expression: impl cast as array function

errors.toml

@@ -1696,6 +1696,21 @@ error = '''
 Invalid data type for JSON data in argument %d to function %s; a JSON string or JSON type is required.
 '''
 
+["expression:3752"]
+error = '''
+Value is out of range for expression index '%s' at row %d
+'''
+
+["expression:3903"]
+error = '''
+Invalid JSON value for CAST for expression index '%s'
+'''
+
+["expression:3907"]
+error = '''
+Data too long for expression index '%s'
+'''
+
 ["expression:8128"]
 error = '''
 Invalid TABLESAMPLE: %s

executor/insert_common.go

@@ -388,7 +388,7 @@ func (e *InsertValues) evalRow(ctx context.Context, list []expression.Expression
 		e.evalBuffer.SetDatum(offset, val1)
 	}
 	// Row may lack of generated column, autoIncrement column, empty column here.
-	return e.fillRow(ctx, row, hasValue)
+	return e.fillRow(ctx, row, hasValue, rowIdx)
 }
 
 var emptyRow chunk.Row
@@ -422,7 +422,7 @@ func (e *InsertValues) fastEvalRow(ctx context.Context, list []expression.Expres
 		offset := e.insertColumns[i].Offset
 		row[offset], hasValue[offset] = val1, true
 	}
-	return e.fillRow(ctx, row, hasValue)
+	return e.fillRow(ctx, row, hasValue, rowIdx)
 }
 
 // setValueForRefColumn set some default values for the row to eval the row value with other columns,
@@ -562,7 +562,7 @@ func (e *InsertValues) getRow(ctx context.Context, vals []types.Datum) ([]types.
 		hasValue[offset] = true
 	}
 
-	return e.fillRow(ctx, row, hasValue)
+	return e.fillRow(ctx, row, hasValue, 0)
 }
 
 // getColDefaultValue gets the column default value.
@@ -647,7 +647,7 @@ func (e *InsertValues) fillColValue(ctx context.Context, datum types.Datum, idx
 // `insert|replace values` can guarantee consecutive autoID in a batch.
 // Other statements like `insert select from` don't guarantee consecutive autoID.
 // https://dev.mysql.com/doc/refman/8.0/en/innodb-auto-increment-handling.html
-func (e *InsertValues) fillRow(ctx context.Context, row []types.Datum, hasValue []bool) ([]types.Datum, error) {
+func (e *InsertValues) fillRow(ctx context.Context, row []types.Datum, hasValue []bool, rowIdx int) ([]types.Datum, error) {
 	gCols := make([]*table.Column, 0)
 	tCols := e.Table.Cols()
 	if e.hasExtraHandle {
@@ -693,6 +693,9 @@ func (e *InsertValues) fillRow(ctx context.Context, row []types.Datum, hasValue
 	for i, gCol := range gCols {
 		colIdx := gCol.ColumnInfo.Offset
 		val, err := e.GenExprs[i].Eval(chunk.MutRowFromDatums(row).ToRow())
+		if err != nil && gCol.FieldType.IsArray() {
+			return nil, completeError(tbl, gCol.Offset, rowIdx, err)
+		}
 		if e.ctx.GetSessionVars().StmtCtx.HandleTruncate(err) != nil {
 			return nil, err
 		}
@@ -708,6 +711,29 @@ func (e *InsertValues) fillRow(ctx context.Context, row []types.Datum, hasValue
 	return row, nil
 }
 
+func completeError(tbl *model.TableInfo, offset int, rowIdx int, err error) error {
+	name := "expression_index"
+	for _, idx := range tbl.Indices {
+		for _, column := range idx.Columns {
+			if column.Offset == offset {
+				name = idx.Name.O
+				break
+			}
+		}
+	}
+
+	if expression.ErrInvalidJSONForFuncIndex.Equal(err) {
+		return expression.ErrInvalidJSONForFuncIndex.GenWithStackByArgs(name)
+	}
+	if types.ErrOverflow.Equal(err) {
+		return expression.ErrDataOutOfRangeFuncIndex.GenWithStackByArgs(name, rowIdx+1)
+	}
+	if types.ErrDataTooLong.Equal(err) {
+		return expression.ErrFuncIndexDataIsTooLong.GenWithStackByArgs(name)
+	}
+	return err
+}
+
 // isAutoNull can help judge whether a datum is AutoIncrement Null quickly.
 // This used to help lazyFillAutoIncrement to find consecutive N datum backwards for batch autoID alloc.
 func (e *InsertValues) isAutoNull(ctx context.Context, d types.Datum, col *table.Column) bool {

expression/builtin_cast.go

@@ -36,6 +36,7 @@ import (
 	"github.com/pingcap/tidb/parser/mysql"
 	"github.com/pingcap/tidb/parser/terror"
 	"github.com/pingcap/tidb/sessionctx"
+	"github.com/pingcap/tidb/sessionctx/stmtctx"
 	"github.com/pingcap/tidb/sessionctx/variable"
 	"github.com/pingcap/tidb/types"
 	"github.com/pingcap/tidb/util/chunk"
@@ -420,7 +421,7 @@ func (c *castAsArrayFunctionClass) verifyArgs(args []Expression) error {
 	}
 
 	if args[0].GetType().EvalType() != types.ETJson {
-		return types.ErrInvalidJSONData.GenWithStackByArgs("1", "cast_as_array")
+		return ErrInvalidTypeForJSON.GenWithStackByArgs(1, "cast_as_array")
 	}
 
 	return nil
@@ -467,9 +468,80 @@ func (b *castJSONAsArrayFunctionSig) evalJSON(row chunk.Row) (res types.BinaryJS
 		return types.BinaryJSON{}, false, ErrNotSupportedYet.GenWithStackByArgs("CAST-ing Non-JSON Array type to array")
 	}
 
-	// TODO: impl the cast(... as ... array) function
+	arrayVals := make([]any, 0, len(b.args))
+	ft := b.tp.ArrayType()
+	f := convertJSON2Tp(ft.EvalType())
+	if f == nil {
+		return types.BinaryJSON{}, false, ErrNotSupportedYet.GenWithStackByArgs("CAS-ing JSON to the target type")
+	}
+	sc := b.ctx.GetSessionVars().StmtCtx
+	originalOverflowAsWarning := sc.OverflowAsWarning
+	originIgnoreTruncate := sc.IgnoreTruncate
+	originTruncateAsWarning := sc.TruncateAsWarning
+	sc.OverflowAsWarning = false
+	sc.IgnoreTruncate = false
+	sc.TruncateAsWarning = false
+	defer func() {
+		sc.OverflowAsWarning = originalOverflowAsWarning
+		sc.IgnoreTruncate = originIgnoreTruncate
+		sc.TruncateAsWarning = originTruncateAsWarning
+	}()
+	for i := 0; i < val.GetElemCount(); i++ {
+		item, err := f(sc, val.ArrayGetElem(i), ft)
+		if err != nil {
+			return types.BinaryJSON{}, false, err
+		}
+		arrayVals = append(arrayVals, item)
+	}
+	return types.CreateBinaryJSON(arrayVals), false, nil
+}
 
-	return types.BinaryJSON{}, false, nil
+func convertJSON2Tp(eval types.EvalType) func(*stmtctx.StatementContext, types.BinaryJSON, *types.FieldType) (any, error) {
+	switch eval {
+	case types.ETString:
+		return func(sc *stmtctx.StatementContext, item types.BinaryJSON, tp *types.FieldType) (any, error) {
+			if item.TypeCode != types.JSONTypeCodeString {
+				return nil, ErrInvalidJSONForFuncIndex
+			}
+			return types.ProduceStrWithSpecifiedTp(string(item.GetString()), tp, sc, false)
+		}
+	case types.ETInt:
+		return func(sc *stmtctx.StatementContext, item types.BinaryJSON, tp *types.FieldType) (any, error) {
+			if item.TypeCode != types.JSONTypeCodeInt64 && item.TypeCode != types.JSONTypeCodeUint64 {
+				return nil, ErrInvalidJSONForFuncIndex
+			}
+			return types.ConvertJSONToInt(sc, item, mysql.HasUnsignedFlag(tp.GetFlag()), tp.GetType())
+		}
+	case types.ETReal, types.ETDecimal:
+		return func(sc *stmtctx.StatementContext, item types.BinaryJSON, tp *types.FieldType) (any, error) {
+			if item.TypeCode != types.JSONTypeCodeInt64 && item.TypeCode != types.JSONTypeCodeUint64 && item.TypeCode != types.JSONTypeCodeFloat64 {
+				return nil, ErrInvalidJSONForFuncIndex
+			}
+			return types.ConvertJSONToFloat(sc, item)
+		}
+	case types.ETDatetime:
+		return func(sc *stmtctx.StatementContext, item types.BinaryJSON, tp *types.FieldType) (any, error) {
+			if (tp.GetType() == mysql.TypeDatetime && item.TypeCode != types.JSONTypeCodeDatetime) || (tp.GetType() == mysql.TypeDate && item.TypeCode != types.JSONTypeCodeDate) {
+				return nil, ErrInvalidJSONForFuncIndex
+			}
+			res := item.GetTime()
+			res.SetType(tp.GetType())
+			if tp.GetType() == mysql.TypeDate {
+				// Truncate hh:mm:ss part if the type is Date.
+				res.SetCoreTime(types.FromDate(res.Year(), res.Month(), res.Day(), 0, 0, 0, 0))
+			}
+			return res, nil
+		}
+	case types.ETDuration:
+		return func(sc *stmtctx.StatementContext, item types.BinaryJSON, tp *types.FieldType) (any, error) {
+			if item.TypeCode != types.JSONTypeCodeDuration {
+				return nil, ErrInvalidJSONForFuncIndex
+			}
+			return item.GetDuration(), nil
+		}
+	default:
+		return nil
+	}
 }
 
 type castAsJSONFunctionClass struct {

expression/builtin_cast_test.go

@@ -1619,3 +1619,73 @@ func TestCastBinaryStringAsJSONSig(t *testing.T) {
 		require.Equal(t, tt.resultStr, res.String())
 	}
 }
+
+func TestCastArrayFunc(t *testing.T) {
+	ctx := createContext(t)
+	tbl := []struct {
+		input            interface{}
+		expected         interface{}
+		tp               *types.FieldType
+		success          bool
+		buildFuncSuccess bool
+	}{
+		{
+			[]interface{}{int64(-1), int64(2), int64(3)},
+			[]interface{}{int64(-1), int64(2), int64(3)},
+			types.NewFieldTypeBuilder().SetType(mysql.TypeLonglong).SetCharset(charset.CharsetBin).SetCollate(charset.CollationBin).SetArray(true).BuildP(),
+			true,
+			true,
+		},
+		{
+			[]interface{}{int64(-1), int64(2), int64(3)},
+			nil,
+			types.NewFieldTypeBuilder().SetType(mysql.TypeString).SetCharset(charset.CharsetUTF8MB4).SetCollate(charset.CollationUTF8MB4).SetArray(true).BuildP(),
+			false,
+			true,
+		},
+		{
+			[]interface{}{"1"},
+			nil,
+			types.NewFieldTypeBuilder().SetType(mysql.TypeLonglong).SetCharset(charset.CharsetBin).SetCollate(charset.CharsetBin).SetArray(true).BuildP(),
+			false,
+			true,
+		},
+		{
+			[]interface{}{"1", "2"},
+			nil,
+			types.NewFieldTypeBuilder().SetType(mysql.TypeDouble).SetCharset(charset.CharsetBin).SetCollate(charset.CharsetBin).SetArray(true).BuildP(),
+			false,
+			true,
+		},
+		{
+			[]interface{}{int64(-1), 2.1, int64(3)},
+			[]interface{}{int64(-1), 2.1, int64(3)},
+			types.NewFieldTypeBuilder().SetType(mysql.TypeDouble).SetCharset(charset.CharsetBin).SetCollate(charset.CharsetBin).SetArray(true).BuildP(),
+			true,
+			true,
+		},
+	}
+	for _, tt := range tbl {
+		f, err := BuildCastFunctionWithCheck(ctx, datumsToConstants(types.MakeDatums(types.CreateBinaryJSON(tt.input)))[0], tt.tp)
+		if tt.buildFuncSuccess {
+			require.NoError(t, err, tt.input)
+		} else {
+			require.Error(t, err, tt.input)
+			continue
+		}
+
+		val, isNull, err := f.EvalJSON(ctx, chunk.Row{})
+		if tt.success {
+			require.NoError(t, err, tt.input)
+			if tt.expected == nil {
+				require.True(t, isNull, tt.input)
+			} else {
+				j1 := types.CreateBinaryJSON(tt.expected)
+				cmp := types.CompareBinaryJSON(j1, val)
+				require.Equal(t, 0, cmp, tt.input)
+			}
+		} else {
+			require.Error(t, err, tt.input)
+		}
+	}
+}

expression/errors.go

@@ -38,6 +38,9 @@ var (
 	ErrInternal                    = dbterror.ClassOptimizer.NewStd(mysql.ErrInternal)
 	ErrNoDB                        = dbterror.ClassOptimizer.NewStd(mysql.ErrNoDB)
 	ErrNotSupportedYet             = dbterror.ClassExpression.NewStd(mysql.ErrNotSupportedYet)
+	ErrInvalidJSONForFuncIndex     = dbterror.ClassExpression.NewStd(mysql.ErrInvalidJSONValueForFuncIndex)
+	ErrDataOutOfRangeFuncIndex     = dbterror.ClassExpression.NewStd(mysql.ErrDataOutOfRangeFunctionalIndex)
+	ErrFuncIndexDataIsTooLong      = dbterror.ClassExpression.NewStd(mysql.ErrFunctionalIndexDataIsTooLong)
 
 	// All the un-exported errors are defined here:
 	errFunctionNotExists             = dbterror.ClassExpression.NewStd(mysql.ErrSpDoesNotExist)