Merge pull request #52 from statisticsnorway/ref-tuesday

fixed a reference week check

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ssb-arbmark-fagfunksjoner"
-version = "0.0.17"
+version = "0.0.18"
 description = "SSB Arbeidsmarked og lønn Fag-fellesfunksjoner"
 authors = ["Jan Sebastian Rothe <[email protected]>"]
 license = "MIT"

src/arbmark/functions/reference.py

@@ -118,13 +118,12 @@ def ref_week(
         )
     )
 
-    # Calculate the week numbers using pandas with Monday as the starting day
-    from_weeks = from_dates.dt.isocalendar().week
-    to_weeks = to_dates.dt.isocalendar().week
-    ref_weeks = ref_days.dt.isocalendar().week
+    # Determine the start and end of the reference week
+    start_of_week = ref_days - pd.to_timedelta(ref_days.dt.weekday, unit="d")
+    end_of_week = start_of_week + pd.Timedelta(days=6)
 
-    # Check if any of the weeks between from_dates and to_dates is the reference week
-    result = np.logical_and(from_weeks <= ref_weeks, ref_weeks <= to_weeks)
+    # Check if the date range overlaps with the reference week
+    result = np.logical_and(from_dates <= end_of_week, to_dates >= start_of_week)
 
     # Return the result as a series of boolean values
     return pd.Series(result, dtype="boolean")

tests/test_reference.py

@@ -24,9 +24,13 @@ def test_ref_day_outside_range() -> None:
 
 
 def test_ref_week_within_range() -> None:
-    from_dates = pd.Series(pd.to_datetime(["2023-01-22", "2023-04-01"]))
-    to_dates = pd.Series(pd.to_datetime(["2023-01-31", "2023-04-15"]))
-    expected = pd.Series([True, True])
+    from_dates = pd.Series(
+        pd.to_datetime(["2022-02-20", "2022-04-01", "2022-01-01", "2022-12-01"])
+    )
+    to_dates = pd.Series(
+        pd.to_datetime(["2022-02-28", "2022-04-11", "2022-01-31", "2022-12-31"])
+    )
+    expected = pd.Series([True, True, True, True])
     assert (
         ref_week(from_dates, to_dates) == expected
     ).all(), "Reference week within range test failed"