test_core_symlog.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Tests for muse/core/symlog.py — Phase 1 core storage layer. |
| 2 | |
| 3 | Covers SL_01 through SL_12. |
| 4 | """ |
| 5 | |
| 6 | from __future__ import annotations |
| 7 | |
| 8 | import datetime |
| 9 | import pathlib |
| 10 | import time |
| 11 | |
| 12 | import pytest |
| 13 | |
| 14 | # --------------------------------------------------------------------------- |
| 15 | # Helpers that will be imported once the module exists |
| 16 | # --------------------------------------------------------------------------- |
| 17 | |
| 18 | from muse.core.symlog import ( |
| 19 | NULL_CONTENT_ID, |
| 20 | SymlogEntry, |
| 21 | append_symlog, |
| 22 | delete_symlog_entry, |
| 23 | expire_symlog, |
| 24 | is_null_content_id, |
| 25 | list_symlog_symbols, |
| 26 | list_symlog_symbols_for_file, |
| 27 | read_symlog, |
| 28 | symlog_path, |
| 29 | ) |
| 30 | |
| 31 | # Test fixtures — distinct content IDs (sha256: + 64-hex) |
| 32 | _CID_A = "sha256:" + "a" * 64 |
| 33 | _CID_B = "sha256:" + "b" * 64 |
| 34 | _CID_C = "sha256:" + "c" * 64 |
| 35 | _CID_D = "sha256:" + "d" * 64 |
| 36 | _COMMIT_X = "sha256:" + "1" * 64 |
| 37 | _COMMIT_Y = "sha256:" + "2" * 64 |
| 38 | _COMMIT_Z = "sha256:" + "3" * 64 |
| 39 | |
| 40 | |
| 41 | # =========================================================================== |
| 42 | # SL_11 — NULL_CONTENT_ID and is_null_content_id |
| 43 | # =========================================================================== |
| 44 | |
| 45 | |
| 46 | class TestNullContentId: |
| 47 | def test_null_content_id_format(self) -> None: |
| 48 | """NULL_CONTENT_ID is sha256: followed by 64 zero hex chars.""" |
| 49 | assert NULL_CONTENT_ID == "sha256:" + "0" * 64 |
| 50 | |
| 51 | def test_is_null_content_id_true(self) -> None: |
| 52 | assert is_null_content_id(NULL_CONTENT_ID) is True |
| 53 | |
| 54 | def test_is_null_content_id_false_for_real_id(self) -> None: |
| 55 | assert is_null_content_id(_CID_A) is False |
| 56 | |
| 57 | def test_is_null_content_id_false_for_bare_zeros(self) -> None: |
| 58 | """Bare 64-char zeros without prefix are NOT the null ID.""" |
| 59 | assert is_null_content_id("0" * 64) is False |
| 60 | |
| 61 | |
| 62 | # =========================================================================== |
| 63 | # SL_01 — SymlogEntry dataclass |
| 64 | # =========================================================================== |
| 65 | |
| 66 | |
| 67 | class TestSymlogEntry: |
| 68 | def _make_entry( |
| 69 | self, |
| 70 | old_content_id: str = _CID_A, |
| 71 | new_content_id: str = _CID_B, |
| 72 | commit_id: str = _COMMIT_X, |
| 73 | author: str = "claude-code", |
| 74 | operation: str = "symbol-modified: add rounding", |
| 75 | ts: datetime.datetime | None = None, |
| 76 | ) -> SymlogEntry: |
| 77 | return SymlogEntry( |
| 78 | old_content_id=old_content_id, |
| 79 | new_content_id=new_content_id, |
| 80 | commit_id=commit_id, |
| 81 | author=author, |
| 82 | timestamp=ts or datetime.datetime.now(tz=datetime.timezone.utc), |
| 83 | operation=operation, |
| 84 | ) |
| 85 | |
| 86 | def test_has_seven_fields(self) -> None: |
| 87 | """SymlogEntry exposes the seven required fields.""" |
| 88 | e = self._make_entry() |
| 89 | assert hasattr(e, "old_content_id") |
| 90 | assert hasattr(e, "new_content_id") |
| 91 | assert hasattr(e, "commit_id") |
| 92 | assert hasattr(e, "author") |
| 93 | assert hasattr(e, "timestamp") |
| 94 | assert hasattr(e, "operation") |
| 95 | assert hasattr(e, "born_from") |
| 96 | |
| 97 | def test_frozen(self) -> None: |
| 98 | e = self._make_entry() |
| 99 | with pytest.raises((AttributeError, TypeError)): |
| 100 | e.author = "other" # type: ignore[misc] |
| 101 | |
| 102 | def test_born_from_none_for_regular_operation(self) -> None: |
| 103 | e = self._make_entry(operation="symbol-modified: fix rounding") |
| 104 | assert e.born_from is None |
| 105 | |
| 106 | def test_born_from_parsed_for_born_from_operation(self) -> None: |
| 107 | """born_from is parsed from the operation string.""" |
| 108 | e = self._make_entry(operation="symbol-born-from: src/billing.py::compute_total") |
| 109 | assert e.born_from == "src/billing.py::compute_total" |
| 110 | |
| 111 | def test_born_from_none_for_symbol_created(self) -> None: |
| 112 | e = self._make_entry( |
| 113 | old_content_id=NULL_CONTENT_ID, |
| 114 | operation="symbol-created: compute_total", |
| 115 | ) |
| 116 | assert e.born_from is None |
| 117 | |
| 118 | def test_born_from_none_for_symbol_renamed_to(self) -> None: |
| 119 | """symbol-renamed-to is a terminal entry on the OLD path — born_from stays None.""" |
| 120 | e = self._make_entry( |
| 121 | new_content_id=NULL_CONTENT_ID, |
| 122 | operation="symbol-renamed-to: src/billing.py::compute_invoice_total", |
| 123 | ) |
| 124 | assert e.born_from is None |
| 125 | |
| 126 | def test_born_from_extracts_full_address(self) -> None: |
| 127 | addr = "src/deep/module.py::SomeClass.method_name" |
| 128 | e = self._make_entry(operation=f"symbol-born-from: {addr}") |
| 129 | assert e.born_from == addr |
| 130 | |
| 131 | |
| 132 | # =========================================================================== |
| 133 | # SL_04 — Path encoding |
| 134 | # =========================================================================== |
| 135 | |
| 136 | |
| 137 | class TestPathEncoding: |
| 138 | def test_simple_symbol_maps_correctly(self, tmp_path: pathlib.Path) -> None: |
| 139 | """src/billing.py::compute_total → .muse/symlogs/src/billing.py/compute_total""" |
| 140 | p = symlog_path(tmp_path, "src/billing.py::compute_total") |
| 141 | assert p == tmp_path / ".muse" / "symlogs" / "src" / "billing.py" / "compute_total" |
| 142 | |
| 143 | def test_nested_path_maps_correctly(self, tmp_path: pathlib.Path) -> None: |
| 144 | p = symlog_path(tmp_path, "tests/test_billing.py::test_compute_total") |
| 145 | assert p == tmp_path / ".muse" / "symlogs" / "tests" / "test_billing.py" / "test_compute_total" |
| 146 | |
| 147 | def test_special_chars_in_symbol_name_are_percent_encoded(self, tmp_path: pathlib.Path) -> None: |
| 148 | """Characters outside [a-zA-Z0-9._-] in the symbol name are percent-encoded.""" |
| 149 | p = symlog_path(tmp_path, "src/billing.py::compute total") |
| 150 | leaf = p.name |
| 151 | assert " " not in leaf |
| 152 | assert "%" in leaf # percent-encoded |
| 153 | |
| 154 | def test_colon_in_symbol_name_is_percent_encoded(self, tmp_path: pathlib.Path) -> None: |
| 155 | """The :: separator splits file path from symbol name; colons IN the symbol name are encoded.""" |
| 156 | p = symlog_path(tmp_path, "src/billing.py::SomeClass:method") |
| 157 | leaf = p.name |
| 158 | assert ":" not in leaf |
| 159 | |
| 160 | def test_dotdot_in_file_path_raises(self, tmp_path: pathlib.Path) -> None: |
| 161 | with pytest.raises(ValueError, match=r"\.\.|traversal"): |
| 162 | symlog_path(tmp_path, "../escape.py::bad") |
| 163 | |
| 164 | def test_dotdot_in_symbol_name_raises(self, tmp_path: pathlib.Path) -> None: |
| 165 | with pytest.raises(ValueError, match=r"\.\.|traversal"): |
| 166 | symlog_path(tmp_path, "src/billing.py::../../../etc/passwd") |
| 167 | |
| 168 | def test_missing_double_colon_raises(self, tmp_path: pathlib.Path) -> None: |
| 169 | with pytest.raises(ValueError, match=r"::|separator"): |
| 170 | symlog_path(tmp_path, "src/billing.py") |
| 171 | |
| 172 | def test_empty_symbol_name_raises(self, tmp_path: pathlib.Path) -> None: |
| 173 | with pytest.raises(ValueError, match=r"empty|symbol"): |
| 174 | symlog_path(tmp_path, "src/billing.py::") |
| 175 | |
| 176 | def test_plain_symbol_name_unchanged(self, tmp_path: pathlib.Path) -> None: |
| 177 | """Names with only [a-zA-Z0-9._-] chars pass through unchanged.""" |
| 178 | p = symlog_path(tmp_path, "src/m.py::compute_total-v2.helper") |
| 179 | assert p.name == "compute_total-v2.helper" |
| 180 | |
| 181 | |
| 182 | # =========================================================================== |
| 183 | # Regression: muse#61 — long/multi-byte symbol names must not exceed the |
| 184 | # OS filename limit (255 bytes). Previously this raised |
| 185 | # OSError(ENAMETOOLONG) from append_symlog's open() call, silently caught |
| 186 | # and downgraded to a warning by the caller (muse/cli/commands/commit.py) — |
| 187 | # the commit succeeded but the journal entry for that symbol was |
| 188 | # permanently missing, with no test ever exercising the boundary. |
| 189 | # =========================================================================== |
| 190 | |
| 191 | |
| 192 | class TestLongSymbolNameFilenameLimit: |
| 193 | #: A markdown-heading-shaped symbol name long enough that percent-encoding |
| 194 | #: (each non-safe byte -> 3 chars, "%XX") comfortably exceeds 255 bytes — |
| 195 | #: this is the exact shape that kept recurring in real issue-doc commits. |
| 196 | _LONG_NAME = ( |
| 197 | "Empty-directory sentinel: one root design flaw, four discovered " |
| 198 | "symptoms — comprehensive fix roadmap.Problem statement — the one " |
| 199 | "root flaw behind all four.Blast radius — every place trusting the " |
| 200 | "ambiguous discriminator" |
| 201 | ) |
| 202 | |
| 203 | def test_long_name_produces_filename_within_os_limit(self, tmp_path: pathlib.Path) -> None: |
| 204 | p = symlog_path(tmp_path, f"docs/issue.md::{self._LONG_NAME}") |
| 205 | assert len(p.name.encode("utf-8")) <= 255, ( |
| 206 | f"leaf filename is {len(p.name.encode('utf-8'))} bytes — exceeds " |
| 207 | "the POSIX 255-byte-per-component limit" |
| 208 | ) |
| 209 | |
| 210 | def test_long_name_does_not_raise_on_append(self, tmp_path: pathlib.Path) -> None: |
| 211 | """The exact failure mode from production: append_symlog must not |
| 212 | raise OSError for a long name — it silently corrupted history before.""" |
| 213 | append_symlog( |
| 214 | tmp_path, f"docs/issue.md::{self._LONG_NAME}", |
| 215 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 216 | commit_id=_COMMIT_X, author="claude-code", |
| 217 | operation="symbol-created: long heading", |
| 218 | ) # must not raise |
| 219 | |
| 220 | def test_long_name_round_trips_via_list_symlog_symbols(self, tmp_path: pathlib.Path) -> None: |
| 221 | addr = f"docs/issue.md::{self._LONG_NAME}" |
| 222 | append_symlog( |
| 223 | tmp_path, addr, |
| 224 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 225 | commit_id=_COMMIT_X, author="claude-code", |
| 226 | operation="symbol-created: long heading", |
| 227 | ) |
| 228 | assert addr in list_symlog_symbols(tmp_path) |
| 229 | |
| 230 | def test_long_name_round_trips_via_list_symlog_symbols_for_file( |
| 231 | self, tmp_path: pathlib.Path |
| 232 | ) -> None: |
| 233 | addr = f"docs/issue.md::{self._LONG_NAME}" |
| 234 | append_symlog( |
| 235 | tmp_path, addr, |
| 236 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 237 | commit_id=_COMMIT_X, author="claude-code", |
| 238 | operation="symbol-created: long heading", |
| 239 | ) |
| 240 | assert addr in list_symlog_symbols_for_file(tmp_path, "docs/issue.md") |
| 241 | |
| 242 | def test_long_name_history_readable_via_read_symlog(self, tmp_path: pathlib.Path) -> None: |
| 243 | """The actual end-user-visible symptom: muse code symbol-log must |
| 244 | return real history for a long-name symbol, not silently nothing.""" |
| 245 | addr = f"docs/issue.md::{self._LONG_NAME}" |
| 246 | append_symlog( |
| 247 | tmp_path, addr, |
| 248 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 249 | commit_id=_COMMIT_X, author="claude-code", |
| 250 | operation="symbol-created: long heading", |
| 251 | ) |
| 252 | entries = read_symlog(tmp_path, addr, limit=10) |
| 253 | assert len(entries) == 1 |
| 254 | assert entries[0].new_content_id == _CID_A |
| 255 | |
| 256 | def test_two_different_long_names_do_not_collide(self, tmp_path: pathlib.Path) -> None: |
| 257 | addr_a = f"docs/issue.md::{self._LONG_NAME} A" |
| 258 | addr_b = f"docs/issue.md::{self._LONG_NAME} B" |
| 259 | append_symlog( |
| 260 | tmp_path, addr_a, |
| 261 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 262 | commit_id=_COMMIT_X, author="claude-code", |
| 263 | operation="symbol-created: A", |
| 264 | ) |
| 265 | append_symlog( |
| 266 | tmp_path, addr_b, |
| 267 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_B, |
| 268 | commit_id=_COMMIT_X, author="claude-code", |
| 269 | operation="symbol-created: B", |
| 270 | ) |
| 271 | result = list_symlog_symbols_for_file(tmp_path, "docs/issue.md") |
| 272 | assert addr_a in result |
| 273 | assert addr_b in result |
| 274 | assert len(result) == 2 |
| 275 | |
| 276 | def test_boundary_at_255_bytes_encoded(self, tmp_path: pathlib.Path) -> None: |
| 277 | """A symbol name whose plain-ASCII percent-encoded form lands exactly |
| 278 | at 255 bytes must still produce a valid, round-trippable filename.""" |
| 279 | # Plain ASCII safe chars pass through 1:1, so an N-char safe-ASCII |
| 280 | # name encodes to exactly N bytes. |
| 281 | name_255 = "x" * 255 |
| 282 | addr = f"docs/issue.md::{name_255}" |
| 283 | append_symlog( |
| 284 | tmp_path, addr, |
| 285 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 286 | commit_id=_COMMIT_X, author="claude-code", |
| 287 | operation="symbol-created: boundary", |
| 288 | ) |
| 289 | assert addr in list_symlog_symbols_for_file(tmp_path, "docs/issue.md") |
| 290 | |
| 291 | def test_short_name_unaffected_by_long_name_fix(self, tmp_path: pathlib.Path) -> None: |
| 292 | """Existing short-name behavior (direct percent-encoding, no hashing, |
| 293 | no index file) must be completely unchanged.""" |
| 294 | p = symlog_path(tmp_path, "src/billing.py::compute_total") |
| 295 | assert p.name == "compute_total" |
| 296 | |
| 297 | |
| 298 | # =========================================================================== |
| 299 | # SL_02 + SL_10 — append_symlog and sanitization |
| 300 | # =========================================================================== |
| 301 | |
| 302 | |
| 303 | class TestAppendSymlog: |
| 304 | def test_creates_parent_dirs_and_file(self, tmp_path: pathlib.Path) -> None: |
| 305 | append_symlog( |
| 306 | tmp_path, |
| 307 | "src/billing.py::compute_total", |
| 308 | old_content_id=NULL_CONTENT_ID, |
| 309 | new_content_id=_CID_A, |
| 310 | commit_id=_COMMIT_X, |
| 311 | author="claude-code", |
| 312 | operation="symbol-created: compute_total", |
| 313 | ) |
| 314 | p = symlog_path(tmp_path, "src/billing.py::compute_total") |
| 315 | assert p.exists() |
| 316 | assert p.is_file() |
| 317 | |
| 318 | def test_one_line_written_per_call(self, tmp_path: pathlib.Path) -> None: |
| 319 | addr = "src/billing.py::compute_total" |
| 320 | for i in range(3): |
| 321 | append_symlog( |
| 322 | tmp_path, addr, |
| 323 | old_content_id=_CID_A, |
| 324 | new_content_id=_CID_B, |
| 325 | commit_id=_COMMIT_X, |
| 326 | author="claude-code", |
| 327 | operation=f"symbol-modified: change {i}", |
| 328 | ) |
| 329 | p = symlog_path(tmp_path, addr) |
| 330 | lines = [l for l in p.read_text().splitlines() if l.strip()] |
| 331 | assert len(lines) == 3 |
| 332 | |
| 333 | def test_line_format(self, tmp_path: pathlib.Path) -> None: |
| 334 | """Each line is tab-delimited between metadata and operation.""" |
| 335 | addr = "src/billing.py::compute_total" |
| 336 | append_symlog( |
| 337 | tmp_path, addr, |
| 338 | old_content_id=_CID_A, |
| 339 | new_content_id=_CID_B, |
| 340 | commit_id=_COMMIT_X, |
| 341 | author="claude-code", |
| 342 | operation="symbol-modified: add rounding", |
| 343 | ) |
| 344 | p = symlog_path(tmp_path, addr) |
| 345 | line = p.read_text().strip() |
| 346 | assert "\t" in line |
| 347 | meta, op = line.split("\t", 1) |
| 348 | tokens = meta.split() |
| 349 | # old_content_id new_content_id commit_id author ts tz |
| 350 | assert tokens[0] == _CID_A |
| 351 | assert tokens[1] == _CID_B |
| 352 | assert tokens[2] == _COMMIT_X |
| 353 | assert op == "symbol-modified: add rounding" |
| 354 | |
| 355 | def test_author_newline_stripped(self, tmp_path: pathlib.Path) -> None: |
| 356 | """SL_10: newlines in author are stripped before write.""" |
| 357 | addr = "src/billing.py::compute_total" |
| 358 | append_symlog( |
| 359 | tmp_path, addr, |
| 360 | old_content_id=_CID_A, new_content_id=_CID_B, |
| 361 | commit_id=_COMMIT_X, |
| 362 | author="bad\nauthor", |
| 363 | operation="symbol-modified: x", |
| 364 | ) |
| 365 | entries = read_symlog(tmp_path, addr) |
| 366 | assert "\n" not in entries[0].author |
| 367 | |
| 368 | def test_author_tab_stripped(self, tmp_path: pathlib.Path) -> None: |
| 369 | addr = "src/billing.py::compute_total" |
| 370 | append_symlog( |
| 371 | tmp_path, addr, |
| 372 | old_content_id=_CID_A, new_content_id=_CID_B, |
| 373 | commit_id=_COMMIT_X, |
| 374 | author="bad\tauthor", |
| 375 | operation="symbol-modified: x", |
| 376 | ) |
| 377 | entries = read_symlog(tmp_path, addr) |
| 378 | assert "\t" not in entries[0].author |
| 379 | |
| 380 | def test_operation_newline_stripped(self, tmp_path: pathlib.Path) -> None: |
| 381 | """SL_10: newlines in operation are stripped to prevent line injection.""" |
| 382 | addr = "src/billing.py::compute_total" |
| 383 | append_symlog( |
| 384 | tmp_path, addr, |
| 385 | old_content_id=_CID_A, new_content_id=_CID_B, |
| 386 | commit_id=_COMMIT_X, |
| 387 | author="claude-code", |
| 388 | operation="symbol-modified: injected\nfake entry", |
| 389 | ) |
| 390 | entries = read_symlog(tmp_path, addr) |
| 391 | assert "\n" not in entries[0].operation |
| 392 | |
| 393 | def test_different_symbols_same_file_get_separate_files(self, tmp_path: pathlib.Path) -> None: |
| 394 | for name in ("compute_total", "validate_invoice"): |
| 395 | append_symlog( |
| 396 | tmp_path, f"src/billing.py::{name}", |
| 397 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 398 | commit_id=_COMMIT_X, author="claude-code", |
| 399 | operation=f"symbol-created: {name}", |
| 400 | ) |
| 401 | p1 = symlog_path(tmp_path, "src/billing.py::compute_total") |
| 402 | p2 = symlog_path(tmp_path, "src/billing.py::validate_invoice") |
| 403 | assert p1.exists() |
| 404 | assert p2.exists() |
| 405 | assert p1 != p2 |
| 406 | |
| 407 | |
| 408 | # =========================================================================== |
| 409 | # SL_03 + SL_09 — read_symlog (newest-first, limit, follow, size cap) |
| 410 | # =========================================================================== |
| 411 | |
| 412 | |
| 413 | class TestReadSymlog: |
| 414 | def _write_n_entries( |
| 415 | self, |
| 416 | tmp_path: pathlib.Path, |
| 417 | addr: str, |
| 418 | n: int, |
| 419 | *, |
| 420 | old_cid: str = _CID_A, |
| 421 | new_cid: str = _CID_B, |
| 422 | ) -> None: |
| 423 | for i in range(n): |
| 424 | append_symlog( |
| 425 | tmp_path, addr, |
| 426 | old_content_id=old_cid, new_content_id=new_cid, |
| 427 | commit_id=_COMMIT_X, author="claude-code", |
| 428 | operation=f"symbol-modified: change {i}", |
| 429 | ) |
| 430 | |
| 431 | def test_empty_log_returns_empty_list(self, tmp_path: pathlib.Path) -> None: |
| 432 | entries = read_symlog(tmp_path, "src/billing.py::compute_total") |
| 433 | assert entries == [] |
| 434 | |
| 435 | def test_entries_returned_newest_first(self, tmp_path: pathlib.Path) -> None: |
| 436 | addr = "src/billing.py::compute_total" |
| 437 | append_symlog(tmp_path, addr, _CID_A, _CID_B, _COMMIT_X, "claude-code", "symbol-modified: first") |
| 438 | time.sleep(0.01) |
| 439 | append_symlog(tmp_path, addr, _CID_B, _CID_C, _COMMIT_Y, "claude-code", "symbol-modified: second") |
| 440 | entries = read_symlog(tmp_path, addr) |
| 441 | assert entries[0].operation == "symbol-modified: second" |
| 442 | assert entries[1].operation == "symbol-modified: first" |
| 443 | |
| 444 | def test_limit_respected(self, tmp_path: pathlib.Path) -> None: |
| 445 | addr = "src/billing.py::compute_total" |
| 446 | self._write_n_entries(tmp_path, addr, 10) |
| 447 | entries = read_symlog(tmp_path, addr, limit=3) |
| 448 | assert len(entries) == 3 |
| 449 | |
| 450 | def test_all_fields_round_trip(self, tmp_path: pathlib.Path) -> None: |
| 451 | addr = "src/billing.py::compute_total" |
| 452 | append_symlog( |
| 453 | tmp_path, addr, |
| 454 | old_content_id=_CID_A, new_content_id=_CID_B, |
| 455 | commit_id=_COMMIT_X, author="claude-code", |
| 456 | operation="symbol-modified: round-trip test", |
| 457 | ) |
| 458 | entries = read_symlog(tmp_path, addr) |
| 459 | assert len(entries) == 1 |
| 460 | e = entries[0] |
| 461 | assert e.old_content_id == _CID_A |
| 462 | assert e.new_content_id == _CID_B |
| 463 | assert e.commit_id == _COMMIT_X |
| 464 | assert e.author == "claude-code" |
| 465 | assert e.operation == "symbol-modified: round-trip test" |
| 466 | assert isinstance(e.timestamp, datetime.datetime) |
| 467 | assert e.timestamp.tzinfo is not None |
| 468 | |
| 469 | def test_born_from_parsed_on_read(self, tmp_path: pathlib.Path) -> None: |
| 470 | addr = "src/billing.py::compute_invoice_total" |
| 471 | old_addr = "src/billing.py::compute_total" |
| 472 | append_symlog( |
| 473 | tmp_path, addr, |
| 474 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_B, |
| 475 | commit_id=_COMMIT_X, author="claude-code", |
| 476 | operation=f"symbol-born-from: {old_addr}", |
| 477 | ) |
| 478 | entries = read_symlog(tmp_path, addr) |
| 479 | assert entries[0].born_from == old_addr |
| 480 | |
| 481 | def test_follow_traverses_rename_chain(self, tmp_path: pathlib.Path) -> None: |
| 482 | """SL_03: follow=True reads prior symbol's log when oldest entry is born-from.""" |
| 483 | old_addr = "src/billing.py::compute_total" |
| 484 | new_addr = "src/billing.py::compute_invoice_total" |
| 485 | |
| 486 | # Write two entries in old symbol's log |
| 487 | append_symlog(tmp_path, old_addr, NULL_CONTENT_ID, _CID_A, _COMMIT_X, "claude-code", "symbol-created: compute_total") |
| 488 | append_symlog(tmp_path, old_addr, _CID_A, _CID_B, _COMMIT_Y, "claude-code", "symbol-modified: improve") |
| 489 | # Terminal rename entry on old path |
| 490 | append_symlog(tmp_path, old_addr, _CID_B, NULL_CONTENT_ID, _COMMIT_Z, "claude-code", f"symbol-renamed-to: {new_addr}") |
| 491 | # Born-from entry on new path |
| 492 | append_symlog(tmp_path, new_addr, NULL_CONTENT_ID, _CID_C, _COMMIT_Z, "claude-code", f"symbol-born-from: {old_addr}") |
| 493 | |
| 494 | entries = read_symlog(tmp_path, new_addr, follow=True) |
| 495 | # Should have new entry + old entries (created, modified, renamed-to) |
| 496 | assert len(entries) >= 4 |
| 497 | # The new_addr entry should be first (newest) |
| 498 | assert entries[0].operation.startswith("symbol-born-from:") |
| 499 | |
| 500 | def test_file_size_cap_warns_and_returns(self, tmp_path: pathlib.Path, caplog: pytest.LogCaptureFixture) -> None: |
| 501 | """SL_09: files over _MAX_SYMLOG_BYTES trigger a warning.""" |
| 502 | import muse.core.symlog as symlog_mod |
| 503 | addr = "src/billing.py::compute_total" |
| 504 | p = symlog_path(tmp_path, addr) |
| 505 | p.parent.mkdir(parents=True, exist_ok=True) |
| 506 | |
| 507 | original_cap = symlog_mod._MAX_SYMLOG_BYTES |
| 508 | try: |
| 509 | symlog_mod._MAX_SYMLOG_BYTES = 10 # tiny cap for test |
| 510 | p.write_text("x" * 20) |
| 511 | import logging |
| 512 | with caplog.at_level(logging.WARNING, logger="muse.core.symlog"): |
| 513 | result = read_symlog(tmp_path, addr) |
| 514 | assert any("MiB" in r.message or "cap" in r.message.lower() or "exceeds" in r.message.lower() for r in caplog.records) |
| 515 | finally: |
| 516 | symlog_mod._MAX_SYMLOG_BYTES = original_cap |
| 517 | |
| 518 | def test_follow_false_does_not_traverse(self, tmp_path: pathlib.Path) -> None: |
| 519 | """follow=False (default) reads only the requested symbol's log.""" |
| 520 | old_addr = "src/billing.py::compute_total" |
| 521 | new_addr = "src/billing.py::compute_invoice_total" |
| 522 | append_symlog(tmp_path, old_addr, NULL_CONTENT_ID, _CID_A, _COMMIT_X, "claude-code", "symbol-created: compute_total") |
| 523 | append_symlog(tmp_path, new_addr, NULL_CONTENT_ID, _CID_B, _COMMIT_Y, "claude-code", f"symbol-born-from: {old_addr}") |
| 524 | entries = read_symlog(tmp_path, new_addr, follow=False) |
| 525 | assert len(entries) == 1 # only the new_addr entry, no old_addr entries |
| 526 | |
| 527 | |
| 528 | # =========================================================================== |
| 529 | # SL_05 + SL_06 — list_symlog_symbols and list_symlog_symbols_for_file |
| 530 | # =========================================================================== |
| 531 | |
| 532 | |
| 533 | class TestListSymlogSymbols: |
| 534 | def _write(self, tmp_path: pathlib.Path, addr: str) -> None: |
| 535 | append_symlog( |
| 536 | tmp_path, addr, |
| 537 | old_content_id=NULL_CONTENT_ID, new_content_id=_CID_A, |
| 538 | commit_id=_COMMIT_X, author="claude-code", |
| 539 | operation="symbol-created: x", |
| 540 | ) |
| 541 | |
| 542 | def test_empty_repo_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 543 | assert list_symlog_symbols(tmp_path) == [] |
| 544 | |
| 545 | def test_returns_all_symbol_addresses_sorted(self, tmp_path: pathlib.Path) -> None: |
| 546 | self._write(tmp_path, "src/billing.py::compute_total") |
| 547 | self._write(tmp_path, "src/billing.py::validate_invoice") |
| 548 | self._write(tmp_path, "src/auth.py::login") |
| 549 | result = list_symlog_symbols(tmp_path) |
| 550 | assert result == sorted(result) |
| 551 | assert "src/billing.py::compute_total" in result |
| 552 | assert "src/billing.py::validate_invoice" in result |
| 553 | assert "src/auth.py::login" in result |
| 554 | |
| 555 | def test_skips_symlinks(self, tmp_path: pathlib.Path) -> None: |
| 556 | self._write(tmp_path, "src/billing.py::compute_total") |
| 557 | real_p = symlog_path(tmp_path, "src/billing.py::compute_total") |
| 558 | link_p = real_p.parent / "symlink_fn" |
| 559 | link_p.symlink_to(real_p) |
| 560 | result = list_symlog_symbols(tmp_path) |
| 561 | # Only the real file — symlink excluded |
| 562 | decoded = [r.split("::")[-1] for r in result] |
| 563 | assert "symlink_fn" not in decoded |
| 564 | |
| 565 | def test_list_for_file_returns_only_that_files_symbols(self, tmp_path: pathlib.Path) -> None: |
| 566 | self._write(tmp_path, "src/billing.py::compute_total") |
| 567 | self._write(tmp_path, "src/billing.py::validate_invoice") |
| 568 | self._write(tmp_path, "src/auth.py::login") |
| 569 | result = list_symlog_symbols_for_file(tmp_path, "src/billing.py") |
| 570 | assert set(result) == { |
| 571 | "src/billing.py::compute_total", |
| 572 | "src/billing.py::validate_invoice", |
| 573 | } |
| 574 | assert "src/auth.py::login" not in result |
| 575 | |
| 576 | def test_list_for_file_empty_dir_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 577 | result = list_symlog_symbols_for_file(tmp_path, "src/billing.py") |
| 578 | assert result == [] |
| 579 | |
| 580 | def test_list_for_file_skips_symlinks(self, tmp_path: pathlib.Path) -> None: |
| 581 | self._write(tmp_path, "src/billing.py::compute_total") |
| 582 | real_p = symlog_path(tmp_path, "src/billing.py::compute_total") |
| 583 | link_p = real_p.parent / "symlink_fn" |
| 584 | link_p.symlink_to(real_p) |
| 585 | result = list_symlog_symbols_for_file(tmp_path, "src/billing.py") |
| 586 | decoded = [r.split("::")[-1] for r in result] |
| 587 | assert "symlink_fn" not in decoded |
| 588 | |
| 589 | |
| 590 | # =========================================================================== |
| 591 | # SL_07 — expire_symlog |
| 592 | # =========================================================================== |
| 593 | |
| 594 | |
| 595 | class TestExpireSymlog: |
| 596 | def _write_old(self, tmp_path: pathlib.Path, addr: str, age_days: float = 100.0) -> None: |
| 597 | """Write an entry with a timestamp in the past.""" |
| 598 | p = symlog_path(tmp_path, addr) |
| 599 | p.parent.mkdir(parents=True, exist_ok=True) |
| 600 | old_ts = int(time.time()) - int(age_days * 86400) |
| 601 | line = f"{_CID_A} {_CID_B} {_COMMIT_X} claude-code {old_ts} +0000\tsymbol-modified: old\n" |
| 602 | with p.open("a") as fh: |
| 603 | fh.write(line) |
| 604 | |
| 605 | def _write_recent(self, tmp_path: pathlib.Path, addr: str) -> None: |
| 606 | append_symlog( |
| 607 | tmp_path, addr, |
| 608 | old_content_id=_CID_A, new_content_id=_CID_C, |
| 609 | commit_id=_COMMIT_Y, author="claude-code", |
| 610 | operation="symbol-modified: recent", |
| 611 | ) |
| 612 | |
| 613 | def test_no_log_file_returns_zero_zero(self, tmp_path: pathlib.Path) -> None: |
| 614 | expired, kept = expire_symlog(tmp_path, "src/billing.py::compute_total", expire_days=90) |
| 615 | assert (expired, kept) == (0, 0) |
| 616 | |
| 617 | def test_old_entries_expired(self, tmp_path: pathlib.Path) -> None: |
| 618 | addr = "src/billing.py::compute_total" |
| 619 | self._write_old(tmp_path, addr, age_days=100) |
| 620 | self._write_recent(tmp_path, addr) |
| 621 | expired, kept = expire_symlog(tmp_path, addr, expire_days=90) |
| 622 | assert expired == 1 |
| 623 | assert kept == 1 |
| 624 | |
| 625 | def test_all_expired_file_deleted(self, tmp_path: pathlib.Path) -> None: |
| 626 | addr = "src/billing.py::compute_total" |
| 627 | self._write_old(tmp_path, addr, age_days=200) |
| 628 | expire_symlog(tmp_path, addr, expire_days=90) |
| 629 | assert not symlog_path(tmp_path, addr).exists() |
| 630 | |
| 631 | def test_dry_run_writes_nothing(self, tmp_path: pathlib.Path) -> None: |
| 632 | addr = "src/billing.py::compute_total" |
| 633 | self._write_old(tmp_path, addr, age_days=200) |
| 634 | expired, kept = expire_symlog(tmp_path, addr, expire_days=90, dry_run=True) |
| 635 | assert expired == 1 |
| 636 | assert symlog_path(tmp_path, addr).exists() # still there |
| 637 | |
| 638 | def test_atomic_write(self, tmp_path: pathlib.Path) -> None: |
| 639 | """After expire the .lock temp file must not exist.""" |
| 640 | addr = "src/billing.py::compute_total" |
| 641 | self._write_old(tmp_path, addr, age_days=200) |
| 642 | self._write_recent(tmp_path, addr) |
| 643 | expire_symlog(tmp_path, addr, expire_days=90) |
| 644 | lock = symlog_path(tmp_path, addr).with_suffix(".lock") |
| 645 | assert not lock.exists() |
| 646 | |
| 647 | def test_recent_entries_kept(self, tmp_path: pathlib.Path) -> None: |
| 648 | addr = "src/billing.py::compute_total" |
| 649 | self._write_recent(tmp_path, addr) |
| 650 | expired, kept = expire_symlog(tmp_path, addr, expire_days=90) |
| 651 | assert expired == 0 |
| 652 | assert kept == 1 |
| 653 | assert symlog_path(tmp_path, addr).exists() |
| 654 | |
| 655 | |
| 656 | # =========================================================================== |
| 657 | # SL_08 — delete_symlog_entry |
| 658 | # =========================================================================== |
| 659 | |
| 660 | |
| 661 | class TestDeleteSymlogEntry: |
| 662 | def _write_entries(self, tmp_path: pathlib.Path, addr: str, n: int) -> None: |
| 663 | for i in range(n): |
| 664 | append_symlog( |
| 665 | tmp_path, addr, |
| 666 | old_content_id=_CID_A, new_content_id=_CID_B, |
| 667 | commit_id=_COMMIT_X, author="claude-code", |
| 668 | operation=f"symbol-modified: entry {i}", |
| 669 | ) |
| 670 | |
| 671 | def test_delete_all_removes_file(self, tmp_path: pathlib.Path) -> None: |
| 672 | addr = "src/billing.py::compute_total" |
| 673 | self._write_entries(tmp_path, addr, 3) |
| 674 | deleted, remaining = delete_symlog_entry(tmp_path, addr, index=None) |
| 675 | assert deleted == 3 |
| 676 | assert remaining == 0 |
| 677 | assert not symlog_path(tmp_path, addr).exists() |
| 678 | |
| 679 | def test_delete_index_zero_removes_newest(self, tmp_path: pathlib.Path) -> None: |
| 680 | addr = "src/billing.py::compute_total" |
| 681 | self._write_entries(tmp_path, addr, 3) |
| 682 | deleted, remaining = delete_symlog_entry(tmp_path, addr, index=0) |
| 683 | assert deleted == 1 |
| 684 | assert remaining == 2 |
| 685 | entries = read_symlog(tmp_path, addr) |
| 686 | assert len(entries) == 2 |
| 687 | # @{0} was entry 2 (newest), so now entries 0 and 1 remain |
| 688 | assert all("entry 2" not in e.operation for e in entries) |
| 689 | |
| 690 | def test_delete_middle_entry(self, tmp_path: pathlib.Path) -> None: |
| 691 | addr = "src/billing.py::compute_total" |
| 692 | self._write_entries(tmp_path, addr, 3) |
| 693 | delete_symlog_entry(tmp_path, addr, index=1) |
| 694 | entries = read_symlog(tmp_path, addr) |
| 695 | assert len(entries) == 2 |
| 696 | ops = [e.operation for e in entries] |
| 697 | # Middle entry (index 1, which was "entry 1" in the file) is gone |
| 698 | assert not any("entry 1" in op for op in ops) |
| 699 | |
| 700 | def test_out_of_bounds_raises_index_error(self, tmp_path: pathlib.Path) -> None: |
| 701 | addr = "src/billing.py::compute_total" |
| 702 | self._write_entries(tmp_path, addr, 3) |
| 703 | with pytest.raises(IndexError): |
| 704 | delete_symlog_entry(tmp_path, addr, index=99) |
| 705 | |
| 706 | def test_missing_log_all_is_graceful(self, tmp_path: pathlib.Path) -> None: |
| 707 | deleted, remaining = delete_symlog_entry(tmp_path, "src/billing.py::compute_total", index=None) |
| 708 | assert deleted == 0 |
| 709 | assert remaining == 0 |
| 710 | |
| 711 | def test_missing_log_by_index_raises(self, tmp_path: pathlib.Path) -> None: |
| 712 | with pytest.raises(FileNotFoundError): |
| 713 | delete_symlog_entry(tmp_path, "src/billing.py::compute_total", index=0) |
| 714 | |
| 715 | def test_atomic_write_no_lock_file_left(self, tmp_path: pathlib.Path) -> None: |
| 716 | addr = "src/billing.py::compute_total" |
| 717 | self._write_entries(tmp_path, addr, 3) |
| 718 | delete_symlog_entry(tmp_path, addr, index=0) |
| 719 | lock = symlog_path(tmp_path, addr).with_suffix(".lock") |
| 720 | assert not lock.exists() |
| 721 | |
| 722 | def test_delete_last_entry_removes_file(self, tmp_path: pathlib.Path) -> None: |
| 723 | addr = "src/billing.py::compute_total" |
| 724 | self._write_entries(tmp_path, addr, 1) |
| 725 | deleted, remaining = delete_symlog_entry(tmp_path, addr, index=0) |
| 726 | assert deleted == 1 |
| 727 | assert remaining == 0 |
| 728 | assert not symlog_path(tmp_path, addr).exists() |
| 729 | |
| 730 | |
| 731 | # =========================================================================== |
| 732 | # SL_12 — Integration test |
| 733 | # =========================================================================== |
| 734 | |
| 735 | |
| 736 | class TestIntegration: |
| 737 | def test_five_entries_round_trip_newest_first(self, tmp_path: pathlib.Path) -> None: |
| 738 | """SL_12: Write 5 entries, read back newest-first, verify all fields round-trip.""" |
| 739 | addr = "src/billing.py::compute_total" |
| 740 | ops = [f"symbol-modified: change {i}" for i in range(5)] |
| 741 | for op in ops: |
| 742 | append_symlog( |
| 743 | tmp_path, addr, |
| 744 | old_content_id=_CID_A, |
| 745 | new_content_id=_CID_B, |
| 746 | commit_id=_COMMIT_X, |
| 747 | author="claude-code", |
| 748 | operation=op, |
| 749 | ) |
| 750 | entries = read_symlog(tmp_path, addr, limit=100) |
| 751 | assert len(entries) == 5 |
| 752 | # Newest first — last written operation is at index 0 |
| 753 | assert entries[0].operation == "symbol-modified: change 4" |
| 754 | assert entries[4].operation == "symbol-modified: change 0" |
| 755 | for e in entries: |
| 756 | assert e.old_content_id == _CID_A |
| 757 | assert e.new_content_id == _CID_B |
| 758 | assert e.commit_id == _COMMIT_X |
| 759 | assert e.author == "claude-code" |
| 760 | assert e.born_from is None |
| 761 | |
| 762 | def test_two_symbols_same_file_independent(self, tmp_path: pathlib.Path) -> None: |
| 763 | """SL_12: Two symbols in same file; list_symlog_symbols_for_file returns both.""" |
| 764 | addr1 = "src/billing.py::compute_total" |
| 765 | addr2 = "src/billing.py::validate_invoice" |
| 766 | append_symlog(tmp_path, addr1, NULL_CONTENT_ID, _CID_A, _COMMIT_X, "claude-code", "symbol-created: compute_total") |
| 767 | append_symlog(tmp_path, addr2, NULL_CONTENT_ID, _CID_B, _COMMIT_Y, "claude-code", "symbol-created: validate_invoice") |
| 768 | |
| 769 | syms = list_symlog_symbols_for_file(tmp_path, "src/billing.py") |
| 770 | assert addr1 in syms |
| 771 | assert addr2 in syms |
| 772 | assert len(syms) == 2 |
| 773 | |
| 774 | e1 = read_symlog(tmp_path, addr1) |
| 775 | e2 = read_symlog(tmp_path, addr2) |
| 776 | assert len(e1) == 1 |
| 777 | assert len(e2) == 1 |
| 778 | assert e1[0].new_content_id == _CID_A |
| 779 | assert e2[0].new_content_id == _CID_B |
| 780 | |
| 781 | def test_created_deleted_sentinel_fields(self, tmp_path: pathlib.Path) -> None: |
| 782 | """Sentinel entries use NULL_CONTENT_ID correctly and is_null_content_id detects them.""" |
| 783 | addr = "src/billing.py::compute_total" |
| 784 | # Created |
| 785 | append_symlog(tmp_path, addr, NULL_CONTENT_ID, _CID_A, _COMMIT_X, "claude-code", "symbol-created: compute_total") |
| 786 | # Modified |
| 787 | append_symlog(tmp_path, addr, _CID_A, _CID_B, _COMMIT_Y, "claude-code", "symbol-modified: improve") |
| 788 | # Deleted |
| 789 | append_symlog(tmp_path, addr, _CID_B, NULL_CONTENT_ID, _COMMIT_Z, "claude-code", "symbol-deleted: compute_total") |
| 790 | |
| 791 | entries = read_symlog(tmp_path, addr, limit=100) |
| 792 | assert len(entries) == 3 |
| 793 | deleted_entry = entries[0] # newest first |
| 794 | created_entry = entries[2] |
| 795 | assert is_null_content_id(deleted_entry.new_content_id) |
| 796 | assert is_null_content_id(created_entry.old_content_id) |
| 797 | assert not is_null_content_id(entries[1].old_content_id) |
| 798 | assert not is_null_content_id(entries[1].new_content_id) |