gravity.py
python
sha256:18b983389ee1b55900fcd799bfbb496552d2e3ecded9d18cefbfef188947a12e
chore: remove blob-debug test marker file
Sonnet 4.6
1 day ago
| 1 | """muse code gravity — structural weight of every symbol. |
| 2 | |
| 3 | "Gravity" answers the question every architect dreads: |
| 4 | |
| 5 | **"If I change the contract of this symbol, how much of the codebase breaks?"** |
| 6 | |
| 7 | Not just the direct callers — the full transitive closure. If ``read_object`` |
| 8 | changes, that breaks ``read_commit``, which breaks ``resolve_commit_ref``, which |
| 9 | breaks every route handler, which breaks every CLI command that touches the |
| 10 | repo. That is gravity: the fraction of the production codebase that lives |
| 11 | downstream of a symbol in the call graph. |
| 12 | |
| 13 | Why this matters |
| 14 | ---------------- |
| 15 | * High-gravity symbols are **structural pillars**. Treat their public |
| 16 | contracts as APIs — design changes carefully, communicate broadly, bump |
| 17 | MAJOR. |
| 18 | * Low-gravity symbols are **safe to refactor**. They sit at the edge of the |
| 19 | dependency graph with few downstream effects. |
| 20 | * **Gravity ≠ hotspot.** A frequently-changed symbol with low gravity is a |
| 21 | quick iteration loop. A rarely-changed symbol with gravity 90% is a time |
| 22 | bomb — the moment its contract shifts, everything breaks. |
| 23 | |
| 24 | How it works |
| 25 | ------------ |
| 26 | 1. Loads the committed HEAD snapshot and builds the production call graph |
| 27 | (AST-based, same approach as ``blast-risk`` and ``semantic-test-coverage``). |
| 28 | 2. Inverts the graph: for each symbol, BFS through the reverse call graph to |
| 29 | find every symbol that transitively depends on it. |
| 30 | 3. ``gravity_pct = transitive_dependents / total_production_symbols × 100``. |
| 31 | |
| 32 | Test files are excluded from the dependency graph by default (a function being |
| 33 | called by 50 test functions does not mean it has production structural weight). |
| 34 | Pass ``--include-tests`` to count test callers as well. |
| 35 | |
| 36 | Usage:: |
| 37 | |
| 38 | muse code gravity |
| 39 | muse code gravity --top 20 |
| 40 | muse code gravity --min-gravity 50 |
| 41 | muse code gravity --kind function |
| 42 | muse code gravity --file core/store.py |
| 43 | muse code gravity --sort depth |
| 44 | muse code gravity --depth 3 |
| 45 | muse code gravity --explain "muse/core/object_store.py::read_object" |
| 46 | muse code gravity --include-tests |
| 47 | muse code gravity --json |
| 48 | |
| 49 | Default output:: |
| 50 | |
| 51 | Symbol gravity — HEAD (378 production symbols) |
| 52 | Structural weight: fraction of production codebase that transitively depends on each symbol. |
| 53 | |
| 54 | # ADDRESS GRAVITY DIRECT DEPTH |
| 55 | 1 muse/core/object_store.py::read_object 94.2% 28 8 |
| 56 | 2 muse/core/store.py::resolve_commit_ref 87.1% 47 6 |
| 57 | 3 muse/core/errors.py::ExitCode 75.4% 31 5 |
| 58 | |
| 59 | ⚠️ High-gravity symbols are structural pillars. Treat their contracts as APIs. |
| 60 | |
| 61 | ``--explain`` output:: |
| 62 | |
| 63 | Gravity breakdown — muse/core/object_store.py::read_object |
| 64 | |
| 65 | Kind: function |
| 66 | Total gravity: 356 / 378 (94.2%) |
| 67 | Direct callers: 28 |
| 68 | Max depth: 8 |
| 69 | |
| 70 | Depth distribution: |
| 71 | depth 1 (direct): 28 callers ████████████████████ |
| 72 | depth 2: 89 callers ████████████████████████████████████████ |
| 73 | depth 3: 127 callers ████████████████████████████████████████████████████ |
| 74 | depth 4: 82 callers ██████████████████████████████████ |
| 75 | depth 5+: 30 callers ████████████ |
| 76 | |
| 77 | Deepest callers (depth 8): |
| 78 | muse/api/routes/wire.py::push_handler |
| 79 | muse/api/routes/wire.py::pull_handler |
| 80 | muse/cli/commands/log.py::run |
| 81 | |
| 82 | A breaking change here propagates through 8 layers of the codebase. |
| 83 | |
| 84 | JSON output (``--json``):: |
| 85 | |
| 86 | { |
| 87 | "ref": "HEAD", |
| 88 | "snapshot_id": "a3f2c9e1ab2c", |
| 89 | "total_production_symbols": 378, |
| 90 | "max_depth": 0, |
| 91 | "include_tests": false, |
| 92 | "filters": { "kind": null, "file": null, "min_gravity": 0.0, "top": 20 }, |
| 93 | "symbols": [ |
| 94 | { |
| 95 | "address": "muse/core/object_store.py::read_object", |
| 96 | "name": "read_object", |
| 97 | "kind": "function", |
| 98 | "file": "muse/core/object_store.py", |
| 99 | "gravity_pct": 94.2, |
| 100 | "direct_dependents": 28, |
| 101 | "transitive_dependents": 356, |
| 102 | "max_depth": 8, |
| 103 | "depth_distribution": { "1": 28, "2": 89, "3": 127, "4": 82, "5": 30 } |
| 104 | } |
| 105 | ] |
| 106 | } |
| 107 | |
| 108 | Security note |
| 109 | ------------- |
| 110 | Symbol addresses are validated to contain ``::`` before any store access. |
| 111 | File paths from the manifest are not passed to the shell — all access goes |
| 112 | through the object store's content-addressed lookup. |
| 113 | """ |
| 114 | |
| 115 | import argparse |
| 116 | import ast |
| 117 | import json |
| 118 | import logging |
| 119 | import pathlib |
| 120 | import sys |
| 121 | from typing import Literal, TypedDict |
| 122 | |
| 123 | from muse.core.envelope import EnvelopeJson, make_envelope |
| 124 | from muse.core.errors import ExitCode |
| 125 | from muse.core.repo import require_repo |
| 126 | from muse.core.timing import start_timer |
| 127 | from muse.core.types import Manifest |
| 128 | from muse.core.refs import read_current_branch |
| 129 | from muse.core.commits import resolve_commit_ref |
| 130 | from muse.core.snapshots import get_commit_snapshot_manifest |
| 131 | from muse.plugins.code._callgraph import ( |
| 132 | ForwardGraph, |
| 133 | ReverseGraph, |
| 134 | call_name, |
| 135 | find_func_node, |
| 136 | transitive_callers, |
| 137 | ) |
| 138 | from muse.plugins.code._query import is_test_file, symbols_for_snapshot |
| 139 | from muse.plugins.code.ast_parser import SymbolRecord, parse_symbols |
| 140 | from muse.core.validation import clamp_int, MAX_AST_BYTES, sanitize_display |
| 141 | |
| 142 | type _BlobMap = dict[str, bytes] |
| 143 | type _SymbolIndex = dict[str, SymbolRecord] |
| 144 | type _CounterMap = dict[str, int] |
| 145 | |
| 146 | logger = logging.getLogger(__name__) |
| 147 | |
| 148 | # ── Constants ────────────────────────────────────────────────────────────────── |
| 149 | |
| 150 | _DEFAULT_TOP = 20 |
| 151 | _DEFAULT_DEPTH = 0 # 0 = unlimited |
| 152 | _PY_SUFFIXES: frozenset[str] = frozenset({".py", ".pyi"}) |
| 153 | |
| 154 | _IMPORT_KIND = "import" |
| 155 | _TRACKED_KINDS: frozenset[str] = frozenset( |
| 156 | {"function", "async_function", "method", "async_method", "class"} |
| 157 | ) |
| 158 | |
| 159 | SortKey = Literal["gravity", "direct", "depth"] |
| 160 | _SORT_CHOICES: tuple[SortKey, ...] = ("gravity", "direct", "depth") |
| 161 | |
| 162 | _BAR_MAX_WIDTH = 40 |
| 163 | |
| 164 | # ── TypedDicts ───────────────────────────────────────────────────────────────── |
| 165 | |
| 166 | class _SymbolGravity(TypedDict): |
| 167 | """Gravity record for a single production symbol. |
| 168 | |
| 169 | Nested inside the ``symbols`` list of :class:`_JsonOut`. |
| 170 | |
| 171 | Fields |
| 172 | ------ |
| 173 | address Full symbol address (``file.py::ClassName.method``). |
| 174 | name Short symbol name (last component of address). |
| 175 | kind Symbol kind: ``"function"``, ``"class"``, ``"method"``, etc. |
| 176 | file Source file path relative to the repo root. |
| 177 | gravity_pct Percentage of production symbols that transitively depend |
| 178 | on this one (0.0–100.0). Higher = more dangerous to change. |
| 179 | direct_dependents Number of symbols that directly call or import this one. |
| 180 | transitive_dependents Total count of symbols reachable via the reverse call graph. |
| 181 | max_depth Maximum BFS depth reached in the reverse dependency traversal. |
| 182 | depth_distribution Count of dependent symbols per BFS depth layer |
| 183 | (``{"1": N, "2": M, …}``). |
| 184 | """ |
| 185 | |
| 186 | address: str |
| 187 | name: str |
| 188 | kind: str |
| 189 | file: str |
| 190 | gravity_pct: float |
| 191 | direct_dependents: int |
| 192 | transitive_dependents: int |
| 193 | max_depth: int |
| 194 | depth_distribution: _CounterMap |
| 195 | |
| 196 | class _FilterSpec(TypedDict): |
| 197 | """Active filter values echoed back in gravity JSON output. |
| 198 | |
| 199 | Fields |
| 200 | ------ |
| 201 | kind Symbol kind filter applied, or ``None`` for all kinds. |
| 202 | file File path filter applied, or ``None`` for all files. |
| 203 | min_gravity Minimum ``gravity_pct`` threshold — symbols below this were excluded. |
| 204 | top Maximum number of symbols returned (leaderboard cap). |
| 205 | """ |
| 206 | |
| 207 | kind: str | None |
| 208 | file: str | None |
| 209 | min_gravity: float |
| 210 | top: int |
| 211 | |
| 212 | class _JsonOut(EnvelopeJson): |
| 213 | """JSON output for ``muse code gravity --json`` (leaderboard mode). |
| 214 | |
| 215 | Inherits the 6 standard envelope fields from :class:`~muse.core.envelope.EnvelopeJson`. |
| 216 | |
| 217 | Fields |
| 218 | ------ |
| 219 | ref Branch ref at which analysis was performed. |
| 220 | snapshot_id Short commit ID for the HEAD snapshot. |
| 221 | total_production_symbols Total production symbols in the snapshot. |
| 222 | max_depth BFS depth cap (0 = unlimited). |
| 223 | include_tests Whether test-file callers were counted. |
| 224 | filters Active filter values (kind, file, min_gravity, top). |
| 225 | symbols Ranked list of symbol gravity records. |
| 226 | """ |
| 227 | |
| 228 | ref: str |
| 229 | snapshot_id: str |
| 230 | total_production_symbols: int |
| 231 | max_depth: int |
| 232 | include_tests: bool |
| 233 | filters: _FilterSpec |
| 234 | symbols: list[_SymbolGravity] |
| 235 | |
| 236 | class _GravityExplainJson(EnvelopeJson): |
| 237 | """JSON output for ``muse code gravity --explain ADDR --json``. |
| 238 | |
| 239 | Inherits the 6 standard envelope fields from :class:`~muse.core.envelope.EnvelopeJson`. |
| 240 | |
| 241 | Fields |
| 242 | ------ |
| 243 | address Symbol address being explained. |
| 244 | name Symbol short name. |
| 245 | kind Symbol kind (function, class, method, …). |
| 246 | file Source file path. |
| 247 | gravity_pct Fraction of production codebase that transitively |
| 248 | depends on this symbol, expressed as a percentage. |
| 249 | direct_dependents Number of symbols that directly call this one. |
| 250 | transitive_dependents Total transitive dependent count. |
| 251 | max_depth Maximum depth reached in the reverse BFS. |
| 252 | depth_distribution Count of dependents per BFS depth layer. |
| 253 | """ |
| 254 | |
| 255 | address: str |
| 256 | name: str |
| 257 | kind: str |
| 258 | file: str |
| 259 | gravity_pct: float |
| 260 | direct_dependents: int |
| 261 | transitive_dependents: int |
| 262 | max_depth: int |
| 263 | depth_distribution: _CounterMap |
| 264 | |
| 265 | # ── Helpers ──────────────────────────────────────────────────────────────────── |
| 266 | |
| 267 | def _load_py_blobs( |
| 268 | root: pathlib.Path, |
| 269 | manifest: Manifest, |
| 270 | include_tests: bool, |
| 271 | ) -> _BlobMap: |
| 272 | """Load Python file blobs from the object store into memory. |
| 273 | |
| 274 | Args: |
| 275 | root: Repository root. |
| 276 | manifest: Snapshot manifest. |
| 277 | include_tests: When False (default), test files are excluded. |
| 278 | |
| 279 | Returns: |
| 280 | Mapping ``{file_path: raw_bytes}`` for qualifying Python files. |
| 281 | """ |
| 282 | from muse.core.object_store import read_object |
| 283 | |
| 284 | blobs: _BlobMap = {} |
| 285 | for file_path, obj_id in manifest.items(): |
| 286 | if pathlib.PurePosixPath(file_path).suffix.lower() not in _PY_SUFFIXES: |
| 287 | continue |
| 288 | if not include_tests and is_test_file(file_path): |
| 289 | continue |
| 290 | raw = read_object(root, obj_id) |
| 291 | if raw is not None: |
| 292 | blobs[file_path] = raw |
| 293 | return blobs |
| 294 | |
| 295 | def _build_forward_graph( |
| 296 | blobs: _BlobMap, |
| 297 | ) -> ForwardGraph: |
| 298 | """Build the forward call graph from pre-loaded blobs. |
| 299 | |
| 300 | Scans every Python file in *blobs*, walking each callable symbol's body |
| 301 | for ``ast.Call`` nodes. Returns ``{caller_address: frozenset[callee_bare_name]}``. |
| 302 | |
| 303 | Building from pre-loaded blobs avoids a second object-store round-trip. |
| 304 | """ |
| 305 | graph: ForwardGraph = {} |
| 306 | for file_path, raw in blobs.items(): |
| 307 | try: |
| 308 | if len(raw) > MAX_AST_BYTES: |
| 309 | return {} |
| 310 | tree = ast.parse(raw) |
| 311 | except SyntaxError: |
| 312 | continue |
| 313 | sym_tree = parse_symbols(raw, file_path) |
| 314 | for addr, rec in sym_tree.items(): |
| 315 | if rec["kind"] not in {"function", "async_function", "method", "async_method"}: |
| 316 | continue |
| 317 | func_node = find_func_node(tree.body, rec["qualified_name"].split(".")) |
| 318 | if func_node is None: |
| 319 | continue |
| 320 | callees: set[str] = set() |
| 321 | for node in ast.walk(func_node): |
| 322 | if isinstance(node, ast.Call): |
| 323 | n = call_name(node.func) |
| 324 | if n: |
| 325 | callees.add(n) |
| 326 | graph[addr] = frozenset(callees) |
| 327 | return graph |
| 328 | |
| 329 | def _invert_graph(forward: ForwardGraph) -> ReverseGraph: |
| 330 | """Invert the forward call graph to get ``{callee_bare_name: [caller_addr, ...]}``. |
| 331 | |
| 332 | Produces a sorted list of callers for each callee name so output is |
| 333 | deterministic across runs. |
| 334 | """ |
| 335 | reverse: ReverseGraph = {} |
| 336 | for caller_addr, callee_names in forward.items(): |
| 337 | for name in callee_names: |
| 338 | reverse.setdefault(name, []).append(caller_addr) |
| 339 | for name in reverse: |
| 340 | reverse[name].sort() |
| 341 | return reverse |
| 342 | |
| 343 | def _gravity_bar(count: int, max_count: int, width: int = _BAR_MAX_WIDTH) -> str: |
| 344 | """Return a unicode block bar proportional to *count* / *max_count*.""" |
| 345 | if max_count <= 0: |
| 346 | return "" |
| 347 | filled = round(count / max_count * width) |
| 348 | return "█" * filled |
| 349 | |
| 350 | # ── Core algorithm ───────────────────────────────────────────────────────────── |
| 351 | |
| 352 | def _compute_gravity_for_symbol( |
| 353 | bare_name: str, |
| 354 | reverse: ReverseGraph, |
| 355 | max_depth: int, |
| 356 | ) -> tuple[dict[int, list[str]], int]: |
| 357 | """Compute transitive dependents for a symbol with the given bare name. |
| 358 | |
| 359 | Uses the existing ``transitive_callers`` BFS from ``_callgraph``. |
| 360 | |
| 361 | Args: |
| 362 | bare_name: Bare callee name (last component of the symbol address). |
| 363 | reverse: Reverse call graph from ``_invert_graph``. |
| 364 | max_depth: BFS depth limit; ``0`` = unlimited. |
| 365 | |
| 366 | Returns: |
| 367 | ``(depth_map, max_reached_depth)`` where depth_map is |
| 368 | ``{depth: [caller_address, ...]}``. |
| 369 | """ |
| 370 | depth_map = transitive_callers(bare_name, reverse, max_depth=max_depth) |
| 371 | max_reached = max(depth_map) if depth_map else 0 |
| 372 | return depth_map, max_reached |
| 373 | |
| 374 | def _build_gravity_records( |
| 375 | prod_symbols: _SymbolIndex, |
| 376 | reverse: ReverseGraph, |
| 377 | max_depth: int, |
| 378 | file_filter: str | None, |
| 379 | kind_filter: str | None, |
| 380 | min_gravity: float, |
| 381 | top: int, |
| 382 | sort_key: SortKey, |
| 383 | ) -> tuple[list[_SymbolGravity], int]: |
| 384 | """Compute gravity for every qualifying production symbol. |
| 385 | |
| 386 | Args: |
| 387 | prod_symbols: Flat ``{address: SymbolRecord}`` for all production symbols. |
| 388 | reverse: Reverse call graph. |
| 389 | max_depth: BFS depth cap (``0`` = unlimited). |
| 390 | file_filter: Optional path suffix filter. |
| 391 | kind_filter: Optional symbol kind filter. |
| 392 | min_gravity: Minimum gravity percentage threshold. |
| 393 | top: Maximum number of records to return (``0`` = unlimited). |
| 394 | sort_key: Sort dimension. |
| 395 | |
| 396 | Returns: |
| 397 | ``(records, total_prod_symbols)`` — records are sorted and truncated |
| 398 | to *top*. ``total_prod_symbols`` is the unfiltered production symbol |
| 399 | count used as the denominator for gravity %. |
| 400 | """ |
| 401 | total = len(prod_symbols) |
| 402 | if total == 0: |
| 403 | return [], 0 |
| 404 | |
| 405 | records: list[_SymbolGravity] = [] |
| 406 | |
| 407 | for addr, rec in prod_symbols.items(): |
| 408 | kind = rec["kind"] |
| 409 | if kind == _IMPORT_KIND or kind not in _TRACKED_KINDS: |
| 410 | continue |
| 411 | file_path = addr.split("::")[0] |
| 412 | if file_filter and file_filter not in file_path: |
| 413 | continue |
| 414 | if kind_filter and kind != kind_filter: |
| 415 | continue |
| 416 | |
| 417 | bare_name = rec["name"] |
| 418 | depth_map, max_reached = _compute_gravity_for_symbol(bare_name, reverse, max_depth) |
| 419 | |
| 420 | direct = len(depth_map.get(1, [])) |
| 421 | total_trans = sum(len(v) for v in depth_map.values()) |
| 422 | denom = max(1, total - 1) # exclude self |
| 423 | pct = round(total_trans / denom * 100, 1) |
| 424 | |
| 425 | if pct < min_gravity: |
| 426 | continue |
| 427 | |
| 428 | # Flatten depth distribution to str-keyed dict for JSON. |
| 429 | dist: _CounterMap = {str(d): len(addrs) for d, addrs in depth_map.items()} |
| 430 | |
| 431 | records.append( |
| 432 | _SymbolGravity( |
| 433 | address=addr, |
| 434 | name=bare_name, |
| 435 | kind=kind, |
| 436 | file=file_path, |
| 437 | gravity_pct=pct, |
| 438 | direct_dependents=direct, |
| 439 | transitive_dependents=total_trans, |
| 440 | max_depth=max_reached, |
| 441 | depth_distribution=dist, |
| 442 | ) |
| 443 | ) |
| 444 | |
| 445 | # Sort. |
| 446 | if sort_key == "direct": |
| 447 | records.sort(key=lambda r: (r["direct_dependents"], r["gravity_pct"]), reverse=True) |
| 448 | elif sort_key == "depth": |
| 449 | records.sort(key=lambda r: (r["max_depth"], r["gravity_pct"]), reverse=True) |
| 450 | else: # gravity (default) |
| 451 | records.sort(key=lambda r: (r["gravity_pct"], r["transitive_dependents"]), reverse=True) |
| 452 | |
| 453 | if top > 0: |
| 454 | records = records[:top] |
| 455 | |
| 456 | return records, total |
| 457 | |
| 458 | def _find_deepest_callers( |
| 459 | bare_name: str, |
| 460 | reverse: ReverseGraph, |
| 461 | max_depth: int, |
| 462 | show_count: int = 5, |
| 463 | ) -> tuple[list[str], int]: |
| 464 | """Return the addresses of the deepest callers (longest dependency chains). |
| 465 | |
| 466 | Args: |
| 467 | bare_name: Target symbol's bare name. |
| 468 | reverse: Reverse call graph. |
| 469 | max_depth: BFS depth cap; 0 = unlimited. |
| 470 | show_count: Maximum number of deepest-caller addresses to return. |
| 471 | |
| 472 | Returns: |
| 473 | ``(addresses, depth)`` — the deepest addresses found, and the depth at |
| 474 | which they were found. |
| 475 | """ |
| 476 | depth_map, max_reached = _compute_gravity_for_symbol(bare_name, reverse, max_depth) |
| 477 | if not depth_map: |
| 478 | return [], 0 |
| 479 | deepest = depth_map[max_reached] |
| 480 | return deepest[:show_count], max_reached |
| 481 | |
| 482 | # ── Formatters ───────────────────────────────────────────────────────────────── |
| 483 | |
| 484 | def _print_leaderboard( |
| 485 | records: list[_SymbolGravity], |
| 486 | total_prod: int, |
| 487 | max_depth_cap: int, |
| 488 | sort_key: SortKey, |
| 489 | include_tests: bool, |
| 490 | ) -> None: |
| 491 | """Print the human-readable gravity leaderboard.""" |
| 492 | scope = "all symbols" if include_tests else "production symbols" |
| 493 | cap_str = f" · depth cap: {max_depth_cap}" if max_depth_cap > 0 else "" |
| 494 | print(f"\n Symbol gravity — HEAD ({total_prod} {scope}{cap_str})") |
| 495 | print( |
| 496 | f" Structural weight: fraction of {scope} that transitively depends" |
| 497 | " on each symbol.\n" |
| 498 | ) |
| 499 | |
| 500 | if not records: |
| 501 | print(" No symbols match the current filters.\n") |
| 502 | return |
| 503 | |
| 504 | # Column widths. |
| 505 | addr_w = min(60, max(len(r["address"]) for r in records) + 2) |
| 506 | print( |
| 507 | f" {'#':>4} {'ADDRESS':<{addr_w}} {'GRAVITY':>8} {'DIRECT':>7} {'DEPTH':>6}" |
| 508 | ) |
| 509 | print(f" {'─' * 4} {'─' * addr_w} {'─' * 8} {'─' * 7} {'─' * 6}") |
| 510 | |
| 511 | for i, rec in enumerate(records, 1): |
| 512 | addr = rec["address"] |
| 513 | if len(addr) > addr_w: |
| 514 | addr = f"…{addr[-(addr_w - 1):]}" |
| 515 | pct_str = f"{rec['gravity_pct']:5.1f}%" |
| 516 | print( |
| 517 | f" {i:>4} {addr:<{addr_w}} {pct_str:>8}" |
| 518 | f" {rec['direct_dependents']:>7}" |
| 519 | f" {rec['max_depth']:>6}" |
| 520 | ) |
| 521 | |
| 522 | print() |
| 523 | if records: |
| 524 | top = records[0] |
| 525 | print( |
| 526 | f" ⚠️ {top['name']} carries {top['gravity_pct']:.1f}% structural weight" |
| 527 | " — treat its contract as an API." |
| 528 | ) |
| 529 | print() |
| 530 | |
| 531 | def _print_explain( |
| 532 | rec: _SymbolGravity, |
| 533 | reverse: ReverseGraph, |
| 534 | max_depth: int, |
| 535 | total_prod: int, |
| 536 | ) -> None: |
| 537 | """Print the detailed gravity breakdown for a single symbol.""" |
| 538 | print(f"\n Gravity breakdown — {sanitize_display(rec['address'])}\n") |
| 539 | print(f" Kind: {sanitize_display(rec['kind'])}") |
| 540 | print(f" File: {sanitize_display(rec['file'])}") |
| 541 | print( |
| 542 | f" Total gravity: {rec['transitive_dependents']} / {total_prod}" |
| 543 | f" ({rec['gravity_pct']:.1f}%)" |
| 544 | ) |
| 545 | print(f" Direct callers: {rec['direct_dependents']}") |
| 546 | print(f" Max depth: {rec['max_depth']}") |
| 547 | print() |
| 548 | |
| 549 | if rec["depth_distribution"]: |
| 550 | # Re-fetch depth_map for the distribution bars. |
| 551 | depth_map, _ = _compute_gravity_for_symbol(rec["name"], reverse, max_depth) |
| 552 | max_at_level = max(len(v) for v in depth_map.values()) if depth_map else 1 |
| 553 | |
| 554 | print(" Depth distribution:") |
| 555 | for depth in sorted(depth_map): |
| 556 | callers = depth_map[depth] |
| 557 | count = len(callers) |
| 558 | bar = _gravity_bar(count, max_at_level, width=30) |
| 559 | label = f"depth {depth} (direct):" if depth == 1 else f"depth {depth}: " |
| 560 | print(f" {label} {count:>5} callers {bar}") |
| 561 | print() |
| 562 | |
| 563 | # Deepest callers. |
| 564 | deepest_addrs, deepest_depth = _find_deepest_callers(rec["name"], reverse, max_depth) |
| 565 | if deepest_addrs: |
| 566 | print(f" Deepest callers (depth {deepest_depth}):") |
| 567 | for addr in deepest_addrs: |
| 568 | print(f" {sanitize_display(addr)}") |
| 569 | print() |
| 570 | print( |
| 571 | f" A breaking change here propagates through" |
| 572 | f" {deepest_depth} layers of the codebase." |
| 573 | ) |
| 574 | print() |
| 575 | |
| 576 | # ── Entry point ──────────────────────────────────────────────────────────────── |
| 577 | |
| 578 | def run(args: argparse.Namespace) -> None: |
| 579 | """Compute structural gravity — transitive downstream weight — for every symbol. |
| 580 | |
| 581 | Builds the AST call graph from HEAD, inverts it, and computes for each |
| 582 | production symbol what fraction of the entire codebase transitively depends |
| 583 | on it. High-gravity symbols are structural pillars; changing them has broad |
| 584 | blast radius. Use ``--explain ADDR`` for a deep-dive into one symbol. |
| 585 | |
| 586 | Agent quickstart |
| 587 | ---------------- |
| 588 | :: |
| 589 | |
| 590 | muse code gravity --json |
| 591 | muse code gravity --top 10 --json |
| 592 | muse code gravity --explain "src/billing.py::compute_total" --json |
| 593 | muse code gravity --kind function --min-gravity 0.1 --json |
| 594 | |
| 595 | JSON fields (leaderboard) |
| 596 | ------------------------- |
| 597 | ref Commit ref analysed (``"HEAD"``). |
| 598 | total_production_symbols Total non-test symbols in scope. |
| 599 | symbols Ranked list: ``address``, ``gravity_pct``, |
| 600 | ``transitive_callers``, ``kind``, ``file``. |
| 601 | |
| 602 | JSON fields (``--explain`` mode) |
| 603 | --------------------------------- |
| 604 | address Symbol explained. |
| 605 | gravity_pct Fraction of codebase depending on this symbol. |
| 606 | depth_distribution Map of depth → caller count. |
| 607 | deepest_callers Farthest callers by hop distance. |
| 608 | |
| 609 | Exit codes |
| 610 | ---------- |
| 611 | 0 Analysis complete. |
| 612 | 1 Symbol not found or invalid arguments. |
| 613 | 2 Not inside a Muse repository. |
| 614 | """ |
| 615 | elapsed = start_timer() |
| 616 | root = require_repo() |
| 617 | |
| 618 | # ── Argument validation ──────────────────────────────────────────────────── |
| 619 | top: int = clamp_int(args.top, 0, 10000, 'top') |
| 620 | if top < 0: |
| 621 | print("❌ --top must be >= 0 (0 = unlimited).", file=sys.stderr) |
| 622 | raise SystemExit(ExitCode.USER_ERROR) |
| 623 | |
| 624 | max_depth: int = clamp_int(args.depth, 0, 50, 'depth') |
| 625 | if max_depth < 0: |
| 626 | print("❌ --depth must be >= 0 (0 = unlimited).", file=sys.stderr) |
| 627 | raise SystemExit(ExitCode.USER_ERROR) |
| 628 | |
| 629 | min_gravity: float = args.min_gravity |
| 630 | if not (0.0 <= min_gravity <= 100.0): |
| 631 | print("❌ --min-gravity must be between 0.0 and 100.0.", file=sys.stderr) |
| 632 | raise SystemExit(ExitCode.USER_ERROR) |
| 633 | |
| 634 | sort_key: SortKey = args.sort |
| 635 | kind_filter: str | None = args.kind or None |
| 636 | file_filter: str | None = args.file or None |
| 637 | include_tests: bool = args.include_tests |
| 638 | explain_addr: str | None = args.explain or None |
| 639 | |
| 640 | if explain_addr is not None and "::" not in explain_addr: |
| 641 | print( |
| 642 | "❌ --explain ADDRESS must be in ``file::Symbol`` format.", |
| 643 | file=sys.stderr, |
| 644 | ) |
| 645 | raise SystemExit(ExitCode.USER_ERROR) |
| 646 | |
| 647 | # ── Resolve HEAD ─────────────────────────────────────────────────────────── |
| 648 | branch = read_current_branch(root) |
| 649 | |
| 650 | head = resolve_commit_ref(root, branch, None) |
| 651 | if head is None: |
| 652 | print("❌ HEAD commit not found — is this an empty repository?", file=sys.stderr) |
| 653 | raise SystemExit(ExitCode.USER_ERROR) |
| 654 | |
| 655 | manifest = get_commit_snapshot_manifest(root, head.commit_id) or {} |
| 656 | |
| 657 | # ── Single bulk read of all qualifying Python blobs ──────────────────────── |
| 658 | blobs = _load_py_blobs(root, manifest, include_tests) |
| 659 | |
| 660 | # ── Symbol extraction ────────────────────────────────────────────────────── |
| 661 | all_trees = symbols_for_snapshot(root, manifest) |
| 662 | |
| 663 | # Partition: test files vs. production (for symbol scope). |
| 664 | prod_trees = ( |
| 665 | all_trees |
| 666 | if include_tests |
| 667 | else {fp: t for fp, t in all_trees.items() if not is_test_file(fp)} |
| 668 | ) |
| 669 | |
| 670 | # Flat production symbol index (imports excluded). |
| 671 | prod_symbols: _SymbolIndex = {} |
| 672 | for sym_tree in prod_trees.values(): |
| 673 | for addr, rec in sym_tree.items(): |
| 674 | if rec["kind"] != _IMPORT_KIND and rec["kind"] in _TRACKED_KINDS: |
| 675 | prod_symbols[addr] = rec |
| 676 | |
| 677 | if not prod_symbols: |
| 678 | print(" No production symbols found in HEAD snapshot.\n") |
| 679 | return |
| 680 | |
| 681 | # ── Build call graphs ────────────────────────────────────────────────────── |
| 682 | forward = _build_forward_graph(blobs) |
| 683 | reverse = _invert_graph(forward) |
| 684 | |
| 685 | # ── --explain: single-symbol deep dive ──────────────────────────────────── |
| 686 | if explain_addr is not None: |
| 687 | if explain_addr not in prod_symbols: |
| 688 | print( |
| 689 | f"❌ {explain_addr!r} not found in HEAD production symbols.", |
| 690 | file=sys.stderr, |
| 691 | ) |
| 692 | print( |
| 693 | " Use ``muse code symbols`` to list available addresses.", |
| 694 | file=sys.stderr, |
| 695 | ) |
| 696 | raise SystemExit(ExitCode.USER_ERROR) |
| 697 | |
| 698 | rec_explain = prod_symbols[explain_addr] |
| 699 | bare = rec_explain["name"] |
| 700 | depth_map, max_reached = _compute_gravity_for_symbol(bare, reverse, max_depth) |
| 701 | direct = len(depth_map.get(1, [])) |
| 702 | total_trans = sum(len(v) for v in depth_map.values()) |
| 703 | denom = max(1, len(prod_symbols) - 1) |
| 704 | pct = round(total_trans / denom * 100, 1) |
| 705 | dist = {str(d): len(addrs) for d, addrs in depth_map.items()} |
| 706 | |
| 707 | single_rec = _SymbolGravity( |
| 708 | address=explain_addr, |
| 709 | name=bare, |
| 710 | kind=rec_explain["kind"], |
| 711 | file=explain_addr.split("::")[0], |
| 712 | gravity_pct=pct, |
| 713 | direct_dependents=direct, |
| 714 | transitive_dependents=total_trans, |
| 715 | max_depth=max_reached, |
| 716 | depth_distribution=dist, |
| 717 | ) |
| 718 | |
| 719 | if args.json_out: |
| 720 | print(json.dumps(_GravityExplainJson(**make_envelope(elapsed), **single_rec))) |
| 721 | return |
| 722 | |
| 723 | _print_explain(single_rec, reverse, max_depth, len(prod_symbols)) |
| 724 | return |
| 725 | |
| 726 | # ── Leaderboard ──────────────────────────────────────────────────────────── |
| 727 | records, total_prod = _build_gravity_records( |
| 728 | prod_symbols, |
| 729 | reverse, |
| 730 | max_depth, |
| 731 | file_filter, |
| 732 | kind_filter, |
| 733 | min_gravity, |
| 734 | top, |
| 735 | sort_key, |
| 736 | ) |
| 737 | |
| 738 | if args.json_out: |
| 739 | print(json.dumps(_JsonOut( |
| 740 | **make_envelope(elapsed), |
| 741 | ref="HEAD", |
| 742 | snapshot_id=head.commit_id, |
| 743 | total_production_symbols=total_prod, |
| 744 | max_depth=max_depth, |
| 745 | include_tests=include_tests, |
| 746 | filters=_FilterSpec( |
| 747 | kind=kind_filter, |
| 748 | file=file_filter, |
| 749 | min_gravity=min_gravity, |
| 750 | top=top, |
| 751 | ), |
| 752 | symbols=records, |
| 753 | ))) |
| 754 | return |
| 755 | |
| 756 | _print_leaderboard(records, total_prod, max_depth, sort_key, include_tests) |
| 757 | |
| 758 | # ── CLI registration ─────────────────────────────────────────────────────────── |
| 759 | |
| 760 | def register( |
| 761 | sub: argparse._SubParsersAction[argparse.ArgumentParser], |
| 762 | ) -> None: |
| 763 | """Register ``gravity`` under the ``code`` subcommand group. |
| 764 | |
| 765 | Args: |
| 766 | sub: The subparser action from the ``code`` command group. |
| 767 | """ |
| 768 | p = sub.add_parser( |
| 769 | "gravity", |
| 770 | help=( |
| 771 | "Structural weight — how much of the production codebase" |
| 772 | " transitively depends on each symbol." |
| 773 | ), |
| 774 | description=__doc__, |
| 775 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 776 | ) |
| 777 | p.add_argument( |
| 778 | "--explain", |
| 779 | metavar="ADDRESS", |
| 780 | help=( |
| 781 | "Deep-dive into a specific symbol's gravity:" |
| 782 | " depth distribution and deepest callers." |
| 783 | ), |
| 784 | ) |
| 785 | p.add_argument( |
| 786 | "--top", |
| 787 | type=int, |
| 788 | default=_DEFAULT_TOP, |
| 789 | metavar="N", |
| 790 | help=f"Show only the top N symbols by gravity (default: {_DEFAULT_TOP}; 0 = all).", |
| 791 | ) |
| 792 | p.add_argument( |
| 793 | "--sort", |
| 794 | metavar="KEY", |
| 795 | choices=list(_SORT_CHOICES), |
| 796 | default="gravity", |
| 797 | help=( |
| 798 | "Sort dimension: ``gravity`` (default, transitive %%)" |
| 799 | ", ``direct`` (direct callers)" |
| 800 | ", ``depth`` (max dependency chain length)." |
| 801 | ), |
| 802 | ) |
| 803 | p.add_argument( |
| 804 | "--depth", |
| 805 | type=int, |
| 806 | default=_DEFAULT_DEPTH, |
| 807 | metavar="D", |
| 808 | help=( |
| 809 | f"Maximum BFS depth for transitive analysis (default: {_DEFAULT_DEPTH} = unlimited)." |
| 810 | " Use a small value (e.g. 3) for large repos to bound runtime." |
| 811 | ), |
| 812 | ) |
| 813 | p.add_argument( |
| 814 | "--kind", |
| 815 | metavar="KIND", |
| 816 | choices=["function", "async_function", "method", "async_method", "class"], |
| 817 | help="Filter symbols by kind.", |
| 818 | ) |
| 819 | p.add_argument( |
| 820 | "--file", |
| 821 | metavar="SUFFIX", |
| 822 | help="Scope analysis to symbols in files whose path contains SUFFIX.", |
| 823 | ) |
| 824 | p.add_argument( |
| 825 | "--min-gravity", |
| 826 | type=float, |
| 827 | default=0.0, |
| 828 | metavar="PCT", |
| 829 | help="Show only symbols with gravity >= PCT%%.", |
| 830 | ) |
| 831 | p.add_argument( |
| 832 | "--include-tests", |
| 833 | action="store_true", |
| 834 | help=( |
| 835 | "Count test-file callers in the dependency graph" |
| 836 | " (default: test callers are excluded)." |
| 837 | ), |
| 838 | ) |
| 839 | p.add_argument( |
| 840 | "--json", "-j", |
| 841 | action="store_true", dest="json_out", |
| 842 | help="Emit JSON instead of human-readable text.", |
| 843 | ) |
| 844 | p.set_defaults(func=run) |
File History
7 commits
sha256:18b983389ee1b55900fcd799bfbb496552d2e3ecded9d18cefbfef188947a12e
chore: remove blob-debug test marker file
Sonnet 4.6
1 day ago
sha256:e452ad9a6ace6ccc6d875a35e06caf9da5576a970c1c36133b69a891ce5fefa8
chore: prebuild timing test
Sonnet 4.6
8 days ago
sha256:0008ab6695e3e064b3e236b24fd19e538fef6a588eb0d211622f4466d919c0b1
merge: pull staging/dev — advance to 0.2.0rc12
Sonnet 4.6
patch
10 days ago
sha256:9c33d61749fff814c5226d5386aa2af7064c2c02788594a25fdd709358132eea
fix: _PROPOSAL_PREFIX_RESOLVE_LIMIT 200 → 100 to match hub …
Sonnet 4.6
21 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
24 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
30 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
30 days ago