refactor(tests): split into tests/unit and tests/tools

tests/tools/test_scroll_trainer.py

@@ -0,0 +1,86 @@
+"""Tests for scroll training pipeline."""
+from __future__ import annotations
+
+import json
+import math
+from pathlib import Path
+
+import numpy as np
+import pytest
+
+from tools.scroll.trainer import load_scroll_data, train_scroll, _augment_scroll
+
+
+def _make_synthetic_scroll_trace(mode="target"):
+    """Create a synthetic scroll trace."""
+    distance = {"target": 1500, "fast": 5000, "precise": 400}[mode]
+    direction = "down"
+    start = 2000
+    target = start + distance
+
+    events = []
+    n_events = 20
+    for i in range(n_events):
+        frac = (i + 1) / n_events
+        delta = int(distance / n_events * (1 + 0.2 * np.random.randn()))
+        delta = max(20, delta)
+        t = int(frac * 800 + np.random.normal(0, 10))
+        events.append({"deltaY": delta, "deltaMode": 0, "t": max(0, t)})
+
+    events.sort(key=lambda e: e["t"])
+    events[0]["t"] = 0
+
+    return {
+        "meta": {
+            "mode": mode,
+            "start_scrollY": start,
+            "target_scrollY": target,
+            "end_scrollY": target + 5,
+            "distance": distance,
+            "direction": direction,
+            "duration_ms": events[-1]["t"],
+            "viewport_height": 900,
+        },
+        "events": events,
+    }
+
+
+@pytest.fixture
+def synthetic_scroll_file(tmp_path):
+    traces_path = tmp_path / "scroll_traces.jsonl"
+    lines = []
+    for mode in ["target", "fast", "precise"]:
+        for _ in range(10):
+            lines.append(json.dumps(_make_synthetic_scroll_trace(mode)))
+    traces_path.write_text("\n".join(lines), encoding="utf-8")
+    return traces_path
+
+
+class TestLoadScrollData:
+    def test_returns_correct_shapes(self, synthetic_scroll_file):
+        seq, cond = load_scroll_data(synthetic_scroll_file, seq_len=32)
+        assert seq.shape[1] == 32
+        assert seq.shape[2] == 2  # (delta_norm, log_dt)
+        assert cond.shape[1] == 7
+        assert len(seq) > 0
+
+
+class TestAugment:
+    def test_4x_augmentation(self, synthetic_scroll_file):
+        seq, cond = load_scroll_data(synthetic_scroll_file, seq_len=32)
+        n = len(seq)
+        seq_aug, cond_aug = _augment_scroll(seq, cond)
+        assert len(seq_aug) == n * 4
+
+
+class TestTrainScroll:
+    def test_produces_model_files(self, synthetic_scroll_file, tmp_path):
+        output_dir = tmp_path / "scroll_models"
+        train_scroll(
+            data_path=synthetic_scroll_file,
+            output_dir=output_dir,
+            epochs=3,
+            batch_size=8,
+        )
+        assert (output_dir / "scroll_model.pt").exists()
+        assert (output_dir / "scroll_config.json").exists()