fix: Fix feature view __getitem__ for feature services (feast-dev#2769)

* fix: Fix feature view __getitem__ for feature services

Signed-off-by: Achal Shah <[email protected]>

* more fixes

Signed-off-by: Achal Shah <[email protected]>

Author: achals

sdk/python/feast/base_feature_view.py

@@ -110,18 +110,20 @@ def __str__(self):
         return str(MessageToJson(self.to_proto()))
 
     def __hash__(self):
-        return hash((self.name))
+        return hash(self.name)
 
     def __getitem__(self, item):
         assert isinstance(item, list)
 
-        referenced_features = []
-        for feature in self.features:
-            if feature.name in item:
-                referenced_features.append(feature)
-
         cp = self.__copy__()
-        cp.projection.features = referenced_features
+        if self.features:
+            referenced_features = []
+            for feature in self.features:
+                if feature.name in item:
+                    referenced_features.append(feature)
+            cp.projection.features = referenced_features
+        else:
+            cp.projection.desired_features = item
 
         return cp
 

sdk/python/feast/feature_service.py

@@ -102,22 +102,36 @@ def __init__(
         self.created_timestamp = None
         self.last_updated_timestamp = None
         self.logging_config = logging_config
-        self.infer_features()
+        for feature_grouping in self._features:
+            if isinstance(feature_grouping, BaseFeatureView):
+                self.feature_view_projections.append(feature_grouping.projection)
 
     def infer_features(self, fvs_to_update: Optional[Dict[str, FeatureView]] = None):
-        self.feature_view_projections = []
         for feature_grouping in self._features:
             if isinstance(feature_grouping, BaseFeatureView):
                 # For feature services that depend on an unspecified feature view, apply inferred schema
-                if (
-                    fvs_to_update is not None
-                    and len(feature_grouping.projection.features) == 0
-                    and feature_grouping.name in fvs_to_update
-                ):
-                    feature_grouping.projection.features = fvs_to_update[
-                        feature_grouping.name
-                    ].features
-                self.feature_view_projections.append(feature_grouping.projection)
+                if fvs_to_update and feature_grouping.name in fvs_to_update:
+                    if feature_grouping.projection.desired_features:
+                        desired_features = set(
+                            feature_grouping.projection.desired_features
+                        )
+                        actual_features = set(
+                            [
+                                f.name
+                                for f in fvs_to_update[feature_grouping.name].features
+                            ]
+                        )
+                        assert desired_features.issubset(actual_features)
+                        # We need to set the features for the projection at this point so we ensure we're starting with
+                        # an empty list.
+                        feature_grouping.projection.features = []
+                        for f in fvs_to_update[feature_grouping.name].features:
+                            if f.name in desired_features:
+                                feature_grouping.projection.features.append(f)
+                    else:
+                        feature_grouping.projection.features = fvs_to_update[
+                            feature_grouping.name
+                        ].features
             else:
                 raise ValueError(
                     f"The feature service {self.name} has been provided with an invalid type "

sdk/python/feast/feature_view_projection.py

@@ -27,6 +27,7 @@ class FeatureViewProjection:
 
     name: str
     name_alias: Optional[str]
+    desired_features: List[str]
     features: List[Field]
     join_key_map: Dict[str, str] = {}
 
@@ -51,6 +52,7 @@ def from_proto(proto: FeatureViewProjectionProto):
             name_alias=proto.feature_view_name_alias,
             features=[],
             join_key_map=dict(proto.join_key_map),
+            desired_features=[],
         )
         for feature_column in proto.feature_columns:
             feature_view_projection.features.append(Field.from_proto(feature_column))
@@ -63,6 +65,7 @@ def from_definition(base_feature_view: "BaseFeatureView"):
             name=base_feature_view.name,
             name_alias=None,
             features=base_feature_view.features,
+            desired_features=[],
         )
 
     def get_feature(self, feature_name: str) -> Field:

sdk/python/feast/field.py

@@ -50,6 +50,9 @@ def __init__(
         self.tags = tags or {}
 
     def __eq__(self, other):
+        if type(self) != type(other):
+            return False
+
         if (
             self.name != other.name
             or self.dtype != other.dtype

sdk/python/feast/inference.py

@@ -27,7 +27,7 @@ def update_data_sources_with_inferred_event_timestamp_col(
             data_source = data_source.batch_source
         if data_source.timestamp_field is None or data_source.timestamp_field == "":
             # prepare right match pattern for data source
-            ts_column_type_regex_pattern = ""
+            ts_column_type_regex_pattern: str
             # TODO(adchia): Move Spark source inference out of this logic
             if (
                 isinstance(data_source, FileSource)

sdk/python/tests/integration/registration/test_inference.py

@@ -407,20 +407,37 @@ def test_update_feature_services_with_inferred_features(simple_dataset_1):
         feature_view_1 = FeatureView(
             name="test1", entities=[entity1], source=file_source,
         )
-        feature_service = FeatureService(name="fs_1", features=[feature_view_1])
-        assert len(feature_service.feature_view_projections) == 1
+        feature_view_2 = FeatureView(
+            name="test2", entities=[entity1], source=file_source,
+        )
+
+        feature_service = FeatureService(
+            name="fs_1", features=[feature_view_1[["string_col"]], feature_view_2]
+        )
+        assert len(feature_service.feature_view_projections) == 2
         assert len(feature_service.feature_view_projections[0].features) == 0
+        assert len(feature_service.feature_view_projections[0].desired_features) == 1
+        assert len(feature_service.feature_view_projections[1].features) == 0
+        assert len(feature_service.feature_view_projections[1].desired_features) == 0
 
         update_feature_views_with_inferred_features_and_entities(
-            [feature_view_1], [entity1], RepoConfig(provider="local", project="test")
+            [feature_view_1, feature_view_2],
+            [entity1],
+            RepoConfig(provider="local", project="test"),
         )
         feature_service.infer_features(
-            fvs_to_update={feature_view_1.name: feature_view_1}
+            fvs_to_update={
+                feature_view_1.name: feature_view_1,
+                feature_view_2.name: feature_view_2,
+            }
         )
 
         assert len(feature_view_1.schema) == 0
         assert len(feature_view_1.features) == 3
-        assert len(feature_service.feature_view_projections[0].features) == 3
+        assert len(feature_view_2.schema) == 0
+        assert len(feature_view_2.features) == 3
+        assert len(feature_service.feature_view_projections[0].features) == 1
+        assert len(feature_service.feature_view_projections[1].features) == 3
 
 
 # TODO(felixwang9817): Add tests that interact with field mapping.