grep.py
file-level
1
files
1
commits
0
hotspots
0
π§ dead
0
π₯ blast risk
| 1 | """muse code grep -- semantic symbol search across the symbol graph. |
| 2 | |
| 3 | Unlike ``git grep`` which searches raw text lines, ``muse code grep`` searches |
| 4 | the *typed symbol graph* -- only returning actual symbol declarations with |
| 5 | their kind, file, line number, and stable content hash. |
| 6 | |
| 7 | No false positives from comments, string literals, or call sites. Every |
| 8 | result is a real symbol that exists in the repository. |
| 9 | |
| 10 | By default, PATTERN is matched case-insensitively against the bare symbol |
| 11 | name. When PATTERN contains a ``.`` or ``::`` it is also matched against the |
| 12 | fully-qualified name, so ``Invoice.validate`` finds only that specific method |
| 13 | rather than every symbol named ``validate``. |
| 14 | |
| 15 | Usage:: |
| 16 | |
| 17 | muse code grep "validate" # symbols whose name contains "validate" |
| 18 | muse code grep "Invoice.validate" # exact qualified-name match |
| 19 | muse code grep "^handle" --regex # names matching regex "^handle" |
| 20 | muse code grep "Invoice" --kind class # only class symbols |
| 21 | muse code grep "compute" --language go # only Go symbols (case-insensitive) |
| 22 | muse code grep "total" --file billing # scope to one file (fast) |
| 23 | muse code grep "total" --commit HEAD~5 # search a historical snapshot |
| 24 | muse code grep "validate" --count # just the total count |
| 25 | muse code grep "validate" --json # machine-readable output for agents |
| 26 | |
| 27 | Output:: |
| 28 | |
| 29 | muse/billing.py::validate_amount fn line 8 |
| 30 | muse/auth.py::validate_token fn line 14 |
| 31 | muse/auth.py::Validator class line 22 |
| 32 | muse/auth.py::Validator.validate method line 28 |
| 33 | |
| 34 | 4 match(es) across 2 file(s) |
| 35 | |
| 36 | Security note: patterns are capped at 512 characters to prevent ReDoS. |
| 37 | Invalid regex syntax is caught and reported as exit 1 rather than crashing. |
| 38 | """ |
| 39 | |
| 40 | from __future__ import annotations |
| 41 | |
| 42 | import argparse |
| 43 | import json |
| 44 | import logging |
| 45 | import pathlib |
| 46 | import re |
| 47 | import sys |
| 48 | |
| 49 | from muse.core.errors import ExitCode |
| 50 | from muse.core.repo import read_repo_id, require_repo |
| 51 | from muse.core.store import get_commit_snapshot_manifest, read_current_branch, resolve_commit_ref |
| 52 | from muse.plugins.code._query import language_of, symbols_for_snapshot |
| 53 | from muse.plugins.code._query import _SUFFIX_LANG # for case-insensitive lang normalisation |
| 54 | from muse.plugins.code.ast_parser import SymbolRecord |
| 55 | from muse.core.validation import sanitize_display |
| 56 | from muse.core.store import Manifest |
| 57 | |
| 58 | |
| 59 | type _IconMap = dict[str, str] |
| 60 | logger = logging.getLogger(__name__) |
| 61 | |
| 62 | # Guard against ReDoS: reject patterns longer than this before compiling. |
| 63 | _MAX_PATTERN_LEN: int = 512 |
| 64 | |
| 65 | _KIND_ICON: _IconMap = { |
| 66 | "function": "fn", |
| 67 | "async_function": "fn~", |
| 68 | "class": "class", |
| 69 | "method": "method", |
| 70 | "async_method": "method~", |
| 71 | "variable": "var", |
| 72 | "import": "import", |
| 73 | } |
| 74 | |
| 75 | # Canonical map for case-insensitive --language matching. |
| 76 | _LANG_CANONICAL: _IconMap = {lang.lower(): lang for lang in set(_SUFFIX_LANG.values())} |
| 77 | |
| 78 | |
| 79 | def _normalise_language(lang: str) -> str: |
| 80 | return _LANG_CANONICAL.get(lang.strip().lower(), lang.strip()) |
| 81 | |
| 82 | |
| 83 | # --------------------------------------------------------------------------- |
| 84 | # Repository helpers |
| 85 | # --------------------------------------------------------------------------- |
| 86 | |
| 87 | |
| 88 | |
| 89 | # --------------------------------------------------------------------------- |
| 90 | # File-filter helpers (same as symbols.py) |
| 91 | # --------------------------------------------------------------------------- |
| 92 | |
| 93 | |
| 94 | def _file_matches(file_path: str, file_filter: str) -> bool: |
| 95 | """True if *file_path* equals or ends with ``/<file_filter>``.""" |
| 96 | if file_path == file_filter: |
| 97 | return True |
| 98 | normalized = file_filter.replace("\\", "/") |
| 99 | return file_path.endswith("/" + normalized) |
| 100 | |
| 101 | |
| 102 | def _resolve_file_filter( |
| 103 | file_filter: str, |
| 104 | manifest: Manifest, |
| 105 | ) -> str | None: |
| 106 | """Resolve a partial path suffix to the exact manifest key. |
| 107 | |
| 108 | Exits non-zero on ambiguity; returns ``None`` when there is no match |
| 109 | (caller handles the empty result). |
| 110 | """ |
| 111 | matching = [p for p in sorted(manifest) if _file_matches(p, file_filter)] |
| 112 | if len(matching) == 1: |
| 113 | return matching[0] |
| 114 | if len(matching) > 1: |
| 115 | print( |
| 116 | f"β '{file_filter}' is ambiguous β matches {len(matching)} files. " |
| 117 | "Use a more specific path:", |
| 118 | file=sys.stderr, |
| 119 | ) |
| 120 | for m in matching[:10]: |
| 121 | print(f" {m}", file=sys.stderr) |
| 122 | if len(matching) > 10: |
| 123 | print(f" β¦ and {len(matching) - 10} more", file=sys.stderr) |
| 124 | raise SystemExit(ExitCode.USER_ERROR) |
| 125 | return None # no match β caller handles empty result |
| 126 | |
| 127 | |
| 128 | # --------------------------------------------------------------------------- |
| 129 | # Argument parser registration |
| 130 | # --------------------------------------------------------------------------- |
| 131 | |
| 132 | |
| 133 | def register(subparsers: "argparse._SubParsersAction[argparse.ArgumentParser]") -> None: |
| 134 | """Register the grep subcommand.""" |
| 135 | parser = subparsers.add_parser( |
| 136 | "grep", |
| 137 | help="Search the symbol graph by name β not file text.", |
| 138 | description=__doc__, |
| 139 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 140 | ) |
| 141 | parser.add_argument( |
| 142 | "pattern", metavar="PATTERN", |
| 143 | help="Name pattern to search for.", |
| 144 | ) |
| 145 | parser.add_argument( |
| 146 | "--regex", "-e", action="store_true", dest="use_regex", |
| 147 | help="Treat PATTERN as a regular expression (default: substring match).", |
| 148 | ) |
| 149 | parser.add_argument( |
| 150 | "--kind", "-k", default=None, metavar="KIND", dest="kind_filter", |
| 151 | help="Restrict to symbols of this kind (function, class, method, β¦).", |
| 152 | ) |
| 153 | parser.add_argument( |
| 154 | "--language", "-l", default=None, metavar="LANG", dest="language_filter", |
| 155 | help="Restrict to symbols from files of this language (case-insensitive).", |
| 156 | ) |
| 157 | parser.add_argument( |
| 158 | "--file", "-f", default=None, metavar="PATH", dest="file_filter", |
| 159 | help=( |
| 160 | "Scope to a single file. Accepts an exact path or a unique suffix " |
| 161 | "(e.g. 'billing.py' matches 'src/billing.py'). Up to 24x faster." |
| 162 | ), |
| 163 | ) |
| 164 | parser.add_argument( |
| 165 | "--commit", "-c", default=None, metavar="REF", dest="ref", |
| 166 | help="Search a historical commit instead of the working tree.", |
| 167 | ) |
| 168 | parser.add_argument( |
| 169 | "--hashes", action="store_true", dest="show_hashes", |
| 170 | help="Include content hashes in output.", |
| 171 | ) |
| 172 | |
| 173 | output_group = parser.add_mutually_exclusive_group() |
| 174 | output_group.add_argument( |
| 175 | "--count", action="store_true", dest="count_only", |
| 176 | help="Print only the total match count.", |
| 177 | ) |
| 178 | output_group.add_argument( |
| 179 | "--json", action="store_true", dest="as_json", |
| 180 | help="Emit results as structured JSON.", |
| 181 | ) |
| 182 | |
| 183 | parser.set_defaults(func=run) |
| 184 | |
| 185 | |
| 186 | # --------------------------------------------------------------------------- |
| 187 | # Command entry point |
| 188 | # --------------------------------------------------------------------------- |
| 189 | |
| 190 | |
| 191 | def run(args: argparse.Namespace) -> None: |
| 192 | """Search the symbol graph by name β not file text. |
| 193 | |
| 194 | ``muse code grep`` searches the typed, content-addressed symbol graph. |
| 195 | Every result is a real symbol declaration β no false positives from |
| 196 | comments, string literals, or call sites. |
| 197 | |
| 198 | The ``--regex`` flag enables full Python regex syntax. Without it, |
| 199 | PATTERN is matched as a case-insensitive substring. |
| 200 | |
| 201 | When PATTERN contains a ``.`` or ``::`` it is also matched against the |
| 202 | fully-qualified symbol name, so ``Invoice.validate`` returns exactly that |
| 203 | method rather than every symbol named ``validate``. |
| 204 | |
| 205 | Use ``--file`` to scope the search to one file β this is up to 24x faster |
| 206 | than a full-codebase scan and sufficient for most targeted lookups. |
| 207 | |
| 208 | When ``--commit`` is omitted the working tree is searched, reflecting |
| 209 | edits not yet committed. |
| 210 | |
| 211 | Patterns are capped at 512 characters to guard against ReDoS. |
| 212 | """ |
| 213 | pattern: str = args.pattern |
| 214 | use_regex: bool = args.use_regex |
| 215 | kind_filter: str | None = args.kind_filter |
| 216 | language_filter: str | None = args.language_filter |
| 217 | file_filter: str | None = args.file_filter |
| 218 | ref: str | None = args.ref |
| 219 | show_hashes: bool = args.show_hashes |
| 220 | count_only: bool = args.count_only |
| 221 | as_json: bool = args.as_json |
| 222 | |
| 223 | # ββ Input validation ββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 224 | |
| 225 | if len(pattern) > _MAX_PATTERN_LEN: |
| 226 | print( |
| 227 | f"β Pattern too long ({len(pattern)} chars) β maximum is {_MAX_PATTERN_LEN}.", |
| 228 | file=sys.stderr, |
| 229 | ) |
| 230 | raise SystemExit(ExitCode.USER_ERROR) |
| 231 | |
| 232 | if language_filter is not None: |
| 233 | language_filter = _normalise_language(language_filter) |
| 234 | |
| 235 | # When pattern contains a separator, also search qualified names. |
| 236 | search_qualified = "." in pattern or "::" in pattern |
| 237 | |
| 238 | try: |
| 239 | regex = ( |
| 240 | re.compile(pattern, re.IGNORECASE) |
| 241 | if use_regex |
| 242 | else re.compile(re.escape(pattern), re.IGNORECASE) |
| 243 | ) |
| 244 | except re.error as exc: |
| 245 | print(f"β Invalid regex pattern: {exc}", file=sys.stderr) |
| 246 | raise SystemExit(ExitCode.USER_ERROR) |
| 247 | |
| 248 | # ββ Repo / commit resolution ββββββββββββββββββββββββββββββββββββββββββββββ |
| 249 | |
| 250 | root = require_repo() |
| 251 | repo_id = read_repo_id(root) |
| 252 | branch = read_current_branch(root) |
| 253 | |
| 254 | commit = resolve_commit_ref(root, repo_id, branch, ref) |
| 255 | if commit is None: |
| 256 | print(f"β Commit '{ref or 'HEAD'}' not found.", file=sys.stderr) |
| 257 | raise SystemExit(ExitCode.USER_ERROR) |
| 258 | |
| 259 | manifest = get_commit_snapshot_manifest(root, commit.commit_id) or {} |
| 260 | |
| 261 | # ββ File-filter resolution ββββββββββββββββββββββββββββββββββββββββββββββββ |
| 262 | |
| 263 | resolved_file_filter = file_filter |
| 264 | if file_filter is not None: |
| 265 | found = _resolve_file_filter(file_filter, manifest) |
| 266 | if found is not None: |
| 267 | resolved_file_filter = found |
| 268 | # None β no match; pass original so symbols_for_snapshot returns {} |
| 269 | |
| 270 | # ββ Working-tree vs object-store mode ββββββββββββββββββββββββββββββββββββ |
| 271 | |
| 272 | working_tree = ref is None |
| 273 | workdir = root if working_tree else None |
| 274 | source_ref = "working-tree" if working_tree else commit.commit_id[:8] |
| 275 | |
| 276 | # ββ Symbol extraction βββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 277 | |
| 278 | symbol_map = symbols_for_snapshot( |
| 279 | root, manifest, |
| 280 | kind_filter=kind_filter, |
| 281 | file_filter=resolved_file_filter, |
| 282 | language_filter=language_filter, |
| 283 | workdir=workdir, |
| 284 | ) |
| 285 | |
| 286 | # ββ Pattern matching ββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 287 | |
| 288 | matches: list[tuple[str, str, SymbolRecord]] = [] |
| 289 | for file_path, tree in sorted(symbol_map.items()): |
| 290 | for addr, rec in sorted(tree.items(), key=lambda kv: kv[1]["lineno"]): |
| 291 | name_hit = regex.search(rec["name"]) |
| 292 | qual_hit = search_qualified and regex.search(rec["qualified_name"]) |
| 293 | if name_hit or qual_hit: |
| 294 | matches.append((file_path, addr, rec)) |
| 295 | |
| 296 | # ββ Output ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 297 | |
| 298 | if count_only: |
| 299 | print(f"{len(matches)} match(es)") |
| 300 | return |
| 301 | |
| 302 | if as_json: |
| 303 | results: list[dict[str, str | int | bool]] = [] |
| 304 | for _fp, addr, rec in matches: |
| 305 | results.append({ |
| 306 | "address": addr, |
| 307 | "kind": rec["kind"], |
| 308 | "name": rec["name"], |
| 309 | "qualified_name": rec["qualified_name"], |
| 310 | "file": addr.split("::")[0], |
| 311 | "lineno": rec["lineno"], |
| 312 | "language": language_of(addr.split("::")[0]), |
| 313 | "content_id": rec["content_id"], |
| 314 | }) |
| 315 | print(json.dumps({ |
| 316 | "source_ref": source_ref, |
| 317 | "working_tree": working_tree, |
| 318 | "pattern": pattern, |
| 319 | "total_matches": len(matches), |
| 320 | "results": results, |
| 321 | }, indent=2)) |
| 322 | return |
| 323 | |
| 324 | if not matches: |
| 325 | print(f" (no symbols matching '{sanitize_display(pattern)}')") |
| 326 | return |
| 327 | |
| 328 | files_seen: set[str] = set() |
| 329 | for file_path, addr, rec in matches: |
| 330 | files_seen.add(file_path) |
| 331 | icon = _KIND_ICON.get(rec["kind"], rec["kind"]) |
| 332 | line = rec["lineno"] |
| 333 | hash_part = f" {rec['content_id'][:8]}.." if show_hashes else "" |
| 334 | print(f" {sanitize_display(addr):<60} {icon:<10} line {line:>4}{hash_part}") |
| 335 | |
| 336 | print(f"\n{len(matches)} match(es) across {len(files_seen)} file(s)") |