feat(v0.2.0): data pipeline

tests/unit/test_data/test_validators.py

@@ -0,0 +1,97 @@
+"""Tests for data validators."""
+
+import pandas as pd
+import pytest
+
+from src.core.enums import Timeframe
+from src.data.validators import check_continuity, detect_outliers, validate_ohlcv
+
+
+@pytest.fixture
+def valid_ohlcv_data():
+    """Create valid OHLCV DataFrame."""
+    dates = pd.date_range("2024-01-01 03:00", periods=100, freq="1min")
+    return pd.DataFrame(
+        {
+            "timestamp": dates,
+            "open": [100.0 + i * 0.1 for i in range(100)],
+            "high": [100.5 + i * 0.1 for i in range(100)],
+            "low": [99.5 + i * 0.1 for i in range(100)],
+            "close": [100.2 + i * 0.1 for i in range(100)],
+            "volume": [1000] * 100,
+        }
+    )
+
+
+@pytest.fixture
+def invalid_ohlcv_data():
+    """Create invalid OHLCV DataFrame."""
+    dates = pd.date_range("2024-01-01 03:00", periods=10, freq="1min")
+    df = pd.DataFrame(
+        {
+            "timestamp": dates,
+            "open": [100.0] * 10,
+            "high": [99.0] * 10,  # Invalid: high < low
+            "low": [99.5] * 10,
+            "close": [100.2] * 10,
+        }
+    )
+    return df
+
+
+def test_validate_ohlcv_valid(valid_ohlcv_data):
+    """Test validation with valid data."""
+    df = validate_ohlcv(valid_ohlcv_data)
+    assert len(df) == 100
+
+
+def test_validate_ohlcv_invalid(invalid_ohlcv_data):
+    """Test validation with invalid data."""
+    with pytest.raises(Exception):  # Should raise ValidationError
+        validate_ohlcv(invalid_ohlcv_data)
+
+
+def test_validate_ohlcv_missing_columns():
+    """Test validation with missing columns."""
+    df = pd.DataFrame({"timestamp": pd.date_range("2024-01-01", periods=10)})
+    with pytest.raises(Exception):  # Should raise ValidationError
+        validate_ohlcv(df)
+
+
+def test_check_continuity(valid_ohlcv_data):
+    """Test continuity check."""
+    is_continuous, gaps = check_continuity(valid_ohlcv_data, Timeframe.M1)
+    assert is_continuous
+    assert len(gaps) == 0
+
+
+def test_check_continuity_with_gaps():
+    """Test continuity check with gaps."""
+    # Create data with gaps
+    dates = pd.date_range("2024-01-01 03:00", periods=10, freq="1min")
+    # Remove some dates to create gaps
+    dates = dates[[0, 1, 2, 5, 6, 7, 8, 9]]  # Gap between index 2 and 5
+
+    df = pd.DataFrame(
+        {
+            "timestamp": dates,
+            "open": [100.0] * len(dates),
+            "high": [100.5] * len(dates),
+            "low": [99.5] * len(dates),
+            "close": [100.2] * len(dates),
+        }
+    )
+
+    is_continuous, gaps = check_continuity(df, Timeframe.M1)
+    assert not is_continuous
+    assert len(gaps) > 0
+
+
+def test_detect_outliers(valid_ohlcv_data):
+    """Test outlier detection."""
+    # Add an outlier
+    df = valid_ohlcv_data.copy()
+    df.loc[50, "close"] = 200.0  # Extreme value
+
+    outliers = detect_outliers(df, columns=["close"], method="iqr", threshold=3.0)
+    assert outliers["is_outlier"].sum() > 0