ensure blob_str enters as bytes

Author: matt-bernstein

label_studio/io_storages/azure_blob/models.py

@@ -222,7 +222,7 @@ def get_data(self, key) -> list[StorageObjectParams]:
 
         container = self.get_container()
         blob = container.download_blob(key)
-        blob_str = blob.content_as_text()
+        blob_str = blob.content_as_bytes()
         return load_tasks_json(blob_str, key)
 
     def scan_and_create_links(self):

label_studio/io_storages/localfiles/models.py

@@ -93,10 +93,8 @@ def get_data(self, key) -> list[StorageObjectParams]:
 
         try:
             with open(path, 'rb') as f:
-                blob_str = f.read().decode('utf-8')
+                blob_str = f.read()
                 return load_tasks_json(blob_str, key)
-        except UnicodeDecodeError as e:
-            raise ValueError(f'Failed to decode file {path} as UTF-8: {str(e)}')
         except OSError as e:
             raise ValueError(f'Failed to read file {path}: {str(e)}')
 

label_studio/io_storages/redis/models.py

@@ -95,7 +95,8 @@ def get_data(self, key) -> list[StorageObjectParams]:
         value_str = client.get(key)
         if not value_str:
             return []
-        return load_tasks_json(value_str, key)
+        blob_str = value_str.encode('utf-8')
+        return load_tasks_json(blob_str, key)
 
     def scan_and_create_links(self):
         return self._scan_and_create_links(RedisImportStorageLink)

label_studio/io_storages/s3/models.py

@@ -227,7 +227,7 @@ def get_data(self, key) -> list[StorageObjectParams]:
         # read task json from bucket and validate it
         _, s3 = self.get_client_and_resource()
         bucket = s3.Bucket(self.bucket)
-        obj = s3.Object(bucket.name, key).get()['Body'].read().decode('utf-8')
+        obj = s3.Object(bucket.name, key).get()['Body'].read()
         return load_tasks_json(obj, key)
 
     @catch_and_reraise_from_none

label_studio/io_storages/utils.py

@@ -139,12 +139,12 @@ def bulk_create(
         ]
 
 
-def _load_tasks_json(blob_str: str, key: str) -> list[StorageObjectParams]:
+def _load_tasks_json(blob_str: bytes, key: str) -> list[StorageObjectParams]:
     """
     Parse blob_str containing task JSON(s) and return the validated result or raise an error.
 
     Args:
-        blob_str (str): The blob string to parse.
+        blob_str (bytes): The blob string to parse.
         key (str): The key of the blob. Used for error messages.
 
     Returns: