| 1 | """CLI tests for ``sway compare`` (S11).""" |
| 2 | |
| 3 | from __future__ import annotations |
| 4 | |
| 5 | import json |
| 6 | from pathlib import Path |
| 7 | |
| 8 | from typer.testing import CliRunner |
| 9 | |
| 10 | from dlm_sway.cli.app import app |
| 11 | |
| 12 | |
| 13 | def _write_run(path: Path, *, timestamp: str, score: float, probe_score: float) -> None: |
| 14 | """Write a minimal result JSON at ``path``. |
| 15 | |
| 16 | The payload only needs the fields ``report.from_json`` and ``compare`` |
| 17 | actually read — probes (name + score), overall score, timestamp. |
| 18 | """ |
| 19 | payload = { |
| 20 | "schema_version": 1, |
| 21 | "sway_version": "0.1.0.dev0", |
| 22 | "base_model_id": "base", |
| 23 | "adapter_id": "adp", |
| 24 | "started_at": timestamp, |
| 25 | "finished_at": timestamp, |
| 26 | "score": { |
| 27 | "overall": score, |
| 28 | "band": "healthy", |
| 29 | "components": {}, |
| 30 | "findings": [], |
| 31 | }, |
| 32 | "probes": [ |
| 33 | { |
| 34 | "name": "dk", |
| 35 | "kind": "delta_kl", |
| 36 | "verdict": "pass", |
| 37 | "score": probe_score, |
| 38 | "message": f"dk score={probe_score}", |
| 39 | }, |
| 40 | ], |
| 41 | } |
| 42 | path.write_text(json.dumps(payload), encoding="utf-8") |
| 43 | |
| 44 | |
| 45 | class TestCompareCli: |
| 46 | def test_two_files_terminal_default(self, tmp_path: Path) -> None: |
| 47 | a = tmp_path / "a.json" |
| 48 | b = tmp_path / "b.json" |
| 49 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 50 | _write_run(b, timestamp="2026-01-02T12:00:00+00:00", score=0.82, probe_score=0.82) |
| 51 | result = CliRunner().invoke(app, ["compare", str(a), str(b)]) |
| 52 | assert result.exit_code == 0, result.stdout |
| 53 | assert "sway compare" in result.stdout |
| 54 | assert "dk" in result.stdout |
| 55 | assert "composite" in result.stdout |
| 56 | |
| 57 | def test_markdown_format(self, tmp_path: Path) -> None: |
| 58 | a = tmp_path / "a.json" |
| 59 | b = tmp_path / "b.json" |
| 60 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 61 | _write_run(b, timestamp="2026-01-02T12:00:00+00:00", score=0.82, probe_score=0.82) |
| 62 | result = CliRunner().invoke(app, ["compare", str(a), str(b), "--format", "md"]) |
| 63 | assert result.exit_code == 0, result.stdout |
| 64 | # Markdown header + table pipes. |
| 65 | assert "# sway compare" in result.stdout |
| 66 | assert "| probe |" in result.stdout |
| 67 | |
| 68 | def test_json_format(self, tmp_path: Path) -> None: |
| 69 | a = tmp_path / "a.json" |
| 70 | b = tmp_path / "b.json" |
| 71 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 72 | _write_run(b, timestamp="2026-01-02T12:00:00+00:00", score=0.82, probe_score=0.82) |
| 73 | result = CliRunner().invoke(app, ["compare", str(a), str(b), "--format", "json"]) |
| 74 | assert result.exit_code == 0, result.stdout |
| 75 | parsed = json.loads(result.stdout) |
| 76 | assert parsed["labels"] == ["a", "b"] |
| 77 | assert "dk" in parsed["scores"] |
| 78 | |
| 79 | def test_fewer_than_two_files_exits_2(self, tmp_path: Path) -> None: |
| 80 | a = tmp_path / "a.json" |
| 81 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 82 | result = CliRunner().invoke(app, ["compare", str(a)]) |
| 83 | assert result.exit_code == 2 |
| 84 | assert "at least two" in result.stderr + result.stdout |
| 85 | |
| 86 | def test_unreadable_file_exits_2(self, tmp_path: Path) -> None: |
| 87 | a = tmp_path / "a.json" |
| 88 | b = tmp_path / "b.json" |
| 89 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 90 | b.write_text("{ not valid json", encoding="utf-8") |
| 91 | result = CliRunner().invoke(app, ["compare", str(a), str(b)]) |
| 92 | assert result.exit_code == 2 |
| 93 | |
| 94 | def test_fail_on_regression_exits_0_when_improving(self, tmp_path: Path) -> None: |
| 95 | a = tmp_path / "a.json" |
| 96 | b = tmp_path / "b.json" |
| 97 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 98 | _write_run(b, timestamp="2026-01-02T12:00:00+00:00", score=0.85, probe_score=0.85) |
| 99 | result = CliRunner().invoke( |
| 100 | app, ["compare", str(a), str(b), "--fail-on-regression", "0.10"] |
| 101 | ) |
| 102 | assert result.exit_code == 0, result.stdout |
| 103 | |
| 104 | def test_fail_on_regression_exits_1_when_probe_drops(self, tmp_path: Path) -> None: |
| 105 | a = tmp_path / "a.json" |
| 106 | b = tmp_path / "b.json" |
| 107 | # dk dropped 0.20 — above the 0.10 threshold. |
| 108 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 109 | _write_run(b, timestamp="2026-01-02T12:00:00+00:00", score=0.60, probe_score=0.60) |
| 110 | result = CliRunner().invoke( |
| 111 | app, ["compare", str(a), str(b), "--fail-on-regression", "0.10"] |
| 112 | ) |
| 113 | assert result.exit_code == 1 |
| 114 | |
| 115 | def test_fail_on_regression_zero_disables_gate(self, tmp_path: Path) -> None: |
| 116 | a = tmp_path / "a.json" |
| 117 | b = tmp_path / "b.json" |
| 118 | # Drop ≥0.10 but threshold=0 → no gate. |
| 119 | _write_run(a, timestamp="2026-01-01T12:00:00+00:00", score=0.80, probe_score=0.80) |
| 120 | _write_run(b, timestamp="2026-01-02T12:00:00+00:00", score=0.50, probe_score=0.50) |
| 121 | result = CliRunner().invoke(app, ["compare", str(a), str(b), "--fail-on-regression", "0"]) |
| 122 | assert result.exit_code == 0, result.stdout |