Merge pull request #152 from open-data/fix/misc-exception-handling

Improved Exception Handling on File Upload

Author: JVickery-TBS

ckanext/recombinant/read_excel.py

@@ -8,6 +8,8 @@
 from werkzeug.datastructures import FileStorage as FlaskFileStorage
 from cgi import FieldStorage
 
+from ckan.plugins.toolkit import _
+
 from ckanext.recombinant.datatypes import canonicalize
 from ckanext.recombinant.errors import BadExcelData
 
@@ -17,7 +19,8 @@
 
 
 def read_excel(f: Union[str, FlaskFileStorage, FieldStorage],
-               expected_sheet_names: List[str],
+               expected_sheet_names: List[str] = [],
+               bad_sheet_names: List[str] = [],
                file_contents: Optional[str] = None) -> Iterator[Any]:
     """
     Return a generator that opens the excel file f (name or file object)
@@ -36,6 +39,12 @@ def read_excel(f: Union[str, FlaskFileStorage, FieldStorage],
     for sheetname in wb.sheetnames:
         if sheetname == 'reference':
             return
+        if sheetname in bad_sheet_names:
+            raise BadExcelData(_('Invalid file for this data type. ' +
+                                 'Sheet must be labeled "{0}", ' +
+                                 'but you supplied a sheet labeled "{1}"').format(
+                                 '"/"'.join(sorted(expected_sheet_names)),
+                                 sheetname))
         if sheetname not in expected_sheet_names:
             # NOTE: some Excel extensions and Macros create fully hidden
             #       worksheets that act as a sort of database/index cache

ckanext/recombinant/views.py

@@ -750,8 +750,16 @@ def _process_upload_file(lc: LocalCKAN,
         (resource['name'], resource['id'])
         for resource in dataset['resources'])
 
+    bad_types: List[str] = h.recombinant_get_types()
+    bad_types.remove(dataset['type'])
+    bad_sheet_names = []
+    for bt in bad_types:
+        brs = get_geno(bt).get('resources', [])
+        bad_sheet_names += [br['resource_name'] for br in brs]
+
     # type_ignore_reason: incomplete typing
-    upload_data = read_excel(upload_file, expected_sheet_names.keys())  # type: ignore
+    upload_data = read_excel(upload_file, expected_sheet_names.keys(),  # type: ignore
+                             bad_sheet_names)
     total_records = 0
     # type_ignore_reason: incomplete typing
     backend: DatastorePostgresqlBackend = DatastoreBackend.\