api_surface.py
python
| 1 | """muse api-surface — public API surface tracking. |
| 2 | |
| 3 | Shows which symbols in a snapshot are part of the public API, and how the |
| 4 | public API changed between two commits. |
| 5 | |
| 6 | A symbol is **public** when all of the following hold: |
| 7 | |
| 8 | * ``kind`` is one of: ``function``, ``async_function``, ``class``, |
| 9 | ``method``, ``async_method`` |
| 10 | * ``name`` does not start with ``_`` (Python convention for private/internal) |
| 11 | * ``kind`` is not ``import`` |
| 12 | |
| 13 | Git cannot answer "what changed in the public API between v1.0 and v1.1?" |
| 14 | without an external diffing tool. Muse answers this in O(1) against committed |
| 15 | snapshots — no checkout required, no working-tree needed. |
| 16 | |
| 17 | Usage:: |
| 18 | |
| 19 | muse api-surface |
| 20 | muse api-surface --commit HEAD~5 |
| 21 | muse api-surface --diff main |
| 22 | muse api-surface --language Python |
| 23 | muse api-surface --json |
| 24 | |
| 25 | With ``--diff REF``, shows a three-section report:: |
| 26 | |
| 27 | Public API surface — commit a1b2c3d4 vs commit e5f6a7b8 |
| 28 | ────────────────────────────────────────────────────────────── |
| 29 | |
| 30 | Added (3): |
| 31 | + src/billing.py::compute_tax function |
| 32 | + src/auth.py::refresh_token function |
| 33 | + src/models.py::User.to_json method |
| 34 | |
| 35 | Removed (1): |
| 36 | - src/billing.py::compute_total function |
| 37 | |
| 38 | Changed (2): |
| 39 | ~ src/billing.py::Invoice.pay method (signature_change) |
| 40 | ~ src/auth.py::validate_token function (impl_only) |
| 41 | |
| 42 | Flags: |
| 43 | |
| 44 | ``--commit, -c REF`` |
| 45 | Show or compare from this commit (default: HEAD). |
| 46 | |
| 47 | ``--diff REF`` |
| 48 | Compare the commit from ``--commit`` against this ref. |
| 49 | |
| 50 | ``--language LANG`` |
| 51 | Filter to symbols in files of this language. |
| 52 | |
| 53 | ``--json`` |
| 54 | Emit results as JSON with a ``schema_version`` wrapper. |
| 55 | """ |
| 56 | |
| 57 | from __future__ import annotations |
| 58 | |
| 59 | import argparse |
| 60 | import json |
| 61 | import logging |
| 62 | import pathlib |
| 63 | import sys |
| 64 | |
| 65 | from muse._version import __version__ |
| 66 | from muse.core.errors import ExitCode |
| 67 | from muse.core.repo import require_repo |
| 68 | from muse.core.store import get_commit_snapshot_manifest, read_current_branch, resolve_commit_ref |
| 69 | from muse.plugins.code._query import language_of, symbols_for_snapshot |
| 70 | from muse.plugins.code.ast_parser import SymbolRecord |
| 71 | |
| 72 | logger = logging.getLogger(__name__) |
| 73 | |
| 74 | _PUBLIC_KINDS: frozenset[str] = frozenset({ |
| 75 | "function", "async_function", "class", "method", "async_method", |
| 76 | }) |
| 77 | |
| 78 | |
| 79 | def _read_repo_id(root: pathlib.Path) -> str: |
| 80 | return str(json.loads((root / ".muse" / "repo.json").read_text())["repo_id"]) |
| 81 | |
| 82 | |
| 83 | def _read_branch(root: pathlib.Path) -> str: |
| 84 | return read_current_branch(root) |
| 85 | |
| 86 | |
| 87 | def _is_public(name: str, kind: str) -> bool: |
| 88 | return kind in _PUBLIC_KINDS and not name.split(".")[-1].startswith("_") |
| 89 | |
| 90 | |
| 91 | def _public_symbols( |
| 92 | root: pathlib.Path, |
| 93 | manifest: dict[str, str], |
| 94 | language_filter: str | None, |
| 95 | ) -> dict[str, SymbolRecord]: |
| 96 | """Return all public symbols from *manifest* as a flat address → SymbolRecord dict.""" |
| 97 | result: dict[str, SymbolRecord] = {} |
| 98 | sym_map = symbols_for_snapshot(root, manifest, language_filter=language_filter) |
| 99 | for _file, tree in sym_map.items(): |
| 100 | for address, rec in tree.items(): |
| 101 | if _is_public(rec["name"], rec["kind"]): |
| 102 | result[address] = rec |
| 103 | return result |
| 104 | |
| 105 | |
| 106 | def _classify_change(old: SymbolRecord, new: SymbolRecord) -> str: |
| 107 | """Return a human-readable classification of what changed.""" |
| 108 | if old["content_id"] == new["content_id"]: |
| 109 | return "unchanged" |
| 110 | if old["signature_id"] != new["signature_id"]: |
| 111 | if old["body_hash"] != new["body_hash"]: |
| 112 | return "signature+impl" |
| 113 | return "signature_change" |
| 114 | return "impl_only" |
| 115 | |
| 116 | |
| 117 | class _ApiEntry: |
| 118 | def __init__(self, address: str, rec: SymbolRecord, language: str) -> None: |
| 119 | self.address = address |
| 120 | self.rec = rec |
| 121 | self.language = language |
| 122 | |
| 123 | def to_dict(self) -> dict[str, str]: |
| 124 | return { |
| 125 | "address": self.address, |
| 126 | "kind": self.rec["kind"], |
| 127 | "name": self.rec["name"], |
| 128 | "qualified_name": self.rec["qualified_name"], |
| 129 | "language": self.language, |
| 130 | "content_id": self.rec["content_id"][:8], |
| 131 | "signature_id": self.rec["signature_id"][:8], |
| 132 | "body_hash": self.rec["body_hash"][:8], |
| 133 | } |
| 134 | |
| 135 | |
| 136 | def register(subparsers: "argparse._SubParsersAction[argparse.ArgumentParser]") -> None: |
| 137 | """Register the api-surface subcommand.""" |
| 138 | parser = subparsers.add_parser( |
| 139 | "api-surface", |
| 140 | help="Show the public API surface and how it changed between two commits.", |
| 141 | description=__doc__, |
| 142 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 143 | ) |
| 144 | parser.add_argument( |
| 145 | "--commit", "-c", default=None, metavar="REF", dest="ref", |
| 146 | help="Show surface at this commit (default: HEAD).", |
| 147 | ) |
| 148 | parser.add_argument( |
| 149 | "--diff", default=None, metavar="REF", dest="diff_ref", |
| 150 | help="Compare HEAD (or --commit) against this ref.", |
| 151 | ) |
| 152 | parser.add_argument( |
| 153 | "--language", "-l", default=None, metavar="LANG", dest="language", |
| 154 | help="Filter to this language (Python, Go, Rust, …).", |
| 155 | ) |
| 156 | parser.add_argument( |
| 157 | "--json", action="store_true", dest="as_json", |
| 158 | help="Emit results as JSON.", |
| 159 | ) |
| 160 | parser.set_defaults(func=run) |
| 161 | |
| 162 | |
| 163 | def run(args: argparse.Namespace) -> None: |
| 164 | """Show the public API surface and how it changed between two commits. |
| 165 | |
| 166 | A symbol is public when its kind is function/class/method (not import) and |
| 167 | its bare name does not start with ``_``. |
| 168 | |
| 169 | With ``--diff REF``, shows three sections: Added, Removed, Changed. |
| 170 | Without ``--diff``, lists all public symbols at the given commit. |
| 171 | |
| 172 | This command runs against committed snapshots only — no working-tree |
| 173 | parsing, no test execution. |
| 174 | """ |
| 175 | ref: str | None = args.ref |
| 176 | diff_ref: str | None = args.diff_ref |
| 177 | language: str | None = args.language |
| 178 | as_json: bool = args.as_json |
| 179 | |
| 180 | root = require_repo() |
| 181 | repo_id = _read_repo_id(root) |
| 182 | branch = _read_branch(root) |
| 183 | |
| 184 | commit = resolve_commit_ref(root, repo_id, branch, ref) |
| 185 | if commit is None: |
| 186 | print(f"❌ Commit '{ref or 'HEAD'}' not found.", file=sys.stderr) |
| 187 | raise SystemExit(ExitCode.USER_ERROR) |
| 188 | |
| 189 | manifest = get_commit_snapshot_manifest(root, commit.commit_id) or {} |
| 190 | current_surface = _public_symbols(root, manifest, language) |
| 191 | |
| 192 | if diff_ref is None: |
| 193 | # Just list the current surface. |
| 194 | entries = [ |
| 195 | _ApiEntry(addr, rec, language_of(addr.split("::")[0])) |
| 196 | for addr, rec in sorted(current_surface.items()) |
| 197 | ] |
| 198 | if as_json: |
| 199 | print(json.dumps( |
| 200 | { |
| 201 | "schema_version": __version__, |
| 202 | "commit": commit.commit_id[:8], |
| 203 | "language_filter": language, |
| 204 | "total": len(entries), |
| 205 | "symbols": [e.to_dict() for e in entries], |
| 206 | }, |
| 207 | indent=2, |
| 208 | )) |
| 209 | return |
| 210 | |
| 211 | print(f"\nPublic API surface — commit {commit.commit_id[:8]}") |
| 212 | if language: |
| 213 | print(f" (language: {language})") |
| 214 | print("─" * 62) |
| 215 | if not entries: |
| 216 | print(" (no public symbols found)") |
| 217 | return |
| 218 | max_addr = max(len(e.address) for e in entries) |
| 219 | for e in entries: |
| 220 | print(f" {e.address:<{max_addr}} {e.rec['kind']}") |
| 221 | print(f"\n {len(entries)} public symbol(s)") |
| 222 | return |
| 223 | |
| 224 | # Diff mode. |
| 225 | base_commit = resolve_commit_ref(root, repo_id, branch, diff_ref) |
| 226 | if base_commit is None: |
| 227 | print(f"❌ Diff ref '{diff_ref}' not found.", file=sys.stderr) |
| 228 | raise SystemExit(ExitCode.USER_ERROR) |
| 229 | |
| 230 | base_manifest = get_commit_snapshot_manifest(root, base_commit.commit_id) or {} |
| 231 | base_surface = _public_symbols(root, base_manifest, language) |
| 232 | |
| 233 | added = {a: r for a, r in current_surface.items() if a not in base_surface} |
| 234 | removed = {a: r for a, r in base_surface.items() if a not in current_surface} |
| 235 | changed: dict[str, tuple[SymbolRecord, SymbolRecord, str]] = {} |
| 236 | for addr in current_surface: |
| 237 | if addr in base_surface: |
| 238 | cls = _classify_change(base_surface[addr], current_surface[addr]) |
| 239 | if cls != "unchanged": |
| 240 | changed[addr] = (base_surface[addr], current_surface[addr], cls) |
| 241 | |
| 242 | if as_json: |
| 243 | print(json.dumps( |
| 244 | { |
| 245 | "schema_version": __version__, |
| 246 | "commit": commit.commit_id[:8], |
| 247 | "base_commit": base_commit.commit_id[:8], |
| 248 | "language_filter": language, |
| 249 | "added": [ |
| 250 | _ApiEntry(a, r, language_of(a.split("::")[0])).to_dict() |
| 251 | for a, r in sorted(added.items()) |
| 252 | ], |
| 253 | "removed": [ |
| 254 | _ApiEntry(a, r, language_of(a.split("::")[0])).to_dict() |
| 255 | for a, r in sorted(removed.items()) |
| 256 | ], |
| 257 | "changed": [ |
| 258 | {**_ApiEntry(a, new, language_of(a.split("::")[0])).to_dict(), |
| 259 | "change": cls} |
| 260 | for a, (_, new, cls) in sorted(changed.items()) |
| 261 | ], |
| 262 | }, |
| 263 | indent=2, |
| 264 | )) |
| 265 | return |
| 266 | |
| 267 | print( |
| 268 | f"\nPublic API surface — commit {commit.commit_id[:8]} vs {base_commit.commit_id[:8]}" |
| 269 | ) |
| 270 | if language: |
| 271 | print(f" (language: {language})") |
| 272 | print("─" * 62) |
| 273 | |
| 274 | all_addrs = sorted(set(list(added) + list(removed) + list(changed))) |
| 275 | max_addr = max((len(a) for a in all_addrs), default=40) |
| 276 | |
| 277 | if added: |
| 278 | print(f"\nAdded ({len(added)}):") |
| 279 | for addr, rec in sorted(added.items()): |
| 280 | print(f" + {addr:<{max_addr}} {rec['kind']}") |
| 281 | |
| 282 | if removed: |
| 283 | print(f"\nRemoved ({len(removed)}):") |
| 284 | for addr, rec in sorted(removed.items()): |
| 285 | print(f" - {addr:<{max_addr}} {rec['kind']}") |
| 286 | |
| 287 | if changed: |
| 288 | print(f"\nChanged ({len(changed)}):") |
| 289 | for addr, (_, new, cls) in sorted(changed.items()): |
| 290 | print(f" ~ {addr:<{max_addr}} {new['kind']} ({cls})") |
| 291 | |
| 292 | if not added and not removed and not changed: |
| 293 | print("\n ✅ No public API changes detected.") |
| 294 | else: |
| 295 | n = len(added) + len(removed) + len(changed) |
| 296 | print(f"\n {n} public API change(s)") |