test_core_blame.py
python
| 1 | """Tests for muse/core/blame.py — line-level text attribution.""" |
| 2 | |
| 3 | from __future__ import annotations |
| 4 | |
| 5 | import hashlib |
| 6 | import json |
| 7 | import pathlib |
| 8 | |
| 9 | import pytest |
| 10 | |
| 11 | from muse.core.blame import BlameLine, blame_file |
| 12 | |
| 13 | |
| 14 | # --------------------------------------------------------------------------- |
| 15 | # Helpers |
| 16 | # --------------------------------------------------------------------------- |
| 17 | |
| 18 | |
| 19 | def _sha256(content: bytes) -> str: |
| 20 | return hashlib.sha256(content).hexdigest() |
| 21 | |
| 22 | |
| 23 | def _write_object(repo: pathlib.Path, content: bytes) -> str: |
| 24 | sha = _sha256(content) |
| 25 | obj_dir = repo / ".muse" / "objects" / sha[:2] |
| 26 | obj_dir.mkdir(parents=True, exist_ok=True) |
| 27 | (obj_dir / sha[2:]).write_bytes(content) |
| 28 | return sha |
| 29 | |
| 30 | |
| 31 | def _write_snapshot(repo: pathlib.Path, snap_id: str, manifest: dict[str, str]) -> None: |
| 32 | snap_dir = repo / ".muse" / "snapshots" |
| 33 | snap_dir.mkdir(parents=True, exist_ok=True) |
| 34 | (snap_dir / f"{snap_id}.json").write_text( |
| 35 | json.dumps({"snapshot_id": snap_id, "manifest": manifest}) |
| 36 | ) |
| 37 | |
| 38 | |
| 39 | def _write_commit( |
| 40 | repo: pathlib.Path, |
| 41 | commit_id: str, |
| 42 | snap_id: str, |
| 43 | message: str = "test", |
| 44 | parent: str | None = None, |
| 45 | author: str = "Author", |
| 46 | ) -> None: |
| 47 | import datetime |
| 48 | |
| 49 | commit_dir = repo / ".muse" / "commits" |
| 50 | commit_dir.mkdir(parents=True, exist_ok=True) |
| 51 | rec = { |
| 52 | "commit_id": commit_id, |
| 53 | "repo_id": "test-repo", |
| 54 | "branch": "main", |
| 55 | "snapshot_id": snap_id, |
| 56 | "message": message, |
| 57 | "committed_at": datetime.datetime.now(datetime.timezone.utc).isoformat(), |
| 58 | "parent_commit_id": parent, |
| 59 | "parent2_commit_id": None, |
| 60 | "author": author, |
| 61 | "metadata": {}, |
| 62 | } |
| 63 | (commit_dir / f"{commit_id}.json").write_text(json.dumps(rec)) |
| 64 | |
| 65 | |
| 66 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 67 | muse = tmp_path / ".muse" |
| 68 | for d in ("objects", "commits", "snapshots", "refs/heads"): |
| 69 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 70 | (muse / "repo.json").write_text(json.dumps({"repo_id": "test-repo"})) |
| 71 | (muse / "HEAD").write_text("refs/heads/main\n") |
| 72 | return tmp_path |
| 73 | |
| 74 | |
| 75 | # --------------------------------------------------------------------------- |
| 76 | # Tests |
| 77 | # --------------------------------------------------------------------------- |
| 78 | |
| 79 | |
| 80 | def test_blame_returns_none_for_missing_file(tmp_path: pathlib.Path) -> None: |
| 81 | repo = _make_repo(tmp_path) |
| 82 | snap_id = "s" * 64 |
| 83 | commit_id = "c" * 64 |
| 84 | _write_snapshot(repo, snap_id, {}) # empty manifest |
| 85 | _write_commit(repo, commit_id, snap_id) |
| 86 | |
| 87 | result = blame_file(repo, "nonexistent.txt", commit_id) |
| 88 | assert result is None |
| 89 | |
| 90 | |
| 91 | def test_blame_single_commit_all_lines_attributed(tmp_path: pathlib.Path) -> None: |
| 92 | repo = _make_repo(tmp_path) |
| 93 | content = b"line one\nline two\nline three\n" |
| 94 | obj_id = _write_object(repo, content) |
| 95 | snap_id = "s" * 64 |
| 96 | commit_id = "c" * 64 |
| 97 | _write_snapshot(repo, snap_id, {"readme.txt": obj_id}) |
| 98 | _write_commit(repo, commit_id, snap_id, message="initial commit", author="Alice") |
| 99 | |
| 100 | result = blame_file(repo, "readme.txt", commit_id) |
| 101 | assert result is not None |
| 102 | assert len(result) == 3 |
| 103 | for line in result: |
| 104 | assert isinstance(line, BlameLine) |
| 105 | assert line.commit_id == commit_id |
| 106 | |
| 107 | |
| 108 | def test_blame_line_numbers_are_1_indexed(tmp_path: pathlib.Path) -> None: |
| 109 | repo = _make_repo(tmp_path) |
| 110 | content = b"a\nb\nc\n" |
| 111 | obj_id = _write_object(repo, content) |
| 112 | snap_id = "s" * 64 |
| 113 | commit_id = "c" * 64 |
| 114 | _write_snapshot(repo, snap_id, {"f.txt": obj_id}) |
| 115 | _write_commit(repo, commit_id, snap_id) |
| 116 | |
| 117 | result = blame_file(repo, "f.txt", commit_id) |
| 118 | assert result is not None |
| 119 | assert [bl.lineno for bl in result] == [1, 2, 3] |
| 120 | |
| 121 | |
| 122 | def test_blame_content_matches_file(tmp_path: pathlib.Path) -> None: |
| 123 | repo = _make_repo(tmp_path) |
| 124 | content = b"hello\nworld\n" |
| 125 | obj_id = _write_object(repo, content) |
| 126 | snap_id = "s" * 64 |
| 127 | commit_id = "c" * 64 |
| 128 | _write_snapshot(repo, snap_id, {"f.txt": obj_id}) |
| 129 | _write_commit(repo, commit_id, snap_id) |
| 130 | |
| 131 | result = blame_file(repo, "f.txt", commit_id) |
| 132 | assert result is not None |
| 133 | assert result[0].content == "hello" |
| 134 | assert result[1].content == "world" |
| 135 | |
| 136 | |
| 137 | def test_blame_empty_file_returns_empty_list(tmp_path: pathlib.Path) -> None: |
| 138 | repo = _make_repo(tmp_path) |
| 139 | content = b"" |
| 140 | obj_id = _write_object(repo, content) |
| 141 | snap_id = "s" * 64 |
| 142 | commit_id = "c" * 64 |
| 143 | _write_snapshot(repo, snap_id, {"empty.txt": obj_id}) |
| 144 | _write_commit(repo, commit_id, snap_id) |
| 145 | |
| 146 | result = blame_file(repo, "empty.txt", commit_id) |
| 147 | assert result == [] |
| 148 | |
| 149 | |
| 150 | def test_blame_two_commits_attributes_older_lines_correctly(tmp_path: pathlib.Path) -> None: |
| 151 | """Lines present in both commits should be attributed to the older commit.""" |
| 152 | repo = _make_repo(tmp_path) |
| 153 | |
| 154 | # Commit 1: file with two lines. |
| 155 | content1 = b"original line 1\noriginal line 2\n" |
| 156 | obj1 = _write_object(repo, content1) |
| 157 | snap1 = "1" * 64 |
| 158 | commit1 = "a" * 64 |
| 159 | _write_snapshot(repo, snap1, {"f.txt": obj1}) |
| 160 | _write_commit(repo, commit1, snap1, message="initial", author="Alice") |
| 161 | |
| 162 | # Commit 2: same two lines + one new line. |
| 163 | content2 = b"original line 1\noriginal line 2\nnew line 3\n" |
| 164 | obj2 = _write_object(repo, content2) |
| 165 | snap2 = "2" * 64 |
| 166 | commit2 = "b" * 64 |
| 167 | _write_snapshot(repo, snap2, {"f.txt": obj2}) |
| 168 | _write_commit(repo, commit2, snap2, message="add line 3", parent=commit1, author="Bob") |
| 169 | |
| 170 | result = blame_file(repo, "f.txt", commit2) |
| 171 | assert result is not None |
| 172 | assert len(result) == 3 |
| 173 | # Lines 1 and 2 should be attributed to commit1 (they existed before commit2). |
| 174 | assert result[0].commit_id == commit1 |
| 175 | assert result[1].commit_id == commit1 |
| 176 | # Line 3 was added by commit2. |
| 177 | assert result[2].commit_id == commit2 |
| 178 | |
| 179 | |
| 180 | def test_blame_author_populated(tmp_path: pathlib.Path) -> None: |
| 181 | repo = _make_repo(tmp_path) |
| 182 | obj_id = _write_object(repo, b"line\n") |
| 183 | snap_id = "s" * 64 |
| 184 | commit_id = "c" * 64 |
| 185 | _write_snapshot(repo, snap_id, {"f.txt": obj_id}) |
| 186 | _write_commit(repo, commit_id, snap_id, author="Carol") |
| 187 | |
| 188 | result = blame_file(repo, "f.txt", commit_id) |
| 189 | assert result is not None |
| 190 | assert result[0].author == "Carol" |
| 191 | |
| 192 | |
| 193 | def test_blame_message_is_first_line_of_commit_message(tmp_path: pathlib.Path) -> None: |
| 194 | repo = _make_repo(tmp_path) |
| 195 | obj_id = _write_object(repo, b"line\n") |
| 196 | snap_id = "s" * 64 |
| 197 | commit_id = "c" * 64 |
| 198 | _write_snapshot(repo, snap_id, {"f.txt": obj_id}) |
| 199 | _write_commit(repo, commit_id, snap_id, message="feat: add feature\n\nLong body here.") |
| 200 | |
| 201 | result = blame_file(repo, "f.txt", commit_id) |
| 202 | assert result is not None |
| 203 | assert result[0].message == "feat: add feature" |
| 204 | |
| 205 | |
| 206 | # --------------------------------------------------------------------------- |
| 207 | # Stress |
| 208 | # --------------------------------------------------------------------------- |
| 209 | |
| 210 | |
| 211 | def test_blame_stress_100_line_file(tmp_path: pathlib.Path) -> None: |
| 212 | """Blame should handle a 100-line file without errors.""" |
| 213 | repo = _make_repo(tmp_path) |
| 214 | content = "\n".join(f"line {i}" for i in range(100)).encode() + b"\n" |
| 215 | obj_id = _write_object(repo, content) |
| 216 | snap_id = "s" * 64 |
| 217 | commit_id = "c" * 64 |
| 218 | _write_snapshot(repo, snap_id, {"big.txt": obj_id}) |
| 219 | _write_commit(repo, commit_id, snap_id) |
| 220 | |
| 221 | result = blame_file(repo, "big.txt", commit_id) |
| 222 | assert result is not None |
| 223 | assert len(result) == 100 |
| 224 | assert all(bl.commit_id == commit_id for bl in result) |