Adjust DWD Mosmix and DMO kml reader to parse all parameters

CHANGELOG.rst

@@ -5,6 +5,7 @@ Development
 ***********
 
 - Fix `dropna` argument for DWD Mosmix and DMO
+- Adjust DWD Mosmix and DMO kml reader to parse all parameters
 
 0.84.0 (15.05.2024)
 *******************

wetterdienst/provider/dwd/dmo/api.py

@@ -1084,20 +1084,8 @@ def __init__(self, stations_result: StationsResult) -> None:
         """
         super().__init__(stations_result=stations_result)
 
-        parameter_base = self.sr.stations._parameter_base
-        dataset_accessor = self.sr.stations._dataset_accessor
-
-        parameter_ = []
-        for parameter, dataset in self.sr.parameter:
-            if parameter == dataset:
-                parameter = [par.value for par in parameter_base[dataset_accessor][dataset_accessor]]
-                parameter_.extend(parameter)
-            else:
-                parameter_.append(parameter.value)
-
         self.kml = KMLReader(
             station_ids=self.sr.station_id.to_list(),
-            parameters=parameter_,
             settings=self.sr.stations.settings,
         )
 
@@ -1126,10 +1114,22 @@ def query(self) -> Iterator[ValuesResult]:
         self.stations_counter = 0
         self.stations_collected = []
 
+        parameter_base = self.sr.stations._parameter_base
+        dataset_accessor = self.sr.stations._dataset_accessor
+
+        parameter_names = set()
+        for parameter, dataset in self.sr.parameter:
+            if parameter == dataset:
+                parameter = [par.value for par in parameter_base[dataset_accessor][dataset_accessor]]
+                parameter_names.update(parameter)
+            else:
+                parameter_names.add(parameter.value)
+
         for df in self._collect_station_parameter():
             df = self._tidy_up_df(df)
             station_id = df.get_column(Columns.STATION_ID.value).gather(0).item()
             df = self._organize_df_columns(df, station_id, self.sr.stations.dmo_type)
+            df = df.filter(pl.col(Columns.PARAMETER.value).is_in(parameter_names))
 
             if self.sr.humanize:
                 df = self._humanize(df, hpm)

wetterdienst/provider/dwd/mosmix/access.py

@@ -38,9 +38,8 @@
 class KMLReader:
     """Read DWD XML Weather Forecast File of Type KML."""
 
-    def __init__(self, station_ids: list[str], parameters: list[str], settings: Settings) -> None:
+    def __init__(self, station_ids: list[str], settings: Settings) -> None:
         self.station_ids = station_ids
-        self.parameters = parameters
         self.metadata = {}
         self.timesteps = []
         self.nsmap = None
@@ -169,14 +168,13 @@ def get_forecasts(self) -> Iterator[pl.DataFrame]:
 
             for measurement_item in measurement_list:
                 measurement_parameter = measurement_item.get(f"{{{self.nsmap['dwd']}}}elementName")
-                if measurement_parameter.lower() in self.parameters:
-                    measurement_string = measurement_item.getchildren()[0].text
-                    measurement_values = " ".join(measurement_string.split()).split(" ")
-                    measurement_values = [None if i == "-" else float(i) for i in measurement_values]
-                    assert len(measurement_values) == len(  # noqa:S101
-                        self.timesteps,
-                    ), "Number of time steps does not match number of measurement values"
-                    data_dict[measurement_parameter.lower()] = measurement_values
+                measurement_string = measurement_item.getchildren()[0].text
+                measurement_values = " ".join(measurement_string.split()).split(" ")
+                measurement_values = [None if i == "-" else float(i) for i in measurement_values]
+                assert len(measurement_values) == len(  # noqa:S101
+                    self.timesteps,
+                ), "Number of time steps does not match number of measurement values"
+                data_dict[measurement_parameter.lower()] = measurement_values
 
             station_forecast.clear()
             yield pl.DataFrame(data_dict)

wetterdienst/provider/dwd/mosmix/api.py

@@ -1025,20 +1025,8 @@ def __init__(self, stations_result: StationsResult) -> None:
         """
         super().__init__(stations_result=stations_result)
 
-        parameter_base = self.sr.stations._parameter_base
-        dataset_accessor = self.sr.stations._dataset_accessor
-
-        parameter_ = []
-        for parameter, dataset in self.sr.parameter:
-            if parameter == dataset:
-                parameter = [par.value for par in parameter_base[dataset_accessor][dataset_accessor]]
-                parameter_.extend(parameter)
-            else:
-                parameter_.append(parameter.value)
-
         self.kml = KMLReader(
             station_ids=self.sr.station_id.to_list(),
-            parameters=parameter_,
             settings=self.sr.stations.settings,
         )
 
@@ -1061,10 +1049,22 @@ def query(self) -> Iterator[ValuesResult]:
         self.stations_counter = 0
         self.stations_collected = []
 
+        parameter_base = self.sr.stations._parameter_base
+        dataset_accessor = self.sr.stations._dataset_accessor
+
+        parameter_names = set()
+        for parameter, dataset in self.sr.parameter:
+            if parameter == dataset:
+                parameter = [par.value for par in parameter_base[dataset_accessor][dataset_accessor]]
+                parameter_names.update(parameter)
+            else:
+                parameter_names.add(parameter.value)
+
         for df in self._collect_station_parameter():
             df = self._tidy_up_df(df)
             station_id = df.get_column(Columns.STATION_ID.value).gather(0).item()
             df = self._organize_df_columns(df, station_id, self.sr.stations.mosmix_type)
+            df = df.filter(pl.col(Columns.PARAMETER.value).is_in(parameter_names))
 
             if self.sr.humanize:
                 df = self._humanize(df, hpm)