refactor: update load_stream method to directly yield file chunks (#9806)

api/extensions/storage/aliyun_oss_storage.py

@@ -36,13 +36,10 @@ class AliyunOssStorage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         obj = self.client.get_object(self.__wrapper_folder_filename(filename))
         while chunk := obj.read(4096):
             yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         self.client.get_object_to_file(self.__wrapper_folder_filename(filename), target_filepath)
 

api/extensions/storage/aws_s3_storage.py

@@ -62,7 +62,6 @@ class AwsS3Storage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         try:
             response = self.client.get_object(Bucket=self.bucket_name, Key=filename)
             yield from response["Body"].iter_chunks()
@@ -72,8 +71,6 @@ class AwsS3Storage(BaseStorage):
             else:
                 raise
 
-        return generate()
-
     def download(self, filename, target_filepath):
         self.client.download_file(self.bucket_name, filename, target_filepath)
 

api/extensions/storage/azure_blob_storage.py

@@ -32,14 +32,10 @@ class AzureBlobStorage(BaseStorage):
 
     def load_stream(self, filename: str) -> Generator:
         client = self._sync_client()
-
-        def generate(filename: str = filename) -> Generator:
         blob = client.get_blob_client(container=self.bucket_name, blob=filename)
         blob_data = blob.download_blob()
         yield from blob_data.chunks()
 
-        return generate(filename)
-
     def download(self, filename, target_filepath):
         client = self._sync_client()
 

api/extensions/storage/baidu_obs_storage.py

@@ -39,13 +39,10 @@ class BaiduObsStorage(BaseStorage):
         return response.data.read()
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         response = self.client.get_object(bucket_name=self.bucket_name, key=filename).data
         while chunk := response.read(4096):
             yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         self.client.get_object_to_file(bucket_name=self.bucket_name, key=filename, file_name=target_filepath)
 

api/extensions/storage/google_cloud_storage.py

@@ -39,15 +39,12 @@ class GoogleCloudStorage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         bucket = self.client.get_bucket(self.bucket_name)
         blob = bucket.get_blob(filename)
         with blob.open(mode="rb") as blob_stream:
             while chunk := blob_stream.read(4096):
                 yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         bucket = self.client.get_bucket(self.bucket_name)
         blob = bucket.get_blob(filename)

api/extensions/storage/huawei_obs_storage.py

@@ -27,13 +27,10 @@ class HuaweiObsStorage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         response = self.client.getObject(bucketName=self.bucket_name, objectKey=filename)["body"].response
         while chunk := response.read(4096):
             yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         self.client.getObject(bucketName=self.bucket_name, objectKey=filename, downloadPath=target_filepath)
 

api/extensions/storage/local_fs_storage.py

@@ -40,16 +40,12 @@ class LocalFsStorage(BaseStorage):
 
     def load_stream(self, filename: str) -> Generator:
         filepath = self._build_filepath(filename)
-
-        def generate() -> Generator:
         if not os.path.exists(filepath):
             raise FileNotFoundError("File not found")
         with open(filepath, "rb") as f:
             while chunk := f.read(4096):  # Read in chunks of 4KB
                 yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         filepath = self._build_filepath(filename)
         if not os.path.exists(filepath):

api/extensions/storage/oracle_oci_storage.py

@@ -36,7 +36,6 @@ class OracleOCIStorage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         try:
             response = self.client.get_object(Bucket=self.bucket_name, Key=filename)
             yield from response["Body"].iter_chunks()
@@ -46,8 +45,6 @@ class OracleOCIStorage(BaseStorage):
             else:
                 raise
 
-        return generate()
-
     def download(self, filename, target_filepath):
         self.client.download_file(self.bucket_name, filename, target_filepath)
 

api/extensions/storage/supabase_storage.py

@@ -36,17 +36,14 @@ class SupabaseStorage(BaseStorage):
         return content
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         result = self.client.storage.from_(self.bucket_name).download(filename)
         byte_stream = io.BytesIO(result)
         while chunk := byte_stream.read(4096):  # Read in chunks of 4KB
             yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         result = self.client.storage.from_(self.bucket_name).download(filename)
-        Path(result).write_bytes(result)
+        Path(target_filepath).write_bytes(result)
 
     def exists(self, filename):
         result = self.client.storage.from_(self.bucket_name).list(filename)

api/extensions/storage/tencent_cos_storage.py

@@ -29,12 +29,9 @@ class TencentCosStorage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         response = self.client.get_object(Bucket=self.bucket_name, Key=filename)
         yield from response["Body"].get_stream(chunk_size=4096)
 
-        return generate()
-
     def download(self, filename, target_filepath):
         response = self.client.get_object(Bucket=self.bucket_name, Key=filename)
         response["Body"].get_stream_to_file(target_filepath)

api/extensions/storage/volcengine_tos_storage.py

@@ -27,13 +27,10 @@ class VolcengineTosStorage(BaseStorage):
         return data
 
     def load_stream(self, filename: str) -> Generator:
-        def generate(filename: str = filename) -> Generator:
         response = self.client.get_object(bucket=self.bucket_name, key=filename)
         while chunk := response.read(4096):
             yield chunk
 
-        return generate()
-
     def download(self, filename, target_filepath):
         self.client.get_object_to_file(bucket=self.bucket_name, key=filename, file_path=target_filepath)