Error on text dataset file not found (#1534)

llmfoundry/command_utils/data_prep/convert_text_to_mds.py

@@ -32,6 +32,7 @@
     CannotUnicodeDecodeFile,
     DatasetTooSmallError,
     InputFolderMissingDataError,
+    InputFolderNotFound,
     OutputFolderNotEmptyError,
 )
 
@@ -125,11 +126,15 @@ def get_object_names(input_folder: str) -> list[str]:
     object_store = maybe_create_object_store_from_uri(input_folder)
     if object_store is not None:
         _, _, folder_prefix = parse_uri(input_folder)
-        names = [
-            name for name in object_store.list_objects(folder_prefix)
-            if name.endswith('.txt')
-        ]
-        log.info(f'Found {len(names)} text files in remote storage')
+        try:
+            names = [
+                name for name in object_store.list_objects(folder_prefix)
+                if name.endswith('.txt')
+            ]
+            log.info(f'Found {len(names)} text files in remote storage')
+        except FileNotFoundError:
+            raise InputFolderNotFound(folder_prefix)
+
     else:
         # input_folder is a local folder
         names = [

llmfoundry/utils/exceptions.py

@@ -348,6 +348,17 @@ def __init__(self, input_folder: str) -> None:
         super().__init__(message, input_folder=input_folder)
 
 
+class InputFolderNotFound(UserError):
+    """Error thrown when the a folder is not found."""
+
+    def __init__(self, folder_that_was_not_found: str) -> None:
+        message = f'{folder_that_was_not_found} not found.'
+        super().__init__(
+            message,
+            folder_that_was_not_found=folder_that_was_not_found,
+        )
+
+
 class CannotUnicodeDecodeFile(UserError):
     """Error thrown when the input folder is missing data."""