strategies.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Knowtation-specific merge strategy implementations. |
| 2 | |
| 3 | These functions are called by the Knowtation merger (Phase 2.4) when |
| 4 | :func:`~muse.core.attributes.resolve_strategy` returns a Knowtation-specific |
| 5 | strategy string. They share a common signature:: |
| 6 | |
| 7 | fn(ours: bytes, theirs: bytes, base: bytes) -> bytes |
| 8 | |
| 9 | *ours* — the current-branch (left) version of the note. |
| 10 | *theirs* — the incoming-branch (right) version of the note. |
| 11 | *base* — the common merge-base version of the note. |
| 12 | |
| 13 | All three arguments are raw Markdown bytes (UTF-8, LF line endings normalised |
| 14 | by the caller). Return value is the resolved content as raw bytes. |
| 15 | |
| 16 | Strategies defined here |
| 17 | ----------------------- |
| 18 | |
| 19 | ``prefer-newer-date`` → :func:`prefer_newer_date` |
| 20 | ``union-sorted`` → :func:`union_sorted` |
| 21 | ``knowtation-3way`` → :func:`knowtation_3way` |
| 22 | |
| 23 | Security notes |
| 24 | -------------- |
| 25 | - Frontmatter is parsed with ``yaml.safe_load`` only. |
| 26 | - Date comparisons use ``datetime.date.fromisoformat`` — invalid strings are |
| 27 | caught and treated as missing rather than propagating exceptions. |
| 28 | - Input size is not capped here (the merge engine is responsible for that); |
| 29 | however, none of the implementations do anything super-linear on input size. |
| 30 | |
| 31 | OBA review note (prefer-newer-date fallback) |
| 32 | -------------------------------------------- |
| 33 | The fallback hierarchy was chosen after analysis of the three failure modes: |
| 34 | |
| 35 | 1. **Both dates missing**: No temporal signal — default to ``ours`` (current |
| 36 | branch is the authoritative capture point). |
| 37 | 2. **One date missing**: The versioned note carries deliberate metadata; |
| 38 | prefer the versioned side regardless of which branch it is on. |
| 39 | 3. **Dates equal**: Tie-break with ``ours`` (current branch wins; import |
| 40 | pipelines that produce the same timestamp for the same event will never |
| 41 | trigger an unwanted remote override). |
| 42 | 4. **Date malformed** (e.g. ``date: "yesterday"``): Treat as missing; do not |
| 43 | crash — SPEC §2.2 allows free-form date strings for human notes. |
| 44 | """ |
| 45 | |
| 46 | from __future__ import annotations |
| 47 | |
| 48 | import logging |
| 49 | from datetime import date, datetime |
| 50 | from typing import Any, Callable |
| 51 | |
| 52 | logger = logging.getLogger(__name__) |
| 53 | |
| 54 | # Strategy function type. |
| 55 | StrategyFn = Callable[[bytes, bytes, bytes], bytes] |
| 56 | |
| 57 | |
| 58 | # --------------------------------------------------------------------------- |
| 59 | # Internal helpers |
| 60 | # --------------------------------------------------------------------------- |
| 61 | |
| 62 | |
| 63 | def _parse_date(fm_date: str | None) -> date | None: |
| 64 | """Parse a SPEC §2.1 date string → ``datetime.date``, or ``None``. |
| 65 | |
| 66 | Accepts ISO 8601 datetime strings and plain ``YYYY-MM-DD`` dates. |
| 67 | Returns ``None`` for any string that cannot be parsed. |
| 68 | |
| 69 | Args: |
| 70 | fm_date: Value of the ``date`` frontmatter field, or ``None``. |
| 71 | |
| 72 | Returns: |
| 73 | A :class:`datetime.date` on success, or ``None`` on failure. |
| 74 | """ |
| 75 | if not fm_date: |
| 76 | return None |
| 77 | # Try plain date first (most common). |
| 78 | try: |
| 79 | return date.fromisoformat(str(fm_date).strip()[:10]) |
| 80 | except (ValueError, TypeError): |
| 81 | pass |
| 82 | # Try full ISO 8601 datetime (e.g. "2025-01-15T14:30:00Z"). |
| 83 | try: |
| 84 | return datetime.fromisoformat(str(fm_date).strip()[:19]).date() |
| 85 | except (ValueError, TypeError): |
| 86 | return None |
| 87 | |
| 88 | |
| 89 | def _parse_fm_safe(content: bytes) -> dict[str, Any] | None: |
| 90 | """Extract raw frontmatter dict from *content* without raising. |
| 91 | |
| 92 | Returns ``None`` when the note has no frontmatter or the YAML is invalid. |
| 93 | """ |
| 94 | from muse.plugins.knowtation.parser import _extract_frontmatter_block |
| 95 | |
| 96 | import yaml |
| 97 | |
| 98 | raw = _extract_frontmatter_block(content) |
| 99 | if raw is None: |
| 100 | return None |
| 101 | try: |
| 102 | data = yaml.safe_load(raw) |
| 103 | except yaml.YAMLError: |
| 104 | return None |
| 105 | return data if isinstance(data, dict) else None |
| 106 | |
| 107 | |
| 108 | def _reconstruct( |
| 109 | original: bytes, |
| 110 | fm_data: dict[str, Any], |
| 111 | ) -> bytes: |
| 112 | """Serialise *fm_data* back into a YAML frontmatter block and re-attach body. |
| 113 | |
| 114 | The note body (everything after the closing ``---``) is preserved verbatim |
| 115 | from *original*. |
| 116 | |
| 117 | Args: |
| 118 | original: Original note bytes (used to extract the body). |
| 119 | fm_data: Frontmatter dict to serialise. |
| 120 | |
| 121 | Returns: |
| 122 | Bytes with rebuilt frontmatter followed by the original body. |
| 123 | """ |
| 124 | import yaml |
| 125 | |
| 126 | try: |
| 127 | text = original.decode("utf-8", errors="replace") |
| 128 | except Exception: |
| 129 | text = "" |
| 130 | |
| 131 | # Extract body: everything after the closing --- delimiter. |
| 132 | body = "" |
| 133 | if text.startswith("---"): |
| 134 | rest = text[3:] |
| 135 | for delim in ("\n---", "\n..."): |
| 136 | pos = rest.find(delim) |
| 137 | if pos != -1: |
| 138 | body = rest[pos + len(delim):] |
| 139 | break |
| 140 | |
| 141 | fm_yaml = yaml.dump( |
| 142 | fm_data, |
| 143 | allow_unicode=True, |
| 144 | default_flow_style=False, |
| 145 | sort_keys=False, |
| 146 | ) |
| 147 | return f"---\n{fm_yaml}---\n{body}".encode("utf-8") |
| 148 | |
| 149 | |
| 150 | def _coerce_list(value: Any) -> list[str]: |
| 151 | """Coerce a YAML value to a sorted list of strings (set semantics).""" |
| 152 | from muse.plugins.knowtation.parser import _coerce_str_or_list |
| 153 | return _coerce_str_or_list(value) |
| 154 | |
| 155 | |
| 156 | # Set-valued frontmatter fields that support union/union-sorted semantics. |
| 157 | _SET_FIELDS: frozenset[str] = frozenset( |
| 158 | {"tags", "entity", "follows", "summarizes", "attachments"} |
| 159 | ) |
| 160 | |
| 161 | |
| 162 | # --------------------------------------------------------------------------- |
| 163 | # prefer-newer-date |
| 164 | # --------------------------------------------------------------------------- |
| 165 | |
| 166 | |
| 167 | def prefer_newer_date(ours: bytes, theirs: bytes, base: bytes) -> bytes: |
| 168 | """Return whichever note has the later ``date`` frontmatter field. |
| 169 | |
| 170 | Implements the ``prefer-newer-date`` strategy. The *base* argument is |
| 171 | accepted for interface compatibility but is not used — the decision is |
| 172 | purely between *ours* and *theirs*. |
| 173 | |
| 174 | Fallback hierarchy (see module docstring for OBA rationale): |
| 175 | |
| 176 | 1. Both parseable, *theirs* is strictly later → return *theirs*. |
| 177 | 2. Otherwise → return *ours*. |
| 178 | |
| 179 | Args: |
| 180 | ours: Current-branch note bytes. |
| 181 | theirs: Incoming-branch note bytes. |
| 182 | base: Common merge-base note bytes (not used). |
| 183 | |
| 184 | Returns: |
| 185 | The resolved note bytes. |
| 186 | """ |
| 187 | fm_ours = _parse_fm_safe(ours) |
| 188 | fm_theirs = _parse_fm_safe(theirs) |
| 189 | |
| 190 | date_ours = _parse_date(fm_ours.get("date") if fm_ours else None) |
| 191 | date_theirs = _parse_date(fm_theirs.get("date") if fm_theirs else None) |
| 192 | |
| 193 | # Case: only theirs has a parseable date → prefer theirs (more intentional). |
| 194 | if date_ours is None and date_theirs is not None: |
| 195 | logger.debug("prefer-newer-date: ours has no parseable date; returning theirs") |
| 196 | return theirs |
| 197 | |
| 198 | # Case: only ours has a parseable date, or both missing → ours. |
| 199 | if date_theirs is None: |
| 200 | return ours |
| 201 | |
| 202 | # Both parseable: strictly later → theirs; equal or earlier → ours. |
| 203 | if date_theirs > date_ours: |
| 204 | logger.debug( |
| 205 | "prefer-newer-date: theirs (%s) > ours (%s); returning theirs", |
| 206 | date_theirs, |
| 207 | date_ours, |
| 208 | ) |
| 209 | return theirs |
| 210 | |
| 211 | return ours |
| 212 | |
| 213 | |
| 214 | # --------------------------------------------------------------------------- |
| 215 | # union-sorted |
| 216 | # --------------------------------------------------------------------------- |
| 217 | |
| 218 | |
| 219 | def union_sorted(ours: bytes, theirs: bytes, base: bytes) -> bytes: |
| 220 | """Return a union of both notes' set-valued frontmatter fields, sorted. |
| 221 | |
| 222 | Implements the ``union-sorted`` strategy: |
| 223 | |
| 224 | - **Set fields** (``tags``, ``entity``, ``follows``, ``summarizes``, |
| 225 | ``attachments``): union of both sides, deduplicated and sorted |
| 226 | alphabetically. |
| 227 | - **Scalar fields**: ours wins (current-branch is authoritative for |
| 228 | non-set values). |
| 229 | - **Body**: ours is returned verbatim (body merging is handled by the |
| 230 | section-level merger in Phase 2.4). |
| 231 | |
| 232 | Args: |
| 233 | ours: Current-branch note bytes. |
| 234 | theirs: Incoming-branch note bytes. |
| 235 | base: Common merge-base note bytes (used to detect deletions). |
| 236 | |
| 237 | Returns: |
| 238 | Resolved note bytes with sorted set-valued frontmatter. |
| 239 | """ |
| 240 | fm_ours = _parse_fm_safe(ours) |
| 241 | fm_theirs = _parse_fm_safe(theirs) |
| 242 | |
| 243 | if fm_ours is None: |
| 244 | return ours |
| 245 | if fm_theirs is None: |
| 246 | return ours |
| 247 | |
| 248 | fm_base = _parse_fm_safe(base) or {} |
| 249 | |
| 250 | # Build merged frontmatter starting from ours. |
| 251 | merged: dict[str, Any] = dict(fm_ours) |
| 252 | |
| 253 | for field_name in _SET_FIELDS: |
| 254 | set_ours = set(_coerce_list(fm_ours.get(field_name))) |
| 255 | set_theirs = set(_coerce_list(fm_theirs.get(field_name))) |
| 256 | set_base = set(_coerce_list(fm_base.get(field_name))) |
| 257 | |
| 258 | # Union of additions from both sides; honour deletions agreed by both. |
| 259 | ours_deletions = set_base - set_ours |
| 260 | theirs_deletions = set_base - set_theirs |
| 261 | both_deleted = ours_deletions & theirs_deletions |
| 262 | |
| 263 | union = (set_ours | set_theirs) - both_deleted |
| 264 | |
| 265 | if union: |
| 266 | merged[field_name] = sorted(union) |
| 267 | elif field_name in merged: |
| 268 | del merged[field_name] |
| 269 | |
| 270 | return _reconstruct(ours, merged) |
| 271 | |
| 272 | |
| 273 | # --------------------------------------------------------------------------- |
| 274 | # knowtation-3way |
| 275 | # --------------------------------------------------------------------------- |
| 276 | |
| 277 | |
| 278 | def knowtation_3way(ours: bytes, theirs: bytes, base: bytes) -> bytes: |
| 279 | """Delegate to the Knowtation structured three-way merger. |
| 280 | |
| 281 | Implements the ``knowtation-3way`` strategy. When the |
| 282 | :mod:`~muse.plugins.knowtation.merger` module is available (Phase 2.4), |
| 283 | calls its ``merge_notes()`` entry point. Falls back to ``ours`` with a |
| 284 | warning when the merger is not yet available, so that Phase 2.2 tests and |
| 285 | Phase 2.3 policies can be exercised before Phase 2.4 lands. |
| 286 | |
| 287 | Args: |
| 288 | ours: Current-branch note bytes. |
| 289 | theirs: Incoming-branch note bytes. |
| 290 | base: Common merge-base note bytes. |
| 291 | |
| 292 | Returns: |
| 293 | Resolved note bytes from the structured merger, or *ours* on fallback. |
| 294 | """ |
| 295 | try: |
| 296 | from muse.plugins.knowtation.merger import merge_notes |
| 297 | |
| 298 | return merge_notes(ours=ours, theirs=theirs, base=base) |
| 299 | except ImportError: |
| 300 | logger.warning( |
| 301 | "knowtation-3way: merger.py (Phase 2.4) is not yet available; " |
| 302 | "falling back to 'ours'." |
| 303 | ) |
| 304 | return ours |
| 305 | except Exception as exc: |
| 306 | logger.warning( |
| 307 | "knowtation-3way: merger raised %s; falling back to 'ours'.", |
| 308 | exc, |
| 309 | ) |
| 310 | return ours |
| 311 | |
| 312 | |
| 313 | # --------------------------------------------------------------------------- |
| 314 | # Dispatch table |
| 315 | # --------------------------------------------------------------------------- |
| 316 | |
| 317 | #: Maps each Knowtation-specific strategy name to its implementation function. |
| 318 | #: Used by the merger and any caller that needs to dispatch by name. |
| 319 | STRATEGY_DISPATCH: dict[str, StrategyFn] = { |
| 320 | "prefer-newer-date": prefer_newer_date, |
| 321 | "union-sorted": union_sorted, |
| 322 | "knowtation-3way": knowtation_3way, |
| 323 | } |
| 324 | |
| 325 | |
| 326 | def apply_strategy( |
| 327 | strategy: str, |
| 328 | ours: bytes, |
| 329 | theirs: bytes, |
| 330 | base: bytes, |
| 331 | ) -> bytes | None: |
| 332 | """Apply a Knowtation strategy by name. |
| 333 | |
| 334 | Returns the resolved bytes when *strategy* is a Knowtation-specific |
| 335 | strategy, or ``None`` when the strategy is not in |
| 336 | :data:`STRATEGY_DISPATCH` (meaning it should be handled by the generic |
| 337 | merge engine instead). |
| 338 | |
| 339 | Args: |
| 340 | strategy: Strategy string (e.g. ``"prefer-newer-date"``). |
| 341 | ours: Current-branch bytes. |
| 342 | theirs: Incoming-branch bytes. |
| 343 | base: Common merge-base bytes. |
| 344 | |
| 345 | Returns: |
| 346 | Resolved bytes, or ``None`` when the strategy is not Knowtation-specific. |
| 347 | """ |
| 348 | fn = STRATEGY_DISPATCH.get(strategy) |
| 349 | if fn is None: |
| 350 | return None |
| 351 | return fn(ours, theirs, base) |