fixed tests

saqc/funcs/rolling.py

@@ -42,7 +42,9 @@ class RollingMixin:
         """
         Calculate a rolling-window function on the data.
 
-        Note, that the data gets assigned the worst flag present in the original data.
+        Note, that the new data gets assigned the worst flag present in the window it was aggregated from.
+
+        Note, That you also can select multiple fields to get a rolling calculation over those.
 
         Parameters
         ----------
@@ -178,9 +180,24 @@ def _hroll(
     center: bool = True,
     **kwargs,
 ):
+
+    if isinstance(window, str):
+        freq = getFreqDelta(data[field].to_pandas().index)
+        if freq is None:
+            raise ValueError(
+                f"Rolling over more than one column is only supported if either the data has a unitary"
+                f'sampling rate, or window is an Integer. "{window}" was passed and combined {field} '
+                f"index is not unitarily sampled"
+            )
+        else:
+            window = int(np.floor(pd.Timedelta(window) / freq))
+
     views = np.lib.stride_tricks.sliding_window_view(
         data[field].to_pandas(), (window, len(field))
     )
+    f_views = np.lib.stride_tricks.sliding_window_view(
+        pd.DataFrame({f: flags[f] for f in field}), (window, len(field))
+    )
     frame = pd.DataFrame(
         views.reshape(views.shape[0], views.shape[1] * views.shape[2] * views.shape[3])
     )
@@ -191,20 +208,12 @@ def _hroll(
 
     insuff_periods_mask = ~(~frame.isna()).sum(axis=1) >= min_periods
     result[insuff_periods_mask] = np.nan
-    out = pd.Series(np.nan, index=data[field].to_pandas().index)
-    out[window - 1 :] = result
-    if center:
-        out = out.shift(-window // 2)
-
-    f_views = np.lib.stride_tricks.sliding_window_view(
-        pd.DataFrame({f: flags[f] for f in field}), (window, len(field))
-    )
     f_result = f_views.max(axis=(2, 3)).squeeze()
 
     d_out = pd.Series(np.nan, index=data[field].to_pandas().index)
     d_out[window - 1 :] = result
     if center:
-        d_out = d_out.shift(-window // 2)
+        d_out = d_out.shift(-int(np.floor(window / 2)))
 
     f_out = pd.Series(np.nan, index=data[field].to_pandas().index)
     f_out[window - 1 :] = f_result

tests/funcs/test_proc_functions.py

@@ -20,6 +20,27 @@ from saqc.lib.ts_operators import linearInterpolation, polynomialInterpolation
 from tests.fixtures import char_dict, course_3, course_5  # noqa, todo: fix fixtures
 
 
+@pytest.mark.parametrize(
+    ("window", "center", "expected"),
+    [
+        (1, True, [3, 2, 3, 2]),
+        (2, False, [np.nan, 5, 5, 5]),
+        (3, True, [np.nan, 8, 7, np.nan]),
+        ("20min", True, [5, 5, 5, np.nan]),
+    ],
+)
+def test_multivariateRolling(window, center, expected):
+    data = pd.DataFrame(
+        {"a": [1, np.nan, 3, 4], "b": [1, 2, 3, 4], "c": [1, 2, 3, np.nan]},
+        index=pd.date_range("2000", periods=4, freq="10min"),
+    )
+    qc = saqc.SaQC(data)
+    qc = qc.rolling(
+        ["a", "b", "c"], func="count", target="count", window=window, center=center
+    )
+    assert np.array_equal(qc.data["count"].values, expected, equal_nan=True)
+
+
 def test_rollingInterpolateMissing(course_5):
     data, characteristics = course_5(periods=10, nan_slice=[5, 6])
     field = data.columns[0]