improve output variable names from generate_output_path

ocean_data_parser/batch/default-batch-config.yaml

@@ -39,11 +39,11 @@ ioos_qc: {}
 manual_qc: {}
 
 # Outputs
-file_output:
+output:
   path: null
-  source: null
+  file_name: null
   file_preffix: ""
-  file_suffix: _test
+  file_suffix: ""
   output_format: .nc
 
 upload_to_database:

ocean_data_parser/batch/utils.py

@@ -10,12 +10,12 @@
 
 def generate_output_path(
     ds: xarray.Dataset,
-    source: str = None,
     path: Union[str, Path] = None,
-    defaults: dict = None,
+    file_name: str = None,
     file_preffix: str = "",
     file_suffix: str = "",
     output_format: str = ".nc",
+    defaults: dict = None,
 ) -> Path:
     """Generate output path where to save Dataset.
 
@@ -42,10 +42,9 @@ def generate_output_path(
 
     # handle defaults
     original_source = Path(ds.attrs.get("source")) if ds.attrs.get("source") else None
-    if source is None and original_source:
-        source = str(original_source.stem)
-
-    if source is None:
+    if file_name is None and original_source:
+        file_name = str(original_source.stem)
+    elif file_name is None:
         raise RuntimeError("No output source available. Please define source output.")
 
     if path is None and ds.attrs.get("source"):
@@ -83,19 +82,19 @@ def generate_output_path(
 
     # Generate path and file name
     output_path = Path(path.format(**path_generation_inputs))
-    source = source.format(**path_generation_inputs)
+    file_name = file_name.format(**path_generation_inputs)
 
     # Retrieve output_format if given in source
 
-    if "." in source and not output_format:
-        source, output_format = source.rsplit(".", 1)
+    if "." in file_name and not output_format:
+        file_name, output_format = file_name.rsplit(".", 1)
     assert (
         output_format
     ), "Unknown output file format extension: define the format through the path or output_format inputs"
 
     # Generate path
     return Path(output_path) / (
-        f"{file_preffix or ''}{source}{file_suffix or ''}{output_format}"
+        f"{file_preffix or ''}{file_name}{file_suffix or ''}{output_format}"
     )
 
 

tests/test_batch.py

@@ -214,10 +214,28 @@ def test_generate_output_from_source_attribute(self):
         assert isinstance(name, Path)
         assert str(name) == "source_file.nc"
 
-    def test_generate_filename_with_path(self):
+    @pytest.mark.parametrize(
+        "input,expected_path",
+        (
+            ({"path": "output"}, "output/source_file.nc"),
+            ({"file_name": "test"}, "test.nc"),
+            ({"file_preffix": "test_"}, "test_source_file.nc"),
+            ({"file_suffix": "_test"}, "source_file_test.nc"),
+        ),
+    )
+    def test_generate_filename_with_unique_input(self, input, expected_path):
         name = generate_output_path(
             self._get_test_dataset(),
-            source="{organization}_{instrument}_test",
+            **input,
+            output_format=".nc",
+        )
+        assert isinstance(name, Path)
+        assert name == Path(expected_path)
+
+    def test_generate_filename_with_file_name(self):
+        name = generate_output_path(
+            self._get_test_dataset(),
+            file_name="{organization}_{instrument}_test",
             output_format=".nc",
         )
         assert isinstance(name, Path)
@@ -226,7 +244,7 @@ def test_generate_filename_with_path(self):
     def test_generate_filename_with_time(self):
         name = generate_output_path(
             self._get_test_dataset(),
-            source="{organization}_{instrument}_{time_min:%Y%m%d}-{time_max:%Y%m%d}",
+            file_name="{organization}_{instrument}_{time_min:%Y%m%d}-{time_max:%Y%m%d}",
             output_format=".nc",
         )
         assert isinstance(name, Path)
@@ -235,7 +253,7 @@ def test_generate_filename_with_time(self):
     def test_generate_filename_with_variable_attribute(self):
         name = generate_output_path(
             self._get_test_dataset(),
-            source="{organization}_{instrument}_{variable_time_timezone}",
+            file_name="{organization}_{instrument}_{variable_time_timezone}",
             output_format=".nc",
         )
         assert isinstance(name, Path)
@@ -264,7 +282,7 @@ def test_generate_filename_with_prefix_and_suffix(self):
     def test_generate_filename_with_defaults(self):
         name = generate_output_path(
             self._get_test_dataset(),
-            source="test_{missing_global}",
+            file_name="test_{missing_global}",
             defaults={"missing_global": "this-is-the-default"},
         )
         assert str(name) == "test_this-is-the-default.nc"