planner: clean the one side filters for outer join reorder which is not done in #44409

planner/core/main_test.go

@@ -35,6 +35,7 @@ func TestMain(m *testing.M) {
 	testDataMap.LoadTestSuiteData("testdata", "plan_suite_unexported")
 	testDataMap.LoadTestSuiteData("testdata", "index_merge_suite")
 	testDataMap.LoadTestSuiteData("testdata", "runtime_filter_generator_suite")
+	testDataMap.LoadTestSuiteData("testdata", "join_reorder_suite")
 
 	indexMergeSuiteData = testDataMap["index_merge_suite"]
 	planSuiteUnexportedData = testDataMap["plan_suite_unexported"]
@@ -62,3 +63,7 @@ func GetIndexMergeSuiteData() testdata.TestData {
 func GetRuntimeFilterGeneratorData() testdata.TestData {
 	return testDataMap["runtime_filter_generator_suite"]
 }
+
+func GetJoinReorderData() testdata.TestData {
+	return testDataMap["join_reorder_suite"]
+}

planner/core/rule_join_reorder.go

@@ -537,12 +537,17 @@ func (s *baseSingleGroupJoinOrderSolver) makeJoin(leftPlan, rightPlan LogicalPla
 	remainOtherConds, otherConds = expression.FilterOutInPlace(remainOtherConds, func(expr expression.Expression) bool {
 		return expression.ExprFromSchema(expr, mergedSchema)
 	})
-	if (joinType.JoinType == LeftOuterJoin || joinType.JoinType == RightOuterJoin || joinType.JoinType == LeftOuterSemiJoin || joinType.JoinType == AntiLeftOuterSemiJoin) && len(otherConds) > 0 {
+
+	if joinType.JoinType == LeftOuterJoin || joinType.JoinType == RightOuterJoin || joinType.JoinType == LeftOuterSemiJoin || joinType.JoinType == AntiLeftOuterSemiJoin {
 		// the original outer join's other conditions has been bound to the outer join Edge,
 		// these remained other condition here shouldn't be appended to it because on-mismatch
 		// logic will produce more append-null rows which is banned in original semantic.
 		remainOtherConds = append(remainOtherConds, otherConds...) // nozero
+		remainOtherConds = append(remainOtherConds, leftConds...)  // nozero
+		remainOtherConds = append(remainOtherConds, rightConds...) // nozero
 		otherConds = otherConds[:0]
+		leftConds = leftConds[:0]
+		rightConds = rightConds[:0]
 	}
 	if len(joinType.outerBindCondition) > 0 {
 		remainOBOtherConds := make([]expression.Expression, len(joinType.outerBindCondition))

planner/core/rule_join_reorder_test.go

@@ -17,7 +17,9 @@ package core_test
 import (
 	"testing"
 
+	"github.com/pingcap/tidb/planner/core"
 	"github.com/pingcap/tidb/testkit"
+	"github.com/pingcap/tidb/testkit/testdata"
 	"github.com/stretchr/testify/require"
 )
 
@@ -82,21 +84,28 @@ func TestAdditionOtherConditionsRemained4OuterJoin(t *testing.T) {
 	tk.MustExec("INSERT  INTO queries_program(`id`, `identifier_id`) values(8, 13), (9, 14);")
 	tk.MustExec("INSERT  INTO queries_channel(`id`, `identifier_id`) values(5, 13);")
 
-	tk.MustQuery("SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` LEFT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC;").Check(testkit.Rows("" +
-		"13 i1"))
-	tk.MustQuery("SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` RIGHT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC;").Check(testkit.Rows("" +
-		"13 i1"))
-	tk.MustQuery("explain format = 'brief' SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` LEFT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC;").Check(testkit.Rows(""+
-		"Sort 2.50 root  test.queries_identifier.id",
-		"└─Projection 2.50 root  test.queries_identifier.id, test.queries_identifier.name",
-		"  └─Selection 2.50 root  or(and(eq(test.queries_channel.id, 5), eq(test.queries_program.id, 9)), eq(test.queries_program.id, 8))",
-		"    └─IndexJoin 3.12 root  left outer join, inner:IndexReader, outer key:test.queries_identifier.id, inner key:test.queries_channel.identifier_id, equal cond:eq(test.queries_identifier.id, test.queries_channel.identifier_id)",
-		"      ├─IndexHashJoin(Build) 2.50 root  inner join, inner:TableReader, outer key:test.queries_program.identifier_id, inner key:test.queries_identifier.id, equal cond:eq(test.queries_program.identifier_id, test.queries_identifier.id)",
-		"      │ ├─Batch_Point_Get(Build) 2.00 root table:queries_program handle:[8 9], keep order:false, desc:false",
-		"      │ └─TableReader(Probe) 2.00 root  data:TableRangeScan",
-		"      │   └─TableRangeScan 2.00 cop[tikv] table:queries_identifier range: decided by [test.queries_program.identifier_id], keep order:false, stats:pseudo",
-		"      └─IndexReader(Probe) 2.50 root  index:IndexRangeScan",
-		"        └─IndexRangeScan 2.50 cop[tikv] table:queries_channel, index:identifier_id(identifier_id) range: decided by [eq(test.queries_channel.identifier_id, test.queries_identifier.id)], keep order:false, stats:pseudo"))
+	tk.MustExec("create table t(a int)")
+	tk.MustExec("create table t1(a int, b int)")
+	tk.MustExec("create table t2(a int, b int, c int)")
+	tk.MustExec("create table t3(a int, b int)")
+	tk.MustExec("create table t4(a int, b int)")
+
+	testData := core.GetJoinReorderData()
+	var (
+		input  []string
+		output []struct {
+			SQL    string
+			Output []string
+		}
+	)
+	testData.LoadTestCases(t, &input, &output)
+	for i, sql := range input {
+		testdata.OnRecord(func() {
+			output[i].SQL = sql
+			output[i].Output = testdata.ConvertRowsToStrings(tk.MustQuery(sql).Rows())
+		})
+		tk.MustQuery(sql).Check(testkit.Rows(output[i].Output...))
+	}
 }
 
 func TestOuterJoinWIthEqCondCrossInnerJoin(t *testing.T) {

planner/core/testdata/join_reorder_suite_in.json

@@ -0,0 +1,12 @@
+[
+  {
+    "name": "TestAdditionOtherConditionsRemained4OuterJoin",
+    "cases": [
+        "SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` LEFT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC",
+        "SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` RIGHT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC",
+        "explain format = 'brief' SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` LEFT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC;",
+        // The where clause should be a Selection out of joins.
+        "explain format='brief' select * from t left join t1 on t.a=t1.a inner join t2 on t.a=t2.a and t2.c = 100 left join t3 on t2.a=t3.a and t3.b > 1 left join t4 on t2.a = t4.a where (t2.b > 100 or t.a > 10 or t1.b < 10)"
+    ]
+  }
+]

planner/core/testdata/join_reorder_suite_out.json

@@ -0,0 +1,60 @@
+[
+  {
+    "Name": "TestAdditionOtherConditionsRemained4OuterJoin",
+    "Cases": [
+      {
+        "SQL": "SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` LEFT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC",
+        "Output": [
+          "13 i1"
+        ]
+      },
+      {
+        "SQL": "SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` RIGHT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC",
+        "Output": [
+          "13 i1"
+        ]
+      },
+      {
+        "SQL": "explain format = 'brief' SELECT `queries_identifier`.`id`, `queries_identifier`.`name` FROM `queries_identifier` LEFT OUTER JOIN `queries_channel` ON (`queries_identifier`.`id` = `queries_channel`.`identifier_id`) INNER JOIN `queries_program` ON (`queries_identifier`.`id` = `queries_program`.`identifier_id`) WHERE ((`queries_channel`.`id` = 5 AND `queries_program`.`id` = 9) OR `queries_program`.`id` = 8) ORDER BY `queries_identifier`.`id` ASC;",
+        "Output": [
+          "Sort 2.50 root  test.queries_identifier.id",
+          "└─Projection 2.50 root  test.queries_identifier.id, test.queries_identifier.name",
+          "  └─Selection 2.50 root  or(and(eq(test.queries_channel.id, 5), eq(test.queries_program.id, 9)), eq(test.queries_program.id, 8))",
+          "    └─IndexJoin 3.12 root  left outer join, inner:IndexReader, outer key:test.queries_identifier.id, inner key:test.queries_channel.identifier_id, equal cond:eq(test.queries_identifier.id, test.queries_channel.identifier_id)",
+          "      ├─IndexHashJoin(Build) 2.50 root  inner join, inner:TableReader, outer key:test.queries_program.identifier_id, inner key:test.queries_identifier.id, equal cond:eq(test.queries_program.identifier_id, test.queries_identifier.id)",
+          "      │ ├─Batch_Point_Get(Build) 2.00 root table:queries_program handle:[8 9], keep order:false, desc:false",
+          "      │ └─TableReader(Probe) 2.00 root  data:TableRangeScan",
+          "      │   └─TableRangeScan 2.00 cop[tikv] table:queries_identifier range: decided by [test.queries_program.identifier_id], keep order:false, stats:pseudo",
+          "      └─IndexReader(Probe) 2.50 root  index:IndexRangeScan",
+          "        └─IndexRangeScan 2.50 cop[tikv] table:queries_channel, index:identifier_id(identifier_id) range: decided by [eq(test.queries_channel.identifier_id, test.queries_identifier.id)], keep order:false, stats:pseudo"
+        ]
+      },
+      {
+        "SQL": "explain format='brief' select * from t left join t1 on t.a=t1.a inner join t2 on t.a=t2.a and t2.c = 100 left join t3 on t2.a=t3.a and t3.b > 1 left join t4 on t2.a = t4.a where (t2.b > 100 or t.a > 10 or t1.b < 10)",
+        "Output": [
+          "Projection 19.51 root  test.t.a, test.t1.a, test.t1.b, test.t2.a, test.t2.b, test.t2.c, test.t3.a, test.t3.b, test.t4.a, test.t4.b",
+          "└─Selection 19.51 root  or(gt(test.t2.b, 100), or(gt(test.t.a, 10), lt(test.t1.b, 10))), or(gt(test.t2.b, 100), or(gt(test.t2.a, 10), lt(test.t1.b, 10)))",
+          "  └─HashJoin 24.39 root  left outer join, equal:[eq(test.t2.a, test.t4.a)]",
+          "    ├─HashJoin(Build) 19.51 root  left outer join, equal:[eq(test.t.a, test.t1.a)]",
+          "    │ ├─HashJoin(Build) 15.61 root  inner join, equal:[eq(test.t2.a, test.t.a)]",
+          "    │ │ ├─HashJoin(Build) 12.49 root  left outer join, equal:[eq(test.t2.a, test.t3.a)]",
+          "    │ │ │ ├─TableReader(Build) 9.99 root  data:Selection",
+          "    │ │ │ │ └─Selection 9.99 cop[tikv]  eq(test.t2.c, 100), not(isnull(test.t2.a))",
+          "    │ │ │ │   └─TableFullScan 10000.00 cop[tikv] table:t2 keep order:false, stats:pseudo",
+          "    │ │ │ └─TableReader(Probe) 3330.00 root  data:Selection",
+          "    │ │ │   └─Selection 3330.00 cop[tikv]  gt(test.t3.b, 1), not(isnull(test.t3.a))",
+          "    │ │ │     └─TableFullScan 10000.00 cop[tikv] table:t3 keep order:false, stats:pseudo",
+          "    │ │ └─TableReader(Probe) 9990.00 root  data:Selection",
+          "    │ │   └─Selection 9990.00 cop[tikv]  not(isnull(test.t.a))",
+          "    │ │     └─TableFullScan 10000.00 cop[tikv] table:t keep order:false, stats:pseudo",
+          "    │ └─TableReader(Probe) 9990.00 root  data:Selection",
+          "    │   └─Selection 9990.00 cop[tikv]  not(isnull(test.t1.a))",
+          "    │     └─TableFullScan 10000.00 cop[tikv] table:t1 keep order:false, stats:pseudo",
+          "    └─TableReader(Probe) 9990.00 root  data:Selection",
+          "      └─Selection 9990.00 cop[tikv]  not(isnull(test.t4.a))",
+          "        └─TableFullScan 10000.00 cop[tikv] table:t4 keep order:false, stats:pseudo"
+        ]
+      }
+    ]
+  }
+]