Merge pull request #93 from cognitedata/making_indsl_user_warnings

feat: adding specific IndslUserWarning [AH-2555]

indsl/equipment/volume_vessel.py

@@ -10,6 +10,7 @@
 from indsl.exceptions import FLUIDS_REQUIRED
 from indsl.resample.auto_align import auto_align
 from indsl.type_check import check_types
+from indsl.warnings import IndslUserWarning
 
 
 NUMBA_DISABLED = os.environ.get("NUMBA_DISABLE_JIT") == "1"
@@ -29,7 +30,8 @@
             import fluids.vectorized as fv
         except ImportError as e:
             warnings.warn(
-                f"Couldn't import fluids.numba_vectorized: {e!s}. Default to import fluids.vectorized.", UserWarning
+                f"Couldn't import fluids.numba_vectorized: {e!s}. Default to import fluids.vectorized.",
+                category=IndslUserWarning,
             )
             fv = None  # Only core dependencies available. Will raise an error later
 

indsl/resample/interpolate.py

@@ -12,6 +12,7 @@
 from indsl.exceptions import UserValueError
 from indsl.ts_utils.ts_utils import functional_mean, is_na_all
 from indsl.type_check import check_types
+from indsl.warnings import IndslUserWarning
 
 from . import interpolate_v1  # noqa
 
@@ -65,7 +66,7 @@ def interpolate(
     """
     # Check if all values are NaN
     if is_na_all(data):
-        warnings.warn("All values in the time series are NaN.", UserWarning)
+        warnings.warn("All values in the time series are NaN.", category=IndslUserWarning)
         return data
 
     # Allow for other ways of defining forward filling for stepwise functions
@@ -92,7 +93,7 @@ def interpolate(
 
     # Check for empty time series
     if len(observations) < 2:
-        warnings.warn("The time series contains less than two values.", UserWarning)
+        warnings.warn("The time series contains less than two values.", category=IndslUserWarning)
         return data
 
     # x and y datapoints used to construct linear piecewise function

indsl/resample/resample.py

@@ -13,6 +13,7 @@
 from indsl.ts_utils.ts_utils import fill_gaps, get_fixed_freq, is_na_all
 from indsl.type_check import check_types
 from indsl.validations import validate_series_has_time_index, validate_series_is_not_empty
+from indsl.warnings import IndslUserWarning
 
 from . import resample_v1  # noqa
 
@@ -102,7 +103,8 @@ def resample(
         if not granularity_current:
             # TODO: pick max resolution and apply to the rest of the timeseries?
             warnings.warn(
-                "Can't infer time series resolution with missing data. Please provide resolution", UserWarning
+                "Can't infer time series resolution with missing data. Please provide resolution",
+                category=IndslUserWarning,
             )
             return data
 

indsl/resample/resample_v1.py

@@ -14,6 +14,7 @@
 from indsl.ts_utils.ts_utils import fill_gaps, get_fixed_freq, is_na_all
 from indsl.type_check import check_types
 from indsl.validations import validate_series_has_time_index
+from indsl.warnings import IndslUserWarning
 
 
 @versioning.register(version="1.0", deprecated=True)
@@ -102,7 +103,8 @@ def resample(
         if not granularity_current:
             # TODO: pick max resolution and apply to the rest of the timeseries?
             warnings.warn(
-                "Can't infer time series resolution with missing data. Please provide resolution", UserWarning
+                "Can't infer time series resolution with missing data. Please provide resolution",
+                category=IndslUserWarning,
             )
             return data
 

indsl/signals/generator.py

@@ -11,6 +11,7 @@
 from indsl.ts_utils.utility_functions import TimeUnits
 from indsl.type_check import check_types
 from indsl.validations import validate_series_has_time_index
+from indsl.warnings import IndslUserWarning
 
 
 @check_types
@@ -574,7 +575,9 @@ def _make_index(
     # Catch strange sample frequency inputs
     if not isinstance(freq, pd.Timedelta):
         freq = pd.Timedelta(1, "m")
-        warnings.warn("Can't recognize the sample frequency, setting it to the '1 m' default.")
+        warnings.warn(
+            "Can't recognize the sample frequency, setting it to the '1 m' default.", category=IndslUserWarning
+        )
 
     if freq.total_seconds() <= 0:
         raise UserValueError(

indsl/statistics/outliers_v1.py

@@ -10,6 +10,7 @@
 from indsl.exceptions import CSAPS_REQUIRED, KNEED_REQUIRED, SCIKIT_LEARN_REQUIRED, UserValueError
 from indsl.type_check import check_types
 from indsl.validations import validate_series_has_time_index, validate_series_is_not_empty
+from indsl.warnings import IndslUserWarning
 
 
 @versioning.register(version="1.0", deprecated=True)
@@ -220,7 +221,7 @@ def _get_outlier_indices(
     if str.isdigit(time_window):  # if user gives '60' it will be considered as '60min'
         warnings.warn(
             f"Missing time unit in argument 'time_window' in remove_outliers function, assuming {time_window}min.",
-            UserWarning,
+            category=IndslUserWarning,
         )
         time_window = str(time_window) + "min"
 

indsl/ts_utils/ts_utils.py

@@ -11,6 +11,7 @@
 
 from indsl.exceptions import UserRuntimeError, UserValueError
 from indsl.type_check import check_types
+from indsl.warnings import IndslUserWarning
 
 
 _unit_in_ms_without_week = {"s": 1000, "m": 60000, "h": 3600000, "d": 86400000}
@@ -292,8 +293,8 @@ def time_parse(time_window: str = "60min", function_name: str = "") -> pd.Timede
     """
     if str.isdigit(time_window):  # if user gives '60' it will be considered as '60min'
         warnings.warn(
-            f"Missing time unit in argument 'time_window' in {function_name} function, assuming {time_window}min.",
-            UserWarning,
+            f"Missing time unit in argument 'time_window' in {function_name} function, assuming {time_window} min.",
+            category=IndslUserWarning,
         )
         time_window = str(time_window) + "min"
     try:

indsl/warnings.py

@@ -0,0 +1,2 @@
+class IndslUserWarning(UserWarning):
+    """Warning that will be shown to the user."""

tests/resample/test_interpolate.py

@@ -4,6 +4,7 @@
 import pytest
 
 from indsl.resample.interpolate import interpolate
+from indsl.warnings import IndslUserWarning
 
 from ..generate_data import create_non_uniform_data, create_uniform_data, set_na_random_data
 
@@ -17,14 +18,14 @@
     ],
 )
 def test_too_few_data_warnings(input):
-    with pytest.warns(UserWarning):
+    with pytest.warns(IndslUserWarning):
         interpolated_data = interpolate(input)
         assert len(interpolated_data) == len(input)
 
 
 # Test for all NaN data
 def test_all_nan_data():
-    with pytest.warns(UserWarning):
+    with pytest.warns(IndslUserWarning):
         test_data = create_uniform_data(np.ones(10) * np.nan)
         interpolated_data = interpolate(test_data)
         assert len(interpolated_data) == len(test_data)

tests/signals/test_signal_generator.py

@@ -16,7 +16,7 @@
     perturb_timestamp,
     sine_wave,
 )
-
+from indsl.warnings import IndslUserWarning
 
 test_data_4index = [
     ("1975/5/9", "1975-5-10", pd.Timedelta("1s"), 1.0, pd.Timedelta("1 day")),
@@ -33,7 +33,7 @@
 
 @pytest.mark.parametrize("bad_rate", [(-99999), (None)])
 def test_wrong_sample_rate(bad_rate):
-    with pytest.warns(UserWarning):
+    with pytest.warns(IndslUserWarning):
         _make_index(freq=bad_rate)
 
 
@@ -65,7 +65,7 @@ def test_index_generator(start_date, end_date, sample_rate, expected_mean, expec
     test_data_warns,
 )
 def test_sample_freq_warns(start_date, end_date, sample_rate, expected_mean, expected_dt):
-    with pytest.warns(UserWarning):
+    with pytest.warns(IndslUserWarning):
         # Test that the generated DatetimeIndex has the correct sampling frequency and duration based on input parameters
         idx = _make_index(start=start_date, end=end_date, freq=sample_rate)
         # Convert to time, where t=0 is idx[0]