def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pd.DataFrame, str],
registry: Registry,
project: str,
full_feature_names: bool = False,
) -> RetrievalJob:
# TODO: Add entity_df validation in order to fail before interacting with BigQuery
assert isinstance(config.offline_store, BigQueryOfflineStoreConfig)
client = _get_bigquery_client(project=config.offline_store.project_id)
assert isinstance(config.offline_store, BigQueryOfflineStoreConfig)
table_reference = _get_table_reference_for_new_entity(
client, client.project, config.offline_store.dataset
)
entity_schema = _upload_entity_df_and_get_entity_schema(
client=client, table_name=table_reference, entity_df=entity_df,
)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema
)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry
)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys, entity_df_event_timestamp_col
)
# Build a query context containing all information required to template the BigQuery SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs, feature_views, registry, project,
)
# Generate the BigQuery SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_reference,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
return BigQueryRetrievalJob(
query=query,
client=client,
config=config,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry
),
)
def query_generator() -> Iterator[str]:
table_name = offline_utils.get_temp_entity_table_name()
entity_schema = _upload_entity_df_and_get_entity_schema(
entity_df, redshift_client, config, s3_resource, table_name)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys,
entity_df_event_timestamp_col)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
redshift_client,
config,
table_name,
)
# Build a query context containing all information required to template the Redshift SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
# Generate the Redshift SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_name,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
try:
yield query
finally:
# Always clean up the uploaded Redshift table
aws_utils.execute_redshift_statement(
redshift_client,
config.offline_store.cluster_id,
config.offline_store.database,
config.offline_store.user,
f"DROP TABLE IF EXISTS {table_name}",
)
def query_generator() -> Iterator[str]:
entity_schema = _upload_entity_df_and_get_entity_schema(
client=client,
table_name=table_reference,
entity_df=entity_df,
)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys,
entity_df_event_timestamp_col)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
client,
table_reference,
)
# Build a query context containing all information required to template the BigQuery SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
# Generate the BigQuery SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_reference,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
try:
yield query
finally:
# Asynchronously clean up the uploaded Bigquery table, which will expire
# if cleanup fails
client.delete_table(table=table_reference, not_found_ok=True)
def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pd.DataFrame, str],
registry: Registry,
project: str,
full_feature_names: bool = False,
user: str = "user",
auth: Optional[Authentication] = None,
http_scheme: Optional[str] = None,
) -> TrinoRetrievalJob:
if not isinstance(config.offline_store, TrinoOfflineStoreConfig):
raise ValueError(
f"This function should be used with a TrinoOfflineStoreConfig object. Instead we have config.offline_store being '{type(config.offline_store)}'"
)
client = _get_trino_client(config=config,
user=user,
auth=auth,
http_scheme=http_scheme)
table_reference = _get_table_reference_for_new_entity(
catalog=config.offline_store.catalog,
dataset_name=config.offline_store.dataset,
)
entity_schema = _upload_entity_df_and_get_entity_schema(
client=client,
table_name=table_reference,
entity_df=entity_df,
connector=config.offline_store.connector,
)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema=entity_schema)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df=entity_df,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
client=client,
)
expected_join_keys = offline_utils.get_expected_join_keys(
project=project, feature_views=feature_views, registry=registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema=entity_schema,
join_keys=expected_join_keys,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
)
# Build a query context containing all information required to template the Trino SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs=feature_refs,
feature_views=feature_views,
registry=registry,
project=project,
entity_df_timestamp_range=entity_df_event_timestamp_range,
)
# Generate the Trino SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_reference,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
return TrinoRetrievalJob(
query=query,
client=client,
config=config,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry),
metadata=RetrievalMetadata(
features=feature_refs,
keys=list(
set(entity_schema.keys()) -
{entity_df_event_timestamp_col}),
min_event_timestamp=entity_df_event_timestamp_range[0],
max_event_timestamp=entity_df_event_timestamp_range[1],
),
)
def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pd.DataFrame, str],
registry: Registry,
project: str,
full_feature_names: bool = False,
) -> RetrievalJob:
assert isinstance(config.offline_store, SnowflakeOfflineStoreConfig)
snowflake_conn = get_snowflake_conn(config.offline_store)
entity_schema = _get_entity_schema(entity_df, snowflake_conn, config)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
snowflake_conn,
)
@contextlib.contextmanager
def query_generator() -> Iterator[str]:
table_name = offline_utils.get_temp_entity_table_name()
_upload_entity_df(entity_df, snowflake_conn, config, table_name)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys,
entity_df_event_timestamp_col)
# Build a query context containing all information required to template the Snowflake SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
query_context = _fix_entity_selections_identifiers(query_context)
# Generate the Snowflake SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_name,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
yield query
return SnowflakeRetrievalJob(
query=query_generator,
snowflake_conn=snowflake_conn,
config=config,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry),
metadata=RetrievalMetadata(
features=feature_refs,
keys=list(entity_schema.keys() -
{entity_df_event_timestamp_col}),
min_event_timestamp=entity_df_event_timestamp_range[0],
max_event_timestamp=entity_df_event_timestamp_range[1],
),
)
def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pd.DataFrame, str],
registry: Registry,
project: str,
full_feature_names: bool = False,
) -> RetrievalJob:
# TODO: Add entity_df validation in order to fail before interacting with BigQuery
assert isinstance(config.offline_store, BigQueryOfflineStoreConfig)
client = _get_bigquery_client(
project=config.offline_store.project_id,
location=config.offline_store.location,
)
assert isinstance(config.offline_store, BigQueryOfflineStoreConfig)
table_reference = _get_table_reference_for_new_entity(
client,
client.project,
config.offline_store.dataset,
config.offline_store.location,
)
entity_schema = _get_entity_schema(
client=client,
entity_df=entity_df,
)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
client,
)
@contextlib.contextmanager
def query_generator() -> Iterator[str]:
_upload_entity_df(
client=client,
table_name=table_reference,
entity_df=entity_df,
)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys,
entity_df_event_timestamp_col)
# Build a query context containing all information required to template the BigQuery SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
# Generate the BigQuery SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_reference,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
try:
yield query
finally:
# Asynchronously clean up the uploaded Bigquery table, which will expire
# if cleanup fails
client.delete_table(table=table_reference, not_found_ok=True)
return BigQueryRetrievalJob(
query=query_generator,
client=client,
config=config,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry),
metadata=RetrievalMetadata(
features=feature_refs,
keys=list(entity_schema.keys() -
{entity_df_event_timestamp_col}),
min_event_timestamp=entity_df_event_timestamp_range[0],
max_event_timestamp=entity_df_event_timestamp_range[1],
),
)
def query_generator() -> Iterator[str]:
table_name = None
if isinstance(entity_df, pd.DataFrame):
table_name = offline_utils.get_temp_entity_table_name()
entity_schema = df_to_postgres_table(config.offline_store,
entity_df, table_name)
df_query = table_name
elif isinstance(entity_df, str):
df_query = f"({entity_df}) AS sub"
entity_schema = get_query_schema(config.offline_store,
df_query)
else:
raise TypeError(entity_df)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys,
entity_df_event_timestamp_col)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
config,
df_query,
)
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
query_context_dict = [asdict(context) for context in query_context]
# Hack for query_context.entity_selections to support uppercase in columns
for context in query_context_dict:
context["entity_selections"] = [
f'''"{entity_selection.replace(' AS ', '" AS "')}\"'''
for entity_selection in context["entity_selections"]
]
try:
yield build_point_in_time_query(
query_context_dict,
left_table_query_string=df_query,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
finally:
if table_name:
with _get_conn(config.offline_store) as conn, conn.cursor(
) as cur:
cur.execute(
sql.SQL("""
DROP TABLE IF EXISTS {};
""").format(sql.Identifier(table_name)), )
def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pandas.DataFrame, str],
registry: Registry,
project: str,
full_feature_names: bool = False,
) -> RetrievalJob:
assert isinstance(config.offline_store, SparkOfflineStoreConfig)
warnings.warn(
"The spark offline store is an experimental feature in alpha development. "
"Some functionality may still be unstable so functionality can change in the future.",
RuntimeWarning,
)
spark_session = get_spark_session_or_start_new_with_repoconfig(
store_config=config.offline_store)
tmp_entity_df_table_name = offline_utils.get_temp_entity_table_name()
entity_schema = _get_entity_schema(
spark_session=spark_session,
entity_df=entity_df,
)
event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema=entity_schema, )
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
event_timestamp_col,
spark_session,
)
_upload_entity_df(
spark_session=spark_session,
table_name=tmp_entity_df_table_name,
entity_df=entity_df,
event_timestamp_col=event_timestamp_col,
)
expected_join_keys = offline_utils.get_expected_join_keys(
project=project, feature_views=feature_views, registry=registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema=entity_schema,
join_keys=expected_join_keys,
entity_df_event_timestamp_col=event_timestamp_col,
)
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
query = offline_utils.build_point_in_time_query(
feature_view_query_contexts=query_context,
left_table_query_string=tmp_entity_df_table_name,
entity_df_event_timestamp_col=event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
return SparkRetrievalJob(
spark_session=spark_session,
query=query,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry),
metadata=RetrievalMetadata(
features=feature_refs,
keys=list(set(entity_schema.keys()) - {event_timestamp_col}),
min_event_timestamp=entity_df_event_timestamp_range[0],
max_event_timestamp=entity_df_event_timestamp_range[1],
),
)
def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pd.DataFrame, str],
registry: Registry,
project: str,
full_feature_names: bool = False,
) -> RetrievalJob:
assert isinstance(config.offline_store, RedshiftOfflineStoreConfig)
redshift_client = aws_utils.get_redshift_data_client(
config.offline_store.region)
s3_resource = aws_utils.get_s3_resource(config.offline_store.region)
entity_schema = _get_entity_schema(entity_df, redshift_client, config,
s3_resource)
entity_df_event_timestamp_col = offline_utils.infer_event_timestamp_from_entity_df(
entity_schema)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
redshift_client,
config,
)
@contextlib.contextmanager
def query_generator() -> Iterator[str]:
table_name = offline_utils.get_temp_entity_table_name()
_upload_entity_df(entity_df, redshift_client, config, s3_resource,
table_name)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys,
entity_df_event_timestamp_col)
# Build a query context containing all information required to template the Redshift SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
# Generate the Redshift SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_name,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
try:
yield query
finally:
# Always clean up the uploaded Redshift table
aws_utils.execute_redshift_statement(
redshift_client,
config.offline_store.cluster_id,
config.offline_store.database,
config.offline_store.user,
f"DROP TABLE IF EXISTS {table_name}",
)
return RedshiftRetrievalJob(
query=query_generator,
redshift_client=redshift_client,
s3_resource=s3_resource,
config=config,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry),
metadata=RetrievalMetadata(
features=feature_refs,
keys=list(entity_schema.keys() -
{entity_df_event_timestamp_col}),
min_event_timestamp=entity_df_event_timestamp_range[0],
max_event_timestamp=entity_df_event_timestamp_range[1],
),
)
