Add Germany-Berlin fetcher updated with ruff formating (Fixes #62)

Author: thiagovmdon

rivretrieve/__init__.py

@@ -7,6 +7,7 @@
 from .chile import ChileFetcher
 from .czech import CzechFetcher
 from .france import FranceFetcher
+from .germany_berlin import GermanyBerlinFetcher
 from .japan import JapanFetcher
 from .poland import PolandFetcher
 from .portugal import PortugalFetcher
@@ -16,6 +17,5 @@
 from .uk_ea import UKEAFetcher
 from .uk_nrfa import UKNRFAFetcher
 from .usa import USAFetcher
-from .germany_berlin import GermanyBerlinFetcher
 
 __version__ = "0.1.0"

rivretrieve/germany_berlin.py

@@ -2,7 +2,7 @@
 
 import logging
 from io import StringIO
-from typing import Any, Dict, List, Optional
+from typing import Optional
 
 import pandas as pd
 import requests
@@ -13,6 +13,7 @@
 
 logger = logging.getLogger(__name__)
 
+
 class GermanyBerlinFetcher(base.RiverDataFetcher):
     """Fetches river gauge data from Wasserportal Berlin.
 
@@ -96,15 +97,15 @@ def get_metadata(self) -> pd.DataFrame:
         except Exception as e:
             logger.error(f"Failed to fetch metadata: {e}")
             return pd.DataFrame(columns=keep_cols).set_index(constants.GAUGE_ID)
-    
+
     @staticmethod
     def get_available_variables() -> tuple[str, ...]:
         return (
             constants.STAGE_DAILY_MEAN,
             constants.DISCHARGE_DAILY_MEAN,
             constants.WATER_TEMPERATURE_DAILY_MEAN,
             constants.STAGE_INSTANT,
-            constants.DISCHARGE_INSTANT
+            constants.DISCHARGE_INSTANT,
         )
 
     def _download_data(
@@ -117,24 +118,21 @@ def _download_data(
         """Downloads CSV data for a gauge and variable."""
         thema_map = {
             # Daily
-            constants.STAGE_DAILY_MEAN: ("ows", "tw"),             # Wasserstand (cm)
-            constants.DISCHARGE_DAILY_MEAN: ("odf", "tw"),         # Durchfluss (m³/s)
-            constants.WATER_TEMPERATURE_DAILY_MEAN: ("owt", "tw"), # Wassertemperatur (°C)
-
+            constants.STAGE_DAILY_MEAN: ("ows", "tw"),  # Wasserstand (cm)
+            constants.DISCHARGE_DAILY_MEAN: ("odf", "tw"),  # Durchfluss (m³/s)
+            constants.WATER_TEMPERATURE_DAILY_MEAN: ("owt", "tw"),  # Wassertemperatur (°C)
             # Instantaneous
             constants.STAGE_INSTANT: ("ows", "ew"),
             constants.DISCHARGE_INSTANT: ("odf", "ew"),
-
         }
 
         if variable not in thema_map:
             raise ValueError(f"Unsupported variable: {variable}")
-        
+
         thema, frequency = thema_map[variable]
         start_date_fmt = pd.to_datetime(start_date).strftime("%d.%m.%Y")
         url = self.BASE_URL.format(id=gauge_id, thema=thema, frequency=frequency, start_date=start_date_fmt)
 
-
         logger.info(f"Fetching {variable} for {gauge_id} from {url}")
         r = requests.get(url, timeout=20)
         r.raise_for_status()
@@ -150,15 +148,17 @@ def _download_data(
         except Exception as e:
             logger.error(f"Error parsing CSV for {gauge_id}: {e}")
             return pd.DataFrame()
-    
+
     def _parse_data(self, gauge_id: str, raw_data: pd.DataFrame, variable: str) -> pd.DataFrame:
         """Parses Wasserportal CSV to standardized DataFrame."""
         if raw_data.empty:
             return pd.DataFrame(columns=[constants.TIME_INDEX, variable])
 
         raw_data.columns = [c.strip().lower() for c in raw_data.columns]
         time_col = next((c for c in raw_data.columns if "datum" in c or "zeit" in c), raw_data.columns[0])
-        val_col = next((c for c in raw_data.columns if c not in [time_col] and raw_data[c].dtype != "O"), raw_data.columns[1])
+        val_col = next(
+            (c for c in raw_data.columns if c not in [time_col] and raw_data[c].dtype != "O"), raw_data.columns[1]
+        )
 
         raw_data[constants.TIME_INDEX] = pd.to_datetime(raw_data[time_col], dayfirst=True, errors="coerce")
         raw_data[variable] = pd.to_numeric(raw_data[val_col], errors="coerce")
@@ -173,7 +173,7 @@ def _parse_data(self, gauge_id: str, raw_data: pd.DataFrame, variable: str) -> p
             .set_index(constants.TIME_INDEX)
         )
         return df
-    
+
     def get_data(
         self,
         gauge_id: str,

tests/test_germany_berlin.py

@@ -1,7 +1,8 @@
 import logging
+
 import matplotlib.pyplot as plt
-from rivretrieve import constants
-from rivretrieve import GermanyBerlinFetcher
+
+from rivretrieve import GermanyBerlinFetcher, constants
 
 logging.basicConfig(level=logging.INFO)
 
@@ -53,4 +54,4 @@
 plot_path = "berlin_fetcher_plot.png"
 plt.savefig(plot_path)
 
-#print(fetcher.get_metadata())
+# print(fetcher.get_metadata())