address comments

pyalex · pyalex · commit 14ad6cdee1b1 · 2022-05-20T13:38:20.000-07:00
Signed-off-by: Oleksii Moskalenko &lt;moskalenko.alexey@gmail.com&gt;
diff --git a/protos/feast/core/Registry.proto b/protos/feast/core/Registry.proto
@@ -33,7 +33,7 @@ import "feast/core/SavedDataset.proto";
 import "feast/core/ValidationProfile.proto";
 import "google/protobuf/timestamp.proto";
 
-// Next id: 13
+// Next id: 14
 message Registry {
     repeated Entity entities = 1;
     repeated FeatureTable feature_tables = 2;
diff --git a/protos/feast/core/ValidationProfile.proto b/protos/feast/core/ValidationProfile.proto
@@ -45,14 +45,18 @@ message ValidationReference {
   string reference_dataset_name = 2;
   // Name of Feast project that this object source belongs to
   string project = 3;
+  // Description of the validation reference
+  string description = 4;
+  // User defined metadata
+  map<string,string> tags = 5;
 
   // validation profiler
   oneof profiler {
-    GEValidationProfiler ge_profiler = 4;
+    GEValidationProfiler ge_profiler = 6;
   }
 
   // (optional) cached validation profile (to avoid constant recalculation)
   oneof cached_profile {
-    GEValidationProfile ge_profile = 5;
+    GEValidationProfile ge_profile = 7;
   }
 }
diff --git a/sdk/python/feast/saved_dataset.py b/sdk/python/feast/saved_dataset.py
@@ -218,12 +218,21 @@ def get_profile(self, profiler: Profiler) -> Profile:
 class ValidationReference:
     name: str
     dataset_name: str
+    description: str
+    tags: Dict[str, str]
     profiler: Profiler
 
     _profile: Optional[Profile] = None
     _dataset: Optional[SavedDataset] = None
 
-    def __init__(self, name: str, dataset_name: str, profiler: Profiler):
+    def __init__(
+        self,
+        name: str,
+        dataset_name: str,
+        profiler: Profiler,
+        description: str = "",
+        tags: Optional[Dict[str, str]] = None,
+    ):
         """
         Validation reference combines a reference dataset (currently only a saved dataset object can be used as
         a reference) and a profiler function to generate a validation profile.
@@ -234,13 +243,17 @@ def __init__(self, name: str, dataset_name: str, profiler: Profiler):
         must be unique within one project.
 
         Args:
-            name: unique name
-            dataset_name: name of a saved dataset
-            profiler: profiler function used to generate profile from a saved dataset
+            name: the unique name for validation reference
+            dataset_name: the name of the saved dataset used as a reference
+            description: a human-readable description
+            tags: a dictionary of key-value pairs to store arbitrary metadata
+            profiler: the profiler function used to generate profile from the saved dataset
         """
         self.name = name
         self.dataset_name = dataset_name
         self.profiler = profiler
+        self.description = description
+        self.tags = tags or {}
 
     @classmethod
     def from_saved_dataset(cls, name: str, dataset: SavedDataset, profiler: Profiler):
@@ -289,6 +302,8 @@ def from_proto(cls, proto: ValidationReferenceProto) -> "ValidationReference":
             name=proto.name,
             dataset_name=proto.reference_dataset_name,
             profiler=profiler,
+            description=proto.description,
+            tags=dict(proto.tags),
         )
         ref._profile = profile
 
@@ -300,6 +315,8 @@ def to_proto(self) -> ValidationReferenceProto:
         proto = ValidationReferenceProto(
             name=self.name,
             reference_dataset_name=self.dataset_name,
+            tags=self.tags,
+            description=self.description,
             ge_profiler=self.profiler.to_proto()
             if isinstance(self.profiler, GEProfiler)
             else None,

Original file line number	Diff line number	Diff line change
`@@ -45,14 +45,18 @@ message ValidationReference {`
`45`	`45`	`string reference_dataset_name = 2;`
`46`	`46`	`// Name of Feast project that this object source belongs to`
`47`	`47`	`string project = 3;`
	`48`	`+ // Description of the validation reference`
	`49`	`+ string description = 4;`
	`50`	`+ // User defined metadata`
	`51`	`+ map<string,string> tags = 5;`
`48`	`52`
`49`	`53`	`// validation profiler`
`50`	`54`	`oneof profiler {`
`51`		`- GEValidationProfiler ge_profiler = 4;`
	`55`	`+ GEValidationProfiler ge_profiler = 6;`
`52`	`56`	`}`
`53`	`57`
`54`	`58`	`// (optional) cached validation profile (to avoid constant recalculation)`
`55`	`59`	`oneof cached_profile {`
`56`		`- GEValidationProfile ge_profile = 5;`
	`60`	`+ GEValidationProfile ge_profile = 7;`
`57`	`61`	`}`
`58`	`62`	`}`