add init and cleanup of long lived resources

Signed-off-by: Rob Howley <howley.robert@gmail.com>

docs/getting-started/concepts/permission.md

@@ -36,7 +36,7 @@ The permission model is based on the following components:
 The `Permission` class identifies a single permission configured on the feature store and is identified by these attributes:
 - `name`: The permission name.
 - `types`: The list of protected resource  types. Defaults to all managed types, e.g. the `ALL_RESOURCE_TYPES` alias. All sub-classes are included in the resource match.
-- `name_pattern`: A regex to match the resource name. Defaults to `None`, meaning that no name filtering is applied
+- `name_patterns`: A list of regex patterns to match resource names. If any regex matches, the `Permission` policy is applied. Defaults to `[]`, meaning no name filtering is applied.
 - `required_tags`: Dictionary of key-value pairs that must match the resource tags. Defaults to `None`, meaning that no tags filtering is applied.
 - `actions`: The actions authorized by this permission. Defaults to `ALL_VALUES`, an alias defined in the `action` module.
 - `policy`: The policy to be applied to validate a client request.
@@ -95,7 +95,7 @@ The following permission grants authorization to read the offline store of all t
 Permission(
     name="reader",
     types=[FeatureView],
-    name_pattern=".*risky.*",
+    name_patterns=".*risky.*", # Accepts both `str` or `list[str]` types
     policy=RoleBasedPolicy(roles=["trusted"]),
     actions=[AuthzedAction.READ_OFFLINE],
 )

docs/reference/feast-cli-commands.md

@@ -172,9 +172,10 @@ Options:
 
 ```text
 +-----------------------+-------------+-----------------------+-----------+----------------+-------------------------+
-| NAME                  | TYPES       | NAME_PATTERN          | ACTIONS   | ROLES          | REQUIRED_TAGS           |
+| NAME                  | TYPES       | NAME_PATTERNS         | ACTIONS   | ROLES          | REQUIRED_TAGS           |
 +=======================+=============+=======================+===========+================+================+========+
 | reader_permission1234 | FeatureView | transformed_conv_rate | DESCRIBE  | reader         | -                       |
+|                       |             | driver_hourly_stats   | DESCRIBE  | reader         | -                       |
 +-----------------------+-------------+-----------------------+-----------+----------------+-------------------------+
 | writer_permission1234 | FeatureView | transformed_conv_rate | CREATE    | writer         | -                       |
 +-----------------------+-------------+-----------------------+-----------+----------------+-------------------------+

java/serving/pom.xml

@@ -287,7 +287,7 @@
     <dependency>
       <groupId>org.apache.avro</groupId>
       <artifactId>avro</artifactId>
-      <version>1.11.3</version>
+      <version>1.11.4</version>
     </dependency>
 
     <!-- https://mvnrepository.com/artifact/org.apache.arrow/arrow-java-root -->

protos/feast/core/Permission.proto

@@ -50,7 +50,7 @@ message PermissionSpec {
 
   repeated Type types = 3;
 
-  string name_pattern = 4;
+  repeated string name_patterns = 4;
 
   map<string, string> required_tags = 5;
 

sdk/python/feast/cli.py

@@ -1211,7 +1211,7 @@ def feast_permissions_list_command(ctx: click.Context, verbose: bool, tags: list
                 headers=[
                     "NAME",
                     "TYPES",
-                    "NAME_PATTERN",
+                    "NAME_PATTERNS",
                     "ACTIONS",
                     "ROLES",
                     "REQUIRED_TAGS",

sdk/python/feast/cli_utils.py

@@ -196,7 +196,7 @@ def handle_not_verbose_permissions_command(
         [
             p.name,
             _to_multi_line([t.__name__ for t in p.types]),  # type: ignore[union-attr, attr-defined]
-            p.name_pattern,
+            _to_multi_line(p.name_patterns),
             _to_multi_line([a.value.upper() for a in p.actions]),
             _to_multi_line(sorted(roles)),
             _dict_to_multi_line(p.required_tags),

sdk/python/feast/feature_server.py

@@ -100,8 +100,10 @@ def async_refresh():
     @asynccontextmanager
     async def lifespan(app: FastAPI):
         async_refresh()
+        await store.initialize()
         yield
         stop_refresh()
+        await store.close()
 
     app = FastAPI(lifespan=lifespan)
 

sdk/python/feast/feature_store.py

@@ -713,7 +713,7 @@ def plan(
             >>> fs = FeatureStore(repo_path="project/feature_repo")
             >>> driver = Entity(name="driver_id", description="driver id")
             >>> driver_hourly_stats = FileSource(
-            ...     path="project/feature_repo/data/driver_stats.parquet",
+            ...     path="data/driver_stats.parquet",
             ...     timestamp_field="event_timestamp",
             ...     created_timestamp_column="created",
             ... )
@@ -827,7 +827,7 @@ def apply(
             >>> fs = FeatureStore(repo_path="project/feature_repo")
             >>> driver = Entity(name="driver_id", description="driver id")
             >>> driver_hourly_stats = FileSource(
-            ...     path="project/feature_repo/data/driver_stats.parquet",
+            ...     path="data/driver_stats.parquet",
             ...     timestamp_field="event_timestamp",
             ...     created_timestamp_column="created",
             ... )
@@ -2078,6 +2078,14 @@ def list_saved_datasets(
             self.project, allow_cache=allow_cache, tags=tags
         )
 
+    async def initialize(self) -> None:
+        """Initialize long-lived clients and/or resources needed for accessing datastores"""
+        await self._get_provider().initialize(self.config)
+
+    async def close(self) -> None:
+        """Cleanup any long-lived clients and/or resources"""
+        await self._get_provider().close()
+
 
 def _print_materialization_log(
     start_date, end_date, num_feature_views: int, online_store: str

sdk/python/feast/infra/offline_stores/dask.py

@@ -57,6 +57,7 @@ def __init__(
         self,
         evaluation_function: Callable,
         full_feature_names: bool,
+        repo_path: str,
         on_demand_feature_views: Optional[List[OnDemandFeatureView]] = None,
         metadata: Optional[RetrievalMetadata] = None,
     ):
@@ -67,6 +68,7 @@ def __init__(
         self._full_feature_names = full_feature_names
         self._on_demand_feature_views = on_demand_feature_views or []
         self._metadata = metadata
+        self.repo_path = repo_path
 
     @property
     def full_feature_names(self) -> bool:
@@ -99,8 +101,13 @@ def persist(
         if not allow_overwrite and os.path.exists(storage.file_options.uri):
             raise SavedDatasetLocationAlreadyExists(location=storage.file_options.uri)
 
+        if not Path(storage.file_options.uri).is_absolute():
+            absolute_path = Path(self.repo_path) / storage.file_options.uri
+        else:
+            absolute_path = Path(storage.file_options.uri)
+
         filesystem, path = FileSource.create_filesystem_and_path(
-            storage.file_options.uri,
+            str(absolute_path),
             storage.file_options.s3_endpoint_override,
         )
 
@@ -243,7 +250,9 @@ def evaluate_historical_retrieval():
 
                 all_join_keys = list(set(all_join_keys + join_keys))
 
-                df_to_join = _read_datasource(feature_view.batch_source)
+                df_to_join = _read_datasource(
+                    feature_view.batch_source, config.repo_path
+                )
 
                 df_to_join, timestamp_field = _field_mapping(
                     df_to_join,
@@ -297,6 +306,7 @@ def evaluate_historical_retrieval():
                 min_event_timestamp=entity_df_event_timestamp_range[0],
                 max_event_timestamp=entity_df_event_timestamp_range[1],
             ),
+            repo_path=str(config.repo_path),
         )
         return job
 
@@ -316,7 +326,7 @@ def pull_latest_from_table_or_query(
 
         # Create lazy function that is only called from the RetrievalJob object
         def evaluate_offline_job():
-            source_df = _read_datasource(data_source)
+            source_df = _read_datasource(data_source, config.repo_path)
 
             source_df = _normalize_timestamp(
                 source_df, timestamp_field, created_timestamp_column
@@ -377,6 +387,7 @@ def evaluate_offline_job():
         return DaskRetrievalJob(
             evaluation_function=evaluate_offline_job,
             full_feature_names=False,
+            repo_path=str(config.repo_path),
         )
 
     @staticmethod
@@ -420,8 +431,13 @@ def write_logged_features(
             # Since this code will be mostly used from Go-created thread, it's better to avoid producing new threads
             data = pyarrow.parquet.read_table(data, use_threads=False, pre_buffer=False)
 
+        if config.repo_path is not None and not Path(destination.path).is_absolute():
+            absolute_path = config.repo_path / destination.path
+        else:
+            absolute_path = Path(destination.path)
+
         filesystem, path = FileSource.create_filesystem_and_path(
-            destination.path,
+            str(absolute_path),
             destination.s3_endpoint_override,
         )
 
@@ -456,8 +472,14 @@ def offline_write_batch(
             )
 
         file_options = feature_view.batch_source.file_options
+
+        if config.repo_path is not None and not Path(file_options.uri).is_absolute():
+            absolute_path = config.repo_path / file_options.uri
+        else:
+            absolute_path = Path(file_options.uri)
+
         filesystem, path = FileSource.create_filesystem_and_path(
-            file_options.uri, file_options.s3_endpoint_override
+            str(absolute_path), file_options.s3_endpoint_override
         )
         prev_table = pyarrow.parquet.read_table(
             path, filesystem=filesystem, memory_map=True
@@ -493,7 +515,7 @@ def _get_entity_df_event_timestamp_range(
     )
 
 
-def _read_datasource(data_source) -> dd.DataFrame:
+def _read_datasource(data_source, repo_path) -> dd.DataFrame:
     storage_options = (
         {
             "client_kwargs": {
@@ -504,8 +526,12 @@ def _read_datasource(data_source) -> dd.DataFrame:
         else None
     )
 
+    if not Path(data_source.path).is_absolute():
+        path = repo_path / data_source.path
+    else:
+        path = data_source.path
     return dd.read_parquet(
-        data_source.path,
+        path,
         storage_options=storage_options,
     )
 

sdk/python/feast/infra/offline_stores/duckdb.py

@@ -27,7 +27,7 @@
 from feast.repo_config import FeastConfigBaseModel, RepoConfig
 
 
-def _read_data_source(data_source: DataSource) -> Table:
+def _read_data_source(data_source: DataSource, repo_path: str) -> Table:
     assert isinstance(data_source, FileSource)
 
     if isinstance(data_source.file_format, ParquetFormat):
@@ -43,21 +43,32 @@ def _read_data_source(data_source: DataSource) -> Table:
 def _write_data_source(
     table: Table,
     data_source: DataSource,
+    repo_path: str,
     mode: str = "append",
     allow_overwrite: bool = False,
 ):
     assert isinstance(data_source, FileSource)
 
     file_options = data_source.file_options
 
-    if mode == "overwrite" and not allow_overwrite and os.path.exists(file_options.uri):
+    if not Path(file_options.uri).is_absolute():
+        absolute_path = Path(repo_path) / file_options.uri
+    else:
+        absolute_path = Path(file_options.uri)
+
+    if (
+        mode == "overwrite"
+        and not allow_overwrite
+        and os.path.exists(str(absolute_path))
+    ):
         raise SavedDatasetLocationAlreadyExists(location=file_options.uri)
 
     if isinstance(data_source.file_format, ParquetFormat):
         if mode == "overwrite":
             table = table.to_pyarrow()
+
             filesystem, path = FileSource.create_filesystem_and_path(
-                file_options.uri,
+                str(absolute_path),
                 file_options.s3_endpoint_override,
             )
 

sdk/python/feast/infra/offline_stores/file_source.py

@@ -1,3 +1,4 @@
+from pathlib import Path
 from typing import Callable, Dict, Iterable, List, Optional, Tuple
 
 import pyarrow
@@ -154,8 +155,16 @@ def source_datatype_to_feast_value_type() -> Callable[[str], ValueType]:
     def get_table_column_names_and_types(
         self, config: RepoConfig
     ) -> Iterable[Tuple[str, str]]:
+        if (
+            config.repo_path is not None
+            and not Path(self.file_options.uri).is_absolute()
+        ):
+            absolute_path = config.repo_path / self.file_options.uri
+        else:
+            absolute_path = Path(self.file_options.uri)
+
         filesystem, path = FileSource.create_filesystem_and_path(
-            self.path, self.file_options.s3_endpoint_override
+            str(absolute_path), self.file_options.s3_endpoint_override
         )
 
         # TODO why None check necessary