Optimized getStatVarSummaries query

internal/sqldb/statements.go

@@ -86,63 +86,72 @@ var statements = struct {
 		ORDER BY date ASC;
 	`,
 	getStatVarSummaries: `
-		WITH entity_types
-			AS (SELECT 
-					o.variable               AS variable,
-					t.object_id              AS entity_type,
-					Count(DISTINCT o.entity) AS entity_count,
-					Min(o.value + 0.0)       AS min_value,
-					Max(o.value + 0.0)       AS max_value
-				FROM   
-					observations o
-					JOIN triples t
-					ON o.entity = t.subject_id
-				WHERE  
-					o.variable IN (:variables)
-					AND t.predicate = 'typeOf'
-				GROUP  BY variable, entity_type
-				ORDER  BY entity_count DESC),
-			entities
-			AS (SELECT 
-					DISTINCT o.variable   variable,
-					t.object_id  entity_type,
-					t.subject_id entity_id
-				FROM   
-					triples t
-					JOIN observations o
-					ON o.entity = t.subject_id
-				WHERE  
-					t.predicate = 'typeOf'
-					AND t.object_id IN (SELECT entity_type FROM entity_types)
-					AND o.variable IN (SELECT DISTINCT variable FROM entity_types)),
-			sample_entities
-			AS (SELECT variable, entity_type, entity_id
-				FROM   (SELECT 
-							*,
-							Row_number() OVER (partition BY variable, entity_type) AS row_num
-						FROM   entities) AS entities_with_row_num
-				WHERE  row_num <= 3),
-			grouped_entities
-			AS (SELECT 
-					variable,
-					entity_type,
-					Group_concat(entity_id) AS sample_entity_ids
-				FROM   sample_entities
-				GROUP  BY variable, entity_type),
-			aggregate
-			AS (SELECT 
-					variable,
-					entity_type,
-					entity_count,
-					min_value,
-					max_value,
-					sample_entity_ids
-				FROM   
-					entity_types
-					JOIN grouped_entities using(variable, entity_type))
+		WITH entity_observations_triples AS (
+			SELECT
+				o.variable,
+				t.object_id AS entity_type,
+				t.subject_id AS entity_id,
+				o.value + 0.0 AS numeric_value
+			FROM
+				observations o
+			JOIN
+				triples t ON o.entity = t.subject_id
+			WHERE
+				t.predicate = 'typeOf'
+				AND o.variable IN (:variables)
+		),
+		entity_types AS (
+			SELECT
+				variable,
+				entity_type,
+				COUNT(DISTINCT entity_id) AS entity_count,
+				MIN(numeric_value) AS min_value,
+				MAX(numeric_value) AS max_value
+			FROM
+				entity_observations_triples
+			GROUP BY
+				variable, entity_type
+		),
+		sample_entities AS (
+			SELECT
+				variable,
+				entity_type,
+				entity_id
+			FROM (
+				SELECT
+					*,
+					ROW_NUMBER() OVER (PARTITION BY variable, entity_type ORDER BY entity_id) AS row_num
+				FROM
+					entity_observations_triples
+			) AS entities_with_row_num
+			WHERE row_num <= 3
+		),
+		grouped_entities AS (
+			SELECT
+				variable,
+				entity_type,
+				GROUP_CONCAT(entity_id ORDER BY entity_id) AS sample_entity_ids
+			FROM
+				sample_entities
+			GROUP BY
+				variable, entity_type
+		),
+		aggregate AS (
+			SELECT
+				et.variable,
+				et.entity_type,
+				et.entity_count,
+				et.min_value,
+				et.max_value,
+				ge.sample_entity_ids
+			FROM
+				entity_types et
+			JOIN
+				grouped_entities ge USING (variable, entity_type)
+		)
 		SELECT *
-		FROM   aggregate;
-	`,
+		FROM aggregate;
+		`,
 	getKeyValue: `
 		SELECT value
 		FROM key_value_store