_hub): |
| 6 | |
| 7 | return _post_json(url, body, token) |
| 8 | |
| 9 | The raw hub response is returned without validation. run_pull then does: |
| 10 | |
| 11 | pulled_records: list[dict] = result.get("records", []) |
| 12 | cursor: int = result.get("cursor", 0) |
| 13 | |
| 14 | When hub returns {"records": null}: |
| 15 | - result.get("records", []) → None (key EXISTS, default [] not used) |
| 16 | - if pulled_records: → False (None is falsy, so _write_remote_records skipped) |
| 17 | - len(pulled_records) → TypeError: object of type 'NoneType' has no len() |
| 18 | |
| 19 | When hub returns {"records": ["a", "b"]} (list of strings not dicts): |
| 20 | - _write_remote_records iterates; rec.get("kind") → AttributeError on str |
| 21 | |
| 22 | When hub returns {"cursor": null}: |
| 23 | - cursor = None; json.dumps({"cursor": None}) → "cursor": null in output |
| 24 | - text mode: f"cursor: {cursor}" → "cursor: None" — contract violated |
| 25 | |
| 26 | When hub returns {"cursor": "malicious"}: |
| 27 | - cursor = "malicious"; propagated verbatim to JSON output |
| 28 | |
| 29 | None of these are CoordBusError. run_pull only catches CoordBusError. |
| 30 | TypeError/AttributeError escape as raw tracebacks. |
| 31 | |
| 32 | Fix location: pull_from_hub in coord_bus.py — validate response before returning. |
| 33 | |
| 34 | Coverage: |
| 35 | Unit — pull_from_hub directly, all bad-value variants |
| 36 | Integration — run_pull with bad hub response, two layers deep |
| 37 | End-to-end — CLI output is valid JSON with no tracebacks |
| 38 | Stress — 50 consecutive pulls mixing good and bad responses |
| 39 | Performance — bad response path not slower than good path |
| 40 | Security — hub cannot inject arbitrary values into cursor output |
| 41 | Data integrity — cursor and count in output reflect reality |
| 42 | """ |
| 43 | from __future__ import annotations |
| 44 | |
| 45 | import argparse |
| 46 | import io |
| 47 | import itertools |
| 48 | import json |
| 49 | import pathlib |
| 50 | import sys |
| 51 | import time |
| 52 | from unittest.mock import patch |
| 53 | |
| 54 | import pytest |
| 55 | |
| 56 | from muse.core.types import MsgpackDict, MsgpackValue, content_hash |
| 57 | from muse.core.paths import coordination_dir, muse_dir |
| 58 | |
| 59 | # --------------------------------------------------------------------------- |
| 60 | # Helpers |
| 61 | # --------------------------------------------------------------------------- |
| 62 | |
| 63 | _FUTURE_TS = "2099-12-31T23:59:59+00:00" |
| 64 | |
| 65 | _id_seq = itertools.count() |
| 66 | |
| 67 | |
| 68 | def _new_id() -> str: |
| 69 | return content_hash({"seq": next(_id_seq)}) |
| 70 | |
| 71 | |
| 72 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 73 | muse_dir(tmp_path).mkdir(parents=True, exist_ok=True) |
| 74 | return tmp_path |
| 75 | |
| 76 | |
| 77 | def _good_record(i: int = 0) -> MsgpackDict: |
| 78 | return { |
| 79 | "kind": "reservation", |
| 80 | "record_id": _new_id(), |
| 81 | "run_id": f"run-{i}", |
| 82 | "payload": {"reservation_id": f"res-{i:06d}", "expires_at": _FUTURE_TS}, |
| 83 | "expires_at": _FUTURE_TS, |
| 84 | } |
| 85 | |
| 86 | |
| 87 | def _run_pull_with_hub_response( |
| 88 | tmp_path: pathlib.Path, |
| 89 | hub_response: MsgpackDict, |
| 90 | ) -> tuple[int | str | None, str]: |
| 91 | """ |
| 92 | Run run_pull with pull_from_hub mocked to return hub_response. |
| 93 | Returns (exit_code, stdout). |
| 94 | exit_code is None for clean success, int for SystemExit, "CRASH" for unhandled exception. |
| 95 | """ |
| 96 | root = _make_repo(tmp_path) |
| 97 | captured = io.StringIO() |
| 98 | |
| 99 | # Mock _post_json (not pull_from_hub) so the validation in pull_from_hub runs. |
| 100 | with patch("muse.core.coord_bus._post_json", return_value=hub_response), \ |
| 101 | patch("muse.cli.commands.coord_sync.require_repo", return_value=root), \ |
| 102 | patch("muse.cli.commands.coord_sync._resolve_hub_and_signing", |
| 103 | return_value=("https://localhost:1337", "tok")), \ |
| 104 | patch("sys.stdout", captured): |
| 105 | args = argparse.Namespace( |
| 106 | owner="torvalds", slug="linux", |
| 107 | json_out=True, hub_url=None, |
| 108 | since_id=0, limit=1000, kinds=[], |
| 109 | ) |
| 110 | try: |
| 111 | from muse.cli.commands.coord_sync import run_pull |
| 112 | run_pull(args) |
| 113 | except SystemExit as exc: |
| 114 | return (exc.code, captured.getvalue()) |
| 115 | except Exception as exc: |
| 116 | return ("CRASH", f"{type(exc).__name__}: {exc}") |
| 117 | |
| 118 | return (None, captured.getvalue()) |
| 119 | |
| 120 | |
| 121 | # ============================================================================= |
| 122 | # 1. UNIT — pull_from_hub directly |
| 123 | # ============================================================================= |
| 124 | |
| 125 | class TestPullFromHubNullRecordsUnit: |
| 126 | """ |
| 127 | Unit tests on coord_bus.pull_from_hub. |
| 128 | _post_json mocked to return bad responses. |
| 129 | Assert CoordBusError is raised, not TypeError/AttributeError. |
| 130 | """ |
| 131 | |
| 132 | # --- records field --- |
| 133 | |
| 134 | def test_null_records_raises_coord_bus_error(self) -> None: |
| 135 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 136 | with patch("muse.core.coord_bus._post_json", |
| 137 | return_value={"records": None, "cursor": 0}): |
| 138 | with pytest.raises(CoordBusError): |
| 139 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 140 | |
| 141 | def test_null_records_never_raises_raw_typeerror(self) -> None: |
| 142 | """The confirmed crash: len(None) must not escape as TypeError.""" |
| 143 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 144 | with patch("muse.core.coord_bus._post_json", |
| 145 | return_value={"records": None, "cursor": 0}): |
| 146 | try: |
| 147 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 148 | except CoordBusError: |
| 149 | pass # correct |
| 150 | except TypeError as exc: |
| 151 | pytest.fail(f"Raw TypeError escaped pull_from_hub: {exc}") |
| 152 | |
| 153 | @pytest.mark.parametrize("bad_records", [ |
| 154 | "malicious string", |
| 155 | 42, |
| 156 | 3.14, |
| 157 | True, |
| 158 | {"key": "val"}, |
| 159 | ]) |
| 160 | def test_non_list_records_raises_coord_bus_error(self, bad_records: MsgpackValue) -> None: |
| 161 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 162 | with patch("muse.core.coord_bus._post_json", |
| 163 | return_value={"records": bad_records, "cursor": 0}): |
| 164 | with pytest.raises(CoordBusError): |
| 165 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 166 | |
| 167 | def test_list_of_strings_raises_coord_bus_error(self) -> None: |
| 168 | """List of strings would cause AttributeError on .get() — must be CoordBusError.""" |
| 169 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 170 | with patch("muse.core.coord_bus._post_json", |
| 171 | return_value={"records": ["a", "b", "c"], "cursor": 3}): |
| 172 | with pytest.raises(CoordBusError): |
| 173 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 174 | |
| 175 | def test_list_of_strings_never_raises_raw_attributeerror(self) -> None: |
| 176 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 177 | with patch("muse.core.coord_bus._post_json", |
| 178 | return_value={"records": ["a", "b"], "cursor": 2}): |
| 179 | try: |
| 180 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 181 | except CoordBusError: |
| 182 | pass |
| 183 | except AttributeError as exc: |
| 184 | pytest.fail(f"Raw AttributeError escaped pull_from_hub: {exc}") |
| 185 | |
| 186 | def test_missing_records_key_defaults_to_empty_list(self) -> None: |
| 187 | """Hub omits records entirely — must default to [] not crash.""" |
| 188 | from muse.core.coord_bus import pull_from_hub |
| 189 | with patch("muse.core.coord_bus._post_json", |
| 190 | return_value={"cursor": 0}): |
| 191 | result = pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 192 | assert result["records"] == [] |
| 193 | |
| 194 | def test_empty_records_list_is_valid(self) -> None: |
| 195 | from muse.core.coord_bus import pull_from_hub |
| 196 | with patch("muse.core.coord_bus._post_json", |
| 197 | return_value={"records": [], "cursor": 0}): |
| 198 | result = pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 199 | assert result["records"] == [] |
| 200 | assert result["cursor"] == 0 |
| 201 | |
| 202 | def test_valid_records_list_passes_through(self) -> None: |
| 203 | from muse.core.coord_bus import pull_from_hub |
| 204 | records = [_good_record(i) for i in range(5)] |
| 205 | with patch("muse.core.coord_bus._post_json", |
| 206 | return_value={"records": records, "cursor": 5}): |
| 207 | result = pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 208 | assert len(result["records"]) == 5 |
| 209 | assert result["cursor"] == 5 |
| 210 | |
| 211 | # --- cursor field --- |
| 212 | |
| 213 | def test_null_cursor_raises_coord_bus_error(self) -> None: |
| 214 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 215 | with patch("muse.core.coord_bus._post_json", |
| 216 | return_value={"records": [], "cursor": None}): |
| 217 | with pytest.raises(CoordBusError): |
| 218 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 219 | |
| 220 | @pytest.mark.parametrize("bad_cursor", [ |
| 221 | "malicious", |
| 222 | [], |
| 223 | {}, |
| 224 | "123abc", |
| 225 | ]) |
| 226 | def test_non_integer_cursor_raises_coord_bus_error(self, bad_cursor: MsgpackValue) -> None: |
| 227 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 228 | with patch("muse.core.coord_bus._post_json", |
| 229 | return_value={"records": [], "cursor": bad_cursor}): |
| 230 | with pytest.raises(CoordBusError): |
| 231 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 232 | |
| 233 | def test_negative_cursor_raises_coord_bus_error(self) -> None: |
| 234 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 235 | with patch("muse.core.coord_bus._post_json", |
| 236 | return_value={"records": [], "cursor": -1}): |
| 237 | with pytest.raises(CoordBusError): |
| 238 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 239 | |
| 240 | def test_missing_cursor_key_defaults_to_zero(self) -> None: |
| 241 | from muse.core.coord_bus import pull_from_hub |
| 242 | with patch("muse.core.coord_bus._post_json", |
| 243 | return_value={"records": []}): |
| 244 | result = pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 245 | assert result["cursor"] == 0 |
| 246 | |
| 247 | def test_float_cursor_truncated(self) -> None: |
| 248 | """Float cursor from hub is truncated to int — no crash.""" |
| 249 | from muse.core.coord_bus import pull_from_hub |
| 250 | with patch("muse.core.coord_bus._post_json", |
| 251 | return_value={"records": [], "cursor": 7.9}): |
| 252 | result = pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 253 | assert result["cursor"] == 7 |
| 254 | |
| 255 | def test_zero_cursor_valid(self) -> None: |
| 256 | from muse.core.coord_bus import pull_from_hub |
| 257 | with patch("muse.core.coord_bus._post_json", |
| 258 | return_value={"records": [], "cursor": 0}): |
| 259 | result = pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 260 | assert result["cursor"] == 0 |
| 261 | |
| 262 | # --- both null --- |
| 263 | |
| 264 | def test_both_null_raises_coord_bus_error(self) -> None: |
| 265 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 266 | with patch("muse.core.coord_bus._post_json", |
| 267 | return_value={"records": None, "cursor": None}): |
| 268 | with pytest.raises(CoordBusError): |
| 269 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 270 | |
| 271 | |
| 272 | # ============================================================================= |
| 273 | # 2. INTEGRATION — run_pull with bad hub response |
| 274 | # ============================================================================= |
| 275 | |
| 276 | class TestRunPullNullRecordsIntegration: |
| 277 | """ |
| 278 | Integration: run_pull with pull_from_hub mocked at the boundary. |
| 279 | Asserts clean exit, no unhandled exceptions, correct JSON structure. |
| 280 | """ |
| 281 | |
| 282 | def test_null_records_exits_cleanly_not_crash(self, tmp_path: pathlib.Path) -> None: |
| 283 | code, output = _run_pull_with_hub_response( |
| 284 | tmp_path, {"records": None, "cursor": 0} |
| 285 | ) |
| 286 | assert code != "CRASH", f"run_pull crashed: {output}" |
| 287 | |
| 288 | def test_null_cursor_exits_cleanly_not_crash(self, tmp_path: pathlib.Path) -> None: |
| 289 | code, output = _run_pull_with_hub_response( |
| 290 | tmp_path, {"records": [], "cursor": None} |
| 291 | ) |
| 292 | assert code != "CRASH", f"run_pull crashed: {output}" |
| 293 | |
| 294 | def test_both_null_exits_cleanly_not_crash(self, tmp_path: pathlib.Path) -> None: |
| 295 | code, output = _run_pull_with_hub_response( |
| 296 | tmp_path, {"records": None, "cursor": None} |
| 297 | ) |
| 298 | assert code != "CRASH", f"run_pull crashed: {output}" |
| 299 | |
| 300 | def test_string_records_exits_cleanly_not_crash(self, tmp_path: pathlib.Path) -> None: |
| 301 | code, output = _run_pull_with_hub_response( |
| 302 | tmp_path, {"records": "malicious", "cursor": 0} |
| 303 | ) |
| 304 | assert code != "CRASH", f"run_pull crashed: {output}" |
| 305 | |
| 306 | def test_list_of_strings_exits_cleanly_not_crash(self, tmp_path: pathlib.Path) -> None: |
| 307 | code, output = _run_pull_with_hub_response( |
| 308 | tmp_path, {"records": ["a", "b", "c"], "cursor": 3} |
| 309 | ) |
| 310 | assert code != "CRASH", f"run_pull crashed: {output}" |
| 311 | |
| 312 | def test_null_records_exits_with_code_1(self, tmp_path: pathlib.Path) -> None: |
| 313 | code, output = _run_pull_with_hub_response( |
| 314 | tmp_path, {"records": None, "cursor": 0} |
| 315 | ) |
| 316 | assert code == 1, f"expected exit 1 for null records, got {code!r}" |
| 317 | |
| 318 | def test_bad_response_json_output_has_no_traceback(self, tmp_path: pathlib.Path) -> None: |
| 319 | _, output = _run_pull_with_hub_response( |
| 320 | tmp_path, {"records": None, "cursor": None} |
| 321 | ) |
| 322 | assert "Traceback" not in output |
| 323 | assert "TypeError" not in output |
| 324 | assert "AttributeError" not in output |
| 325 | |
| 326 | def test_good_response_exits_cleanly(self, tmp_path: pathlib.Path) -> None: |
| 327 | records = [_good_record(i) for i in range(3)] |
| 328 | code, output = _run_pull_with_hub_response( |
| 329 | tmp_path, {"records": records, "cursor": 3} |
| 330 | ) |
| 331 | assert code in (0, None), f"expected clean exit, got {code!r}" |
| 332 | |
| 333 | def test_good_response_output_has_correct_count(self, tmp_path: pathlib.Path) -> None: |
| 334 | records = [_good_record(i) for i in range(3)] |
| 335 | code, output = _run_pull_with_hub_response( |
| 336 | tmp_path, {"records": records, "cursor": 3} |
| 337 | ) |
| 338 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 339 | summary = json.loads(lines[-1]) |
| 340 | assert summary["count"] == 3 |
| 341 | assert summary["cursor"] == 3 |
| 342 | |
| 343 | def test_empty_records_exits_cleanly(self, tmp_path: pathlib.Path) -> None: |
| 344 | code, output = _run_pull_with_hub_response( |
| 345 | tmp_path, {"records": [], "cursor": 0} |
| 346 | ) |
| 347 | assert code in (0, None) |
| 348 | |
| 349 | def test_text_mode_null_records_does_not_crash(self, tmp_path: pathlib.Path) -> None: |
| 350 | root = _make_repo(tmp_path) |
| 351 | with patch("muse.core.coord_bus._post_json", |
| 352 | return_value={"records": None, "cursor": 0}), \ |
| 353 | patch("muse.cli.commands.coord_sync.require_repo", return_value=root), \ |
| 354 | patch("muse.cli.commands.coord_sync._resolve_hub_and_signing", |
| 355 | return_value=("https://localhost:1337", "tok")), \ |
| 356 | patch("sys.stdout", io.StringIO()): |
| 357 | args = argparse.Namespace( |
| 358 | owner="torvalds", slug="linux", |
| 359 | json_out=False, hub_url=None, |
| 360 | since_id=0, limit=1000, kinds=[], |
| 361 | ) |
| 362 | try: |
| 363 | from muse.cli.commands.coord_sync import run_pull |
| 364 | run_pull(args) |
| 365 | except SystemExit: |
| 366 | pass |
| 367 | except Exception as exc: |
| 368 | pytest.fail(f"text mode crashed: {type(exc).__name__}: {exc}") |
| 369 | |
| 370 | |
| 371 | # ============================================================================= |
| 372 | # 3. END-TO-END — CLI output is always valid JSON with no exception text |
| 373 | # ============================================================================= |
| 374 | |
| 375 | class TestRunPullNullRecordsEndToEnd: |
| 376 | |
| 377 | @pytest.mark.parametrize("bad_response", [ |
| 378 | {"records": None, "cursor": 0}, |
| 379 | {"records": None, "cursor": None}, |
| 380 | {"records": [], "cursor": None}, |
| 381 | {"records": "malicious", "cursor": 0}, |
| 382 | {"records": 42, "cursor": 0}, |
| 383 | {"records": ["a", "b"], "cursor": 2}, |
| 384 | {"records": None}, |
| 385 | {"cursor": 0}, |
| 386 | {}, |
| 387 | ]) |
| 388 | def test_every_bad_response_produces_valid_json_output(self, tmp_path: pathlib.Path, bad_response: MsgpackDict) -> None: |
| 389 | code, output = _run_pull_with_hub_response(tmp_path, bad_response) |
| 390 | assert code != "CRASH", f"crashed on {bad_response}: {output}" |
| 391 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 392 | assert lines, f"no output for {bad_response}" |
| 393 | for line in lines: |
| 394 | try: |
| 395 | json.loads(line) |
| 396 | except json.JSONDecodeError: |
| 397 | pytest.fail(f"non-JSON output for {bad_response!r}: {line!r}") |
| 398 | |
| 399 | def test_output_never_contains_exception_class_names(self, tmp_path: pathlib.Path) -> None: |
| 400 | for bad in [None, "malicious", [], 42]: |
| 401 | _, output = _run_pull_with_hub_response( |
| 402 | tmp_path, {"records": bad, "cursor": 0} |
| 403 | ) |
| 404 | for forbidden in ("TypeError", "AttributeError", "Traceback", |
| 405 | "most recent call", "ValueError"): |
| 406 | assert forbidden not in output, ( |
| 407 | f"{forbidden!r} leaked for records={bad!r}:\n{output}" |
| 408 | ) |
| 409 | |
| 410 | def test_output_json_schema_complete_on_bad_response(self, tmp_path: pathlib.Path) -> None: |
| 411 | """All required keys present in output even on error.""" |
| 412 | _, output = _run_pull_with_hub_response( |
| 413 | tmp_path, {"records": None, "cursor": 0} |
| 414 | ) |
| 415 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 416 | # At minimum there should be an error line |
| 417 | assert lines, "no output at all" |
| 418 | # The last line must be parseable JSON |
| 419 | summary = json.loads(lines[-1]) |
| 420 | assert isinstance(summary, dict) |
| 421 | |
| 422 | |
| 423 | # ============================================================================= |
| 424 | # 4. STRESS — many pulls mixing good and bad responses |
| 425 | # ============================================================================= |
| 426 | |
| 427 | class TestRunPullNullRecordsStress: |
| 428 | |
| 429 | def _pull(self, tmp_path_subdir: pathlib.Path, hub_response: MsgpackDict) -> tuple[int | str | None, str]: |
| 430 | return _run_pull_with_hub_response(tmp_path_subdir, hub_response) |
| 431 | |
| 432 | def test_50_consecutive_null_records_no_crash(self, tmp_path: pathlib.Path) -> None: |
| 433 | for i in range(50): |
| 434 | code, output = self._pull( |
| 435 | tmp_path / str(i), |
| 436 | {"records": None, "cursor": i} |
| 437 | ) |
| 438 | assert code != "CRASH", f"crashed on iteration {i}: {output}" |
| 439 | |
| 440 | def test_alternating_good_and_null_no_crash(self, tmp_path: pathlib.Path) -> None: |
| 441 | for i in range(20): |
| 442 | if i % 2 == 0: |
| 443 | response = {"records": [_good_record(i)], "cursor": i + 1} |
| 444 | else: |
| 445 | response = {"records": None, "cursor": None} |
| 446 | code, output = self._pull(tmp_path / str(i), response) |
| 447 | assert code != "CRASH", f"crashed on iteration {i}: {output}" |
| 448 | |
| 449 | def test_all_bad_response_types_in_sequence_no_crash(self, tmp_path: pathlib.Path) -> None: |
| 450 | bad_responses = [ |
| 451 | {"records": None, "cursor": 0}, |
| 452 | {"records": None, "cursor": None}, |
| 453 | {"records": "string", "cursor": 0}, |
| 454 | {"records": 42, "cursor": 0}, |
| 455 | {"records": True, "cursor": 0}, |
| 456 | {"records": ["str1", "str2"], "cursor": 2}, |
| 457 | {"records": {}, "cursor": 0}, |
| 458 | {"records": [], "cursor": None}, |
| 459 | {"records": [], "cursor": "bad"}, |
| 460 | {"records": [], "cursor": -1}, |
| 461 | ] |
| 462 | for i, response in enumerate(bad_responses): |
| 463 | code, output = self._pull(tmp_path / str(i), response) |
| 464 | assert code != "CRASH", f"crashed on response {response}: {output}" |
| 465 | assert code == 1, f"expected exit 1 for {response}, got {code!r}" |
| 466 | |
| 467 | |
| 468 | # ============================================================================= |
| 469 | # 5. PERFORMANCE — bad response path overhead is negligible |
| 470 | # ============================================================================= |
| 471 | |
| 472 | class TestRunPullNullRecordsPerformance: |
| 473 | |
| 474 | def _time_pull(self, tmp_path: pathlib.Path, response: MsgpackDict) -> float: |
| 475 | t0 = time.monotonic() |
| 476 | _run_pull_with_hub_response(tmp_path, response) |
| 477 | return time.monotonic() - t0 |
| 478 | |
| 479 | def test_null_response_not_slower_than_good_response(self, tmp_path: pathlib.Path) -> None: |
| 480 | # warm up |
| 481 | self._time_pull(tmp_path / "w1", {"records": [], "cursor": 0}) |
| 482 | self._time_pull(tmp_path / "w2", {"records": None, "cursor": 0}) |
| 483 | |
| 484 | good = self._time_pull(tmp_path / "g", {"records": [], "cursor": 0}) |
| 485 | bad = self._time_pull(tmp_path / "b", {"records": None, "cursor": 0}) |
| 486 | |
| 487 | assert bad < max(good * 10, 0.100), ( |
| 488 | f"null response ({bad:.4f}s) unexpectedly slower than good ({good:.4f}s)" |
| 489 | ) |
| 490 | |
| 491 | def test_50_null_pulls_under_1s(self, tmp_path: pathlib.Path) -> None: |
| 492 | t0 = time.monotonic() |
| 493 | for i in range(50): |
| 494 | _run_pull_with_hub_response( |
| 495 | tmp_path / str(i), {"records": None, "cursor": 0} |
| 496 | ) |
| 497 | elapsed = time.monotonic() - t0 |
| 498 | assert elapsed < 1.0, f"50 null-response pulls took {elapsed:.3f}s (> 1s)" |
| 499 | |
| 500 | |
| 501 | # ============================================================================= |
| 502 | # 6. SECURITY — hub cannot inject arbitrary cursor values into output |
| 503 | # ============================================================================= |
| 504 | |
| 505 | class TestRunPullNullRecordsSecurity: |
| 506 | |
| 507 | @pytest.mark.parametrize("attack_cursor", [ |
| 508 | "__import__('os').system('id')", |
| 509 | "${7*7}", |
| 510 | "{{7*7}}", |
| 511 | "' OR 1=1 --", |
| 512 | "\x00\x01\x02", |
| 513 | "9" * 10000, |
| 514 | "1e308", |
| 515 | "inf", |
| 516 | -9999999999, |
| 517 | ]) |
| 518 | def test_attack_cursor_raises_coord_bus_error_not_exec(self, attack_cursor: str | int) -> None: |
| 519 | from muse.core.coord_bus import pull_from_hub, CoordBusError |
| 520 | with patch("muse.core.coord_bus._post_json", |
| 521 | return_value={"records": [], "cursor": attack_cursor}): |
| 522 | try: |
| 523 | pull_from_hub("https://localhost:1337", "torvalds", "linux") |
| 524 | except CoordBusError: |
| 525 | pass |
| 526 | except Exception as exc: |
| 527 | pytest.fail( |
| 528 | f"Attack cursor {attack_cursor!r} escaped as " |
| 529 | f"{type(exc).__name__}: {exc}" |
| 530 | ) |
| 531 | |
| 532 | def test_null_cursor_never_appears_in_output_as_none_string(self, tmp_path: pathlib.Path) -> None: |
| 533 | """'None' must not appear in CLI output — it means Python None leaked.""" |
| 534 | _, output = _run_pull_with_hub_response( |
| 535 | tmp_path, {"records": [], "cursor": None} |
| 536 | ) |
| 537 | assert '"cursor": null' not in output or True # null is ok in error JSON |
| 538 | # What must NOT happen: the Python repr "None" appearing as a string value |
| 539 | for line in output.strip().splitlines(): |
| 540 | if not line.strip(): |
| 541 | continue |
| 542 | parsed = json.loads(line) |
| 543 | # If cursor appears in output it must be an int or absent |
| 544 | if "cursor" in parsed: |
| 545 | assert isinstance(parsed["cursor"], int) or parsed.get("failed"), ( |
| 546 | f"cursor in output is not an int: {parsed['cursor']!r}" |
| 547 | ) |
| 548 | |
| 549 | def test_extremely_large_cursor_rejected(self, tmp_path: pathlib.Path) -> None: |
| 550 | """Hub claiming cursor=2^63 must not be accepted verbatim.""" |
| 551 | huge = 2**63 |
| 552 | code, output = _run_pull_with_hub_response( |
| 553 | tmp_path, {"records": [], "cursor": huge} |
| 554 | ) |
| 555 | assert code != "CRASH" |
| 556 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 557 | summary = json.loads(lines[-1]) |
| 558 | if not summary.get("failed"): |
| 559 | # If it succeeded, cursor must be sane |
| 560 | assert summary.get("cursor", 0) <= 10**15, ( |
| 561 | f"2^63 cursor accepted verbatim: {summary}" |
| 562 | ) |
| 563 | |
| 564 | |
| 565 | # ============================================================================= |
| 566 | # 7. DATA INTEGRITY — count and cursor in output reflect reality |
| 567 | # ============================================================================= |
| 568 | |
| 569 | class TestRunPullNullRecordsDataIntegrity: |
| 570 | |
| 571 | def test_count_is_zero_when_records_null(self, tmp_path: pathlib.Path) -> None: |
| 572 | """count in output must be 0 (not a crash) when hub returns null records.""" |
| 573 | code, output = _run_pull_with_hub_response( |
| 574 | tmp_path, {"records": None, "cursor": 0} |
| 575 | ) |
| 576 | assert code != "CRASH" |
| 577 | # Output must contain some line indicating failure |
| 578 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 579 | assert lines |
| 580 | |
| 581 | def test_count_equals_len_of_returned_records(self, tmp_path: pathlib.Path) -> None: |
| 582 | records = [_good_record(i) for i in range(7)] |
| 583 | _, output = _run_pull_with_hub_response( |
| 584 | tmp_path, {"records": records, "cursor": 7} |
| 585 | ) |
| 586 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 587 | summary = json.loads(lines[-1]) |
| 588 | assert summary["count"] == 7 |
| 589 | |
| 590 | def test_cursor_in_output_matches_hub_cursor(self, tmp_path: pathlib.Path) -> None: |
| 591 | records = [_good_record(i) for i in range(3)] |
| 592 | _, output = _run_pull_with_hub_response( |
| 593 | tmp_path, {"records": records, "cursor": 42} |
| 594 | ) |
| 595 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 596 | summary = json.loads(lines[-1]) |
| 597 | assert summary["cursor"] == 42 |
| 598 | |
| 599 | def test_cursor_zero_when_no_records(self, tmp_path: pathlib.Path) -> None: |
| 600 | _, output = _run_pull_with_hub_response( |
| 601 | tmp_path, {"records": [], "cursor": 0} |
| 602 | ) |
| 603 | lines = [l for l in output.strip().splitlines() if l.strip()] |
| 604 | summary = json.loads(lines[-1]) |
| 605 | assert summary["cursor"] == 0 |
| 606 | |
| 607 | def test_records_written_to_disk_on_good_response(self, tmp_path: pathlib.Path) -> None: |
| 608 | """Pulled records must actually be written to the remote/ directory.""" |
| 609 | root = _make_repo(tmp_path) |
| 610 | records = [_good_record(i) for i in range(3)] |
| 611 | |
| 612 | with patch("muse.core.coord_bus._post_json", |
| 613 | return_value={"records": records, "cursor": 3}), \ |
| 614 | patch("muse.cli.commands.coord_sync.require_repo", return_value=root), \ |
| 615 | patch("muse.cli.commands.coord_sync._resolve_hub_and_signing", |
| 616 | return_value=("https://localhost:1337", "tok")), \ |
| 617 | patch("sys.stdout", io.StringIO()): |
| 618 | args = argparse.Namespace( |
| 619 | owner="torvalds", slug="linux", |
| 620 | json_out=True, hub_url=None, |
| 621 | since_id=0, limit=1000, kinds=[], |
| 622 | ) |
| 623 | try: |
| 624 | from muse.cli.commands.coord_sync import run_pull |
| 625 | run_pull(args) |
| 626 | except SystemExit: |
| 627 | pass |
| 628 | |
| 629 | remote_dir = coordination_dir(root) / "remote" |
| 630 | written = list(remote_dir.rglob("*.json")) |
| 631 | assert len(written) == 3, f"expected 3 files written, got {len(written)}" |
| 632 | |
| 633 | def test_null_records_writes_nothing_to_disk(self, tmp_path: pathlib.Path) -> None: |
| 634 | """When hub returns null records, nothing must be written to remote/.""" |
| 635 | root = _make_repo(tmp_path) |
| 636 | |
| 637 | with patch("muse.core.coord_bus._post_json", |
| 638 | return_value={"records": None, "cursor": 0}), \ |
| 639 | patch("muse.cli.commands.coord_sync.require_repo", return_value=root), \ |
| 640 | patch("muse.cli.commands.coord_sync._resolve_hub_and_signing", |
| 641 | return_value=("https://localhost:1337", "tok")), \ |
| 642 | patch("sys.stdout", io.StringIO()): |
| 643 | args = argparse.Namespace( |
| 644 | owner="torvalds", slug="linux", |
| 645 | json_out=True, hub_url=None, |
| 646 | since_id=0, limit=1000, kinds=[], |
| 647 | ) |
| 648 | try: |
| 649 | from muse.cli.commands.coord_sync import run_pull |
| 650 | run_pull(args) |
| 651 | except SystemExit: |
| 652 | pass |
| 653 | except Exception: |
| 654 | pass |
| 655 | |
| 656 | remote_dir = coordination_dir(root) / "remote" |
| 657 | written = list(remote_dir.rglob("*.json")) if remote_dir.exists() else [] |
| 658 | assert written == [], f"null records caused {len(written)} files to be written" |
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