do not define units if already present

ecmwf-projects · Dec 30, 2024 · 9bdf939 · 9bdf939
1 parent 7597fde
commit 9bdf939
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 10 deletions.
diff --git a/cdsobs/api.py b/cdsobs/api.py
@@ -8,8 +8,8 @@
 from sqlalchemy.orm import Session
 
 from cdsobs.cdm.api import (
-    apply_unit_changes,
     check_cdm_compliance,
+    define_units,
 )
 from cdsobs.config import CDSObsConfig, DatasetConfig
 from cdsobs.ingestion.api import (
@@ -294,11 +294,12 @@ def _read_homogenise_and_partition(
     # Check CDM compliance
     check_cdm_compliance(homogenised_data, dataset_metadata.cdm_tables)
     # Apply unit changes
-    homogenised_data = apply_unit_changes(
-        homogenised_data,
-        service_definition.sources[source],
-        dataset_metadata.cdm_code_tables["observed_variable"],
-    )
+    if "units" not in homogenised_data.columns:
+        homogenised_data = define_units(
+            homogenised_data,
+            service_definition.sources[source],
+            dataset_metadata.cdm_code_tables["observed_variable"],
+        )
     year = time_space_batch.time_batch.year
     lon_tile_size = dataset_config.get_tile_size("lon", source, year)
     lat_tile_size = dataset_config.get_tile_size("lat", source, year)

diff --git a/cdsobs/cdm/api.py b/cdsobs/cdm/api.py
@@ -328,7 +328,7 @@ def check_cdm_compliance(
     return table_field_mappings
 
 
-def apply_unit_changes(
+def define_units(
     homogenised_data: pandas.DataFrame,
     source_definition: SourceDefinition,
     cdm_variable_table: CDMCodeTable,

diff --git a/tests/test_cdm_api.py b/tests/test_cdm_api.py
@@ -1,8 +1,8 @@
 import pytest
 
 from cdsobs.cdm.api import (
-    apply_unit_changes,
     check_cdm_compliance,
+    define_units,
     read_cdm_code_tables,
 )
 from cdsobs.cdm.tables import read_cdm_tables
@@ -49,13 +49,13 @@ def test_apply_variable_unit_change(test_config):
     )
     source_definition = service_definition.sources[source]
     cdm_code_tables = read_cdm_code_tables(test_config.cdm_tables_location)
-    actual = apply_unit_changes(
+    actual = define_units(
         homogenised_data, source_definition, cdm_code_tables["observed_variable"]
     )
     assert "original_units" in actual and "units" in actual and len(actual) > 0
 
     with pytest.raises(RuntimeError):
         source_definition.descriptions["geopotential_height"].units = "wrong"
-        apply_unit_changes(
+        define_units(
             homogenised_data, source_definition, cdm_code_tables["observed_variable"]
         )