contract.py
python
sha256:057be43e401106b2b4d877e8028ae67d16a966562d7429707b71a5eaacd5027b
docs(mwp-master): tick all ACs green; mark MWP-7 complete w…
Sonnet 4.6
20 days ago
| 1 | """muse code contract — infer the implicit behavioral contract of a symbol. |
| 2 | |
| 3 | Every function has two contracts: |
| 4 | |
| 5 | The **explicit** contract — type annotations, docstrings, parameter names. |
| 6 | The **implicit** contract — what callers actually depend on, what tests |
| 7 | actually assert, and how the contract has evolved under real pressure. |
| 8 | |
| 9 | ``muse code contract`` infers the implicit contract by mining four sources |
| 10 | that only Muse has simultaneous access to: |
| 11 | |
| 12 | 1. **Current signature** — parameters, type annotations, defaults, and return |
| 13 | type extracted from the AST. |
| 14 | |
| 15 | 2. **Call-site patterns** — how every direct caller in the production codebase |
| 16 | actually invokes the symbol: which keyword args are used, whether the return |
| 17 | value is stored, discarded, asserted, or returned, and what argument types |
| 18 | appear in practice. |
| 19 | |
| 20 | 3. **Test assertions** — ``assert`` statements extracted from every test |
| 21 | function that exercises this symbol, forming the ground-truth specification |
| 22 | of expected behavior. |
| 23 | |
| 24 | 4. **Commit history** — stability fingerprint (MAJOR/MINOR/PATCH bumps, |
| 25 | signature changes, body rewrites) plus commit messages that contain |
| 26 | contract signals like ``breaking:``, ``invariant:``, ``deprecated:``. |
| 27 | |
| 28 | Combined, these four sources answer questions no single tool can answer alone: |
| 29 | |
| 30 | * "Is this parameter actually required, or do callers always rely on the |
| 31 | default?" |
| 32 | * "Do callers check the return value, or do they silently discard it?" |
| 33 | * "What do tests actually guarantee about the return value?" |
| 34 | * "Has this function been stable for 200 commits, or is it changing every |
| 35 | week?" |
| 36 | * "Does any commit message hint at an implicit invariant?" |
| 37 | |
| 38 | Usage:: |
| 39 | |
| 40 | muse code contract "billing.py::Invoice.compute_total" |
| 41 | muse code contract "muse/core/store.py::resolve_commit_ref" |
| 42 | muse code contract "billing.py::Invoice.compute_total" --max-commits 200 |
| 43 | muse code contract "billing.py::Invoice.compute_total" --json |
| 44 | |
| 45 | Output:: |
| 46 | |
| 47 | Contract: billing.py::Invoice.compute_total |
| 48 | |
| 49 | Signature |
| 50 | def compute_total(items, currency='USD') -> float |
| 51 | |
| 52 | Parameters |
| 53 | items positional always provided · list/comprehension at 11/12 sites |
| 54 | currency keyword omitted at 8/12 sites (default relied on) |
| 55 | observed values: 'USD' (8×) · 'EUR' (4×) |
| 56 | |
| 57 | Return value |
| 58 | → stored at 9/12 call sites (75%) |
| 59 | → discarded at 2/12 call sites ⚠️ possible misuse |
| 60 | → returned at 1/12 call sites |
| 61 | |
| 62 | Inferred preconditions |
| 63 | • items is always a list or list-comprehension |
| 64 | • currency is effectively optional — default 'USD' used by 67% of callers |
| 65 | |
| 66 | Inferred postconditions |
| 67 | • Return value is expected to be meaningful (stored 75% of the time) |
| 68 | • 2 callers silently discard the return — possible side-effect assumption |
| 69 | |
| 70 | Test assertions (3 test functions) |
| 71 | assert result == expected_total |
| 72 | assert result > 0 |
| 73 | assert isinstance(result, (int, float)) |
| 74 | |
| 75 | Commit signals |
| 76 | MINOR "feat: add optional currency parameter" (Feb 03) |
| 77 | PATCH "perf: vectorise invoice computation" (Feb 14) |
| 78 | |
| 79 | Stability |
| 80 | 97 commits · 2 signature changes · 3 body rewrites |
| 81 | 0 MAJOR · 5 MINOR · 11 PATCH |
| 82 | Est. 25% of original implementation remains |
| 83 | Assessment: evolving — contract is extending actively |
| 84 | |
| 85 | ⚠️ Warnings |
| 86 | • No return type annotation — return type inferred by callers |
| 87 | • 2 call sites discard the return value — check for misuse |
| 88 | |
| 89 | JSON output (``--json``):: |
| 90 | |
| 91 | { |
| 92 | "address": "billing.py::Invoice.compute_total", |
| 93 | "name": "compute_total", |
| 94 | "kind": "method", |
| 95 | "signature": "def compute_total(items, currency='USD') -> float", |
| 96 | "parameters": [ |
| 97 | { "name": "items", "annotation": null, "has_default": false, |
| 98 | "default_str": null }, |
| 99 | { "name": "currency", "annotation": null, "has_default": true, |
| 100 | "default_str": "'USD'" } |
| 101 | ], |
| 102 | "return_annotation": "float", |
| 103 | "call_sites": 12, |
| 104 | "caller_files": 4, |
| 105 | "return_dispositions": { |
| 106 | "stored": 9, "discarded": 2, "returned": 1, "asserted": 0, |
| 107 | "compared": 0, "argument": 0, "unknown": 0 |
| 108 | }, |
| 109 | "arg_observations": [ |
| 110 | { "param_index": 0, "categories": {"list": 8, "name": 3, "call": 1}, |
| 111 | "none_count": 0, "always_provided": true }, |
| 112 | { "param_index": 1, "categories": {"str": 4}, "none_count": 0, |
| 113 | "always_provided": false, "omit_count": 8 } |
| 114 | ], |
| 115 | "test_assertions": [ |
| 116 | "assert result == expected_total", |
| 117 | "assert result > 0", |
| 118 | "assert isinstance(result, (int, float))" |
| 119 | ], |
| 120 | "commit_signals": [ |
| 121 | { "bump": "minor", "message": "feat: add optional currency parameter", |
| 122 | "date": "2026-02-03" } |
| 123 | ], |
| 124 | "history": { |
| 125 | "commits_analysed": 97, "truncated": false, |
| 126 | "major_bumps": 0, "minor_bumps": 5, "patch_bumps": 11, |
| 127 | "sig_changes": 2, "impl_changes": 3, "est_survival_pct": 25 |
| 128 | }, |
| 129 | "preconditions": [ "items is always a list or list-comprehension" ], |
| 130 | "postconditions": [ "return value is stored at 75%% of call sites" ], |
| 131 | "warnings": [ |
| 132 | "no return type annotation", |
| 133 | "2 call sites discard the return value" |
| 134 | ], |
| 135 | "stability": "evolving" |
| 136 | } |
| 137 | |
| 138 | Security note |
| 139 | ------------- |
| 140 | Symbol addresses are validated before any store access. Commit messages are |
| 141 | sanitised (control characters stripped) before display. All file access goes |
| 142 | through the content-addressed object store — no user-supplied paths are |
| 143 | opened directly. |
| 144 | """ |
| 145 | |
| 146 | import argparse |
| 147 | import ast |
| 148 | import datetime |
| 149 | import json |
| 150 | import logging |
| 151 | import pathlib |
| 152 | import re |
| 153 | import sys |
| 154 | from collections import Counter |
| 155 | from dataclasses import dataclass |
| 156 | from typing import Literal, TypedDict |
| 157 | |
| 158 | from muse.core.errors import ExitCode |
| 159 | from muse.core.repo import require_repo |
| 160 | from muse.core.envelope import EnvelopeJson, make_envelope |
| 161 | from muse.core.timing import start_timer |
| 162 | from muse.core.refs import read_current_branch |
| 163 | from muse.core.commits import resolve_commit_ref |
| 164 | from muse.core.snapshots import get_commit_snapshot_manifest |
| 165 | from muse.domain import DomainOp |
| 166 | from muse.plugins.code._callgraph import ( |
| 167 | ReverseGraph, |
| 168 | call_name, |
| 169 | find_func_node, |
| 170 | transitive_callers, |
| 171 | ) |
| 172 | from muse.plugins.code._query import flat_symbol_ops, symbols_for_snapshot, walk_commits_bfs |
| 173 | from muse.plugins.code.ast_parser import SymbolRecord, parse_symbols |
| 174 | from muse.core.validation import clamp_int, MAX_AST_BYTES, sanitize_display |
| 175 | |
| 176 | type _BlobMap = dict[str, bytes] |
| 177 | type _SymbolIndex = dict[str, SymbolRecord] |
| 178 | type _CounterMap = dict[str, int] |
| 179 | |
| 180 | logger = logging.getLogger(__name__) |
| 181 | |
| 182 | # ── Constants ────────────────────────────────────────────────────────────────── |
| 183 | |
| 184 | _DEFAULT_MAX_COMMITS = 2_000 |
| 185 | _DEFAULT_MAX_CALL_SITES = 200 # cap to keep analysis bounded |
| 186 | _PY_SUFFIXES: frozenset[str] = frozenset({".py", ".pyi"}) |
| 187 | |
| 188 | _TEST_PATTERNS: tuple[re.Pattern[str], ...] = ( |
| 189 | re.compile(r"(^|/)test_"), |
| 190 | re.compile(r"_test\.py$"), |
| 191 | re.compile(r"(^|/)tests/"), |
| 192 | re.compile(r"(^|/)spec/"), |
| 193 | re.compile(r"(^|/)conftest\.py$"), |
| 194 | ) |
| 195 | |
| 196 | # Keywords in commit messages that signal contract-relevant events. |
| 197 | _CONTRACT_SIGNAL_WORDS: frozenset[str] = frozenset( |
| 198 | { |
| 199 | "breaking", "break", "invariant", "contract", "guarantee", "guarantee:", |
| 200 | "deprecated", "deprecate", "must", "require", "require:", "assert", |
| 201 | "precondition", "postcondition", "never", "always", "invariants", |
| 202 | } |
| 203 | ) |
| 204 | |
| 205 | _CTRL_RE = re.compile(r"[\x00-\x09\x0b-\x1f\x7f-\x9f\x1b]") |
| 206 | |
| 207 | # Op summary keywords for change classification (same as age / narrative). |
| 208 | _IMPL_KW: frozenset[str] = frozenset({"implementation", "modified", "body", "reformatted"}) |
| 209 | _SIG_KW: frozenset[str] = frozenset({"signature"}) |
| 210 | |
| 211 | # ── TypedDicts ───────────────────────────────────────────────────────────────── |
| 212 | |
| 213 | class _ParamInfo(TypedDict): |
| 214 | name: str |
| 215 | annotation: str | None |
| 216 | has_default: bool |
| 217 | default_str: str | None |
| 218 | |
| 219 | class _ArgObs(TypedDict): |
| 220 | """Observed argument patterns at call sites for one positional slot.""" |
| 221 | |
| 222 | param_index: int |
| 223 | categories: _CounterMap # ast category → count |
| 224 | none_count: int |
| 225 | always_provided: bool |
| 226 | omit_count: int |
| 227 | |
| 228 | class _CommitSignal(TypedDict): |
| 229 | bump: str |
| 230 | message: str |
| 231 | date: str |
| 232 | |
| 233 | class _HistorySummary(TypedDict): |
| 234 | commits_analysed: int |
| 235 | truncated: bool |
| 236 | major_bumps: int |
| 237 | minor_bumps: int |
| 238 | patch_bumps: int |
| 239 | sig_changes: int |
| 240 | impl_changes: int |
| 241 | est_survival_pct: int |
| 242 | |
| 243 | class _ContractJson(EnvelopeJson): |
| 244 | address: str |
| 245 | name: str |
| 246 | kind: str |
| 247 | signature: str |
| 248 | parameters: list[_ParamInfo] |
| 249 | return_annotation: str | None |
| 250 | call_sites: int |
| 251 | caller_files: int |
| 252 | return_dispositions: _CounterMap |
| 253 | arg_observations: list[_ArgObs] |
| 254 | test_assertions: list[str] |
| 255 | commit_signals: list[_CommitSignal] |
| 256 | history: _HistorySummary |
| 257 | preconditions: list[str] |
| 258 | postconditions: list[str] |
| 259 | stability: str |
| 260 | |
| 261 | # ── Internal accumulators ────────────────────────────────────────────────────── |
| 262 | |
| 263 | @dataclass |
| 264 | class _CallSiteRecord: |
| 265 | """Data collected from a single call to the target symbol.""" |
| 266 | |
| 267 | caller_addr: str |
| 268 | is_test: bool |
| 269 | positional_count: int |
| 270 | keyword_names: list[str] |
| 271 | disposition: str # stored|returned|asserted|discarded|compared|argument|unknown |
| 272 | arg_categories: list[str] # one entry per positional arg |
| 273 | |
| 274 | # ── Helpers ──────────────────────────────────────────────────────────────────── |
| 275 | |
| 276 | def _is_test_file(fp: str) -> bool: |
| 277 | return any(p.search(fp) for p in _TEST_PATTERNS) |
| 278 | |
| 279 | def _sanitise(s: str, max_len: int = 80) -> str: |
| 280 | clean = _CTRL_RE.sub("", s).strip() |
| 281 | return f"{clean[:max_len - 1]}…" if len(clean) > max_len else clean |
| 282 | |
| 283 | def _arg_category(node: ast.expr) -> str: |
| 284 | """Classify an AST argument expression into a coarse type category.""" |
| 285 | if isinstance(node, ast.Constant): |
| 286 | if node.value is None: |
| 287 | return "None" |
| 288 | return type(node.value).__name__ # "str", "int", "float", "bool" |
| 289 | if isinstance(node, ast.List): |
| 290 | return "list" |
| 291 | if isinstance(node, ast.Dict): |
| 292 | return "dict" |
| 293 | if isinstance(node, ast.Tuple): |
| 294 | return "tuple" |
| 295 | if isinstance(node, ast.Set): |
| 296 | return "set" |
| 297 | if isinstance(node, (ast.ListComp, ast.GeneratorExp, ast.DictComp, ast.SetComp)): |
| 298 | return "comprehension" |
| 299 | if isinstance(node, ast.Call): |
| 300 | return "call" |
| 301 | if isinstance(node, ast.Name): |
| 302 | return f"var" |
| 303 | return "other" |
| 304 | |
| 305 | def _extract_direct_call(node: ast.expr | None, target_name: str) -> ast.Call | None: |
| 306 | """If *node* is a direct ast.Call to *target_name*, return it; else None.""" |
| 307 | if isinstance(node, ast.Call): |
| 308 | n = call_name(node.func) |
| 309 | if n == target_name: |
| 310 | return node |
| 311 | return None |
| 312 | |
| 313 | def _classify_disposition( |
| 314 | stmts: list[ast.stmt], |
| 315 | target_name: str, |
| 316 | ) -> list[tuple[str, list[ast.expr], list[ast.keyword]]]: |
| 317 | """Walk *stmts* and classify how each direct call to *target_name* is used. |
| 318 | |
| 319 | Returns a list of ``(disposition, positional_args, keywords)`` tuples. |
| 320 | """ |
| 321 | results: list[tuple[str, list[ast.expr], list[ast.keyword]]] = [] |
| 322 | |
| 323 | for stmt in stmts: |
| 324 | if isinstance(stmt, ast.Expr): |
| 325 | c = _extract_direct_call(stmt.value, target_name) |
| 326 | if c: |
| 327 | results.append(("discarded", list(c.args), list(c.keywords))) |
| 328 | |
| 329 | elif isinstance(stmt, (ast.Assign, ast.AnnAssign)): |
| 330 | val = stmt.value if isinstance(stmt, ast.Assign) else stmt.value |
| 331 | if val is not None: |
| 332 | c = _extract_direct_call(val, target_name) |
| 333 | if c: |
| 334 | results.append(("stored", list(c.args), list(c.keywords))) |
| 335 | |
| 336 | elif isinstance(stmt, ast.Return) and stmt.value is not None: |
| 337 | c = _extract_direct_call(stmt.value, target_name) |
| 338 | if c: |
| 339 | results.append(("returned", list(c.args), list(c.keywords))) |
| 340 | |
| 341 | elif isinstance(stmt, ast.Assert): |
| 342 | c = _extract_direct_call(stmt.test, target_name) |
| 343 | if c: |
| 344 | results.append(("asserted", list(c.args), list(c.keywords))) |
| 345 | |
| 346 | elif isinstance(stmt, ast.If): |
| 347 | c = _extract_direct_call(stmt.test, target_name) |
| 348 | if c: |
| 349 | results.append(("compared", list(c.args), list(c.keywords))) |
| 350 | results.extend(_classify_disposition(stmt.body, target_name)) |
| 351 | results.extend(_classify_disposition(stmt.orelse, target_name)) |
| 352 | |
| 353 | elif isinstance(stmt, (ast.For, ast.While)): |
| 354 | results.extend(_classify_disposition(stmt.body, target_name)) |
| 355 | |
| 356 | elif isinstance(stmt, ast.With): |
| 357 | results.extend(_classify_disposition(stmt.body, target_name)) |
| 358 | |
| 359 | elif isinstance(stmt, ast.Try): |
| 360 | results.extend(_classify_disposition(stmt.body, target_name)) |
| 361 | for h in stmt.handlers: |
| 362 | results.extend(_classify_disposition(h.body, target_name)) |
| 363 | results.extend(_classify_disposition(stmt.finalbody, target_name)) |
| 364 | |
| 365 | return results |
| 366 | |
| 367 | def _extract_assertions_from_test( |
| 368 | func_node: ast.FunctionDef | ast.AsyncFunctionDef, |
| 369 | target_bare_name: str, |
| 370 | ) -> list[str]: |
| 371 | """Return ``assert`` expressions from a test function that calls *target*. |
| 372 | |
| 373 | Only collects assertions from test functions that contain at least one call |
| 374 | to *target_bare_name*. Caps at 8 assertions to keep output focused. |
| 375 | """ |
| 376 | # Check if this function ever calls target. |
| 377 | has_call = any( |
| 378 | isinstance(node, ast.Call) and call_name(node.func) == target_bare_name |
| 379 | for node in ast.walk(func_node) |
| 380 | ) |
| 381 | if not has_call: |
| 382 | return [] |
| 383 | |
| 384 | assertions: list[str] = [] |
| 385 | for node in ast.walk(func_node): |
| 386 | if isinstance(node, ast.Assert): |
| 387 | try: |
| 388 | assertions.append(ast.unparse(node)) |
| 389 | except Exception: |
| 390 | pass |
| 391 | if len(assertions) >= 8: |
| 392 | break |
| 393 | return assertions |
| 394 | |
| 395 | # ── Signature extraction ─────────────────────────────────────────────────────── |
| 396 | |
| 397 | def _extract_signature( |
| 398 | raw: bytes, |
| 399 | address: str, |
| 400 | ) -> tuple[str, list[_ParamInfo], str | None]: |
| 401 | """Extract signature data for *address* from *raw* Python source. |
| 402 | |
| 403 | Returns: |
| 404 | ``(signature_str, params, return_annotation)`` |
| 405 | |
| 406 | ``signature_str`` is a compact human-readable signature line. |
| 407 | ``params`` is a list of parameter info dicts. |
| 408 | ``return_annotation`` is the return type as a string, or ``None``. |
| 409 | """ |
| 410 | if "::" not in address: |
| 411 | return ("", [], None) |
| 412 | sym_name = address.split("::", 1)[1] |
| 413 | try: |
| 414 | if len(raw) > MAX_AST_BYTES: |
| 415 | return ("", [], None) |
| 416 | tree = ast.parse(raw) |
| 417 | except SyntaxError: |
| 418 | return ("", [], None) |
| 419 | |
| 420 | func_node = find_func_node(tree.body, sym_name.split(".")) |
| 421 | if func_node is None: |
| 422 | return ("", [], None) |
| 423 | |
| 424 | args = func_node.args |
| 425 | all_args = list(args.args) |
| 426 | # Include *args, **kwargs for completeness if present. |
| 427 | num_defaults = len(args.defaults) |
| 428 | num_params = len(all_args) |
| 429 | # Defaults align to the last N params. |
| 430 | default_offset = num_params - num_defaults |
| 431 | |
| 432 | params: list[_ParamInfo] = [] |
| 433 | for i, arg in enumerate(all_args): |
| 434 | ann = ast.unparse(arg.annotation) if arg.annotation else None |
| 435 | has_default = i >= default_offset |
| 436 | default_str: str | None = None |
| 437 | if has_default: |
| 438 | d = args.defaults[i - default_offset] |
| 439 | try: |
| 440 | default_str = ast.unparse(d) |
| 441 | except Exception: |
| 442 | default_str = "..." |
| 443 | # Skip "self" and "cls" from the public contract. |
| 444 | if arg.arg in ("self", "cls"): |
| 445 | continue |
| 446 | params.append( |
| 447 | _ParamInfo( |
| 448 | name=arg.arg, |
| 449 | annotation=ann, |
| 450 | has_default=has_default, |
| 451 | default_str=default_str, |
| 452 | ) |
| 453 | ) |
| 454 | |
| 455 | ret_ann: str | None = None |
| 456 | if func_node.returns: |
| 457 | try: |
| 458 | ret_ann = ast.unparse(func_node.returns) |
| 459 | except Exception: |
| 460 | pass |
| 461 | |
| 462 | # Build compact signature string. |
| 463 | parts: list[str] = [] |
| 464 | for i, arg in enumerate(all_args): |
| 465 | if arg.arg in ("self", "cls"): |
| 466 | continue |
| 467 | ann_str = f": {ast.unparse(arg.annotation)}" if arg.annotation else "" |
| 468 | has_default = (all_args.index(arg)) >= default_offset |
| 469 | def_str = "" |
| 470 | if has_default: |
| 471 | d = args.defaults[all_args.index(arg) - default_offset] |
| 472 | try: |
| 473 | def_str = f"={ast.unparse(d)}" |
| 474 | except Exception: |
| 475 | def_str = "=..." |
| 476 | parts.append(f"{arg.arg}{ann_str}{def_str}") |
| 477 | |
| 478 | bare_name = sym_name.split(".")[-1] |
| 479 | ret_str = f" -> {ret_ann}" if ret_ann else "" |
| 480 | signature_str = f"def {bare_name}({', '.join(parts)}){ret_str}" |
| 481 | |
| 482 | return signature_str, params, ret_ann |
| 483 | |
| 484 | # ── Call-site analysis ───────────────────────────────────────────────────────── |
| 485 | |
| 486 | def _analyze_call_sites( |
| 487 | blobs: _BlobMap, |
| 488 | direct_callers: list[str], |
| 489 | target_bare_name: str, |
| 490 | ) -> list[_CallSiteRecord]: |
| 491 | """Analyse how each direct caller invokes the target symbol. |
| 492 | |
| 493 | Args: |
| 494 | blobs: Pre-loaded Python blobs. |
| 495 | direct_callers: List of caller symbol addresses (direct callers only). |
| 496 | target_bare_name: Bare name of the target symbol. |
| 497 | |
| 498 | Returns: |
| 499 | One ``_CallSiteRecord`` per call to the target found. |
| 500 | """ |
| 501 | records: list[_CallSiteRecord] = [] |
| 502 | |
| 503 | for caller_addr in direct_callers[:_DEFAULT_MAX_CALL_SITES]: |
| 504 | if "::" not in caller_addr: |
| 505 | continue |
| 506 | file_path = caller_addr.split("::")[0] |
| 507 | sym_name = caller_addr.split("::", 1)[1] |
| 508 | raw = blobs.get(file_path) |
| 509 | if raw is None: |
| 510 | continue |
| 511 | |
| 512 | try: |
| 513 | if len(raw) > MAX_AST_BYTES: |
| 514 | continue |
| 515 | tree = ast.parse(raw) |
| 516 | except SyntaxError: |
| 517 | continue |
| 518 | |
| 519 | func_node = find_func_node(tree.body, sym_name.split(".")) |
| 520 | if func_node is None: |
| 521 | continue |
| 522 | |
| 523 | is_test = _is_test_file(file_path) |
| 524 | call_items = _classify_disposition(list(func_node.body), target_bare_name) |
| 525 | |
| 526 | for disposition, pos_args, kw_args in call_items: |
| 527 | arg_cats = [_arg_category(a) for a in pos_args] |
| 528 | kw_names = [k.arg for k in kw_args if k.arg is not None] |
| 529 | records.append( |
| 530 | _CallSiteRecord( |
| 531 | caller_addr=caller_addr, |
| 532 | is_test=is_test, |
| 533 | positional_count=len(pos_args), |
| 534 | keyword_names=kw_names, |
| 535 | disposition=disposition, |
| 536 | arg_categories=arg_cats, |
| 537 | ) |
| 538 | ) |
| 539 | |
| 540 | return records |
| 541 | |
| 542 | def _extract_all_test_assertions( |
| 543 | blobs: _BlobMap, |
| 544 | direct_callers: list[str], |
| 545 | target_bare_name: str, |
| 546 | ) -> list[str]: |
| 547 | """Collect assert statements from test callers of the target. |
| 548 | |
| 549 | Deduplicates across all test functions and caps at 10 total. |
| 550 | """ |
| 551 | seen: set[str] = set() |
| 552 | assertions: list[str] = [] |
| 553 | |
| 554 | for caller_addr in direct_callers: |
| 555 | if "::" not in caller_addr: |
| 556 | continue |
| 557 | file_path = caller_addr.split("::")[0] |
| 558 | if not _is_test_file(file_path): |
| 559 | continue |
| 560 | sym_name = caller_addr.split("::", 1)[1] |
| 561 | raw = blobs.get(file_path) |
| 562 | if raw is None: |
| 563 | continue |
| 564 | |
| 565 | try: |
| 566 | if len(raw) > MAX_AST_BYTES: |
| 567 | continue |
| 568 | tree = ast.parse(raw) |
| 569 | except SyntaxError: |
| 570 | continue |
| 571 | |
| 572 | func_node = find_func_node(tree.body, sym_name.split(".")) |
| 573 | if func_node is None: |
| 574 | continue |
| 575 | |
| 576 | for assertion in _extract_assertions_from_test(func_node, target_bare_name): |
| 577 | if assertion not in seen: |
| 578 | seen.add(assertion) |
| 579 | assertions.append(assertion) |
| 580 | if len(assertions) >= 10: |
| 581 | return assertions |
| 582 | |
| 583 | return assertions |
| 584 | |
| 585 | # ── History analysis ─────────────────────────────────────────────────────────── |
| 586 | |
| 587 | def _collect_history( |
| 588 | root: pathlib.Path, |
| 589 | head_commit_id: str, |
| 590 | address: str, |
| 591 | max_commits: int, |
| 592 | ) -> tuple[_HistorySummary, list[_CommitSignal]]: |
| 593 | """Walk the commit DAG and build a history summary for *address*. |
| 594 | |
| 595 | Returns: |
| 596 | ``(history_summary, commit_signals)`` |
| 597 | """ |
| 598 | commits, truncated = walk_commits_bfs(root, head_commit_id, max_commits) |
| 599 | |
| 600 | major = minor = patch = sig = impl = 0 |
| 601 | signals: list[_CommitSignal] = [] |
| 602 | |
| 603 | for commit in commits: |
| 604 | if commit.structured_delta is None: |
| 605 | continue |
| 606 | ops: list[DomainOp] = commit.structured_delta["ops"] |
| 607 | |
| 608 | touched = False |
| 609 | for op in flat_symbol_ops(ops): |
| 610 | if op["address"] != address: |
| 611 | continue |
| 612 | touched = True |
| 613 | kind = op.get("op", "") |
| 614 | if kind == "replace": |
| 615 | new_sum = str(op.get("new_summary") or "").lower() |
| 616 | old_sum = str(op.get("old_summary") or "").lower() |
| 617 | if any(kw in new_sum for kw in _SIG_KW): |
| 618 | sig += 1 |
| 619 | elif any(kw in new_sum for kw in _IMPL_KW) or any( |
| 620 | kw in old_sum for kw in _IMPL_KW |
| 621 | ): |
| 622 | impl += 1 |
| 623 | else: |
| 624 | impl += 1 # conservative |
| 625 | |
| 626 | if not touched: |
| 627 | continue |
| 628 | |
| 629 | bump = (commit.sem_ver_bump if commit.sem_ver_bump else "none").lower() |
| 630 | if bump == "major": |
| 631 | major += 1 |
| 632 | elif bump == "minor": |
| 633 | minor += 1 |
| 634 | elif bump == "patch": |
| 635 | patch += 1 |
| 636 | |
| 637 | # Extract contract signals from commit message. |
| 638 | msg = _sanitise(commit.message, max_len=100) |
| 639 | msg_lower = msg.lower() |
| 640 | if any(w in msg_lower for w in _CONTRACT_SIGNAL_WORDS): |
| 641 | signals.append( |
| 642 | _CommitSignal( |
| 643 | bump=bump, |
| 644 | message=msg, |
| 645 | date=commit.committed_at.strftime("%Y-%m-%d"), |
| 646 | ) |
| 647 | ) |
| 648 | |
| 649 | est_survival = round(100 / (impl + 1)) |
| 650 | |
| 651 | return ( |
| 652 | _HistorySummary( |
| 653 | commits_analysed=len(commits), |
| 654 | truncated=truncated, |
| 655 | major_bumps=major, |
| 656 | minor_bumps=minor, |
| 657 | patch_bumps=patch, |
| 658 | sig_changes=sig, |
| 659 | impl_changes=impl, |
| 660 | est_survival_pct=est_survival, |
| 661 | ), |
| 662 | signals[:20], # cap signals to keep output readable |
| 663 | ) |
| 664 | |
| 665 | # ── Contract inference ───────────────────────────────────────────────────────── |
| 666 | |
| 667 | def _infer_contract( |
| 668 | params: list[_ParamInfo], |
| 669 | call_records: list[_CallSiteRecord], |
| 670 | history: _HistorySummary, |
| 671 | ) -> tuple[list[str], list[str], list[str], str]: |
| 672 | """Infer preconditions, postconditions, warnings, and stability assessment. |
| 673 | |
| 674 | Returns: |
| 675 | ``(preconditions, postconditions, warnings, stability)`` |
| 676 | """ |
| 677 | pre: list[str] = [] |
| 678 | post: list[str] = [] |
| 679 | warn: list[str] = [] |
| 680 | |
| 681 | n = len(call_records) |
| 682 | |
| 683 | # ── Return value patterns ────────────────────────────────────────────────── |
| 684 | disp_counts: Counter[str] = Counter(r.disposition for r in call_records) |
| 685 | discarded = disp_counts.get("discarded", 0) |
| 686 | stored = disp_counts.get("stored", 0) |
| 687 | returned = disp_counts.get("returned", 0) |
| 688 | asserted = disp_counts.get("asserted", 0) |
| 689 | compared = disp_counts.get("compared", 0) |
| 690 | |
| 691 | if n > 0: |
| 692 | use_count = n - discarded |
| 693 | if use_count > 0: |
| 694 | post.append( |
| 695 | f"Return value is used at {use_count}/{n} call sites" |
| 696 | f" ({round(use_count / n * 100)}%)" |
| 697 | ) |
| 698 | if stored > 0: |
| 699 | post.append( |
| 700 | f"Return value is stored at {stored}/{n} call sites" |
| 701 | f" — callers expect a meaningful result" |
| 702 | ) |
| 703 | if discarded > 0: |
| 704 | warn.append( |
| 705 | f"{discarded} call site{'s' if discarded > 1 else ''} discard" |
| 706 | f" the return value — check for misuse or side-effect assumption" |
| 707 | ) |
| 708 | |
| 709 | # ── Argument patterns ────────────────────────────────────────────────────── |
| 710 | # For each positional param slot, gather observed arg categories. |
| 711 | for slot, param in enumerate(params): |
| 712 | slot_cats: Counter[str] = Counter() |
| 713 | omit_count = 0 |
| 714 | for r in call_records: |
| 715 | if slot < len(r.arg_categories): |
| 716 | slot_cats[r.arg_categories[slot]] += 1 |
| 717 | elif param["has_default"]: |
| 718 | omit_count += 1 |
| 719 | |
| 720 | if param["has_default"] and n > 0: |
| 721 | omit_rate = omit_count / n |
| 722 | if omit_rate > 0.5: |
| 723 | pre.append( |
| 724 | f"``{param['name']}`` defaults relied upon at" |
| 725 | f" {omit_count}/{n} call sites — effectively optional" |
| 726 | ) |
| 727 | elif omit_rate == 0.0: |
| 728 | pre.append( |
| 729 | f"``{param['name']}`` always provided explicitly — callers" |
| 730 | " never rely on the default" |
| 731 | ) |
| 732 | |
| 733 | if slot_cats: |
| 734 | top_cat, top_n = slot_cats.most_common(1)[0] |
| 735 | if top_n == len(call_records) and len(call_records) > 1: |
| 736 | pre.append( |
| 737 | f"``{param['name']}`` is always a {top_cat}" |
| 738 | f" ({top_n} call sites agree)" |
| 739 | ) |
| 740 | elif "None" in slot_cats: |
| 741 | none_count = slot_cats["None"] |
| 742 | pre.append( |
| 743 | f"``{param['name']}`` is sometimes None" |
| 744 | f" ({none_count}/{n} call sites)" |
| 745 | ) |
| 746 | |
| 747 | # ── Annotation gaps ──────────────────────────────────────────────────────── |
| 748 | unannotated = [p["name"] for p in params if not p["annotation"]] |
| 749 | if unannotated: |
| 750 | warn.append( |
| 751 | f"No type annotations on: {', '.join(unannotated)}" |
| 752 | " — contract is implicit" |
| 753 | ) |
| 754 | |
| 755 | # ── Stability assessment ─────────────────────────────────────────────────── |
| 756 | if history["major_bumps"] > 0: |
| 757 | stability = "volatile" |
| 758 | elif history["sig_changes"] > 2 or history["minor_bumps"] > 3: |
| 759 | stability = "evolving" |
| 760 | elif history["impl_changes"] == 0 and history["sig_changes"] == 0: |
| 761 | stability = "dormant" |
| 762 | elif history["impl_changes"] <= 1 and history["sig_changes"] == 0: |
| 763 | stability = "stable" |
| 764 | else: |
| 765 | stability = "evolving" |
| 766 | |
| 767 | return pre, post, warn, stability |
| 768 | |
| 769 | # ── Formatters ───────────────────────────────────────────────────────────────── |
| 770 | |
| 771 | def _pct_bar(n: int, total: int, width: int = 20) -> str: |
| 772 | if total <= 0: |
| 773 | return "" |
| 774 | filled = round(n / total * width) |
| 775 | return f"{'█' * filled}{'░' * (width - filled)}" |
| 776 | |
| 777 | def _print_contract( |
| 778 | address: str, |
| 779 | name: str, |
| 780 | kind: str, |
| 781 | signature_str: str, |
| 782 | params: list[_ParamInfo], |
| 783 | ret_ann: str | None, |
| 784 | call_records: list[_CallSiteRecord], |
| 785 | test_assertions: list[str], |
| 786 | commit_signals: list[_CommitSignal], |
| 787 | history: _HistorySummary, |
| 788 | pre: list[str], |
| 789 | post: list[str], |
| 790 | warn: list[str], |
| 791 | stability: str, |
| 792 | ) -> None: |
| 793 | """Print the human-readable contract report.""" |
| 794 | caller_files = len({r.caller_addr.split("::")[0] for r in call_records}) |
| 795 | n = len(call_records) |
| 796 | |
| 797 | print(f"\n Contract: {sanitize_display(address)}\n") |
| 798 | |
| 799 | # ── Signature ────────────────────────────────────────────────────────────── |
| 800 | print(" Signature") |
| 801 | print(f" {signature_str}\n") |
| 802 | |
| 803 | # ── Parameters ──────────────────────────────────────────────────────────── |
| 804 | if params: |
| 805 | print(" Parameters") |
| 806 | for slot, param in enumerate(params): |
| 807 | ann = f": {param['annotation']}" if param["annotation"] else "" |
| 808 | default = f" = {param['default_str']}" if param["has_default"] else "" |
| 809 | print(f" {sanitize_display(param['name'])}{ann}{default}") |
| 810 | |
| 811 | # Arg usage patterns. |
| 812 | slot_cats: Counter[str] = Counter() |
| 813 | omit_count = 0 |
| 814 | for r in call_records: |
| 815 | if slot < len(r.arg_categories): |
| 816 | slot_cats[r.arg_categories[slot]] += 1 |
| 817 | elif param["has_default"]: |
| 818 | omit_count += 1 |
| 819 | |
| 820 | lines: list[str] = [] |
| 821 | if omit_count > 0 and n > 0: |
| 822 | lines.append(f"omitted at {omit_count}/{n} sites (default relied on)") |
| 823 | if slot_cats: |
| 824 | cats = ", ".join( |
| 825 | f"{cat} ({cnt}×)" for cat, cnt in slot_cats.most_common(3) |
| 826 | ) |
| 827 | lines.append(f"observed: {cats}") |
| 828 | for line in lines: |
| 829 | print(f" {line}") |
| 830 | print() |
| 831 | |
| 832 | # ── Return value ────────────────────────────────────────────────────────── |
| 833 | if n > 0: |
| 834 | print(" Return value") |
| 835 | disp: Counter[str] = Counter(r.disposition for r in call_records) |
| 836 | for label, count in disp.most_common(): |
| 837 | if count == 0: |
| 838 | continue |
| 839 | bar = _pct_bar(count, n, width=16) |
| 840 | warn_flag = " ⚠️" if label == "discarded" and count > 1 else "" |
| 841 | print(f" → {label:<12} {count:>3}/{n} [{bar}]{warn_flag}") |
| 842 | print() |
| 843 | |
| 844 | # ── Preconditions ───────────────────────────────────────────────────────── |
| 845 | if pre: |
| 846 | print(" Inferred preconditions") |
| 847 | for p in pre: |
| 848 | print(f" • {p}") |
| 849 | print() |
| 850 | |
| 851 | # ── Postconditions ──────────────────────────────────────────────────────── |
| 852 | if post: |
| 853 | print(" Inferred postconditions") |
| 854 | for p in post: |
| 855 | print(f" • {p}") |
| 856 | print() |
| 857 | |
| 858 | # ── Test assertions ─────────────────────────────────────────────────────── |
| 859 | test_sites = sum(1 for r in call_records if r.is_test) |
| 860 | if test_assertions: |
| 861 | print(f" Test assertions ({test_sites} test call site{'s' if test_sites != 1 else ''})") |
| 862 | for a in test_assertions: |
| 863 | print(f" {a}") |
| 864 | print() |
| 865 | |
| 866 | # ── Commit signals ──────────────────────────────────────────────────────── |
| 867 | if commit_signals: |
| 868 | print(" Commit signals") |
| 869 | for sig in commit_signals[:5]: |
| 870 | print(f" {sig['bump'].upper():<8} \"{sanitize_display(sig['message'][:60])}\" ({sig['date']})") |
| 871 | print() |
| 872 | |
| 873 | # ── Stability ───────────────────────────────────────────────────────────── |
| 874 | print(" Stability") |
| 875 | h = history |
| 876 | print( |
| 877 | f" {h['commits_analysed']} commits" |
| 878 | f" · {h['sig_changes']} signature change{'s' if h['sig_changes'] != 1 else ''}" |
| 879 | f" · {h['impl_changes']} body rewrite{'s' if h['impl_changes'] != 1 else ''}" |
| 880 | ) |
| 881 | print( |
| 882 | f" {h['major_bumps']} MAJOR" |
| 883 | f" · {h['minor_bumps']} MINOR" |
| 884 | f" · {h['patch_bumps']} PATCH" |
| 885 | ) |
| 886 | print(f" Est. {h['est_survival_pct']}% of original implementation remains") |
| 887 | stability_desc = { |
| 888 | "stable": "stable — contract has been consistent", |
| 889 | "evolving": "evolving — contract is extending actively", |
| 890 | "volatile": "volatile — MAJOR contract breaks in history", |
| 891 | "dormant": "dormant — rarely changed", |
| 892 | }.get(stability, stability) |
| 893 | print(f" Assessment: {stability_desc}\n") |
| 894 | |
| 895 | # ── Warnings ────────────────────────────────────────────────────────────── |
| 896 | if warn: |
| 897 | print(" ⚠️ Warnings") |
| 898 | for w in warn: |
| 899 | print(f" • {w}") |
| 900 | print() |
| 901 | |
| 902 | # ── Entry point ──────────────────────────────────────────────────────────────── |
| 903 | |
| 904 | def run(args: argparse.Namespace) -> None: |
| 905 | """Infer the behavioural contract of a symbol from its call sites and history. |
| 906 | |
| 907 | Analyses all call sites, return value dispositions, observed argument |
| 908 | categories, test assertions, and commit signals to produce a rich contract |
| 909 | profile: parameters, stability, preconditions, postconditions, and history |
| 910 | summary. |
| 911 | |
| 912 | Agent quickstart |
| 913 | ---------------- |
| 914 | :: |
| 915 | |
| 916 | muse code contract "billing.py::compute_tax" --json |
| 917 | muse code contract "billing.py::compute_tax" --max-commits 200 --json |
| 918 | |
| 919 | JSON fields |
| 920 | ----------- |
| 921 | address Symbol address analysed. |
| 922 | name Symbol name. |
| 923 | kind Symbol kind (``"function"``, ``"method"``, etc.). |
| 924 | signature Full signature string. |
| 925 | parameters List of parameter objects with inferred type hints. |
| 926 | return_annotation Return type annotation or ``null``. |
| 927 | call_sites Total number of call sites found. |
| 928 | caller_files Number of distinct caller files. |
| 929 | return_dispositions Map of disposition → count (``"stored"``, ``"returned"``, etc.). |
| 930 | arg_observations List of observed argument category distributions. |
| 931 | test_assertions List of assertion patterns found in test call sites. |
| 932 | commit_signals List of signals from commit history. |
| 933 | history Summary: ``total_commits``, ``avg_churn``, ``first_seen``, ``last_modified``. |
| 934 | preconditions Inferred precondition strings. |
| 935 | postconditions Inferred postcondition strings. |
| 936 | stability Stability label: ``"stable"``, ``"volatile"``, ``"new"``. |
| 937 | |
| 938 | Exit codes |
| 939 | ---------- |
| 940 | 0 Analysis complete. |
| 941 | 1 Invalid address format or symbol not found. |
| 942 | 2 Not inside a Muse repository. |
| 943 | """ |
| 944 | elapsed = start_timer() |
| 945 | root = require_repo() |
| 946 | |
| 947 | # ── Argument validation ──────────────────────────────────────────────────── |
| 948 | address: str = args.address.strip() |
| 949 | if "::" not in address: |
| 950 | print( |
| 951 | "❌ ADDRESS must be in ``file::Symbol`` format.", |
| 952 | file=sys.stderr, |
| 953 | ) |
| 954 | raise SystemExit(ExitCode.USER_ERROR) |
| 955 | |
| 956 | max_commits: int = clamp_int(args.max_commits, 1, 100000, 'max_commits') |
| 957 | if max_commits < 1: |
| 958 | print("❌ --max-commits must be >= 1.", file=sys.stderr) |
| 959 | raise SystemExit(ExitCode.USER_ERROR) |
| 960 | |
| 961 | # ── Resolve HEAD ─────────────────────────────────────────────────────────── |
| 962 | branch = read_current_branch(root) |
| 963 | |
| 964 | head = resolve_commit_ref(root, branch, None) |
| 965 | if head is None: |
| 966 | print("❌ HEAD commit not found — is this an empty repository?", file=sys.stderr) |
| 967 | raise SystemExit(ExitCode.USER_ERROR) |
| 968 | |
| 969 | manifest = get_commit_snapshot_manifest(root, head.commit_id) or {} |
| 970 | |
| 971 | # ── Load all Python blobs in one pass ───────────────────────────────────── |
| 972 | from muse.core.object_store import read_object |
| 973 | |
| 974 | blobs: _BlobMap = {} |
| 975 | for fp, obj_id in manifest.items(): |
| 976 | if pathlib.PurePosixPath(fp).suffix.lower() in _PY_SUFFIXES: |
| 977 | raw = read_object(root, obj_id) |
| 978 | if raw is not None: |
| 979 | blobs[fp] = raw |
| 980 | |
| 981 | # ── Verify symbol exists at HEAD ─────────────────────────────────────────── |
| 982 | all_trees = symbols_for_snapshot(root, manifest) |
| 983 | addr_to_rec: _SymbolIndex = {} |
| 984 | for sym_tree in all_trees.values(): |
| 985 | addr_to_rec.update(sym_tree) |
| 986 | |
| 987 | if address not in addr_to_rec: |
| 988 | print(f"❌ {sanitize_display(address)!r} not found in HEAD snapshot.", file=sys.stderr) |
| 989 | print(" Use ``muse code symbols`` to list available addresses.", file=sys.stderr) |
| 990 | raise SystemExit(ExitCode.USER_ERROR) |
| 991 | |
| 992 | rec = addr_to_rec[address] |
| 993 | name = rec["name"] |
| 994 | kind = rec["kind"] |
| 995 | file_path = address.split("::")[0] |
| 996 | |
| 997 | # ── Extract current signature ────────────────────────────────────────────── |
| 998 | raw_file = blobs.get(file_path) |
| 999 | params: list[_ParamInfo] |
| 1000 | if raw_file is None: |
| 1001 | signature_str, params, ret_ann = "", [], None |
| 1002 | else: |
| 1003 | signature_str, params, ret_ann = _extract_signature(raw_file, address) |
| 1004 | |
| 1005 | # ── Build forward + reverse call graph from production blobs ────────────── |
| 1006 | from muse.plugins.code._callgraph import ForwardGraph |
| 1007 | |
| 1008 | forward: ForwardGraph = {} |
| 1009 | for fp, raw in blobs.items(): |
| 1010 | if _is_test_file(fp): |
| 1011 | continue |
| 1012 | try: |
| 1013 | if len(raw) > MAX_AST_BYTES: |
| 1014 | continue |
| 1015 | tree_ast = ast.parse(raw) |
| 1016 | except SyntaxError: |
| 1017 | continue |
| 1018 | sym_tree = parse_symbols(raw, fp) |
| 1019 | for addr2, rec2 in sym_tree.items(): |
| 1020 | if rec2["kind"] not in {"function", "async_function", "method", "async_method"}: |
| 1021 | continue |
| 1022 | fn = find_func_node(tree_ast.body, rec2["qualified_name"].split(".")) |
| 1023 | if fn is None: |
| 1024 | continue |
| 1025 | callees: set[str] = set() |
| 1026 | for node in ast.walk(fn): |
| 1027 | if isinstance(node, ast.Call): |
| 1028 | n2 = call_name(node.func) |
| 1029 | if n2: |
| 1030 | callees.add(n2) |
| 1031 | forward[addr2] = frozenset(callees) |
| 1032 | |
| 1033 | # Invert → reverse graph. |
| 1034 | reverse: ReverseGraph = {} |
| 1035 | for caller_addr, callee_names in forward.items(): |
| 1036 | for cn in callee_names: |
| 1037 | reverse.setdefault(cn, []).append(caller_addr) |
| 1038 | for cn in reverse: |
| 1039 | reverse[cn].sort() |
| 1040 | |
| 1041 | # Find direct callers only (depth=1). |
| 1042 | depth_map = transitive_callers(name, reverse, max_depth=1) |
| 1043 | direct_callers: list[str] = list(depth_map.get(1, [])) |
| 1044 | |
| 1045 | # Also include test-file callers for test assertion extraction. |
| 1046 | # Build a separate test reverse graph including test files. |
| 1047 | test_forward: ForwardGraph = {} |
| 1048 | for fp, raw in blobs.items(): |
| 1049 | if not _is_test_file(fp): |
| 1050 | continue |
| 1051 | try: |
| 1052 | if len(raw) > MAX_AST_BYTES: |
| 1053 | continue |
| 1054 | tree_ast = ast.parse(raw) |
| 1055 | except SyntaxError: |
| 1056 | continue |
| 1057 | sym_tree = parse_symbols(raw, fp) |
| 1058 | for addr2, rec2 in sym_tree.items(): |
| 1059 | if rec2["kind"] not in {"function", "async_function", "method", "async_method"}: |
| 1060 | continue |
| 1061 | fn = find_func_node(tree_ast.body, rec2["qualified_name"].split(".")) |
| 1062 | if fn is None: |
| 1063 | continue |
| 1064 | callees = set() |
| 1065 | for node in ast.walk(fn): |
| 1066 | if isinstance(node, ast.Call): |
| 1067 | n2 = call_name(node.func) |
| 1068 | if n2: |
| 1069 | callees.add(n2) |
| 1070 | test_forward[addr2] = frozenset(callees) |
| 1071 | |
| 1072 | test_reverse: ReverseGraph = {} |
| 1073 | for addr2, callee_names in test_forward.items(): |
| 1074 | for cn in callee_names: |
| 1075 | test_reverse.setdefault(cn, []).append(addr2) |
| 1076 | |
| 1077 | test_callers = list(transitive_callers(name, test_reverse, max_depth=1).get(1, [])) |
| 1078 | all_direct = direct_callers + test_callers |
| 1079 | |
| 1080 | # ── Analyse call sites ───────────────────────────────────────────────────── |
| 1081 | call_records = _analyze_call_sites(blobs, all_direct, name) |
| 1082 | |
| 1083 | # ── Extract test assertions ──────────────────────────────────────────────── |
| 1084 | test_assertions = _extract_all_test_assertions(blobs, test_callers, name) |
| 1085 | |
| 1086 | # ── Commit history ──────────────────────────────────────────────────────── |
| 1087 | history, commit_signals = _collect_history(root, head.commit_id, address, max_commits) |
| 1088 | |
| 1089 | # ── Infer contract ──────────────────────────────────────────────────────── |
| 1090 | pre, post, warn, stability = _infer_contract(params, call_records, history) |
| 1091 | |
| 1092 | # ── Output ──────────────────────────────────────────────────────────────── |
| 1093 | if args.json_out: |
| 1094 | disp_counts: _CounterMap = { |
| 1095 | "stored": 0, "returned": 0, "asserted": 0, |
| 1096 | "discarded": 0, "compared": 0, "argument": 0, "unknown": 0, |
| 1097 | } |
| 1098 | for r in call_records: |
| 1099 | disp_counts[r.disposition] = disp_counts.get(r.disposition, 0) + 1 |
| 1100 | |
| 1101 | # Arg observations per param slot. |
| 1102 | arg_obs: list[_ArgObs] = [] |
| 1103 | for slot, param in enumerate(params): |
| 1104 | cats: Counter[str] = Counter() |
| 1105 | omit = 0 |
| 1106 | for r in call_records: |
| 1107 | if slot < len(r.arg_categories): |
| 1108 | cats[r.arg_categories[slot]] += 1 |
| 1109 | elif param["has_default"]: |
| 1110 | omit += 1 |
| 1111 | arg_obs.append( |
| 1112 | _ArgObs( |
| 1113 | param_index=slot, |
| 1114 | categories=dict(cats), |
| 1115 | none_count=cats.get("None", 0), |
| 1116 | always_provided=omit == 0, |
| 1117 | omit_count=omit, |
| 1118 | ) |
| 1119 | ) |
| 1120 | |
| 1121 | caller_files = len({r.caller_addr.split("::")[0] for r in call_records}) |
| 1122 | print(json.dumps(_ContractJson( |
| 1123 | **make_envelope(elapsed, warnings=warn), |
| 1124 | address=address, |
| 1125 | name=name, |
| 1126 | kind=kind, |
| 1127 | signature=signature_str, |
| 1128 | parameters=params, |
| 1129 | return_annotation=ret_ann, |
| 1130 | call_sites=len(call_records), |
| 1131 | caller_files=caller_files, |
| 1132 | return_dispositions=disp_counts, |
| 1133 | arg_observations=arg_obs, |
| 1134 | test_assertions=test_assertions, |
| 1135 | commit_signals=commit_signals, |
| 1136 | history=history, |
| 1137 | preconditions=pre, |
| 1138 | postconditions=post, |
| 1139 | stability=stability, |
| 1140 | ))) |
| 1141 | return |
| 1142 | |
| 1143 | _print_contract( |
| 1144 | address, name, kind, signature_str, params, ret_ann, |
| 1145 | call_records, test_assertions, commit_signals, |
| 1146 | history, pre, post, warn, stability, |
| 1147 | ) |
| 1148 | |
| 1149 | # ── CLI registration ─────────────────────────────────────────────────────────── |
| 1150 | |
| 1151 | def register( |
| 1152 | sub: argparse._SubParsersAction[argparse.ArgumentParser], |
| 1153 | ) -> None: |
| 1154 | """Register ``contract`` under the ``code`` subcommand group.""" |
| 1155 | p = sub.add_parser( |
| 1156 | "contract", |
| 1157 | help=( |
| 1158 | "Infer the implicit behavioral contract of a symbol from its" |
| 1159 | " call sites, test assertions, and commit history." |
| 1160 | ), |
| 1161 | description=__doc__, |
| 1162 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 1163 | ) |
| 1164 | p.add_argument( |
| 1165 | "address", |
| 1166 | metavar="ADDRESS", |
| 1167 | help=( |
| 1168 | "Full symbol address" |
| 1169 | " (e.g. billing.py::Invoice.compute_total)." |
| 1170 | ), |
| 1171 | ) |
| 1172 | p.add_argument( |
| 1173 | "--max-commits", |
| 1174 | type=int, |
| 1175 | default=_DEFAULT_MAX_COMMITS, |
| 1176 | metavar="N", |
| 1177 | help=f"Maximum commits to walk for history analysis (default: {_DEFAULT_MAX_COMMITS}).", |
| 1178 | ) |
| 1179 | p.add_argument( |
| 1180 | "--json", "-j", |
| 1181 | action="store_true", |
| 1182 | dest="json_out", |
| 1183 | help="Emit JSON instead of human-readable text.", |
| 1184 | ) |
| 1185 | p.set_defaults(func=run) |
File History
1 commit
sha256:057be43e401106b2b4d877e8028ae67d16a966562d7429707b71a5eaacd5027b
docs(mwp-master): tick all ACs green; mark MWP-7 complete w…
Sonnet 4.6
20 days ago