test_history.py
python
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
32 days ago
| 1 | """Persistent test-run history indexed by pytest node ID. |
| 2 | |
| 3 | Every time ``muse code test`` executes a test suite it appends a |
| 4 | :class:`RunRecord` to the history. The history is stored in |
| 5 | ``.muse/test_history.msgpack`` — a msgpack-encoded list of run records, |
| 6 | one per ``run_tests`` invocation. |
| 7 | |
| 8 | Why msgpack? |
| 9 | ------------ |
| 10 | * **Consistency** — the rest of the Muse object store uses msgpack. |
| 11 | * **Performance** — msgpack encodes/decodes the history in <5 ms for |
| 12 | 10,000 test-result records. |
| 13 | * **Correctness** — binary-safe, no ambiguous JSON quoting. |
| 14 | |
| 15 | What the history enables |
| 16 | ------------------------ |
| 17 | * **Flaky-test detection** — a test that sometimes passes and sometimes |
| 18 | fails across the last N runs is flagged as flaky. |
| 19 | * **Failure-streak tracking** — how many consecutive runs ended in failure |
| 20 | for a given test? High streaks signal systemic breakage. |
| 21 | * **Duration trend** — is a test getting slower over time? |
| 22 | * **Smart test ordering** — sort slowest/most-recently-failed tests to run |
| 23 | first so failures surface as early as possible in a parallel run. |
| 24 | |
| 25 | Security |
| 26 | -------- |
| 27 | The history file is written atomically (rename-after-write) to prevent |
| 28 | partial writes from corrupting the index. All data originates from pytest |
| 29 | subprocess output (JSON report); no user-supplied data is executed. |
| 30 | """ |
| 31 | |
| 32 | from __future__ import annotations |
| 33 | |
| 34 | import logging |
| 35 | import os |
| 36 | import pathlib |
| 37 | import time |
| 38 | import uuid |
| 39 | from collections.abc import Sequence |
| 40 | from typing import Literal, NotRequired, TypedDict |
| 41 | |
| 42 | import msgpack |
| 43 | |
| 44 | from muse.core._types import MsgpackDict |
| 45 | from muse.core.store import MsgpackValue, _int_val, _str_list, _str_or_none, _str_val, read_msgpack_file |
| 46 | |
| 47 | logger = logging.getLogger(__name__) |
| 48 | |
| 49 | # --------------------------------------------------------------------------- |
| 50 | # Public type definitions |
| 51 | # --------------------------------------------------------------------------- |
| 52 | |
| 53 | Outcome = Literal["passed", "failed", "error", "skipped"] |
| 54 | |
| 55 | |
| 56 | class CaseRecord(TypedDict): |
| 57 | """Result of a single test function within a run.""" |
| 58 | |
| 59 | node_id: str |
| 60 | """Pytest node ID, e.g. ``"tests/test_foo.py::TestBar::test_baz"``.""" |
| 61 | |
| 62 | outcome: Outcome |
| 63 | """Test outcome as reported by pytest.""" |
| 64 | |
| 65 | duration_ms: float |
| 66 | """Wall-clock execution time in milliseconds.""" |
| 67 | |
| 68 | symbol_addresses: list[str] |
| 69 | """Production symbol addresses this test is known to cover (may be empty |
| 70 | when selection did not produce coverage data).""" |
| 71 | |
| 72 | longrepr: NotRequired[str] |
| 73 | """Short failure representation from pytest (omitted when passing).""" |
| 74 | |
| 75 | |
| 76 | class RunRecord(TypedDict): |
| 77 | """A single ``muse code test`` invocation.""" |
| 78 | |
| 79 | run_id: str |
| 80 | """UUID v4 string identifying this specific run.""" |
| 81 | |
| 82 | timestamp: str |
| 83 | """ISO 8601 UTC timestamp of the run start, e.g. ``"2026-03-26T14:05:00Z"``.""" |
| 84 | |
| 85 | commit_id: str | None |
| 86 | """HEAD commit ID at the time of the run, or ``None`` if repo has no commits.""" |
| 87 | |
| 88 | branch: str | None |
| 89 | """Current branch name at run time, or ``None`` for detached HEAD.""" |
| 90 | |
| 91 | results: list[CaseRecord] |
| 92 | """Individual test-case outcomes within this run.""" |
| 93 | |
| 94 | total: int |
| 95 | """Total number of test cases.""" |
| 96 | |
| 97 | passed: int |
| 98 | """Number of passing test cases.""" |
| 99 | |
| 100 | failed: int |
| 101 | """Number of failing test cases.""" |
| 102 | |
| 103 | errored: int |
| 104 | """Number of test cases that raised an unexpected error.""" |
| 105 | |
| 106 | skipped: int |
| 107 | """Number of skipped test cases.""" |
| 108 | |
| 109 | |
| 110 | class HistorySummary(TypedDict): |
| 111 | """Per-test-function aggregated history summary.""" |
| 112 | |
| 113 | node_id: str |
| 114 | """Pytest node ID.""" |
| 115 | |
| 116 | total_runs: int |
| 117 | """Number of times this test has been seen across all recorded runs.""" |
| 118 | |
| 119 | pass_count: int |
| 120 | """Runs where the test passed.""" |
| 121 | |
| 122 | fail_count: int |
| 123 | """Runs where the test failed or errored.""" |
| 124 | |
| 125 | skip_count: int |
| 126 | """Runs where the test was skipped.""" |
| 127 | |
| 128 | flaky: bool |
| 129 | """True when pass_count > 0 **and** fail_count > 0 across recorded runs.""" |
| 130 | |
| 131 | avg_duration_ms: float |
| 132 | """Mean execution time across all non-skipped runs, in milliseconds.""" |
| 133 | |
| 134 | last_outcome: Outcome | None |
| 135 | """Most recent outcome for this test, or ``None`` if never recorded.""" |
| 136 | |
| 137 | last_run_timestamp: str | None |
| 138 | """ISO 8601 timestamp of the most recent run that included this test.""" |
| 139 | |
| 140 | fail_streak: int |
| 141 | """Number of consecutive most-recent runs in which the test failed/errored.""" |
| 142 | |
| 143 | |
| 144 | # --------------------------------------------------------------------------- |
| 145 | # Storage path |
| 146 | # --------------------------------------------------------------------------- |
| 147 | |
| 148 | _HISTORY_FILE = "test_history.msgpack" |
| 149 | _HISTORY_VERSION = 1 |
| 150 | |
| 151 | type _SummaryMap = dict[str, "HistorySummary"] |
| 152 | type _MutableSummaryMap = dict[str, "_MutableSummary"] |
| 153 | |
| 154 | |
| 155 | # --------------------------------------------------------------------------- |
| 156 | # Internal serialisation TypedDicts |
| 157 | # --------------------------------------------------------------------------- |
| 158 | |
| 159 | |
| 160 | class _TestCaseDoc(TypedDict): |
| 161 | """Msgpack document shape for a single test-case result.""" |
| 162 | |
| 163 | node_id: str |
| 164 | outcome: str |
| 165 | duration_ms: float |
| 166 | symbol_addresses: list[str] |
| 167 | longrepr: str |
| 168 | |
| 169 | |
| 170 | class _RunDoc(TypedDict): |
| 171 | """Msgpack document shape for a single run record.""" |
| 172 | |
| 173 | run_id: str |
| 174 | timestamp: str |
| 175 | commit_id: str | None |
| 176 | branch: str | None |
| 177 | total: int |
| 178 | passed: int |
| 179 | failed: int |
| 180 | errored: int |
| 181 | skipped: int |
| 182 | results: list[_TestCaseDoc] |
| 183 | |
| 184 | |
| 185 | class _HistoryDoc(TypedDict): |
| 186 | """Top-level msgpack document shape for the history file.""" |
| 187 | |
| 188 | version: int |
| 189 | runs: list[_RunDoc] |
| 190 | |
| 191 | |
| 192 | def _history_path(root: pathlib.Path) -> pathlib.Path: |
| 193 | """Return the path to the test-history msgpack file inside ``.muse/``.""" |
| 194 | return root / ".muse" / _HISTORY_FILE |
| 195 | |
| 196 | |
| 197 | # --------------------------------------------------------------------------- |
| 198 | # Serialisation helpers |
| 199 | # --------------------------------------------------------------------------- |
| 200 | |
| 201 | |
| 202 | def _record_to_msgpack(record: RunRecord) -> _RunDoc: |
| 203 | """Serialise a :class:`RunRecord` to a msgpack-compatible :class:`_RunDoc`.""" |
| 204 | return _RunDoc( |
| 205 | run_id=record["run_id"], |
| 206 | timestamp=record["timestamp"], |
| 207 | commit_id=record.get("commit_id"), |
| 208 | branch=record.get("branch"), |
| 209 | total=record["total"], |
| 210 | passed=record["passed"], |
| 211 | failed=record["failed"], |
| 212 | errored=record["errored"], |
| 213 | skipped=record["skipped"], |
| 214 | results=[ |
| 215 | _TestCaseDoc( |
| 216 | node_id=r["node_id"], |
| 217 | outcome=r["outcome"], |
| 218 | duration_ms=r["duration_ms"], |
| 219 | symbol_addresses=r["symbol_addresses"], |
| 220 | longrepr=r.get("longrepr", ""), |
| 221 | ) |
| 222 | for r in record["results"] |
| 223 | ], |
| 224 | ) |
| 225 | |
| 226 | |
| 227 | def _record_from_msgpack(raw: MsgpackValue) -> RunRecord | None: |
| 228 | """Deserialise a msgpack-decoded value into a :class:`RunRecord`. |
| 229 | |
| 230 | Returns ``None`` on any structural mismatch so a single corrupt entry |
| 231 | does not abort the entire history load. |
| 232 | """ |
| 233 | if not isinstance(raw, dict): |
| 234 | logger.debug("test_history: skipping non-dict run record") |
| 235 | return None |
| 236 | try: |
| 237 | results: list[CaseRecord] = [] |
| 238 | raw_results = raw.get("results", []) |
| 239 | if not isinstance(raw_results, list): |
| 240 | return None |
| 241 | for r in raw_results: |
| 242 | if not isinstance(r, dict): |
| 243 | continue |
| 244 | r_dict: MsgpackDict = r |
| 245 | node_id = _str_val(r_dict, "node_id", "") |
| 246 | raw_outcome = _str_val(r_dict, "outcome", "error") |
| 247 | if not node_id: |
| 248 | continue |
| 249 | if raw_outcome == "passed": |
| 250 | outcome: Outcome = "passed" |
| 251 | elif raw_outcome == "failed": |
| 252 | outcome = "failed" |
| 253 | elif raw_outcome == "skipped": |
| 254 | outcome = "skipped" |
| 255 | else: |
| 256 | outcome = "error" |
| 257 | longrepr = _str_val(r_dict, "longrepr", "") |
| 258 | duration_raw = r_dict.get("duration_ms", 0.0) |
| 259 | duration_ms = float(duration_raw) if isinstance(duration_raw, (int, float)) else 0.0 |
| 260 | rec = CaseRecord( |
| 261 | node_id=node_id, |
| 262 | outcome=outcome, |
| 263 | duration_ms=duration_ms, |
| 264 | symbol_addresses=_str_list(r_dict, "symbol_addresses"), |
| 265 | ) |
| 266 | if longrepr: |
| 267 | rec["longrepr"] = longrepr |
| 268 | results.append(rec) |
| 269 | |
| 270 | raw_dict: MsgpackDict = raw |
| 271 | |
| 272 | run_id_str = _str_val(raw_dict, "run_id", "") or str(uuid.uuid4()) |
| 273 | timestamp_str = _str_val(raw_dict, "timestamp", "") |
| 274 | commit_id_str = _str_or_none(raw_dict, "commit_id") |
| 275 | branch_str = _str_or_none(raw_dict, "branch") |
| 276 | |
| 277 | return RunRecord( |
| 278 | run_id=run_id_str, |
| 279 | timestamp=timestamp_str, |
| 280 | commit_id=commit_id_str, |
| 281 | branch=branch_str, |
| 282 | results=results, |
| 283 | total=_int_val(raw_dict, "total", len(results)), |
| 284 | passed=_int_val(raw_dict, "passed", 0), |
| 285 | failed=_int_val(raw_dict, "failed", 0), |
| 286 | errored=_int_val(raw_dict, "errored", 0), |
| 287 | skipped=_int_val(raw_dict, "skipped", 0), |
| 288 | ) |
| 289 | except (KeyError, TypeError, ValueError) as exc: |
| 290 | logger.debug("test_history: failed to deserialise run record: %s", exc) |
| 291 | return None |
| 292 | |
| 293 | |
| 294 | # --------------------------------------------------------------------------- |
| 295 | # Public I/O |
| 296 | # --------------------------------------------------------------------------- |
| 297 | |
| 298 | |
| 299 | def load_history(root: pathlib.Path) -> list[RunRecord]: |
| 300 | """Load and return all run records from ``.muse/test_history.msgpack``. |
| 301 | |
| 302 | Returns an empty list if the file does not exist or cannot be parsed. |
| 303 | Individual corrupt records are silently skipped so one bad entry never |
| 304 | prevents history from loading. |
| 305 | """ |
| 306 | path = _history_path(root) |
| 307 | if not path.exists(): |
| 308 | return [] |
| 309 | try: |
| 310 | doc = read_msgpack_file(path) |
| 311 | except Exception as exc: |
| 312 | logger.warning("⚠️ test_history: could not load %s: %s", path, exc) |
| 313 | return [] |
| 314 | |
| 315 | if not isinstance(doc, dict): |
| 316 | return [] |
| 317 | |
| 318 | entries = doc.get("runs", []) |
| 319 | if not isinstance(entries, list): |
| 320 | return [] |
| 321 | |
| 322 | records: list[RunRecord] = [] |
| 323 | for entry in entries: |
| 324 | parsed = _record_from_msgpack(entry) |
| 325 | if parsed is not None: |
| 326 | records.append(parsed) |
| 327 | return records |
| 328 | |
| 329 | |
| 330 | def save_history(root: pathlib.Path, records: list[RunRecord]) -> None: |
| 331 | """Atomically overwrite ``.muse/test_history.msgpack`` with *records*. |
| 332 | |
| 333 | Uses rename-after-write to guarantee the file is never left in a |
| 334 | partially written state. |
| 335 | """ |
| 336 | path = _history_path(root) |
| 337 | path.parent.mkdir(parents=True, exist_ok=True) |
| 338 | |
| 339 | doc = _HistoryDoc( |
| 340 | version=_HISTORY_VERSION, |
| 341 | runs=[_record_to_msgpack(r) for r in records], |
| 342 | ) |
| 343 | packed = msgpack.packb(doc, use_bin_type=True) |
| 344 | |
| 345 | tmp = path.with_suffix(".tmp") |
| 346 | try: |
| 347 | tmp.write_bytes(packed) |
| 348 | os.replace(tmp, path) |
| 349 | except OSError as exc: |
| 350 | logger.error("❌ test_history: failed to write %s: %s", path, exc) |
| 351 | tmp.unlink(missing_ok=True) |
| 352 | raise |
| 353 | |
| 354 | |
| 355 | def append_run(root: pathlib.Path, record: RunRecord) -> None: |
| 356 | """Append a single :class:`RunRecord` to the history. |
| 357 | |
| 358 | Loads the existing history, appends *record*, and saves atomically. |
| 359 | Concurrent appends from parallel workers may interleave; the history is |
| 360 | not a CRDT but the worst-case outcome is a duplicate entry which is |
| 361 | harmless for the analytics use-cases. |
| 362 | """ |
| 363 | records = load_history(root) |
| 364 | records.append(record) |
| 365 | save_history(root, records) |
| 366 | |
| 367 | |
| 368 | def make_run_id() -> str: |
| 369 | """Return a fresh UUID v4 string to identify a new test run.""" |
| 370 | return str(uuid.uuid4()) |
| 371 | |
| 372 | |
| 373 | def iso_now() -> str: |
| 374 | """Return the current UTC time as an ISO 8601 string (seconds precision).""" |
| 375 | t = time.gmtime() |
| 376 | return ( |
| 377 | f"{t.tm_year:04d}-{t.tm_mon:02d}-{t.tm_mday:02d}T" |
| 378 | f"{t.tm_hour:02d}:{t.tm_min:02d}:{t.tm_sec:02d}Z" |
| 379 | ) |
| 380 | |
| 381 | |
| 382 | # --------------------------------------------------------------------------- |
| 383 | # Analytics |
| 384 | # --------------------------------------------------------------------------- |
| 385 | |
| 386 | |
| 387 | def summarize(records: Sequence[RunRecord]) -> _SummaryMap: |
| 388 | """Aggregate *records* into a per-test summary map. |
| 389 | |
| 390 | Args: |
| 391 | records: Run records as returned by :func:`load_history`. |
| 392 | |
| 393 | Returns: |
| 394 | Dict mapping pytest node ID → :class:`HistorySummary`. |
| 395 | """ |
| 396 | summaries: _MutableSummaryMap = {} |
| 397 | |
| 398 | for run in records: |
| 399 | for result in run["results"]: |
| 400 | nid = result["node_id"] |
| 401 | if nid not in summaries: |
| 402 | summaries[nid] = _MutableSummary( |
| 403 | node_id=nid, |
| 404 | outcomes=[], |
| 405 | durations=[], |
| 406 | timestamps=[], |
| 407 | ) |
| 408 | summaries[nid]["outcomes"].append(result["outcome"]) |
| 409 | summaries[nid]["durations"].append(result["duration_ms"]) |
| 410 | summaries[nid]["timestamps"].append(run["timestamp"]) |
| 411 | |
| 412 | out: _SummaryMap = {} |
| 413 | for nid, ms in summaries.items(): |
| 414 | outcomes = ms["outcomes"] |
| 415 | durations = ms["durations"] |
| 416 | timestamps = ms["timestamps"] |
| 417 | |
| 418 | pass_count = sum(1 for o in outcomes if o == "passed") |
| 419 | fail_count = sum(1 for o in outcomes if o in {"failed", "error"}) |
| 420 | skip_count = sum(1 for o in outcomes if o == "skipped") |
| 421 | |
| 422 | non_skip_durations = [ |
| 423 | d for d, o in zip(durations, outcomes) if o != "skipped" |
| 424 | ] |
| 425 | avg_ms = ( |
| 426 | sum(non_skip_durations) / len(non_skip_durations) |
| 427 | if non_skip_durations |
| 428 | else 0.0 |
| 429 | ) |
| 430 | |
| 431 | # Failure streak: count consecutive failures from the most recent run. |
| 432 | streak = 0 |
| 433 | for o in reversed(outcomes): |
| 434 | if o in {"failed", "error"}: |
| 435 | streak += 1 |
| 436 | else: |
| 437 | break |
| 438 | |
| 439 | last_timestamp = timestamps[-1] if timestamps else None |
| 440 | last_outcome: Outcome | None = outcomes[-1] if outcomes else None |
| 441 | |
| 442 | out[nid] = HistorySummary( |
| 443 | node_id=nid, |
| 444 | total_runs=len(outcomes), |
| 445 | pass_count=pass_count, |
| 446 | fail_count=fail_count, |
| 447 | skip_count=skip_count, |
| 448 | flaky=pass_count > 0 and fail_count > 0, |
| 449 | avg_duration_ms=avg_ms, |
| 450 | last_outcome=last_outcome, |
| 451 | last_run_timestamp=last_timestamp, |
| 452 | fail_streak=streak, |
| 453 | ) |
| 454 | |
| 455 | return out |
| 456 | |
| 457 | |
| 458 | def flaky_tests(records: Sequence[RunRecord]) -> list[HistorySummary]: |
| 459 | """Return :class:`HistorySummary` entries for tests that are flaky. |
| 460 | |
| 461 | A test is flaky when it has both at least one pass and at least one |
| 462 | failure across the recorded history. Results are sorted by |
| 463 | ``fail_count`` descending so the most problematic tests appear first. |
| 464 | """ |
| 465 | sums = summarize(records) |
| 466 | flaky = [s for s in sums.values() if s["flaky"]] |
| 467 | flaky.sort(key=lambda s: s["fail_count"], reverse=True) |
| 468 | return flaky |
| 469 | |
| 470 | |
| 471 | def prioritize_targets( |
| 472 | node_ids: list[str], |
| 473 | records: Sequence[RunRecord], |
| 474 | ) -> list[str]: |
| 475 | """Re-order *node_ids* so highest-risk tests run first. |
| 476 | |
| 477 | Risk ordering (highest first): |
| 478 | |
| 479 | 1. Tests with a failure streak > 0 (currently broken). |
| 480 | 2. Tests that are flaky (historically unreliable). |
| 481 | 3. Tests that have never been recorded (unknown risk — run early). |
| 482 | 4. Tests sorted by average duration descending (slow tests surface |
| 483 | failures earlier in a parallel run). |
| 484 | |
| 485 | Returns the same node IDs in a new order. |
| 486 | """ |
| 487 | sums = summarize(records) |
| 488 | |
| 489 | def _sort_key(nid: str) -> tuple[int, int, float]: |
| 490 | s = sums.get(nid) |
| 491 | if s is None: |
| 492 | # Unknown: moderate priority between streaky and healthy. |
| 493 | return (1, 0, 0.0) |
| 494 | streak_score = 0 if s["fail_streak"] == 0 else 2 |
| 495 | flaky_score = 1 if s["flaky"] else 0 |
| 496 | # Negate duration so slower tests come first (we sort ascending). |
| 497 | return (-(streak_score + flaky_score), 0, -s["avg_duration_ms"]) |
| 498 | |
| 499 | return sorted(node_ids, key=_sort_key) |
| 500 | |
| 501 | |
| 502 | # --------------------------------------------------------------------------- |
| 503 | # Internal mutable accumulation type (not exported) |
| 504 | # --------------------------------------------------------------------------- |
| 505 | |
| 506 | |
| 507 | class _MutableSummary(TypedDict): |
| 508 | """Temporary accumulator used inside :func:`summarize`.""" |
| 509 | |
| 510 | node_id: str |
| 511 | outcomes: list[Outcome] |
| 512 | durations: list[float] |
| 513 | timestamps: list[str] |
File History
2 commits
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
32 days ago
sha256:d8316ffae901be06347e16ab55be11868eb519dd16ade3e8aa16a99e662f7e62
baseline: rc14 re-baseline after rc3 store corruption recovery
Human
patch
32 days ago