query_history.py
python
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
33 days ago
| 1 | """muse code query-history — temporal symbol search across commit history. |
| 2 | |
| 3 | Searches the commit history for symbols matching a predicate expression, |
| 4 | bounded by a commit range. Unlike ``muse code query --all-commits``, this |
| 5 | command is focused on *change events* — it shows when each symbol first |
| 6 | appeared, when it was last seen, how many commits it survived, and how many |
| 7 | distinct implementations it had. |
| 8 | |
| 9 | It answers questions that are impossible in Git: |
| 10 | |
| 11 | * "Find all public Python functions introduced after tag v1.0" |
| 12 | * "Show me every class whose signature changed in the last 50 commits" |
| 13 | * "Which functions were present in tag v1.0 but are gone in tag v2.0?" |
| 14 | * "Find the most volatile symbols in the last 100 commits" |
| 15 | * "What was deleted between the last two releases?" |
| 16 | |
| 17 | Usage:: |
| 18 | |
| 19 | muse code query-history "kind=function" "language=Python" |
| 20 | muse code query-history "name~=validate" --from v1.0 --to HEAD |
| 21 | muse code query-history "kind=class" --from abc12345 |
| 22 | muse code query-history "file~=billing" "kind=function" --json |
| 23 | muse code query-history "kind=function" --changed-only --sort changes |
| 24 | muse code query-history "kind=function" --removed-only --from v1.0 --to v2.0 |
| 25 | muse code query-history "kind=function" --introduced-only --from v1.0 |
| 26 | |
| 27 | Output:: |
| 28 | |
| 29 | Symbol history — kind=function language=Python (42 commits) |
| 30 | ────────────────────────────────────────────────────────────── |
| 31 | |
| 32 | src/billing.py::compute_total function [12 commits] 3 versions 2026-01-01..2026-03-10 |
| 33 | src/billing.py::compute_tax function [ 8 commits] stable 2026-01-15..2026-03-10 |
| 34 | └─ introduced: a1b2c3d4 2026-01-15 |
| 35 | └─ last seen: f7a8b9c0 2026-03-10 |
| 36 | |
| 37 | Flags: |
| 38 | |
| 39 | ``--from REF`` |
| 40 | Start of the commit range (exclusive; default: initial commit). |
| 41 | |
| 42 | ``--to REF`` |
| 43 | End of the commit range (inclusive; default: HEAD). |
| 44 | |
| 45 | ``--changed-only`` |
| 46 | Show only symbols that changed at least once (change_count > 1). |
| 47 | |
| 48 | ``--introduced-only`` |
| 49 | Show only symbols that exist in ``--to`` but not in ``--from`` (net-new). |
| 50 | |
| 51 | ``--removed-only`` |
| 52 | Show only symbols that exist in ``--from`` but not in ``--to`` (deleted). |
| 53 | |
| 54 | ``--sort FIELD`` |
| 55 | Sort results by: address (default), commits, changes, first, last. |
| 56 | |
| 57 | ``--min-changes N`` |
| 58 | Only show symbols with at least N distinct versions. |
| 59 | |
| 60 | ``--count`` |
| 61 | Emit only the count of matching symbols. |
| 62 | |
| 63 | ``--limit N`` |
| 64 | Cap the number of results returned. |
| 65 | |
| 66 | ``--max-commits N`` |
| 67 | Cap the number of commits walked (default: 10000). |
| 68 | |
| 69 | ``--json`` |
| 70 | Emit results as JSON. |
| 71 | """ |
| 72 | |
| 73 | from __future__ import annotations |
| 74 | |
| 75 | import argparse |
| 76 | import collections.abc |
| 77 | import json |
| 78 | import logging |
| 79 | import pathlib |
| 80 | import sys |
| 81 | |
| 82 | from muse._version import __version__ |
| 83 | from muse.core.errors import ExitCode |
| 84 | from muse.core.repo import read_repo_id, require_repo |
| 85 | from muse.core.store import ( |
| 86 | CommitRecord, |
| 87 | get_commit_snapshot_manifest, |
| 88 | read_current_branch, |
| 89 | resolve_commit_ref, |
| 90 | walk_commits_between, |
| 91 | ) |
| 92 | from muse.core.symbol_cache import SymbolCache, load_symbol_cache |
| 93 | from muse.plugins.code._predicate import Predicate, PredicateError, parse_query |
| 94 | from muse.plugins.code._query import language_of, symbols_for_snapshot |
| 95 | from muse.plugins.code.ast_parser import SymbolRecord |
| 96 | from muse.core.validation import clamp_int, sanitize_display |
| 97 | |
| 98 | type _HistoryDict = dict[str, str | int | None] |
| 99 | type _SymbolDict = dict[str, str | int] |
| 100 | type _AddrRecordMap = dict[str, tuple[SymbolRecord, str]] |
| 101 | type _HistoryIndex = dict[str, "_SymbolHistory"] |
| 102 | |
| 103 | logger = logging.getLogger(__name__) |
| 104 | |
| 105 | _VALID_SORT_FIELDS = frozenset({"address", "commits", "changes", "first", "last"}) |
| 106 | |
| 107 | |
| 108 | |
| 109 | class _SymbolHistory: |
| 110 | """Accumulated history of one symbol across a commit range.""" |
| 111 | |
| 112 | def __init__(self, address: str, kind: str, language: str) -> None: |
| 113 | self.address = address |
| 114 | self.kind = kind |
| 115 | self.language = language |
| 116 | self.first_commit_id: str = "" |
| 117 | self.first_committed_at: str = "" |
| 118 | self.last_commit_id: str = "" |
| 119 | self.last_committed_at: str = "" |
| 120 | self.commit_count: int = 0 |
| 121 | self.content_ids: set[str] = set() |
| 122 | |
| 123 | @property |
| 124 | def change_count(self) -> int: |
| 125 | """Number of distinct content_ids seen — 1 means body never changed.""" |
| 126 | return len(self.content_ids) |
| 127 | |
| 128 | def record(self, commit_id: str, committed_at: str, content_id: str) -> None: |
| 129 | """Record the symbol's presence in one commit.""" |
| 130 | if not self.first_commit_id: |
| 131 | self.first_commit_id = commit_id |
| 132 | self.first_committed_at = committed_at |
| 133 | self.last_commit_id = commit_id |
| 134 | self.last_committed_at = committed_at |
| 135 | self.commit_count += 1 |
| 136 | self.content_ids.add(content_id) |
| 137 | |
| 138 | def to_dict(self) -> _HistoryDict: |
| 139 | return { |
| 140 | "address": self.address, |
| 141 | "kind": self.kind, |
| 142 | "language": self.language, |
| 143 | "commit_count": self.commit_count, |
| 144 | "change_count": self.change_count, |
| 145 | "first_commit_id": self.first_commit_id, |
| 146 | "first_commit_id_short": self.first_commit_id[:8], |
| 147 | "first_committed_at": self.first_committed_at[:10], |
| 148 | "last_commit_id": self.last_commit_id, |
| 149 | "last_commit_id_short": self.last_commit_id[:8], |
| 150 | "last_committed_at": self.last_committed_at[:10], |
| 151 | "stable": self.change_count == 1, |
| 152 | } |
| 153 | |
| 154 | |
| 155 | class _RemovedSymbol: |
| 156 | """A symbol present in the --from snapshot but absent in --to.""" |
| 157 | |
| 158 | def __init__(self, address: str, rec: SymbolRecord, language: str) -> None: |
| 159 | self.address = address |
| 160 | self.rec = rec |
| 161 | self.language = language |
| 162 | |
| 163 | def to_dict(self) -> _SymbolDict: |
| 164 | return { |
| 165 | "address": self.address, |
| 166 | "kind": self.rec["kind"], |
| 167 | "language": self.language, |
| 168 | "status": "removed", |
| 169 | } |
| 170 | |
| 171 | |
| 172 | class _IntroducedSymbol: |
| 173 | """A symbol present in the --to snapshot but absent in --from.""" |
| 174 | |
| 175 | def __init__(self, address: str, rec: SymbolRecord, language: str) -> None: |
| 176 | self.address = address |
| 177 | self.rec = rec |
| 178 | self.language = language |
| 179 | |
| 180 | def to_dict(self) -> _SymbolDict: |
| 181 | return { |
| 182 | "address": self.address, |
| 183 | "kind": self.rec["kind"], |
| 184 | "language": self.language, |
| 185 | "status": "introduced", |
| 186 | } |
| 187 | |
| 188 | |
| 189 | def _collect_addresses( |
| 190 | root: pathlib.Path, |
| 191 | commit_id: str, |
| 192 | predicate: Predicate, |
| 193 | cache: SymbolCache | None = None, |
| 194 | ) -> _AddrRecordMap: |
| 195 | """Return {address: (rec, language)} for all symbols in *commit_id* matching *predicate*.""" |
| 196 | manifest = get_commit_snapshot_manifest(root, commit_id) or {} |
| 197 | sym_map = symbols_for_snapshot(root, manifest, cache=cache) |
| 198 | result: _AddrRecordMap = {} |
| 199 | for file_path, tree in sym_map.items(): |
| 200 | lang = language_of(file_path) |
| 201 | for addr, rec in tree.items(): |
| 202 | if predicate(file_path, rec): |
| 203 | result[addr] = (rec, lang) |
| 204 | return result |
| 205 | |
| 206 | |
| 207 | def _sort_key_fn( |
| 208 | sort_by: str, |
| 209 | ) -> collections.abc.Callable[[_SymbolHistory], tuple[str | int, ...]]: |
| 210 | """Return a sort key for ``_SymbolHistory`` objects.""" |
| 211 | if sort_by == "commits": |
| 212 | return lambda h: (-h.commit_count, h.address) |
| 213 | if sort_by == "changes": |
| 214 | return lambda h: (-h.change_count, h.address) |
| 215 | if sort_by == "first": |
| 216 | return lambda h: (h.first_committed_at, h.address) |
| 217 | if sort_by == "last": |
| 218 | return lambda h: (h.last_committed_at, h.address) |
| 219 | return lambda h: (h.address,) |
| 220 | |
| 221 | |
| 222 | def register( |
| 223 | subparsers: "argparse._SubParsersAction[argparse.ArgumentParser]", |
| 224 | ) -> None: |
| 225 | """Register the query-history subcommand.""" |
| 226 | parser = subparsers.add_parser( |
| 227 | "query-history", |
| 228 | help="Search commit history for symbols matching a predicate expression.", |
| 229 | description=__doc__, |
| 230 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 231 | ) |
| 232 | parser.add_argument( |
| 233 | "predicates", |
| 234 | nargs="+", |
| 235 | metavar="PREDICATE", |
| 236 | help='One or more predicates, e.g. "kind=function" "language=Python".', |
| 237 | ) |
| 238 | parser.add_argument( |
| 239 | "--from", |
| 240 | dest="from_ref", |
| 241 | default=None, |
| 242 | metavar="REF", |
| 243 | help="Start of range (exclusive; default: initial commit).", |
| 244 | ) |
| 245 | parser.add_argument( |
| 246 | "--to", |
| 247 | dest="to_ref", |
| 248 | default=None, |
| 249 | metavar="REF", |
| 250 | help="End of range (inclusive; default: HEAD).", |
| 251 | ) |
| 252 | parser.add_argument( |
| 253 | "--changed-only", |
| 254 | action="store_true", |
| 255 | help="Show only symbols with more than one distinct implementation (change_count > 1).", |
| 256 | ) |
| 257 | parser.add_argument( |
| 258 | "--introduced-only", |
| 259 | action="store_true", |
| 260 | help=( |
| 261 | "Show only symbols present in --to but absent in --from" |
| 262 | " (net-new symbols)." |
| 263 | ), |
| 264 | ) |
| 265 | parser.add_argument( |
| 266 | "--removed-only", |
| 267 | action="store_true", |
| 268 | help=( |
| 269 | "Show only symbols present in --from but absent in --to" |
| 270 | " (deleted symbols)." |
| 271 | ), |
| 272 | ) |
| 273 | parser.add_argument( |
| 274 | "--sort", |
| 275 | default="address", |
| 276 | metavar="FIELD", |
| 277 | choices=sorted(_VALID_SORT_FIELDS), |
| 278 | help=( |
| 279 | f"Sort results by field:" |
| 280 | f" {', '.join(sorted(_VALID_SORT_FIELDS))} (default: address)." |
| 281 | ), |
| 282 | ) |
| 283 | parser.add_argument( |
| 284 | "--min-changes", |
| 285 | type=int, |
| 286 | default=1, |
| 287 | metavar="N", |
| 288 | help="Only show symbols with at least N distinct versions (default: 1).", |
| 289 | ) |
| 290 | parser.add_argument( |
| 291 | "--count", |
| 292 | action="store_true", |
| 293 | help="Emit only the count of matching symbols.", |
| 294 | ) |
| 295 | parser.add_argument( |
| 296 | "--limit", |
| 297 | type=int, |
| 298 | default=0, |
| 299 | metavar="N", |
| 300 | help="Cap the number of results returned (0 = unlimited).", |
| 301 | ) |
| 302 | parser.add_argument( |
| 303 | "--max-commits", |
| 304 | type=int, |
| 305 | default=10_000, |
| 306 | metavar="N", |
| 307 | help="Cap the number of commits walked (default: 10000).", |
| 308 | ) |
| 309 | parser.add_argument( |
| 310 | "--json", |
| 311 | dest="as_json", |
| 312 | action="store_true", |
| 313 | help="Emit results as JSON.", |
| 314 | ) |
| 315 | parser.set_defaults(func=run) |
| 316 | |
| 317 | |
| 318 | def run(args: argparse.Namespace) -> None: |
| 319 | """Search commit history for symbols matching a predicate expression. |
| 320 | |
| 321 | Walks the commit range ``(--from, --to]`` oldest-first, accumulating |
| 322 | presence data for every matching symbol. |
| 323 | |
| 324 | The predicate grammar is identical to ``muse code query`` — supports OR, |
| 325 | NOT, parentheses, and all field keys including ``size_gt`` / ``size_lt``. |
| 326 | |
| 327 | Modes:: |
| 328 | |
| 329 | (default) All symbols seen anywhere in the range. |
| 330 | --changed-only Symbols whose body changed at least once. |
| 331 | --introduced-only Symbols present in --to but not in --from (born). |
| 332 | --removed-only Symbols present in --from but not in --to (deleted). |
| 333 | |
| 334 | Examples:: |
| 335 | |
| 336 | muse code query-history "kind=function" "language=Python" |
| 337 | muse code query-history "name~=validate" --from v1.0 --to HEAD |
| 338 | muse code query-history "kind=class" --json |
| 339 | muse code query-history "kind=function" --changed-only --sort changes |
| 340 | muse code query-history "kind=function" --removed-only --from v1.0 --to v2.0 |
| 341 | """ |
| 342 | predicates: list[str] = args.predicates |
| 343 | from_ref: str | None = args.from_ref |
| 344 | to_ref: str | None = args.to_ref |
| 345 | as_json: bool = args.as_json |
| 346 | changed_only: bool = args.changed_only |
| 347 | introduced_only: bool = args.introduced_only |
| 348 | removed_only: bool = args.removed_only |
| 349 | sort_by: str = args.sort |
| 350 | min_changes: int = clamp_int(args.min_changes, 0, 100000, 'min_changes') |
| 351 | count_only: bool = args.count |
| 352 | limit: int = clamp_int(args.limit, 0, 10000, 'limit') |
| 353 | max_commits: int = clamp_int(args.max_commits, 1, 100000, 'max_commits') |
| 354 | |
| 355 | # Validate mutually exclusive mode flags. |
| 356 | mode_flags = sum([changed_only, introduced_only, removed_only]) |
| 357 | if mode_flags > 1: |
| 358 | print( |
| 359 | "❌ --changed-only, --introduced-only, and --removed-only" |
| 360 | " are mutually exclusive.", |
| 361 | file=sys.stderr, |
| 362 | ) |
| 363 | raise SystemExit(ExitCode.USER_ERROR) |
| 364 | |
| 365 | if limit < 0: |
| 366 | print("❌ --limit must be >= 0.", file=sys.stderr) |
| 367 | raise SystemExit(ExitCode.USER_ERROR) |
| 368 | |
| 369 | if min_changes < 1: |
| 370 | print("❌ --min-changes must be >= 1.", file=sys.stderr) |
| 371 | raise SystemExit(ExitCode.USER_ERROR) |
| 372 | |
| 373 | if max_commits < 1: |
| 374 | print("❌ --max-commits must be >= 1.", file=sys.stderr) |
| 375 | raise SystemExit(ExitCode.USER_ERROR) |
| 376 | |
| 377 | root = require_repo() |
| 378 | repo_id = read_repo_id(root) |
| 379 | branch = read_current_branch(root) |
| 380 | |
| 381 | if not predicates: |
| 382 | print("❌ At least one predicate is required.", file=sys.stderr) |
| 383 | raise SystemExit(ExitCode.USER_ERROR) |
| 384 | |
| 385 | try: |
| 386 | predicate = parse_query(predicates) |
| 387 | except PredicateError as exc: |
| 388 | print(f"❌ {exc}", file=sys.stderr) |
| 389 | raise SystemExit(ExitCode.USER_ERROR) |
| 390 | |
| 391 | # Resolve range endpoints. |
| 392 | to_commit = resolve_commit_ref(root, repo_id, branch, to_ref) |
| 393 | if to_commit is None: |
| 394 | print(f"❌ --to ref '{to_ref or 'HEAD'}' not found.", file=sys.stderr) |
| 395 | raise SystemExit(ExitCode.USER_ERROR) |
| 396 | |
| 397 | from_commit_id: str | None = None |
| 398 | if from_ref is not None: |
| 399 | from_c = resolve_commit_ref(root, repo_id, branch, from_ref) |
| 400 | if from_c is None: |
| 401 | print(f"❌ --from ref '{from_ref}' not found.", file=sys.stderr) |
| 402 | raise SystemExit(ExitCode.USER_ERROR) |
| 403 | from_commit_id = from_c.commit_id |
| 404 | |
| 405 | # Load the symbol cache once — shared across all snapshot operations in this |
| 406 | # run. On a warm cache this eliminates O(n × 170ms) disk I/O in favour of |
| 407 | # a single load + single save, regardless of how many snapshots are walked. |
| 408 | shared_cache: SymbolCache = load_symbol_cache(root) |
| 409 | |
| 410 | # ── --introduced-only / --removed-only: snapshot diff mode ─────────────── |
| 411 | if introduced_only or removed_only: |
| 412 | to_addrs = _collect_addresses( |
| 413 | root, to_commit.commit_id, predicate, cache=shared_cache |
| 414 | ) |
| 415 | |
| 416 | # For --from: use from_commit_id if provided, else fall back to the |
| 417 | # oldest commit reachable from to_commit. |
| 418 | if from_commit_id is not None: |
| 419 | from_addrs = _collect_addresses( |
| 420 | root, from_commit_id, predicate, cache=shared_cache |
| 421 | ) |
| 422 | else: |
| 423 | # No --from: treat the very first commit as the baseline. |
| 424 | all_in_range: list[CommitRecord] = sorted( |
| 425 | walk_commits_between(root, to_commit.commit_id, None), |
| 426 | key=lambda c: c.committed_at, |
| 427 | ) |
| 428 | if all_in_range: |
| 429 | from_addrs = _collect_addresses( |
| 430 | root, all_in_range[0].commit_id, predicate, cache=shared_cache |
| 431 | ) |
| 432 | else: |
| 433 | from_addrs = {} |
| 434 | |
| 435 | # All snapshot loading for diff mode is complete — persist any new entries. |
| 436 | shared_cache.save() |
| 437 | |
| 438 | if introduced_only: |
| 439 | introduced_addrs = set(to_addrs) - set(from_addrs) |
| 440 | intro_symbols: list[_IntroducedSymbol] = sorted( |
| 441 | [ |
| 442 | _IntroducedSymbol(addr, rec, lang) |
| 443 | for addr, (rec, lang) in to_addrs.items() |
| 444 | if addr in introduced_addrs |
| 445 | ], |
| 446 | key=lambda s: s.address, |
| 447 | ) |
| 448 | if limit > 0: |
| 449 | intro_symbols = intro_symbols[:limit] |
| 450 | if count_only: |
| 451 | print(len(intro_symbols)) |
| 452 | return |
| 453 | if as_json: |
| 454 | print( |
| 455 | json.dumps( |
| 456 | { |
| 457 | "schema_version": __version__, |
| 458 | "mode": "introduced-only", |
| 459 | "to_commit": to_commit.commit_id[:8], |
| 460 | "from_commit": from_commit_id[:8] |
| 461 | if from_commit_id |
| 462 | else None, |
| 463 | "symbols_found": len(intro_symbols), |
| 464 | "results": [s.to_dict() for s in intro_symbols], |
| 465 | }, |
| 466 | indent=2, |
| 467 | ) |
| 468 | ) |
| 469 | return |
| 470 | pred_display = " AND ".join(predicates) |
| 471 | from_label = from_commit_id[:8] if from_commit_id else "initial" |
| 472 | print( |
| 473 | f"\n Introduced symbols — {pred_display}" |
| 474 | f" ({from_label}..{to_commit.commit_id[:8]})\n" |
| 475 | ) |
| 476 | if not intro_symbols: |
| 477 | print(" (no new symbols found)") |
| 478 | return |
| 479 | for sym in intro_symbols: |
| 480 | print(f" + {sanitize_display(sym.address)} [{sym.rec['kind']}]") |
| 481 | print(f"\n {len(intro_symbols)} symbol(s) introduced") |
| 482 | return |
| 483 | |
| 484 | # removed_only |
| 485 | removed_addrs = set(from_addrs) - set(to_addrs) |
| 486 | _removed_list: list[_RemovedSymbol] = [ |
| 487 | _RemovedSymbol(addr, rec, lang) |
| 488 | for addr, (rec, lang) in from_addrs.items() |
| 489 | if addr in removed_addrs |
| 490 | ] |
| 491 | _removed_list.sort(key=lambda s: s.address) |
| 492 | removed_symbols: list[_RemovedSymbol] = ( |
| 493 | _removed_list[:limit] if limit > 0 else _removed_list |
| 494 | ) |
| 495 | if count_only: |
| 496 | print(len(removed_symbols)) |
| 497 | return |
| 498 | if as_json: |
| 499 | print( |
| 500 | json.dumps( |
| 501 | { |
| 502 | "schema_version": __version__, |
| 503 | "mode": "removed-only", |
| 504 | "to_commit": to_commit.commit_id[:8], |
| 505 | "from_commit": from_commit_id[:8] |
| 506 | if from_commit_id |
| 507 | else None, |
| 508 | "symbols_found": len(removed_symbols), |
| 509 | "results": [s.to_dict() for s in removed_symbols], |
| 510 | }, |
| 511 | indent=2, |
| 512 | ) |
| 513 | ) |
| 514 | return |
| 515 | pred_display = " AND ".join(predicates) |
| 516 | from_label = from_commit_id[:8] if from_commit_id else "initial" |
| 517 | print( |
| 518 | f"\n Removed symbols — {pred_display}" |
| 519 | f" ({from_label}..{to_commit.commit_id[:8]})\n" |
| 520 | ) |
| 521 | if not removed_symbols: |
| 522 | print(" (no symbols removed in this range)") |
| 523 | return |
| 524 | for rem in removed_symbols: |
| 525 | print(f" - {sanitize_display(rem.address)} [{rem.rec['kind']}]") |
| 526 | print(f"\n {len(removed_symbols)} symbol(s) removed") |
| 527 | return |
| 528 | |
| 529 | # ── Default / --changed-only: walk commit range ─────────────────────────── |
| 530 | raw_commits: list[CommitRecord] = sorted( |
| 531 | walk_commits_between(root, to_commit.commit_id, from_commit_id), |
| 532 | key=lambda c: c.committed_at, |
| 533 | ) |
| 534 | |
| 535 | truncated = len(raw_commits) > max_commits |
| 536 | commits = raw_commits[:max_commits] |
| 537 | |
| 538 | # Accumulate per-symbol history — deduplicate on snapshot_id. |
| 539 | history: _HistoryIndex = {} |
| 540 | seen_snapshots: set[str] = set() |
| 541 | |
| 542 | for commit in commits: |
| 543 | if commit.snapshot_id in seen_snapshots: |
| 544 | continue |
| 545 | seen_snapshots.add(commit.snapshot_id) |
| 546 | |
| 547 | manifest = get_commit_snapshot_manifest(root, commit.commit_id) or {} |
| 548 | sym_map = symbols_for_snapshot(root, manifest, cache=shared_cache) |
| 549 | for file_path, tree in sym_map.items(): |
| 550 | for addr, rec in tree.items(): |
| 551 | if not predicate(file_path, rec): |
| 552 | continue |
| 553 | if addr not in history: |
| 554 | history[addr] = _SymbolHistory( |
| 555 | address=addr, |
| 556 | kind=rec["kind"], |
| 557 | language=language_of(file_path), |
| 558 | ) |
| 559 | history[addr].record( |
| 560 | commit.commit_id, |
| 561 | commit.committed_at.isoformat(), |
| 562 | rec["content_id"], |
| 563 | ) |
| 564 | |
| 565 | # All snapshot loading for walk mode is complete — persist any new entries. |
| 566 | shared_cache.save() |
| 567 | |
| 568 | results: list[_SymbolHistory] = list(history.values()) |
| 569 | |
| 570 | # Apply filters. |
| 571 | if changed_only: |
| 572 | results = [r for r in results if r.change_count > 1] |
| 573 | if min_changes > 1: |
| 574 | results = [r for r in results if r.change_count >= min_changes] |
| 575 | |
| 576 | # Sort. |
| 577 | results.sort(key=_sort_key_fn(sort_by)) |
| 578 | |
| 579 | # Apply limit. |
| 580 | limited = limit > 0 and len(results) > limit |
| 581 | if limited: |
| 582 | results = results[:limit] |
| 583 | |
| 584 | # Count-only output. |
| 585 | if count_only: |
| 586 | print(len(results)) |
| 587 | return |
| 588 | |
| 589 | # JSON output. |
| 590 | if as_json: |
| 591 | print( |
| 592 | json.dumps( |
| 593 | { |
| 594 | "schema_version": __version__, |
| 595 | "mode": "changed-only" if changed_only else "default", |
| 596 | "to_commit": to_commit.commit_id[:8], |
| 597 | "from_commit": from_commit_id[:8] if from_commit_id else None, |
| 598 | "commits_scanned": len(commits), |
| 599 | "truncated": truncated, |
| 600 | "symbols_found": len(results), |
| 601 | "results": [r.to_dict() for r in results], |
| 602 | }, |
| 603 | indent=2, |
| 604 | ) |
| 605 | ) |
| 606 | return |
| 607 | |
| 608 | # Human-readable output. |
| 609 | pred_display = " AND ".join(predicates) |
| 610 | trunc_note = f" ⚠️ capped at {max_commits}" if truncated else "" |
| 611 | print( |
| 612 | f"\nSymbol history — {pred_display}" |
| 613 | f" ({len(commits)} commit(s) scanned{trunc_note})" |
| 614 | ) |
| 615 | print("─" * 62) |
| 616 | |
| 617 | if not results: |
| 618 | print(" (no matching symbols found in range)") |
| 619 | return |
| 620 | |
| 621 | max_addr = max(len(r.address) for r in results) |
| 622 | for r in results: |
| 623 | versions = ( |
| 624 | f"{r.change_count} version(s)" |
| 625 | if r.change_count > 1 |
| 626 | else "stable" |
| 627 | ) |
| 628 | span = f"{r.first_committed_at[:10]}..{r.last_committed_at[:10]}" |
| 629 | print( |
| 630 | f" {r.address:<{max_addr}} {r.kind:<14} " |
| 631 | f"[{r.commit_count:>3} commit(s)] {versions:<14} {span}" |
| 632 | ) |
| 633 | if r.first_commit_id: |
| 634 | print(f" └─ introduced: {r.first_commit_id[:8]}") |
| 635 | if r.first_commit_id != r.last_commit_id: |
| 636 | print(f" └─ last seen: {r.last_commit_id[:8]}") |
| 637 | |
| 638 | lim_note = f" (limited to {limit})" if limited else "" |
| 639 | print(f"\n {len(results)} symbol(s) found{lim_note}") |
File History
2 commits
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
33 days ago
sha256:d8316ffae901be06347e16ab55be11868eb519dd16ade3e8aa16a99e662f7e62
baseline: rc14 re-baseline after rc3 store corruption recovery
Human
patch
33 days ago