structure saas with tools

.venv/lib/python3.10/site-packages/vertexai/resources/__init__.py

@@ -0,0 +1,182 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+"""The vertexai resources module."""
+
+from google.cloud.aiplatform import initializer
+
+from google.cloud.aiplatform.datasets import (
+    ImageDataset,
+    TabularDataset,
+    TextDataset,
+    TimeSeriesDataset,
+    VideoDataset,
+)
+from google.cloud.aiplatform import explain
+from google.cloud.aiplatform import gapic
+from google.cloud.aiplatform import hyperparameter_tuning
+from google.cloud.aiplatform.featurestore import (
+    EntityType,
+    Feature,
+    Featurestore,
+)
+from google.cloud.aiplatform.matching_engine import (
+    MatchingEngineIndex,
+    MatchingEngineIndexEndpoint,
+)
+from google.cloud.aiplatform import metadata
+from google.cloud.aiplatform.tensorboard import uploader_tracker
+from google.cloud.aiplatform.models import DeploymentResourcePool
+from google.cloud.aiplatform.models import Endpoint
+from google.cloud.aiplatform.models import PrivateEndpoint
+from google.cloud.aiplatform.models import Model
+from google.cloud.aiplatform.models import ModelRegistry
+from google.cloud.aiplatform.model_evaluation import ModelEvaluation
+from google.cloud.aiplatform.jobs import (
+    BatchPredictionJob,
+    CustomJob,
+    HyperparameterTuningJob,
+    ModelDeploymentMonitoringJob,
+)
+from google.cloud.aiplatform.pipeline_jobs import PipelineJob
+from google.cloud.aiplatform.pipeline_job_schedules import (
+    PipelineJobSchedule,
+)
+from google.cloud.aiplatform.tensorboard import (
+    Tensorboard,
+    TensorboardExperiment,
+    TensorboardRun,
+    TensorboardTimeSeries,
+)
+from google.cloud.aiplatform.training_jobs import (
+    CustomTrainingJob,
+    CustomContainerTrainingJob,
+    CustomPythonPackageTrainingJob,
+    AutoMLTabularTrainingJob,
+    AutoMLForecastingTrainingJob,
+    SequenceToSequencePlusForecastingTrainingJob,
+    TemporalFusionTransformerForecastingTrainingJob,
+    TimeSeriesDenseEncoderForecastingTrainingJob,
+    AutoMLImageTrainingJob,
+    AutoMLTextTrainingJob,
+    AutoMLVideoTrainingJob,
+)
+
+from google.cloud.aiplatform import helpers
+
+"""
+Usage:
+import vertexai
+
+vertexai.init(project='my_project')
+"""
+init = initializer.global_config.init
+
+get_pipeline_df = metadata.metadata._LegacyExperimentService.get_pipeline_df
+
+log_params = metadata.metadata._experiment_tracker.log_params
+log_metrics = metadata.metadata._experiment_tracker.log_metrics
+log_classification_metrics = (
+    metadata.metadata._experiment_tracker.log_classification_metrics
+)
+log_model = metadata.metadata._experiment_tracker.log_model
+get_experiment_df = metadata.metadata._experiment_tracker.get_experiment_df
+start_run = metadata.metadata._experiment_tracker.start_run
+autolog = metadata.metadata._experiment_tracker.autolog
+start_execution = metadata.metadata._experiment_tracker.start_execution
+log = metadata.metadata._experiment_tracker.log
+log_time_series_metrics = metadata.metadata._experiment_tracker.log_time_series_metrics
+end_run = metadata.metadata._experiment_tracker.end_run
+
+upload_tb_log = uploader_tracker._tensorboard_tracker.upload_tb_log
+start_upload_tb_log = uploader_tracker._tensorboard_tracker.start_upload_tb_log
+end_upload_tb_log = uploader_tracker._tensorboard_tracker.end_upload_tb_log
+
+save_model = metadata._models.save_model
+get_experiment_model = metadata.schema.google.artifact_schema.ExperimentModel.get
+
+Experiment = metadata.experiment_resources.Experiment
+ExperimentRun = metadata.experiment_run_resource.ExperimentRun
+Artifact = metadata.artifact.Artifact
+Execution = metadata.execution.Execution
+Context = metadata.context.Context
+
+
+__all__ = (
+    "end_run",
+    "explain",
+    "gapic",
+    "init",
+    "helpers",
+    "hyperparameter_tuning",
+    "log",
+    "log_params",
+    "log_metrics",
+    "log_classification_metrics",
+    "log_model",
+    "log_time_series_metrics",
+    "get_experiment_df",
+    "get_pipeline_df",
+    "start_run",
+    "start_execution",
+    "save_model",
+    "get_experiment_model",
+    "autolog",
+    "upload_tb_log",
+    "start_upload_tb_log",
+    "end_upload_tb_log",
+    "Artifact",
+    "AutoMLImageTrainingJob",
+    "AutoMLTabularTrainingJob",
+    "AutoMLForecastingTrainingJob",
+    "AutoMLTextTrainingJob",
+    "AutoMLVideoTrainingJob",
+    "BatchPredictionJob",
+    "CustomJob",
+    "CustomTrainingJob",
+    "CustomContainerTrainingJob",
+    "CustomPythonPackageTrainingJob",
+    "DeploymentResourcePool",
+    "Endpoint",
+    "EntityType",
+    "Execution",
+    "Experiment",
+    "ExperimentRun",
+    "Feature",
+    "Featurestore",
+    "MatchingEngineIndex",
+    "MatchingEngineIndexEndpoint",
+    "ImageDataset",
+    "HyperparameterTuningJob",
+    "Model",
+    "ModelRegistry",
+    "ModelEvaluation",
+    "ModelDeploymentMonitoringJob",
+    "PipelineJob",
+    "PipelineJobSchedule",
+    "PrivateEndpoint",
+    "SequenceToSequencePlusForecastingTrainingJob",
+    "TabularDataset",
+    "Tensorboard",
+    "TensorboardExperiment",
+    "TensorboardRun",
+    "TensorboardTimeSeries",
+    "TextDataset",
+    "TemporalFusionTransformerForecastingTrainingJob",
+    "TimeSeriesDataset",
+    "TimeSeriesDenseEncoderForecastingTrainingJob",
+    "VideoDataset",
+)

.venv/lib/python3.10/site-packages/vertexai/resources/preview/__init__.py

@@ -0,0 +1,91 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+"""The vertexai resources preview module."""
+
+from google.cloud.aiplatform.preview.jobs import (
+    CustomJob,
+    HyperparameterTuningJob,
+)
+from google.cloud.aiplatform.preview.models import (
+    Prediction,
+    DeploymentResourcePool,
+    Endpoint,
+    Model,
+)
+from google.cloud.aiplatform.preview.featurestore.entity_type import (
+    EntityType,
+)
+from google.cloud.aiplatform.preview.persistent_resource import (
+    PersistentResource,
+)
+from google.cloud.aiplatform.preview.pipelinejobschedule.pipeline_job_schedules import (
+    PipelineJobSchedule,
+)
+
+from vertexai.resources.preview.feature_store import (
+    Feature,
+    FeatureGroup,
+    FeatureGroupBigQuerySource,
+    FeatureMonitor,
+    FeatureOnlineStore,
+    FeatureOnlineStoreType,
+    FeatureView,
+    FeatureViewBigQuerySource,
+    FeatureViewReadResponse,
+    FeatureViewRegistrySource,
+    FeatureViewVertexRagSource,
+    IndexConfig,
+    TreeAhConfig,
+    BruteForceConfig,
+    DistanceMeasureType,
+    AlgorithmConfig,
+)
+
+from vertexai.resources.preview.ml_monitoring import (
+    ModelMonitor,
+    ModelMonitoringJob,
+)
+
+__all__ = (
+    "CustomJob",
+    "HyperparameterTuningJob",
+    "Prediction",
+    "DeploymentResourcePool",
+    "Endpoint",
+    "Model",
+    "PersistentResource",
+    "EntityType",
+    "PipelineJobSchedule",
+    "Feature",
+    "FeatureGroup",
+    "FeatureGroupBigQuerySource",
+    "FeatureMonitor",
+    "FeatureOnlineStoreType",
+    "FeatureOnlineStore",
+    "FeatureView",
+    "FeatureViewBigQuerySource",
+    "FeatureViewReadResponse",
+    "FeatureViewVertexRagSource",
+    "FeatureViewRegistrySource",
+    "IndexConfig",
+    "TreeAhConfig",
+    "BruteForceConfig",
+    "DistanceMeasureType",
+    "AlgorithmConfig",
+    "ModelMonitor",
+    "ModelMonitoringJob",
+)

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/__init__.py

@@ -0,0 +1,71 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+"""The vertexai resources preview module."""
+
+from vertexai.resources.preview.feature_store.feature import (
+    Feature,
+)
+
+from vertexai.resources.preview.feature_store.feature_group import (
+    FeatureGroup,
+)
+
+from vertexai.resources.preview.feature_store.feature_monitor import (
+    FeatureMonitor,
+)
+
+from vertexai.resources.preview.feature_store.feature_online_store import (
+    FeatureOnlineStore,
+    FeatureOnlineStoreType,
+)
+
+from vertexai.resources.preview.feature_store.feature_view import (
+    FeatureView,
+)
+
+from vertexai.resources.preview.feature_store.utils import (
+    FeatureGroupBigQuerySource,
+    FeatureViewBigQuerySource,
+    FeatureViewReadResponse,
+    FeatureViewVertexRagSource,
+    FeatureViewRegistrySource,
+    IndexConfig,
+    TreeAhConfig,
+    BruteForceConfig,
+    DistanceMeasureType,
+    AlgorithmConfig,
+)
+
+__all__ = (
+    Feature,
+    FeatureGroup,
+    FeatureGroupBigQuerySource,
+    FeatureMonitor,
+    FeatureOnlineStoreType,
+    FeatureOnlineStore,
+    FeatureView,
+    FeatureViewBigQuerySource,
+    FeatureViewReadResponse,
+    FeatureViewVertexRagSource,
+    FeatureViewRegistrySource,
+    IndexConfig,
+    IndexConfig,
+    TreeAhConfig,
+    BruteForceConfig,
+    DistanceMeasureType,
+    AlgorithmConfig,
+)

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/_offline_store_impl.py

@@ -0,0 +1,190 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import textwrap
+from dataclasses import dataclass
+from typing import Optional, List
+
+
+@dataclass
+class DataSource:
+    """An object to represent a data source - both entity DataFrame and any feature data.
+
+    Contains helpers for use with SQL templating.
+    """
+
+    def __init__(
+        self,
+        qualifying_name: str,
+        sql: str,
+        data_columns: List[str],
+        timestamp_column: str,
+        entity_id_columns: Optional[List[str]] = None,
+    ):
+        """Initialize DataSource object.
+
+        Args:
+            qualifying_name:
+                A unique name used to qualify the data in the PITL query.
+            sql:
+                SQL query representing the data_source.
+            data_columns:
+                Columns other than entity ID column(s) and timestamp column.
+            timestamp_column:
+                The column that holds feature timestamp data.
+            entity_id_columns:
+                The column(s) that holds entity IDs. Shouldn't be populated for
+                entity_df.
+        """
+        self.qualifying_name = qualifying_name
+        self._sql = sql
+        self.data_columns = data_columns
+        self.timestamp_column = timestamp_column
+        self.entity_id_columns = entity_id_columns
+
+    def copy_with_pitl_suffix(self) -> "DataSource":
+        import copy
+
+        data_source = copy.copy(self)
+        data_source.qualifying_name += "_pitl"
+        return data_source
+
+    @property
+    def sql(self):
+        return self._sql
+
+    @property
+    def comma_separated_qualified_data_columns(self):
+        return ", ".join(
+            [self.qualifying_name + "." + col for col in self.data_columns]
+        )
+
+    @property
+    def comma_separated_name_qualified_all_non_timestamp_columns(self):
+        """Same as `comma_separated_qualified_data_columns` but including entity ID column."""
+        all_columns = self.data_columns.copy()
+        if self.entity_id_columns:
+            all_columns += self.entity_id_columns
+        return ", ".join([self.qualifying_name + "." + col for col in all_columns])
+
+    @property
+    def qualified_timestamp_column(self) -> str:
+        """Returns name qualified timestamp column e.g. `name.feature_timestamp`."""
+        return f"{self.qualifying_name}.{self.timestamp_column}"
+
+
+def _generate_eid_check(entity_data: DataSource, feature: DataSource):
+    """Generate equality check for entity columns of feature against matching columns in entity_data."""
+    e_cols = set(entity_data.data_columns)
+    f_cols = feature.entity_id_columns
+    assert f_cols
+
+    equal_statements = []
+    for col in f_cols:
+        if col not in e_cols:
+            raise ValueError(
+                f"Feature entity ID column '{col}' should be a column in the entity DataFrame."
+            )
+        equal_statements.append(
+            f"{entity_data.qualifying_name}.{col} = {feature.qualifying_name}.{col}"
+        )
+
+    statement = " AND\n".join(equal_statements)
+
+    return statement
+
+
+# Args:
+#   textwrap: Module
+#   generate_eid_check: function (above)
+#   entity_data: DataSource
+#   feature_data: List[DataSource]
+_PITL_QUERY_TEMPLATE_RAW = """WITH
+  {{ entity_data.qualifying_name }}_without_row_num AS (
+{{ textwrap.indent(entity_data.sql, ' ' * 4) }}
+  ),
+  {{ entity_data.qualifying_name }} AS (
+    SELECT *, ROW_NUMBER() OVER() AS row_num,
+    FROM entity_df_without_row_num
+  ),
+
+  # Features
+  {% for feature_data_elem in feature_data %}
+  {{ feature_data_elem.qualifying_name }} AS (
+{{ textwrap.indent(feature_data_elem.sql, ' ' * 4) }}
+  ),
+  {% endfor %}
+
+  # Features with PITL
+  {% for feature_data_elem in feature_data %}
+  {{ feature_data_elem.qualifying_name }}_pitl AS (
+    SELECT
+      {{ entity_data.qualifying_name }}.row_num,
+      {{ feature_data_elem.comma_separated_qualified_data_columns }},
+    FROM {{ entity_data.qualifying_name }}
+    LEFT JOIN {{ feature_data_elem.qualifying_name }}
+    ON (
+{{ textwrap.indent(generate_eid_check(entity_data, feature_data_elem) + ' AND', ' ' * 6) }}
+      CAST({{ feature_data_elem.qualified_timestamp_column }} AS TIMESTAMP) <= CAST({{ entity_data.qualified_timestamp_column }} AS TIMESTAMP)
+    )
+    QUALIFY ROW_NUMBER() OVER (PARTITION BY {{ entity_data.qualifying_name }}.row_num ORDER BY {{ feature_data_elem.qualified_timestamp_column }} DESC) = 1
+  ){{ ',' if not loop.last else '' }}
+  {% endfor %}
+
+
+SELECT
+  {{ entity_data.comma_separated_name_qualified_all_non_timestamp_columns }},
+  {% for feature_data_elem in feature_data %}
+  {% set feature_pitl = feature_data_elem.copy_with_pitl_suffix() %}
+  {{ feature_pitl.comma_separated_qualified_data_columns }},
+  {% endfor %}
+  {{ entity_data.qualified_timestamp_column }}
+
+FROM {{ entity_data.qualifying_name }}
+{% for feature_data_elem in feature_data %}
+JOIN {{ feature_data_elem.qualifying_name }}_pitl USING (row_num)
+{% endfor %}
+"""
+
+
+def pitl_query_template():
+    try:
+        import jinja2
+    except ImportError as exc:
+        raise ImportError(
+            "`Jinja2` is not installed but required for this functionality."
+        ) from exc
+
+    return jinja2.Environment(
+        loader=jinja2.BaseLoader, lstrip_blocks=True, trim_blocks=True
+    ).from_string(_PITL_QUERY_TEMPLATE_RAW)
+
+
+def render_pitl_query(entity_data: DataSource, feature_data: List[DataSource]):
+    """Return the PITL query jinja template.
+
+    The args for the query are as follows:
+      textwrap: The python textwrap module.
+      entity_data[DataSource]: The entity data(frame) as SQL source.
+      feature_data[List[DataSource]]:
+    """
+    return pitl_query_template().render(
+        textwrap=textwrap,
+        generate_eid_check=_generate_eid_check,
+        entity_data=entity_data,
+        feature_data=feature_data,
+    )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/feature.py

@@ -0,0 +1,151 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import re
+from typing import List, Optional
+from google.auth import credentials as auth_credentials
+from google.cloud.aiplatform import base
+from google.cloud.aiplatform import utils
+from google.cloud.aiplatform.compat.types import (
+    feature as gca_feature,
+    feature_monitor_v1beta1 as gca_feature_monitor,
+    feature_v1beta1 as gca_feature_v1beta1,
+    featurestore_service_v1beta1 as gca_featurestore_service_v1beta1,
+)
+
+
+class Feature(base.VertexAiResourceNounWithFutureManager):
+    """Class for managing Feature resources."""
+
+    client_class = utils.FeatureRegistryClientWithOverride
+
+    _resource_noun = "features"
+    _getter_method = "get_feature"
+    _list_method = "list_features"
+    _delete_method = "delete_feature"
+    _parse_resource_name_method = "parse_feature_path"
+    _format_resource_name_method = "feature_path"
+    _gca_resource: gca_feature.Feature
+
+    def __init__(
+        self,
+        name: str,
+        feature_group_id: Optional[str] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        latest_stats_count: Optional[int] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ):
+        """Retrieves an existing managed feature.
+
+        Args:
+            name:
+                The resource name
+                (`projects/.../locations/.../featureGroups/.../features/...`) or
+                ID.
+            feature_group_id:
+                The feature group ID. Must be passed in if name is an ID and not
+                a resource path.
+            project:
+                Project to retrieve feature from. If not set, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to retrieve feature from. If not set, the location set
+                in aiplatform.init will be used.
+            gca_feature_arg:
+                The GCA feature object.
+                Only set when calling from get_feature with latest_stats_count set.
+            credentials:
+                Custom credentials to use to retrieve this feature. Overrides
+                credentials set in aiplatform.init.
+        """
+
+        super().__init__(
+            project=project,
+            location=location,
+            credentials=credentials,
+            resource_name=name,
+        )
+
+        if re.fullmatch(
+            r"projects/.+/locations/.+/featureGroups/.+/features/.+",
+            name,
+        ):
+            if feature_group_id:
+                raise ValueError(
+                    f"Since feature '{name}' is provided as a path, feature_group_id should not be specified."
+                )
+            feature = name
+        else:
+            from .feature_group import FeatureGroup
+
+            # Construct the feature path using feature group ID if  only the
+            # feature group ID is provided.
+            if not feature_group_id:
+                raise ValueError(
+                    f"Since feature '{name}' is not provided as a path, please specify feature_group_id."
+                )
+
+            feature_group_path = utils.full_resource_name(
+                resource_name=feature_group_id,
+                resource_noun=FeatureGroup._resource_noun,
+                parse_resource_name_method=FeatureGroup._parse_resource_name,
+                format_resource_name_method=FeatureGroup._format_resource_name,
+            )
+
+            feature = f"{feature_group_path}/features/{name}"
+
+        if latest_stats_count is not None:
+            api_client = self.__class__._instantiate_client(
+                location=location, credentials=credentials
+            )
+
+            feature_obj: gca_feature_v1beta1.Feature = api_client.select_version(
+                "v1beta1"
+            ).get_feature(
+                request=gca_featurestore_service_v1beta1.GetFeatureRequest(
+                    name=f"{feature}",
+                    feature_stats_and_anomaly_spec=gca_feature_monitor.FeatureStatsAndAnomalySpec(
+                        latest_stats_count=latest_stats_count
+                    ),
+                )
+            )
+            self._gca_resource = feature_obj
+        else:
+            self._gca_resource = self._get_gca_resource(resource_name=feature)
+
+    @property
+    def version_column_name(self) -> str:
+        """The name of the BigQuery Table/View column hosting data for this version."""
+        return self._gca_resource.version_column_name
+
+    @property
+    def description(self) -> str:
+        """The description of the feature."""
+        return self._gca_resource.description
+
+    @property
+    def point_of_contact(self) -> str:
+        """The point of contact for the feature."""
+        return self._gca_resource.point_of_contact
+
+    @property
+    def feature_stats_and_anomalies(
+        self,
+    ) -> List[gca_feature_monitor.FeatureStatsAndAnomaly]:
+        """The number of latest stats to return. Only present when gca_feature is set."""
+        return self._gca_resource.feature_stats_and_anomaly

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/feature_group.py

@@ -0,0 +1,592 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from typing import Dict, List, Optional, Sequence, Tuple
+from google.auth import credentials as auth_credentials
+from google.cloud.aiplatform import base, initializer
+from google.cloud.aiplatform import utils
+from google.cloud.aiplatform.compat.types import (
+    feature as gca_feature,
+    feature_group as gca_feature_group,
+    io as gca_io,
+    feature_monitor_v1beta1 as gca_feature_monitor,
+)
+from vertexai.resources.preview.feature_store.utils import (
+    FeatureGroupBigQuerySource,
+)
+from vertexai.resources.preview.feature_store import (
+    Feature,
+)
+from vertexai.resources.preview.feature_store.feature_monitor import (
+    FeatureMonitor,
+)
+
+
+_LOGGER = base.Logger(__name__)
+
+
+class FeatureGroup(base.VertexAiResourceNounWithFutureManager):
+    """Class for managing Feature Group resources."""
+
+    client_class = utils.FeatureRegistryClientWithOverride
+
+    _resource_noun = "feature_groups"
+    _getter_method = "get_feature_group"
+    _list_method = "list_feature_groups"
+    _delete_method = "delete_feature_group"
+    _parse_resource_name_method = "parse_feature_group_path"
+    _format_resource_name_method = "feature_group_path"
+    _gca_resource: gca_feature_group.FeatureGroup
+
+    def __init__(
+        self,
+        name: str,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ):
+        """Retrieves an existing managed feature group.
+
+        Args:
+            name:
+                The resource name
+                (`projects/.../locations/.../featureGroups/...`) or ID.
+            project:
+                Project to retrieve feature group from. If unset, the
+                project set in aiplatform.init will be used.
+            location:
+                Location to retrieve feature group from. If not set,
+                location set in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to retrieve this feature group.
+                Overrides credentials set in aiplatform.init.
+        """
+
+        super().__init__(
+            project=project,
+            location=location,
+            credentials=credentials,
+            resource_name=name,
+        )
+
+        self._gca_resource = self._get_gca_resource(resource_name=name)
+
+    @classmethod
+    def create(
+        cls,
+        name: str,
+        source: FeatureGroupBigQuerySource = None,
+        labels: Optional[Dict[str, str]] = None,
+        description: Optional[str] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+        sync: bool = True,
+    ) -> "FeatureGroup":
+        """Creates a new feature group.
+
+        Args:
+            name: The name of the feature group.
+            source: The BigQuery source of the feature group.
+            labels:
+                The labels with user-defined metadata to organize your
+                FeatureGroup.
+
+                Label keys and values can be no longer than 64
+                characters (Unicode codepoints), can only
+                contain lowercase letters, numeric characters,
+                underscores and dashes. International characters
+                are allowed.
+
+                See https://goo.gl/xmQnxf for more information
+                on and examples of labels. No more than 64 user
+                labels can be associated with one
+                FeatureGroup(System labels are excluded)."
+                System reserved label keys are prefixed with
+                "aiplatform.googleapis.com/" and are immutable.
+            description: Description of the FeatureGroup.
+            project:
+                Project to create feature group in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to create feature group in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature group.
+                Overrides credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+            sync:
+                Whether to execute this creation synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+
+        Returns:
+            FeatureGroup - the FeatureGroup resource object.
+        """
+
+        if not source:
+            raise ValueError("Please specify a valid source.")
+
+        # Only BigQuery source is supported right now.
+        if not isinstance(source, FeatureGroupBigQuerySource):
+            raise ValueError("Only FeatureGroupBigQuerySource is a supported source.")
+
+        # BigQuery source validation.
+        if not source.uri:
+            raise ValueError("Please specify URI in BigQuery source.")
+
+        if not source.entity_id_columns:
+            _LOGGER.info(
+                "No entity ID columns specified in BigQuery source. Defaulting to ['entity_id']."
+            )
+            entity_id_columns = ["entity_id"]
+        else:
+            entity_id_columns = source.entity_id_columns
+
+        gapic_feature_group = gca_feature_group.FeatureGroup(
+            big_query=gca_feature_group.FeatureGroup.BigQuery(
+                big_query_source=gca_io.BigQuerySource(input_uri=source.uri),
+                entity_id_columns=entity_id_columns,
+            ),
+            name=name,
+            description=description,
+        )
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature_group.labels = labels
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        api_client = cls._instantiate_client(location=location, credentials=credentials)
+
+        create_feature_group_lro = api_client.create_feature_group(
+            parent=initializer.global_config.common_location_path(
+                project=project, location=location
+            ),
+            feature_group=gapic_feature_group,
+            feature_group_id=name,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        _LOGGER.log_create_with_lro(cls, create_feature_group_lro)
+
+        created_feature_group = create_feature_group_lro.result()
+
+        _LOGGER.log_create_complete(cls, created_feature_group, "feature_group")
+
+        feature_group_obj = cls(
+            name=created_feature_group.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return feature_group_obj
+
+    @base.optional_sync()
+    def delete(self, force: bool = False, sync: bool = True) -> None:
+        """Deletes this feature group.
+
+        WARNING: This deletion is permanent.
+
+        Args:
+            force:
+                If set to True, all features under this online store will be
+                deleted prior to online store deletion. Otherwise, deletion
+                will only succeed if the online store has no FeatureViews.
+
+                If set to true, any Features under this FeatureGroup will also
+                be deleted. (Otherwise, the request will only work if the
+                FeatureGroup has no Features.)
+            sync:
+                Whether to execute this deletion synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+        """
+
+        lro = getattr(self.api_client, self._delete_method)(
+            name=self.resource_name,
+            force=force,
+        )
+        _LOGGER.log_delete_with_lro(self, lro)
+        lro.result()
+        _LOGGER.log_delete_complete(self)
+
+    def get_feature(
+        self,
+        feature_id: str,
+        latest_stats_count: Optional[int] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> Feature:
+        """Retrieves an existing managed feature.
+
+        Args:
+            feature_id: The ID of the feature.
+            latest_stats_count:
+                The number of latest stats to retrieve. Only returns stats if
+                Feature Monitor is created, and historical stats were generated.
+            credentials:
+                Custom credentials to use to retrieve the feature under this
+                feature group. The order of which credentials are used is as
+                follows: (1) this parameter (2) credentials passed to FeatureGroup
+                constructor (3) credentials set in aiplatform.init.
+
+        Returns:
+            Feature - the Feature resource object under this feature group.
+        """
+        credentials = (
+            credentials or self.credentials or initializer.global_config.credentials
+        )
+        if latest_stats_count is not None:
+            return Feature(
+                name=f"{self.resource_name}/features/{feature_id}",
+                latest_stats_count=latest_stats_count,
+                credentials=credentials,
+            )
+        return Feature(
+            f"{self.resource_name}/features/{feature_id}", credentials=credentials
+        )
+
+    def create_feature(
+        self,
+        name: str,
+        version_column_name: Optional[str] = None,
+        description: Optional[str] = None,
+        labels: Optional[Dict[str, str]] = None,
+        point_of_contact: Optional[str] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+        sync: bool = True,
+    ) -> Feature:
+        """Creates a new feature.
+
+        Args:
+            name: The name of the feature.
+            version_column_name:
+                The name of the BigQuery Table/View column hosting data for this
+                version. If no value is provided, will use feature_id.
+            description: Description of the feature.
+            labels:
+                The labels with user-defined metadata to organize your Features.
+                Label keys and values can be no longer than 64 characters
+                (Unicode codepoints), can only contain lowercase letters,
+                numeric characters, underscores and dashes. International
+                characters are allowed.
+
+                See https://goo.gl/xmQnxf for more information on and examples
+                of labels. No more than 64 user labels can be associated with
+                one Feature (System labels are excluded)." System reserved label
+                keys are prefixed with "aiplatform.googleapis.com/" and are
+                immutable.
+            point_of_contact:
+                Entity responsible for maintaining this feature. Can be comma
+                separated list of email addresses or URIs.
+            project:
+                Project to create feature in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to create feature in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature. Overrides
+                credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+            sync:
+                Whether to execute this creation synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+
+        Returns:
+            Feature - the Feature resource object.
+        """
+
+        gapic_feature = gca_feature.Feature()
+
+        if version_column_name:
+            gapic_feature.version_column_name = version_column_name
+
+        if description:
+            gapic_feature.description = description
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature.labels = labels
+
+        if point_of_contact:
+            gapic_feature.point_of_contact = point_of_contact
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        api_client = self.__class__._instantiate_client(
+            location=location, credentials=credentials
+        )
+
+        create_feature_lro = api_client.create_feature(
+            parent=self.resource_name,
+            feature=gapic_feature,
+            feature_id=name,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        _LOGGER.log_create_with_lro(Feature, create_feature_lro)
+
+        created_feature = create_feature_lro.result()
+
+        _LOGGER.log_create_complete(Feature, created_feature, "feature")
+
+        feature_obj = Feature(
+            name=created_feature.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return feature_obj
+
+    def list_features(
+        self,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> List[Feature]:
+        """Lists features under this feature group.
+
+        Args:
+            project:
+                Project to list features in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to list features in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to list features. Overrides
+                credentials set in aiplatform.init.
+
+        Returns:
+            List of features under this feature group.
+        """
+
+        return Feature.list(
+            parent=self.resource_name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+    def get_feature_monitor(
+        self,
+        feature_monitor_id: str,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> FeatureMonitor:
+        """Retrieves an existing feature monitor.
+
+        Args:
+            feature_monitor_id: The ID of the feature monitor.
+            credentials:
+                Custom credentials to use to retrieve the feature monitor under this
+                feature group. The order of which credentials are used is as
+                follows: (1) this parameter (2) credentials passed to FeatureGroup
+                constructor (3) credentials set in aiplatform.init.
+
+        Returns:
+            FeatureMonitor - the Feature Monitor resource object under this
+            feature group.
+        """
+        credentials = (
+            credentials or self.credentials or initializer.global_config.credentials
+        )
+        return FeatureMonitor(
+            f"{self.resource_name}/featureMonitors/{feature_monitor_id}",
+            credentials=credentials,
+        )
+
+    def create_feature_monitor(
+        self,
+        name: str,
+        description: Optional[str] = None,
+        labels: Optional[Dict[str, str]] = None,
+        schedule_config: Optional[str] = None,
+        feature_selection_configs: Optional[List[Tuple[str, float]]] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+    ) -> FeatureMonitor:
+        """Creates a new feature monitor.
+
+        Args:
+            name: The name of the feature monitor.
+            description: Description of the feature monitor.
+            labels:
+                The labels with user-defined metadata to organize your FeatureMonitors.
+                Label keys and values can be no longer than 64 characters
+                (Unicode codepoints), can only contain lowercase letters,
+                numeric characters, underscores and dashes. International
+                characters are allowed.
+
+                See https://goo.gl/xmQnxf for more information on and examples
+                of labels. No more than 64 user labels can be associated with
+                one FeatureMonitor (System labels are excluded)." System reserved label
+                keys are prefixed with "aiplatform.googleapis.com/" and are
+                immutable.
+            schedule_config:
+                Configures when data is to be monitored for this
+                FeatureMonitor. At the end of the scheduled time,
+                the stats and drift are generated for the selected features.
+                Example format: "TZ=America/New_York 0 9 * * *" (monitors
+                daily at 9 AM EST).
+            feature_selection_configs:
+                List of tuples of feature id and monitoring threshold. If unset,
+                all features in the feature group will be monitored, and the
+                default thresholds 0.3 will be used.
+            project:
+                Project to create feature in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to create feature in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature. Overrides
+                credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+
+        Returns:
+            FeatureMonitor - the FeatureMonitor resource object.
+        """
+
+        gapic_feature_monitor = gca_feature_monitor.FeatureMonitor()
+
+        if description:
+            gapic_feature_monitor.description = description
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature_monitor.labels = labels
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        if schedule_config:
+            gapic_feature_monitor.schedule_config = gca_feature_monitor.ScheduleConfig(
+                cron=schedule_config
+            )
+
+        if feature_selection_configs is None:
+            raise ValueError(
+                "Please specify feature_configs: features to be monitored and"
+                " their thresholds."
+            )
+
+        if feature_selection_configs is not None:
+            gapic_feature_monitor.feature_selection_config.feature_configs = [
+                gca_feature_monitor.FeatureSelectionConfig.FeatureConfig(
+                    feature_id=feature_id,
+                    drift_threshold=threshold if threshold else 0.3,
+                )
+                for feature_id, threshold in feature_selection_configs
+            ]
+
+        api_client = self.__class__._instantiate_client(
+            location=location, credentials=credentials
+        )
+
+        create_feature_monitor_lro = api_client.select_version(
+            "v1beta1"
+        ).create_feature_monitor(
+            parent=self.resource_name,
+            feature_monitor=gapic_feature_monitor,
+            feature_monitor_id=name,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        _LOGGER.log_create_with_lro(FeatureMonitor, create_feature_monitor_lro)
+
+        created_feature_monitor = create_feature_monitor_lro.result()
+
+        _LOGGER.log_create_complete(
+            FeatureMonitor, created_feature_monitor, "feature_monitor"
+        )
+
+        feature_monitor_obj = FeatureMonitor(
+            name=created_feature_monitor.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return feature_monitor_obj
+
+    def list_feature_monitors(
+        self,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> List[FeatureMonitor]:
+        """Lists features monitors under this feature group.
+
+        Args:
+            project:
+                Project to list feature monitors in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to list feature monitors in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to list feature monitors. Overrides
+                credentials set in aiplatform.init.
+
+        Returns:
+            List of feature monitors under this feature group.
+        """
+
+        return FeatureMonitor.list(
+            parent=self.resource_name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+    @property
+    def source(self) -> FeatureGroupBigQuerySource:
+        return FeatureGroupBigQuerySource(
+            uri=self._gca_resource.big_query.big_query_source.input_uri,
+            entity_id_columns=self._gca_resource.big_query.entity_id_columns,
+        )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/feature_monitor.py

@@ -0,0 +1,335 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import re
+from typing import List, Dict, Optional, Tuple, Sequence
+from google.auth import credentials as auth_credentials
+from google.cloud.aiplatform import base, initializer
+from google.cloud.aiplatform import utils
+from google.cloud.aiplatform.compat.types import (
+    feature_monitor_v1beta1 as gca_feature_monitor,
+    feature_monitor_job_v1beta1 as gca_feature_monitor_job,
+)
+
+_LOGGER = base.Logger(__name__)
+
+
+class FeatureMonitor(base.VertexAiResourceNounWithFutureManager):
+    """Class for managing Feature Monitor resources."""
+
+    client_class = utils.FeatureRegistryClientV1Beta1WithOverride
+
+    _resource_noun = "feature_monitors"
+    _getter_method = "get_feature_monitor"
+    _list_method = "list_feature_monitors"
+    _delete_method = "delete_feature_monitor"
+    _parse_resource_name_method = "parse_feature_monitor_path"
+    _format_resource_name_method = "feature_monitor_path"
+    _gca_resource: gca_feature_monitor.FeatureMonitor
+
+    def __init__(
+        self,
+        name: str,
+        feature_group_id: Optional[str] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ):
+        """Retrieves an existing managed feature.
+
+        Args:
+            name:
+                The resource name
+                (`projects/.../locations/.../featureGroups/.../featureMonitors/...`) or
+                ID.
+            feature_group_id:
+                The feature group ID. Must be passed in if name is an ID and not
+                a resource path.
+            project:
+                Project to retrieve feature from. If not set, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to retrieve feature from. If not set, the location set
+                in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to retrieve this feature. Overrides
+                credentials set in aiplatform.init.
+        """
+
+        super().__init__(
+            project=project,
+            location=location,
+            credentials=credentials,
+            resource_name=name,
+        )
+
+        if re.fullmatch(
+            r"projects/.+/locations/.+/featureGroups/.+/featureMonitors/.+",
+            name,
+        ):
+            if feature_group_id:
+                raise ValueError(
+                    f"Since feature monitor '{name}' is provided as a path, feature_group_id should not be specified."
+                )
+            feature_monitor = name
+        else:
+            from .feature_group import FeatureGroup
+
+            # Construct the feature path using feature group ID if  only the
+            # feature group ID is provided.
+            if not feature_group_id:
+                raise ValueError(
+                    f"Since feature monitor '{name}' is not provided as a path, please specify feature_group_id."
+                )
+
+            feature_group_path = utils.full_resource_name(
+                resource_name=feature_group_id,
+                resource_noun=FeatureGroup._resource_noun,
+                parse_resource_name_method=FeatureGroup._parse_resource_name,
+                format_resource_name_method=FeatureGroup._format_resource_name,
+            )
+
+            feature_monitor = f"{feature_group_path}/featureMonitors/{name}"
+
+        self._gca_resource = self._get_gca_resource(resource_name=feature_monitor)
+
+    @property
+    def description(self) -> str:
+        """The description of the feature monitor."""
+        return self._gca_resource.description
+
+    @property
+    def schedule_config(self) -> str:
+        """The schedule config of the feature monitor."""
+        return self._gca_resource.schedule_config.cron
+
+    @property
+    def feature_selection_configs(self) -> List[Tuple[str, float]]:
+        """The feature and it's drift threshold configs of the feature monitor."""
+        configs: List[Tuple[str, float]] = []
+        for (
+            feature_config
+        ) in self._gca_resource.feature_selection_config.feature_configs:
+            configs.append(
+                (
+                    feature_config.feature_id,
+                    feature_config.drift_threshold
+                    if feature_config.drift_threshold
+                    else 0.3,
+                )
+            )
+        return configs
+
+    class FeatureMonitorJob(base.VertexAiResourceNounWithFutureManager):
+        """Class for managing Feature Monitor Job resources."""
+
+        client_class = utils.FeatureRegistryClientV1Beta1WithOverride
+
+        _resource_noun = "featureMonitorJobs"
+        _getter_method = "get_feature_monitor_job"
+        _list_method = "list_feature_monitor_jobs"
+        _delete_method = "delete_feature_monitor_job"
+        _parse_resource_name_method = "parse_feature_monitor_job_path"
+        _format_resource_name_method = "feature_monitor_job_path"
+        _gca_resource: gca_feature_monitor_job.FeatureMonitorJob
+
+        def __init__(
+            self,
+            name: str,
+            project: Optional[str] = None,
+            location: Optional[str] = None,
+            credentials: Optional[auth_credentials.Credentials] = None,
+        ):
+            """Retrieves an existing managed feature monitor job.
+
+            Args:
+                name: The resource name
+                  (`projects/.../locations/.../featureGroups/.../featureMonitors/.../featureMonitorJobs/...`)
+                project: Project to retrieve the feature monitor job from. If
+                  unset, the project set in aiplatform.init will be used.
+                location: Location to retrieve the feature monitor job from. If
+                  not set, location set in aiplatform.init will be used.
+                credentials: Custom credentials to use to retrieve this feature
+                  monitor job. Overrides credentials set in aiplatform.init.
+            """
+            super().__init__(
+                project=project,
+                location=location,
+                credentials=credentials,
+                resource_name=name,
+            )
+
+            if not re.fullmatch(
+                r"projects/.+/locations/.+/featureGroups/.+/featureMonitors/.+/featureMonitorJobs/.+",
+                name,
+            ):
+                raise ValueError(
+                    "name need to specify the fully qualified"
+                    + " feature monitor job resource path."
+                )
+
+            self._gca_resource = self._get_gca_resource(resource_name=name)
+
+        @property
+        def description(self) -> str:
+            """The description of the feature monitor."""
+            return self._gca_resource.description
+
+        @property
+        def feature_stats_and_anomalies(
+            self,
+        ) -> List[gca_feature_monitor.FeatureStatsAndAnomaly]:
+            """The feature stats and anomaly of the feature monitor job."""
+            if self._gca_resource.job_summary:
+                return self._gca_resource.job_summary.feature_stats_and_anomalies
+            return []
+
+    def create_feature_monitor_job(
+        self,
+        description: Optional[str] = None,
+        labels: Optional[Dict[str, str]] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+    ) -> FeatureMonitorJob:
+        """Creates a new feature monitor job.
+
+        Args:
+            description: Description of the feature monitor job.
+            labels:
+                The labels with user-defined metadata to organize your
+                FeatureMonitorJobs.
+                Label keys and values can be no longer than 64 characters
+                (Unicode codepoints), can only contain lowercase letters,
+                numeric characters, underscores and dashes. International
+                characters are allowed.
+
+                See https://goo.gl/xmQnxf for more information on and examples
+                of labels. No more than 64 user labels can be associated with
+                one FeatureMonitor (System labels are excluded)." System reserved label
+                keys are prefixed with "aiplatform.googleapis.com/" and are
+                immutable.
+            project:
+                Project to create feature in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to create feature in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature. Overrides
+                credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+
+        Returns:
+            FeatureMonitorJob - the FeatureMonitorJob resource object.
+        """
+
+        gapic_feature_monitor_job = gca_feature_monitor_job.FeatureMonitorJob()
+
+        if description:
+            gapic_feature_monitor_job.description = description
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature_monitor_job.labels = labels
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        api_client = self.__class__._instantiate_client(
+            location=location, credentials=credentials
+        )
+
+        created_feature_monitor_job = api_client.select_version(
+            "v1beta1"
+        ).create_feature_monitor_job(
+            parent=self.resource_name,
+            feature_monitor_job=gapic_feature_monitor_job,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        feature_monitor_job_obj = self.FeatureMonitorJob(
+            name=created_feature_monitor_job.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return feature_monitor_job_obj
+
+    def get_feature_monitor_job(
+        self,
+        feature_monitor_job_id: str,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> FeatureMonitorJob:
+        """Retrieves an existing feature monitor.
+
+        Args:
+            feature_monitor_job_id: The ID of the feature monitor job.
+            credentials:
+                Custom credentials to use to retrieve the feature monitor job under this
+                feature monitor. The order of which credentials are used is as
+                follows - (1) this parameter (2) credentials passed to FeatureMonitor
+                constructor (3) credentials set in aiplatform.init.
+
+        Returns:
+            FeatureMonitorJob - the Feature Monitor Job resource object under this
+            feature monitor.
+        """
+        credentials = (
+            credentials or self.credentials or initializer.global_config.credentials
+        )
+        return FeatureMonitor.FeatureMonitorJob(
+            f"{self.resource_name}/featureMonitorJobs/{feature_monitor_job_id}",
+            credentials=credentials,
+        )
+
+    def list_feature_monitor_jobs(
+        self,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> List[FeatureMonitorJob]:
+        """Lists features monitor jobs under this feature monitor.
+
+        Args:
+            project:
+                Project to list feature monitors in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to list feature monitors in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to list feature monitors. Overrides
+                credentials set in aiplatform.init.
+
+        Returns:
+            List of feature monitor jobs under this feature monitor.
+        """
+
+        return FeatureMonitor.FeatureMonitorJob.list(
+            parent=self.resource_name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/feature_online_store.py

@@ -0,0 +1,645 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import enum
+from typing import (
+    Dict,
+    List,
+    Optional,
+    Sequence,
+    Tuple,
+    Union,
+)
+
+from google.auth import credentials as auth_credentials
+from google.cloud.aiplatform import (
+    base,
+    initializer,
+    utils,
+)
+from google.cloud.aiplatform.compat.types import (
+    feature_online_store as gca_feature_online_store,
+    service_networking as gca_service_networking,
+    feature_view as gca_feature_view,
+)
+from vertexai.resources.preview.feature_store.feature_view import (
+    FeatureView,
+)
+from vertexai.resources.preview.feature_store.utils import (
+    IndexConfig,
+    FeatureViewBigQuerySource,
+    FeatureViewVertexRagSource,
+    FeatureViewRegistrySource,
+)
+
+
+_LOGGER = base.Logger(__name__)
+
+
+@enum.unique
+class FeatureOnlineStoreType(enum.Enum):
+    UNKNOWN = 0
+    BIGTABLE = 1
+    OPTIMIZED = 2
+
+
+class FeatureOnlineStore(base.VertexAiResourceNounWithFutureManager):
+    """Class for managing Feature Online Store resources."""
+
+    client_class = utils.FeatureOnlineStoreAdminClientWithOverride
+
+    _resource_noun = "feature_online_stores"
+    _getter_method = "get_feature_online_store"
+    _list_method = "list_feature_online_stores"
+    _delete_method = "delete_feature_online_store"
+    _parse_resource_name_method = "parse_feature_online_store_path"
+    _format_resource_name_method = "feature_online_store_path"
+    _gca_resource: gca_feature_online_store.FeatureOnlineStore
+
+    def __init__(
+        self,
+        name: str,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ):
+        """Retrieves an existing managed feature online store.
+
+        Args:
+            name:
+                The resource name
+                (`projects/.../locations/.../featureOnlineStores/...`) or ID.
+            project:
+                Project to retrieve feature online store from. If unset, the
+                project set in aiplatform.init will be used.
+            location:
+                Location to retrieve feature online store from. If not set,
+                location set in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to retrieve this feature online store.
+                Overrides credentials set in aiplatform.init.
+        """
+
+        super().__init__(
+            project=project,
+            location=location,
+            credentials=credentials,
+            resource_name=name,
+        )
+        self._gca_resource = self._get_gca_resource(resource_name=name)
+
+    @classmethod
+    @base.optional_sync()
+    def create_bigtable_store(
+        cls,
+        name: str,
+        min_node_count: Optional[int] = 1,
+        max_node_count: Optional[int] = 1,
+        cpu_utilization_target: Optional[int] = 50,
+        labels: Optional[Dict[str, str]] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+        sync: bool = True,
+    ) -> "FeatureOnlineStore":
+        """Creates a Bigtable online store.
+
+        Example Usage:
+
+            my_fos = vertexai.preview.FeatureOnlineStore.create_bigtable_store('my_fos')
+
+        Args:
+            name: The name of the feature online store.
+            min_node_count:
+                The minimum number of Bigtable nodes to scale down to.  Must be
+                greater than or equal to 1.
+            max_node_count:
+                The maximum number of Bigtable nodes to scale up to.  Must
+                satisfy min_node_count <= max_node_count <= (10 *
+                min_node_count).
+            cpu_utilization_target:
+                A percentage of the cluster's CPU capacity. Can be from 10% to
+                80%. When a cluster's CPU utilization exceeds the target that
+                you have set, Bigtable immediately adds nodes to the cluster.
+                When CPU utilization is substantially lower than the target,
+                Bigtable removes nodes. If not set will default to 50%.
+            labels:
+                The labels with user-defined metadata to organize your feature
+                online store. Label keys and values can be no longer than 64
+                characters (Unicode codepoints), can only contain lowercase
+                letters, numeric characters, underscores and dashes.
+                International characters are allowed. See https://goo.gl/xmQnxf
+                for more information on and examples of labels. No more than 64
+                user labels can be associated with one feature online store
+                (System labels are excluded)." System reserved label keys are
+                prefixed with "aiplatform.googleapis.com/" and are immutable.
+            project:
+                Project to create feature online store in. If unset, the project
+                set in aiplatform.init will be used.
+            location:
+                Location to create feature online store in. If not set, location
+                set in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature online store.
+                Overrides credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+            sync:
+                Whether to execute this creation synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+
+        Returns:
+            FeatureOnlineStore - the FeatureOnlineStore resource object.
+        """
+
+        if min_node_count < 1:
+            raise ValueError("min_node_count must be greater than or equal to 1")
+
+        if max_node_count < min_node_count:
+            raise ValueError(
+                "max_node_count must be greater than or equal to min_node_count"
+            )
+        elif 10 * min_node_count < max_node_count:
+            raise ValueError(
+                "max_node_count must be less than or equal to 10 * min_node_count"
+            )
+
+        if cpu_utilization_target < 10 or cpu_utilization_target > 80:
+            raise ValueError("cpu_utilization_target must be between 10 and 80")
+
+        gapic_feature_online_store = gca_feature_online_store.FeatureOnlineStore(
+            bigtable=gca_feature_online_store.FeatureOnlineStore.Bigtable(
+                auto_scaling=gca_feature_online_store.FeatureOnlineStore.Bigtable.AutoScaling(
+                    min_node_count=min_node_count,
+                    max_node_count=max_node_count,
+                    cpu_utilization_target=cpu_utilization_target,
+                ),
+            ),
+        )
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature_online_store.labels = labels
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        api_client = cls._instantiate_client(location=location, credentials=credentials)
+
+        create_online_store_lro = api_client.create_feature_online_store(
+            parent=initializer.global_config.common_location_path(
+                project=project, location=location
+            ),
+            feature_online_store=gapic_feature_online_store,
+            feature_online_store_id=name,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        _LOGGER.log_create_with_lro(cls, create_online_store_lro)
+
+        created_online_store = create_online_store_lro.result()
+
+        _LOGGER.log_create_complete(cls, created_online_store, "feature_online_store")
+
+        online_store_obj = cls(
+            name=created_online_store.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return online_store_obj
+
+    @classmethod
+    @base.optional_sync()
+    def create_optimized_store(
+        cls,
+        name: str,
+        enable_private_service_connect: bool = False,
+        project_allowlist: Optional[Sequence[str]] = None,
+        labels: Optional[Dict[str, str]] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+        sync: bool = True,
+    ) -> "FeatureOnlineStore":
+        """Creates an Optimized online store.
+
+        Example Usage:
+
+            ```
+            # Create optimized store with public endpoint.
+            my_fos = vertexai.preview.FeatureOnlineStore.create_optimized_store('my_fos')
+            ```
+
+            ```
+            # Create optimized online store with private service connect.
+            my_fos = vertexai.preview.FeatureOnlineStore.create_optimized_store(
+                'my_fos',
+                enable_private_service_connect=True,
+                project_allowlist=['my-project'],
+            )
+            ```
+
+        Args:
+            name: The name of the feature online store.
+            enable_private_service_connect:
+                Optional. If true, expose the optimized online store
+                via private service connect. Otherwise the optimized online
+                store will be accessible through public endpoint.
+            project_allowlist:
+                A list of Projects from which the forwarding
+                rule will target the service attachment. Only needed when
+                `enable_private_service_connect` is set to true.
+            labels:
+                The labels with user-defined metadata to organize your feature
+                online store. Label keys and values can be no longer than 64
+                characters (Unicode codepoints), can only contain lowercase
+                letters, numeric characters, underscores and dashes.
+                International characters are allowed. See https://goo.gl/xmQnxf
+                for more information on and examples of labels. No more than 64
+                user labels can be associated with one feature online store
+                (System labels are excluded)." System reserved label keys are
+                prefixed with "aiplatform.googleapis.com/" and are immutable.
+            project:
+                Project to create feature online store in. If unset, the project
+                set in aiplatform.init will be used.
+            location:
+                Location to create feature online store in. If not set, location
+                set in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature online store.
+                Overrides credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+            sync:
+                Whether to execute this creation synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+
+        Returns:
+            FeatureOnlineStore - the FeatureOnlineStore resource object.
+        """
+        if enable_private_service_connect:
+            if not project_allowlist:
+                raise ValueError(
+                    "`project_allowlist` cannot be empty when `enable_private_service_connect` is set to true."
+                )
+
+            dedicated_serving_endpoint = gca_feature_online_store.FeatureOnlineStore.DedicatedServingEndpoint(
+                private_service_connect_config=gca_service_networking.PrivateServiceConnectConfig(
+                    enable_private_service_connect=True,
+                    project_allowlist=project_allowlist,
+                ),
+            )
+        else:
+            dedicated_serving_endpoint = (
+                gca_feature_online_store.FeatureOnlineStore.DedicatedServingEndpoint()
+            )
+
+        gapic_feature_online_store = gca_feature_online_store.FeatureOnlineStore(
+            optimized=gca_feature_online_store.FeatureOnlineStore.Optimized(),
+            dedicated_serving_endpoint=dedicated_serving_endpoint,
+        )
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature_online_store.labels = labels
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        api_client = cls._instantiate_client(location=location, credentials=credentials)
+
+        create_online_store_lro = api_client.create_feature_online_store(
+            parent=initializer.global_config.common_location_path(
+                project=project, location=location
+            ),
+            feature_online_store=gapic_feature_online_store,
+            feature_online_store_id=name,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        _LOGGER.log_create_with_lro(cls, create_online_store_lro)
+
+        created_online_store = create_online_store_lro.result()
+
+        _LOGGER.log_create_complete(cls, created_online_store, "feature_online_store")
+
+        online_store_obj = cls(
+            name=created_online_store.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return online_store_obj
+
+    @base.optional_sync()
+    def delete(self, force: bool = False, sync: bool = True) -> None:
+        """Deletes this online store.
+
+        WARNING: This deletion is permanent.
+
+        Args:
+            force:
+                If set to True, all feature views under this online store will
+                be deleted prior to online store deletion. Otherwise, deletion
+                will only succeed if the online store has no FeatureViews.
+            sync:
+                Whether to execute this deletion synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+        """
+
+        lro = getattr(self.api_client, self._delete_method)(
+            name=self.resource_name,
+            force=force,
+        )
+        _LOGGER.log_delete_with_lro(self, lro)
+        lro.result()
+        _LOGGER.log_delete_complete(self)
+
+    @property
+    def feature_online_store_type(self) -> FeatureOnlineStoreType:
+        if self._gca_resource.bigtable:
+            return FeatureOnlineStoreType.BIGTABLE
+        # Optimized is an empty proto, so self._gca_resource.optimized is always false.
+        elif hasattr(self.gca_resource, "optimized"):
+            return FeatureOnlineStoreType.OPTIMIZED
+        else:
+            raise ValueError(
+                f"Online store does not have type or is unsupported by SDK: {self._gca_resource}."
+            )
+
+    @property
+    def labels(self) -> Dict[str, str]:
+        return self._gca_resource.labels
+
+    @base.optional_sync()
+    def create_feature_view(
+        self,
+        name: str,
+        source: Union[
+            FeatureViewBigQuerySource,
+            FeatureViewVertexRagSource,
+            FeatureViewRegistrySource,
+        ],
+        labels: Optional[Dict[str, str]] = None,
+        sync_config: Optional[str] = None,
+        index_config: Optional[IndexConfig] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+        request_metadata: Optional[Sequence[Tuple[str, str]]] = None,
+        create_request_timeout: Optional[float] = None,
+        sync: bool = True,
+    ) -> FeatureView:
+        """Creates a FeatureView from a BigQuery source.
+
+        Example Usage:
+        ```
+        existing_fos = FeatureOnlineStore('my_fos')
+        new_fv = existing_fos.create_feature_view(
+                'my_fos',
+                BigQuerySource(
+                    uri='bq://my-proj/dataset/table',
+                    entity_id_columns=['entity_id'],
+                )
+        )
+        # Example for how to create an embedding FeatureView.
+        embedding_fv = existing_fos.create_feature_view(
+                'my_fos',
+                BigQuerySource(
+                    uri='bq://my-proj/dataset/table',
+                    entity_id_columns=['entity_id'],
+                )
+                index_config=IndexConfig(
+                    embedding_column="embedding",
+                    filter_column=["currency_code", "gender",
+                    crowding_column="crowding",
+                    dimentions=1536,
+                    distance_measure_type=DistanceMeasureType.SQUARED_L2_DISTANCE,
+                    algorithm_config=TreeAhConfig(),
+                )
+            )
+        ```
+        Args:
+            name: The name of the feature view.
+            source:
+                The source to load data from when a feature view sync runs.
+                Currently supports a BigQuery source, Vertex RAG source, Registry source.
+            labels:
+                The labels with user-defined metadata to organize your
+                FeatureViews.
+
+                Label keys and values can be no longer than 64 characters
+                (Unicode codepoints), can only contain lowercase letters,
+                numeric characters, underscores and dashes. International
+                characters are allowed.
+
+                See https://goo.gl/xmQnxf for more information on and examples
+                of labels. No more than 64 user labels can be associated with
+                one FeatureOnlineStore(System labels are excluded)." System
+                reserved label keys are prefixed with
+                "aiplatform.googleapis.com/" and are immutable.
+            sync_config:
+                Configures when data is to be synced/updated for this
+                FeatureView. At the end of the sync the latest feature values
+                for each entity ID of this FeatureView are made ready for online
+                serving. Example format: "TZ=America/New_York 0 9 * * *" (sync
+                daily at 9 AM EST).
+            index_config:
+                Configuration for index preparation for vector search. It
+                contains the required configurations to create an index from
+                source data, so that approximate nearest neighbor (a.k.a ANN)
+                algorithms search can be performed during online serving.
+            project:
+                Project to create feature view in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to create feature view in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to create this feature view.
+                Overrides credentials set in aiplatform.init.
+            request_metadata:
+                Strings which should be sent along with the request as metadata.
+            create_request_timeout:
+                The timeout for the create request in seconds.
+            sync:
+                Whether to execute this creation synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+
+        Returns:
+            FeatureView - the FeatureView resource object.
+        """
+        if not source:
+            raise ValueError("Please specify a valid source.")
+
+        big_query_source = None
+        vertex_rag_source = None
+        feature_registry_source = None
+
+        if isinstance(source, FeatureViewBigQuerySource):
+            if not source.uri:
+                raise ValueError("Please specify URI in BigQuery source.")
+
+            if not source.entity_id_columns:
+                raise ValueError("Please specify entity ID columns in BigQuery source.")
+
+            big_query_source = gca_feature_view.FeatureView.BigQuerySource(
+                uri=source.uri,
+                entity_id_columns=source.entity_id_columns,
+            )
+        elif isinstance(source, FeatureViewVertexRagSource):
+            if not source.uri:
+                raise ValueError("Please specify URI in Vertex RAG source.")
+
+            vertex_rag_source = gca_feature_view.FeatureView.VertexRagSource(
+                uri=source.uri,
+                rag_corpus_id=source.rag_corpus_id or None,
+            )
+        elif isinstance(source, FeatureViewRegistrySource):
+            if not source.features:
+                raise ValueError(
+                    "Please specify features in Registry Source in format `<feature_group_id>.<feature_id>`."
+                )
+            feature_group_mappings = {}
+            for feature in source.features:
+                feature_group_id, feature_id = feature.split(".")
+                if not feature_id or not feature_group_id:
+                    raise ValueError(
+                        "Please specify features in Registry Source in format `<feature_group_id>.<feature_id>`."
+                    )
+                if feature_group_id in feature_group_mappings:
+                    feature_group_mappings[feature_group_id].append(feature_id)
+                else:
+                    feature_group_mappings[feature_group_id] = [feature_id]
+            feature_groups = []
+            for feature_group_id in feature_group_mappings:
+                feature_ids = feature_group_mappings[feature_group_id]
+                feature_groups.append(
+                    gca_feature_view.FeatureView.FeatureRegistrySource.FeatureGroup(
+                        feature_group_id=feature_group_id,
+                        feature_ids=feature_ids,
+                    )
+                )
+            feature_registry_source = (
+                gca_feature_view.FeatureView.FeatureRegistrySource(
+                    feature_groups=feature_groups,
+                    project_number=source.project_number or None,
+                )
+            )
+        else:
+            raise ValueError(
+                "Only FeatureViewBigQuerySource, FeatureViewVertexRagSource and FeatureViewRegistrySource are supported sources."
+            )
+
+        gapic_feature_view = gca_feature_view.FeatureView(
+            big_query_source=big_query_source,
+            vertex_rag_source=vertex_rag_source,
+            feature_registry_source=feature_registry_source,
+            sync_config=gca_feature_view.FeatureView.SyncConfig(cron=sync_config)
+            if sync_config
+            else None,
+        )
+
+        if labels:
+            utils.validate_labels(labels)
+            gapic_feature_view.labels = labels
+
+        if request_metadata is None:
+            request_metadata = ()
+
+        if index_config:
+            gapic_feature_view.index_config = gca_feature_view.FeatureView.IndexConfig(
+                index_config.as_dict()
+            )
+
+        api_client = self.__class__._instantiate_client(
+            location=location, credentials=credentials
+        )
+
+        create_feature_view_lro = api_client.create_feature_view(
+            parent=self.resource_name,
+            feature_view=gapic_feature_view,
+            feature_view_id=name,
+            metadata=request_metadata,
+            timeout=create_request_timeout,
+        )
+
+        _LOGGER.log_create_with_lro(FeatureView, create_feature_view_lro)
+
+        created_feature_view = create_feature_view_lro.result()
+
+        _LOGGER.log_create_complete(FeatureView, created_feature_view, "feature_view")
+
+        feature_view_obj = FeatureView(
+            name=created_feature_view.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+
+        return feature_view_obj
+
+    def list_feature_views(
+        self,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> List[FeatureView]:
+        """Lists feature views under this feature online store.
+
+        Args:
+            project:
+                Project to list feature views in. If unset, the project set in
+                aiplatform.init will be used.
+            location:
+                Location to list feature views in. If not set, location set in
+                aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to list feature views. Overrides
+                credentials set in aiplatform.init.
+
+        Returns:
+            List of feature views under this feature online store.
+        """
+
+        return FeatureView.list(
+            feature_online_store_id=self.name,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/feature_view.py

@@ -0,0 +1,539 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import re
+from typing import List, Dict, Optional
+from google.cloud.aiplatform import initializer
+from google.auth import credentials as auth_credentials
+from google.cloud.aiplatform import base
+from google.cloud.aiplatform import utils
+from google.cloud.aiplatform.compat.types import (
+    feature_view_sync as gca_feature_view_sync,
+    feature_view as gca_feature_view,
+    feature_online_store_service as fos_service,
+)
+import vertexai.resources.preview.feature_store.utils as fs_utils
+
+_LOGGER = base.Logger(__name__)
+
+
+class FeatureView(base.VertexAiResourceNounWithFutureManager):
+    """Class for managing Feature View resources."""
+
+    client_class = utils.FeatureOnlineStoreAdminClientWithOverride
+
+    _resource_noun = "featureViews"
+    _getter_method = "get_feature_view"
+    _list_method = "list_feature_views"
+    _delete_method = "delete_feature_view"
+    _parse_resource_name_method = "parse_feature_view_path"
+    _format_resource_name_method = "feature_view_path"
+    _gca_resource: gca_feature_view.FeatureView
+    _online_store_client: utils.FeatureOnlineStoreClientWithOverride
+
+    _online_store_clients_with_connection_options: Dict[
+        fs_utils.ConnectionOptions, utils.FeatureOnlineStoreClientWithOverride
+    ] = None
+
+    def __init__(
+        self,
+        name: str,
+        feature_online_store_id: Optional[str] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ):
+        """Retrieves an existing managed feature view.
+
+        Args:
+            name:
+                The resource name
+                (`projects/.../locations/.../featureOnlineStores/.../featureViews/...`)
+                or ID.
+            feature_online_store_id:
+                The feature online store ID. Must be passed in if name is an ID
+                and not a resource path.
+            project:
+                Project to retrieve the feature view from. If unset, the project
+                set in aiplatform.init will be used.
+            location:
+                Location to retrieve the feature view from. If not set, location
+                set in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to retrieve this feature view.
+                Overrides credentials set in aiplatform.init.
+        """
+
+        super().__init__(
+            project=project,
+            location=location,
+            credentials=credentials,
+            resource_name=name,
+        )
+
+        if re.fullmatch(
+            r"projects/.+/locations/.+/featureOnlineStores/.+/featureViews/.+",
+            name,
+        ):
+            feature_view = name
+        else:
+            from .feature_online_store import FeatureOnlineStore
+
+            # Construct the feature view path using feature online store ID if
+            # only the feature view ID is provided.
+            if not feature_online_store_id:
+                raise ValueError(
+                    "Since feature view is not provided as a path, please specify"
+                    + " feature_online_store_id."
+                )
+
+            feature_online_store_path = utils.full_resource_name(
+                resource_name=feature_online_store_id,
+                resource_noun=FeatureOnlineStore._resource_noun,
+                parse_resource_name_method=FeatureOnlineStore._parse_resource_name,
+                format_resource_name_method=FeatureOnlineStore._format_resource_name,
+            )
+
+            feature_view = f"{feature_online_store_path}/featureViews/{name}"
+
+        self._gca_resource = self._get_gca_resource(resource_name=feature_view)
+
+    def _get_online_store_client(
+        self, connection_options: Optional[fs_utils.ConnectionOptions] = None
+    ) -> utils.FeatureOnlineStoreClientWithOverride:
+        """Return the online store client.
+
+        Also sets the `_online_store_client` attr if not set yet. Note that if
+        `connection_options` is passed in, the `_online_store_client` attr will
+        not be set - only the client will be returned. If the same
+        `connection_options` is passed in, this code will return the same
+        (cached) client as previously built.
+        """
+        if getattr(self, "_online_store_client", None):
+            return self._online_store_client
+
+        fos_name = fs_utils.get_feature_online_store_name(self.resource_name)
+        from .feature_online_store import FeatureOnlineStore
+
+        fos = FeatureOnlineStore(name=fos_name)
+
+        if connection_options:
+            # Check if we have a previously client created for these
+            # connection_options.
+            if self._online_store_clients_with_connection_options is None:
+                self._online_store_clients_with_connection_options = {}
+            if connection_options in self._online_store_clients_with_connection_options:
+                return self._online_store_clients_with_connection_options[
+                    connection_options
+                ]
+            host = connection_options.host
+
+            if isinstance(
+                connection_options.transport,
+                fs_utils.ConnectionOptions.InsecureGrpcChannel,
+            ):
+                import grpc
+                from google.cloud.aiplatform_v1.services import (
+                    feature_online_store_service as feature_online_store_service_v1,
+                )
+                from google.cloud.aiplatform_v1beta1.services import (
+                    feature_online_store_service as feature_online_store_service_v1beta1,
+                )
+
+                gapic_client_class = (
+                    utils.FeatureOnlineStoreClientWithOverride.get_gapic_client_class()
+                )
+                gapic_client_class_to_transport_class = {
+                    feature_online_store_service_v1.client.FeatureOnlineStoreServiceClient: (
+                        feature_online_store_service_v1.transports.grpc.FeatureOnlineStoreServiceGrpcTransport
+                    ),
+                    feature_online_store_service_v1beta1.client.FeatureOnlineStoreServiceClient: (
+                        feature_online_store_service_v1beta1.transports.grpc.FeatureOnlineStoreServiceGrpcTransport
+                    ),
+                }
+                if gapic_client_class not in gapic_client_class_to_transport_class:
+                    raise ValueError(
+                        f"Unexpected gapic class '{gapic_client_class}' used by internal client."
+                    )
+
+                transport_class = gapic_client_class_to_transport_class[
+                    gapic_client_class
+                ]
+
+                client = gapic_client_class(
+                    transport=transport_class(
+                        channel=grpc.insecure_channel(host + ":10002")
+                    ),
+                )
+
+                self._online_store_clients_with_connection_options[
+                    connection_options
+                ] = client
+                return client
+            else:
+                raise ValueError(
+                    f"Unsupported connection transport type, got transport: {connection_options.transport}"
+                )
+
+        if fos._gca_resource.bigtable.auto_scaling:
+            # This is Bigtable online store.
+            _LOGGER.info(f"Connecting to Bigtable online store name {fos_name}")
+            self._online_store_client = initializer.global_config.create_client(
+                client_class=utils.FeatureOnlineStoreClientWithOverride,
+                credentials=self.credentials,
+                location_override=self.location,
+            )
+            return self._online_store_client
+
+        if (
+            fos._gca_resource.dedicated_serving_endpoint.private_service_connect_config.enable_private_service_connect
+        ):
+            raise ValueError(
+                "Use `connection_options` to specify an IP address. Required for optimized online store with private service connect."
+            )
+
+        # From here, optimized serving with public endpoint.
+        if not fos._gca_resource.dedicated_serving_endpoint.public_endpoint_domain_name:
+            raise fs_utils.PublicEndpointNotFoundError(
+                "Public endpoint is not created yet for the optimized online store:"
+                f"{fos_name}. Please run sync and wait for it to complete."
+            )
+
+        _LOGGER.info(
+            f"Public endpoint for the optimized online store {fos_name} is"
+            f" {fos._gca_resource.dedicated_serving_endpoint.public_endpoint_domain_name}"
+        )
+        self._online_store_client = initializer.global_config.create_client(
+            client_class=utils.FeatureOnlineStoreClientWithOverride,
+            credentials=self.credentials,
+            location_override=self.location,
+            prediction_client=True,
+            api_path_override=fos._gca_resource.dedicated_serving_endpoint.public_endpoint_domain_name,
+        )
+        return self._online_store_client
+
+    @classmethod
+    def list(
+        cls,
+        feature_online_store_id: str,
+        filter: Optional[str] = None,
+        project: Optional[str] = None,
+        location: Optional[str] = None,
+        credentials: Optional[auth_credentials.Credentials] = None,
+    ) -> List["FeatureView"]:
+        """List all feature view under feature_online_store_id.
+
+        Example Usage:
+        ```
+        feature_views = vertexai.preview.FeatureView.list(
+            feature_online_store_id="my_fos",
+            filter=labels.label_key=label_value)
+        ```
+        Args:
+            feature_online_store_id:
+                Parentfeature online store ID.
+            filter:
+                Filter to apply on the returned feature online store.
+            project:
+                Project to use to get a list of feature views. If unset, the
+                project set in aiplatform.init will be used.
+            location:
+                Location to use to get a list feature views. If not set,
+                location set in aiplatform.init will be used.
+            credentials:
+                Custom credentials to use to get a list of feature views.
+                Overrides credentials set in aiplatform.init.
+
+        Returns:
+            List[FeatureView] - list of FeatureView resource object.
+        """
+        from .feature_online_store import FeatureOnlineStore
+
+        fos = FeatureOnlineStore(
+            name=feature_online_store_id,
+            project=project,
+            location=location,
+            credentials=credentials,
+        )
+        return cls._list(
+            filter=filter, credentials=credentials, parent=fos.resource_name
+        )
+
+    @base.optional_sync()
+    def delete(self, sync: bool = True) -> None:
+        """Deletes this feature view.
+
+        WARNING: This deletion is permanent.
+
+        Args:
+            sync:
+                Whether to execute this deletion synchronously. If False, this
+                method will be executed in concurrent Future and any downstream
+                object will be immediately returned and synced when the Future
+                has completed.
+        """
+        lro = getattr(self.api_client, self._delete_method)(name=self.resource_name)
+        _LOGGER.log_delete_with_lro(self, lro)
+        lro.result()
+        _LOGGER.log_delete_complete(self)
+
+    def sync(self) -> "FeatureViewSync":
+        """Starts an on-demand Sync for the FeatureView.
+
+        Args: None
+
+        Returns:
+            "FeatureViewSync" - FeatureViewSync instance
+        """
+        sync_method = getattr(self.api_client, self.FeatureViewSync.sync_method())
+
+        sync_request = {
+            "feature_view": self.resource_name,
+        }
+        sync_response = sync_method(request=sync_request)
+
+        return self.FeatureViewSync(name=sync_response.feature_view_sync)
+
+    def get_sync(self, name) -> "FeatureViewSync":
+        """Gets the FeatureViewSync resource for the given name.
+
+        Args:
+            name: The resource ID
+
+        Returns:
+            "FeatureViewSync" - FeatureViewSync instance
+        """
+        feature_view_path = self.resource_name
+        feature_view_sync = f"{feature_view_path}/featureViewSyncs/{name}"
+        return self.FeatureViewSync(name=feature_view_sync)
+
+    def list_syncs(
+        self,
+        filter: Optional[str] = None,
+    ) -> List["FeatureViewSync"]:
+        """List all feature view under this FeatureView.
+
+        Args:
+            parent_resource_name: Fully qualified name of the parent FeatureView
+              resource.
+            filter: Filter to apply on the returned feature online store.
+
+        Returns:
+            List[FeatureViewSync] - list of FeatureViewSync resource object.
+        """
+
+        return self.FeatureViewSync._list(
+            filter=filter, credentials=self.credentials, parent=self.resource_name
+        )
+
+    def read(
+        self,
+        key: List[str],
+        connection_options: Optional[fs_utils.ConnectionOptions] = None,
+        request_timeout: Optional[float] = None,
+    ) -> fs_utils.FeatureViewReadResponse:
+        """Read the feature values from FeatureView.
+
+          Example Usage:
+            Read feature view. Use this for Bigtable online stores and for
+            Optimized online stores that use public endpoint.
+            ```
+            data = vertexai.preview.FeatureView(
+                name='feature_view_name', feature_online_store_id='fos_name')
+                .read(key=[12345, 6789])
+                .to_dict()
+            ```
+
+            Read feature view using IP with an insecure gRPC channel. Use this
+            for optimized online stores using private service connect.
+            ```
+            data = vertexai.preview.FeatureView(
+                name='feature_view_name', feature_online_store_id='fos_name')
+                .read(
+                    key=[12345, 6789],
+                    connection_options=fs_utils.ConnectionOptions(
+                        host="<ip>",
+                        transport=fs_utils.ConnectionOptions.InsecureGrpcChannel()))
+                .to_dict()
+            ```
+        Args:
+            key: The request key to read feature values for.
+            connection_options:
+                If specified, use these options to connect to a host for sending
+                requests instead of the default
+                `<region>-aiplatform.googleapis.com` or the feature online
+                store's public endpoint.
+
+        Returns:
+            "FeatureViewReadResponse" - FeatureViewReadResponse object. It is
+            intermediate class that can be further converted by to_dict() or
+            to_proto().
+        """
+        self.wait()
+
+        online_store_client = self._get_online_store_client(
+            connection_options=connection_options
+        )
+
+        response = online_store_client.fetch_feature_values(
+            feature_view=self.resource_name,
+            data_key=fos_service.FeatureViewDataKey(
+                composite_key=fos_service.FeatureViewDataKey.CompositeKey(parts=key)
+            ),
+            timeout=request_timeout,
+        )
+        return fs_utils.FeatureViewReadResponse(response)
+
+    def search(
+        self,
+        entity_id: Optional[str] = None,
+        embedding_value: Optional[List[float]] = None,
+        neighbor_count: Optional[int] = None,
+        string_filters: Optional[
+            List[fos_service.NearestNeighborQuery.StringFilter]
+        ] = None,
+        per_crowding_attribute_neighbor_count: Optional[int] = None,
+        return_full_entity: bool = False,
+        approximate_neighbor_candidates: Optional[int] = None,
+        leaf_nodes_search_fraction: Optional[float] = None,
+        request_timeout: Optional[float] = None,
+    ) -> fs_utils.SearchNearestEntitiesResponse:
+        """Search the nearest entities from FeatureView.
+
+        Example Usage:
+        ```
+          data = vertexai.preview.FeatureView(
+              name='feature_view_name', feature_online_store_id='fos_name')
+            .search(entity_id='sample_entity')
+            .to_dict()
+        ```
+        Args:
+            entity_id: The entity id whose similar entities should be searched
+              for.
+            embedding_value: The embedding vector that be used for similar
+              search.
+            neighbor_count: The number of similar entities to be retrieved
+              from feature view for each query.
+            string_filters: The list of string filters.
+            per_crowding_attribute_neighbor_count: Crowding is a constraint on a
+            neighbor list produced by nearest neighbor search requiring that
+              no more than sper_crowding_attribute_neighbor_count of the k
+              neighbors returned have the same value of crowding_attribute.
+              It's used for improving result diversity.
+            return_full_entity: If true, return full entities including the
+              features other than embeddings.
+            approximate_neighbor_candidates: The number of neighbors to find via
+              approximate search before exact reordering is performed; if set,
+              this value must be > neighbor_count.
+            leaf_nodes_search_fraction: The fraction of the number of leaves to
+              search, set at query time allows user to tune search performance.
+              This value increase result in both search accuracy and latency
+              increase. The value should be between 0.0 and 1.0.
+
+        Returns:
+            "SearchNearestEntitiesResponse" - SearchNearestEntitiesResponse
+            object. It is intermediate class that can be further converted by
+            to_dict() or to_proto()
+        """
+        self.wait()
+        if entity_id:
+            embedding = None
+        elif embedding_value:
+            embedding = fos_service.NearestNeighborQuery.Embedding(
+                value=embedding_value
+            )
+        else:
+            raise ValueError(
+                "Either entity_id or embedding_value needs to be provided for search."
+            )
+        response = self._get_online_store_client().search_nearest_entities(
+            request=fos_service.SearchNearestEntitiesRequest(
+                feature_view=self.resource_name,
+                query=fos_service.NearestNeighborQuery(
+                    entity_id=entity_id,
+                    embedding=embedding,
+                    neighbor_count=neighbor_count,
+                    string_filters=string_filters,
+                    per_crowding_attribute_neighbor_count=per_crowding_attribute_neighbor_count,  # pylint: disable=line-too-long
+                    parameters=fos_service.NearestNeighborQuery.Parameters(
+                        approximate_neighbor_candidates=approximate_neighbor_candidates,
+                        leaf_nodes_search_fraction=leaf_nodes_search_fraction,
+                    ),
+                ),
+                return_full_entity=return_full_entity,
+            ),
+            timeout=request_timeout,
+        )
+        return fs_utils.SearchNearestEntitiesResponse(response)
+
+    class FeatureViewSync(base.VertexAiResourceNounWithFutureManager):
+        """Class for managing Feature View Sync resources."""
+
+        client_class = utils.FeatureOnlineStoreAdminClientWithOverride
+
+        _resource_noun = "featureViewSyncs"
+        _getter_method = "get_feature_view_sync"
+        _list_method = "list_feature_view_syncs"
+        _delete_method = "delete_feature_view"
+        _sync_method = "sync_feature_view"
+        _parse_resource_name_method = "parse_feature_view_sync_path"
+        _format_resource_name_method = "feature_view_sync_path"
+        _gca_resource: gca_feature_view_sync.FeatureViewSync
+
+        def __init__(
+            self,
+            name: str,
+            project: Optional[str] = None,
+            location: Optional[str] = None,
+            credentials: Optional[auth_credentials.Credentials] = None,
+        ):
+            """Retrieves an existing managed feature view sync.
+
+            Args:
+                name: The resource name
+                  (`projects/.../locations/.../featureOnlineStores/.../featureViews/.../featureViewSyncs/...`)
+                project: Project to retrieve the feature view from. If unset, the
+                  project set in aiplatform.init will be used.
+                location: Location to retrieve the feature view from. If not set,
+                  location set in aiplatform.init will be used.
+                credentials: Custom credentials to use to retrieve this feature view.
+                  Overrides credentials set in aiplatform.init.
+            """
+            super().__init__(
+                project=project,
+                location=location,
+                credentials=credentials,
+                resource_name=name,
+            )
+
+            if not re.fullmatch(
+                r"projects/.+/locations/.+/featureOnlineStores/.+/featureViews/.+/featureViewSyncs/.+",
+                name,
+            ):
+                raise ValueError(
+                    "name need to specify the fully qualified"
+                    + " feature_view_sync resource path."
+                )
+
+            self._gca_resource = getattr(self.api_client, self._getter_method)(
+                name=name, retry=base._DEFAULT_RETRY
+            )
+
+        @classmethod
+        def sync_method(cls) -> str:
+            """Returns the sync method."""
+            return cls._sync_method

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/offline_store.py

@@ -0,0 +1,291 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import datetime
+import re
+
+from typing import Optional, List, Tuple, Union, TYPE_CHECKING
+from google.auth import credentials as auth_credentials
+from vertexai.resources.preview.feature_store import (
+    FeatureGroup,
+    Feature,
+)
+from google.cloud.aiplatform import initializer, __version__
+
+from . import _offline_store_impl as impl
+
+
+if TYPE_CHECKING:
+    try:
+        import bigframes
+    except ImportError:
+        bigframes = None
+
+    try:
+        import pandas as pd
+    except ImportError:
+        pd = None
+
+
+def _try_import_bigframes():
+    """Try to import `bigframes` and return it if successful - otherwise raise an import error."""
+    try:
+        import bigframes
+        import bigframes.pandas
+
+        return bigframes
+    except ImportError as exc:
+        raise ImportError(
+            "`bigframes` is not installed but required for this functionality."
+        ) from exc
+
+
+def _get_feature_group_from_feature(
+    feature: Feature, credentials: auth_credentials.Credentials
+):
+    """Given a feature, return the feature group resource."""
+    result = re.fullmatch(
+        r"projects/(?P<project>.+)/locations/(?P<location>.+)/featureGroups/(?P<feature_group>.+)/features/.+",
+        feature.resource_name,
+    )
+
+    if not result:
+        raise ValueError("Couldn't find feature group in feature.")
+
+    project = feature.project
+    location = feature.location
+    feature_group = result.group("feature_group")
+
+    return FeatureGroup(
+        feature_group, project=project, location=location, credentials=credentials
+    )
+
+
+def _extract_feature_from_str_repr(
+    str_feature: str, credentials: auth_credentials.Credentials
+) -> Tuple[FeatureGroup, Feature]:
+    """Given a feature in string representation, return the feature and feature group."""
+    # TODO: compile expr + place it in a constant
+    result = re.fullmatch(
+        r"((?P<project>.*)\.)?(?P<feature_group>.*)\.(?P<feature>.*)",
+        str_feature,
+    )
+    if not result:
+        raise ValueError(
+            f"Feature '{str_feature}' is a string but not in expected format 'feature_group.feature' or 'project.feature_group.feature'."
+        )
+
+    feature_group = FeatureGroup(
+        result.group("feature_group"),
+        project=result.group("project"),  # None if no match.
+        credentials=credentials,
+    )
+    feature = feature_group.get_feature(result.group("feature"))
+
+    return (feature_group, feature)
+
+
+def _feature_to_data_source(
+    feature_group: FeatureGroup, feature: Feature
+) -> impl.DataSource:
+    qualifying_name = f"{feature_group.name}__{feature.name}"
+    gbq_column = feature.version_column_name
+    assert gbq_column
+
+    column_name = feature.name
+    assert column_name
+
+    timestamp_column = "feature_timestamp"
+
+    # TODO: Expose entity_id_columns as a property in FeatureGroup
+    entity_id_columns = feature_group._gca_resource.big_query.entity_id_columns
+    assert entity_id_columns
+
+    bq_uri = feature_group._gca_resource.big_query.big_query_source.input_uri
+    assert bq_uri
+
+    fully_qualified_table = bq_uri.lstrip("bq://")
+    assert fully_qualified_table
+
+    query = (
+        f"SELECT\n"
+        f'  {", ".join(entity_id_columns)},\n'
+        f"  {gbq_column} AS {column_name},\n"
+        f"  {timestamp_column}\n"
+        f"FROM {fully_qualified_table}"
+    )
+
+    return impl.DataSource(
+        qualifying_name=qualifying_name,
+        sql=query,
+        data_columns=[column_name],
+        # TODO: this will be parameterized in the future
+        timestamp_column=timestamp_column,
+        entity_id_columns=entity_id_columns,
+    )
+
+
+class _DataFrameToBigQueryDataFramesConverter:
+    @classmethod
+    def to_bigquery_dataframe(
+        cls, df: "pd.DataFrame", session: "Optional[bigframes.session.Session]" = None
+    ) -> "bigframes.pandas.DataFrame":
+        bigframes = _try_import_bigframes()
+        return bigframes.pandas.DataFrame(data=df, session=session)
+
+
+def fetch_historical_feature_values(
+    entity_df: "bigframes.pandas.DataFrame",
+    # TODO: Add support for FeatureView | FeatureGroup | bigframes.pandas.DataFrame
+    features: List[Union[str, Feature]],
+    # TODO: Add support for feature_age_threshold
+    feature_age_threshold: Optional[datetime.timedelta] = None,
+    dry_run: bool = False,
+    project: Optional[str] = None,
+    location: Optional[str] = None,
+    credentials: Optional[auth_credentials.Credentials] = None,
+) -> "Union[bigframes.pandas.DataFrame, None]":
+    """Fetch historical data at the timestamp specified for each entity.
+
+    This runs a Point-In-Time Lookup (PITL) query in BigQuery across all
+    features and returns the historical feature values. Feature data will be
+    joined by matching their entity_id_column(s) with corresponding columns in
+    the entity data frame.
+
+    Args:
+      entity_df:
+        An entity DataFrame where one/multiple columns have entity ID.
+        One column should have a timestamp (used for feature lookup). Other
+        columns may have feature data. Entity IDs may be repeated with
+        different timestamp values (in the timestamp column) to lookup data for
+        entities at different points in time.
+      features:
+        Feature data will be joined with the entity data frame.
+         * If `str` is given use `project.feature_group.feature` as the format.
+          `project_id.feature_group_id.feature_id` may be used if features are
+          in another project.
+         * If `FeatureView` is given, the *sources* of the FeatureView will be
+           used - but data will be read from the backing BigQuery table.
+      feature_age_threshold:
+        How far back from the timestamp to look for features values. If no
+        feature values are found, empty/null value will be populated.
+      dry_run:
+        Build the Point-In-Time Lookup (PITL) query but don't run it. The PITL
+        query will be printed to stdout.
+      project:
+        The project to use for feature lookup and running the Point-In-Time
+        Lookup (PITL) query in BigQuery. If unset, the project set in
+        aiplatform.init will be used.
+      location:
+        The location to use for feature lookup and running the Point-In-Time
+        Lookup (PITL) query in BigQuery. If unset, the project set in
+        aiplatform.init will be used.
+      credentials:
+        Custom credentials to use for feature lookup and running the
+        Point-In-Time Lookup (PITL) query in BigQuery. Overrides credentials
+        set in aiplatform.init.
+
+    Returns:
+      A `bigframes.pandas.DataFrame` with the historical feature values. `None`
+      if in `dry_run` mode.
+    """
+
+    bigframes = _try_import_bigframes()
+    project = project or initializer.global_config.project
+    location = location or initializer.global_config.location
+    credentials = credentials or initializer.global_config.credentials
+    application_name = (
+        f"vertexai-offline-store/{__version__}+fetch-historical-feature-values"
+    )
+    session_options = bigframes.BigQueryOptions(
+        credentials=credentials,
+        project=project,
+        location=location,
+        application_name=application_name,
+    )
+    session = bigframes.connect(session_options)
+
+    if feature_age_threshold is not None:
+        raise NotImplementedError("feature_age_threshold is not yet supported.")
+
+    if not features:
+        raise ValueError("Please specify a non-empty list of features.")
+
+    # Convert to bigframe if needed.
+    if not isinstance(entity_df, bigframes.pandas.DataFrame):
+        entity_df = _DataFrameToBigQueryDataFramesConverter.to_bigquery_dataframe(
+            df=entity_df,
+            session=session,
+        )
+
+    # Ensure one timestamp column is present in the entity DataFrame.
+    ts_cols = entity_df.select_dtypes(include=["datetime"]).columns
+    if len(ts_cols) > 1:
+        # TODO: Support multiple timestamp columns by specifying feature_timestamp column in an override.
+        raise ValueError(
+            'Multiple timestamp columns ("datetime" dtype) found in entity DataFrame. '
+            "Only one timestamp column is allowed. "
+            f"Timestamp columns: {', '.join([col for col in ts_cols])}"
+        )
+    elif len(ts_cols) == 0:
+        raise ValueError(
+            'No timestamp column ("datetime" dtype) found in entity DataFrame.'
+        )
+    entity_df_ts_col = ts_cols[0]
+    entity_df_non_ts_cols = [c for c in entity_df.columns if c != entity_df_ts_col]
+    entity_data_source = impl.DataSource(
+        qualifying_name="entity_df",
+        sql=entity_df.sql,
+        data_columns=entity_df_non_ts_cols,
+        timestamp_column=entity_df_ts_col,
+    )
+
+    feature_data: List[impl.DataSource] = []
+    for feature in features:
+        if isinstance(feature, Feature):
+            feature_group = _get_feature_group_from_feature(feature, credentials)
+            feature_data.append(_feature_to_data_source(feature_group, feature))
+        elif isinstance(feature, str):
+            feature_group, feature = _extract_feature_from_str_repr(
+                feature, credentials
+            )
+            feature_data.append(_feature_to_data_source(feature_group, feature))
+        else:
+            raise ValueError(
+                f"Unsupported feature type {type(feature)} found in feature list. Feature: {feature}"
+            )
+
+    # TODO: Verify `feature_data`.
+    #  * Ensure that qualifying_names are not interfering.
+    #  * Ensure that feature names are not interfering.
+    #  * Ensure that entity id columns of all features are present in the entity DF.
+
+    query = impl.render_pitl_query(
+        entity_data=entity_data_source,
+        feature_data=feature_data,
+    )
+
+    if dry_run:
+        print("--- Dry run mode: PITL QUERY BEGIN ---")
+        print(query)
+        print("--- Dry run mode: PITL QUERY END ---")
+        return None
+
+    return session.read_gbq_query(
+        query,
+        index_col=bigframes.enums.DefaultIndexKind.NULL,
+    )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/feature_store/utils.py

@@ -0,0 +1,233 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import abc
+from dataclasses import dataclass
+from dataclasses import field
+import enum
+from typing import Any, Dict, List, Optional, Union
+from google.cloud.aiplatform.compat.types import (
+    feature_online_store_service as fos_service,
+)
+import proto
+from typing_extensions import override
+
+
+def get_feature_online_store_name(online_store_name: str) -> str:
+    """Extract Feature Online Store's name from FeatureView's full resource name.
+
+    Args:
+        online_store_name: Full resource name is projects/project_number/
+          locations/us-central1/featureOnlineStores/fos_name/featureViews/fv_name
+
+    Returns:
+        str: feature online store name.
+    """
+    arr = online_store_name.split("/")
+    return arr[5]
+
+
+class PublicEndpointNotFoundError(RuntimeError):
+    """Public endpoint has not been created yet."""
+
+
+@dataclass
+class FeatureViewBigQuerySource:
+    uri: str
+    entity_id_columns: List[str]
+
+
+@dataclass
+class FeatureViewVertexRagSource:
+    uri: str
+    rag_corpus_id: Optional[str] = None
+
+
+@dataclass
+class FeatureViewRegistrySource:
+    """Configuration options for Feature View being registered with Feature Registry features.
+
+    Attributes:
+        features : Use `<feature_group_id>.<feature_id>` as
+          the format for each feature.
+        project_number : Optional. The project number of the project that owns the
+          Feature Registry if in a different project.
+    """
+
+    features: List[str]
+    project_number: Optional[int] = None
+
+
+@dataclass(frozen=True)
+class ConnectionOptions:
+    """Represents connection options used for sending RPCs to the online store."""
+
+    @dataclass(frozen=True)
+    class InsecureGrpcChannel:
+        """Use an insecure gRPC channel to connect to the host."""
+
+        pass
+
+    host: str  # IP address or DNS.
+    transport: Union[
+        InsecureGrpcChannel
+    ]  # Currently only insecure gRPC channel is supported.
+
+    def __eq__(self, other):
+        if self.host != other.host:
+            return False
+
+        if isinstance(self.transport, ConnectionOptions.InsecureGrpcChannel):
+            # Insecure grpc channel has no other parameters to check.
+            if isinstance(other.transport, ConnectionOptions.InsecureGrpcChannel):
+                return True
+
+            # Otherwise, can't compare against a different transport type.
+            raise ValueError(
+                f"Transport '{self.transport}' cannot be compared to transport '{other.transport}'."
+            )
+
+        # Currently only InsecureGrpcChannel is supported.
+        raise ValueError(f"Unsupported transport supplied: {self.transport}")
+
+
+@dataclass
+class FeatureViewReadResponse:
+    _response: fos_service.FetchFeatureValuesResponse
+
+    def __init__(self, response: fos_service.FetchFeatureValuesResponse):
+        self._response = response
+
+    def to_dict(self) -> Dict[str, Any]:
+        return proto.Message.to_dict(self._response.key_values)
+
+    def to_proto(self) -> fos_service.FetchFeatureValuesResponse:
+        return self._response
+
+
+@dataclass
+class SearchNearestEntitiesResponse:
+    _response: fos_service.SearchNearestEntitiesResponse
+
+    def __init__(self, response: fos_service.SearchNearestEntitiesResponse):
+        self._response = response
+
+    def to_dict(self) -> Dict[str, Any]:
+        return proto.Message.to_dict(self._response.nearest_neighbors)
+
+    def to_proto(self) -> fos_service.SearchNearestEntitiesResponse:
+        return self._response
+
+
+class DistanceMeasureType(enum.Enum):
+    """The distance measure used in nearest neighbor search."""
+
+    DISTANCE_MEASURE_TYPE_UNSPECIFIED = 0
+    # Euclidean (L_2) Distance.
+    SQUARED_L2_DISTANCE = 1
+    # Cosine Distance. Defined as 1 - cosine similarity.
+    COSINE_DISTANCE = 2
+    # Dot Product Distance. Defined as a negative of the dot product.
+    DOT_PRODUCT_DISTANCE = 3
+
+
+class AlgorithmConfig(abc.ABC):
+    """Base class for configuration options for matching algorithm."""
+
+    def as_dict(self) -> Dict:
+        """Returns the configuration as a dictionary.
+
+        Returns:
+            Dict[str, Any]
+        """
+        pass
+
+
+@dataclass
+class TreeAhConfig(AlgorithmConfig):
+    """Configuration options for using the tree-AH algorithm (Shallow tree + Asymmetric Hashing).
+
+    Please refer to this paper for more details: https://arxiv.org/abs/1908.10396
+
+    Args:
+        leaf_node_embedding_count (int): Optional. Number of embeddings on each
+          leaf node. The default value is 1000 if not set.
+    """
+
+    leaf_node_embedding_count: Optional[int] = None
+
+    @override
+    def as_dict(self) -> Dict:
+        return {"leaf_node_embedding_count": self.leaf_node_embedding_count}
+
+
+@dataclass
+class BruteForceConfig(AlgorithmConfig):
+    """Configuration options for using brute force search.
+
+    It simply implements the standard linear search in the database for each
+    query.
+    """
+
+    @override
+    def as_dict(self) -> Dict[str, Any]:
+        return {"bruteForceConfig": {}}
+
+
+@dataclass
+class IndexConfig:
+    """Configuration options for the Vertex FeatureView for embedding."""
+
+    embedding_column: str
+    dimensions: int
+    algorithm_config: AlgorithmConfig = field(default_factory=TreeAhConfig())
+    filter_columns: Optional[List[str]] = None
+    crowding_column: Optional[str] = None
+    distance_measure_type: Optional[DistanceMeasureType] = None
+
+    def as_dict(self) -> Dict[str, Any]:
+        """Returns the configuration as a dictionary.
+
+        Returns:
+            Dict[str, Any]
+        """
+        config = {
+            "embedding_column": self.embedding_column,
+            "embedding_dimension": self.dimensions,
+        }
+        if self.distance_measure_type is not None:
+            config["distance_measure_type"] = self.distance_measure_type.value
+        if self.filter_columns is not None:
+            config["filter_columns"] = self.filter_columns
+        if self.crowding_column is not None:
+            config["crowding_column"] = self.crowding_column
+
+        if isinstance(self.algorithm_config, TreeAhConfig):
+            config["tree_ah_config"] = self.algorithm_config.as_dict()
+        else:
+            config["brute_force_config"] = self.algorithm_config.as_dict()
+        return config
+
+
+@dataclass
+class FeatureGroupBigQuerySource:
+    """BigQuery source for the Feature Group."""
+
+    # The URI for the BigQuery table/view.
+    uri: str
+    # The entity ID columns. If not specified, defaults to ['entity_id'].
+    entity_id_columns: Optional[List[str]] = None

.venv/lib/python3.10/site-packages/vertexai/resources/preview/ml_monitoring/__init__.py

@@ -0,0 +1,26 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2022 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from vertexai.resources.preview.ml_monitoring.model_monitors import (
+    ModelMonitor,
+    ModelMonitoringJob,
+)
+
+__all__ = (
+    "ModelMonitor",
+    "ModelMonitoringJob",
+)

.venv/lib/python3.10/site-packages/vertexai/resources/preview/ml_monitoring/spec/__init__.py

@@ -0,0 +1,46 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2022 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from vertexai.resources.preview.ml_monitoring.spec.notification import (
+    NotificationSpec,
+)
+from vertexai.resources.preview.ml_monitoring.spec.objective import (
+    FeatureAttributionSpec,
+    DataDriftSpec,
+    MonitoringInput,
+    ObjectiveSpec,
+    TabularObjective,
+)
+from vertexai.resources.preview.ml_monitoring.spec.output import (
+    OutputSpec,
+)
+from vertexai.resources.preview.ml_monitoring.spec.schema import (
+    FieldSchema,
+    ModelMonitoringSchema,
+)
+
+__all__ = (
+    "NotificationSpec",
+    "OutputSpec",
+    "ObjectiveSpec",
+    "FeatureAttributionSpec",
+    "DataDriftSpec",
+    "MonitoringInput",
+    "TabularObjective",
+    "FieldSchema",
+    "ModelMonitoringSchema",
+)

.venv/lib/python3.10/site-packages/vertexai/resources/preview/ml_monitoring/spec/notification.py

@@ -0,0 +1,74 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2022 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from typing import Optional, List
+from google.cloud.aiplatform.compat.types import (
+    model_monitoring_spec_v1beta1 as model_monitoring_spec,
+)
+
+
+class NotificationSpec:
+    """Initializer for NotificationSpec.
+
+    Args:
+        user_emails (List[str]):
+            Optional. The email addresses to send the alert to.
+        notification_channels (List[str]):
+            Optional. The notification channels to send the alert to.
+            Format: ``projects/{project}/notificationChannels/{channel}``
+        enable_cloud_logging (bool):
+            Optional. If dump the anomalies to Cloud Logging. The anomalies will
+            be put to json payload. This can be further sinked to Pub/Sub or any
+            other services supported by Cloud Logging.
+    """
+
+    def __init__(
+        self,
+        user_emails: Optional[List[str]] = None,
+        notification_channels: Optional[List[str]] = None,
+        enable_cloud_logging: Optional[bool] = False,
+    ):
+        self.user_emails = user_emails
+        self.notification_channels = notification_channels
+        self.enable_cloud_logging = enable_cloud_logging
+
+    def _as_proto(self) -> model_monitoring_spec.ModelMonitoringNotificationSpec:
+        """Converts ModelMonitoringNotificationSpec to a proto message.
+
+        Returns:
+           The GAPIC representation of the notification alert config.
+        """
+        user_email_config = None
+        if self.user_emails is not None:
+            user_email_config = (
+                model_monitoring_spec.ModelMonitoringNotificationSpec.EmailConfig(
+                    user_emails=self.user_emails
+                )
+            )
+        user_notification_channel_config = []
+        if self.notification_channels:
+            for notification_channel in self.notification_channels:
+                user_notification_channel_config.append(
+                    model_monitoring_spec.ModelMonitoringNotificationSpec.NotificationChannelConfig(
+                        notification_channel=notification_channel
+                    )
+                )
+        return model_monitoring_spec.ModelMonitoringNotificationSpec(
+            email_config=user_email_config,
+            notification_channel_configs=user_notification_channel_config,
+            enable_cloud_logging=self.enable_cloud_logging,
+        )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/ml_monitoring/spec/objective.py

@@ -0,0 +1,522 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2022 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from typing import Dict, List, Optional
+
+from google.cloud.aiplatform.compat.types import (
+    explanation_v1beta1 as explanation,
+    machine_resources_v1beta1 as machine_resources,
+    model_monitoring_alert_v1beta1 as model_monitoring_alert,
+    model_monitoring_spec_v1beta1 as model_monitoring_spec,
+)
+
+from google.protobuf import timestamp_pb2
+from google.type import interval_pb2
+
+
+TF_RECORD = "tf-record"
+CSV = "csv"
+JSONL = "jsonl"
+JENSEN_SHANNON_DIVERGENCE = "jensen_shannon_divergence"
+L_INFINITY = "l_infinity"
+SUPPORTED_NUMERIC_METRICS = [JENSEN_SHANNON_DIVERGENCE]
+SUPPORTED_CATEGORICAL_METRICS = [JENSEN_SHANNON_DIVERGENCE, L_INFINITY]
+
+
+class DataDriftSpec:
+    """Data drift monitoring spec.
+
+    Data drift measures the distribution distance between the current dataset
+    and a baseline dataset. A typical use case is to detect data drift between
+    the recent production serving dataset and the training dataset, or to
+    compare the recent production dataset with a dataset from a previous period.
+
+    Example:
+        feature_drift_spec=DataDriftSpec(
+                features=["feature1"]
+                categorical_metric_type="l_infinity",
+                numeric_metric_type="jensen_shannon_divergence",
+                default_categorical_alert_threshold=0.01,
+                default_numeric_alert_threshold=0.02,
+                feature_alert_thresholds={"feature1":0.02, "feature2":0.01},
+        )
+
+    Attributes:
+        features (List[str]):
+            Optional. Feature names / Prediction output names interested in
+            monitoring. These should be a subset of the input feature names or
+            prediction output names specified in the monitoring schema.
+            If not specified, all features / prediction outputs outlied in the
+            monitoring schema will be used.
+        categorical_metric_type (str):
+            Optional. Supported metrics type: l_infinity, jensen_shannon_divergence
+        numeric_metric_type (str):
+            Optional. Supported metrics type: jensen_shannon_divergence
+        default_categorical_alert_threshold (float):
+            Optional. Default alert threshold for all the categorical features.
+        default_numeric_alert_threshold (float):
+            Optional. Default alert threshold for all the numeric features.
+        feature_alert_thresholds (Dict[str, float]):
+            Optional. Per feature alert threshold will override default alert
+            threshold.
+    """
+
+    def __init__(
+        self,
+        features: Optional[List[str]] = None,
+        categorical_metric_type: Optional[str] = L_INFINITY,
+        numeric_metric_type: Optional[str] = JENSEN_SHANNON_DIVERGENCE,
+        default_categorical_alert_threshold: Optional[float] = None,
+        default_numeric_alert_threshold: Optional[float] = None,
+        feature_alert_thresholds: Optional[Dict[str, float]] = None,
+    ):
+        self.features = features
+        self.categorical_metric_type = categorical_metric_type
+        self.numeric_metric_type = numeric_metric_type
+        self.default_categorical_alert_threshold = default_categorical_alert_threshold
+        self.default_numeric_alert_threshold = default_numeric_alert_threshold
+        self.feature_alert_thresholds = feature_alert_thresholds
+
+    def _as_proto(
+        self,
+    ) -> model_monitoring_spec.ModelMonitoringObjectiveSpec.DataDriftSpec:
+        """Converts DataDriftSpec to a proto message.
+
+        Returns:
+           The GAPIC representation of the data drift spec.
+        """
+        user_default_categorical_alert_threshold = None
+        user_default_numeric_alert_threshold = None
+        user_alert_thresholds = None
+        user_features = None
+        if self.numeric_metric_type not in SUPPORTED_NUMERIC_METRICS:
+            raise ValueError(
+                f"The numeric metric type is not supported"
+                f" {self.numeric_metric_type}"
+            )
+        user_numeric_metric_type = self.numeric_metric_type
+        if self.categorical_metric_type not in SUPPORTED_CATEGORICAL_METRICS:
+            raise ValueError(
+                f"The categorical metric type is not supported"
+                f" {self.categorical_metric_type}"
+            )
+        user_categorical_metric_type = self.categorical_metric_type
+        if self.default_categorical_alert_threshold:
+            user_default_categorical_alert_threshold = (
+                model_monitoring_alert.ModelMonitoringAlertCondition(
+                    threshold=self.default_categorical_alert_threshold
+                )
+            )
+        if self.default_numeric_alert_threshold:
+            user_default_numeric_alert_threshold = (
+                model_monitoring_alert.ModelMonitoringAlertCondition(
+                    threshold=self.default_numeric_alert_threshold
+                )
+            )
+        if self.feature_alert_thresholds:
+            user_alert_thresholds = {}
+            for feature in self.feature_alert_thresholds:
+                user_alert_thresholds.update(
+                    {
+                        feature: model_monitoring_alert.ModelMonitoringAlertCondition(
+                            threshold=self.feature_alert_thresholds[feature]
+                        )
+                    }
+                )
+        if self.features:
+            user_features = self.features
+        return model_monitoring_spec.ModelMonitoringObjectiveSpec.DataDriftSpec(
+            default_categorical_alert_condition=user_default_categorical_alert_threshold,
+            default_numeric_alert_condition=user_default_numeric_alert_threshold,
+            categorical_metric_type=user_categorical_metric_type,
+            numeric_metric_type=user_numeric_metric_type,
+            feature_alert_conditions=user_alert_thresholds,
+            features=user_features,
+        )
+
+
+class FeatureAttributionSpec:
+    """Feature attribution spec.
+
+    Example:
+        feature_attribution_spec=FeatureAttributionSpec(
+                features=["feature1"]
+                default_alert_threshold=0.01,
+                feature_alert_thresholds={"feature1":0.02, "feature2":0.01},
+                batch_dedicated_resources=BatchDedicatedResources(
+                    starting_replica_count=1,
+                    max_replica_count=2,
+                    machine_spec=my_machine_spec,
+                ),
+        )
+
+    Attributes:
+        features (List[str]):
+            Optional. Input feature names interested in monitoring. These should
+            be a subset of the input feature names specified in the monitoring
+            schema.
+            If not specified, all features outlied in the monitoring schema will
+            be used.
+        default_alert_threshold (float):
+            Optional. Default alert threshold for all the features.
+        feature_alert_thresholds (Dict[str, float]):
+            Optional. Per feature alert threshold will override default alert
+            threshold.
+        batch_dedicated_resources (machine_resources.BatchDedicatedResources):
+            Optional. The config of resources used by the Model Monitoring during
+            the batch explanation for non-AutoML models. If not set, `n1-standard-2`
+            machine type will be used by default.
+    """
+
+    def __init__(
+        self,
+        features: Optional[List[str]] = None,
+        default_alert_threshold: Optional[float] = None,
+        feature_alert_thresholds: Optional[Dict[str, float]] = None,
+        batch_dedicated_resources: Optional[
+            machine_resources.BatchDedicatedResources
+        ] = None,
+    ):
+        self.features = features
+        self.default_alert_threshold = default_alert_threshold
+        self.feature_alert_thresholds = feature_alert_thresholds
+        self.batch_dedicated_resources = batch_dedicated_resources
+
+    def _as_proto(
+        self,
+    ) -> model_monitoring_spec.ModelMonitoringObjectiveSpec.FeatureAttributionSpec:
+        """Converts FeatureAttributionSpec to a proto message.
+
+        Returns:
+           The GAPIC representation of the feature attribution spec.
+        """
+        user_default_alert_threshold = None
+        user_alert_thresholds = None
+        user_features = None
+        if self.default_alert_threshold:
+            user_default_alert_threshold = (
+                model_monitoring_alert.ModelMonitoringAlertCondition(
+                    threshold=self.default_alert_threshold
+                )
+            )
+        if self.feature_alert_thresholds:
+            user_alert_thresholds = {}
+            for feature in self.feature_alert_thresholds:
+                user_alert_thresholds.update(
+                    {
+                        feature: model_monitoring_alert.ModelMonitoringAlertCondition(
+                            threshold=self.feature_alert_thresholds[feature]
+                        )
+                    }
+                )
+        if self.features:
+            user_features = self.features
+        return (
+            model_monitoring_spec.ModelMonitoringObjectiveSpec.FeatureAttributionSpec(
+                default_alert_condition=user_default_alert_threshold,
+                feature_alert_conditions=user_alert_thresholds,
+                features=user_features,
+                batch_explanation_dedicated_resources=self.batch_dedicated_resources,
+            )
+        )
+
+
+class MonitoringInput:
+    """Model monitoring data input spec.
+
+    Attributes:
+        vertex_dataset (str):
+            Optional. Resource name of the Vertex AI managed dataset.
+            Format: ``projects/{project}/locations/{location}/datasets/{dataset}``
+            At least one source of dataset should be provided, and if one of the
+            fields is set, no need to set other sources
+            (vertex_dataset, gcs_uri, table_uri, query, batch_prediction_job,
+            endpoints).
+        gcs_uri (str):
+            Optional. Google Cloud Storage URI to the input file(s). May contain
+            wildcards.
+        data_format (str):
+            Optional. Data format of Google Cloud Storage file(s). Should be
+            provided if a gcs_uri is set.
+            Supported formats:
+                "csv", "jsonl", "tf-record"
+        table_uri (str):
+            Optonal. BigQuery URI to a table, up to 2000 characters long.
+            All the columns in the table will be selected. Accepted forms:
+
+            -  BigQuery path. For example:
+                ``bq://projectId.bqDatasetId.bqTableId``.
+        query (str):
+            Optional. Standard SQL for BigQuery to be used instead of the
+            ``table_uri``.
+        timestamp_field (str):
+            Optional. The timestamp field in the dataset.
+            the ``timestamp_field`` must be specified if you'd like to use
+            ``start_time``, ``end_time``, ``offset`` or ``window``.
+            If you use ``query`` to specify the dataset, make sure the
+            ``timestamp_field`` is in the selection fields.
+        batch_prediction_job (str):
+            Optional. Vertex AI Batch Prediction Job resource name.
+            Format: ``projects/{project}/locations/{location}/batchPredictionJobs/{batch_prediction_job}``
+        endpoints (List[str]):
+            Optional. List of Vertex AI Endpoint resource names.
+            Format: ``projects/{project}/locations/{location}/endpoints/{endpoint}``
+        start_time (timestamp_pb2.Timestamp):
+            Optional. Inclusive start of the time interval for which results
+            should be returned. Should be set together with ``end_time``.
+        end_time (timestamp_pb2.Timestamp):
+            Optional. Exclusive end of the time interval for which results
+            should be returned. Should be set together with ``start_time`.`
+        offset (str):
+            Optional. Offset is the time difference from the cut-off time.
+            For scheduled jobs, the cut-off time is the scheduled time.
+            For non-scheduled jobs, it's the time when the job was created.
+            Currently we support the following format:
+            'w|W': Week, 'd|D': Day, 'h|H': Hour
+            E.g. '1h' stands for 1 hour, '2d' stands for 2 days.
+        window (str):
+            Optional. Window refers to the scope of data selected for analysis.
+            It allows you to specify the quantity of data you wish to examine.
+            It refers to the data time window prior to the cut-off time or the
+            cut-off time minus the offset.
+            Currently we support the following format:
+            'w|W': Week, 'd|D': Day, 'h|H': Hour
+            E.g. '1h' stands for 1 hour, '2d' stands for 2 days.
+    """
+
+    def __init__(
+        self,
+        vertex_dataset: Optional[str] = None,
+        gcs_uri: Optional[str] = None,
+        data_format: Optional[str] = None,
+        table_uri: Optional[str] = None,
+        query: Optional[str] = None,
+        timestamp_field: Optional[str] = None,
+        batch_prediction_job: Optional[str] = None,
+        endpoints: Optional[List[str]] = None,
+        start_time: Optional[timestamp_pb2.Timestamp] = None,
+        end_time: Optional[timestamp_pb2.Timestamp] = None,
+        offset: Optional[str] = None,
+        window: Optional[str] = None,
+    ):
+        self.vertex_dataset = vertex_dataset
+        self.gcs_uri = gcs_uri
+        self.data_format = data_format
+        self.table_uri = table_uri
+        self.query = query
+        self.timestamp_field = timestamp_field
+        self.batch_prediction_job = batch_prediction_job
+        self.endpoints = endpoints
+        self.start_time = start_time
+        self.end_time = end_time
+        self.offset = offset
+        self.window = window
+
+    def _as_proto(self) -> model_monitoring_spec.ModelMonitoringInput:
+        """Converts ModelMonitoringInput to a proto message.
+
+        Returns:
+           The GAPIC representation of the model monitoring input.
+        """
+        user_time_interval = None
+        user_time_spec = None
+        if self.offset or self.window:
+            user_time_spec = model_monitoring_spec.ModelMonitoringInput.TimeOffset(
+                offset=self.offset if self.offset else None,
+                window=self.window if self.window else None,
+            )
+        elif self.start_time or self.end_time:
+            user_time_interval = interval_pb2.Interval(
+                start_time=self.start_time if self.start_time else None,
+                end_time=self.end_time if self.end_time else None,
+            )
+        if self.vertex_dataset or self.gcs_uri or self.table_uri or self.query:
+            user_vertex_dataset = None
+            user_gcs_source = None
+            user_bigquery_source = None
+            if self.vertex_dataset:
+                user_vertex_dataset = self.vertex_dataset
+            elif self.gcs_uri:
+                if not self.data_format:
+                    raise ValueError("`data_format` must be provided with gcs uri.")
+                if self.data_format == CSV:
+                    user_data_format = (
+                        model_monitoring_spec.ModelMonitoringInput.ModelMonitoringDataset.ModelMonitoringGcsSource.DataFormat.CSV
+                    )
+                elif self.data_format == JSONL:
+                    user_data_format = (
+                        model_monitoring_spec.ModelMonitoringInput.ModelMonitoringDataset.ModelMonitoringGcsSource.DataFormat.JSONL
+                    )
+                elif self.data_format == TF_RECORD:
+                    user_data_format = (
+                        model_monitoring_spec.ModelMonitoringInput.ModelMonitoringDataset.ModelMonitoringGcsSource.DataFormat.TF_RECORD
+                    )
+                else:
+                    raise ValueError(
+                        (
+                            "Unsupported value in data format. `data_format` "
+                            "must be one of %s, %s, or %s"
+                        )
+                        % (TF_RECORD, CSV, JSONL)
+                    )
+                user_gcs_source = model_monitoring_spec.ModelMonitoringInput.ModelMonitoringDataset.ModelMonitoringGcsSource(
+                    gcs_uri=self.gcs_uri,
+                    format_=user_data_format,
+                )
+            elif self.table_uri or self.query:
+                user_bigquery_source = model_monitoring_spec.ModelMonitoringInput.ModelMonitoringDataset.ModelMonitoringBigQuerySource(
+                    table_uri=self.table_uri,
+                    query=self.query,
+                )
+            else:
+                raise ValueError(
+                    ("At least one source of dataset must" " be provided.")
+                )
+            user_model_monitoring_dataset = (
+                model_monitoring_spec.ModelMonitoringInput.ModelMonitoringDataset(
+                    vertex_dataset=user_vertex_dataset,
+                    gcs_source=user_gcs_source,
+                    bigquery_source=user_bigquery_source,
+                    timestamp_field=self.timestamp_field,
+                )
+            )
+            return model_monitoring_spec.ModelMonitoringInput(
+                columnized_dataset=user_model_monitoring_dataset,
+                time_offset=user_time_spec,
+                time_interval=user_time_interval,
+            )
+        elif self.batch_prediction_job:
+            user_batch_prediction_output = (
+                model_monitoring_spec.ModelMonitoringInput.BatchPredictionOutput(
+                    batch_prediction_job=self.batch_prediction_job,
+                )
+            )
+            return model_monitoring_spec.ModelMonitoringInput(
+                batch_prediction_output=user_batch_prediction_output,
+                time_offset=user_time_spec,
+                time_interval=user_time_interval,
+            )
+        elif self.endpoints:
+            user_vertex_endpoint_logs = (
+                model_monitoring_spec.ModelMonitoringInput.VertexEndpointLogs(
+                    endpoints=self.endpoints,
+                )
+            )
+            return model_monitoring_spec.ModelMonitoringInput(
+                vertex_endpoint_logs=user_vertex_endpoint_logs,
+                time_offset=user_time_spec,
+                time_interval=user_time_interval,
+            )
+        else:
+            raise ValueError("At least one source of dataInput must be provided.")
+
+
+class TabularObjective:
+    """Initializer for TabularObjective.
+
+    Attributes:
+        feature_drift_spec (DataDriftSpec):
+            Optional. Input feature distribution drift monitoring spec.
+        prediction_output_drift_spec (DataDriftSpec):
+            Optional. Prediction output distribution drift monitoring spec.
+        feature_attribution_spec (FeatureAttributionSpec):
+            Optional. Feature attribution monitoring spec.
+    """
+
+    def __init__(
+        self,
+        feature_drift_spec: Optional[DataDriftSpec] = None,
+        prediction_output_drift_spec: Optional[DataDriftSpec] = None,
+        feature_attribution_spec: Optional[FeatureAttributionSpec] = None,
+    ):
+        self.feature_drift_spec = feature_drift_spec
+        self.prediction_output_drift_spec = prediction_output_drift_spec
+        self.feature_attribution_spec = feature_attribution_spec
+
+    def _as_proto(
+        self,
+    ) -> model_monitoring_spec.ModelMonitoringObjectiveSpec.TabularObjective:
+        """Converts TabularObjective to a proto message.
+
+        Returns:
+           The GAPIC representation of the model monitoring tabular objective.
+        """
+        user_feature_drift_spec = None
+        user_prediction_output_drift_spec = None
+        user_feature_attribution_spec = None
+        if self.feature_drift_spec:
+            user_feature_drift_spec = self.feature_drift_spec._as_proto()
+        if self.prediction_output_drift_spec:
+            user_prediction_output_drift_spec = (
+                self.prediction_output_drift_spec._as_proto()
+            )
+        if self.feature_attribution_spec:
+            user_feature_attribution_spec = self.feature_attribution_spec._as_proto()
+        return model_monitoring_spec.ModelMonitoringObjectiveSpec.TabularObjective(
+            feature_drift_spec=user_feature_drift_spec,
+            prediction_output_drift_spec=user_prediction_output_drift_spec,
+            feature_attribution_spec=user_feature_attribution_spec,
+        )
+
+
+class ObjectiveSpec:
+    """Initializer for ObjectiveSpec.
+
+    Args:
+        baseline_dataset (MonitoringInput):
+            Required. Baseline datasets that are used by all the monitoring
+            objectives. It could be the training dataset or production serving
+            dataset from a previous period.
+        target_dataset (MonitoringInput):
+            Required. Target dataset for monitoring analysis, it's used by all
+            the monitoring objectives.
+        tabular_objective (TabularObjective):
+            Optional. The tabular monitoring objective.
+        explanation_spec (explanation.ExplanationSpec):
+            Optional. The explanation spec. This spec is required when the
+            objectives spec includes feature attribution objectives.
+    """
+
+    def __init__(
+        self,
+        baseline_dataset: MonitoringInput,
+        target_dataset: MonitoringInput,
+        tabular_objective: Optional[TabularObjective] = None,
+        explanation_spec: Optional[explanation.ExplanationSpec] = None,
+    ):
+        self.baseline = baseline_dataset
+        self.target = target_dataset
+        self.tabular_objective = tabular_objective
+        self.explanation_spec = explanation_spec
+
+    def _as_proto(self) -> model_monitoring_spec.ModelMonitoringObjectiveSpec:
+        """Converts ModelMonitoringObjectiveSpec to a proto message.
+
+        Returns:
+           The GAPIC representation of the model monitoring objective config.
+        """
+        user_tabular_objective = None
+        if not self.baseline or not self.target:
+            raise ValueError("At least one objective must be provided.")
+        if self.tabular_objective:
+            user_tabular_objective = self.tabular_objective._as_proto()
+        return model_monitoring_spec.ModelMonitoringObjectiveSpec(
+            tabular_objective=user_tabular_objective,
+            explanation_spec=self.explanation_spec if self.explanation_spec else None,
+            target_dataset=self.target._as_proto(),
+            baseline_dataset=self.baseline._as_proto(),
+        )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/ml_monitoring/spec/output.py

@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2022 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+from google.cloud.aiplatform.compat.types import (
+    io_v1beta1 as io,
+    model_monitoring_spec_v1beta1 as model_monitoring_spec,
+)
+
+
+class OutputSpec:
+    """Initializer for OutputSpec.
+
+    Args:
+        data_source (str):
+            Optional. Google Cloud Storage base folder path for metrics, error
+            logs, etc.
+    """
+
+    def __init__(
+        self,
+        gcs_base_dir: str,
+    ):
+        self.gcs_base_dir = gcs_base_dir
+
+    def _as_proto(self) -> model_monitoring_spec.ModelMonitoringOutputSpec:
+        """Converts ModelMonitoringOutputSpec to a proto message.
+
+        Returns:
+           The GAPIC representation of the notification alert config.
+        """
+        user_gcs_base_dir = io.GcsDestination(output_uri_prefix=self.gcs_base_dir)
+        return model_monitoring_spec.ModelMonitoringOutputSpec(
+            gcs_base_directory=user_gcs_base_dir,
+        )

.venv/lib/python3.10/site-packages/vertexai/resources/preview/ml_monitoring/spec/schema.py

@@ -0,0 +1,441 @@
+# -*- coding: utf-8 -*-
+
+# Copyright 2022 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+import json
+import logging
+import os
+from typing import Dict, List, MutableSequence, Optional
+from google.cloud import bigquery
+from google.cloud.aiplatform.compat.types import (
+    model_monitor_v1beta1 as model_monitor,
+)
+
+try:
+    import pandas as pd
+except ImportError:
+    pd = None
+try:
+    import tensorflow as tf
+except ImportError:
+    tf = None
+
+
+class FieldSchema:
+    """Field Schema.
+
+    The class identifies the data type of a single feature,
+    which combines together to form the Schema for different fields in
+    ModelMonitoringSchema.
+
+    Attributes:
+        name (str):
+            Required. Field name.
+        data_type (str):
+            Required. Supported data types are: ``float``, ``integer``
+            ``boolean``, ``string``, ``categorical``.
+        repeated (bool):
+            Optional. Describes if the schema field is an array of given data
+            type.
+    """
+
+    def __init__(
+        self,
+        name: str,
+        data_type: str,
+        repeated: Optional[bool] = False,
+    ):
+        self.name = name
+        self.data_type = data_type
+        self.repeated = repeated
+
+    def _as_proto(self) -> model_monitor.ModelMonitoringSchema.FieldSchema:
+        """Converts ModelMonitoringSchema.FieldSchema to a proto message.
+
+        Returns:
+           The GAPIC representation of the model monitoring field schema.
+        """
+        return model_monitor.ModelMonitoringSchema.FieldSchema(
+            name=self.name,
+            data_type=self.data_type,
+            repeated=self.repeated,
+        )
+
+
+class ModelMonitoringSchema:
+    """Initializer for ModelMonitoringSchema.
+
+    Args:
+        feature_fields (MutableSequence[FieldSchema]):
+            Required. Feature names of the model. Vertex AI will try to match
+            the features from your dataset as follows:
+            * For 'csv' files, the header names are required, and we will
+              extract thecorresponding feature values when the header names
+              align with the feature names.
+            * For 'jsonl' files, we will extract the corresponding feature
+              values if the key names match the feature names. Note: Nested
+              features are not supported, so please ensure your features are
+              flattened. Ensure the feature values are scalar or an array of
+              scalars.
+            * For 'bigquery' dataset, we will extract the corresponding feature
+              values if the column names match the feature names.
+              Note: The column type can be a scalar or an array of scalars.
+              STRUCT or JSON types are not supported. You may use SQL queries to
+              select or aggregate the relevant features from your original
+              table. However, ensure that the 'schema' of the query results
+              meets our requirements.
+            * For the Vertex AI Endpoint Request Response Logging table or
+              Vertex AI Batch Prediction Job results. If the prediction
+              instance format is an array, ensure that the sequence in
+              ``feature_fields`` matches the order of features in the prediction
+              instance. We will match the feature with the array in the order
+              specified in ``feature_fields``.
+        prediction_fields (MutableSequence[FieldSchema]):
+            Optional. Prediction output names of the model. The requirements are
+            the same as the ``feature_fields``.
+            For AutoML Tables, the prediction output name presented in schema
+            will be: `predicted_{target_column}`, the `target_column` is the one
+            you specified when you train the model.
+            For Prediction output drift analysis:
+            * AutoML Classification, the distribution of the argmax label will
+              be analyzed.
+            * AutoML Regression, the distribution of the value will be analyzed.
+        ground_truth_fields (MutableSequence[FieldSchema]):
+            Optional. Target /ground truth names of the model.
+    """
+
+    def __init__(
+        self,
+        feature_fields: MutableSequence[FieldSchema],
+        ground_truth_fields: Optional[MutableSequence[FieldSchema]] = None,
+        prediction_fields: Optional[MutableSequence[FieldSchema]] = None,
+    ):
+        self.feature_fields = feature_fields
+        self.prediction_fields = prediction_fields
+        self.ground_truth_fields = ground_truth_fields
+
+    def _as_proto(self) -> model_monitor.ModelMonitoringSchema:
+        """Converts ModelMonitoringSchema to a proto message.
+
+        Returns:
+           The GAPIC representation of the model monitoring schema.
+        """
+        user_feature_fields = list()
+        user_prediction_fields = list()
+        user_ground_truth_fields = list()
+        for field in self.feature_fields:
+            user_feature_fields.append(field._as_proto())
+        if self.prediction_fields:
+            for field in self.prediction_fields:
+                user_prediction_fields.append(field._as_proto())
+        if self.ground_truth_fields:
+            for field in self.ground_truth_fields:
+                user_ground_truth_fields.append(field._as_proto())
+        return model_monitor.ModelMonitoringSchema(
+            feature_fields=user_feature_fields,
+            prediction_fields=user_prediction_fields
+            if self.prediction_fields
+            else None,
+            ground_truth_fields=user_ground_truth_fields
+            if self.ground_truth_fields
+            else None,
+        )
+
+    def to_json(self, output_dir: Optional[str] = None) -> str:
+        """Transform ModelMonitoringSchema to json format.
+
+        Args:
+            output_dir (str):
+                Optional. The output directory that the transformed json file
+                would be put into.
+        """
+        result = model_monitor.ModelMonitoringSchema.to_json(self._as_proto())
+        if output_dir:
+            result_path = os.path.join(output_dir, "model_monitoring_schema.json")
+            with tf.io.gfile.GFile(result_path, "w") as f:
+                json.dump(result, f)
+                f.close()
+            logging.info("Transformed schema to json file: %s", result_path)
+        return result
+
+
+def _check_duplicate(
+    field: str,
+    feature_fields: Optional[List[str]] = None,
+    ground_truth_fields: Optional[List[str]] = None,
+    prediction_fields: Optional[List[str]] = None,
+) -> bool:
+    """Check if a field appears in two field lists."""
+    feature = True
+    ground_truth = True
+    prediction = True
+    if not feature_fields or field not in feature_fields:
+        feature = False
+    if not ground_truth_fields or field not in ground_truth_fields:
+        ground_truth = False
+    if not prediction_fields or field not in prediction_fields:
+        prediction = False
+    return feature if (feature == ground_truth) else prediction
+
+
+def _transform_schema_pandas(
+    dataset: Dict[str, str],
+    feature_fields: Optional[List[str]] = None,
+    ground_truth_fields: Optional[List[str]] = None,
+    prediction_fields: Optional[List[str]] = None,
+) -> ModelMonitoringSchema:
+    """Transforms the pandas schema to model monitoring schema."""
+    ground_truth_fields_list = list()
+    prediction_fields_list = list()
+    feature_fields_list = list()
+    pandas_integer_types = ["integer", "Int32", "Int64", "UInt32", "UInt64"]
+    pandas_string_types = [
+        "string",
+        "bytes",
+        "date",
+        "time",
+        "datetime64",
+        "datetime",
+        "mixed-integer",
+        "inteval",
+        "Interval",
+    ]
+    pandas_float_types = [
+        "floating",
+        "decimal",
+        "mixed-integer-float",
+        "Float32",
+        "Float64",
+    ]
+    for field in dataset:
+        infer_type = dataset[field]
+        if infer_type in pandas_string_types:
+            data_type = "string"
+        elif infer_type in pandas_integer_types:
+            data_type = "integer"
+        elif infer_type in pandas_float_types:
+            data_type = "float"
+        elif infer_type == "boolean":
+            data_type = "boolean"
+        elif infer_type == "categorical" or infer_type == "category":
+            data_type = "categorical"
+        else:
+            raise ValueError(f"Unsupported data type: {infer_type}")
+        if _check_duplicate(
+            field, feature_fields, ground_truth_fields, prediction_fields
+        ):
+            raise ValueError(f"The field {field} specified in two or more field lists")
+        if ground_truth_fields and field in ground_truth_fields:
+            ground_truth_fields_list.append(
+                FieldSchema(
+                    name=field,
+                    data_type=data_type,
+                )
+            )
+        elif prediction_fields and field in prediction_fields:
+            prediction_fields_list.append(
+                FieldSchema(
+                    name=field,
+                    data_type=data_type,
+                )
+            )
+        elif (feature_fields and field in feature_fields) or not feature_fields:
+            feature_fields_list.append(
+                FieldSchema(
+                    name=field,
+                    data_type=data_type,
+                )
+            )
+    return ModelMonitoringSchema(
+        ground_truth_fields=ground_truth_fields_list if ground_truth_fields else None,
+        prediction_fields=prediction_fields_list if prediction_fields else None,
+        feature_fields=feature_fields_list,
+    )
+
+
+def transform_schema_from_bigquery(
+    feature_fields: Optional[List[str]] = None,
+    ground_truth_fields: Optional[List[str]] = None,
+    prediction_fields: Optional[List[str]] = None,
+    table: Optional[str] = None,
+    query: Optional[str] = None,
+) -> ModelMonitoringSchema:
+    """Transform the existing dataset to ModelMonitoringSchema as model monitor
+    could accept.
+
+    Args:
+        feature_fields (List[str]):
+            Optional. The input feature fields for given dataset.
+            By default all features we find would be the input features.
+        ground_truth_fields (List[str]):
+            Optional. The ground truth fields for given dataset.
+            By default all features we find would be the input features.
+        prediction_fields (List[str]):
+            Optional. The prediction output field for given dataset.
+            By default all features we find would be the input features.
+        table (str):
+            Optional. The BigQuery table uri.
+        query (str):
+            Optional. The BigQuery query.
+    """
+    ground_truth_fields_list = list()
+    prediction_fields_list = list()
+    feature_fields_list = list()
+    bq_string_types = [
+        "STRING",
+        "BYTES",
+        "DATE",
+        "TIME",
+        "GEOGRAPHY",
+        "DATETIME",
+        "JSON",
+        "INTEVAL",
+        "RANGE",
+    ]
+    bq_integer_types = ["INTEGER", "INT64", "TIMESTAMP"]
+    bq_float_types = ["FLOAT", "DOUBLE", "FLOAT64", "NUMERIC", "BIGNUMERIC"]
+    if table:
+        if table.startswith("bq://"):
+            table = table[len("bq://") :]
+        try:
+            client = bigquery.Client()
+            table = client.get_table(table)
+            bq_schema = table.schema
+        except Exception as e:
+            raise ValueError("Failed to get table from bq address provided.") from e
+    elif query:
+        try:
+            client = bigquery.Client()
+            bq_schema = client.query(
+                query=query, job_config=bigquery.job.QueryJobConfig(dry_run=True)
+            ).schema
+        except Exception as e:
+            raise ValueError("Failed to get query from bq address provided.") from e
+    else:
+        raise ValueError("Either table or query must be provided.")
+    for field in bq_schema:
+        if field.field_type in bq_string_types:
+            data_type = "string"
+        elif field.field_type in bq_integer_types:
+            data_type = "integer"
+        elif field.field_type in bq_float_types:
+            data_type = "float"
+        elif field.field_type == "BOOLEAN" or field.field_type == "BOOL":
+            data_type = "boolean"
+        else:
+            raise ValueError(f"Unsupported data type: {field.field_type}")
+        if _check_duplicate(
+            field.name, feature_fields, ground_truth_fields, prediction_fields
+        ):
+            raise ValueError(
+                f"The field {field.name} specified in two or more field lists"
+            )
+        if ground_truth_fields and field.name in ground_truth_fields:
+            ground_truth_fields_list.append(
+                FieldSchema(
+                    name=field.name,
+                    data_type=data_type,
+                    repeated=True if field.mode == "REPEATED" else False,
+                )
+            )
+        elif prediction_fields and field.name in prediction_fields:
+            prediction_fields_list.append(
+                FieldSchema(
+                    name=field.name,
+                    data_type=data_type,
+                    repeated=True if field.mode == "REPEATED" else False,
+                )
+            )
+        elif (feature_fields and field.name in feature_fields) or not feature_fields:
+            feature_fields_list.append(
+                FieldSchema(
+                    name=field.name,
+                    data_type=data_type,
+                    repeated=True if field.mode == "REPEATED" else False,
+                )
+            )
+    return ModelMonitoringSchema(
+        ground_truth_fields=ground_truth_fields_list if ground_truth_fields else None,
+        prediction_fields=prediction_fields_list if prediction_fields else None,
+        feature_fields=feature_fields_list,
+    )
+
+
+def transform_schema_from_csv(
+    file_path: str,
+    feature_fields: Optional[List[str]] = None,
+    ground_truth_fields: Optional[List[str]] = None,
+    prediction_fields: Optional[List[str]] = None,
+) -> ModelMonitoringSchema:
+    """Transform the existing dataset to ModelMonitoringSchema as model monitor could accept.
+
+    Args:
+        file_path (str):
+            Required. The dataset file path.
+        feature_fields (List[str]):
+            Optional. The input feature fields for given dataset.
+            By default all features we find would be the input features.
+        ground_truth_fields (List[str]):
+            Optional. The ground truth fields for given dataset.
+            By default all features we find would be the input features.
+        prediction_fields (List[str]):s
+            Optional. The prediction output field for given dataset.
+            By default all features we find would be the input features.
+    """
+    with tf.io.gfile.GFile(file_path, "r") as f:
+        input_dataset = pd.read_csv(f)
+        dict_dataset = dict()
+        for field in input_dataset.columns:
+            dict_dataset[field] = input_dataset.convert_dtypes().dtypes[field]
+        monitoring_schema = _transform_schema_pandas(
+            dict_dataset, feature_fields, ground_truth_fields, prediction_fields
+        )
+        f.close()
+    return monitoring_schema
+
+
+def transform_schema_from_json(
+    file_path: str,
+    feature_fields: Optional[List[str]] = None,
+    ground_truth_fields: Optional[List[str]] = None,
+    prediction_fields: Optional[List[str]] = None,
+) -> ModelMonitoringSchema:
+    """Transform the existing dataset to ModelMonitoringSchema as model monitor
+    could accept.
+
+    Args:
+        file_path (str):
+            Required. The dataset file path.
+        feature_fields (List[str]):
+            Optional. The input feature fields for given dataset.
+            By default all features we find would be the input features.
+        ground_truth_fields (List[str]):
+            Optional. The ground truth fields for given dataset.
+            By default all features we find would be the input features.
+        prediction_fields (List[str]):
+            Optional. The prediction output field for given dataset.
+            By default all features we find would be the input features.
+    """
+    with tf.io.gfile.GFile(file_path, "r") as f:
+        input_dataset = pd.read_json(f, lines=True)
+        dict_dataset = dict()
+        for field in input_dataset.columns:
+            dict_dataset[field] = input_dataset.convert_dtypes().dtypes[field]
+        monitoring_schema = _transform_schema_pandas(
+            dict_dataset, feature_fields, ground_truth_fields, prediction_fields
+        )
+        f.close()
+    return monitoring_schema