feat: GenAI SDK client(multimodal) - Add to_batch_job_source and get_batch_job_destination to MultimodalDataset

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

@@ -81,6 +81,15 @@ def mock_generate_multimodal_dataset_display_name():
         yield mock_generate
 
 
+@pytest.fixture
+def mock_get_batch_job_unique_name():
+    with mock.patch.object(
+        _datasets_utils, "get_batch_job_unique_name"
+    ) as mock_unique_name:
+        mock_unique_name.return_value = "12345678901234_abcde"
+        yield mock_unique_name
+
+
 def test_create_dataset(client):
     create_dataset_operation = client.datasets._create_multimodal_dataset(
         name="projects/vertex-sdk-dev/locations/us-central1",

tests/unit/vertexai/genai/test_multimodal_datasets_genai.py

@@ -36,6 +36,15 @@ def mock_import_bigframes():
         yield mock_import_bigframes
 
 
+@pytest.fixture
+def mock_get_batch_job_unique_name():
+    with mock.patch.object(
+        _datasets_utils, "get_batch_job_unique_name"
+    ) as mock_unique_name:
+        mock_unique_name.return_value = "12345678901234_abcde"
+        yield mock_unique_name
+
+
 class TestMultimodalDataset:
 
     def test_read_config(self):
@@ -157,6 +166,28 @@ def test_to_bigframes(self, mock_import_bigframes):
             "project.dataset.table"
         )
 
+    def test_get_batch_job_destination(self, mock_get_batch_job_unique_name):
+        dataset = types.MultimodalDataset(
+            name="projects/vertex-sdk-dev/locations/us-central1/datasets/12345",
+            display_name="test_multimodal_dataset",
+            metadata={
+                "inputConfig": {
+                    "bigquerySource": {
+                        "uri": "bq://target_project.target_dataset.target_table"
+                    },
+                },
+            },
+        )
+        destination = dataset.get_batch_job_destination()
+        assert (
+            destination.vertex_dataset.display_name
+            == "test_multimodal_dataset_batch_output_12345678901234_abcde"
+        )
+        assert (
+            destination.vertex_dataset.bigquery_destination
+            == "bq://target_project.target_dataset.target_table_batch_output_12345678901234_abcde"
+        )
+
 
 class TestGeminiRequestReadConfig:
     def test_single_turn_template(self):

vertexai/_genai/_datasets_utils.py

@@ -242,6 +242,13 @@ def generate_multimodal_dataset_display_name() -> str:
     return f"MultimodalDataset {datetime.datetime.now().isoformat(sep=' ')}"
 
 
+def get_batch_job_unique_name() -> str:
+    """Generates a unique name suffix for a batch job destination."""
+    timestamp = datetime.datetime.now().strftime("%Y%m%d%H%M%S")
+    unique_id = uuid.uuid4().hex[0:5]
+    return f"{timestamp}_{unique_id}"
+
+
 def save_dataframe_to_bigquery(
     dataframe: "bigframes.pandas.DataFrame",  # type: ignore # noqa: F821
     target_table_id: str,

vertexai/_genai/types/common.py

@@ -15334,6 +15334,28 @@ def to_bigframes(
             raise ValueError("Multimodal dataset bigquery source uri is not set.")
         return bigframes.pandas.read_gbq_table(self.bigquery_uri.removeprefix("bq://"))
 
+    def to_batch_job_source(self) -> "genai_types.BatchJobSource":
+        """Converts the dataset to a BatchJobSource."""
+        return genai_types.BatchJobSource(
+            vertex_dataset_name=self.name,
+        )
+
+    def get_batch_job_destination(self) -> "genai_types.BatchJobDestination":
+        """Converts the dataset to a BatchJobDestination."""
+        from .. import _datasets_utils
+
+        unique_name = _datasets_utils.get_batch_job_unique_name()
+        bigquery_uri = self.bigquery_uri
+        if bigquery_uri is None:
+            raise ValueError("Multimodal dataset bigquery source uri is not set.")
+        curr_display_name = self.display_name or "genai_batch_job"
+        return genai_types.BatchJobDestination(
+            vertex_dataset=genai_types.VertexMultimodalDatasetDestination(
+                display_name=f"{curr_display_name}_batch_output_{unique_name}",
+                bigquery_destination=f"{bigquery_uri}_batch_output_{unique_name}",
+            )
+        )
+
 
 class MultimodalDatasetDict(TypedDict, total=False):
     """Represents a multimodal dataset."""