Fixing subdataset step

preprocessor/preprocessor/steps/subdataset.py

 import os
-from os.path import join, splitext, basename
+from os.path import join, splitext, basename, dirname
 from glob import glob
-from typing import Set
+from typing import Dict
 
 from osgeo import gdal
 
 from ..util import replace_ext
 
 
-def extract_subdataset_step(source_dir: os.PathLike, target_dir: os.PathLike, data_file_glob: str, subdataset_types: Set[str]=None):
-    for filename in glob(join(source_dir, data_file_glob)):
+gdal.UseExceptions()
+
+
+def extract_subdataset_step(source_dir: os.PathLike, target_dir: os.PathLike, data_file_glob: str, subdataset_types: Dict[str, str]=None):
+    datafiles = glob(join(source_dir, data_file_glob))
+    if not datafiles:
+        raise Exception('No datafiles were matched by the provided glob')
+
+    for filename in datafiles:
         extract_subdatasets(
             filename,
-            join(target_dir, basename(replace_ext(filename, '.tif'))),
+            target_dir,
             subdataset_types
         )
 
 
-def extract_subdatasets(source_filename: os.PathLike, target_filename: os.PathLike, subdataset_types: Set[str]=None):
+def extract_subdatasets(source_filename: os.PathLike, target_dir: os.PathLike, subdataset_types: Dict[str, str]=None):
     ds = gdal.Open(source_filename)
 
     sub_datasets = []
     for locator, _ in ds.GetSubDatasets():
         _, _, sd_type = locator.split(':')
         if subdataset_types is None or sd_type in subdataset_types:
-            sub_datasets.append(locator)
-
-    # combine as VRT
-    out_vrt_name = join(source_dir, replace_ext(filename, '.vrt'))
-    gdal.BuildVRT(out_vrt_name, sub_datasets, # TODO: options
-    )
-
-    # TODO: maybe just translate here?
-    gdal.Warp(
-        target_filename,
-        out_vrt_name
-    )
+            sub_datasets.append((locator, subdataset_types[sd_type]))
+
+    if not sub_datasets:
+        raise Exception('No subdatasets were matched by the provided types')
+
+    for locator, suffix in sub_datasets:
+        target_filename = join(target_dir, basename(replace_ext(source_filename, '%s.tif' % suffix)))
+        gdal.Translate(target_filename, locator, format='GTiff')