Gen4: add __sq_has_values for in/not in subqueries

go/test/endtoend/vtgate/misc_test.go

@@ -753,6 +753,19 @@ func TestSimpleOrderBy(t *testing.T) {
 	assertMatches(t, conn, `SELECT id2 FROM t1 ORDER BY id2 ASC`, `[[INT64(5)] [INT64(6)] [INT64(7)] [INT64(8)] [INT64(9)] [INT64(10)]]`)
 }
 
+func TestSubqueriesHasValues(t *testing.T) {
+	defer cluster.PanicHandler(t)
+	ctx := context.Background()
+	conn, err := mysql.Connect(ctx, &vtParams)
+	require.NoError(t, err)
+	defer conn.Close()
+
+	defer exec(t, conn, `delete from t1`)
+	exec(t, conn, "insert into t1(id1, id2) values (0,1),(1,2),(2,3),(3,4),(4,5),(5,6)")
+	assertMatches(t, conn, `SELECT id2 FROM t1 WHERE id1 IN (SELECT id1 FROM t1 WHERE id1 > 10)`, `[]`)
+	assertMatches(t, conn, `SELECT id2 FROM t1 WHERE id1 NOT IN (SELECT id1 FROM t1 WHERE id1 > 10) ORDER BY id2`, `[[INT64(1)] [INT64(2)] [INT64(3)] [INT64(4)] [INT64(5)] [INT64(6)]]`)
+}
+
 func TestSelectEqualUniqueOuterJoinRightPredicate(t *testing.T) {
 	defer cluster.PanicHandler(t)
 	ctx := context.Background()

go/vt/sqlparser/analyzer.go

@@ -326,8 +326,23 @@ func SplitAndExpression(filters []Expr, node Expr) []Expr {
 	return append(filters, node)
 }
 
-// AndExpressions ands together two expression, minimising the expr when possible
-func AndExpressions(exprs ...Expr) Expr {
+// SplitOrExpression breaks up the Expr into OR-separated conditions
+// and appends them to filters. Outer parenthesis are removed. Precedence
+// should be taken into account if expressions are recombined.
+func SplitOrExpression(filters []Expr, node Expr) []Expr {
+	if node == nil {
+		return filters
+	}
+	switch node := node.(type) {
+	case *OrExpr:
+		filters = SplitOrExpression(filters, node.Left)
+		return SplitOrExpression(filters, node.Right)
+	}
+	return append(filters, node)
+}
+
+// joinExpressions join together a list of Expr using the baseType as operator (either AndExpr or OrExpr).
+func joinExpressions(baseType Expr, exprs ...Expr) Expr {
 	switch len(exprs) {
 	case 0:
 		return nil
@@ -336,6 +351,9 @@ func AndExpressions(exprs ...Expr) Expr {
 	default:
 		result := (Expr)(nil)
 		for i, expr := range exprs {
+			if expr == nil {
+				continue
+			}
 			if result == nil {
 				result = expr
 			} else {
@@ -347,12 +365,28 @@ func AndExpressions(exprs ...Expr) Expr {
 					}
 				}
 				if !found {
-					result = &AndExpr{Left: result, Right: expr}
+					switch baseType.(type) {
+					case *AndExpr:
+						result = &AndExpr{Left: result, Right: expr}
+					case *OrExpr:
+						result = &OrExpr{Left: result, Right: expr}
+					}
 				}
 			}
 		}
 		return result
 	}
+
+}
+
+// AndExpressions ands together two or more expressions, minimising the expr when possible
+func AndExpressions(exprs ...Expr) Expr {
+	return joinExpressions(&AndExpr{}, exprs...)
+}
+
+// OrExpressions ors together two or more expressions, minimising the expr when possible
+func OrExpressions(exprs ...Expr) Expr {
+	return joinExpressions(&OrExpr{}, exprs...)
 }
 
 // TableFromStatement returns the qualified table name for the query.

go/vt/sqlparser/ast_rewriting.go

@@ -101,6 +101,22 @@ func (r *ReservedVars) ReserveSubQuery() string {
 	}
 }
 
+// ReserveSubQueryWithHasValues returns the next argument name to replace subquery with pullout value.
+func (r *ReservedVars) ReserveSubQueryWithHasValues() (string, string) {
+	for {
+		r.sqNext++
+		joinVar := strconv.AppendInt(subQueryBaseArgName, r.sqNext, 10)
+		hasValuesJoinVar := strconv.AppendInt(HasValueSubQueryBaseName, r.sqNext, 10)
+		_, joinVarOK := r.reserved[string(joinVar)]
+		_, hasValuesJoinVarOK := r.reserved[string(hasValuesJoinVar)]
+		if !joinVarOK && !hasValuesJoinVarOK {
+			r.reserved[string(joinVar)] = struct{}{}
+			r.reserved[string(hasValuesJoinVar)] = struct{}{}
+			return string(joinVar), string(hasValuesJoinVar)
+		}
+	}
+}
+
 // ReserveHasValuesSubQuery returns the next argument name to replace subquery with has value.
 func (r *ReservedVars) ReserveHasValuesSubQuery() string {
 	for {

go/vt/vtgate/planbuilder/abstract/operator.go

@@ -34,7 +34,6 @@ type (
 	//  *  SubQuery - Represents a query that encapsulates one or more sub-queries (SubQueryInner).
 	//  *  Vindex - Represents a query that selects from vindex tables.
 	//  *  Concatenate - Represents concatenation of the outputs of all the input sources
-	//  *  Distinct - Represents elimination of duplicates from the output of the input source
 	Operator interface {
 		// TableID returns a TableSet of the tables contained within
 		TableID() semantics.TableSet
@@ -211,10 +210,13 @@ func createOperatorFromSelect(sel *sqlparser.Select, semTable *semantics.SemTabl
 				return nil, err
 			}
 			resultantOp.Inner = append(resultantOp.Inner, &SubQueryInner{
-				SelectStatement: subquerySelectStatement,
-				Inner:           opInner,
-				Type:            sq.OpCode,
-				ArgName:         sq.ArgName,
+				SelectStatement:  subquerySelectStatement,
+				Inner:            opInner,
+				Type:             sq.OpCode,
+				ArgName:          sq.ArgName,
+				HasValues:        sq.HasValues,
+				ExprsNeedReplace: sq.ExprsNeedReplace,
+				ReplaceBy:        sq.ReplaceBy,
 			})
 		}
 	}

go/vt/vtgate/planbuilder/abstract/subquery.go

@@ -34,10 +34,35 @@ var _ Operator = (*SubQuery)(nil)
 
 // SubQueryInner stores the subquery information for a select statement
 type SubQueryInner struct {
-	Inner           Operator
-	Type            engine.PulloutOpcode
+	// Inner is the Operator inside the parenthesis of the subquery.
+	// i.e: select (select 1 union select 1), the Inner here would be
+	// of type Concatenate since we have a Union.
+	Inner Operator
+
+	// Type represents the type of the subquery (value, in, not in, exists)
+	Type engine.PulloutOpcode
+
+	// SelectStatement is the inner's select
 	SelectStatement *sqlparser.Select
-	ArgName         string
+
+	// ArgName is the substitution argument string for the subquery.
+	// Subquery argument name looks like: `__sq1`, with `1` being an
+	// unique identifier. This is used when we wish to replace the
+	// subquery by an argument for PullOut subqueries.
+	ArgName string
+
+	// HasValues is a string of form `__sq_has_values1` with `1` being
+	// a unique identifier that matches the one used in ArgName.
+	// We use `__sq_has_values` for in and not in subqueries.
+	HasValues string
+
+	// ExprsNeedReplace is a slice of all the expressions that were
+	// introduced by the rewrite of the subquery and that potentially
+	// need to be re-replace if we can merge the subquery into a route.
+	// An expression that contains at least all of ExprsNeedReplace will
+	// be replaced by the expression in ReplaceBy.
+	ExprsNeedReplace []sqlparser.Expr
+	ReplaceBy        sqlparser.Expr
 }
 
 // TableID implements the Operator interface

go/vt/vtgate/planbuilder/gen4_planner.go

@@ -178,10 +178,11 @@ func newBuildSelectPlan(selStmt sqlparser.SelectStatement, reservedVars *sqlpars
 
 func newPlanningContext(reservedVars *sqlparser.ReservedVars, semTable *semantics.SemTable, vschema ContextVSchema) *planningContext {
 	ctx := &planningContext{
-		reservedVars: reservedVars,
-		semTable:     semTable,
-		vschema:      vschema,
-		sqToReplace:  map[string]*sqlparser.Select{},
+		reservedVars:          reservedVars,
+		semTable:              semTable,
+		vschema:               vschema,
+		argToReplaceBySelect:  map[string]*sqlparser.Select{},
+		exprToReplaceBySqExpr: map[sqlparser.Expr]sqlparser.Expr{},
 	}
 	return ctx
 }
@@ -216,7 +217,7 @@ func planHorizon(ctx *planningContext, plan logicalPlan, in sqlparser.SelectStat
 			sel: node,
 		}
 
-		replaceSubQuery(ctx.sqToReplace, node)
+		replaceSubQuery(ctx.exprToReplaceBySqExpr, node)
 		var err error
 		plan, err = hp.planHorizon(ctx, plan)
 		if err != nil {

go/vt/vtgate/planbuilder/querytree_transformers.go

@@ -98,7 +98,7 @@ func transformSubqueryTree(ctx *planningContext, n *subqueryTree) (logicalPlan,
 		return nil, err
 	}
 
-	plan := newPulloutSubquery(n.opcode, n.argName, "", innerPlan)
+	plan := newPulloutSubquery(n.opcode, n.argName, n.hasValues, innerPlan)
 	outerPlan, err := transformToLogicalPlan(ctx, n.outer)
 	if err != nil {
 		return nil, err
@@ -324,7 +324,7 @@ func transformRoutePlan(ctx *planningContext, n *routeTree) (*route, error) {
 		Comments:    ctx.semTable.Comments,
 	}
 
-	replaceSubQuery(ctx.sqToReplace, sel)
+	replaceSubQuery(ctx.exprToReplaceBySqExpr, sel)
 
 	// TODO clean up when gen4 is the only planner
 	var condition sqlparser.Expr
@@ -439,39 +439,56 @@ func relToTableExpr(t relation) (sqlparser.TableExpr, error) {
 }
 
 type subQReplacer struct {
-	sqToReplace map[string]*sqlparser.Select
-	err         error
-	replaced    bool
+	exprToReplaceBySqExpr map[sqlparser.Expr]sqlparser.Expr
+	replaced              bool
 }
 
 func (sqr *subQReplacer) replacer(cursor *sqlparser.Cursor) bool {
-	argName := argumentName(cursor.Node())
-	if argName == "" {
+	var exprs []sqlparser.Expr
+	switch node := cursor.Node().(type) {
+	case *sqlparser.AndExpr:
+		exprs = sqlparser.SplitAndExpression(nil, node)
+	case *sqlparser.OrExpr:
+		exprs = sqlparser.SplitOrExpression(nil, node)
+	case sqlparser.Argument:
+		exprs = append(exprs, node)
+	case sqlparser.ListArg:
+		exprs = append(exprs, node)
+	case *sqlparser.ExistsExpr:
+		exprs = append(exprs, node)
+	default:
 		return true
 	}
 
-	var node sqlparser.SQLNode
-	subqSelect, exists := sqr.sqToReplace[argName]
-	if !exists {
-		sqr.err = vterrors.Errorf(vtrpcpb.Code_INTERNAL, "[BUG] unable to find subquery with argument: %s", argName)
-		return false
+	var replaceBy sqlparser.Expr
+	var remainder sqlparser.Expr
+	for _, expr := range exprs {
+		found := false
+		for sqExprToReplace, replaceByExpr := range sqr.exprToReplaceBySqExpr {
+			if sqlparser.EqualsExpr(expr, sqExprToReplace) {
+				allReplaceByExprs := sqlparser.SplitAndExpression(nil, replaceBy)
+				allReplaceByExprs = append(allReplaceByExprs, replaceByExpr)
+				replaceBy = sqlparser.AndExpressions(allReplaceByExprs...)
+				found = true
+				break
+			}
+		}
+		if !found {
+			remainder = sqlparser.AndExpressions(remainder, expr)
+		}
 	}
-	sq := &sqlparser.Subquery{Select: subqSelect}
-	node = sq
-
-	// if the subquery is in an EXISTS, e.g. "__sq_has_values1"
-	// then we encapsulate the subquery in an exists expression.
-	if strings.HasPrefix(argName, string(sqlparser.HasValueSubQueryBaseName)) {
-		node = &sqlparser.ExistsExpr{Subquery: sq}
+	if replaceBy == nil {
+		return true
 	}
-	cursor.Replace(node)
+	newNode := sqlparser.AndExpressions(remainder, replaceBy)
+	cursor.Replace(newNode)
 	sqr.replaced = true
 	return false
 }
 
-func replaceSubQuery(sqToReplace map[string]*sqlparser.Select, sel *sqlparser.Select) {
-	if len(sqToReplace) > 0 {
-		sqr := &subQReplacer{sqToReplace: sqToReplace}
+func replaceSubQuery(exprToReplaceBySqExpr map[sqlparser.Expr]sqlparser.Expr, sel *sqlparser.Select) {
+	if len(exprToReplaceBySqExpr) > 0 {
+		sqr := &subQReplacer{exprToReplaceBySqExpr: exprToReplaceBySqExpr}
 		sqlparser.Rewrite(sel, sqr.replacer, nil)
 		for sqr.replaced {
 			// to handle subqueries inside subqueries, we need to do this again and again until no replacements are left