update testing so that theres only one gpm test file

podaac/subsetter/gpm_cleanup.py

@@ -30,7 +30,7 @@ def compute_new_time_data(time_group, nc_dataset):
     return new_time_list, time_unit_out
 
 
-def change_var_dims(nc_dataset, variables=None):
+def change_var_dims(nc_dataset, variables=None, time_name="_timeMidScan"):
     """
     Go through each variable and get the dimension names from attribute "DimensionNames
     If the name is unique, add it as a dimension to the netCDF4 dataset. Then change the
@@ -86,14 +86,14 @@ def change_var_dims(nc_dataset, variables=None):
                 # copy the data to the new variable with dimension names
                 new_mapped_var[var_name][:] = var[:]
 
-    if not any("timeMidScan" in var for var in var_list):
+    if not any(time_name in var for var in var_list):
         # if there isn't any timeMidScan variables, create one
         scan_time_groups = ["__".join(i.split('__')[:-1]) for i in var_list if 'ScanTime' in i]
         for time_group in list(set(scan_time_groups)):
             # get the seconds since Jan 6, 1980
             time_data, time_unit = compute_new_time_data(time_group, nc_dataset)
             # make a new variable for each ScanTime group
-            new_time_var_name = time_group+'__timeMidScan'
+            new_time_var_name = time_group+time_name
             # copy dimensions from the Year variable
             var_dims = nc_dataset.variables[time_group+'__Year'].dimensions
             comp_args = {"zlib": True, "complevel": 1}

tests/test_subset.py

@@ -2287,46 +2287,27 @@ def test_get_unique_groups():
 
     assert expected_groups_single == unique_groups_single
     assert expected_diff_counts_single == diff_counts_single
-
-def test_gpm_dimension_map(data_dir, subset_output_dir, request):
-    """Test GPM files for dimension mapping and returns the expected netCDF
-       dataset without the phony dimensions"""
 
-    gpm_dir = join(data_dir, 'GPM')
-    gpm_file = 'GPM_test_file.HDF5'
-    bbox = np.array(((-180, 180), (-90, 90)))
-    shutil.copyfile(
-        os.path.join(gpm_dir, gpm_file),
-        os.path.join(subset_output_dir, gpm_file)
-    )
-
-    nc_dataset, has_groups, file_extension = subset.open_as_nc_dataset(join(subset_output_dir, gpm_file))
-
-    nc_dataset = gc.change_var_dims(nc_dataset)
-
-    for var_name, var in nc_dataset.variables.items():
-        dims = list(var.dimensions)
-
-        for dim in dims:
-            assert 'phony' not in dim
 
 def test_gpm_compute_new_var_data(data_dir, subset_output_dir, request):
     """Test GPM files that have scantime variable to compute the time for seconds
     since 1980-01-06"""
 
     gpm_dir = join(data_dir, 'GPM')
     gpm_file = 'GPM_test_file_2.HDF5'
-    bbox = np.array(((-180, 180), (-90, 90)))
     shutil.copyfile(
         os.path.join(gpm_dir, gpm_file),
         os.path.join(subset_output_dir, gpm_file)
     )
 
     nc_dataset, has_groups, file_extension = subset.open_as_nc_dataset(join(subset_output_dir, gpm_file))
 
-    del nc_dataset.variables["__FS__ScanTime__timeMidScan"]
-    del nc_dataset.variables["__HS__ScanTime__timeMidScan"]
 
-    nc_dataset = gc.change_var_dims(nc_dataset, variables=None)
+    nc_dataset_new = gc.change_var_dims(nc_dataset, variables=None, time_name='__test_time')
+    assert int(nc_dataset_new.variables["__FS__ScanTime__test_time"][:][0]) == 1306403820
 
-    assert int(nc_dataset["__FS__ScanTime__timeMidScan"][:][0]) == 1306403820
+    for var_name, var in nc_dataset.variables.items():
+        dims = list(var.dimensions)
+
+        for dim in dims:
+            assert 'phony' not in dim