Deduplicate rows in subquery (feast-dev#409)

Chen Zhiling · feast-ci-bot · commit 57b3887817e8 · 2020-01-06T18:52:04.000+08:00
* Deduplicate rows in subquery

* Rename columns

* Use correct feature names

* Select k.* instead of iterating through fields
diff --git a/serving/src/main/resources/templates/single_featureset_pit_join.sql b/serving/src/main/resources/templates/single_featureset_pit_join.sql
@@ -15,7 +15,7 @@ SELECT
   {{ featureSet.entities | join(', ')}},
   false AS is_entity_table
 FROM `{{projectId}}.{{datasetId}}.{{ featureSet.project }}_{{ featureSet.name }}_v{{ featureSet.version }}` WHERE event_timestamp <= '{{maxTimestamp}}' AND event_timestamp >= Timestamp_sub(TIMESTAMP '{{ minTimestamp }}', interval {{ featureSet.maxAge }} second)
-)
+), joined AS (
 SELECT
   uuid,
   event_timestamp,
@@ -44,4 +44,12 @@ SELECT
   {% endfor %}
 FROM `{{projectId}}.{{datasetId}}.{{ featureSet.project }}_{{ featureSet.name }}_v{{ featureSet.version }}` WHERE event_timestamp <= '{{maxTimestamp}}' AND event_timestamp >= Timestamp_sub(TIMESTAMP '{{ minTimestamp }}', interval {{ featureSet.maxAge }} second)
 ) USING ({{ featureSet.project }}_{{ featureSet.name }}_v{{ featureSet.version }}_feature_timestamp, created_timestamp, {{ featureSet.entities | join(', ')}})
-WHERE is_entity_table
+WHERE is_entity_table
+)
+SELECT
+  k.*
+FROM (
+  SELECT ARRAY_AGG(row LIMIT 1)[OFFSET(0)] k
+  FROM joined row
+  GROUP BY uuid
+)
