From 1353e82528e8d6d266f17f4cce013a6d8c65f7bd Mon Sep 17 00:00:00 2001
From: Bo Xu <shifucun@users.noreply.github.com>
Date: Fri, 9 Jun 2023 21:29:07 +0000
Subject: [PATCH] Improve BQ query efficiency for stat var observation related
 SPARQL (#1160)

---
 internal/translator/translate.go      | 23 ++++++++++++++++++++---
 internal/translator/translate_test.go | 24 ++++++++++++++++++++++--
 2 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/internal/translator/translate.go b/internal/translator/translate.go
index e7fc060f7..e6d856ae2 100644
--- a/internal/translator/translate.go
+++ b/internal/translator/translate.go
@@ -28,6 +28,8 @@ import (
 	"google.golang.org/grpc/status"
 )
 
+const svProp = "variable_measured"
+
 // Binding contains a query and mapping object which bind together.
 type Binding struct {
 	Query   *types.Query
@@ -812,8 +814,17 @@ func getSQL(
 
 	// Sort to get deterministic result.
 	sort.SliceStable(whereConstraints, func(i, j int) bool {
-		return strings.Compare(
-			whereConstraints[i].LHS.String(), whereConstraints[j].LHS.String()) < 0
+		l := whereConstraints[i].LHS
+		r := whereConstraints[j].LHS
+		// Put "variable_measured" constraints at the beginning to better use
+		// StatVarObservation's key.
+		if l.Name == svProp {
+			return true
+		}
+		if r.Name == svProp {
+			return false
+		}
+		return strings.Compare(l.String(), r.String()) < 0
 	})
 	for idx, c := range whereConstraints {
 		if idx == 0 {
@@ -904,7 +915,13 @@ func Translate(
 		queryOptions = &types.QueryOptions{}
 	}
 
-	sql, prov, err := getSQL(nodes, constraints, constNode, ProvInfo{queryProv, tableProv}, queryOptions)
+	sql, prov, err := getSQL(
+		nodes,
+		constraints,
+		constNode,
+		ProvInfo{queryProv, tableProv},
+		queryOptions,
+	)
 	if err != nil {
 		return nil, err
 	}
diff --git a/internal/translator/translate_test.go b/internal/translator/translate_test.go
index 29f688fa3..513da7e9c 100644
--- a/internal/translator/translate_test.go
+++ b/internal/translator/translate_test.go
@@ -1234,8 +1234,28 @@ func TestStatVarObs(t *testing.T) {
 				"FROM `dc_v3.Place` AS _dc_v3_Place_1\n" +
 				"JOIN `dc_v3.StatVarObservation` AS _dc_v3_StatVarObservation_0\n" +
 				"ON _dc_v3_Place_1.id = _dc_v3_StatVarObservation_0.observation_about\n" +
-				"WHERE _dc_v3_Place_1.type = \"Country\"\n" +
-				"AND _dc_v3_StatVarObservation_0.variable_measured = \"Amount_EconomicActivity_GrossNationalIncome_PurchasingPowerParity_PerCapita\"\n",
+				"WHERE _dc_v3_StatVarObservation_0.variable_measured = \"Amount_EconomicActivity_GrossNationalIncome_PurchasingPowerParity_PerCapita\"\n" +
+				"AND _dc_v3_Place_1.type = \"Country\"\n",
+		},
+		{
+			"browser-observation",
+			`
+			SELECT ?dcid ?mmethod ?obsPeriod ?obsDate
+				WHERE {
+					?svObservation typeOf StatVarObservation .
+					?svObservation variableMeasured Count_Person .
+					?svObservation observationAbout country/USA .
+					?svObservation dcid ?dcid .
+					?svObservation measurementMethod ?mmethod .
+					?svObservation observationPeriod ?obsPeriod .
+				}
+				`,
+			"SELECT _dc_v3_StatVarObservation_0.id AS dcid,\n" +
+				"_dc_v3_StatVarObservation_0.measurement_method AS mmethod,\n" +
+				"_dc_v3_StatVarObservation_0.observation_period AS obsPeriod,\n\n" +
+				"FROM `dc_v3.StatVarObservation` AS _dc_v3_StatVarObservation_0\n" +
+				"WHERE _dc_v3_StatVarObservation_0.variable_measured = \"Count_Person\"\n" +
+				"AND _dc_v3_StatVarObservation_0.observation_about = \"country/USA\"\n",
 		},
 	} {
 		nodes, queries, _, err := sparql.ParseQuery(c.queryStr)