test_phase5_store_linear_walks.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
125 days ago
| 1 | """TDD — Phase 5: store.py linear walks become iter_ancestors wrappers. |
| 2 | |
| 3 | Phase 5 of issue #6 (generic DAG walker). |
| 4 | |
| 5 | ``store.py`` contains two linear first-parent walkers that predate |
| 6 | ``graph.py``: |
| 7 | |
| 8 | - ``walk_commits_between_result`` — bounded range walk, returns WalkResult |
| 9 | - ``get_commits_for_branch`` — branch HEAD walk with optional max_count |
| 10 | |
| 11 | Both use the same inline pattern: a ``while commit_id`` loop that follows |
| 12 | ``parent_commit_id`` one step at a time. Neither uses ``iter_ancestors``. |
| 13 | |
| 14 | Fix: make both functions thin wrappers over |
| 15 | ``iter_ancestors(root, start, first_parent_only=True)``. |
| 16 | Behaviour must be identical — same commit order (newest-first), same |
| 17 | ``from_commit_id`` stop condition, same ``max_commits`` / ``max_count`` |
| 18 | cap, same ``truncated`` signalling. |
| 19 | |
| 20 | ``walk_commits_between`` is a one-liner wrapper over |
| 21 | ``walk_commits_between_result`` — it stays as-is (no inline loop). |
| 22 | |
| 23 | Coverage |
| 24 | -------- |
| 25 | P5-1 Structural — ``walk_commits_between_result`` source contains |
| 26 | ``iter_ancestors`` and no inline ``while commit_id`` loop |
| 27 | P5-2 Structural — ``get_commits_for_branch`` source contains |
| 28 | ``iter_ancestors`` and no inline ``while commit_id`` loop |
| 29 | P5-3 Behavioural — ``walk_commits_between_result`` linear chain matches |
| 30 | old behaviour: commits newest-first, stops before from_commit_id |
| 31 | P5-4 Behavioural — ``walk_commits_between_result`` truncated flag fires at cap |
| 32 | P5-5 Behavioural — ``walk_commits_between_result`` full walk (no from_commit) |
| 33 | P5-6 Behavioural — ``get_commits_for_branch`` returns commits newest-first |
| 34 | P5-7 Behavioural — ``get_commits_for_branch`` respects max_count |
| 35 | """ |
| 36 | from __future__ import annotations |
| 37 | |
| 38 | import datetime |
| 39 | import inspect |
| 40 | import json |
| 41 | import pathlib |
| 42 | |
| 43 | import pytest |
| 44 | |
| 45 | from muse._version import __version__ |
| 46 | from muse.core.object_store import write_object |
| 47 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 48 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 49 | from muse.core.types import blob_id |
| 50 | from muse.core.paths import muse_dir |
| 51 | |
| 52 | |
| 53 | # --------------------------------------------------------------------------- |
| 54 | # Repo fixture helpers |
| 55 | # --------------------------------------------------------------------------- |
| 56 | |
| 57 | def _repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 58 | dot_muse = muse_dir(tmp_path) |
| 59 | for d in ("commits", "snapshots", "objects", "refs/heads", "remotes"): |
| 60 | (dot_muse / d).mkdir(parents=True, exist_ok=True) |
| 61 | (dot_muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 62 | (dot_muse / "repo.json").write_text( |
| 63 | json.dumps({"repo_id": "test-repo", "schema_version": __version__, "domain": "code"}) |
| 64 | ) |
| 65 | monkeypatch.setenv("MUSE_REPO_ROOT", str(tmp_path)) |
| 66 | monkeypatch.chdir(tmp_path) |
| 67 | return tmp_path |
| 68 | |
| 69 | |
| 70 | def _write_obj(root: pathlib.Path, content: bytes) -> str: |
| 71 | oid = blob_id(content) |
| 72 | write_object(root, oid, content) |
| 73 | return oid |
| 74 | |
| 75 | |
| 76 | def _make_commit( |
| 77 | root: pathlib.Path, |
| 78 | manifest: dict[str, str], |
| 79 | parent_id: str | None = None, |
| 80 | *, |
| 81 | message: str = "test", |
| 82 | ) -> CommitRecord: |
| 83 | snap_id = compute_snapshot_id(manifest) |
| 84 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 85 | ts = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 86 | cid = compute_commit_id( |
| 87 | parent_ids=[parent_id] if parent_id else [], |
| 88 | snapshot_id=snap_id, |
| 89 | message=message, |
| 90 | committed_at_iso=ts.isoformat(), |
| 91 | ) |
| 92 | rec = CommitRecord( |
| 93 | repo_id="test-repo", |
| 94 | commit_id=cid, |
| 95 | branch="main", |
| 96 | snapshot_id=snap_id, |
| 97 | message=message, |
| 98 | committed_at=ts, |
| 99 | parent_commit_id=parent_id, |
| 100 | ) |
| 101 | write_commit(root, rec) |
| 102 | return rec |
| 103 | |
| 104 | |
| 105 | def _linear_chain(root: pathlib.Path, n: int) -> list[CommitRecord]: |
| 106 | """Build a linear chain of n commits, return oldest-first.""" |
| 107 | oid = _write_obj(root, b"data") |
| 108 | commits: list[CommitRecord] = [] |
| 109 | parent_id: str | None = None |
| 110 | for i in range(n): |
| 111 | c = _make_commit(root, {"f.py": oid}, parent_id, message=f"commit {i}") |
| 112 | commits.append(c) |
| 113 | parent_id = c.commit_id |
| 114 | return commits |
| 115 | |
| 116 | |
| 117 | # --------------------------------------------------------------------------- |
| 118 | # P5-1 Structural — walk_commits_between_result uses iter_ancestors |
| 119 | # --------------------------------------------------------------------------- |
| 120 | |
| 121 | def test_p5_1_walk_commits_between_result_uses_iter_ancestors() -> None: |
| 122 | """walk_commits_between_result must delegate to iter_ancestors. |
| 123 | |
| 124 | The inline ``while commit_id`` pattern must not appear in the function |
| 125 | body — it predates graph.py and is now replaced by iter_ancestors with |
| 126 | first_parent_only=True. |
| 127 | """ |
| 128 | from muse.core import store as store_mod |
| 129 | |
| 130 | src = inspect.getsource(store_mod.walk_commits_between_result) |
| 131 | |
| 132 | assert "iter_ancestors" in src, ( |
| 133 | "walk_commits_between_result must delegate to iter_ancestors. " |
| 134 | "Replace the inline while-loop with iter_ancestors(first_parent_only=True)." |
| 135 | ) |
| 136 | assert "while commit_id" not in src, ( |
| 137 | "walk_commits_between_result still has an inline while-loop. " |
| 138 | "Replace with iter_ancestors(first_parent_only=True)." |
| 139 | ) |
| 140 | |
| 141 | |
| 142 | # --------------------------------------------------------------------------- |
| 143 | # P5-2 Structural — get_commits_for_branch uses iter_ancestors |
| 144 | # --------------------------------------------------------------------------- |
| 145 | |
| 146 | def test_p5_2_get_commits_for_branch_uses_iter_ancestors() -> None: |
| 147 | """get_commits_for_branch must delegate to iter_ancestors.""" |
| 148 | from muse.core import store as store_mod |
| 149 | |
| 150 | src = inspect.getsource(store_mod.get_commits_for_branch) |
| 151 | |
| 152 | assert "iter_ancestors" in src, ( |
| 153 | "get_commits_for_branch must delegate to iter_ancestors. " |
| 154 | "Replace the inline while-loop with iter_ancestors(first_parent_only=True)." |
| 155 | ) |
| 156 | assert "while commit_id" not in src, ( |
| 157 | "get_commits_for_branch still has an inline while-loop. " |
| 158 | "Replace with iter_ancestors(first_parent_only=True)." |
| 159 | ) |
| 160 | |
| 161 | |
| 162 | # --------------------------------------------------------------------------- |
| 163 | # P5-3 Behavioural — walk_commits_between_result range stop |
| 164 | # --------------------------------------------------------------------------- |
| 165 | |
| 166 | def test_p5_3_walk_commits_between_result_stops_before_from_commit( |
| 167 | tmp_path: pathlib.Path, |
| 168 | monkeypatch: pytest.MonkeyPatch, |
| 169 | ) -> None: |
| 170 | """walk_commits_between_result returns commits from to_commit up to but |
| 171 | not including from_commit, newest-first. |
| 172 | |
| 173 | Chain (oldest → newest): C1 → C2 → C3 → C4 |
| 174 | Call: walk_commits_between_result(root, C4, from_commit_id=C1) |
| 175 | Expected: [C4, C3, C2] — C1 is the exclusive lower bound. |
| 176 | """ |
| 177 | from muse.core.store import walk_commits_between_result |
| 178 | |
| 179 | root = _repo(tmp_path, monkeypatch) |
| 180 | chain = _linear_chain(root, 4) |
| 181 | c1, c2, c3, c4 = chain |
| 182 | |
| 183 | result = walk_commits_between_result(root, c4.commit_id, from_commit_id=c1.commit_id) |
| 184 | |
| 185 | assert result["truncated"] is False |
| 186 | ids = [c.commit_id for c in result["commits"]] |
| 187 | assert ids == [c4.commit_id, c3.commit_id, c2.commit_id], ( |
| 188 | f"Expected [C4, C3, C2] (C1 excluded), got {[i[:12] for i in ids]}" |
| 189 | ) |
| 190 | |
| 191 | |
| 192 | # --------------------------------------------------------------------------- |
| 193 | # P5-4 Behavioural — walk_commits_between_result truncated flag |
| 194 | # --------------------------------------------------------------------------- |
| 195 | |
| 196 | def test_p5_4_walk_commits_between_result_truncated_flag( |
| 197 | tmp_path: pathlib.Path, |
| 198 | monkeypatch: pytest.MonkeyPatch, |
| 199 | ) -> None: |
| 200 | """truncated=True when max_commits is hit before the chain is exhausted.""" |
| 201 | from muse.core.store import walk_commits_between_result |
| 202 | |
| 203 | root = _repo(tmp_path, monkeypatch) |
| 204 | chain = _linear_chain(root, 5) |
| 205 | |
| 206 | result = walk_commits_between_result(root, chain[-1].commit_id, max_commits=2) |
| 207 | |
| 208 | assert result["truncated"] is True |
| 209 | assert len(result["commits"]) == 2 |
| 210 | |
| 211 | |
| 212 | # --------------------------------------------------------------------------- |
| 213 | # P5-5 Behavioural — walk_commits_between_result full walk (no from_commit) |
| 214 | # --------------------------------------------------------------------------- |
| 215 | |
| 216 | def test_p5_5_walk_commits_between_result_full_walk( |
| 217 | tmp_path: pathlib.Path, |
| 218 | monkeypatch: pytest.MonkeyPatch, |
| 219 | ) -> None: |
| 220 | """When from_commit_id is None, walk all the way to the initial commit.""" |
| 221 | from muse.core.store import walk_commits_between_result |
| 222 | |
| 223 | root = _repo(tmp_path, monkeypatch) |
| 224 | chain = _linear_chain(root, 4) |
| 225 | |
| 226 | result = walk_commits_between_result(root, chain[-1].commit_id) |
| 227 | |
| 228 | assert result["truncated"] is False |
| 229 | assert result["count"] == 4 |
| 230 | ids = [c.commit_id for c in result["commits"]] |
| 231 | expected = [c.commit_id for c in reversed(chain)] |
| 232 | assert ids == expected, "Full walk must return all commits newest-first" |
| 233 | |
| 234 | |
| 235 | # --------------------------------------------------------------------------- |
| 236 | # P5-6 Behavioural — get_commits_for_branch returns newest-first |
| 237 | # --------------------------------------------------------------------------- |
| 238 | |
| 239 | def test_p5_6_get_commits_for_branch_newest_first( |
| 240 | tmp_path: pathlib.Path, |
| 241 | monkeypatch: pytest.MonkeyPatch, |
| 242 | ) -> None: |
| 243 | """get_commits_for_branch returns commits newest-first for a branch.""" |
| 244 | from muse.core.store import get_commits_for_branch |
| 245 | |
| 246 | root = _repo(tmp_path, monkeypatch) |
| 247 | chain = _linear_chain(root, 3) |
| 248 | |
| 249 | # Point the branch ref at the tip. |
| 250 | (muse_dir(root) / "refs" / "heads" / "main").write_text( |
| 251 | chain[-1].commit_id + "\n" |
| 252 | ) |
| 253 | |
| 254 | result = get_commits_for_branch(root, "test-repo", "main") |
| 255 | |
| 256 | ids = [c.commit_id for c in result] |
| 257 | expected = [c.commit_id for c in reversed(chain)] |
| 258 | assert ids == expected, "Commits must be newest-first" |
| 259 | |
| 260 | |
| 261 | # --------------------------------------------------------------------------- |
| 262 | # P5-7 Behavioural — get_commits_for_branch respects max_count |
| 263 | # --------------------------------------------------------------------------- |
| 264 | |
| 265 | def test_p5_7_get_commits_for_branch_max_count( |
| 266 | tmp_path: pathlib.Path, |
| 267 | monkeypatch: pytest.MonkeyPatch, |
| 268 | ) -> None: |
| 269 | """get_commits_for_branch stops after max_count commits.""" |
| 270 | from muse.core.store import get_commits_for_branch |
| 271 | |
| 272 | root = _repo(tmp_path, monkeypatch) |
| 273 | chain = _linear_chain(root, 5) |
| 274 | |
| 275 | (muse_dir(root) / "refs" / "heads" / "main").write_text( |
| 276 | chain[-1].commit_id + "\n" |
| 277 | ) |
| 278 | |
| 279 | result = get_commits_for_branch(root, "test-repo", "main", max_count=2) |
| 280 | |
| 281 | assert len(result) == 2 |
| 282 | assert result[0].commit_id == chain[-1].commit_id # newest first |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
125 days ago