Bump polars to >=1.0.0

earthobservations · Sep 10, 2024 · 43248f7 · 43248f7
1 parent c1c51d9
commit 43248f7
Show file tree

Hide file tree

Showing 38 changed files with 219 additions and 193 deletions.
diff --git a/CHANGELOG.rst b/CHANGELOG.rst
@@ -4,6 +4,8 @@ Changelog
 Development
 ***********
 
+- Bump polars to `>=1.0.0`
+
 0.95.1 (04.09.2024)
 *******************
 

diff --git a/pyproject.toml b/pyproject.toml
@@ -100,7 +100,7 @@ dependencies = [
   "pillow>=10.3",
   "pint>=0.17",
   "platformdirs",
-  "polars>=0.20.10",
+  "polars>=1.0.0",
   "portion>=2.4.2",
   "pyarrow>=13",
   "pydantic>=2.7.3",

diff --git a/tests/core/timeseries/test_interpolation.py b/tests/core/timeseries/test_interpolation.py
@@ -24,12 +24,12 @@
 def df_interpolated_empty():
     return pl.DataFrame(
         schema={
-            Columns.STATION_ID.value: pl.Utf8,
-            Columns.PARAMETER.value: pl.Utf8,
+            Columns.STATION_ID.value: pl.String,
+            Columns.PARAMETER.value: pl.String,
             Columns.DATE.value: pl.Datetime(time_zone="UTC"),
             Columns.VALUE.value: pl.Float64,
             Columns.DISTANCE_MEAN.value: pl.Float64,
-            Columns.TAKEN_STATION_IDS.value: pl.List(pl.Utf8),
+            Columns.TAKEN_STATION_IDS.value: pl.List(pl.String),
         },
     )
 

diff --git a/tests/core/timeseries/test_io.py b/tests/core/timeseries/test_io.py
@@ -153,9 +153,9 @@ def df_values():
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,
@@ -177,12 +177,12 @@ def df_interpolated_values():
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "distance_mean": pl.Float64,
-            "taken_station_ids": pl.List(pl.Utf8),
+            "taken_station_ids": pl.List(pl.String),
         },
     )
 
@@ -201,12 +201,12 @@ def df_summarized_values():
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "distance": pl.Float64,
-            "taken_station_id": pl.Utf8,
+            "taken_station_id": pl.String,
         },
     )
 

diff --git a/tests/core/timeseries/test_request.py b/tests/core/timeseries/test_request.py
@@ -52,13 +52,13 @@ def expected_stations_df():
             ),
         ],
         schema={
-            "station_id": pl.Utf8,
+            "station_id": pl.String,
             "start_date": pl.Datetime(time_zone="UTC"),
             "latitude": pl.Float64,
             "longitude": pl.Float64,
             "height": pl.Float64,
-            "name": pl.Utf8,
-            "state": pl.Utf8,
+            "name": pl.String,
+            "state": pl.String,
             "distance": pl.Float64,
         },
     )

diff --git a/tests/core/timeseries/test_summary.py b/tests/core/timeseries/test_summary.py
@@ -57,12 +57,12 @@ def test_not_summarizable_dataset(default_settings):
     assert given_df.drop_nulls().shape[0] == 0
     expected_df = pl.DataFrame(
         schema={
-            Columns.STATION_ID.value: pl.Utf8,
-            Columns.PARAMETER.value: pl.Utf8,
+            Columns.STATION_ID.value: pl.String,
+            Columns.PARAMETER.value: pl.String,
             Columns.DATE.value: pl.Datetime(time_zone="UTC"),
             Columns.VALUE.value: pl.Float64,
             Columns.DISTANCE.value: pl.Float64,
-            Columns.TAKEN_STATION_ID.value: pl.Utf8,
+            Columns.TAKEN_STATION_ID.value: pl.String,
         },
     )
     assert_frame_equal(

diff --git a/tests/provider/dwd/observation/test_api_data.py b/tests/provider/dwd/observation/test_api_data.py
@@ -220,9 +220,9 @@ def test_dwd_observation_data_result_missing_data(default_settings):
             "quality": [None],
         },
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,
@@ -302,8 +302,8 @@ def test_dwd_observation_data_result_wide_single_dataset(
             "qn_tgk": [None, None],
         },
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "fx": pl.Float64,
             "qn_fx": pl.Float64,
@@ -375,8 +375,8 @@ def test_dwd_observation_data_result_wide_single_parameter(
             "qn_rsk": [None, 1.0],
         },
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "rsk": pl.Float64,
             "qn_rsk": pl.Float64,
@@ -440,8 +440,8 @@ def test_dwd_observation_data_result_wide_si(
             "qn_tgk": [None, None],
         },
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "fx": pl.Float64,
             "qn_fx": pl.Float64,
@@ -659,8 +659,8 @@ def test_dwd_observation_data_result_wide_two_datasets(
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "climate_summary_fx": pl.Float64,
             "qn_climate_summary_fx": pl.Float64,
@@ -952,8 +952,8 @@ def test_dwd_observation_data_result_tidy_si(settings_humanize_false):
             ],
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
             "parameter": pl.Utf8,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,

diff --git a/tests/provider/dwd/observation/test_parser.py b/tests/provider/dwd/observation/test_parser.py
@@ -39,22 +39,22 @@ def test_parse_dwd_data():
                 dt.datetime(1937, 1, 1, tzinfo=ZoneInfo("UTC")),
                 dt.datetime(1986, 6, 30, tzinfo=ZoneInfo("UTC")),
             ],
-            "qn_3": pl.Series(values=[None, None], dtype=pl.Utf8),
-            "fx": pl.Series(values=[None, None], dtype=pl.Utf8),
-            "fm": pl.Series(values=[None, None], dtype=pl.Utf8),
+            "qn_3": pl.Series(values=[None, None], dtype=pl.String),
+            "fx": pl.Series(values=[None, None], dtype=pl.String),
+            "fm": pl.Series(values=[None, None], dtype=pl.String),
             "qn_4": ["5", "10"],
             "rsk": ["0.0", "0.0"],
             "rskf": ["0", "0"],
-            "sdk": pl.Series(values=[None, None], dtype=pl.Utf8),
+            "sdk": pl.Series(values=[None, None], dtype=pl.String),
             "shk_tag": ["0", "0"],
             "nm": ["6.3", "0.3"],
             "vpm": [None, "13.9"],
-            "pm": pl.Series(values=[None, None], dtype=pl.Utf8),
+            "pm": pl.Series(values=[None, None], dtype=pl.String),
             "tmk": ["-0.5", "19.8"],
             "upm": [None, "60.00"],
             "txk": ["2.5", "24.8"],
             "tnk": ["-1.6", "14.4"],
-            "tgk": pl.Series(values=[None, None], dtype=pl.Utf8),
+            "tgk": pl.Series(values=[None, None], dtype=pl.String),
         },
     )
     assert_frame_equal(

diff --git a/tests/provider/imgw/hydrology/test_api.py b/tests/provider/imgw/hydrology/test_api.py
@@ -22,14 +22,14 @@ def df_expected_station():
             "state": None,
         },
         schema={
-            "station_id": pl.Utf8,
+            "station_id": pl.String,
             "start_date": pl.Datetime(time_zone="UTC"),
             "end_date": pl.Datetime(time_zone="UTC"),
             "latitude": pl.Float64,
             "longitude": pl.Float64,
             "height": pl.Float64,
-            "name": pl.Utf8,
-            "state": pl.Utf8,
+            "name": pl.String,
+            "state": pl.String,
         },
     )
 
@@ -71,9 +71,9 @@ def test_imgw_hydrology_api_daily(df_expected_station):
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,
@@ -167,9 +167,9 @@ def test_imgw_hydrology_api_monthly(df_expected_station):
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,

diff --git a/tests/provider/imgw/meteorology/test_api.py b/tests/provider/imgw/meteorology/test_api.py
@@ -29,14 +29,14 @@ def test_imgw_meteorology_api_daily():
             "state": "Drawa",
         },
         schema={
-            "station_id": pl.Utf8,
+            "station_id": pl.String,
             "start_date": pl.Datetime(time_zone="UTC"),
             "end_date": pl.Datetime(time_zone="UTC"),
             "latitude": pl.Float64,
             "longitude": pl.Float64,
             "height": pl.Float64,
-            "name": pl.Utf8,
-            "state": pl.Utf8,
+            "name": pl.String,
+            "state": pl.String,
         },
     )
     assert_frame_equal(request.df, df_expected_station)
@@ -117,9 +117,9 @@ def test_imgw_meteorology_api_daily():
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,
@@ -147,14 +147,14 @@ def test_imgw_meteorology_api_monthly():
             "state": "Bia³a",
         },
         schema={
-            "station_id": pl.Utf8,
+            "station_id": pl.String,
             "start_date": pl.Datetime(time_zone="UTC"),
             "end_date": pl.Datetime(time_zone="UTC"),
             "latitude": pl.Float64,
             "longitude": pl.Float64,
             "height": pl.Float64,
-            "name": pl.Utf8,
-            "state": pl.Utf8,
+            "name": pl.String,
+            "state": pl.String,
         },
     )
     assert_frame_equal(request.df, df_expected_station)
@@ -299,9 +299,9 @@ def test_imgw_meteorology_api_monthly():
             },
         ],
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,

diff --git a/tests/provider/noaa/ghcn/test_api_data.py b/tests/provider/noaa/ghcn/test_api_data.py
@@ -37,9 +37,9 @@ def test_api_amsterdam(start_date, end_date, default_settings):
             "quality": [None],
         },
         schema={
-            "station_id": pl.Utf8,
-            "dataset": pl.Utf8,
-            "parameter": pl.Utf8,
+            "station_id": pl.String,
+            "dataset": pl.String,
+            "parameter": pl.String,
             "date": pl.Datetime(time_zone="UTC"),
             "value": pl.Float64,
             "quality": pl.Float64,