Add tests for CLI help output, empty/invalid durations, and

ScenarioGroup radd

Author: Евгений Блинов

tests/cli/test_scenario_cli.py

@@ -120,7 +120,7 @@ def test_help_mentions_number(self) -> None:
     def test_help_mentions_max_mean(self) -> None:
         proc = run_script(scenario_script(), '--help')
         combined = proc.stdout + proc.stderr
-        assert 'max-mean' in combined.lower() or 'max_mean' in combined.lower()
+        assert '--max-mean' in combined
 
     def test_help_does_not_run_benchmark(self) -> None:
         proc = run_script(scenario_script(), '--help')

tests/units/test_benchmark_result.py

@@ -34,14 +34,33 @@ def test_durations_is_tuple(self) -> None:
         result = make_result((0.1, 0.2, 0.3))
         assert isinstance(result.durations, tuple)
 
+    def test_empty_durations_raises(self) -> None:
+        # BenchmarkResult does not validate durations length;
+        # creating with empty tuple causes ZeroDivisionError in __post_init__
+        s = Scenario(lambda: None, name='s', number=1)
+        with pytest.raises((ZeroDivisionError, ValueError)):
+            BenchmarkResult(scenario=s, durations=(), is_primary=True)
+
+    def test_inf_durations_fields(self) -> None:
+        result = make_result((float('inf'), 1.0, 2.0))
+        assert math.isinf(result.worst)
+        assert math.isinf(result.mean)
+        assert result.best == 1.0
+
+    def test_nan_durations_fields(self) -> None:
+        result = make_result((float('nan'),))
+        assert math.isnan(result.mean)
+        assert math.isnan(result.best)
+        assert math.isnan(result.worst)
+
     def test_mean_computed_correctly(self) -> None:
         result = make_result((1.0, 2.0, 3.0))
         expected = math.fsum([1.0, 2.0, 3.0]) / 3
         assert result.mean == expected
 
     def test_mean_uses_fsum_precision(self) -> None:
-        # fsum handles cancellation correctly; plain sum loses precision:
-        # sum([1e20, 1.0, -1e20]) == 0.0, fsum == 1.0
+        # fsum handles cancellation correctly; plain sum loses precision
+        # for (1e20, 1.0, -1e20): fsum=1.0, sum=0.0 due to IEEE 754
         durations = (1e20, 1.0, -1e20)
         result = make_result(durations)
         assert result.mean == pytest.approx(1.0 / 3)
@@ -337,6 +356,12 @@ def test_percentile_single_element(self) -> None:
         assert trimmed.durations == (5.0,)
         assert trimmed.is_primary is False
 
+    def test_percentile_100_single_element(self) -> None:
+        result = make_result((5.0,))
+        trimmed = result.percentile(100)
+        assert trimmed.durations == (5.0,)
+        assert trimmed.is_primary is False
+
     def test_from_json_missing_is_primary_raises(self) -> None:
         payload = json.dumps({'durations': [0.1, 0.2]})
         with pytest.raises(ValueError, match='required fields'):

tests/units/test_scenario.py

@@ -167,12 +167,13 @@ def fn() -> None:
         assert counter[0] == 5
 
     def test_run_uses_custom_timer(self) -> None:
-        # timer produces: 0.000, 0.001, 0.002, 0.003, 0.004, 0.005, ...
+        # timer produces: 0.000, 0.001, 0.002, 0.003, ...  (infinite)
         # each measured interval: end - start = 0.001
-        values = iter(t * 0.001 for t in range(200))
+        import itertools  # noqa: PLC0415
+        counter = itertools.count(0)
 
         def fake_timer() -> float:
-            return next(values)
+            return next(counter) * 0.001
 
         s = Scenario(lambda: None, name='s', number=3, timer=fake_timer)
         result = s.run()

tests/units/test_scenario_group.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+import pytest
+
 from microbenchmark import BenchmarkResult, Scenario, ScenarioGroup
 
 
@@ -80,6 +82,19 @@ def test_radd_scenario_to_group(self) -> None:
         assert isinstance(group, ScenarioGroup)
         assert len(group.run()) == 2
 
+    def test_radd_group_to_group(self) -> None:
+        s1, s2, s3 = make_scenario('s1'), make_scenario('s2'), make_scenario('s3')
+        g1 = ScenarioGroup(s1, s2)
+        g2 = ScenarioGroup(s3)
+        # g2.__radd__(g1) = ScenarioGroup(*g1._scenarios, *g2._scenarios) = [s1, s2, s3]
+        group = g2.__radd__(g1)
+        assert isinstance(group, ScenarioGroup)
+        results = group.run()
+        assert len(results) == 3
+        assert results[0].scenario is s1
+        assert results[1].scenario is s2
+        assert results[2].scenario is s3
+
     def test_duplicate_scenarios(self) -> None:
         s = make_scenario('s')
         group = s + s
@@ -156,3 +171,13 @@ def test_run_warmup_different_numbers(self) -> None:
         results = g.run(warmup=2)
         assert len(results[0].durations) == 3
         assert len(results[1].durations) == 7
+
+    def test_run_propagates_exception_from_scenario(self) -> None:
+        def bad() -> None:
+            raise RuntimeError('scenario failed')
+
+        s1 = make_scenario('s1')
+        s2 = Scenario(bad, name='s2', number=1)
+        g = ScenarioGroup(s1, s2)
+        with pytest.raises(RuntimeError, match='scenario failed'):
+            g.run()