compare.py
python
sha256:37af4ab271a64f11d7bf14b4871ebaaf5dd68fb0247521007c6b034d35aa01be
Merge branch 'fix/hub-user-read-update-path' into dev
Human
11 days ago
| 1 | """muse code compare — semantic comparison between any two historical snapshots. |
| 2 | |
| 3 | ``muse diff`` compares the working tree to HEAD. ``muse code compare`` |
| 4 | compares any two historical commits — a full semantic diff between a release |
| 5 | tag and the current HEAD, between the start and end of a sprint, between two |
| 6 | branches. |
| 7 | |
| 8 | Unlike a line-level diff, ``muse code compare`` tracks: |
| 9 | |
| 10 | - Functions / classes **added** or **removed** |
| 11 | - Symbols **renamed** or **moved** across files (not just deleted + re-added) |
| 12 | - **Signature changes** vs **implementation-only** changes |
| 13 | |
| 14 | Pass ``--semver`` to get an automatic MAJOR / MINOR / PATCH recommendation |
| 15 | based on public API changes: removed symbols → MAJOR, added symbols → MINOR, |
| 16 | implementation-only changes → PATCH. |
| 17 | |
| 18 | Usage:: |
| 19 | |
| 20 | muse code compare HEAD~10 HEAD (use commit SHAs, not ~ notation) |
| 21 | muse code compare a3f2c9 cb4afa |
| 22 | muse code compare a3f2c9 cb4afa --kind function |
| 23 | muse code compare a3f2c9 cb4afa --file billing.py |
| 24 | muse code compare a3f2c9 cb4afa --semver |
| 25 | muse code compare a3f2c9 cb4afa --stat |
| 26 | muse code compare a3f2c9 cb4afa --json |
| 27 | |
| 28 | Output:: |
| 29 | |
| 30 | Semantic comparison |
| 31 | From: a3f2c9e1 "Add billing module" |
| 32 | To: cb4afaed "Merge: release v1.0" |
| 33 | |
| 34 | src/billing.py |
| 35 | added compute_invoice_total (renamed from calculate_total) |
| 36 | modified Invoice.to_dict (signature changed) |
| 37 | moved validate_amount → src/validation.py |
| 38 | |
| 39 | src/validation.py (new file) |
| 40 | added validate_amount (moved from src/billing.py) |
| 41 | |
| 42 | 7 symbol change(s) across 3 file(s) |
| 43 | SemVer impact: MINOR |
| 44 | """ |
| 45 | |
| 46 | import argparse |
| 47 | import json |
| 48 | import logging |
| 49 | import pathlib |
| 50 | import sys |
| 51 | from typing import TypedDict |
| 52 | |
| 53 | from muse.core.types import Manifest |
| 54 | from muse.core.errors import ExitCode |
| 55 | from muse.core.repo import require_repo |
| 56 | from muse.core.refs import read_current_branch |
| 57 | from muse.core.commits import resolve_commit_ref |
| 58 | from muse.core.snapshots import get_commit_snapshot_manifest |
| 59 | from muse.core.timing import start_timer |
| 60 | from muse.core.envelope import EnvelopeJson, make_envelope |
| 61 | from muse.core.symbol_cache import load_symbol_cache |
| 62 | from muse.domain import DomainOp |
| 63 | from muse.plugins.code._query import language_of, normalise_language, symbols_for_snapshot |
| 64 | from muse.plugins.code.symbol_diff import build_diff_ops |
| 65 | from muse.core.validation import sanitize_display |
| 66 | |
| 67 | logger = logging.getLogger(__name__) |
| 68 | |
| 69 | class _OpSummary(TypedDict): |
| 70 | op: str |
| 71 | address: str |
| 72 | detail: str |
| 73 | |
| 74 | class _CommitRef(TypedDict): |
| 75 | commit_id: str |
| 76 | message: str |
| 77 | |
| 78 | class _CompareFilters(TypedDict): |
| 79 | kind: str | None |
| 80 | file: str | None |
| 81 | language: str | None |
| 82 | |
| 83 | class _CompareStat(TypedDict): |
| 84 | files_changed: int |
| 85 | symbols_added: int |
| 86 | symbols_removed: int |
| 87 | symbols_modified: int |
| 88 | semver_impact: str |
| 89 | |
| 90 | # ``from`` is a Python keyword, so use the functional TypedDict form for the |
| 91 | # domain payload, then inherit into a class-based TypedDict that also inherits |
| 92 | # EnvelopeJson — giving us proper wire-shape typing without workarounds. |
| 93 | _ComparePayload = TypedDict("_ComparePayload", { |
| 94 | "from": _CommitRef, |
| 95 | "to": _CommitRef, |
| 96 | "filters": _CompareFilters, |
| 97 | "stat": _CompareStat, |
| 98 | "ops": list[_OpSummary], |
| 99 | }) |
| 100 | |
| 101 | class _CompareJson(_ComparePayload, EnvelopeJson): |
| 102 | """Full wire shape for ``muse code compare --json``.""" |
| 103 | |
| 104 | def _first_line(message: str) -> str: |
| 105 | """Return the first non-empty line of a commit message.""" |
| 106 | for line in message.splitlines(): |
| 107 | stripped = line.strip() |
| 108 | if stripped: |
| 109 | return stripped |
| 110 | return message.strip() |
| 111 | |
| 112 | def _is_public_name(address: str) -> bool: |
| 113 | """Return True if the symbol is a public, non-import API symbol. |
| 114 | |
| 115 | Excludes: |
| 116 | - Import pseudo-symbols (``::import::*``) — dependency management, not API. |
| 117 | - Private names starting with ``_``. |
| 118 | """ |
| 119 | if "::import::" in address: |
| 120 | return False |
| 121 | name = address.split("::")[-1] if "::" in address else address |
| 122 | return not name.startswith("_") |
| 123 | |
| 124 | def _assess_semver(ops: list[DomainOp]) -> tuple[str, list[str]]: |
| 125 | """Return (bump_level, reasons) from the diff ops. |
| 126 | |
| 127 | Rules applied to public symbols only (names not starting with ``_``): |
| 128 | |
| 129 | MAJOR — symbol removed, or signature / rename / move change. |
| 130 | MINOR — symbol added (no MAJOR triggers). |
| 131 | PATCH — only implementation changes. |
| 132 | NONE — empty diff. |
| 133 | """ |
| 134 | reasons: list[str] = [] |
| 135 | has_major = False |
| 136 | has_minor = False |
| 137 | has_patch = False |
| 138 | |
| 139 | for op in ops: |
| 140 | if op["op"] == "patch": |
| 141 | for child in op["child_ops"]: |
| 142 | addr: str = child["address"] |
| 143 | if not _is_public_name(addr): |
| 144 | continue |
| 145 | if child["op"] == "delete": |
| 146 | has_major = True |
| 147 | reasons.append(f"removed: {addr}") |
| 148 | elif child["op"] == "insert": |
| 149 | has_minor = True |
| 150 | elif child["op"] == "replace": |
| 151 | ns: str = child["new_summary"] |
| 152 | if any(kw in ns for kw in ("signature", "renamed", "moved")): |
| 153 | has_major = True |
| 154 | reasons.append(f"breaking: {addr} — {ns}") |
| 155 | else: |
| 156 | has_patch = True |
| 157 | elif op["op"] == "delete": |
| 158 | addr2: str = op["address"] |
| 159 | if _is_public_name(addr2): |
| 160 | has_major = True |
| 161 | reasons.append(f"file removed: {addr2}") |
| 162 | elif op["op"] == "insert": |
| 163 | if _is_public_name(op["address"]): |
| 164 | has_minor = True |
| 165 | |
| 166 | if has_major: |
| 167 | return "MAJOR", reasons |
| 168 | if has_minor: |
| 169 | return "MINOR", [] |
| 170 | if has_patch: |
| 171 | return "PATCH", [] |
| 172 | return "NONE", [] |
| 173 | |
| 174 | def _count_ops(ops: list[DomainOp]) -> tuple[int, int, int, int]: |
| 175 | """Return (added, removed, modified, files_changed) from diff ops.""" |
| 176 | added = removed = modified = 0 |
| 177 | files: set[str] = set() |
| 178 | for op in ops: |
| 179 | if op["op"] == "patch": |
| 180 | if op["child_ops"]: |
| 181 | files.add(op["address"]) |
| 182 | for child in op["child_ops"]: |
| 183 | if child["op"] == "insert": |
| 184 | added += 1 |
| 185 | elif child["op"] == "delete": |
| 186 | removed += 1 |
| 187 | elif child["op"] == "replace": |
| 188 | modified += 1 |
| 189 | elif op["op"] == "insert": |
| 190 | files.add(op["address"]) |
| 191 | added += 1 |
| 192 | elif op["op"] == "delete": |
| 193 | files.add(op["address"]) |
| 194 | removed += 1 |
| 195 | elif op["op"] == "replace": |
| 196 | files.add(op["address"]) |
| 197 | modified += 1 |
| 198 | return added, removed, modified, len(files) |
| 199 | |
| 200 | def _format_child_op(op: DomainOp) -> str: |
| 201 | """Return a compact one-line description of a symbol-level op.""" |
| 202 | addr: str = op["address"] |
| 203 | name = addr.split("::")[-1] if "::" in addr else addr |
| 204 | if op["op"] == "insert": |
| 205 | summary: str = op["content_summary"] |
| 206 | moved = ( |
| 207 | f" (moved from {summary.split('moved from')[-1].strip()})" |
| 208 | if "moved from" in summary else "" |
| 209 | ) |
| 210 | return f" added {name}{moved}" |
| 211 | if op["op"] == "delete": |
| 212 | summary2: str = op["content_summary"] |
| 213 | moved2 = ( |
| 214 | f" (moved to {summary2.split('moved to')[-1].strip()})" |
| 215 | if "moved to" in summary2 else "" |
| 216 | ) |
| 217 | return f" removed {name}{moved2}" |
| 218 | if op["op"] == "replace": |
| 219 | ns: str = op["new_summary"] |
| 220 | detail = f" ({ns})" if ns else "" |
| 221 | return f" modified {name}{detail}" |
| 222 | return f" changed {name}" |
| 223 | |
| 224 | def _flatten_ops(ops: list[DomainOp]) -> list[_OpSummary]: |
| 225 | """Flatten all ops to a serialisable summary list.""" |
| 226 | result: list[_OpSummary] = [] |
| 227 | for op in ops: |
| 228 | if op["op"] == "patch": |
| 229 | for child in op["child_ops"]: |
| 230 | if child["op"] == "insert": |
| 231 | detail: str = child["content_summary"] |
| 232 | elif child["op"] == "delete": |
| 233 | detail = child["content_summary"] |
| 234 | elif child["op"] == "replace": |
| 235 | detail = child["new_summary"] |
| 236 | else: |
| 237 | detail = "" |
| 238 | result.append(_OpSummary( |
| 239 | op=child["op"], |
| 240 | address=child["address"], |
| 241 | detail=detail, |
| 242 | )) |
| 243 | elif op["op"] == "insert": |
| 244 | result.append(_OpSummary(op="insert", address=op["address"], detail=op["content_summary"])) |
| 245 | elif op["op"] == "delete": |
| 246 | result.append(_OpSummary(op="delete", address=op["address"], detail=op["content_summary"])) |
| 247 | elif op["op"] == "replace": |
| 248 | result.append(_OpSummary(op="replace", address=op["address"], detail=op["new_summary"])) |
| 249 | else: |
| 250 | result.append(_OpSummary(op=op["op"], address=op["address"], detail="")) |
| 251 | return result |
| 252 | |
| 253 | def _filter_manifest_by_language( |
| 254 | manifest: Manifest, language: str |
| 255 | ) -> Manifest: |
| 256 | return {p: h for p, h in manifest.items() if language_of(p) == language} |
| 257 | |
| 258 | def _filter_ops_by_file(ops: list[DomainOp], file_filter: str) -> list[DomainOp]: |
| 259 | """Keep only ops whose address is or ends with *file_filter*.""" |
| 260 | return [ |
| 261 | op for op in ops |
| 262 | if op.get("address", "") == file_filter |
| 263 | or op.get("address", "").endswith(f"/{file_filter}") |
| 264 | ] |
| 265 | |
| 266 | def register(subparsers: "argparse._SubParsersAction[argparse.ArgumentParser]") -> None: |
| 267 | """Register the compare subcommand.""" |
| 268 | parser = subparsers.add_parser( |
| 269 | "compare", |
| 270 | help="Deep semantic comparison between any two historical snapshots.", |
| 271 | description=__doc__, |
| 272 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 273 | ) |
| 274 | parser.add_argument("ref_a", metavar="REF-A", help="Base commit (older).") |
| 275 | parser.add_argument("ref_b", metavar="REF-B", help="Target commit (newer).") |
| 276 | parser.add_argument( |
| 277 | "--kind", "-k", default=None, metavar="KIND", dest="kind_filter", |
| 278 | help="Restrict to symbols of this kind (function, class, method, …).", |
| 279 | ) |
| 280 | parser.add_argument( |
| 281 | "--file", "-f", default=None, metavar="FILE", dest="file_filter", |
| 282 | help="Only show changes in this file (accepts a path suffix, e.g. 'billing.py').", |
| 283 | ) |
| 284 | parser.add_argument( |
| 285 | "--language", "-l", default=None, metavar="LANG", dest="language_filter", |
| 286 | help="Only show changes in files of this language (case-insensitive).", |
| 287 | ) |
| 288 | parser.add_argument( |
| 289 | "--stat", |
| 290 | action="store_true", |
| 291 | help="Print a summary (counts + SemVer impact) instead of the full listing.", |
| 292 | ) |
| 293 | parser.add_argument( |
| 294 | "--semver", |
| 295 | action="store_true", |
| 296 | help=( |
| 297 | "Append a MAJOR / MINOR / PATCH recommendation based on public API changes: " |
| 298 | "symbol removed → MAJOR, symbol added → MINOR, impl-only → PATCH." |
| 299 | ), |
| 300 | ) |
| 301 | parser.add_argument( |
| 302 | "--json", "-j", action="store_true", dest="json_out", |
| 303 | help="Emit results as JSON.", |
| 304 | ) |
| 305 | parser.set_defaults(func=run) |
| 306 | |
| 307 | def run(args: argparse.Namespace) -> None: |
| 308 | """Deep semantic comparison between any two historical snapshots. |
| 309 | |
| 310 | Reads both commits from the object store, parses AST symbol trees for all |
| 311 | semantic files, and produces a full symbol-level delta: added, removed, |
| 312 | renamed, moved, and modified symbols. Use it to understand the semantic |
| 313 | scope of a release, sprint, or branch divergence at the function level. |
| 314 | |
| 315 | Agent quickstart |
| 316 | ---------------- |
| 317 | :: |
| 318 | |
| 319 | muse code compare HEAD~10 HEAD --json |
| 320 | muse code compare v0.1.0 dev --kind function --json |
| 321 | muse code compare main feat/new --semver --json |
| 322 | |
| 323 | JSON fields |
| 324 | ----------- |
| 325 | ref_a First (older) ref. |
| 326 | ref_b Second (newer) ref. |
| 327 | added List of symbol addresses added in ref_b. |
| 328 | removed List of symbol addresses removed from ref_a. |
| 329 | modified List of symbol addresses present in both but changed. |
| 330 | renamed List of ``{from, to}`` rename objects. |
| 331 | moved List of ``{address, from_file, to_file}`` move objects. |
| 332 | total Total change count. |
| 333 | sem_ver_bump Inferred bump (with ``--semver``): ``"major"``, ``"minor"``, |
| 334 | ``"patch"``, or ``"none"``. |
| 335 | exit_code Integer exit code (see below). |
| 336 | |
| 337 | Exit codes |
| 338 | ---------- |
| 339 | 0 Comparison complete. |
| 340 | 1 Ref not found or invalid arguments. |
| 341 | 2 Not inside a Muse repository. |
| 342 | """ |
| 343 | elapsed = start_timer() |
| 344 | ref_a: str = args.ref_a |
| 345 | ref_b: str = args.ref_b |
| 346 | kind_filter: str | None = args.kind_filter |
| 347 | file_filter: str | None = args.file_filter |
| 348 | language_filter: str | None = args.language_filter |
| 349 | stat_only: bool = args.stat |
| 350 | show_semver: bool = args.semver |
| 351 | json_out: bool = args.json_out |
| 352 | |
| 353 | if language_filter is not None: |
| 354 | language_filter = normalise_language(language_filter) |
| 355 | if kind_filter is not None: |
| 356 | kind_filter = kind_filter.strip().lower() |
| 357 | |
| 358 | root = require_repo() |
| 359 | branch = read_current_branch(root) |
| 360 | |
| 361 | commit_a = resolve_commit_ref(root, branch, ref_a) |
| 362 | if commit_a is None: |
| 363 | print(f"❌ Commit '{ref_a}' not found.", file=sys.stderr) |
| 364 | raise SystemExit(ExitCode.USER_ERROR) |
| 365 | |
| 366 | commit_b = resolve_commit_ref(root, branch, ref_b) |
| 367 | if commit_b is None: |
| 368 | print(f"❌ Commit '{ref_b}' not found.", file=sys.stderr) |
| 369 | raise SystemExit(ExitCode.USER_ERROR) |
| 370 | |
| 371 | manifest_a: Manifest = get_commit_snapshot_manifest(root, commit_a.commit_id) or {} |
| 372 | manifest_b: Manifest = get_commit_snapshot_manifest(root, commit_b.commit_id) or {} |
| 373 | |
| 374 | # Apply language filter at the manifest level so cross-file move detection |
| 375 | # in build_diff_ops only considers the requested language. |
| 376 | if language_filter is not None: |
| 377 | manifest_a = _filter_manifest_by_language(manifest_a, language_filter) |
| 378 | manifest_b = _filter_manifest_by_language(manifest_b, language_filter) |
| 379 | |
| 380 | # Load the symbol cache once and share it across both snapshot loads. |
| 381 | shared_cache = load_symbol_cache(root) |
| 382 | |
| 383 | trees_a = symbols_for_snapshot( |
| 384 | root, manifest_a, kind_filter=kind_filter, cache=shared_cache |
| 385 | ) |
| 386 | trees_b = symbols_for_snapshot( |
| 387 | root, manifest_b, kind_filter=kind_filter, cache=shared_cache |
| 388 | ) |
| 389 | |
| 390 | ops = build_diff_ops(manifest_a, manifest_b, trees_a, trees_b) |
| 391 | |
| 392 | # Post-filter by --file (applied after build_diff_ops so cross-file move |
| 393 | # detection still runs on the full op set). |
| 394 | if file_filter is not None: |
| 395 | ops = _filter_ops_by_file(ops, file_filter) |
| 396 | |
| 397 | semver_bump, semver_reasons = _assess_semver(ops) |
| 398 | added, removed, modified, files_changed = _count_ops(ops) |
| 399 | total_symbols = added + removed + modified |
| 400 | |
| 401 | msg_a = _first_line(commit_a.message) |
| 402 | msg_b = _first_line(commit_b.message) |
| 403 | |
| 404 | if json_out: |
| 405 | print(json.dumps(_CompareJson(**make_envelope(elapsed), **{ |
| 406 | "from": _CommitRef(commit_id=commit_a.commit_id, message=msg_a), |
| 407 | "to": _CommitRef(commit_id=commit_b.commit_id, message=msg_b), |
| 408 | "filters": _CompareFilters(kind=kind_filter, file=file_filter, language=language_filter), |
| 409 | "stat": _CompareStat( |
| 410 | files_changed=files_changed, |
| 411 | symbols_added=added, |
| 412 | symbols_removed=removed, |
| 413 | symbols_modified=modified, |
| 414 | semver_impact=semver_bump, |
| 415 | ), |
| 416 | "ops": list(_flatten_ops(ops)), |
| 417 | }))) |
| 418 | return |
| 419 | |
| 420 | # ── Human-readable ──────────────────────────────────────────────────────── |
| 421 | |
| 422 | print("\nSemantic comparison") |
| 423 | print(f' From: {commit_a.commit_id} "{sanitize_display(msg_a)}"') |
| 424 | print(f' To: {commit_b.commit_id} "{sanitize_display(msg_b)}"') |
| 425 | |
| 426 | if stat_only: |
| 427 | print(f"\n Files changed: {files_changed}") |
| 428 | print(f" Symbols added: {added}") |
| 429 | print(f" Symbols removed: {removed}") |
| 430 | print(f" Symbols modified: {modified}") |
| 431 | print(f" Net change: {added - removed:+d} symbols") |
| 432 | print(f" SemVer impact: {semver_bump}") |
| 433 | if semver_reasons: |
| 434 | for r in semver_reasons[:5]: |
| 435 | print(f" → {r}") |
| 436 | return |
| 437 | |
| 438 | if not ops: |
| 439 | print("\n (no semantic changes between these two commits)") |
| 440 | return |
| 441 | |
| 442 | for op in ops: |
| 443 | if op["op"] == "patch": |
| 444 | fp = op["address"] |
| 445 | child_ops = op.get("child_ops", []) |
| 446 | if not child_ops: |
| 447 | continue |
| 448 | is_new = fp not in manifest_a |
| 449 | is_gone = fp not in manifest_b |
| 450 | suffix = " (new file)" if is_new else (" (removed)" if is_gone else "") |
| 451 | print(f"\n{sanitize_display(fp)}{suffix}") |
| 452 | for child in child_ops: |
| 453 | print(_format_child_op(child)) |
| 454 | else: |
| 455 | fp = op["address"] |
| 456 | if op["op"] == "insert": |
| 457 | print(f"\n{sanitize_display(fp)} (new file)") |
| 458 | print(f" added {sanitize_display(fp)} (file)") |
| 459 | elif op["op"] == "delete": |
| 460 | print(f"\n{sanitize_display(fp)} (removed)") |
| 461 | print(f" removed {sanitize_display(fp)} (file)") |
| 462 | else: |
| 463 | print(f"\n{sanitize_display(fp)}") |
| 464 | print(f" modified {sanitize_display(fp)} (file)") |
| 465 | |
| 466 | print(f"\n{total_symbols} symbol change(s) across {files_changed} file(s)") |
| 467 | |
| 468 | if show_semver or semver_bump in ("MAJOR",): |
| 469 | print(f"SemVer impact: {semver_bump}") |
| 470 | for r in semver_reasons[:5]: |
| 471 | print(f" → {r}") |
File History
11 commits
sha256:37af4ab271a64f11d7bf14b4871ebaaf5dd68fb0247521007c6b034d35aa01be
Merge branch 'fix/hub-user-read-update-path' into dev
Human
11 days ago
sha256:b7be56ec091919a612cffe7f3c8b600209d5155517443fdac0e16954c8c7d81f
Merge 'task/git-export-ignored-file-deletion' into 'dev' — …
Human
13 days ago
sha256:8de4334a98c945aace420969d389ad678aa926d4ab4e886b2ac4c4241cb3bf2b
revert: keep pyproject.toml in canonical PEP 440 form
Sonnet 4.6
patch
17 days ago
sha256:50bb615c573aa4b928fca75b60f82ba2a659910066507cec6a95c412ae22ccb9
docs: add issue docs for push have-negotiation bug (#55) an…
Sonnet 4.6
20 days ago
sha256:d90d175cded68aae1d4ffcf4858917854195d0cd8ce1fe73cee4dbc02541cb74
chore: trigger push to surface null-OID paths
Sonnet 4.6
26 days ago
sha256:e452ad9a6ace6ccc6d875a35e06caf9da5576a970c1c36133b69a891ce5fefa8
chore: prebuild timing test
Sonnet 4.6
36 days ago
sha256:0008ab6695e3e064b3e236b24fd19e538fef6a588eb0d211622f4466d919c0b1
merge: pull staging/dev — advance to 0.2.0rc12
Sonnet 4.6
patch
37 days ago
sha256:9c33d61749fff814c5226d5386aa2af7064c2c02788594a25fdd709358132eea
fix: _PROPOSAL_PREFIX_RESOLVE_LIMIT 200 → 100 to match hub …
Sonnet 4.6
49 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
52 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
58 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
58 days ago