feat(storage): Enable full object checksum PR 1/3 : parse finalize_time and server crc32c in async object stream

packages/google-cloud-storage/google/cloud/storage/asyncio/async_read_object_stream.py

@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import datetime
 from typing import List, Optional, Tuple
 
 from google.api_core.bidi_async import AsyncBidiRpc
@@ -79,6 +80,9 @@ def __init__(
         self.socket_like_rpc: Optional[AsyncBidiRpc] = None
         self._is_stream_open: bool = False
         self.persisted_size: Optional[int] = None
+        self.is_finalized: bool = False
+        self.full_obj_server_crc32c: Optional[int] = None
+        self.object_metadata: Optional[_storage_v2.Object] = None
 
     async def open(self, metadata: Optional[List[Tuple[str, str]]] = None) -> None:
         """Opens the bidi-gRPC connection to read from the object.
@@ -132,6 +136,21 @@ async def open(self, metadata: Optional[List[Tuple[str, str]]] = None) -> None:
                 self.generation_number = response.metadata.generation
             # update persisted size
             self.persisted_size = response.metadata.size
+            self.object_metadata = response.metadata
+            # Since full object checksum validation is only required for finalized objects,
+            # check finalize_time (which is DatetimeWithNanoseconds/datetime in production, or mocked in tests).
+            finalize_time = getattr(response.metadata, "finalize_time", None)
+            if finalize_time:
+                is_finalized_val = False
+                if isinstance(finalize_time, datetime.datetime):
+                    is_finalized_val = True
+                elif hasattr(finalize_time, "seconds") and finalize_time.seconds > 0:
+                    is_finalized_val = True
+
+                if is_finalized_val:
+                    self.is_finalized = True
+                    if hasattr(response.metadata, "checksums") and response.metadata.checksums:
+                        self.full_obj_server_crc32c = response.metadata.checksums.crc32c
 
         if response and response.read_handle:
             self.read_handle = response.read_handle

packages/google-cloud-storage/tests/unit/asyncio/test_async_read_object_stream.py

@@ -38,9 +38,11 @@ async def instantiate_read_obj_stream(mock_client, mock_cls_async_bidi_rpc, open
     socket_like_rpc.open = AsyncMock()
 
     recv_response = mock.MagicMock(spec=_storage_v2.BidiReadObjectResponse)
-    recv_response.metadata = mock.MagicMock(spec=_storage_v2.Object)
+    recv_response.metadata = mock.MagicMock()
     recv_response.metadata.generation = _TEST_GENERATION_NUMBER
     recv_response.metadata.size = _TEST_OBJECT_SIZE
+    recv_response.metadata.finalize_time.seconds = 12345
+    recv_response.metadata.checksums.crc32c = 98765
     recv_response.read_handle = _TEST_READ_HANDLE
     socket_like_rpc.recv = AsyncMock(return_value=recv_response)
 
@@ -130,6 +132,8 @@ async def test_open(mock_client, mock_cls_async_bidi_rpc):
     assert read_obj_stream.generation_number == _TEST_GENERATION_NUMBER
     assert read_obj_stream.read_handle == _TEST_READ_HANDLE
     assert read_obj_stream.persisted_size == _TEST_OBJECT_SIZE
+    assert read_obj_stream.is_finalized is True
+    assert read_obj_stream.full_obj_server_crc32c == 98765
     assert read_obj_stream.is_stream_open
 
 
@@ -381,3 +385,34 @@ async def test_recv_updates_read_handle_on_refresh(
 
     await stream.recv()
     assert stream.read_handle == refreshed_handle
+
+
+@mock.patch("google.cloud.storage.asyncio.async_read_object_stream.AsyncBidiRpc")
+@mock.patch(
+    "google.cloud.storage.asyncio.async_grpc_client.AsyncGrpcClient.grpc_client"
+)
+@pytest.mark.asyncio
+async def test_open_unfinalized_object_skips_checksum(mock_client, mock_cls_async_bidi_rpc):
+    socket_like_rpc = AsyncMock()
+    mock_cls_async_bidi_rpc.return_value = socket_like_rpc
+    socket_like_rpc.open = AsyncMock()
+
+    recv_response = mock.MagicMock(spec=_storage_v2.BidiReadObjectResponse)
+    recv_response.metadata = mock.MagicMock()
+    recv_response.metadata.generation = _TEST_GENERATION_NUMBER
+    recv_response.metadata.size = _TEST_OBJECT_SIZE
+    recv_response.metadata.finalize_time.seconds = 0  # NOT finalized!
+    recv_response.metadata.checksums.crc32c = 98765
+    recv_response.read_handle = _TEST_READ_HANDLE
+    socket_like_rpc.recv = AsyncMock(return_value=recv_response)
+
+    read_obj_stream = _AsyncReadObjectStream(
+        client=mock_client,
+        bucket_name=_TEST_BUCKET_NAME,
+        object_name=_TEST_OBJECT_NAME,
+    )
+
+    await read_obj_stream.open()
+
+    assert read_obj_stream.is_finalized is False
+    assert read_obj_stream.full_obj_server_crc32c is None