test_musehub_muse_cli.py
python
| 1 | """Unit tests for muse_cli snapshot hashing utilities. |
| 2 | |
| 3 | muse_cli/snapshot.py provides deterministic ID functions used by test |
| 4 | fixtures across the suite — yet it had no tests of its own. These |
| 5 | tests lock down the hashing contract so that any accidental change to |
| 6 | the algorithm is immediately caught. |
| 7 | """ |
| 8 | from __future__ import annotations |
| 9 | |
| 10 | import hashlib |
| 11 | |
| 12 | import pytest |
| 13 | |
| 14 | from musehub.muse_cli.snapshot import compute_commit_id, compute_snapshot_id |
| 15 | |
| 16 | |
| 17 | class TestComputeSnapshotId: |
| 18 | def test_empty_manifest_is_deterministic(self) -> None: |
| 19 | result = compute_snapshot_id({}) |
| 20 | assert isinstance(result, str) |
| 21 | assert len(result) == 64 # SHA-256 hex |
| 22 | |
| 23 | def test_single_entry(self) -> None: |
| 24 | result = compute_snapshot_id({"tracks/bass.mid": "sha256:abc"}) |
| 25 | assert len(result) == 64 |
| 26 | |
| 27 | def test_same_manifest_same_id(self) -> None: |
| 28 | manifest = {"a": "1", "b": "2"} |
| 29 | assert compute_snapshot_id(manifest) == compute_snapshot_id(manifest) |
| 30 | |
| 31 | def test_insertion_order_does_not_matter(self) -> None: |
| 32 | m1 = {"a": "1", "b": "2"} |
| 33 | m2 = {"b": "2", "a": "1"} |
| 34 | assert compute_snapshot_id(m1) == compute_snapshot_id(m2) |
| 35 | |
| 36 | def test_different_manifests_different_ids(self) -> None: |
| 37 | assert compute_snapshot_id({"a": "1"}) != compute_snapshot_id({"a": "2"}) |
| 38 | assert compute_snapshot_id({"a": "1"}) != compute_snapshot_id({"b": "1"}) |
| 39 | |
| 40 | def test_known_value(self) -> None: |
| 41 | """Regression: algorithm must not change without updating this test. |
| 42 | |
| 43 | Uses the null-byte separator (\\x00) introduced when the CLI migrated |
| 44 | from the old ``|``/``:`` scheme to prevent separator-injection attacks. |
| 45 | """ |
| 46 | _SEP = "\x00" |
| 47 | manifest = {"tracks/piano.mid": "sha256:deadbeef"} |
| 48 | parts = sorted(f"{k}{_SEP}{v}" for k, v in manifest.items()) |
| 49 | payload = _SEP.join(parts).encode() |
| 50 | expected = hashlib.sha256(payload).hexdigest() |
| 51 | assert compute_snapshot_id(manifest) == expected |
| 52 | |
| 53 | def test_multiple_entries_sorted(self) -> None: |
| 54 | _SEP = "\x00" |
| 55 | m = {"z/file": "oid-z", "a/file": "oid-a", "m/file": "oid-m"} |
| 56 | result = compute_snapshot_id(m) |
| 57 | # Manually compute expected value using null-byte separator |
| 58 | parts = sorted(f"{k}{_SEP}{v}" for k, v in m.items()) |
| 59 | expected = hashlib.sha256(_SEP.join(parts).encode()).hexdigest() |
| 60 | assert result == expected |
| 61 | |
| 62 | |
| 63 | class TestComputeCommitId: |
| 64 | def test_returns_sha256_hex(self) -> None: |
| 65 | cid = compute_commit_id([], "snap-id", "init", "2026-01-01T00:00:00+00:00") |
| 66 | assert len(cid) == 64 |
| 67 | |
| 68 | def test_deterministic(self) -> None: |
| 69 | kwargs = dict( |
| 70 | parent_ids=["p1", "p2"], |
| 71 | snapshot_id="snap-abc", |
| 72 | message="feat: add piano", |
| 73 | committed_at_iso="2026-03-01T12:00:00+00:00", |
| 74 | ) |
| 75 | assert compute_commit_id(**kwargs) == compute_commit_id(**kwargs) |
| 76 | |
| 77 | def test_parent_order_does_not_matter(self) -> None: |
| 78 | base = dict(snapshot_id="s", message="m", committed_at_iso="2026-01-01T00:00:00") |
| 79 | id1 = compute_commit_id(parent_ids=["a", "b"], **base) |
| 80 | id2 = compute_commit_id(parent_ids=["b", "a"], **base) |
| 81 | assert id1 == id2 |
| 82 | |
| 83 | def test_different_messages_different_ids(self) -> None: |
| 84 | base = dict(parent_ids=[], snapshot_id="s", committed_at_iso="2026-01-01T00:00:00") |
| 85 | assert compute_commit_id(message="msg-1", **base) != compute_commit_id(message="msg-2", **base) |
| 86 | |
| 87 | def test_different_timestamps_different_ids(self) -> None: |
| 88 | base = dict(parent_ids=[], snapshot_id="s", message="m") |
| 89 | t1 = compute_commit_id(committed_at_iso="2026-01-01T00:00:00", **base) |
| 90 | t2 = compute_commit_id(committed_at_iso="2026-01-02T00:00:00", **base) |
| 91 | assert t1 != t2 |
| 92 | |
| 93 | def test_different_snapshots_different_ids(self) -> None: |
| 94 | base = dict(parent_ids=[], message="m", committed_at_iso="2026-01-01T00:00:00") |
| 95 | assert compute_commit_id(snapshot_id="s1", **base) != compute_commit_id(snapshot_id="s2", **base) |
| 96 | |
| 97 | def test_known_value(self) -> None: |
| 98 | """Regression: algorithm must not change without updating this test.""" |
| 99 | parents = ["parent-a", "parent-b"] |
| 100 | snap = "snapshot-xyz" |
| 101 | msg = "feat: add groove" |
| 102 | ts = "2026-01-01T00:00:00+00:00" |
| 103 | parts = ["|".join(sorted(parents)), snap, msg, ts] |
| 104 | expected = hashlib.sha256("|".join(parts).encode()).hexdigest() |
| 105 | assert compute_commit_id(parent_ids=parents, snapshot_id=snap, message=msg, committed_at_iso=ts) == expected |