util: fix build ranger's collation (pingcap#31226)

close pingcap#31216

cmd/explaintest/r/collation_check_use_collation.result

@@ -22,4 +22,13 @@ a
 select a as a_col from t where t.a = (select a collate utf8mb4_general_ci from t1);
 a_col
 a
+drop table if exists tbl_2;
+create table tbl_2 ( col_20 bigint not null , col_21 smallint not null , col_22 decimal(24,10) default null , col_23 tinyint default 71 not null , col_24 bigint not null , col_25 tinyint default 18 , col_26 varchar(330) collate utf8_bin not null , col_27 char(77) collate                    utf8mb4_unicode_ci , col_28 char(46) collate utf8_general_ci not null , col_29 smallint unsigned not null , primary key idx_13 ( col_27(5) ) , key idx_14 ( col_24 ) , unique key idx_15 ( col_23,col_21,col_28,col_29,col_24 ) ) collate utf8_bin ;
+insert ignore into tbl_2 values ( 5888267793391993829,5371,94.63,-109,5728076076919247337,89,'WUicqUTgdGJcjbC','SapBPqczTWWSN','xUSwH',49462 );
+select col_25 from tbl_2 where ( tbl_2.col_27 > 'nSWYrpTH' or not( tbl_2.col_27 between 'CsWIuxlSjU' and 'SfwoyjUEzgg' ) ) and ( tbl_2.col_23 <= -95);
+col_25
+89
+select col_25 from tbl_2 use index(primary) where ( tbl_2.col_27 > 'nSWYrpTH' or not( tbl_2.col_27 between 'CsWIuxlSjU' and 'SfwoyjUEzgg' ) ) and ( tbl_2.col_23 <= -95);
+col_25
+89
 use test

cmd/explaintest/t/collation_check_use_collation.test

@@ -19,5 +19,11 @@ select a as a_col from t where t.a <= all (select a collate utf8mb4_general_ci f
 select a as a_col from t where t.a <= any (select a collate utf8mb4_general_ci from t1);
 select a as a_col from t where t.a = (select a collate utf8mb4_general_ci from t1);
 
+drop table if exists tbl_2;
+create table tbl_2 ( col_20 bigint not null , col_21 smallint not null , col_22 decimal(24,10) default null , col_23 tinyint default 71 not null , col_24 bigint not null , col_25 tinyint default 18 , col_26 varchar(330) collate utf8_bin not null , col_27 char(77) collate                    utf8mb4_unicode_ci , col_28 char(46) collate utf8_general_ci not null , col_29 smallint unsigned not null , primary key idx_13 ( col_27(5) ) , key idx_14 ( col_24 ) , unique key idx_15 ( col_23,col_21,col_28,col_29,col_24 ) ) collate utf8_bin ;
+insert ignore into tbl_2 values ( 5888267793391993829,5371,94.63,-109,5728076076919247337,89,'WUicqUTgdGJcjbC','SapBPqczTWWSN','xUSwH',49462 );
+select col_25 from tbl_2 where ( tbl_2.col_27 > 'nSWYrpTH' or not( tbl_2.col_27 between 'CsWIuxlSjU' and 'SfwoyjUEzgg' ) ) and ( tbl_2.col_23 <= -95);
+select col_25 from tbl_2 use index(primary) where ( tbl_2.col_27 > 'nSWYrpTH' or not( tbl_2.col_27 between 'CsWIuxlSjU' and 'SfwoyjUEzgg' ) ) and ( tbl_2.col_23 <= -95);
+
 # cleanup environment
 use test

util/ranger/detacher.go

@@ -529,11 +529,12 @@ func ExtractEqAndInCondition(sctx sessionctx.Context, conditions []expression.Ex
 		}
 		// Multiple Eq/In conditions for one column in CNF, apply intersection on them
 		// Lazily compute the points for the previously visited Eq/In
+		collator := collate.GetCollator(cols[offset].GetType().Collate)
 		if mergedAccesses[offset] == nil {
 			mergedAccesses[offset] = accesses[offset]
-			points[offset] = rb.build(accesses[offset])
+			points[offset] = rb.build(accesses[offset], collator)
 		}
-		points[offset] = rb.intersection(points[offset], rb.build(cond), collate.GetCollator(cols[offset].GetType().Collate))
+		points[offset] = rb.intersection(points[offset], rb.build(cond, collator), collator)
 		if len(points[offset]) == 0 { // Early termination if false expression found
 			if expression.MaybeOverOptimized4PlanCache(sctx, conditions) {
 				// cannot return an empty-range for plan-cache since the range may become non-empty as parameters change
@@ -665,7 +666,7 @@ func (d *rangeDetacher) detachDNFCondAndBuildRangeForIndex(condition *expression
 				hasResidual = true
 				firstColumnChecker.shouldReserve = d.lengths[0] != types.UnspecifiedLength
 			}
-			points := rb.build(item)
+			points := rb.build(item, collate.GetCollator(newTpSlice[0].Collate))
 			ranges, err := points2Ranges(d.sctx, points, newTpSlice[0])
 			if err != nil {
 				return nil, nil, nil, false, errors.Trace(err)

util/ranger/points.go

@@ -185,12 +185,12 @@ type builder struct {
 	sc  *stmtctx.StatementContext
 }
 
-func (r *builder) build(expr expression.Expression) []*point {
+func (r *builder) build(expr expression.Expression, collator collate.Collator) []*point {
 	switch x := expr.(type) {
 	case *expression.Column:
 		return r.buildFromColumn(x)
 	case *expression.ScalarFunction:
-		return r.buildFromScalarFunc(x)
+		return r.buildFromScalarFunc(x, collator)
 	case *expression.Constant:
 		return r.buildFromConstant(x)
 	}
@@ -762,16 +762,14 @@ func (r *builder) buildFromNot(expr *expression.ScalarFunction) []*point {
 	return getFullRange()
 }
 
-func (r *builder) buildFromScalarFunc(expr *expression.ScalarFunction) []*point {
-	_, coll := expr.CharsetAndCollation()
-	collator := collate.GetCollator(coll)
+func (r *builder) buildFromScalarFunc(expr *expression.ScalarFunction, collator collate.Collator) []*point {
 	switch op := expr.FuncName.L; op {
 	case ast.GE, ast.GT, ast.LT, ast.LE, ast.EQ, ast.NE, ast.NullEQ:
 		return r.buildFromBinOp(expr)
 	case ast.LogicAnd:
-		return r.intersection(r.build(expr.GetArgs()[0]), r.build(expr.GetArgs()[1]), collator)
+		return r.intersection(r.build(expr.GetArgs()[0], collator), r.build(expr.GetArgs()[1], collator), collator)
 	case ast.LogicOr:
-		return r.union(r.build(expr.GetArgs()[0]), r.build(expr.GetArgs()[1]), collator)
+		return r.union(r.build(expr.GetArgs()[0], collator), r.build(expr.GetArgs()[1], collator), collator)
 	case ast.IsTruthWithoutNull:
 		return r.buildFromIsTrue(expr, 0, false)
 	case ast.IsTruthWithNull:

util/ranger/ranger.go

@@ -312,7 +312,8 @@ func buildColumnRange(accessConditions []expression.Expression, sctx sessionctx.
 	rb := builder{sc: sctx.GetSessionVars().StmtCtx}
 	rangePoints := getFullRange()
 	for _, cond := range accessConditions {
-		rangePoints = rb.intersection(rangePoints, rb.build(cond), collate.GetCollator(tp.Collate))
+		collator := collate.GetCollator(tp.Collate)
+		rangePoints = rb.intersection(rangePoints, rb.build(cond, collator), collator)
 		if rb.err != nil {
 			return nil, errors.Trace(rb.err)
 		}
@@ -372,7 +373,7 @@ func (d *rangeDetacher) buildCNFIndexRange(newTp []*types.FieldType,
 	}
 	for i := 0; i < eqAndInCount; i++ {
 		// Build ranges for equal or in access conditions.
-		point := rb.build(accessCondition[i])
+		point := rb.build(accessCondition[i], collate.GetCollator(newTp[i].Collate))
 		if rb.err != nil {
 			return nil, errors.Trace(rb.err)
 		}
@@ -388,7 +389,8 @@ func (d *rangeDetacher) buildCNFIndexRange(newTp []*types.FieldType,
 	rangePoints := getFullRange()
 	// Build rangePoints for non-equal access conditions.
 	for i := eqAndInCount; i < len(accessCondition); i++ {
-		rangePoints = rb.intersection(rangePoints, rb.build(accessCondition[i]), collate.GetCollator(newTp[eqAndInCount].Collate))
+		collator := collate.GetCollator(newTp[eqAndInCount].Collate)
+		rangePoints = rb.intersection(rangePoints, rb.build(accessCondition[i], collator), collator)
 		if rb.err != nil {
 			return nil, errors.Trace(rb.err)
 		}