This closes qax-os#1961, add shared formula cell cache for speedup calculation (qax-os#2118)

- Reuse formula shared index when update refer cell formula
- Fix shared cell not been updated on update refer cell formula with new range
- Upgrade dependencies package go-deepcopy to v1.5.2
- Update unit tests

Author: shcabin

cell.go

@@ -689,7 +689,8 @@ func (f *File) getCellFormula(sheet, cell string, transformed bool) (string, err
 			return "", false, nil
 		}
 		if c.F.T == STCellFormulaTypeShared && c.F.Si != nil {
-			return getSharedFormula(x, *c.F.Si, c.R), true, nil
+			formula, err := getSharedFormula(x, *c.F.Si, c.R)
+			return formula, true, err
 		}
 		return c.F.Content, true, nil
 	})
@@ -793,6 +794,7 @@ func (f *File) SetCellFormula(sheet, cell, formula string, opts ...FormulaOpts)
 		return err
 	}
 	if formula == "" {
+		ws.deleteSharedFormula(c)
 		c.F = nil
 		return f.deleteCalcChain(f.getSheetID(sheet), cell)
 	}
@@ -815,7 +817,8 @@ func (f *File) SetCellFormula(sheet, cell, formula string, opts ...FormulaOpts)
 				}
 			}
 			if c.F.T == STCellFormulaTypeShared {
-				if err = ws.setSharedFormula(*opt.Ref); err != nil {
+				ws.deleteSharedFormula(c)
+				if err = ws.setSharedFormula(cell, *opt.Ref); err != nil {
 					return err
 				}
 			}
@@ -890,22 +893,28 @@ func (f *File) setArrayFormulaCells() error {
 }
 
 // setSharedFormula set shared formula for the cells.
-func (ws *xlsxWorksheet) setSharedFormula(ref string) error {
+func (ws *xlsxWorksheet) setSharedFormula(cell, ref string) error {
 	coordinates, err := rangeRefToCoordinates(ref)
 	if err != nil {
 		return err
 	}
 	_ = sortCoordinates(coordinates)
-	cnt := ws.countSharedFormula()
-	for c := coordinates[0]; c <= coordinates[2]; c++ {
-		for r := coordinates[1]; r <= coordinates[3]; r++ {
-			ws.prepareSheetXML(c, r)
-			cell := &ws.SheetData.Row[r-1].C[c-1]
-			if cell.F == nil {
-				cell.F = &xlsxF{}
+	si := ws.countSharedFormula()
+	for col := coordinates[0]; col <= coordinates[2]; col++ {
+		for rol := coordinates[1]; rol <= coordinates[3]; rol++ {
+			ws.prepareSheetXML(col, rol)
+			c := &ws.SheetData.Row[rol-1].C[col-1]
+			if c.F == nil {
+				c.F = &xlsxF{}
 			}
-			cell.F.T = STCellFormulaTypeShared
-			cell.F.Si = &cnt
+			c.F.T = STCellFormulaTypeShared
+			if c.R == cell {
+				if c.F.Ref != "" {
+					si = *c.F.Si
+					continue
+				}
+			}
+			c.F.Si = &si
 		}
 	}
 	return err
@@ -923,6 +932,23 @@ func (ws *xlsxWorksheet) countSharedFormula() (count int) {
 	return
 }
 
+// deleteSharedFormula delete shared formula cell from worksheet shared formula
+// index cache and remove all shared cells formula which refer to the cell which
+// containing the formula.
+func (ws *xlsxWorksheet) deleteSharedFormula(c *xlsxC) {
+	if c.F != nil && c.F.Si != nil && c.F.Ref != "" {
+		si := *c.F.Si
+		ws.formulaSI.Delete(si)
+		for r, row := range ws.SheetData.Row {
+			for c, cell := range row.C {
+				if cell.F != nil && cell.F.Si != nil && *cell.F.Si == si && cell.F.Ref == "" {
+					ws.SheetData.Row[r].C[c].F = nil
+				}
+			}
+		}
+	}
+}
+
 // GetCellHyperLink gets a cell hyperlink based on the given worksheet name and
 // cell reference. If the cell has a hyperlink, it will return 'true' and
 // the link address, otherwise it will return 'false' and an empty link
@@ -1640,18 +1666,27 @@ func isOverlap(rect1, rect2 []int) bool {
 		cellInRange([]int{rect2[2], rect2[3]}, rect1)
 }
 
-// parseSharedFormula generate dynamic part of shared formula for target cell
-// by given column and rows distance and origin shared formula.
-func parseSharedFormula(dCol, dRow int, orig string) string {
+// convertSharedFormula creates a non shared formula from the shared formula
+// counterpart by given cell reference which not containing the formula.
+func (c *xlsxC) convertSharedFormula(cell string) (string, error) {
+	col, row, err := CellNameToCoordinates(cell)
+	if err != nil {
+		return "", err
+	}
+	sharedCol, sharedRow, err := CellNameToCoordinates(c.R)
+	if err != nil {
+		return "", err
+	}
+	dCol, dRow := col-sharedCol, row-sharedRow
 	ps := efp.ExcelParser()
-	tokens := ps.Parse(string(orig))
-	for i := 0; i < len(tokens); i++ {
+	tokens := ps.Parse(c.F.Content)
+	for i := range tokens {
 		token := tokens[i]
 		if token.TType == efp.TokenTypeOperand && token.TSubType == efp.TokenSubTypeRange {
 			tokens[i].TValue = shiftCell(token.TValue, dCol, dRow)
 		}
 	}
-	return ps.Render()
+	return ps.Render(), nil
 }
 
 // getSharedFormula find a cell contains the same formula as another cell,
@@ -1662,21 +1697,23 @@ func parseSharedFormula(dCol, dRow int, orig string) string {
 //
 // Note that this function not validate ref tag to check the cell whether in
 // allow range reference, and always return origin shared formula.
-func getSharedFormula(ws *xlsxWorksheet, si int, cell string) string {
-	for row := 0; row < len(ws.SheetData.Row); row++ {
+func getSharedFormula(ws *xlsxWorksheet, si int, cell string) (string, error) {
+	val, ok := ws.formulaSI.Load(si)
+
+	if ok {
+		return val.(*xlsxC).convertSharedFormula(cell)
+	}
+	for row := range ws.SheetData.Row {
 		r := &ws.SheetData.Row[row]
-		for column := 0; column < len(r.C); column++ {
+		for column := range r.C {
 			c := &r.C[column]
 			if c.F != nil && c.F.Ref != "" && c.F.T == STCellFormulaTypeShared && c.F.Si != nil && *c.F.Si == si {
-				col, row, _ := CellNameToCoordinates(cell)
-				sharedCol, sharedRow, _ := CellNameToCoordinates(c.R)
-				dCol := col - sharedCol
-				dRow := row - sharedRow
-				return parseSharedFormula(dCol, dRow, c.F.Content)
+				ws.formulaSI.Store(si, c)
+				return c.convertSharedFormula(cell)
 			}
 		}
 	}
-	return ""
+	return "", nil
 }
 
 // shiftCell returns the cell shifted according to dCol and dRow taking into

cell_test.go

@@ -563,7 +563,7 @@ func TestGetValueFrom(t *testing.T) {
 	assert.NoError(t, err)
 	value, err := c.getValueFrom(f, sst, false)
 	assert.NoError(t, err)
-	assert.Equal(t, "", value)
+	assert.Empty(t, value)
 
 	c = xlsxC{T: "s", V: " 1 "}
 	value, err = c.getValueFrom(f, &xlsxSST{Count: 1, SI: []xlsxSI{{}, {T: &xlsxT{Val: "s"}}}}, false)
@@ -602,13 +602,17 @@ func TestGetCellFormula(t *testing.T) {
 		formula, err := f.GetCellFormula("Sheet1", "B3")
 		assert.NoError(t, err)
 		assert.Equal(t, expected, formula)
+		// Test get shared formula form cache
+		formula, err = f.GetCellFormula("Sheet1", "B3")
+		assert.NoError(t, err)
+		assert.Equal(t, expected, formula)
 	}
 
 	f.Sheet.Delete("xl/worksheets/sheet1.xml")
 	f.Pkg.Store("xl/worksheets/sheet1.xml", []byte(`<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main"><sheetData><row r="2"><c r="B2"><f t="shared" si="0"></f></c></row></sheetData></worksheet>`))
 	formula, err := f.GetCellFormula("Sheet1", "B2")
 	assert.NoError(t, err)
-	assert.Equal(t, "", formula)
+	assert.Empty(t, formula)
 
 	// Test get array formula with invalid cell range reference
 	f = NewFile()
@@ -628,6 +632,81 @@ func TestGetCellFormula(t *testing.T) {
 	f.Sheet.Delete("xl/worksheets/sheet1.xml")
 	f.Pkg.Store("xl/worksheets/sheet1.xml", MacintoshCyrillicCharset)
 	assert.EqualError(t, f.setArrayFormulaCells(), "XML syntax error on line 1: invalid UTF-8")
+
+	// Test get shared formula after updated refer cell formula, the shared
+	// formula cell reference range covered the previous.
+	f = NewFile()
+	formulaType, ref = STCellFormulaTypeShared, "C2:C6"
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", "=A2+B2", FormulaOpts{Ref: &ref, Type: &formulaType}))
+	formula, err = f.GetCellFormula("Sheet1", "C2")
+	assert.NoError(t, err)
+	assert.Equal(t, "A2+B2", formula)
+	formula, err = f.GetCellFormula("Sheet1", "C6")
+	assert.NoError(t, err)
+	assert.Equal(t, "A6+B6", formula)
+
+	formulaType, ref = STCellFormulaTypeShared, "C2:C8"
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", "=A2*B2", FormulaOpts{Ref: &ref, Type: &formulaType}))
+	formula, err = f.GetCellFormula("Sheet1", "C2")
+	assert.NoError(t, err)
+	assert.Equal(t, "A2*B2", formula)
+	formula, err = f.GetCellFormula("Sheet1", "C8")
+	assert.NoError(t, err)
+	assert.Equal(t, "A8*B8", formula)
+	assert.NoError(t, f.Close())
+
+	// Test get shared formula after updated refer cell formula, the shared
+	// formula cell reference range not over the previous.
+	f = NewFile()
+	formulaType, ref = STCellFormulaTypeShared, "C2:C6"
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", "=A2+B2", FormulaOpts{Ref: &ref, Type: &formulaType}))
+	formula, err = f.GetCellFormula("Sheet1", "C2")
+	assert.NoError(t, err)
+	assert.Equal(t, "A2+B2", formula)
+	formula, err = f.GetCellFormula("Sheet1", "C6")
+	assert.NoError(t, err)
+	assert.Equal(t, "A6+B6", formula)
+
+	formulaType, ref = STCellFormulaTypeShared, "C2:C4"
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", "=A2*B2", FormulaOpts{Ref: &ref, Type: &formulaType}))
+	formula, err = f.GetCellFormula("Sheet1", "C2")
+	assert.NoError(t, err)
+	assert.Equal(t, "A2*B2", formula)
+	formula, err = f.GetCellFormula("Sheet1", "C6")
+	assert.NoError(t, err)
+	assert.Empty(t, formula)
+
+	// Test get shared formula after remove refer cell formula
+	f = NewFile()
+	formulaType, ref = STCellFormulaTypeShared, "C2:C6"
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", "=A2+B2", FormulaOpts{Ref: &ref, Type: &formulaType}))
+
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", ""))
+
+	formula, err = f.GetCellFormula("Sheet1", "C2")
+	assert.NoError(t, err)
+	assert.Empty(t, formula)
+	formula, err = f.GetCellFormula("Sheet1", "C6")
+	assert.NoError(t, err)
+	assert.Empty(t, formula)
+
+	formulaType, ref = STCellFormulaTypeShared, "C2:C8"
+	assert.NoError(t, f.SetCellFormula("Sheet1", "C2", "=A2*B2", FormulaOpts{Ref: &ref, Type: &formulaType}))
+	formula, err = f.GetCellFormula("Sheet1", "C2")
+	assert.NoError(t, err)
+	assert.Equal(t, "A2*B2", formula)
+	formula, err = f.GetCellFormula("Sheet1", "C8")
+	assert.NoError(t, err)
+	assert.Equal(t, "A8*B8", formula)
+	assert.NoError(t, f.Close())
+}
+
+func TestConvertSharedFormula(t *testing.T) {
+	c := xlsxC{R: "A"}
+	_, err := c.convertSharedFormula("A")
+	assert.Equal(t, newCellNameToCoordinatesError("A", newInvalidCellNameError("A")), err)
+	_, err = c.convertSharedFormula("A1")
+	assert.Equal(t, newCellNameToCoordinatesError("A", newInvalidCellNameError("A")), err)
 }
 
 func ExampleFile_SetCellFloat() {

col.go

@@ -782,6 +782,7 @@ func (f *File) RemoveCol(sheet, col string) error {
 	if err != nil {
 		return err
 	}
+	ws.formulaSI.Clear()
 	for rowIdx := range ws.SheetData.Row {
 		rowData := &ws.SheetData.Row[rowIdx]
 		for colIdx := range rowData.C {

datavalidation_test.go

@@ -81,7 +81,7 @@ func TestDataValidation(t *testing.T) {
 		dv.Formula1 = ""
 		assert.NoError(t, dv.SetDropList(listValid),
 			"SetDropList failed for valid input %v", listValid)
-		assert.NotEqual(t, "", dv.Formula1,
+		assert.NotEmpty(t, dv.Formula1,
 			"Formula1 should not be empty for valid input %v", listValid)
 	}
 	assert.Equal(t, `"A<,B>,C"",D	,E',F"`, dv.Formula1)

excelize_test.go

@@ -86,13 +86,13 @@ func TestOpenFile(t *testing.T) {
 
 	f.SetActiveSheet(2)
 	// Test get cell formula with given rows number
-	_, err = f.GetCellFormula("Sheet1", "B19")
+	formula, err := f.GetCellFormula("Sheet1", "B19")
 	assert.NoError(t, err)
+	assert.Equal(t, "SUM(Sheet2!D2,Sheet2!D11)", formula)
 	// Test get cell formula with illegal worksheet name
-	_, err = f.GetCellFormula("Sheet2", "B20")
-	assert.NoError(t, err)
-	_, err = f.GetCellFormula("Sheet1", "B20")
+	formula, err = f.GetCellFormula("Sheet2", "B20")
 	assert.NoError(t, err)
+	assert.Empty(t, formula)
 
 	// Test get cell formula with illegal rows number
 	_, err = f.GetCellFormula("Sheet1", "B")
@@ -1060,7 +1060,7 @@ func TestCopySheetError(t *testing.T) {
 
 func TestGetSheetComments(t *testing.T) {
 	f := NewFile()
-	assert.Equal(t, "", f.getSheetComments("sheet0"))
+	assert.Empty(t, f.getSheetComments("sheet0"))
 }
 
 func TestGetActiveSheetIndex(t *testing.T) {
@@ -1414,7 +1414,7 @@ func TestProtectSheet(t *testing.T) {
 	assert.NoError(t, f.UnprotectSheet(sheetName, "password"))
 	// Test protect worksheet with empty password
 	assert.NoError(t, f.ProtectSheet(sheetName, &SheetProtectionOptions{}))
-	assert.Equal(t, "", ws.SheetProtection.Password)
+	assert.Empty(t, ws.SheetProtection.Password)
 	// Test protect worksheet with password exceeds the limit length
 	assert.EqualError(t, f.ProtectSheet(sheetName, &SheetProtectionOptions{
 		AlgorithmName: "MD4",

go.mod

@@ -5,7 +5,7 @@ go 1.23.0
 require (
 	github.com/richardlehane/mscfb v1.0.4
 	github.com/stretchr/testify v1.10.0
-	github.com/tiendc/go-deepcopy v1.5.1
+	github.com/tiendc/go-deepcopy v1.5.2
 	github.com/xuri/efp v0.0.0-20250227110027-3491fafc2b79
 	github.com/xuri/nfp v0.0.0-20250226145837-86d5fc24b2ba
 	golang.org/x/crypto v0.36.0

go.sum

@@ -9,8 +9,8 @@ github.com/richardlehane/msoleps v1.0.4 h1:WuESlvhX3gH2IHcd8UqyCuFY5yiq/GR/yqaSM
 github.com/richardlehane/msoleps v1.0.4/go.mod h1:BWev5JBpU9Ko2WAgmZEuiz4/u3ZYTKbjLycmwiWUfWg=
 github.com/stretchr/testify v1.10.0 h1:Xv5erBjTwe/5IxqUQTdXv5kgmIvbHo3QQyRwhJsOfJA=
 github.com/stretchr/testify v1.10.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
-github.com/tiendc/go-deepcopy v1.5.1 h1:5ymXIB8ReIywehne6oy3HgywC8LicXYucPBNnj5QQxE=
-github.com/tiendc/go-deepcopy v1.5.1/go.mod h1:toXoeQoUqXOOS/X4sKuiAoSk6elIdqc0pN7MTgOOo2I=
+github.com/tiendc/go-deepcopy v1.5.2 h1:fzTSgAOzxw4MFuDzvyxRDUsdwA7qs7FBTvgXVj28NpQ=
+github.com/tiendc/go-deepcopy v1.5.2/go.mod h1:toXoeQoUqXOOS/X4sKuiAoSk6elIdqc0pN7MTgOOo2I=
 github.com/xuri/efp v0.0.0-20250227110027-3491fafc2b79 h1:78nKszZqigiBRBVcoe/AuPzyLTWW5B+ltBaUX1rlIXA=
 github.com/xuri/efp v0.0.0-20250227110027-3491fafc2b79/go.mod h1:ybY/Jr0T0GTCnYjKqmdwxyxn2BQf2RcQIIvex5QldPI=
 github.com/xuri/nfp v0.0.0-20250226145837-86d5fc24b2ba h1:DhIu6n3qU0joqG9f4IO6a/Gkerd+flXrmlJ+0yX2W8U=

lib_test.go

@@ -95,12 +95,12 @@ func TestColumnNumberToName_OK(t *testing.T) {
 func TestColumnNumberToName_Error(t *testing.T) {
 	out, err := ColumnNumberToName(-1)
 	if assert.Error(t, err) {
-		assert.Equal(t, "", out)
+		assert.Empty(t, out)
 	}
 
 	out, err = ColumnNumberToName(0)
 	if assert.Error(t, err) {
-		assert.Equal(t, "", out)
+		assert.Empty(t, out)
 	}
 
 	_, err = ColumnNumberToName(MaxColumns + 1)

merge_test.go

@@ -35,7 +35,7 @@ func TestMergeCell(t *testing.T) {
 	assert.NoError(t, err)
 	// Merged cell ref is single coordinate
 	value, err = f.GetCellValue("Sheet2", "A6")
-	assert.Equal(t, "", value)
+	assert.Empty(t, value)
 	assert.NoError(t, err)
 	value, err = f.GetCellFormula("Sheet1", "G12")
 	assert.Equal(t, "SUM(Sheet1!B19,Sheet1!C19)", value)
@@ -104,7 +104,7 @@ func TestMergeCellOverlap(t *testing.T) {
 	assert.Len(t, mc, 1)
 	assert.Equal(t, "A1", mc[0].GetStartAxis())
 	assert.Equal(t, "D3", mc[0].GetEndAxis())
-	assert.Equal(t, "", mc[0].GetCellValue())
+	assert.Empty(t, mc[0].GetCellValue())
 	assert.NoError(t, f.Close())
 }
 

rows.go

@@ -632,11 +632,12 @@ func (f *File) RemoveRow(sheet string, row int) error {
 	if err != nil {
 		return err
 	}
+	ws.formulaSI.Clear()
 	if row > len(ws.SheetData.Row) {
 		return f.adjustHelper(sheet, rows, row, -1)
 	}
 	keep := 0
-	for rowIdx := 0; rowIdx < len(ws.SheetData.Row); rowIdx++ {
+	for rowIdx := range ws.SheetData.Row {
 		v := &ws.SheetData.Row[rowIdx]
 		if v.R != row {
 			ws.SheetData.Row[keep] = *v