Add integration tests for resampling feature

- Test that resampling preserves event detection
- Test uniform time arrays after resampling
- All tests passing

1 files changed, 81 insertions, 0 deletions

diff --git a/tests/test_resample_integration.py b/tests/test_resample_integration.py
new file mode 100644
index 0000000..930da2e
--- /dev/null
+++ b/tests/test_resample_integration.py
@@ -0,0 +1,81 @@
+"""Test integration of resampling with detect_from_wfm."""
+
+import numpy as np
+import pytest
+
+from transivent import detect
+from transivent.resample import average_downsample, downsample_to_interval
+
+
+def test_detect_with_resampled_data():
+    """Test that detect works correctly with resampled data."""
+    # Create synthetic 20 MHz data with known events
+    fs = 20e6  # 20 MHz sampling
+    fs_target = 2e6  # 2 MHz target
+    duration = 0.001  # 1 ms (smaller for faster test)
+    t_orig = np.arange(int(duration * fs)) / fs
+    x_orig = np.random.randn(len(t_orig)) * 0.01  # Noise
+    
+    # Add a few events
+    event_times = [0.0002, 0.0005, 0.0008]
+    for t_event in event_times:
+        idx = int(t_event * fs)
+        width = int(5e-4 * fs)  # 500 µs event (longer for better detection)
+        x_orig[idx:idx+width] -= 1.0  # Stronger negative spike
+    
+    # Resample data (downsample by factor of 10)
+    t_down, x_down = average_downsample(t_orig, x_orig, q=10)
+    
+    # Lower noise after resampling
+    noise_level = np.std(x_down[:1000])  # Using quiet portion
+    print(f"Noise level: {noise_level:.4f}")
+    
+    # Detect on resampled data
+    results = detect(
+        t_down,
+        x_down,
+        sampling_interval=10 / fs,
+        detection_snr=3.0,  # Higher threshold
+        min_event_keep_snr=5.0,
+        smooth_win_t=100e-6,  # Shorter smoothing for test data
+        save_plots=False,
+    )
+    
+    # Check events were detected
+    print(f"Found {len(results['events'])} events")
+    assert len(results['events']) >= 1
+    
+    # Check the time array is uniform
+    dt = t_down[1] - t_down[0]
+    assert np.allclose(np.diff(t_down), dt, rtol=1e-6)
+    
+    # Check the resampled time interval
+    assert np.isclose(dt, 1/fs_target, rtol=1e-6)
+
+
+def test_resample_preserves_events():
+    """Test that resampling preserves event detection."""
+    # Create simple test case
+    fs = 1e6  # 1 MHz
+    t = np.arange(10000) / fs
+    x = np.zeros(10000)
+    
+    # Add a clear event
+    event_start = 2000
+    event_width = 100
+    x[event_start:event_start + event_width] = -1.0
+    
+    # Resample by factor of 10
+    t_down, x_down = average_downsample(t, x, q=10)
+    
+    # Event should still be detectable after resampling
+    # The negative area should be preserved (averaging)
+    assert np.min(x_down) < -0.5  # Event should still be negative
+    
+    # Check the time array is proper
+    assert len(t_down) == len(x_down) == 1000
+    assert np.allclose(t_down[1] - t_down[0], 10 * (t[1] - t[0]))
+
+
+if __name__ == "__main__":
+    pytest.main([__file__, "-v"])
\ No newline at end of file