[Storage] Add client-side encryption version 2.0, supporting AES-GCM-256

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

@@ -4,6 +4,10 @@
 
 ### Features Added
 - Added support for service version 2021-08-06.
+- Added a new version of client-side encryption for blobs (version 2.0) which utilizes AES-GCM-256 encryption.
+If you are currently using client-side encryption, it is **highly recommended** to switch to a form of server-side
+encryption (Customer-Provided Key, Encryption Scope, etc.) or version 2.0 of client-side encryption. The encryption
+version can be specified on any client constructor via the `encryption_version` keyword (`encryption_version='2.0'`).
 
 ## 12.12.0 (2022-05-09)
 

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

@@ -331,6 +331,8 @@ Defaults to `False`.
 Use the following keyword arguments when instantiating a client to configure encryption:
 
 * __require_encryption__ (bool): If set to True, will enforce that objects are encrypted and decrypt them.
+* __encryption_version__ (str): Specifies the version of encryption to use. Current options are `'2.0'` or `'1.0'` and
+the default value is `'1.0'`. Version 1.0 is deprecated, and it is **highly recommended** to use version 2.0.
 * __key_encryption_key__ (object): The user-provided key-encryption-key. The instance must implement the following methods:
     - `wrap_key(key)`--wraps the specified key using an algorithm of the user's choice.
     - `get_key_wrap_algorithm()`--returns the algorithm used to wrap the specified symmetric key.

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

@@ -355,11 +355,14 @@ def _upload_blob_options(  # pylint:disable=too-many-statements
             raise ValueError("Encryption required but no key was provided.")
         encryption_options = {
             'required': self.require_encryption,
+            'version': self.encryption_version,
             'key': self.key_encryption_key,
             'resolver': self.key_resolver_function,
         }
         if self.key_encryption_key is not None:
-            cek, iv, encryption_data = generate_blob_encryption_data(self.key_encryption_key)
+            cek, iv, encryption_data = generate_blob_encryption_data(
+                self.key_encryption_key,
+                self.encryption_version)
             encryption_options['cek'] = cek
             encryption_options['vector'] = iv
             encryption_options['data'] = encryption_data
@@ -422,6 +425,8 @@ def _upload_blob_options(  # pylint:disable=too-many-statements
             kwargs['client'] = self._client.block_blob
             kwargs['data'] = data
         elif blob_type == BlobType.PageBlob:
+            if self.encryption_version == '2.0' and (self.require_encryption or self.key_encryption_key is not None):
+                raise ValueError("Encryption version 2.0 does not currently support page blobs.")
             kwargs['client'] = self._client.page_blob
         elif blob_type == BlobType.AppendBlob:
             if self.require_encryption or (self.key_encryption_key is not None):
@@ -4126,5 +4131,5 @@ def _get_container_client(self): # pylint: disable=client-method-missing-kwargs
             "{}://{}".format(self.scheme, self.primary_hostname), container_name=self.container_name,
             credential=self._raw_credential, api_version=self.api_version, _configuration=self._config,
             _pipeline=_pipeline, _location_mode=self._location_mode, _hosts=self._hosts,
-            require_encryption=self.require_encryption, key_encryption_key=self.key_encryption_key,
-            key_resolver_function=self.key_resolver_function)
+            require_encryption=self.require_encryption, encryption_version=self.encryption_version,
+            key_encryption_key=self.key_encryption_key, key_resolver_function=self.key_resolver_function)

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

@@ -683,8 +683,8 @@ def get_container_client(self, container):
             self.url, container_name=container_name,
             credential=self.credential, api_version=self.api_version, _configuration=self._config,
             _pipeline=_pipeline, _location_mode=self._location_mode, _hosts=self._hosts,
-            require_encryption=self.require_encryption, key_encryption_key=self.key_encryption_key,
-            key_resolver_function=self.key_resolver_function)
+            require_encryption=self.require_encryption, encryption_version=self.encryption_version,
+            key_encryption_key=self.key_encryption_key, key_resolver_function=self.key_resolver_function)
 
     def get_blob_client(
             self, container,  # type: Union[ContainerProperties, str]
@@ -736,5 +736,5 @@ def get_blob_client(
             self.url, container_name=container_name, blob_name=blob_name, snapshot=snapshot,
             credential=self.credential, api_version=self.api_version, _configuration=self._config,
             _pipeline=_pipeline, _location_mode=self._location_mode, _hosts=self._hosts,
-            require_encryption=self.require_encryption, key_encryption_key=self.key_encryption_key,
-            key_resolver_function=self.key_resolver_function)
+            require_encryption=self.require_encryption, encryption_version=self.encryption_version,
+            key_encryption_key=self.key_encryption_key, key_resolver_function=self.key_resolver_function)

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

@@ -323,16 +323,16 @@ def _rename_container(self, new_name, **kwargs):
         """
         lease = kwargs.pop('lease', None)
         try:
-            kwargs['source_lease_id'] = lease.id  # type: str
+            kwargs['source_lease_id'] = lease.id
         except AttributeError:
             kwargs['source_lease_id'] = lease
         try:
             renamed_container = ContainerClient(
                 "{}://{}".format(self.scheme, self.primary_hostname), container_name=new_name,
                 credential=self.credential, api_version=self.api_version, _configuration=self._config,
                 _pipeline=self._pipeline, _location_mode=self._location_mode, _hosts=self._hosts,
-                require_encryption=self.require_encryption, key_encryption_key=self.key_encryption_key,
-                key_resolver_function=self.key_resolver_function)
+                require_encryption=self.require_encryption, encryption_version=self.encryption_version,
+                key_encryption_key=self.key_encryption_key, key_resolver_function=self.key_resolver_function)
             renamed_container._client.container.rename(self.container_name, **kwargs)   # pylint: disable = protected-access
             return renamed_container
         except HttpResponseError as error:
@@ -619,8 +619,8 @@ def _get_blob_service_client(self):  # pylint: disable=client-method-missing-kwa
             "{}://{}".format(self.scheme, self.primary_hostname),
             credential=self._raw_credential, api_version=self.api_version, _configuration=self._config,
             _location_mode=self._location_mode, _hosts=self._hosts, require_encryption=self.require_encryption,
-            key_encryption_key=self.key_encryption_key, key_resolver_function=self.key_resolver_function,
-            _pipeline=_pipeline)
+            encryption_version=self.encryption_version, key_encryption_key=self.key_encryption_key,
+            key_resolver_function=self.key_resolver_function, _pipeline=_pipeline)
 
     @distributed_trace
     def get_container_access_policy(self, **kwargs):
@@ -1608,5 +1608,5 @@ def get_blob_client(
             self.url, container_name=self.container_name, blob_name=blob_name, snapshot=snapshot,
             credential=self.credential, api_version=self.api_version, _configuration=self._config,
             _pipeline=_pipeline, _location_mode=self._location_mode, _hosts=self._hosts,
-            require_encryption=self.require_encryption, key_encryption_key=self.key_encryption_key,
-            key_resolver_function=self.key_resolver_function)
+            require_encryption=self.require_encryption, encryption_version=self.encryption_version,
+            key_encryption_key=self.key_encryption_key, key_resolver_function=self.key_resolver_function)

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

@@ -16,31 +16,26 @@
 from azure.core.exceptions import HttpResponseError, ServiceResponseError
 
 from azure.core.tracing.common import with_current_context
-from ._shared.encryption import decrypt_blob
+from ._shared.encryption import (
+    adjust_blob_size_for_encryption,
+    decrypt_blob,
+    get_adjusted_download_range_and_offset,
+    is_encryption_v2,
+    parse_encryption_data
+)
 from ._shared.request_handlers import validate_and_format_range_headers
 from ._shared.response_handlers import process_storage_error, parse_length_from_content_range
-from ._deserialize import get_page_ranges_result
+from ._deserialize import deserialize_blob_properties, get_page_ranges_result
 
 
-def process_range_and_offset(start_range, end_range, length, encryption):
+def process_range_and_offset(start_range, end_range, length, encryption_options, encryption_data):
     start_offset, end_offset = 0, 0
-    if encryption.get("key") is not None or encryption.get("resolver") is not None:
-        if start_range is not None:
-            # Align the start of the range along a 16 byte block
-            start_offset = start_range % 16
-            start_range -= start_offset
-
-            # Include an extra 16 bytes for the IV if necessary
-            # Because of the previous offsetting, start_range will always
-            # be a multiple of 16.
-            if start_range > 0:
-                start_offset += 16
-                start_range -= 16
-
-        if length is not None:
-            # Align the end of the range along a 16 byte block
-            end_offset = 15 - (end_range % 16)
-            end_range += end_offset
+    if encryption_options.get("key") is not None or encryption_options.get("resolver") is not None:
+        return get_adjusted_download_range_and_offset(
+            start_range,
+            end_range,
+            length,
+            encryption_data)
 
     return (start_range, end_range), (start_offset, end_offset)
 
@@ -81,6 +76,7 @@ def __init__(
         parallel=None,
         validate_content=None,
         encryption_options=None,
+        encryption_data=None,
         progress_hook=None,
         **kwargs
     ):
@@ -108,6 +104,7 @@ def __init__(
 
         # Encryption
         self.encryption_options = encryption_options
+        self.encryption_data = encryption_data
 
         # Parameters for each get operation
         self.validate_content = validate_content
@@ -183,7 +180,7 @@ def _do_optimize(self, given_range_start, given_range_end):
 
     def _download_chunk(self, chunk_start, chunk_end):
         download_range, offset = process_range_and_offset(
-            chunk_start, chunk_end, chunk_end, self.encryption_options
+            chunk_start, chunk_end, chunk_end, self.encryption_options, self.encryption_data
         )
 
         # No need to download the empty chunk from server if there's no data in the chunk to be downloaded.
@@ -335,6 +332,10 @@ def __init__(
         self._file_size = None
         self._non_empty_ranges = None
         self._response = None
+        self._encryption_data = None
+
+        if self._encryption_options.get("key") is not None or self._encryption_options.get("resolver") is not None:
+            self._get_encryption_data_request()
 
         # The service only provides transactional MD5s for chunks under 4MB.
         # If validate_content is on, get only self.MAX_CHUNK_GET_SIZE for the first
@@ -349,7 +350,11 @@ def __init__(
             initial_request_end = initial_request_start + self._first_get_size - 1
 
         self._initial_range, self._initial_offset = process_range_and_offset(
-            initial_request_start, initial_request_end, self._end_range, self._encryption_options
+            initial_request_start,
+            initial_request_end,
+            self._end_range,
+            self._encryption_options,
+            self._encryption_data
         )
 
         self._response = self._initial_request()
@@ -376,6 +381,21 @@ def __init__(
     def __len__(self):
         return self.size
 
+    def _get_encryption_data_request(self):
+        # Save current request cls
+        download_cls = self._request_options.pop('cls', None)
+        # Adjust cls for get_properties
+        self._request_options['cls'] = deserialize_blob_properties
+
+        properties = self._clients.blob.get_properties(**self._request_options)
+        # This will return None if there is no encryption metadata or there are parsing errors.
+        # That is acceptable here, the proper error will be caught and surfaced hen attempting
+        # to decrypt the blob.
+        self._encryption_data = parse_encryption_data(properties.metadata)
+
+        # Restore cls for download
+        self._request_options['cls'] = download_cls
+
     def _initial_request(self):
         range_header, range_validation = validate_and_format_range_headers(
             self._initial_range[0],
@@ -405,6 +425,9 @@ def _initial_request(self):
                 # Parse the total file size and adjust the download size if ranges
                 # were specified
                 self._file_size = parse_length_from_content_range(response.properties.content_range)
+                # Remove any extra encryption data size from blob size
+                self._file_size = adjust_blob_size_for_encryption(self._file_size, self._encryption_data)
+
                 if self._end_range is not None:
                     # Use the end range index unless it is over the end of the file
                     self.size = min(self._file_size, self._end_range - self._start_range + 1)
@@ -465,7 +488,8 @@ def _initial_request(self):
 
         # If the file is small, the download is complete at this point.
         # If file size is large, download the rest of the file in chunks.
-        if response.properties.size != self.size:
+        # Use less than here for encryption.
+        if response.properties.size < self.size:
             if self._request_options.get("modified_access_conditions"):
                 self._request_options["modified_access_conditions"].if_match = response.properties.etag
         else:
@@ -494,18 +518,25 @@ def chunks(self):
             if self._end_range is not None:
                 # Use the end range index unless it is over the end of the file
                 data_end = min(self._file_size, self._end_range + 1)
+
+            data_start = self._initial_range[1] + 1  # Start where the first download ended
+            # For encryption V2 only, adjust start to the end of the fetched data rather than download size
+            if is_encryption_v2(self._encryption_data):
+                data_start = (self._start_range or 0) + len(self._current_content)
+
             iter_downloader = _ChunkDownloader(
                 client=self._clients.blob,
                 non_empty_ranges=self._non_empty_ranges,
                 total_size=self.size,
                 chunk_size=self._config.max_chunk_get_size,
                 current_progress=self._first_get_size,
-                start_range=self._initial_range[1] + 1,  # start where the first download ended
+                start_range=data_start,
                 end_range=data_end,
                 stream=None,
                 parallel=False,
                 validate_content=self._validate_content,
                 encryption_options=self._encryption_options,
+                encryption_data=self._encryption_data,
                 use_location=self._location_mode,
                 **self._request_options
             )
@@ -599,18 +630,24 @@ def readinto(self, stream):
             # Use the length unless it is over the end of the file
             data_end = min(self._file_size, self._end_range + 1)
 
+        data_start = self._initial_range[1] + 1  # Start where the first download ended
+        # For encryption V2 only, adjust start to the end of the fetched data rather than download size
+        if is_encryption_v2(self._encryption_data):
+            data_start = (self._start_range or 0) + len(self._current_content)
+
         downloader = _ChunkDownloader(
             client=self._clients.blob,
             non_empty_ranges=self._non_empty_ranges,
             total_size=self.size,
             chunk_size=self._config.max_chunk_get_size,
             current_progress=self._first_get_size,
-            start_range=self._initial_range[1] + 1,  # Start where the first download ended
+            start_range=data_start,
             end_range=data_end,
             stream=stream,
             parallel=parallel,
             validate_content=self._validate_content,
             encryption_options=self._encryption_options,
+            encryption_data=self._encryption_data,
             use_location=self._location_mode,
             progress_hook=self._progress_hook,
             **self._request_options

sdk/storage/azure-storage-blob/azure/storage/blob/_shared/base_client.py

@@ -105,6 +105,7 @@ def __init__(
             self._hosts = {LocationMode.PRIMARY: primary_hostname, LocationMode.SECONDARY: secondary_hostname}
 
         self.require_encryption = kwargs.get("require_encryption", False)
+        self.encryption_version = kwargs.get("encryption_version", "1.0")
         self.key_encryption_key = kwargs.get("key_encryption_key")
         self.key_resolver_function = kwargs.get("key_resolver_function")
         self._config, self._pipeline = self._create_pipeline(self.credential, storage_sdk=service, **kwargs)