test_cmd_content_grep.py
python
| 1 | """Tests for ``muse content-grep``. |
| 2 | |
| 3 | Covers: no match exit-1, pattern found, --files-only, --count, --ignore-case, |
| 4 | --format json, binary skip, multi-file, stress: 100 files. |
| 5 | """ |
| 6 | |
| 7 | from __future__ import annotations |
| 8 | |
| 9 | import datetime |
| 10 | import hashlib |
| 11 | import json |
| 12 | import pathlib |
| 13 | |
| 14 | import pytest |
| 15 | from tests.cli_test_helper import CliRunner |
| 16 | |
| 17 | cli = None # argparse migration — CliRunner ignores this arg |
| 18 | from muse.core.object_store import write_object |
| 19 | from muse.core.snapshot import compute_snapshot_id |
| 20 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 21 | |
| 22 | runner = CliRunner() |
| 23 | |
| 24 | _REPO_ID = "cgrep-test" |
| 25 | |
| 26 | |
| 27 | # --------------------------------------------------------------------------- |
| 28 | # Helpers |
| 29 | # --------------------------------------------------------------------------- |
| 30 | |
| 31 | |
| 32 | def _sha(data: bytes) -> str: |
| 33 | return hashlib.sha256(data).hexdigest() |
| 34 | |
| 35 | |
| 36 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 37 | muse = path / ".muse" |
| 38 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 39 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 40 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 41 | (muse / "repo.json").write_text( |
| 42 | json.dumps({"repo_id": _REPO_ID, "domain": "midi"}), encoding="utf-8" |
| 43 | ) |
| 44 | return path |
| 45 | |
| 46 | |
| 47 | def _env(repo: pathlib.Path) -> dict[str, str]: |
| 48 | return {"MUSE_REPO_ROOT": str(repo)} |
| 49 | |
| 50 | |
| 51 | _counter = 0 |
| 52 | |
| 53 | |
| 54 | def _commit_files(root: pathlib.Path, files: dict[str, bytes]) -> str: |
| 55 | global _counter |
| 56 | _counter += 1 |
| 57 | manifest: dict[str, str] = {} |
| 58 | for rel_path, content in files.items(): |
| 59 | obj_id = _sha(content) |
| 60 | write_object(root, obj_id, content) |
| 61 | manifest[rel_path] = obj_id |
| 62 | snap_id = compute_snapshot_id(manifest) |
| 63 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 64 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 65 | commit_id = _sha(f"{_counter}:{snap_id}:{committed_at.isoformat()}".encode()) |
| 66 | write_commit(root, CommitRecord( |
| 67 | commit_id=commit_id, |
| 68 | repo_id=_REPO_ID, |
| 69 | branch="main", |
| 70 | snapshot_id=snap_id, |
| 71 | message=f"commit {_counter}", |
| 72 | committed_at=committed_at, |
| 73 | )) |
| 74 | (root / ".muse" / "refs" / "heads" / "main").write_text(commit_id, encoding="utf-8") |
| 75 | return commit_id |
| 76 | |
| 77 | |
| 78 | # --------------------------------------------------------------------------- |
| 79 | # Unit: help |
| 80 | # --------------------------------------------------------------------------- |
| 81 | |
| 82 | |
| 83 | def test_content_grep_help() -> None: |
| 84 | result = runner.invoke(cli, ["content-grep", "--help"]) |
| 85 | assert result.exit_code == 0 |
| 86 | assert "--pattern" in result.output or "-p" in result.output |
| 87 | |
| 88 | |
| 89 | # --------------------------------------------------------------------------- |
| 90 | # Unit: no match → exit 1 |
| 91 | # --------------------------------------------------------------------------- |
| 92 | |
| 93 | |
| 94 | def test_content_grep_no_match(tmp_path: pathlib.Path) -> None: |
| 95 | _init_repo(tmp_path) |
| 96 | _commit_files(tmp_path, {"song.txt": b"chord: Am\ntempo: 120\n"}) |
| 97 | result = runner.invoke(cli, ["content-grep", "--pattern", "ZZZNOMATCH"], env=_env(tmp_path)) |
| 98 | assert result.exit_code != 0 |
| 99 | |
| 100 | |
| 101 | # --------------------------------------------------------------------------- |
| 102 | # Unit: match found → exit 0 |
| 103 | # --------------------------------------------------------------------------- |
| 104 | |
| 105 | |
| 106 | def test_content_grep_match_found(tmp_path: pathlib.Path) -> None: |
| 107 | _init_repo(tmp_path) |
| 108 | _commit_files(tmp_path, {"song.txt": b"chord: Cm7\ntempo: 120\n"}) |
| 109 | result = runner.invoke(cli, ["content-grep", "--pattern", "Cm7"], env=_env(tmp_path)) |
| 110 | assert result.exit_code == 0 |
| 111 | assert "song.txt" in result.output |
| 112 | |
| 113 | |
| 114 | # --------------------------------------------------------------------------- |
| 115 | # Unit: --ignore-case |
| 116 | # --------------------------------------------------------------------------- |
| 117 | |
| 118 | |
| 119 | def test_content_grep_ignore_case(tmp_path: pathlib.Path) -> None: |
| 120 | _init_repo(tmp_path) |
| 121 | _commit_files(tmp_path, {"notes.txt": b"VERSE: intro melody\n"}) |
| 122 | result = runner.invoke( |
| 123 | cli, ["content-grep", "--pattern", "verse", "--ignore-case"], env=_env(tmp_path) |
| 124 | ) |
| 125 | assert result.exit_code == 0 |
| 126 | assert "notes.txt" in result.output |
| 127 | |
| 128 | |
| 129 | def test_content_grep_case_sensitive_no_match(tmp_path: pathlib.Path) -> None: |
| 130 | _init_repo(tmp_path) |
| 131 | _commit_files(tmp_path, {"notes.txt": b"VERSE: intro melody\n"}) |
| 132 | result = runner.invoke( |
| 133 | cli, ["content-grep", "--pattern", "verse"], env=_env(tmp_path) |
| 134 | ) |
| 135 | # Case-sensitive: "verse" ≠ "VERSE" → no match. |
| 136 | assert result.exit_code != 0 |
| 137 | |
| 138 | |
| 139 | # --------------------------------------------------------------------------- |
| 140 | # Unit: --files-only |
| 141 | # --------------------------------------------------------------------------- |
| 142 | |
| 143 | |
| 144 | def test_content_grep_files_only(tmp_path: pathlib.Path) -> None: |
| 145 | _init_repo(tmp_path) |
| 146 | _commit_files(tmp_path, { |
| 147 | "a.txt": b"match here\n", |
| 148 | "b.txt": b"match here too\n", |
| 149 | }) |
| 150 | result = runner.invoke( |
| 151 | cli, ["content-grep", "--pattern", "match", "--files-only"], env=_env(tmp_path) |
| 152 | ) |
| 153 | assert result.exit_code == 0 |
| 154 | lines = [l.strip() for l in result.output.strip().split("\n") if l.strip()] |
| 155 | for line in lines: |
| 156 | assert ":" not in line or line.startswith("a.txt") or line.startswith("b.txt") |
| 157 | |
| 158 | |
| 159 | # --------------------------------------------------------------------------- |
| 160 | # Unit: --count |
| 161 | # --------------------------------------------------------------------------- |
| 162 | |
| 163 | |
| 164 | def test_content_grep_count(tmp_path: pathlib.Path) -> None: |
| 165 | _init_repo(tmp_path) |
| 166 | _commit_files(tmp_path, {"multi.txt": b"hit\nhit\nhit\nmiss\n"}) |
| 167 | result = runner.invoke( |
| 168 | cli, ["content-grep", "--pattern", "hit", "--count"], env=_env(tmp_path) |
| 169 | ) |
| 170 | assert result.exit_code == 0 |
| 171 | assert "3" in result.output |
| 172 | |
| 173 | |
| 174 | # --------------------------------------------------------------------------- |
| 175 | # Unit: --format json |
| 176 | # --------------------------------------------------------------------------- |
| 177 | |
| 178 | |
| 179 | def test_content_grep_json_output(tmp_path: pathlib.Path) -> None: |
| 180 | _init_repo(tmp_path) |
| 181 | _commit_files(tmp_path, {"song.midi.txt": b"note: C4\nnote: D4\n"}) |
| 182 | result = runner.invoke( |
| 183 | cli, ["content-grep", "--pattern", "note", "--format", "json"], env=_env(tmp_path) |
| 184 | ) |
| 185 | assert result.exit_code == 0 |
| 186 | data = json.loads(result.output) |
| 187 | assert isinstance(data, list) |
| 188 | assert len(data) >= 1 |
| 189 | assert data[0]["match_count"] >= 2 |
| 190 | |
| 191 | |
| 192 | # --------------------------------------------------------------------------- |
| 193 | # Unit: binary file skipped silently |
| 194 | # --------------------------------------------------------------------------- |
| 195 | |
| 196 | |
| 197 | def test_content_grep_binary_skipped(tmp_path: pathlib.Path) -> None: |
| 198 | _init_repo(tmp_path) |
| 199 | binary_content = b"\x00\x01\x02\x03" * 100 |
| 200 | text_content = b"searchable text here\n" |
| 201 | _commit_files(tmp_path, { |
| 202 | "binary.bin": binary_content, |
| 203 | "text.txt": text_content, |
| 204 | }) |
| 205 | result = runner.invoke( |
| 206 | cli, ["content-grep", "--pattern", "searchable"], env=_env(tmp_path) |
| 207 | ) |
| 208 | assert result.exit_code == 0 |
| 209 | assert "text.txt" in result.output |
| 210 | |
| 211 | |
| 212 | # --------------------------------------------------------------------------- |
| 213 | # Unit: short flags work |
| 214 | # --------------------------------------------------------------------------- |
| 215 | |
| 216 | |
| 217 | def test_content_grep_short_flags(tmp_path: pathlib.Path) -> None: |
| 218 | _init_repo(tmp_path) |
| 219 | _commit_files(tmp_path, {"f.txt": b"hello world\n"}) |
| 220 | result = runner.invoke( |
| 221 | cli, ["content-grep", "-p", "hello", "-i", "-f", "json"], env=_env(tmp_path) |
| 222 | ) |
| 223 | assert result.exit_code == 0 |
| 224 | data = json.loads(result.output) |
| 225 | assert len(data) >= 1 |
| 226 | |
| 227 | |
| 228 | # --------------------------------------------------------------------------- |
| 229 | # Stress: 100 files, pattern matches 50 |
| 230 | # --------------------------------------------------------------------------- |
| 231 | |
| 232 | |
| 233 | def test_content_grep_stress_100_files(tmp_path: pathlib.Path) -> None: |
| 234 | _init_repo(tmp_path) |
| 235 | files: dict[str, bytes] = {} |
| 236 | for i in range(100): |
| 237 | content = b"TARGET_LINE\n" if i % 2 == 0 else b"other content\n" |
| 238 | files[f"file_{i:04d}.txt"] = content |
| 239 | _commit_files(tmp_path, files) |
| 240 | result = runner.invoke( |
| 241 | cli, ["content-grep", "--pattern", "TARGET_LINE", "--format", "json"], env=_env(tmp_path) |
| 242 | ) |
| 243 | assert result.exit_code == 0 |
| 244 | data = json.loads(result.output) |
| 245 | assert len(data) == 50 |