Variable naming refactor, extending fetchers, fixing stage data retrieval in france.

examples/test_australia_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "403213",
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 # Fetch a recent period for testing
 start_date = "2023-10-01"
 end_date = "2024-03-31"
@@ -26,15 +26,15 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker="o",
         )
     else:
         print(f"No data found for {gauge_id}")
 
 plt.xlabel(constants.TIME_INDEX)
-plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
 plt.title(f"Australia River Discharge ({gauge_ids[0]} - Full Time Series)")
 plt.legend()
 plt.grid(True)

examples/test_canada_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "01AD003",
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 
 plt.figure(figsize=(12, 6))
 
@@ -23,7 +23,7 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker=".",
             linestyle="-",
@@ -33,7 +33,7 @@
 
 if "data" in locals() and not data.empty:
     plt.xlabel(constants.TIME_INDEX)
-    plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+    plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
     plt.title(f"Canada River Discharge ({gauge_ids[0]} - Full Time Series)")
     plt.legend()
     plt.grid(True)

examples/test_chile_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "01201005",
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 
 plt.figure(figsize=(12, 6))
 
@@ -23,7 +23,7 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker=".",
             linestyle="-",
@@ -33,7 +33,7 @@
 
 if "data" in locals() and not data.empty:
     plt.xlabel(constants.TIME_INDEX)
-    plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+    plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
     plt.title(f"Chile River Discharge ({gauge_ids[0]} - Full Time Series)")
     plt.legend()
     plt.grid(True)

examples/test_france_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "K027401001",
 ]
-variable = constants.DISCHARGE
+variable = constants.STAGE_DAILY_MAX
 
 plt.figure(figsize=(12, 6))
 
@@ -23,7 +23,7 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[variable],
             label=gauge_id,
             marker=".",
             linestyle="-",
@@ -34,7 +34,7 @@
 
 if not data.empty:
     plt.xlabel(constants.TIME_INDEX)
-    plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+    plt.ylabel(f"{variable} (m3/s)")
     plt.title(f"France River Discharge ({gauge_ids[0]} - Full Time Series)")
     plt.legend()
     plt.grid(True)

examples/test_japan_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "301011281104010",
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 start_date = "2019-01-01"
 end_date = "2019-12-31"  # Fetching a few months to test
 
@@ -25,15 +25,15 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker="o",
         )
     else:
         print(f"No data found for {gauge_id}")
 
 plt.xlabel(constants.TIME_INDEX)
-plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
 plt.title("Japan River Discharge - Full Time Series")
 plt.legend()
 plt.grid(True)

examples/test_poland_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "149180020",  # CHAŁUPKI on Odra
 ]
-variables = [constants.DISCHARGE, constants.STAGE, constants.WATER_TEMPERATURE]
+variables = [constants.DISCHARGE_DAILY_MEAN, constants.STAGE_DAILY_MEAN, constants.WATER_TEMPERATURE_DAILY_MEAN]
 # Fetch a period from Simon's example
 start_date = "1999-01-01"
 end_date = "2001-12-31"

examples/test_slovenia_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "1020",  # Cmurek on Mura
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 
 plt.figure(figsize=(12, 6))
 
@@ -23,7 +23,7 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker="o",
         )
@@ -32,7 +32,7 @@
 
 if "data" in locals() and not data.empty:
     plt.xlabel(constants.TIME_INDEX)
-    plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+    plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
     plt.title(f"Slovenia River Discharge ({gauge_ids[0]} - Full Time Series)")
     plt.legend()
     plt.grid(True)

examples/test_southafrica_fetcher.py

@@ -9,21 +9,21 @@
 gauge_ids = [
     "X3H023",
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 
 plt.figure(figsize=(12, 6))
 
 fetcher = SouthAfricaFetcher()
 for gauge_id in gauge_ids:
     print(f"Fetching data for {gauge_id}...")
-    data = fetcher.get_data(gauge_id=gauge_id, variable=variable)
+    data = fetcher.get_data(gauge_id=gauge_id, variable=variable, start_date="2000-01-01", end_date="2005-12-31")
     if not data.empty:
         print(f"Data for {gauge_id}:")
         print(data.head())
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker=".",
             linestyle="-",
@@ -33,7 +33,7 @@
 
 if "data" in locals() and not data.empty:
     plt.xlabel(constants.TIME_INDEX)
-    plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+    plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
     plt.title(f"South Africa River Discharge ({gauge_ids[0]} - Full Time Series)")
     plt.legend()
     plt.grid(True)

examples/test_uk_ea_fetcher.py

@@ -7,7 +7,7 @@
 ]
 start_date = "2024-01-01"
 end_date = "2024-01-31"
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 
 plt.figure(figsize=(12, 6))
 
@@ -20,14 +20,14 @@
         print(data.head())
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id.split("/")[-1],
         )
     else:
         print(f"No data found for {gauge_id}")
 
 plt.xlabel(constants.TIME_INDEX)
-plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
 plt.title("UK River Discharge - Full Time Series")
 plt.legend()
 plt.grid(True)

examples/test_uk_nrfa_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "1001",  # Sample gauge from issue #34
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 start_date = "2022-01-01"
 end_date = "2022-01-31"
 
@@ -34,7 +34,7 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker=".",
             linestyle="-",
@@ -44,7 +44,7 @@
 
 if "data" in locals() and not data.empty:
     plt.xlabel(constants.TIME_INDEX)
-    plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+    plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
     plt.title(f"UK NRFA River Discharge ({gauge_ids[0]} - {start_date} to {end_date})")
     plt.legend()
     plt.grid(True)

examples/test_usa_fetcher.py

@@ -9,7 +9,7 @@
 gauge_ids = [
     "07374000",
 ]
-variable = constants.DISCHARGE
+variable = constants.DISCHARGE_DAILY_MEAN
 # Fetch a recent period for testing
 start_date = "1950-01-01"
 end_date = None
@@ -26,15 +26,15 @@
         print(f"Time series from {data.index.min()} to {data.index.max()}")
         plt.plot(
             data.index,
-            data[constants.DISCHARGE],
+            data[constants.DISCHARGE_DAILY_MEAN],
             label=gauge_id,
             marker="o",
         )
     else:
         print(f"No data found for {gauge_id}")
 
 plt.xlabel(constants.TIME_INDEX)
-plt.ylabel(f"{constants.DISCHARGE} (m3/s)")
+plt.ylabel(f"{constants.DISCHARGE_DAILY_MEAN} (m3/s)")
 plt.title(f"USA River Discharge ({gauge_ids[0]} - 1950 to Present)")
 plt.legend()
 plt.grid(True)

rivretrieve/australia.py

@@ -25,7 +25,7 @@ def get_gauge_ids() -> pd.DataFrame:
 
     @staticmethod
     def get_available_variables() -> tuple[str, ...]:
-        return (constants.DISCHARGE, constants.STAGE)
+        return (constants.DISCHARGE_DAILY_MEAN, constants.STAGE_DAILY_MEAN)
 
     def _make_bom_request(self, params: Dict[str, Any]) -> Any:
         """Helper function to make requests to the BoM API."""
@@ -51,10 +51,10 @@ def _make_bom_request(self, params: Dict[str, Any]) -> Any:
 
     def _get_timeseries_id(self, gauge_id: str, variable: str) -> Optional[str]:
         """Retrieves the timeseries ID for the given site and variable."""
-        if variable == constants.STAGE:
+        if variable == constants.STAGE_DAILY_MEAN:
             bom_variable = "Water Course Level"
             # ts_name = "H.Merged.DailyMean"
-        elif variable == constants.DISCHARGE:
+        elif variable == constants.DISCHARGE_DAILY_MEAN:
             bom_variable = "Water Course Discharge"
             # ts_name = "Q.Merged.DailyMean"
         else:

rivretrieve/canada.py

@@ -33,7 +33,7 @@ def get_gauge_ids() -> pd.DataFrame:
 
     @staticmethod
     def get_available_variables() -> tuple[str, ...]:
-        return (constants.DISCHARGE, constants.STAGE)
+        return (constants.DISCHARGE_DAILY_MEAN, constants.STAGE_DAILY_MEAN)
 
     def _find_latest_hydat_link(self) -> Optional[str]:
         s = utils.requests_retry_session()
@@ -130,8 +130,8 @@ def get_data(
             raise ValueError(f"Unsupported variable: {variable}")
 
         var_map = {
-            constants.DISCHARGE: {"table": "DLY_FLOWS", "prefix": "FLOW"},
-            constants.STAGE: {"table": "DLY_LEVELS", "prefix": "LEVEL"},
+            constants.DISCHARGE_DAILY_MEAN: {"table": "DLY_FLOWS", "prefix": "FLOW"},
+            constants.STAGE_DAILY_MEAN: {"table": "DLY_LEVELS", "prefix": "LEVEL"},
         }
 
         table = var_map[variable]["table"]

rivretrieve/chile.py

@@ -24,7 +24,7 @@ def get_gauge_ids() -> pd.DataFrame:
 
     @staticmethod
     def get_available_variables() -> tuple[str, ...]:
-        return (constants.DISCHARGE,)
+        return (constants.DISCHARGE_DAILY_MEAN,)
 
     def _download_data(
         self,
@@ -33,9 +33,8 @@ def _download_data(
         start_date: str,
         end_date: str,
     ) -> Optional[pd.DataFrame]:
-        """Downloads the raw CSV data by first finding the download link."""
-        if variable != constants.DISCHARGE:
-            logger.warning(f"ChileFetcher only supports variable='{constants.DISCHARGE}'")
+        if variable != constants.DISCHARGE_DAILY_MEAN:
+            logger.warning(f"ChileFetcher only supports variable='{constants.DISCHARGE_DAILY_MEAN}'")
             return None
 
         # This long URL was extracted from the R code
@@ -118,8 +117,8 @@ def get_data(
         end_date: Optional[str] = None,
     ) -> pd.DataFrame:
         """Fetches and parses Chilean river gauge data."""
-        if variable != constants.DISCHARGE:
-            logger.warning(f"ChileFetcher only supports variable='{constants.DISCHARGE}'")
+        if variable != constants.DISCHARGE_DAILY_MEAN:
+            logger.warning(f"ChileFetcher only supports variable='{constants.DISCHARGE_DAILY_MEAN}'")
             return pd.DataFrame(columns=[constants.TIME_INDEX, variable])
 
         start_date = utils.format_start_date(start_date)