This closes qax-os#2223, limit calculation engine processing to actual data ranges (qax-os#2242)

- Cache formula argument for calculation engine 
- Optimized string concatenation
- Upgrade macos-13 image with macos-15-intel image in GitHub Action

Author: rhinewg

.github/workflows/go.yml

@@ -6,7 +6,7 @@ jobs:
     strategy:
       matrix:
         go-version: [1.24.x, 1.25.x]
-        os: [ubuntu-24.04, macos-13, windows-latest]
+        os: [ubuntu-24.04, macos-15-intel, windows-latest]
         targetplatform: [x86, x64]
 
     runs-on: ${{ matrix.os }}

adjust.go

@@ -75,7 +75,7 @@ func (f *File) adjustHelper(sheet string, dir adjustDirection, num, offset int)
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	sheetID := f.getSheetID(sheet)
 	if dir == rows {
 		err = f.adjustRowDimensions(sheet, ws, num, offset)

calc.go

@@ -193,7 +193,8 @@ var (
 			return fmt.Sprintf("R[%d]C[%d]", row, col), nil
 		},
 	}
-	formulaFormats = []*regexp.Regexp{
+	formulaFnNameReplacer = strings.NewReplacer("_xlfn.", "", ".", "dot")
+	formulaFormats        = []*regexp.Regexp{
 		regexp.MustCompile(`^(\d+)$`),
 		regexp.MustCompile(`^=(.*)$`),
 		regexp.MustCompile(`^<>(.*)$`),
@@ -839,8 +840,8 @@ type formulaFuncs struct {
 //	Z.TEST
 //	ZTEST
 func (f *File) CalcCellValue(sheet, cell string, opts ...Options) (result string, err error) {
-	cacheKey := fmt.Sprintf("%s!%s", sheet, cell)
-	if cachedResult, found := f.calcCache.Load(cacheKey); found {
+	entry := sheet + "!" + cell
+	if cachedResult, ok := f.calcCache.Load(entry); ok {
 		return cachedResult.(string), nil
 	}
 	options := f.getOptions(opts...)
@@ -850,7 +851,7 @@ func (f *File) CalcCellValue(sheet, cell string, opts ...Options) (result string
 		token        formulaArg
 	)
 	if token, err = f.calcCellValue(&calcContext{
-		entry:             fmt.Sprintf("%s!%s", sheet, cell),
+		entry:             entry,
 		maxCalcIterations: options.MaxCalcIterations,
 		iterations:        make(map[string]uint),
 		iterationsCache:   make(map[string]formulaArg),
@@ -866,25 +867,31 @@ func (f *File) CalcCellValue(sheet, cell string, opts ...Options) (result string
 		if precision > 15 {
 			result, err = f.formattedValue(&xlsxC{S: styleIdx, V: strings.ToUpper(strconv.FormatFloat(decimal, 'G', 15, 64))}, rawCellValue, CellTypeNumber)
 			if err == nil {
-				f.calcCache.Store(cacheKey, result)
+				f.calcCache.Store(entry, result)
 			}
 			return
 		}
 		if !strings.HasPrefix(result, "0") {
 			result, err = f.formattedValue(&xlsxC{S: styleIdx, V: strings.ToUpper(strconv.FormatFloat(decimal, 'f', -1, 64))}, rawCellValue, CellTypeNumber)
 		}
 		if err == nil {
-			f.calcCache.Store(cacheKey, result)
+			f.calcCache.Store(entry, result)
 		}
 		return
 	}
 	result, err = f.formattedValue(&xlsxC{S: styleIdx, V: token.Value()}, rawCellValue, CellTypeInlineString)
 	if err == nil {
-		f.calcCache.Store(cacheKey, result)
+		f.calcCache.Store(entry, result)
 	}
 	return
 }
 
+// clearCalcCache clear all calculation related caches.
+func (f *File) clearCalcCache() {
+	f.calcCache.Clear()
+	f.formulaArgCache.Clear()
+}
+
 // calcCellValue calculate cell value by given context, worksheet name and cell
 // reference.
 func (f *File) calcCellValue(ctx *calcContext, sheet, cell string) (result formulaArg, err error) {
@@ -1106,8 +1113,8 @@ func (f *File) evalInfixExpFunc(ctx *calcContext, sheet, cell string, token, nex
 	}
 	prepareEvalInfixExp(opfStack, opftStack, opfdStack, argsStack)
 	// call formula function to evaluate
-	arg := callFuncByName(&formulaFuncs{f: f, sheet: sheet, cell: cell, ctx: ctx}, strings.NewReplacer(
-		"_xlfn.", "", ".", "dot").Replace(opfStack.Peek().(efp.Token).TValue),
+	arg := callFuncByName(&formulaFuncs{f: f, sheet: sheet, cell: cell, ctx: ctx},
+		formulaFnNameReplacer.Replace(opfStack.Peek().(efp.Token).TValue),
 		[]reflect.Value{reflect.ValueOf(argsStack.Peek().(*list.List))})
 	if arg.Type == ArgError && opfStack.Len() == 1 {
 		return arg
@@ -1651,7 +1658,11 @@ func (f *File) cellResolver(ctx *calcContext, sheet, cell string) (formulaArg, e
 		value string
 		err   error
 	)
-	ref := fmt.Sprintf("%s!%s", sheet, cell)
+	ref := sheet + "!" + cell
+	if cached, ok := f.formulaArgCache.Load(ref); ok {
+		return cached.(formulaArg), err
+	}
+
 	if formula, _ := f.getCellFormula(sheet, cell, true); len(formula) != 0 {
 		ctx.mu.Lock()
 		if ctx.entry != ref {
@@ -1660,6 +1671,7 @@ func (f *File) cellResolver(ctx *calcContext, sheet, cell string) (formulaArg, e
 				ctx.mu.Unlock()
 				arg, _ = f.calcCellValue(ctx, sheet, cell)
 				ctx.iterationsCache[ref] = arg
+				f.formulaArgCache.Store(ref, arg)
 				return arg, nil
 			}
 			ctx.mu.Unlock()
@@ -1674,29 +1686,29 @@ func (f *File) cellResolver(ctx *calcContext, sheet, cell string) (formulaArg, e
 	cellType, _ := f.GetCellType(sheet, cell)
 	switch cellType {
 	case CellTypeBool:
-		return arg.ToBool(), err
+		arg = arg.ToBool()
 	case CellTypeNumber, CellTypeUnset:
 		if arg.Value() == "" {
-			return newEmptyFormulaArg(), err
+			arg = newEmptyFormulaArg()
+		} else {
+			arg = arg.ToNumber()
 		}
-		return arg.ToNumber(), err
 	case CellTypeInlineString, CellTypeSharedString:
-		return arg, err
 	case CellTypeFormula:
-		if value != "" {
-			return arg, err
+		if value == "" {
+			arg = newEmptyFormulaArg()
 		}
-		return newEmptyFormulaArg(), err
 	case CellTypeDate:
 		if value, err = f.GetCellValue(sheet, cell); err == nil {
 			if num := newStringFormulaArg(value).ToNumber(); num.Type == ArgNumber {
-				return num, err
+				arg = num
 			}
 		}
-		return arg, err
 	default:
-		return newErrorFormulaArg(value, value), err
+		arg = newErrorFormulaArg(value, value)
 	}
+	f.formulaArgCache.Store(ref, arg)
+	return arg, err
 }
 
 // rangeResolver extract value as string from given reference and range list.
@@ -1735,13 +1747,39 @@ func (f *File) rangeResolver(ctx *calcContext, cellRefs, cellRanges *list.List)
 			return
 		}
 
+		// Detect whole column/row reference, limit to actual data range
+		if valueRange[1] == TotalRows {
+			actualMaxRow := 0
+			for _, rowData := range ws.SheetData.Row {
+				if rowData.R > actualMaxRow {
+					actualMaxRow = rowData.R
+				}
+			}
+			if actualMaxRow > 0 && actualMaxRow < TotalRows {
+				valueRange[1] = actualMaxRow
+			}
+		}
+		if valueRange[3] == MaxColumns {
+			actualMaxCol := 0
+			for _, rowData := range ws.SheetData.Row {
+				for _, cell := range rowData.C {
+					col, _, err := CellNameToCoordinates(cell.R)
+					if err == nil && col > actualMaxCol {
+						actualMaxCol = col
+					}
+				}
+			}
+			if actualMaxCol > 0 && actualMaxCol < MaxColumns {
+				valueRange[3] = actualMaxCol
+			}
+		}
+
 		for row := valueRange[0]; row <= valueRange[1]; row++ {
 			colMax := 0
 			if row <= len(ws.SheetData.Row) {
 				rowData := &ws.SheetData.Row[row-1]
 				colMax = min(valueRange[3], len(rowData.C))
 			}
-
 			var matrixRow []formulaArg
 			for col := valueRange[2]; col <= valueRange[3]; col++ {
 				value := newEmptyFormulaArg()

calc_test.go

@@ -6474,6 +6474,20 @@ func TestCalcRangeResolver(t *testing.T) {
 	cellRefs.PushBack(cellRef{Col: 1, Row: TotalRows + 1, Sheet: "SheetN"})
 	_, err = f.rangeResolver(&calcContext{}, cellRefs, cellRanges)
 	assert.Equal(t, ErrMaxRows, err)
+	t.Run("for_range_resolver_error", func(t *testing.T) {
+		f := NewFile()
+		assert.NoError(t, f.SetCellValue("Sheet1", "A1", "test"))
+		cellRefs := list.New()
+		cellRanges := list.New()
+		cellRanges.PushBack(cellRange{
+			From: cellRef{Col: 1, Row: 1, Sheet: "Sheet1"},
+			To:   cellRef{Col: 1, Row: 1, Sheet: "Sheet1"},
+		})
+		f.SharedStrings = nil
+		f.Pkg.Store(defaultXMLPathSharedStrings, MacintoshCyrillicCharset)
+		_, err := f.rangeResolver(&calcContext{}, cellRefs, cellRanges)
+		assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
+	})
 }
 
 func TestCalcBahttextAppendDigit(t *testing.T) {

cell.go

@@ -182,7 +182,7 @@ func (c *xlsxC) hasValue() bool {
 
 // removeFormula delete formula for the cell.
 func (f *File) removeFormula(c *xlsxC, ws *xlsxWorksheet, sheet string) error {
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	if c.F != nil && c.Vm == nil {
 		sheetID := f.getSheetID(sheet)
 		if err := f.deleteCalcChain(sheetID, c.R); err != nil {
@@ -795,7 +795,7 @@ func (f *File) SetCellFormula(sheet, cell, formula string, opts ...FormulaOpts)
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	if formula == "" {
 		ws.deleteSharedFormula(c)
 		c.F = nil
@@ -1371,7 +1371,7 @@ func (f *File) SetCellRichText(sheet, cell string, runs []RichTextRun) error {
 	if si.R, err = setRichText(runs); err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	for idx, strItem := range sst.SI {
 		if reflect.DeepEqual(strItem, si) {
 			c.T, c.V = "s", strconv.Itoa(idx)

excelize.go

@@ -42,6 +42,7 @@ type File struct {
 	tempFiles        sync.Map
 	xmlAttr          sync.Map
 	calcCache        sync.Map
+	formulaArgCache  sync.Map
 	CalcChain        *xlsxCalcChain
 	CharsetReader    func(charset string, input io.Reader) (rdr io.Reader, err error)
 	Comments         map[string]*xlsxComments

merge.go

@@ -115,7 +115,7 @@ func (f *File) UnmergeCell(sheet, topLeftCell, bottomRightCell string) error {
 	if err = ws.mergeOverlapCells(); err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	i := 0
 	for _, mergeCell := range ws.MergeCells.Cells {
 		if rect2, _ := rangeRefToCoordinates(mergeCell.Ref); isOverlap(rect1, rect2) {

pivotTable.go

@@ -163,7 +163,7 @@ func (f *File) AddPivotTable(opts *PivotTableOptions) error {
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	pivotTableID := f.countPivotTables() + 1
 	pivotCacheID := f.countPivotCache() + 1
 
@@ -1062,7 +1062,7 @@ func (f *File) DeletePivotTable(sheet, name string) error {
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	pivotTableCaches := map[string]int{}
 	pivotTables, _ := f.getPivotTables()
 	for _, sheetPivotTables := range pivotTables {

sheet.go

@@ -384,7 +384,7 @@ func (f *File) SetSheetName(source, target string) error {
 	if target == source {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	wb, _ := f.workbookReader()
 	for k, v := range wb.Sheets.Sheet {
 		if v.Name == source {
@@ -580,7 +580,7 @@ func (f *File) DeleteSheet(sheet string) error {
 	if idx, _ := f.GetSheetIndex(sheet); f.SheetCount == 1 || idx == -1 {
 		return nil
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	wb, _ := f.workbookReader()
 	wbRels, _ := f.relsReader(f.getWorkbookRelsPath())
 	activeSheetName := f.GetSheetName(f.GetActiveSheetIndex())
@@ -769,7 +769,7 @@ func (f *File) copySheet(from, to int) error {
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	worksheet := &xlsxWorksheet{}
 	deepcopy.Copy(worksheet, sheet)
 	toSheetID := strconv.Itoa(f.getSheetID(f.GetSheetName(to)))
@@ -1773,7 +1773,7 @@ func (f *File) SetDefinedName(definedName *DefinedName) error {
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	d := xlsxDefinedName{
 		Name:    definedName.Name,
 		Comment: definedName.Comment,
@@ -1816,7 +1816,7 @@ func (f *File) DeleteDefinedName(definedName *DefinedName) error {
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	if wb.DefinedNames != nil {
 		for idx, dn := range wb.DefinedNames.DefinedName {
 			scope := "Workbook"

table.go

@@ -118,7 +118,7 @@ func (f *File) AddTable(sheet string, table *Table) error {
 		return err
 	}
 	f.addSheetNameSpace(sheet, SourceRelationship)
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	if err = f.addTable(sheet, tableXML, coordinates[0], coordinates[1], coordinates[2], coordinates[3], tableID, options); err != nil {
 		return err
 	}
@@ -178,7 +178,7 @@ func (f *File) DeleteTable(name string) error {
 	if err != nil {
 		return err
 	}
-	f.calcCache.Clear()
+	f.clearCalcCache()
 	for sheet, tables := range tbls {
 		for _, table := range tables {
 			if table.Name != name {