proposal_merge_strategies.py
python
sha256:7281683f5c41e5d88b6d8811fbdafebd3e01a0c9dcd90975cfcb444ba71e8e81
docs: add local source-of-truth for musehub#225, #226, #227…
Sonnet 5
16 hours ago
| 1 | """Merge strategy engine for proposal merges — issue #37 Phase 3. |
| 2 | |
| 3 | Five strategies, each producing a ``MergeResult`` from two branch manifests: |
| 4 | |
| 5 | OVERLAY — from_branch wins all file conflicts (current default) |
| 6 | WEAVE — three-way at file level: ancestor base isolates true conflicts |
| 7 | REPLAY — apply from_branch delta (vs ancestor) on top of to_branch |
| 8 | SELECTIVE — only apply from_branch changes for files in selective_domains |
| 9 | PHASED — merge in dependency-DAG order; each phase is an OVERLAY |
| 10 | of one dep's delta; yields a phase audit trail |
| 11 | |
| 12 | Domain classification uses file extension and path-prefix heuristics. The |
| 13 | ``DomainClassifier`` is the single authority — all strategies call it. |
| 14 | |
| 15 | All functions are pure with respect to DB I/O — they operate on |
| 16 | ``StrDict`` manifests (``dict[str, str]``, path → object_id) that callers |
| 17 | have already loaded from the snapshot service. |
| 18 | """ |
| 19 | |
| 20 | from __future__ import annotations |
| 21 | |
| 22 | import logging |
| 23 | from dataclasses import dataclass, field |
| 24 | from pathlib import PurePosixPath |
| 25 | from typing import Sequence |
| 26 | |
| 27 | from muse.core.merge_engine import STRATEGY_MAP, run_merge |
| 28 | from muse.domain import MergeResult as MuseMergeResult |
| 29 | from musehub.types.json_types import StrDict |
| 30 | |
| 31 | logger = logging.getLogger(__name__) |
| 32 | |
| 33 | # ───────────────────────────────────────────────────────────────────────────── |
| 34 | # Domain classification |
| 35 | # ───────────────────────────────────────────────────────────────────────────── |
| 36 | |
| 37 | # Extension sets per domain — order matters for the classifier (first match wins) |
| 38 | _DOMAIN_EXTENSIONS: dict[str, frozenset[str]] = { |
| 39 | "midi": frozenset({".mid", ".midi", ".smf"}), |
| 40 | "stem": frozenset({".wav", ".aiff", ".aif", ".flac", ".mp3", ".ogg", ".opus", ".m4a"}), |
| 41 | "payment": frozenset({".mpay", ".claim", ".settle"}), |
| 42 | "identity": frozenset({".toml"}) |
| 43 | if False else # .toml is too broad — use path-prefix below |
| 44 | frozenset({".mpay"}), # placeholder; identity matched by path prefix |
| 45 | "code": frozenset({ |
| 46 | ".py", ".ts", ".tsx", ".js", ".jsx", ".go", ".rs", ".java", |
| 47 | ".c", ".cpp", ".h", ".hpp", ".cs", ".rb", ".swift", ".kt", |
| 48 | ".sh", ".bash", ".zsh", ".fish", ".sql", ".yaml", ".yml", |
| 49 | ".toml", ".json", ".md", ".rst", ".txt", ".cfg", ".ini", |
| 50 | ".dockerfile", ".makefile", |
| 51 | }), |
| 52 | } |
| 53 | |
| 54 | # Path prefixes that override extension-based classification |
| 55 | _DOMAIN_PATH_PREFIXES: dict[str, tuple[str, ...]] = { |
| 56 | "identity": ("identity/", ".muse/identity", "keys/", "auth/"), |
| 57 | "payment": ("payments/", "claims/", "ledger/", "settle/"), |
| 58 | "midi": ("midi/", "tracks/", "sequences/"), |
| 59 | "stem": ("stems/", "audio/", "samples/", "recordings/"), |
| 60 | } |
| 61 | |
| 62 | |
| 63 | def classify_domain(path: str) -> str: |
| 64 | """Return the domain name for a file path. |
| 65 | |
| 66 | Priority: path-prefix rules → extension rules → fallback "code". |
| 67 | """ |
| 68 | lower = path.lower() |
| 69 | for domain, prefixes in _DOMAIN_PATH_PREFIXES.items(): |
| 70 | if any(lower.startswith(p) for p in prefixes): |
| 71 | return domain |
| 72 | ext = PurePosixPath(path).suffix.lower() |
| 73 | for domain, exts in _DOMAIN_EXTENSIONS.items(): |
| 74 | if ext in exts: |
| 75 | return domain |
| 76 | return "code" |
| 77 | |
| 78 | |
| 79 | def paths_for_domains(manifest: StrDict, domains: Sequence[str]) -> frozenset[str]: |
| 80 | """Return the set of paths in ``manifest`` that belong to any of ``domains``.""" |
| 81 | domain_set = frozenset(domains) |
| 82 | return frozenset(p for p in manifest if classify_domain(p) in domain_set) |
| 83 | |
| 84 | |
| 85 | # ───────────────────────────────────────────────────────────────────────────── |
| 86 | # Result types |
| 87 | # ───────────────────────────────────────────────────────────────────────────── |
| 88 | |
| 89 | |
| 90 | @dataclass |
| 91 | class ConflictEntry: |
| 92 | """A file where both branches modified the content since the common ancestor.""" |
| 93 | path: str |
| 94 | to_branch_object_id: str |
| 95 | from_branch_object_id: str |
| 96 | resolution: str = "from_wins" # from_wins | to_wins | manual_required |
| 97 | |
| 98 | |
| 99 | @dataclass |
| 100 | class PhaseResult: |
| 101 | """Outcome of one phase in a PHASED merge.""" |
| 102 | phase_index: int |
| 103 | dependency_proposal_id: str |
| 104 | files_applied: int |
| 105 | domains_touched: list[str] |
| 106 | |
| 107 | |
| 108 | @dataclass |
| 109 | class MergeResult: |
| 110 | """The output of a merge strategy execution. |
| 111 | |
| 112 | ``manifest`` is the final merged ``{path: object_id}`` mapping that callers |
| 113 | should persist as a new snapshot. |
| 114 | |
| 115 | ``conflicts`` contains entries where both branches diverged from the ancestor; |
| 116 | for strategies that auto-resolve (OVERLAY, PHASED) these are recorded |
| 117 | but do not block the merge. WEAVE surfaces them for review. |
| 118 | |
| 119 | ``domains_merged`` lists the domains actually touched by from_branch changes. |
| 120 | For SELECTIVE, it reflects only the requested subset. |
| 121 | """ |
| 122 | strategy: str |
| 123 | manifest: StrDict |
| 124 | files_added: int = 0 |
| 125 | files_modified: int = 0 |
| 126 | files_removed: int = 0 |
| 127 | files_skipped: int = 0 |
| 128 | conflicts: list[ConflictEntry] = field(default_factory=list) |
| 129 | domains_merged: list[str] = field(default_factory=list) |
| 130 | phase_results: list[PhaseResult] = field(default_factory=list) |
| 131 | # Populated by STATE_WEAVE: ancestor snapshot_id used for three-way base |
| 132 | ancestor_snapshot_id: str | None = None |
| 133 | |
| 134 | |
| 135 | # ───────────────────────────────────────────────────────────────────────────── |
| 136 | # Shared manifest diff helpers |
| 137 | # ───────────────────────────────────────────────────────────────────────────── |
| 138 | |
| 139 | |
| 140 | def _manifest_delta( |
| 141 | base: StrDict, |
| 142 | head: StrDict, |
| 143 | ) -> tuple[set[str], set[str], set[str]]: |
| 144 | """Return (added, modified, removed) path sets from base→head.""" |
| 145 | base_paths = set(base) |
| 146 | head_paths = set(head) |
| 147 | added = head_paths - base_paths |
| 148 | removed = base_paths - head_paths |
| 149 | modified = {p for p in base_paths & head_paths if base[p] != head[p]} |
| 150 | return added, modified, removed |
| 151 | |
| 152 | |
| 153 | def _apply_delta( |
| 154 | target: StrDict, |
| 155 | delta_head: StrDict, |
| 156 | delta_base: StrDict, |
| 157 | ) -> tuple[StrDict, int, int, int]: |
| 158 | """Apply changes from delta_base→delta_head onto target. |
| 159 | |
| 160 | Returns (merged_manifest, added, modified, removed). |
| 161 | """ |
| 162 | result = dict(target) |
| 163 | added, modified, removed = _manifest_delta(delta_base, delta_head) |
| 164 | for path in added | modified: |
| 165 | result[path] = delta_head[path] |
| 166 | for path in removed: |
| 167 | result.pop(path, None) |
| 168 | return result, len(added), len(modified), len(removed) |
| 169 | |
| 170 | |
| 171 | def _domain_summary(paths: set[str]) -> list[str]: |
| 172 | """Return sorted unique domain names for a set of paths.""" |
| 173 | return sorted({classify_domain(p) for p in paths}) |
| 174 | |
| 175 | |
| 176 | # ───────────────────────────────────────────────────────────────────────────── |
| 177 | # Strategy implementations |
| 178 | # ───────────────────────────────────────────────────────────────────────────── |
| 179 | |
| 180 | |
| 181 | def merge_overlay( |
| 182 | to_manifest: StrDict, |
| 183 | from_manifest: StrDict, |
| 184 | *, |
| 185 | ancestor_manifest: StrDict | None = None, |
| 186 | ) -> MergeResult: |
| 187 | """OVERLAY — from_branch wins all conflicts. |
| 188 | |
| 189 | The simplest and fastest strategy: the merged manifest is |
| 190 | ``{**to_manifest, **from_manifest}``. Files only on to_branch are kept; |
| 191 | files only on from_branch are added; files on both use from_branch's |
| 192 | version. Ancestor is used only to populate the conflicts list for audit. |
| 193 | """ |
| 194 | merged: StrDict = {**to_manifest, **from_manifest} |
| 195 | |
| 196 | added, modified, removed = _manifest_delta(to_manifest, merged) |
| 197 | # removed relative to to_manifest = paths in to but not in from |
| 198 | actual_removed = set(to_manifest) - set(from_manifest) |
| 199 | actual_added = set(from_manifest) - set(to_manifest) |
| 200 | actual_modified = {p for p in set(from_manifest) & set(to_manifest) if from_manifest[p] != to_manifest[p]} |
| 201 | |
| 202 | conflicts: list[ConflictEntry] = [] |
| 203 | if ancestor_manifest is not None: |
| 204 | # Paths changed on BOTH branches since ancestor = true conflicts |
| 205 | _, to_modified, _ = _manifest_delta(ancestor_manifest, to_manifest) |
| 206 | _, from_modified, _ = _manifest_delta(ancestor_manifest, from_manifest) |
| 207 | for path in to_modified & from_modified: |
| 208 | if to_manifest.get(path) != from_manifest.get(path): |
| 209 | conflicts.append(ConflictEntry( |
| 210 | path=path, |
| 211 | to_branch_object_id=to_manifest[path], |
| 212 | from_branch_object_id=from_manifest[path], |
| 213 | resolution="from_wins", |
| 214 | )) |
| 215 | |
| 216 | changed_paths = actual_added | actual_modified |
| 217 | return MergeResult( |
| 218 | strategy="overlay", |
| 219 | manifest=merged, |
| 220 | files_added=len(actual_added), |
| 221 | files_modified=len(actual_modified), |
| 222 | files_removed=len(actual_removed), |
| 223 | conflicts=conflicts, |
| 224 | domains_merged=_domain_summary(changed_paths), |
| 225 | ) |
| 226 | |
| 227 | |
| 228 | def merge_weave( |
| 229 | to_manifest: StrDict, |
| 230 | from_manifest: StrDict, |
| 231 | *, |
| 232 | ancestor_manifest: StrDict, |
| 233 | ) -> MergeResult: |
| 234 | """WEAVE — three-way file-level merge with conflict surfacing. |
| 235 | |
| 236 | Uses the common ancestor to distinguish true conflicts (both sides changed |
| 237 | the same file) from clean additions (only one side changed it). |
| 238 | |
| 239 | Resolution rules per file: |
| 240 | - Only to_branch changed it → keep to_branch version (clean) |
| 241 | - Only from_branch changed it → take from_branch version (clean) |
| 242 | - Both changed it differently → from_branch wins; record ConflictEntry |
| 243 | - Neither changed it → keep ancestor version |
| 244 | - Added only on from_branch → include (clean) |
| 245 | - Added only on to_branch → include (clean) |
| 246 | - Added on both, different → from_branch wins; ConflictEntry |
| 247 | - Removed on from_branch → exclude |
| 248 | - Removed on to_branch → exclude (respect both deletions) |
| 249 | """ |
| 250 | all_paths = set(to_manifest) | set(from_manifest) | set(ancestor_manifest) |
| 251 | merged: StrDict = {} |
| 252 | conflicts: list[ConflictEntry] = [] |
| 253 | added = modified = removed = 0 |
| 254 | |
| 255 | for path in all_paths: |
| 256 | in_anc = path in ancestor_manifest |
| 257 | in_to = path in to_manifest |
| 258 | in_frm = path in from_manifest |
| 259 | |
| 260 | anc_id = ancestor_manifest.get(path) |
| 261 | to_id = to_manifest.get(path) |
| 262 | frm_id = from_manifest.get(path) |
| 263 | |
| 264 | to_changed = in_to and (not in_anc or to_id != anc_id) |
| 265 | frm_changed = in_frm and (not in_anc or frm_id != anc_id) |
| 266 | to_deleted = in_anc and not in_to |
| 267 | frm_deleted = in_anc and not in_frm |
| 268 | |
| 269 | if frm_deleted: |
| 270 | # from_branch deleted it — honour the deletion |
| 271 | if in_anc and in_to and not to_changed: |
| 272 | removed += 1 # only in ancestor + to (unchanged) — delete |
| 273 | # if to_branch also changed it, we still honour the deletion (from wins) |
| 274 | elif in_anc: |
| 275 | removed += 1 |
| 276 | continue # don't include in merged |
| 277 | |
| 278 | if to_deleted and not frm_changed: |
| 279 | # to_branch deleted it, from_branch didn't change it — honour deletion |
| 280 | removed += 1 |
| 281 | continue |
| 282 | |
| 283 | if frm_changed and to_changed and frm_id != to_id: |
| 284 | # True conflict — from_branch wins but record for review |
| 285 | merged[path] = frm_id # type: ignore[assignment] |
| 286 | conflicts.append(ConflictEntry( |
| 287 | path=path, |
| 288 | to_branch_object_id=to_id, # type: ignore[arg-type] |
| 289 | from_branch_object_id=frm_id, # type: ignore[arg-type] |
| 290 | resolution="from_wins", |
| 291 | )) |
| 292 | if in_anc: |
| 293 | modified += 1 |
| 294 | else: |
| 295 | added += 1 |
| 296 | elif frm_changed: |
| 297 | merged[path] = frm_id # type: ignore[assignment] |
| 298 | if in_anc: |
| 299 | modified += 1 |
| 300 | else: |
| 301 | added += 1 |
| 302 | elif to_changed: |
| 303 | merged[path] = to_id # type: ignore[assignment] |
| 304 | if in_anc: |
| 305 | modified += 1 |
| 306 | else: |
| 307 | added += 1 |
| 308 | elif in_anc: |
| 309 | # Neither side changed it — keep ancestor value |
| 310 | merged[path] = anc_id # type: ignore[assignment] |
| 311 | |
| 312 | changed_paths = {c.path for c in conflicts} |
| 313 | changed_paths |= {p for p in merged if p not in ancestor_manifest or merged[p] != ancestor_manifest.get(p)} |
| 314 | |
| 315 | return MergeResult( |
| 316 | strategy="weave", |
| 317 | manifest=merged, |
| 318 | files_added=added, |
| 319 | files_modified=modified, |
| 320 | files_removed=removed, |
| 321 | conflicts=conflicts, |
| 322 | domains_merged=_domain_summary(changed_paths), |
| 323 | ancestor_snapshot_id=None, # caller fills this in |
| 324 | ) |
| 325 | |
| 326 | |
| 327 | def merge_replay( |
| 328 | to_manifest: StrDict, |
| 329 | from_manifest: StrDict, |
| 330 | *, |
| 331 | ancestor_manifest: StrDict, |
| 332 | ) -> MergeResult: |
| 333 | """REPLAY — apply from_branch delta (vs ancestor) on top of to_branch. |
| 334 | |
| 335 | Identifies exactly which files from_branch changed relative to the common |
| 336 | ancestor, then applies only those changes onto the current to_branch state. |
| 337 | Files that to_branch changed but from_branch didn't touch are preserved |
| 338 | untouched — this is the key difference from OVERLAY. |
| 339 | |
| 340 | Conflict rule: if the same file was changed on both sides since the ancestor, |
| 341 | from_branch wins (recorded as ConflictEntry). |
| 342 | """ |
| 343 | from_added, from_modified, from_removed = _manifest_delta(ancestor_manifest, from_manifest) |
| 344 | to_added, to_modified, _ = _manifest_delta(ancestor_manifest, to_manifest) |
| 345 | |
| 346 | merged: StrDict = dict(to_manifest) |
| 347 | conflicts: list[ConflictEntry] = [] |
| 348 | |
| 349 | # Apply from_branch additions and modifications |
| 350 | for path in from_added | from_modified: |
| 351 | if path in to_modified or path in to_added: |
| 352 | if to_manifest.get(path) != from_manifest[path]: |
| 353 | conflicts.append(ConflictEntry( |
| 354 | path=path, |
| 355 | to_branch_object_id=to_manifest.get(path, ""), |
| 356 | from_branch_object_id=from_manifest[path], |
| 357 | resolution="from_wins", |
| 358 | )) |
| 359 | merged[path] = from_manifest[path] |
| 360 | |
| 361 | # Apply from_branch removals |
| 362 | for path in from_removed: |
| 363 | merged.pop(path, None) |
| 364 | |
| 365 | changed = from_added | from_modified |
| 366 | return MergeResult( |
| 367 | strategy="replay", |
| 368 | manifest=merged, |
| 369 | files_added=len(from_added), |
| 370 | files_modified=len(from_modified), |
| 371 | files_removed=len(from_removed), |
| 372 | conflicts=conflicts, |
| 373 | domains_merged=_domain_summary(changed), |
| 374 | ) |
| 375 | |
| 376 | |
| 377 | def merge_selective( |
| 378 | to_manifest: StrDict, |
| 379 | from_manifest: StrDict, |
| 380 | *, |
| 381 | selective_domains: list[str], |
| 382 | ancestor_manifest: StrDict | None = None, |
| 383 | ) -> MergeResult: |
| 384 | """SELECTIVE — only apply from_branch changes for the specified domains. |
| 385 | |
| 386 | Files outside ``selective_domains`` remain exactly as they are on to_branch. |
| 387 | Files inside the domains use from_branch's version (OVERLAY semantics |
| 388 | scoped to the domain subset). |
| 389 | |
| 390 | Conflicts are recorded for files in the selected domains that were changed |
| 391 | on both sides since the ancestor. |
| 392 | """ |
| 393 | if not selective_domains: |
| 394 | raise ValueError("SELECTIVE strategy requires at least one domain in selective_domains") |
| 395 | |
| 396 | selected = frozenset(selective_domains) |
| 397 | merged: StrDict = dict(to_manifest) |
| 398 | conflicts: list[ConflictEntry] = [] |
| 399 | added = modified = removed = skipped = 0 |
| 400 | |
| 401 | # Collect all paths that from_branch wants to change |
| 402 | from_added_raw, from_modified_raw, from_removed_raw = _manifest_delta( |
| 403 | ancestor_manifest or to_manifest, from_manifest |
| 404 | ) |
| 405 | |
| 406 | for path in from_added_raw | from_modified_raw: |
| 407 | if classify_domain(path) in selected: |
| 408 | if path in to_manifest and to_manifest[path] != from_manifest[path]: |
| 409 | if ancestor_manifest: |
| 410 | anc_id = ancestor_manifest.get(path) |
| 411 | to_changed = to_manifest[path] != anc_id |
| 412 | if to_changed: |
| 413 | conflicts.append(ConflictEntry( |
| 414 | path=path, |
| 415 | to_branch_object_id=to_manifest[path], |
| 416 | from_branch_object_id=from_manifest[path], |
| 417 | resolution="from_wins", |
| 418 | )) |
| 419 | modified += 1 |
| 420 | else: |
| 421 | added += 1 |
| 422 | merged[path] = from_manifest[path] |
| 423 | else: |
| 424 | skipped += 1 |
| 425 | |
| 426 | for path in from_removed_raw: |
| 427 | if classify_domain(path) in selected: |
| 428 | merged.pop(path, None) |
| 429 | removed += 1 |
| 430 | else: |
| 431 | skipped += 1 |
| 432 | |
| 433 | return MergeResult( |
| 434 | strategy="selective", |
| 435 | manifest=merged, |
| 436 | files_added=added, |
| 437 | files_modified=modified, |
| 438 | files_removed=removed, |
| 439 | files_skipped=skipped, |
| 440 | conflicts=conflicts, |
| 441 | domains_merged=sorted(selective_domains), |
| 442 | ) |
| 443 | |
| 444 | |
| 445 | def merge_phased( |
| 446 | to_manifest: StrDict, |
| 447 | from_manifest: StrDict, |
| 448 | *, |
| 449 | ancestor_manifest: StrDict | None = None, |
| 450 | dependency_order: list[str] | None = None, |
| 451 | phase_manifests: dict[str, StrDict] | None = None, |
| 452 | ) -> MergeResult: |
| 453 | """PHASED — merge in dependency-DAG topological order. |
| 454 | |
| 455 | When ``phase_manifests`` is provided (a map from proposal_id → manifest for |
| 456 | each dependency), the merge applies each dependency's delta in Kahn order, |
| 457 | building up a cumulative result. This ensures that intermediate state is |
| 458 | consistent at each phase boundary. |
| 459 | |
| 460 | When ``phase_manifests`` is not provided (the common case: all deps already |
| 461 | merged), falls back to STATE_OVERLAY semantics on the final manifests and |
| 462 | records a single phase result covering the entire merge. |
| 463 | |
| 464 | ``dependency_order``: topological order of proposal IDs (from proposal_dag). |
| 465 | """ |
| 466 | phase_results: list[PhaseResult] = [] |
| 467 | |
| 468 | if phase_manifests and dependency_order: |
| 469 | # Full phased execution: apply each dep delta in order |
| 470 | current = dict(to_manifest) |
| 471 | prev_manifest = dict(ancestor_manifest or to_manifest) |
| 472 | total_added = total_modified = total_removed = 0 |
| 473 | all_conflicts: list[ConflictEntry] = [] |
| 474 | |
| 475 | for phase_idx, dep_id in enumerate(dependency_order): |
| 476 | dep_manifest = phase_manifests.get(dep_id, {}) |
| 477 | if not dep_manifest: |
| 478 | continue |
| 479 | |
| 480 | phase_merged, p_added, p_modified, p_removed = _apply_delta( |
| 481 | current, dep_manifest, prev_manifest |
| 482 | ) |
| 483 | delta_paths = set(dep_manifest) - set(prev_manifest) |
| 484 | delta_paths |= {p for p in dep_manifest if dep_manifest.get(p) != prev_manifest.get(p)} |
| 485 | |
| 486 | phase_results.append(PhaseResult( |
| 487 | phase_index=phase_idx, |
| 488 | dependency_proposal_id=dep_id, |
| 489 | files_applied=p_added + p_modified + p_removed, |
| 490 | domains_touched=_domain_summary(delta_paths), |
| 491 | )) |
| 492 | current = phase_merged |
| 493 | prev_manifest = dict(dep_manifest) |
| 494 | total_added += p_added |
| 495 | total_modified += p_modified |
| 496 | total_removed += p_removed |
| 497 | |
| 498 | # Final overlay: apply the proposal's own changes on top |
| 499 | final, f_added, f_modified, f_removed = _apply_delta( |
| 500 | current, from_manifest, prev_manifest |
| 501 | ) |
| 502 | phase_results.append(PhaseResult( |
| 503 | phase_index=len(dependency_order), |
| 504 | dependency_proposal_id="self", |
| 505 | files_applied=f_added + f_modified + f_removed, |
| 506 | domains_touched=_domain_summary(set(from_manifest)), |
| 507 | )) |
| 508 | |
| 509 | changed = {p for p in final if final.get(p) != to_manifest.get(p)} |
| 510 | return MergeResult( |
| 511 | strategy="phased", |
| 512 | manifest=final, |
| 513 | files_added=total_added + f_added, |
| 514 | files_modified=total_modified + f_modified, |
| 515 | files_removed=total_removed + f_removed, |
| 516 | conflicts=all_conflicts, |
| 517 | domains_merged=_domain_summary(changed), |
| 518 | phase_results=phase_results, |
| 519 | ) |
| 520 | |
| 521 | # Fallback: deps already merged into to_branch — single-phase overlay |
| 522 | result = merge_overlay(to_manifest, from_manifest, ancestor_manifest=ancestor_manifest) |
| 523 | result.strategy = "phased" |
| 524 | changed_paths = {p for p in from_manifest if from_manifest.get(p) != to_manifest.get(p)} |
| 525 | changed_paths |= set(from_manifest) - set(to_manifest) |
| 526 | result.phase_results = [PhaseResult( |
| 527 | phase_index=0, |
| 528 | dependency_proposal_id="self", |
| 529 | files_applied=result.files_added + result.files_modified + result.files_removed, |
| 530 | domains_touched=_domain_summary(changed_paths), |
| 531 | )] |
| 532 | return result |
| 533 | |
| 534 | |
| 535 | def merge_cherry_pick( |
| 536 | to_manifest: StrDict, |
| 537 | *, |
| 538 | cherry_pick_commits: list[tuple[StrDict, StrDict]], |
| 539 | ) -> MergeResult: |
| 540 | """CHERRY_PICK — apply the delta of each specified commit onto to_branch. |
| 541 | |
| 542 | Each entry in ``cherry_pick_commits`` is a ``(parent_manifest, commit_manifest)`` |
| 543 | pair. The delta introduced by each commit (relative to its parent) is applied |
| 544 | in order onto the accumulating result. Files not touched by any picked commit |
| 545 | are preserved exactly as they are on to_branch. |
| 546 | |
| 547 | Conflict rule: if a file is both changed by a picked commit and already |
| 548 | differs on to_branch relative to the commit's parent, from_wins is recorded. |
| 549 | """ |
| 550 | if not cherry_pick_commits: |
| 551 | raise ValueError("cherry_pick strategy requires at least one commit in cherry_pick_commits") |
| 552 | |
| 553 | current: StrDict = dict(to_manifest) |
| 554 | conflicts: list[ConflictEntry] = [] |
| 555 | total_added = total_modified = total_removed = 0 |
| 556 | |
| 557 | for parent_manifest, commit_manifest in cherry_pick_commits: |
| 558 | added, modified, removed = _manifest_delta(parent_manifest, commit_manifest) |
| 559 | |
| 560 | for path in added | modified: |
| 561 | frm_id = commit_manifest[path] |
| 562 | cur_id = current.get(path) |
| 563 | par_id = parent_manifest.get(path) |
| 564 | if cur_id is not None and cur_id != frm_id and cur_id != par_id: |
| 565 | conflicts.append(ConflictEntry( |
| 566 | path=path, |
| 567 | to_branch_object_id=cur_id, |
| 568 | from_branch_object_id=frm_id, |
| 569 | resolution="from_wins", |
| 570 | )) |
| 571 | if path not in current: |
| 572 | total_added += 1 |
| 573 | elif cur_id != frm_id: |
| 574 | total_modified += 1 |
| 575 | current[path] = frm_id |
| 576 | |
| 577 | for path in removed: |
| 578 | if path in current: |
| 579 | current.pop(path) |
| 580 | total_removed += 1 |
| 581 | |
| 582 | changed = {p for p in current if current.get(p) != to_manifest.get(p)} |
| 583 | changed |= set(to_manifest) - set(current) |
| 584 | |
| 585 | return MergeResult( |
| 586 | strategy="cherry_pick", |
| 587 | manifest=current, |
| 588 | files_added=total_added, |
| 589 | files_modified=total_modified, |
| 590 | files_removed=total_removed, |
| 591 | conflicts=conflicts, |
| 592 | domains_merged=_domain_summary(changed), |
| 593 | ) |
| 594 | |
| 595 | |
| 596 | # ───────────────────────────────────────────────────────────────────────────── |
| 597 | # Strategy router |
| 598 | # ───────────────────────────────────────────────────────────────────────────── |
| 599 | |
| 600 | |
| 601 | def _to_local_result( |
| 602 | strategy: str, |
| 603 | muse_result: MuseMergeResult, |
| 604 | to_manifest: StrDict, |
| 605 | from_manifest: StrDict, |
| 606 | ) -> "MergeResult": |
| 607 | """Convert a muse.domain.MergeResult to the local MergeResult shape.""" |
| 608 | merged: StrDict = dict(muse_result.merged["files"]) # type: ignore[attr-defined] |
| 609 | to_keys = set(to_manifest) |
| 610 | merged_keys = set(merged) |
| 611 | files_added = len(merged_keys - to_keys) |
| 612 | files_removed = len(to_keys - merged_keys) |
| 613 | files_modified = len({p for p in to_keys & merged_keys if to_manifest[p] != merged[p]}) |
| 614 | conflicts = [ |
| 615 | ConflictEntry( |
| 616 | path=p, |
| 617 | to_branch_object_id=to_manifest.get(p, ""), |
| 618 | from_branch_object_id=from_manifest.get(p, ""), |
| 619 | resolution="manual_required", |
| 620 | ) |
| 621 | for p in muse_result.conflicts # type: ignore[attr-defined] |
| 622 | ] |
| 623 | changed = {p for p in merged if merged.get(p) != to_manifest.get(p)} |
| 624 | changed |= to_keys - merged_keys |
| 625 | return MergeResult( |
| 626 | strategy=strategy, |
| 627 | manifest=merged, |
| 628 | files_added=files_added, |
| 629 | files_modified=files_modified, |
| 630 | files_removed=files_removed, |
| 631 | conflicts=conflicts, |
| 632 | domains_merged=_domain_summary(changed), |
| 633 | ) |
| 634 | |
| 635 | |
| 636 | def execute_merge_strategy( |
| 637 | strategy: str, |
| 638 | to_manifest: StrDict, |
| 639 | from_manifest: StrDict, |
| 640 | *, |
| 641 | ancestor_manifest: StrDict | None = None, |
| 642 | selective_domains: list[str] | None = None, |
| 643 | dependency_order: list[str] | None = None, |
| 644 | phase_manifests: dict[str, StrDict] | None = None, |
| 645 | cherry_pick_commits: list[tuple[StrDict, StrDict]] | None = None, |
| 646 | ) -> MergeResult: |
| 647 | """Route to the correct strategy implementation. |
| 648 | |
| 649 | Args: |
| 650 | strategy: One of the MergeStrategy values. |
| 651 | to_manifest: Current to_branch manifest. |
| 652 | from_manifest: Current from_branch manifest. |
| 653 | ancestor_manifest: Common ancestor manifest. Required for weave |
| 654 | and replay; used for conflict detection in |
| 655 | overlay and selective when provided. |
| 656 | selective_domains: Required for selective. |
| 657 | dependency_order: Topological order of dep IDs for phased. |
| 658 | phase_manifests: Per-dep manifests for phased (optional). |
| 659 | cherry_pick_commits: Ordered (parent_manifest, commit_manifest) pairs |
| 660 | for cherry_pick. |
| 661 | |
| 662 | Raises: |
| 663 | ValueError: Unknown strategy or missing required parameters. |
| 664 | """ |
| 665 | # Canonical strategies route through STRATEGY_MAP and run_merge() — single |
| 666 | # source of truth shared with muse.core.merge_engine. |
| 667 | engine = STRATEGY_MAP.get(strategy) |
| 668 | if engine is not None: |
| 669 | # musehub#182: "overlay"'s diff_unit is "snapshot", which run_merge() |
| 670 | # forces to an EMPTY base regardless of what ancestor_manifest we |
| 671 | # pass — i.e. "from_branch wins" for every path it happens to carry, |
| 672 | # even ones it never touched since the fork. For a long-lived |
| 673 | # from_branch forked well before to_branch's later, unrelated |
| 674 | # changes, this silently reverts/deletes them (confirmed: real |
| 675 | # incident, 47 files, musehub#144 proposal #9). When a real ancestor |
| 676 | # is available, swap to the "theirs" engine instead — same |
| 677 | # diff_unit family, same "from_branch wins conflicts" resolution, |
| 678 | # but a proper three-way diff against the real ancestor so paths |
| 679 | # from_branch never touched are left alone. Only the untouched-path |
| 680 | # behavior changes; genuine conflicts still resolve to from_branch, |
| 681 | # matching overlay's documented contract. |
| 682 | if strategy == "overlay" and ancestor_manifest is not None: |
| 683 | engine = STRATEGY_MAP["theirs"] |
| 684 | |
| 685 | needs_ancestor = engine.diff_unit in ("three_way", "replay_ours", "replay_theirs") |
| 686 | if needs_ancestor and ancestor_manifest is None: |
| 687 | logger.warning( |
| 688 | "%s requested but no ancestor manifest available; falling back to overlay", |
| 689 | strategy, |
| 690 | ) |
| 691 | result = merge_overlay(to_manifest, from_manifest) |
| 692 | result.strategy = strategy |
| 693 | return result |
| 694 | muse_result = run_merge( |
| 695 | ancestor_manifest or {}, |
| 696 | to_manifest, |
| 697 | from_manifest, |
| 698 | engine, |
| 699 | ) |
| 700 | return _to_local_result(strategy, muse_result, to_manifest, from_manifest) |
| 701 | |
| 702 | # Non-canonical strategies: weave, selective, phased, cherry_pick |
| 703 | if strategy == "weave": |
| 704 | if ancestor_manifest is None: |
| 705 | logger.warning( |
| 706 | "weave requested but no ancestor manifest available; " |
| 707 | "falling back to overlay" |
| 708 | ) |
| 709 | result = merge_overlay(to_manifest, from_manifest) |
| 710 | result.strategy = "weave" |
| 711 | return result |
| 712 | return merge_weave( |
| 713 | to_manifest, from_manifest, ancestor_manifest=ancestor_manifest |
| 714 | ) |
| 715 | elif strategy == "selective": |
| 716 | return merge_selective( |
| 717 | to_manifest, |
| 718 | from_manifest, |
| 719 | selective_domains=selective_domains or [], |
| 720 | ancestor_manifest=ancestor_manifest, |
| 721 | ) |
| 722 | elif strategy == "phased": |
| 723 | return merge_phased( |
| 724 | to_manifest, |
| 725 | from_manifest, |
| 726 | ancestor_manifest=ancestor_manifest, |
| 727 | dependency_order=dependency_order, |
| 728 | phase_manifests=phase_manifests, |
| 729 | ) |
| 730 | elif strategy == "cherry_pick": |
| 731 | return merge_cherry_pick( |
| 732 | to_manifest, |
| 733 | cherry_pick_commits=cherry_pick_commits or [], |
| 734 | ) |
| 735 | else: |
| 736 | raise ValueError(f"Unknown merge strategy: {strategy!r}") |
File History
1 commit
sha256:7281683f5c41e5d88b6d8811fbdafebd3e01a0c9dcd90975cfcb444ba71e8e81
docs: add local source-of-truth for musehub#225, #226, #227…
Sonnet 5
16 hours ago