[Storage] Support metadata on upload_blob_from_url API (#37696)

Author: vincenttran-msft

sdk/storage/azure-storage-blob/CHANGELOG.md

@@ -3,7 +3,7 @@
 ## 12.24.0b1 (Unreleased)
 
 ### Features Added
-
+- Added support for passing metadata to `upload_blob_from_url` via the new `metadata` keyword.
 
 
 ## 12.23.1 (2024-09-25)

sdk/storage/azure-storage-blob/assets.json

@@ -2,5 +2,5 @@
   "AssetsRepo": "Azure/azure-sdk-assets",
   "AssetsRepoPrefixPath": "python",
   "TagPrefix": "python/storage/azure-storage-blob",
-  "Tag": "python/storage/azure-storage-blob_fad0e99de3"
+  "Tag": "python/storage/azure-storage-blob_7df5687d1f"
 }

sdk/storage/azure-storage-blob/azure/storage/blob/_blob_client.py

@@ -320,7 +320,12 @@ def get_account_information(self, **kwargs: Any) -> Dict[str, str]:
             process_storage_error(error)
 
     @distributed_trace
-    def upload_blob_from_url(self, source_url: str, **kwargs: Any) -> Dict[str, Any]:
+    def upload_blob_from_url(
+        self, source_url: str,
+        *,
+        metadata: Optional[Dict[str, str]] = None,
+        **kwargs: Any
+    ) -> Dict[str, Any]:
         """
         Creates a new Block Blob where the content of the blob is read from a given URL.
         The content of an existing blob is overwritten with the new blob.
@@ -337,6 +342,8 @@ def upload_blob_from_url(self, source_url: str, **kwargs: Any) -> Dict[str, Any]
             https://myaccount.blob.core.windows.net/mycontainer/myblob?snapshot=<DateTime>
 
             https://otheraccount.blob.core.windows.net/mycontainer/myblob?sastoken
+        :keyword dict(str, str) metadata:
+            Name-value pairs associated with the blob as metadata.
         :keyword bool overwrite: Whether the blob to be uploaded should overwrite the current data.
             If True, upload_blob will overwrite the existing data. If set to False, the
             operation will fail with ResourceExistsError.
@@ -422,6 +429,7 @@ def upload_blob_from_url(self, source_url: str, **kwargs: Any) -> Dict[str, Any]
             raise ValueError("Customer provided encryption key must be used over HTTPS.")
         options = _upload_blob_from_url_options(
             source_url=source_url,
+            metadata=metadata,
             **kwargs)
         try:
             return cast(Dict[str, Any], self._client.block_blob.put_blob_from_url(**options))

sdk/storage/azure-storage-blob/azure/storage/blob/_blob_client_helpers.py

@@ -188,7 +188,10 @@ def _upload_blob_options(  # pylint:disable=too-many-statements
         raise ValueError(f"Unsupported BlobType: {blob_type}")
     return kwargs
 
-def _upload_blob_from_url_options(source_url: str, **kwargs: Any ) -> Dict[str, Any]:
+def _upload_blob_from_url_options(source_url: str, **kwargs: Any) -> Dict[str, Any]:
+    metadata = kwargs.pop('metadata', None)
+    headers = kwargs.pop('headers', {})
+    headers.update(add_metadata_headers(metadata))
     source_url = _encode_source_url(source_url=source_url)
     tier = kwargs.pop('standard_blob_tier', None)
     overwrite = kwargs.pop('overwrite', False)
@@ -222,7 +225,8 @@ def _upload_blob_from_url_options(source_url: str, **kwargs: Any ) -> Dict[str,
         'tier': tier.value if tier else None,
         'source_modified_access_conditions': get_source_conditions(kwargs),
         'cpk_info': cpk_info,
-        'cpk_scope_info': get_cpk_scope_info(kwargs)
+        'cpk_scope_info': get_cpk_scope_info(kwargs),
+        'headers': headers,
     }
     options.update(kwargs)
     if not overwrite and not _any_conditions(**options): # pylint: disable=protected-access

sdk/storage/azure-storage-blob/azure/storage/blob/aio/_blob_client_async.py

@@ -310,7 +310,12 @@ async def get_account_information(self, **kwargs: Any) -> Dict[str, str]:
             process_storage_error(error)
 
     @distributed_trace_async
-    async def upload_blob_from_url(self, source_url: str, **kwargs: Any) -> Dict[str, Any]:
+    async def upload_blob_from_url(
+        self, source_url: str,
+        *,
+        metadata: Optional[Dict[str, str]] = None,
+        **kwargs: Any
+    ) -> Dict[str, Any]:
         """
         Creates a new Block Blob where the content of the blob is read from a given URL.
         The content of an existing blob is overwritten with the new blob.
@@ -327,6 +332,8 @@ async def upload_blob_from_url(self, source_url: str, **kwargs: Any) -> Dict[str
             https://myaccount.blob.core.windows.net/mycontainer/myblob?snapshot=<DateTime>
 
             https://otheraccount.blob.core.windows.net/mycontainer/myblob?sastoken
+        :keyword dict(str, str) metadata:
+            Name-value pairs associated with the blob as metadata.
         :keyword bool overwrite: Whether the blob to be uploaded should overwrite the current data.
             If True, upload_blob will overwrite the existing data. If set to False, the
             operation will fail with ResourceExistsError.
@@ -412,6 +419,7 @@ async def upload_blob_from_url(self, source_url: str, **kwargs: Any) -> Dict[str
             raise ValueError("Customer provided encryption key must be used over HTTPS.")
         options = _upload_blob_from_url_options(
             source_url=source_url,
+            metadata=metadata,
             **kwargs)
         try:
             return cast(Dict[str, Any], await self._client.block_blob.put_blob_from_url(**options))

sdk/storage/azure-storage-blob/tests/test_block_blob.py

@@ -197,6 +197,38 @@ def test_upload_blob_from_url_with_standard_tier_specified(self, **kwargs):
         # Assert
         assert new_blob_properties.blob_tier == blob_tier
 
+    @BlobPreparer()
+    @recorded_by_proxy
+    def test_upload_blob_from_url_with_metadata(self, **kwargs):
+        storage_account_name = kwargs.pop("storage_account_name")
+        storage_account_key = kwargs.pop("storage_account_key")
+
+        # Arrange
+        self._setup(storage_account_name, storage_account_key, container_name="testcontainer")
+        blob = self._create_blob()
+        self.bsc.get_blob_client(self.container_name, blob.blob_name)
+        sas = self.generate_sas(
+            generate_blob_sas,
+            account_name=storage_account_name,
+            account_key=storage_account_key,
+            container_name=self.container_name,
+            blob_name=blob.blob_name,
+            permission=BlobSasPermissions(read=True),
+            expiry=datetime.utcnow() + timedelta(hours=1)
+        )
+        # Act
+        source_blob = '{0}/{1}/{2}?{3}'.format(
+            self.account_url(storage_account_name, "blob"), self.container_name, blob.blob_name, sas)
+
+        blob_name = self.get_resource_name("blobcopy")
+        new_blob = self.bsc.get_blob_client(self.container_name, blob_name)
+        new_blob.upload_blob_from_url(source_blob, metadata={'blobdata': 'data1'})
+
+        new_blob_properties = new_blob.get_blob_properties()
+
+        # Assert
+        assert new_blob_properties.metadata == {'blobdata': 'data1'}
+
     @BlobPreparer()
     @recorded_by_proxy
     def test_upload_blob_from_url_with_cold_tier_specified(self, **kwargs):

sdk/storage/azure-storage-blob/tests/test_block_blob_async.py

@@ -220,6 +220,38 @@ async def test_upload_blob_from_url_with_standard_tier_specified(self, **kwargs)
         # Assert
         assert new_blob_properties.blob_tier == blob_tier
 
+    @BlobPreparer()
+    @recorded_by_proxy_async
+    async def test_upload_blob_from_url_with_metadata(self, **kwargs):
+        storage_account_name = kwargs.pop("storage_account_name")
+        storage_account_key = kwargs.pop("storage_account_key")
+
+        # Arrange
+        await self._setup(storage_account_name, storage_account_key, container_name="testcontainer")
+        blob = await self._create_blob()
+        self.bsc.get_blob_client(self.container_name, blob.blob_name)
+        sas = self.generate_sas(
+            generate_blob_sas,
+            account_name=storage_account_name,
+            account_key=storage_account_key,
+            container_name=self.container_name,
+            blob_name=blob.blob_name,
+            permission=BlobSasPermissions(read=True),
+            expiry=datetime.utcnow() + timedelta(hours=1)
+        )
+        # Act
+        source_blob = '{0}/{1}/{2}?{3}'.format(
+            self.account_url(storage_account_name, "blob"), self.container_name, blob.blob_name, sas)
+
+        blob_name = self.get_resource_name("blobcopy")
+        new_blob = self.bsc.get_blob_client(self.container_name, blob_name)
+        await new_blob.upload_blob_from_url(source_blob, metadata={'blobdata': 'data1'})
+
+        new_blob_properties = await new_blob.get_blob_properties()
+
+        # Assert
+        assert new_blob_properties.metadata == {'blobdata': 'data1'}
+
     @BlobPreparer()
     @recorded_by_proxy_async
     async def test_upload_blob_from_url_with_cold_tier_specified(self, **kwargs):