feat: GenAI SDK client(multimodal) - Auto-generate display name for multimodal datasets if not provided.

PiperOrigin-RevId: 897780705

Author: cleop-google

tests/unit/vertexai/genai/replays/test_create_multimodal_datasets.py

@@ -72,6 +72,15 @@ def mock_import_bigframes(is_replay_mode):
         yield None
 
 
+@pytest.fixture
+def mock_generate_multimodal_dataset_display_name():
+    with mock.patch.object(
+        _datasets_utils, "generate_multimodal_dataset_display_name"
+    ) as mock_generate:
+        mock_generate.return_value = "test-generated-name"
+        yield mock_generate
+
+
 def test_create_dataset(client):
     create_dataset_operation = client.datasets._create_multimodal_dataset(
         name="projects/vertex-sdk-dev/locations/us-central1",
@@ -106,6 +115,21 @@ def test_create_dataset_from_bigquery(client):
     )
 
 
+@pytest.mark.usefixtures("mock_generate_multimodal_dataset_display_name")
+def test_create_dataset_from_bigquery_no_display_name(client):
+    dataset = client.datasets.create_from_bigquery(
+        multimodal_dataset={
+            "metadata": {
+                "inputConfig": {
+                    "bigquerySource": {"uri": f"bq://{BIGQUERY_TABLE_NAME}"},
+                },
+            },
+        }
+    )
+    assert isinstance(dataset, types.MultimodalDataset)
+    assert dataset.display_name == "test-generated-name"
+
+
 @pytest.mark.usefixtures("mock_bigquery_client", "mock_import_bigframes")
 def test_create_dataset_from_pandas(client, is_replay_mode):
     dataframe = pd.DataFrame(
@@ -230,6 +254,22 @@ async def test_create_dataset_from_bigquery_async(client):
     )
 
 
+@pytest.mark.asyncio
+@pytest.mark.usefixtures("mock_generate_multimodal_dataset_display_name")
+async def test_create_dataset_from_bigquery_no_display_name_async(client):
+    dataset = await client.aio.datasets.create_from_bigquery(
+        multimodal_dataset={
+            "metadata": {
+                "inputConfig": {
+                    "bigquerySource": {"uri": f"bq://{BIGQUERY_TABLE_NAME}"},
+                },
+            },
+        }
+    )
+    assert isinstance(dataset, types.MultimodalDataset)
+    assert dataset.display_name == "test-generated-name"
+
+
 @pytest.mark.asyncio
 async def test_create_dataset_from_bigquery_async_with_timeout(client):
     dataset = await client.aio.datasets.create_from_bigquery(

vertexai/_genai/_datasets_utils.py

@@ -15,6 +15,7 @@
 """Utility functions for multimodal dataset."""
 
 import asyncio
+import datetime
 from typing import Any, Type, TypeVar
 import uuid
 
@@ -227,6 +228,11 @@ def _generate_target_table_id(dataset_id: str) -> str:
     return f"{dataset_id}.{_DEFAULT_BQ_TABLE_PREFIX}_{str(uuid.uuid4())}"
 
 
+def generate_multimodal_dataset_display_name() -> str:
+    """Generates a display name with a timestamp."""
+    return f"MultimodalDataset {datetime.datetime.now().isoformat(sep=' ')}"
+
+
 def save_dataframe_to_bigquery(
     dataframe: "bigframes.pandas.DataFrame",  # type: ignore # noqa: F821
     target_table_id: str,

vertexai/_genai/datasets.py

@@ -933,9 +933,14 @@ def create_from_bigquery(
         elif not config:
             config = types.CreateMultimodalDatasetConfig()
 
+        display_name = (
+            multimodal_dataset.display_name
+            if multimodal_dataset.display_name is not None
+            else _datasets_utils.generate_multimodal_dataset_display_name()
+        )
         multimodal_dataset_operation = self._create_multimodal_dataset(
             config=config,
-            display_name=multimodal_dataset.display_name,
+            display_name=display_name,
             metadata_schema_uri=_datasets_utils.METADATA_SCHEMA_URI,
             metadata=multimodal_dataset.metadata,
         )
@@ -2150,9 +2155,14 @@ async def create_from_bigquery(
         elif not config:
             config = types.CreateMultimodalDatasetConfig()
 
+        display_name = (
+            multimodal_dataset.display_name
+            if multimodal_dataset.display_name is not None
+            else _datasets_utils.generate_multimodal_dataset_display_name()
+        )
         multimodal_dataset_operation = await self._create_multimodal_dataset(
             config=config,
-            display_name=multimodal_dataset.display_name,
+            display_name=display_name,
             metadata_schema_uri=_datasets_utils.METADATA_SCHEMA_URI,
             metadata=multimodal_dataset.metadata,
         )