Only download data within requested time range

Author: bouweandela

scripts/fetch_test_data.py

@@ -165,12 +165,29 @@ def process_sample_data_request(
         facets=dict(
             source_id="ACCESS-ESM1-5",
             frequency=["fx", "mon"],
-            variable_id=["areacella", "tas", "tos", "rsut", "rlut", "rsdt"],
+            variable_id=[
+                "areacella",
+                "tas",
+                "tos",
+                "rsut",
+                "rsdt",
+            ],
             experiment_id=["ssp126", "historical"],
         ),
         remove_ensembles=True,
         time_span=("2000", "2025"),
     ),
+    CMIP6Request(
+        id="example-metric",
+        facets=dict(
+            source_id="ACCESS-ESM1-5",
+            frequency=["mon"],
+            variable_id=["rlut"],
+            experiment_id=["historical"],
+        ),
+        remove_ensembles=True,
+        time_span=("2000", "2025"),
+    ),
     # ESMValTool Climate at global warmings levels data
     CMIP6Request(
         id="esmvaltool-climate-at-global-warmings-levels",

src/ref_sample_data/data_request/base.py

@@ -79,9 +79,14 @@ class IntakeESGFDataRequest(DataRequest):
 
     def fetch_datasets(self) -> pd.DataFrame:
         """Fetch the datasets from the ESGF."""
+        facets = dict(self.facets)
+        if self.time_span:
+            facets["file_start"] = self.time_span[0]
+            facets["file_end"] = self.time_span[1]
+
         cat = ESGFCatalog()
 
-        cat.search(**self.facets)
+        cat.search(**facets)
         if self.remove_ensembles:
             cat.remove_ensembles()