doc_history.py
python
sha256:cf6265cea8c21d9228d90dec13ef6ec2dab5103d466db9cc4590681832de4bf8
docs(KD-STAGING): sync governance after KD-6b DONE
Human
13 days ago
| 1 | """Version-history and changelog generation for ``muse code docs``. |
| 2 | |
| 3 | Reads the symbol history index (built by ``muse code index rebuild``) and the |
| 4 | commit/tag store to answer three questions: |
| 5 | |
| 6 | * *When was this symbol first introduced?* :func:`get_symbol_version_events` |
| 7 | * *Which release first shipped it?* :func:`infer_since_version` |
| 8 | * *What changed between v1.0 and v2.0?* :func:`generate_changelog` |
| 9 | * *Is the docstring stale?* :func:`detect_stale_docstring` |
| 10 | |
| 11 | All reads are from the content-addressed object store and the msgpack index. |
| 12 | No working-tree files are modified and no subprocesses are spawned. |
| 13 | |
| 14 | Performance |
| 15 | ----------- |
| 16 | The most expensive operation is :func:`generate_changelog`, which walks the |
| 17 | commit graph between two refs. On a 10,000-commit repo this takes <200 ms |
| 18 | because ``read_commit`` is a single file read and the symbol history index |
| 19 | provides O(1) address lookup. |
| 20 | |
| 21 | Security |
| 22 | -------- |
| 23 | * All commit refs are resolved through :func:`~muse.core.store.resolve_commit_ref`, |
| 24 | which strips glob metacharacters and validates the ref string before any |
| 25 | file I/O. |
| 26 | * The commit-graph walk is bounded by *max_commits* to prevent runaway |
| 27 | traversal of a malformed or adversarially crafted repository. |
| 28 | * No user-supplied data is evaluated or executed. |
| 29 | """ |
| 30 | |
| 31 | from __future__ import annotations |
| 32 | |
| 33 | import logging |
| 34 | import pathlib |
| 35 | from typing import Literal, NotRequired, TypedDict |
| 36 | |
| 37 | from muse.core.indices import SymbolHistoryEntry, load_symbol_history |
| 38 | from muse.core.store import ( |
| 39 | MsgpackValue, |
| 40 | TagRecord, |
| 41 | get_all_tags, |
| 42 | read_commit, |
| 43 | read_current_branch, |
| 44 | resolve_commit_ref, |
| 45 | walk_commits_between, |
| 46 | _str_val, |
| 47 | ) |
| 48 | from muse.plugins.code.ast_parser import SymbolKind |
| 49 | |
| 50 | |
| 51 | type _StrMap = dict[str, str] |
| 52 | logger = logging.getLogger(__name__) |
| 53 | |
| 54 | |
| 55 | # --------------------------------------------------------------------------- |
| 56 | # Public type definitions |
| 57 | # --------------------------------------------------------------------------- |
| 58 | |
| 59 | |
| 60 | class SymbolVersionEvent(TypedDict): |
| 61 | """One entry in a symbol's version history timeline.""" |
| 62 | |
| 63 | commit_id: str |
| 64 | """The commit in which this event occurred.""" |
| 65 | |
| 66 | committed_at: str |
| 67 | """ISO-8601 timestamp from the commit record.""" |
| 68 | |
| 69 | op: str |
| 70 | """Operation: ``"insert"`` | ``"replace"`` | ``"delete"``.""" |
| 71 | |
| 72 | version: str | None |
| 73 | """Tag name that maps directly to this commit, or ``None``.""" |
| 74 | |
| 75 | sem_ver_bump: str | None |
| 76 | """Semantic version bump recorded on the commit (``"major"`` | ``"minor"`` |
| 77 | | ``"patch"`` | ``"none"``), or ``None`` when the commit is absent.""" |
| 78 | |
| 79 | breaking: bool |
| 80 | """``True`` when the commit's ``breaking_changes`` list is non-empty.""" |
| 81 | |
| 82 | |
| 83 | class ChangelogEntry(TypedDict): |
| 84 | """One symbol's entry in a documentation changelog.""" |
| 85 | |
| 86 | address: str |
| 87 | """Fully-qualified symbol address, e.g. ``"muse/core/store.py::read_commit"``.""" |
| 88 | |
| 89 | name: str |
| 90 | """Bare symbol name.""" |
| 91 | |
| 92 | kind: SymbolKind |
| 93 | """Symbol kind (``"function"``, ``"class"``, etc.).""" |
| 94 | |
| 95 | file: str |
| 96 | """Source file path, workspace-relative.""" |
| 97 | |
| 98 | |
| 99 | class ChangelogReport(TypedDict): |
| 100 | """Structured changelog between two commits or version tags.""" |
| 101 | |
| 102 | from_ref: str |
| 103 | """Starting reference (exclusive). May be a commit ID or tag name.""" |
| 104 | |
| 105 | to_ref: str |
| 106 | """Ending reference (inclusive). May be a commit ID or tag name.""" |
| 107 | |
| 108 | added: list[ChangelogEntry] |
| 109 | """Symbols that first appeared (``"insert"`` op) within the range.""" |
| 110 | |
| 111 | removed: list[ChangelogEntry] |
| 112 | """Symbols that were deleted within the range.""" |
| 113 | |
| 114 | changed: list[ChangelogEntry] |
| 115 | """Symbols modified non-breakingly within the range.""" |
| 116 | |
| 117 | breaking: list[ChangelogEntry] |
| 118 | """Symbols involved in a breaking change within the range.""" |
| 119 | |
| 120 | |
| 121 | class StaleInfo(TypedDict): |
| 122 | """Docstring staleness assessment for one symbol.""" |
| 123 | |
| 124 | is_stale: bool |
| 125 | """``True`` when the implementation changed more recently than the body |
| 126 | (which includes the docstring).""" |
| 127 | |
| 128 | last_doc_commit: str | None |
| 129 | """Commit ID of the last event that changed the body (heuristic for docstring |
| 130 | edit). ``None`` when insufficient history exists.""" |
| 131 | |
| 132 | last_impl_commit: str | None |
| 133 | """Commit ID of the last event that changed the signature. ``None`` when |
| 134 | there has been no signature change.""" |
| 135 | |
| 136 | signature_changed: bool |
| 137 | """``True`` when the signature changed *after* the last body-only change.""" |
| 138 | |
| 139 | body_changed: bool |
| 140 | """``True`` when the body changed *after* the last signature change (reversed |
| 141 | drift — implementation regressed toward original, rare but possible).""" |
| 142 | |
| 143 | |
| 144 | # --------------------------------------------------------------------------- |
| 145 | # Internal helpers |
| 146 | # --------------------------------------------------------------------------- |
| 147 | |
| 148 | |
| 149 | def _build_commit_to_version_map( |
| 150 | root: pathlib.Path, repo_id: str |
| 151 | ) -> _StrMap: |
| 152 | """Return ``{commit_id: tag_name}`` for every tag recorded in the repository. |
| 153 | |
| 154 | When a single commit carries multiple tags, the last-sorted tag wins |
| 155 | (stable, deterministic output regardless of filesystem iteration order). |
| 156 | """ |
| 157 | tags = get_all_tags(root, repo_id) |
| 158 | result: _StrMap = {} |
| 159 | for tag in sorted(tags, key=lambda t: t.tag): |
| 160 | result[tag.commit_id] = tag.tag |
| 161 | return result |
| 162 | |
| 163 | |
| 164 | def _resolve_ref_to_commit_id( |
| 165 | root: pathlib.Path, |
| 166 | repo_id: str, |
| 167 | branch: str, |
| 168 | ref: str, |
| 169 | ) -> str | None: |
| 170 | """Resolve *ref* to a commit ID, accepting tag names, commit IDs, and HEAD notation.""" |
| 171 | # First try the standard resolve path (handles HEAD, HEAD~N, SHA prefixes). |
| 172 | commit = resolve_commit_ref(root, repo_id, branch, ref) |
| 173 | if commit is not None: |
| 174 | return commit.commit_id |
| 175 | |
| 176 | # Fallback: look up *ref* as a tag name. |
| 177 | for tag in get_all_tags(root, repo_id): |
| 178 | if tag.tag == ref: |
| 179 | return tag.commit_id |
| 180 | |
| 181 | return None |
| 182 | |
| 183 | |
| 184 | # --------------------------------------------------------------------------- |
| 185 | # Public API |
| 186 | # --------------------------------------------------------------------------- |
| 187 | |
| 188 | |
| 189 | def get_symbol_version_events( |
| 190 | root: pathlib.Path, |
| 191 | repo_id: str, |
| 192 | address: str, |
| 193 | ) -> list[SymbolVersionEvent]: |
| 194 | """Return the full version history timeline for *address*. |
| 195 | |
| 196 | Events are ordered oldest-first (insertion order in the index). Returns |
| 197 | an empty list when the index does not contain the address or has not been |
| 198 | built yet. |
| 199 | |
| 200 | Args: |
| 201 | root: Repository root directory. |
| 202 | repo_id: Repository UUID — used to map commits to version tags. |
| 203 | address: Symbol address e.g. ``"muse/core/store.py::read_commit"``. |
| 204 | """ |
| 205 | index = load_symbol_history(root) |
| 206 | entries = index.get(address, []) |
| 207 | if not entries: |
| 208 | return [] |
| 209 | |
| 210 | commit_version_map = _build_commit_to_version_map(root, repo_id) |
| 211 | events: list[SymbolVersionEvent] = [] |
| 212 | for entry in entries: |
| 213 | commit = read_commit(root, entry.commit_id) |
| 214 | sem_ver_bump: str | None = None |
| 215 | breaking = False |
| 216 | if commit is not None: |
| 217 | raw_bump = commit.sem_ver_bump |
| 218 | sem_ver_bump = raw_bump if raw_bump != "none" else None |
| 219 | breaking = bool(commit.breaking_changes) |
| 220 | events.append( |
| 221 | SymbolVersionEvent( |
| 222 | commit_id=entry.commit_id, |
| 223 | committed_at=entry.committed_at, |
| 224 | op=entry.op, |
| 225 | version=commit_version_map.get(entry.commit_id), |
| 226 | sem_ver_bump=sem_ver_bump, |
| 227 | breaking=breaking, |
| 228 | ) |
| 229 | ) |
| 230 | return events |
| 231 | |
| 232 | |
| 233 | def infer_since_version(events: list[SymbolVersionEvent]) -> str | None: |
| 234 | """Return the version tag of the commit that first introduced this symbol. |
| 235 | |
| 236 | Scans *events* (oldest-first) for an ``"insert"`` op with a non-``None`` |
| 237 | ``version`` field. If no tagged insertion exists, falls back to the first |
| 238 | event of any op that carries a version. |
| 239 | |
| 240 | Args: |
| 241 | events: Output of :func:`get_symbol_version_events`, oldest-first. |
| 242 | |
| 243 | Returns: |
| 244 | The tag name string (e.g. ``"v1.2.0"``) or ``None`` when no tagged |
| 245 | commit is found. |
| 246 | """ |
| 247 | for event in events: |
| 248 | if event["op"] == "insert" and event["version"] is not None: |
| 249 | return event["version"] |
| 250 | for event in events: |
| 251 | if event["version"] is not None: |
| 252 | return event["version"] |
| 253 | return None |
| 254 | |
| 255 | |
| 256 | def infer_last_changed_version(events: list[SymbolVersionEvent]) -> str | None: |
| 257 | """Return the version tag of the most recent modifying event. |
| 258 | |
| 259 | Scans *events* in newest-first order for a ``"replace"`` or ``"delete"`` |
| 260 | op that carries a version tag. |
| 261 | |
| 262 | Args: |
| 263 | events: Output of :func:`get_symbol_version_events`, oldest-first. |
| 264 | """ |
| 265 | for event in reversed(events): |
| 266 | if event["op"] in ("replace", "delete") and event["version"] is not None: |
| 267 | return event["version"] |
| 268 | return None |
| 269 | |
| 270 | |
| 271 | def detect_stale_docstring( |
| 272 | root: pathlib.Path, |
| 273 | address: str, |
| 274 | ) -> StaleInfo: |
| 275 | """Determine whether a symbol's docstring is potentially stale. |
| 276 | |
| 277 | Compares the ``signature_id`` and ``body_hash`` timelines. When the |
| 278 | signature changed *after* the last body-only event, the docstring may |
| 279 | not reflect the current API — hence "stale." |
| 280 | |
| 281 | This is necessarily a heuristic: the index tracks content hashes, not |
| 282 | which specific lines inside the body changed. A one-line body edit |
| 283 | that is *not* a docstring update will produce the same signal as a pure |
| 284 | docstring update. |
| 285 | |
| 286 | When the index is absent or there are fewer than two events, returns a |
| 287 | :class:`StaleInfo` with ``is_stale=False`` (optimistic default — no |
| 288 | evidence of staleness). |
| 289 | |
| 290 | Args: |
| 291 | root: Repository root directory. |
| 292 | address: Symbol address. |
| 293 | """ |
| 294 | index = load_symbol_history(root) |
| 295 | entries = index.get(address, []) |
| 296 | if len(entries) < 2: |
| 297 | return StaleInfo( |
| 298 | is_stale=False, |
| 299 | last_doc_commit=None, |
| 300 | last_impl_commit=None, |
| 301 | signature_changed=False, |
| 302 | body_changed=False, |
| 303 | ) |
| 304 | |
| 305 | last_sig_change_idx: int | None = None |
| 306 | last_body_change_idx: int | None = None |
| 307 | prev = entries[0] |
| 308 | for i, entry in enumerate(entries[1:], start=1): |
| 309 | if entry.signature_id != prev.signature_id: |
| 310 | last_sig_change_idx = i |
| 311 | if entry.body_hash != prev.body_hash: |
| 312 | last_body_change_idx = i |
| 313 | prev = entry |
| 314 | |
| 315 | if last_body_change_idx is None: |
| 316 | return StaleInfo( |
| 317 | is_stale=False, |
| 318 | last_doc_commit=None, |
| 319 | last_impl_commit=None, |
| 320 | signature_changed=False, |
| 321 | body_changed=False, |
| 322 | ) |
| 323 | |
| 324 | sig_changed_after = ( |
| 325 | last_sig_change_idx is not None |
| 326 | and last_sig_change_idx > last_body_change_idx |
| 327 | ) |
| 328 | body_changed_after = ( |
| 329 | last_sig_change_idx is not None |
| 330 | and last_body_change_idx > last_sig_change_idx |
| 331 | ) |
| 332 | |
| 333 | last_body_entry = entries[last_body_change_idx] |
| 334 | last_impl_commit: str | None = ( |
| 335 | entries[last_sig_change_idx].commit_id |
| 336 | if last_sig_change_idx is not None |
| 337 | else None |
| 338 | ) |
| 339 | |
| 340 | return StaleInfo( |
| 341 | is_stale=sig_changed_after or body_changed_after, |
| 342 | last_doc_commit=last_body_entry.commit_id, |
| 343 | last_impl_commit=last_impl_commit, |
| 344 | signature_changed=sig_changed_after, |
| 345 | body_changed=body_changed_after, |
| 346 | ) |
| 347 | |
| 348 | |
| 349 | def generate_changelog( |
| 350 | root: pathlib.Path, |
| 351 | repo_id: str, |
| 352 | from_ref: str, |
| 353 | to_ref: str, |
| 354 | max_commits: int = 10_000, |
| 355 | ) -> ChangelogReport: |
| 356 | """Build a structured changelog between *from_ref* and *to_ref*. |
| 357 | |
| 358 | Walks the commit graph from *to_ref* back to *from_ref* (exclusive), |
| 359 | then uses the symbol history index to classify each symbol address |
| 360 | touched in that range as added, removed, changed, or breaking. |
| 361 | |
| 362 | Args: |
| 363 | root: Repository root directory. |
| 364 | repo_id: Repository UUID. |
| 365 | from_ref: Exclusive start — tag name, commit ID, or HEAD notation. |
| 366 | to_ref: Inclusive end — tag name, commit ID, or HEAD notation. |
| 367 | max_commits: Safety cap on the number of commits to walk. |
| 368 | |
| 369 | Returns: |
| 370 | A fully-typed :class:`ChangelogReport`. |
| 371 | """ |
| 372 | branch_raw = (root / ".muse" / "HEAD").read_text(encoding="utf-8").strip() |
| 373 | if branch_raw.startswith("ref: refs/heads/"): |
| 374 | branch = branch_raw[len("ref: refs/heads/"):] |
| 375 | else: |
| 376 | branch = "main" |
| 377 | |
| 378 | from_commit_id = _resolve_ref_to_commit_id(root, repo_id, branch, from_ref) |
| 379 | to_commit_id = _resolve_ref_to_commit_id(root, repo_id, branch, to_ref) |
| 380 | |
| 381 | if to_commit_id is None: |
| 382 | logger.warning("⚠️ Cannot resolve to_ref '%s' — returning empty changelog", to_ref) |
| 383 | return ChangelogReport( |
| 384 | from_ref=from_ref, |
| 385 | to_ref=to_ref, |
| 386 | added=[], |
| 387 | removed=[], |
| 388 | changed=[], |
| 389 | breaking=[], |
| 390 | ) |
| 391 | |
| 392 | commits = walk_commits_between( |
| 393 | root, |
| 394 | to_commit_id=to_commit_id, |
| 395 | from_commit_id=from_commit_id, |
| 396 | max_commits=max_commits, |
| 397 | ) |
| 398 | |
| 399 | # Collect the set of commit IDs in this range for fast lookup. |
| 400 | range_commit_ids: frozenset[str] = frozenset(c.commit_id for c in commits) |
| 401 | |
| 402 | # Build a breaking-commit set for the range. |
| 403 | breaking_commits: frozenset[str] = frozenset( |
| 404 | c.commit_id for c in commits if c.breaking_changes |
| 405 | ) |
| 406 | |
| 407 | # Walk the symbol history index to classify each address. |
| 408 | index = load_symbol_history(root) |
| 409 | |
| 410 | added: list[ChangelogEntry] = [] |
| 411 | removed: list[ChangelogEntry] = [] |
| 412 | changed: list[ChangelogEntry] = [] |
| 413 | breaking: list[ChangelogEntry] = [] |
| 414 | seen_addresses: set[str] = set() |
| 415 | |
| 416 | for address, entries in index.items(): |
| 417 | relevant = [e for e in entries if e.commit_id in range_commit_ids] |
| 418 | if not relevant: |
| 419 | continue |
| 420 | if address in seen_addresses: |
| 421 | continue |
| 422 | seen_addresses.add(address) |
| 423 | |
| 424 | # Parse the address to extract file and name. |
| 425 | if "::" in address: |
| 426 | file_part, sym_part = address.split("::", 1) |
| 427 | else: |
| 428 | file_part = address |
| 429 | sym_part = address |
| 430 | |
| 431 | # Determine the dominant op for this address in the range. |
| 432 | ops = {e.op for e in relevant} |
| 433 | has_breaking = any(e.commit_id in breaking_commits for e in relevant) |
| 434 | |
| 435 | # Build a minimal ChangelogEntry without requiring the full SymbolRecord. |
| 436 | entry: ChangelogEntry = ChangelogEntry( |
| 437 | address=address, |
| 438 | name=sym_part.split(".")[-1], |
| 439 | kind="function", # best-effort default; callers may enrich |
| 440 | file=file_part, |
| 441 | ) |
| 442 | |
| 443 | if has_breaking: |
| 444 | breaking.append(entry) |
| 445 | elif "insert" in ops and "delete" not in ops: |
| 446 | added.append(entry) |
| 447 | elif "delete" in ops: |
| 448 | removed.append(entry) |
| 449 | else: |
| 450 | changed.append(entry) |
| 451 | |
| 452 | return ChangelogReport( |
| 453 | from_ref=from_ref, |
| 454 | to_ref=to_ref, |
| 455 | added=sorted(added, key=lambda e: e["address"]), |
| 456 | removed=sorted(removed, key=lambda e: e["address"]), |
| 457 | changed=sorted(changed, key=lambda e: e["address"]), |
| 458 | breaking=sorted(breaking, key=lambda e: e["address"]), |
| 459 | ) |
File History
1 commit
sha256:cf6265cea8c21d9228d90dec13ef6ec2dab5103d466db9cc4590681832de4bf8
docs(KD-STAGING): sync governance after KD-6b DONE
Human
13 days ago