parser.py
python
sha256:cf6265cea8c21d9228d90dec13ef6ec2dab5103d466db9cc4590681832de4bf8
docs(KD-STAGING): sync governance after KD-6b DONE
Human
12 days ago
| 1 | """Knowtation frontmatter schema — typed parser, validator, and migrator. |
| 2 | |
| 3 | This module is the **single source of truth** for every frontmatter field that |
| 4 | the Knowtation Muse plugin reads, writes, or validates. It replaces the |
| 5 | lightweight regex key-extractor in :mod:`~muse.plugins.knowtation.detector` for |
| 6 | cases where field *values* matter (diff, merge, schema migration, domain-info). |
| 7 | |
| 8 | Schema version |
| 9 | -------------- |
| 10 | :data:`FRONTMATTER_SCHEMA_VERSION` is a plain integer string (``"1"``, ``"2"``, |
| 11 | …) **independent** of the Muse package semver (which lives in |
| 12 | :mod:`muse._version`). The schema version is bumped when: |
| 13 | |
| 14 | - A field is added with a non-trivial coercion rule or mandatory default, OR |
| 15 | - A field is renamed or removed. |
| 16 | |
| 17 | Adding a new optional field with a ``None`` / empty-list default does **not** |
| 18 | require a version bump (backward-compatible). |
| 19 | |
| 20 | Field reference (SPEC §2) |
| 21 | -------------------------- |
| 22 | §2.1 Common (any note): |
| 23 | |
| 24 | title, project, tags, date, updated |
| 25 | |
| 26 | §2.2 Inbox (notes written by capture/import pipeline): |
| 27 | |
| 28 | source (canonical), source_id (recommended), source_type (import alias) |
| 29 | |
| 30 | §2.3 Intention and temporal (all optional): |
| 31 | |
| 32 | follows, causal_chain_id, entity, episode_id, summarizes, |
| 33 | summarizes_range, state_snapshot |
| 34 | |
| 35 | Phase 1.7 — mist attachment IDs (reserved, schema v2): |
| 36 | |
| 37 | attachments |
| 38 | |
| 39 | §2.4 Reserved for Phase 12 (blockchain; parsers MUST NOT require them): |
| 40 | |
| 41 | network, wallet_address, tx_hash, payment_status |
| 42 | |
| 43 | All other keys are preserved verbatim in :attr:`ParsedFrontmatter.extra` so |
| 44 | that round-trips through ``parse → to_dict`` are lossless. |
| 45 | """ |
| 46 | |
| 47 | from __future__ import annotations |
| 48 | |
| 49 | import logging |
| 50 | import re |
| 51 | from dataclasses import dataclass, field |
| 52 | from typing import Any |
| 53 | |
| 54 | import yaml # pyyaml — safe_load only; no arbitrary object instantiation |
| 55 | |
| 56 | logger = logging.getLogger(__name__) |
| 57 | |
| 58 | |
| 59 | # ────────────────────────────────────────────────────────────────────────────── |
| 60 | # Schema version |
| 61 | # ────────────────────────────────────────────────────────────────────────────── |
| 62 | |
| 63 | #: Current frontmatter schema version. Bumped independently of Muse semver. |
| 64 | #: Phase 1.3 baseline = "1". Phase 1.7 (``attachments`` field) = "2". |
| 65 | FRONTMATTER_SCHEMA_VERSION: str = "2" |
| 66 | |
| 67 | #: All schema version strings this module knows how to migrate through, ordered. |
| 68 | SCHEMA_VERSIONS: tuple[str, ...] = ("1", "2") |
| 69 | |
| 70 | |
| 71 | # ────────────────────────────────────────────────────────────────────────────── |
| 72 | # Slug / tag normalisation (mirrors SPEC §1) |
| 73 | # ────────────────────────────────────────────────────────────────────────────── |
| 74 | |
| 75 | _NON_SLUG_RE: re.Pattern[str] = re.compile(r"[^a-z0-9\-]") |
| 76 | _EDGE_HYPHEN_RE: re.Pattern[str] = re.compile(r"^-+|-+$") |
| 77 | |
| 78 | |
| 79 | def normalize_slug(value: str) -> str: |
| 80 | """Normalise *value* to a SPEC §1 slug. |
| 81 | |
| 82 | Lowercases the string, replaces every character that is not ``[a-z0-9-]`` |
| 83 | with a hyphen, then strips leading/trailing hyphens. |
| 84 | |
| 85 | Args: |
| 86 | value: Arbitrary string (e.g. a project name or tag). |
| 87 | |
| 88 | Returns: |
| 89 | Slug string that satisfies SPEC §1 normalisation. |
| 90 | |
| 91 | Examples:: |
| 92 | |
| 93 | >>> normalize_slug("Born Free") |
| 94 | 'born-free' |
| 95 | >>> normalize_slug(" hello_world! ") |
| 96 | 'hello-world-' # trailing hyphen stripped → 'hello-world' |
| 97 | """ |
| 98 | lowered = value.lower() |
| 99 | slugged = _NON_SLUG_RE.sub("-", lowered) |
| 100 | return _EDGE_HYPHEN_RE.sub("", slugged) |
| 101 | |
| 102 | |
| 103 | # ────────────────────────────────────────────────────────────────────────────── |
| 104 | # Value coercion helpers |
| 105 | # ────────────────────────────────────────────────────────────────────────────── |
| 106 | |
| 107 | |
| 108 | def _coerce_optional_str(value: Any) -> str | None: |
| 109 | """Return *value* as ``str`` when non-``None``, else ``None``.""" |
| 110 | if value is None: |
| 111 | return None |
| 112 | return str(value) |
| 113 | |
| 114 | |
| 115 | def _coerce_str_or_list(value: Any) -> list[str]: |
| 116 | """Coerce a scalar-or-list YAML value to a ``list[str]``. |
| 117 | |
| 118 | Handles: |
| 119 | - ``None`` → ``[]`` |
| 120 | - YAML list → each element stringified |
| 121 | - A bare string → split on commas (``"a, b, c"`` → ``["a", "b", "c"]``) |
| 122 | - Any other scalar → ``[str(value)]`` |
| 123 | |
| 124 | This covers both the ``tags: [foo, bar]`` (YAML list) and |
| 125 | ``tags: foo, bar`` (comma-separated scalar) forms permitted by SPEC §2.1. |
| 126 | """ |
| 127 | if value is None: |
| 128 | return [] |
| 129 | if isinstance(value, list): |
| 130 | return [str(v) for v in value] |
| 131 | if isinstance(value, str): |
| 132 | parts = [p.strip() for p in value.split(",")] |
| 133 | return [p for p in parts if p] |
| 134 | return [str(value)] |
| 135 | |
| 136 | |
| 137 | # ────────────────────────────────────────────────────────────────────────────── |
| 138 | # All recognised top-level keys (used to populate ``extra``) |
| 139 | # ────────────────────────────────────────────────────────────────────────────── |
| 140 | |
| 141 | _KNOWN_KEYS: frozenset[str] = frozenset( |
| 142 | { |
| 143 | # §2.1 Common |
| 144 | "title", |
| 145 | "project", |
| 146 | "tags", |
| 147 | "date", |
| 148 | "updated", |
| 149 | # §2.2 Inbox |
| 150 | "source", |
| 151 | "source_id", |
| 152 | # Import-pipeline alias |
| 153 | "source_type", |
| 154 | # §2.3 Temporal / intention |
| 155 | "follows", |
| 156 | "causal_chain_id", |
| 157 | "entity", |
| 158 | "episode_id", |
| 159 | "summarizes", |
| 160 | "summarizes_range", |
| 161 | "state_snapshot", |
| 162 | # Phase 1.7 — mist attachment IDs |
| 163 | "attachments", |
| 164 | # §2.4 Reserved Phase 12 |
| 165 | "network", |
| 166 | "wallet_address", |
| 167 | "tx_hash", |
| 168 | "payment_status", |
| 169 | } |
| 170 | ) |
| 171 | |
| 172 | |
| 173 | # ────────────────────────────────────────────────────────────────────────────── |
| 174 | # ParsedFrontmatter dataclass |
| 175 | # ────────────────────────────────────────────────────────────────────────────── |
| 176 | |
| 177 | |
| 178 | @dataclass |
| 179 | class ParsedFrontmatter: |
| 180 | """Fully typed representation of a Knowtation note's YAML frontmatter. |
| 181 | |
| 182 | Every field mirrors SPEC §2 (§2.1 common, §2.2 inbox, §2.3 temporal, §2.4 |
| 183 | reserved Phase-12) plus the import-pipeline ``source_type`` alias and the |
| 184 | Phase 1.7 ``attachments`` list. |
| 185 | |
| 186 | All list fields default to ``[]``; optional string fields default to |
| 187 | ``None``; ``state_snapshot`` defaults to ``False``. |
| 188 | |
| 189 | Unknown keys are preserved verbatim in :attr:`extra` so that |
| 190 | ``parse → to_dict → parse`` round-trips are lossless. |
| 191 | """ |
| 192 | |
| 193 | # ── §2.1 Common ────────────────────────────────────────────────────────── |
| 194 | title: str | None = None |
| 195 | project: str | None = None |
| 196 | tags: list[str] = field(default_factory=list) |
| 197 | date: str | None = None |
| 198 | updated: str | None = None |
| 199 | |
| 200 | # ── §2.2 Inbox ─────────────────────────────────────────────────────────── |
| 201 | source: str | None = None |
| 202 | source_id: str | None = None |
| 203 | |
| 204 | #: Import-pipeline alias for ``source``. Kept for backward-compat reading; |
| 205 | #: new notes should use ``source`` only (SPEC §2.2). |
| 206 | source_type: str | None = None |
| 207 | |
| 208 | # ── §2.3 Intention and temporal ────────────────────────────────────────── |
| 209 | follows: list[str] = field(default_factory=list) |
| 210 | causal_chain_id: str | None = None |
| 211 | entity: list[str] = field(default_factory=list) |
| 212 | episode_id: str | None = None |
| 213 | summarizes: list[str] = field(default_factory=list) |
| 214 | summarizes_range: str | None = None |
| 215 | state_snapshot: bool = False |
| 216 | |
| 217 | # ── Phase 1.7 — mist attachment IDs ───────────────────────────────────── |
| 218 | attachments: list[str] = field(default_factory=list) |
| 219 | |
| 220 | # ── §2.4 Reserved Phase 12 ─────────────────────────────────────────────── |
| 221 | #: Parsers MUST NOT require these; stored as-is without validation. |
| 222 | network: str | None = None |
| 223 | wallet_address: str | None = None |
| 224 | tx_hash: str | None = None |
| 225 | payment_status: str | None = None |
| 226 | |
| 227 | # ── Pass-through for unknown / future keys ──────────────────────────────── |
| 228 | extra: dict[str, Any] = field(default_factory=dict) |
| 229 | |
| 230 | # ── Schema metadata ─────────────────────────────────────────────────────── |
| 231 | #: Schema version at parse time, from :data:`FRONTMATTER_SCHEMA_VERSION`. |
| 232 | schema_version: str = FRONTMATTER_SCHEMA_VERSION |
| 233 | |
| 234 | # ── Computed properties ─────────────────────────────────────────────────── |
| 235 | |
| 236 | @property |
| 237 | def effective_source(self) -> str | None: |
| 238 | """Return ``source`` when set, else ``source_type`` (import-pipeline alias). |
| 239 | |
| 240 | Callers that only care *whether* a source is present should use this |
| 241 | rather than checking both fields separately. |
| 242 | """ |
| 243 | return self.source or self.source_type |
| 244 | |
| 245 | @property |
| 246 | def is_inbox_note(self) -> bool: |
| 247 | """Return ``True`` when the note satisfies the SPEC §2.2 inbox contract. |
| 248 | |
| 249 | An inbox note must have both ``source`` (or ``source_type``) and ``date``. |
| 250 | """ |
| 251 | return self.effective_source is not None and self.date is not None |
| 252 | |
| 253 | def to_dict(self) -> dict[str, Any]: |
| 254 | """Serialise back to a plain ``dict`` (suitable for ``yaml.dump``). |
| 255 | |
| 256 | Only non-empty / non-None / non-False fields are included so that the |
| 257 | output remains clean for human readers. ``extra`` keys are merged last |
| 258 | (known fields take precedence if there is a collision). |
| 259 | """ |
| 260 | out: dict[str, Any] = {} |
| 261 | |
| 262 | # §2.1 |
| 263 | if self.title is not None: |
| 264 | out["title"] = self.title |
| 265 | if self.project is not None: |
| 266 | out["project"] = self.project |
| 267 | if self.tags: |
| 268 | out["tags"] = self.tags |
| 269 | if self.date is not None: |
| 270 | out["date"] = self.date |
| 271 | if self.updated is not None: |
| 272 | out["updated"] = self.updated |
| 273 | |
| 274 | # §2.2 |
| 275 | if self.source is not None: |
| 276 | out["source"] = self.source |
| 277 | if self.source_id is not None: |
| 278 | out["source_id"] = self.source_id |
| 279 | if self.source_type is not None: |
| 280 | out["source_type"] = self.source_type |
| 281 | |
| 282 | # §2.3 |
| 283 | if self.follows: |
| 284 | out["follows"] = self.follows |
| 285 | if self.causal_chain_id is not None: |
| 286 | out["causal_chain_id"] = self.causal_chain_id |
| 287 | if self.entity: |
| 288 | out["entity"] = self.entity |
| 289 | if self.episode_id is not None: |
| 290 | out["episode_id"] = self.episode_id |
| 291 | if self.summarizes: |
| 292 | out["summarizes"] = self.summarizes |
| 293 | if self.summarizes_range is not None: |
| 294 | out["summarizes_range"] = self.summarizes_range |
| 295 | if self.state_snapshot: |
| 296 | out["state_snapshot"] = self.state_snapshot |
| 297 | |
| 298 | # Phase 1.7 |
| 299 | if self.attachments: |
| 300 | out["attachments"] = self.attachments |
| 301 | |
| 302 | # §2.4 reserved |
| 303 | if self.network is not None: |
| 304 | out["network"] = self.network |
| 305 | if self.wallet_address is not None: |
| 306 | out["wallet_address"] = self.wallet_address |
| 307 | if self.tx_hash is not None: |
| 308 | out["tx_hash"] = self.tx_hash |
| 309 | if self.payment_status is not None: |
| 310 | out["payment_status"] = self.payment_status |
| 311 | |
| 312 | # Unknown pass-through keys (known fields already written above) |
| 313 | for k, v in self.extra.items(): |
| 314 | if k not in out: |
| 315 | out[k] = v |
| 316 | |
| 317 | return out |
| 318 | |
| 319 | |
| 320 | # ────────────────────────────────────────────────────────────────────────────── |
| 321 | # Raw YAML block extraction |
| 322 | # ────────────────────────────────────────────────────────────────────────────── |
| 323 | |
| 324 | |
| 325 | def _extract_frontmatter_block(content: bytes) -> str | None: |
| 326 | """Return the raw YAML string inside the leading ``---`` delimiters. |
| 327 | |
| 328 | Recognises both ``---`` … ``---`` and ``---`` … ``...`` (YAML end-of-document |
| 329 | marker). Returns ``None`` when no valid frontmatter block is found. |
| 330 | |
| 331 | Args: |
| 332 | content: Raw bytes of the note file. |
| 333 | |
| 334 | Returns: |
| 335 | YAML string between the delimiters, or ``None``. |
| 336 | """ |
| 337 | try: |
| 338 | text = content.decode("utf-8", errors="replace") |
| 339 | except Exception: |
| 340 | return None |
| 341 | |
| 342 | if not text.startswith("---"): |
| 343 | return None |
| 344 | |
| 345 | rest = text[3:] |
| 346 | end = -1 |
| 347 | for delim in ("\n---", "\n..."): |
| 348 | pos = rest.find(delim) |
| 349 | if pos != -1 and (end == -1 or pos < end): |
| 350 | end = pos |
| 351 | |
| 352 | if end == -1: |
| 353 | return None |
| 354 | |
| 355 | return rest[:end] |
| 356 | |
| 357 | |
| 358 | # ────────────────────────────────────────────────────────────────────────────── |
| 359 | # Public parse API |
| 360 | # ────────────────────────────────────────────────────────────────────────────── |
| 361 | |
| 362 | |
| 363 | def parse_frontmatter(content: bytes) -> ParsedFrontmatter | None: |
| 364 | """Parse the YAML frontmatter block of a note into a typed object. |
| 365 | |
| 366 | Uses ``yaml.safe_load`` — no arbitrary Python object instantiation. |
| 367 | |
| 368 | Args: |
| 369 | content: Raw bytes of the note file (Markdown, CRLF or LF, UTF-8). |
| 370 | |
| 371 | Returns: |
| 372 | A :class:`ParsedFrontmatter` when a valid frontmatter block is found, |
| 373 | or ``None`` when the file has no frontmatter, has a YAML parse error, |
| 374 | or the frontmatter is not a mapping. |
| 375 | |
| 376 | Notes: |
| 377 | - ``tags``, ``follows``, ``entity``, ``summarizes``, ``attachments`` |
| 378 | accept either a YAML list or a comma-separated string. |
| 379 | - ``source_type`` is stored as-is; call :attr:`~ParsedFrontmatter.effective_source` |
| 380 | to get the canonical source regardless of which key was used. |
| 381 | - Unknown keys are preserved in :attr:`~ParsedFrontmatter.extra`. |
| 382 | """ |
| 383 | raw = _extract_frontmatter_block(content) |
| 384 | if raw is None: |
| 385 | return None |
| 386 | |
| 387 | try: |
| 388 | data = yaml.safe_load(raw) |
| 389 | except yaml.YAMLError as exc: |
| 390 | logger.debug("YAMLError parsing frontmatter block: %s", exc) |
| 391 | return None |
| 392 | |
| 393 | if not isinstance(data, dict): |
| 394 | return None |
| 395 | |
| 396 | extra: dict[str, Any] = {k: v for k, v in data.items() if k not in _KNOWN_KEYS} |
| 397 | |
| 398 | return ParsedFrontmatter( |
| 399 | # §2.1 |
| 400 | title=_coerce_optional_str(data.get("title")), |
| 401 | project=_coerce_optional_str(data.get("project")), |
| 402 | tags=_coerce_str_or_list(data.get("tags")), |
| 403 | date=_coerce_optional_str(data.get("date")), |
| 404 | updated=_coerce_optional_str(data.get("updated")), |
| 405 | # §2.2 |
| 406 | source=_coerce_optional_str(data.get("source")), |
| 407 | source_id=_coerce_optional_str(data.get("source_id")), |
| 408 | source_type=_coerce_optional_str(data.get("source_type")), |
| 409 | # §2.3 |
| 410 | follows=_coerce_str_or_list(data.get("follows")), |
| 411 | causal_chain_id=_coerce_optional_str(data.get("causal_chain_id")), |
| 412 | entity=_coerce_str_or_list(data.get("entity")), |
| 413 | episode_id=_coerce_optional_str(data.get("episode_id")), |
| 414 | summarizes=_coerce_str_or_list(data.get("summarizes")), |
| 415 | summarizes_range=_coerce_optional_str(data.get("summarizes_range")), |
| 416 | state_snapshot=bool(data.get("state_snapshot", False)), |
| 417 | # Phase 1.7 |
| 418 | attachments=_coerce_str_or_list(data.get("attachments")), |
| 419 | # §2.4 |
| 420 | network=_coerce_optional_str(data.get("network")), |
| 421 | wallet_address=_coerce_optional_str(data.get("wallet_address")), |
| 422 | tx_hash=_coerce_optional_str(data.get("tx_hash")), |
| 423 | payment_status=_coerce_optional_str(data.get("payment_status")), |
| 424 | extra=extra, |
| 425 | schema_version=FRONTMATTER_SCHEMA_VERSION, |
| 426 | ) |
| 427 | |
| 428 | |
| 429 | # ────────────────────────────────────────────────────────────────────────────── |
| 430 | # Mist-ID validation constants (Phase 1.7) |
| 431 | # ────────────────────────────────────────────────────────────────────────────── |
| 432 | |
| 433 | # Mist attachment IDs are the first 12 characters of the base58-encoded |
| 434 | # SHA-256 hash of the blob content, per the Mist object-store manifest spec. |
| 435 | # Base58 alphabet excludes 0/O/I/l to prevent transcription errors. |
| 436 | _BASE58_ALPHABET: str = "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz" |
| 437 | _MIST_ID_RE: re.Pattern[str] = re.compile( |
| 438 | r"^[123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz]{12}$" |
| 439 | ) |
| 440 | |
| 441 | |
| 442 | # ────────────────────────────────────────────────────────────────────────────── |
| 443 | # Validation |
| 444 | # ────────────────────────────────────────────────────────────────────────────── |
| 445 | |
| 446 | |
| 447 | def validate_inbox_note(fm: ParsedFrontmatter) -> list[str]: |
| 448 | """Validate *fm* against the SPEC §2.2 inbox-note required-field contract. |
| 449 | |
| 450 | Args: |
| 451 | fm: A parsed frontmatter object (from :func:`parse_frontmatter`). |
| 452 | |
| 453 | Returns: |
| 454 | List of human-readable error strings. Empty list means valid. |
| 455 | """ |
| 456 | errors: list[str] = [] |
| 457 | if fm.effective_source is None: |
| 458 | errors.append( |
| 459 | "Inbox note MUST have 'source' (or 'source_type') field (SPEC §2.2)." |
| 460 | ) |
| 461 | if fm.date is None: |
| 462 | errors.append("Inbox note MUST have 'date' field (SPEC §2.2).") |
| 463 | return errors |
| 464 | |
| 465 | |
| 466 | def validate_attachment_ids(fm: ParsedFrontmatter) -> list[str]: |
| 467 | """Validate that every entry in *fm.attachments* is a well-formed mist ID. |
| 468 | |
| 469 | A valid mist ID is exactly 12 characters from the base58 alphabet |
| 470 | (``123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz``). |
| 471 | This matches the first 12 characters of the base58-encoded SHA-256 hash |
| 472 | produced by the Mist object-store push command. |
| 473 | |
| 474 | Args: |
| 475 | fm: A parsed frontmatter object. |
| 476 | |
| 477 | Returns: |
| 478 | List of human-readable error strings; empty list means all IDs are |
| 479 | valid (or the ``attachments`` field is empty). |
| 480 | """ |
| 481 | errors: list[str] = [] |
| 482 | for i, mist_id in enumerate(fm.attachments): |
| 483 | if not isinstance(mist_id, str): |
| 484 | errors.append( |
| 485 | f"attachments[{i}]: expected a string, got {type(mist_id).__name__!r}." |
| 486 | ) |
| 487 | elif not _MIST_ID_RE.match(mist_id): |
| 488 | errors.append( |
| 489 | f"attachments[{i}]: {mist_id!r} is not a valid mist ID " |
| 490 | f"(must be exactly 12 base58 characters)." |
| 491 | ) |
| 492 | return errors |
| 493 | |
| 494 | |
| 495 | def validate_slug_field(value: str, field_name: str) -> list[str]: |
| 496 | """Return validation errors if *value* is not a valid SPEC §1 slug. |
| 497 | |
| 498 | Args: |
| 499 | value: The field value to check. |
| 500 | field_name: Name used in error messages (e.g. ``"project"``). |
| 501 | |
| 502 | Returns: |
| 503 | List of error strings; empty means the value is a valid slug. |
| 504 | """ |
| 505 | normalized = normalize_slug(value) |
| 506 | if normalized != value: |
| 507 | return [ |
| 508 | f"Field '{field_name}' value '{value}' is not a valid slug; " |
| 509 | f"normalized form is '{normalized}'." |
| 510 | ] |
| 511 | return [] |
| 512 | |
| 513 | |
| 514 | # ────────────────────────────────────────────────────────────────────────────── |
| 515 | # Schema migration |
| 516 | # ────────────────────────────────────────────────────────────────────────────── |
| 517 | |
| 518 | |
| 519 | def migrate_frontmatter(data: dict[str, Any], from_version: str) -> dict[str, Any]: |
| 520 | """Migrate a raw frontmatter dict from *from_version* to the current schema. |
| 521 | |
| 522 | Migrations are applied sequentially so ``from_version="1"`` will apply |
| 523 | v1→v2, v2→v3, … until :data:`FRONTMATTER_SCHEMA_VERSION` is reached. |
| 524 | |
| 525 | When *from_version* already equals :data:`FRONTMATTER_SCHEMA_VERSION`, the |
| 526 | dict is returned unchanged (no copy). |
| 527 | |
| 528 | Defined migrations |
| 529 | ------------------ |
| 530 | v1 → v2 (stub, Phase 1.7): |
| 531 | Phase 1.7 will make ``attachments`` a first-class field. At that |
| 532 | point this step will normalise any legacy attachment key that lived in |
| 533 | ``extra``. For now the migration is a no-op so callers can exercise |
| 534 | the pipeline end-to-end before the field name is locked. |
| 535 | |
| 536 | Args: |
| 537 | data: Raw frontmatter dict as returned by ``yaml.safe_load``. |
| 538 | from_version: Schema version string stored alongside the data. |
| 539 | |
| 540 | Returns: |
| 541 | Migrated dict, or the original dict when no migration is needed. |
| 542 | |
| 543 | Raises: |
| 544 | ValueError: When *from_version* is not in :data:`SCHEMA_VERSIONS`. |
| 545 | """ |
| 546 | if from_version == FRONTMATTER_SCHEMA_VERSION: |
| 547 | return data |
| 548 | |
| 549 | if from_version not in SCHEMA_VERSIONS: |
| 550 | raise ValueError( |
| 551 | f"Unknown frontmatter schema version '{from_version}'. " |
| 552 | f"Known versions: {SCHEMA_VERSIONS}." |
| 553 | ) |
| 554 | |
| 555 | current = data.copy() |
| 556 | start_idx = SCHEMA_VERSIONS.index(from_version) |
| 557 | target_idx = SCHEMA_VERSIONS.index(FRONTMATTER_SCHEMA_VERSION) |
| 558 | |
| 559 | for step in range(start_idx, target_idx): |
| 560 | step_key = (SCHEMA_VERSIONS[step], SCHEMA_VERSIONS[step + 1]) |
| 561 | migration_fn = _MIGRATION_TABLE.get(step_key) |
| 562 | if migration_fn is not None: |
| 563 | current = migration_fn(current) |
| 564 | |
| 565 | return current |
| 566 | |
| 567 | |
| 568 | def _migrate_v1_to_v2(data: dict[str, Any]) -> dict[str, Any]: |
| 569 | """v1 → v2: promote the ``attachments`` field to first-class status. |
| 570 | |
| 571 | In schema v1, notes that stored mist blob IDs placed them under various |
| 572 | ad-hoc keys (``attachment``, ``mist_id``, ``mist_ids``). v2 normalises |
| 573 | all of these into the canonical ``attachments: [<mist-id>, ...]`` list. |
| 574 | |
| 575 | Migration steps applied (in order): |
| 576 | 1. ``attachment`` (singular) → appended to ``attachments`` list, key removed. |
| 577 | 2. ``mist_id`` (singular) → appended to ``attachments`` list, key removed. |
| 578 | 3. ``mist_ids`` (list) → items merged into ``attachments`` list, key removed. |
| 579 | 4. Existing ``attachments`` value is coerced to a list via |
| 580 | :func:`_coerce_str_or_list` so that a bare scalar is accepted. |
| 581 | |
| 582 | Args: |
| 583 | data: Raw frontmatter dict (from ``yaml.safe_load``). |
| 584 | |
| 585 | Returns: |
| 586 | New dict with the normalised ``attachments`` key. |
| 587 | """ |
| 588 | result = data.copy() |
| 589 | |
| 590 | # Collect attachments from all legacy locations. |
| 591 | merged: list[str] = [] |
| 592 | |
| 593 | existing = result.pop("attachments", None) |
| 594 | if existing is not None: |
| 595 | merged.extend(_coerce_str_or_list(existing)) |
| 596 | |
| 597 | for legacy_key in ("attachment", "mist_id"): |
| 598 | val = result.pop(legacy_key, None) |
| 599 | if val is not None: |
| 600 | merged.extend(_coerce_str_or_list(val)) |
| 601 | |
| 602 | legacy_list = result.pop("mist_ids", None) |
| 603 | if legacy_list is not None: |
| 604 | merged.extend(_coerce_str_or_list(legacy_list)) |
| 605 | |
| 606 | if merged: |
| 607 | result["attachments"] = merged |
| 608 | |
| 609 | return result |
| 610 | |
| 611 | |
| 612 | _MIGRATION_TABLE: dict[tuple[str, str], Any] = { |
| 613 | ("1", "2"): _migrate_v1_to_v2, |
| 614 | } |
File History
1 commit
sha256:cf6265cea8c21d9228d90dec13ef6ec2dab5103d466db9cc4590681832de4bf8
docs(KD-STAGING): sync governance after KD-6b DONE
Human
12 days ago