Support get rich data value rels index from rich value part (#1866)

Author: yunkeweb

excelize.go

@@ -601,6 +601,17 @@ func (f *File) metadataReader() (*xlsxMetadata, error) {
 	return &mataData, nil
 }
 
+// richValueReader provides a function to get the pointer to the structure after
+// deserialization of xl/richData/richvalue.xml.
+func (f *File) richValueReader() (*xlsxRichValueData, error) {
+	var richValue xlsxRichValueData
+	if err := f.xmlNewDecoder(bytes.NewReader(namespaceStrictToTransitional(f.readXML(defaultXMLRichDataRichValue)))).
+		Decode(&richValue); err != nil && err != io.EOF {
+		return &richValue, err
+	}
+	return &richValue, nil
+}
+
 // richValueRelReader provides a function to get the pointer to the structure
 // after deserialization of xl/richData/richValueRel.xml.
 func (f *File) richValueRelReader() (*xlsxRichValueRels, error) {

picture.go

@@ -450,7 +450,8 @@ func (f *File) addMedia(file []byte, ext string) string {
 // GetPictures provides a function to get picture meta info and raw content
 // embed in spreadsheet by given worksheet and cell name. This function
 // returns the image contents as []byte data types. This function is
-// concurrency safe. For example:
+// concurrency safe. Note that, this function doesn't support getting cell image
+// inserted by IMAGE formula function currently. For example:
 //
 //	f, err := excelize.OpenFile("Book1.xlsx")
 //	if err != nil {
@@ -506,7 +507,8 @@ func (f *File) GetPictures(sheet, cell string) ([]Picture, error) {
 }
 
 // GetPictureCells returns all picture cell references in a worksheet by a
-// specific worksheet name.
+// specific worksheet name. Note that, this function doesn't support getting
+// cell image inserted by IMAGE formula function currently.
 func (f *File) GetPictureCells(sheet string) ([]string, error) {
 	f.mu.Lock()
 	ws, err := f.workSheetReader(sheet)
@@ -790,7 +792,7 @@ func (f *File) cellImagesReader() (*decodeCellImages, error) {
 	return f.DecodeCellImages, nil
 }
 
-// getImageCells returns all the Microsoft 365 cell images and the Kingsoft WPS
+// getImageCells returns all the cell images and the Kingsoft WPS
 // Office embedded image cells reference by given worksheet name.
 func (f *File) getImageCells(sheet string) ([]string, error) {
 	var (
@@ -823,7 +825,29 @@ func (f *File) getImageCells(sheet string) ([]string, error) {
 	return cells, err
 }
 
-// getImageCellRel returns the Microsoft 365 cell image relationship.
+// getImageCellRichValueIdx returns index of the cell image rich value by given
+// cell value meta index and meta blocks.
+func (f *File) getImageCellRichValueIdx(vm uint, blocks *xlsxMetadataBlocks) (int, error) {
+	richValueIdx := blocks.Bk[vm-1].Rc[0].V
+	richValue, err := f.richValueReader()
+	if err != nil {
+		return -1, err
+	}
+	if richValueIdx >= len(richValue.Rv) {
+		return -1, err
+	}
+	rv := richValue.Rv[richValueIdx].V
+	if len(rv) != 2 || rv[1] != "5" {
+		return -1, err
+	}
+	richValueRelIdx, err := strconv.Atoi(rv[0])
+	if err != nil {
+		return -1, err
+	}
+	return richValueRelIdx, err
+}
+
+// getImageCellRel returns the cell image relationship.
 func (f *File) getImageCellRel(c *xlsxC) (*xlsxRelationship, error) {
 	var r *xlsxRelationship
 	if c.Vm == nil || c.V != formulaErrorVALUE {
@@ -837,21 +861,25 @@ func (f *File) getImageCellRel(c *xlsxC) (*xlsxRelationship, error) {
 	if vmd == nil || int(*c.Vm) > len(vmd.Bk) || len(vmd.Bk[*c.Vm-1].Rc) == 0 {
 		return r, err
 	}
+	richValueRelIdx, err := f.getImageCellRichValueIdx(*c.Vm, vmd)
+	if err != nil || richValueRelIdx == -1 {
+		return r, err
+	}
 	richValueRel, err := f.richValueRelReader()
 	if err != nil {
 		return r, err
 	}
-	if vmd.Bk[*c.Vm-1].Rc[0].V >= len(richValueRel.Rels) {
+	if richValueRelIdx >= len(richValueRel.Rels) {
 		return r, err
 	}
-	rID := richValueRel.Rels[vmd.Bk[*c.Vm-1].Rc[0].V].ID
+	rID := richValueRel.Rels[richValueRelIdx].ID
 	if r = f.getRichDataRichValueRelRelationships(rID); r != nil && r.Type != SourceRelationshipImage {
 		return nil, err
 	}
 	return r, err
 }
 
-// getCellImages provides a function to get the Microsoft 365 cell images and
+// getCellImages provides a function to get the cell images and
 // the Kingsoft WPS Office embedded cell images by given worksheet name and cell
 // reference.
 func (f *File) getCellImages(sheet, cell string) ([]Picture, error) {

picture_test.go

@@ -452,17 +452,22 @@ func TestGetCellImages(t *testing.T) {
 	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
 	assert.NoError(t, f.Close())
 
-	// Test get the Microsoft 365 cell images
-	f = NewFile()
-	assert.NoError(t, f.AddPicture("Sheet1", "A1", filepath.Join("test", "images", "excel.png"), nil))
-	f.Pkg.Store(defaultXMLMetadata, []byte(`<metadata><valueMetadata count="1"><bk><rc t="1" v="0"/></bk></valueMetadata></metadata>`))
-	f.Pkg.Store(defaultXMLRichDataRichValueRel, []byte(`<richValueRels><rel r:id="rId1"/></richValueRels>`))
-	f.Pkg.Store(defaultXMLRichDataRichValueRelRels, []byte(fmt.Sprintf(`<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="%s" Target="../media/image1.png"/></Relationships>`, SourceRelationshipImage)))
-	f.Sheet.Store("xl/worksheets/sheet1.xml", &xlsxWorksheet{
-		SheetData: xlsxSheetData{Row: []xlsxRow{
-			{R: 1, C: []xlsxC{{R: "A1", T: "e", V: formulaErrorVALUE, Vm: uintPtr(1)}}},
-		}},
-	})
+	// Test get the cell images
+	prepareWorkbook := func() *File {
+		f := NewFile()
+		assert.NoError(t, f.AddPicture("Sheet1", "A1", filepath.Join("test", "images", "excel.png"), nil))
+		f.Pkg.Store(defaultXMLMetadata, []byte(`<metadata><valueMetadata count="1"><bk><rc t="1" v="0"/></bk></valueMetadata></metadata>`))
+		f.Pkg.Store(defaultXMLRichDataRichValue, []byte(`<rvData count="1"><rv s="0"><v>0</v><v>5</v></rv></rvData>`))
+		f.Pkg.Store(defaultXMLRichDataRichValueRel, []byte(`<richValueRels><rel r:id="rId1"/></richValueRels>`))
+		f.Pkg.Store(defaultXMLRichDataRichValueRelRels, []byte(fmt.Sprintf(`<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="%s" Target="../media/image1.png"/></Relationships>`, SourceRelationshipImage)))
+		f.Sheet.Store("xl/worksheets/sheet1.xml", &xlsxWorksheet{
+			SheetData: xlsxSheetData{Row: []xlsxRow{
+				{R: 1, C: []xlsxC{{R: "A1", T: "e", V: formulaErrorVALUE, Vm: uintPtr(1)}}},
+			}},
+		})
+		return f
+	}
+	f = prepareWorkbook()
 	pics, err := f.GetPictures("Sheet1", "A1")
 	assert.NoError(t, err)
 	assert.Equal(t, 1, len(pics))
@@ -471,41 +476,64 @@ func TestGetCellImages(t *testing.T) {
 	assert.NoError(t, err)
 	assert.Equal(t, []string{"A1"}, cells)
 
-	// Test get the Microsoft 365 cell images without image relationships parts
+	// Test get the cell images without image relationships parts
 	f.Relationships.Delete(defaultXMLRichDataRichValueRelRels)
 	f.Pkg.Store(defaultXMLRichDataRichValueRelRels, []byte(fmt.Sprintf(`<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="%s" Target="../media/image1.png"/></Relationships>`, SourceRelationshipHyperLink)))
 	pics, err = f.GetPictures("Sheet1", "A1")
 	assert.NoError(t, err)
 	assert.Empty(t, pics)
-	// Test get the Microsoft 365 cell images with unsupported charset rich data rich value relationships
+	// Test get the cell images with unsupported charset rich data rich value relationships
 	f.Relationships.Delete(defaultXMLRichDataRichValueRelRels)
 	f.Pkg.Store(defaultXMLRichDataRichValueRelRels, MacintoshCyrillicCharset)
 	pics, err = f.GetPictures("Sheet1", "A1")
 	assert.NoError(t, err)
 	assert.Empty(t, pics)
-	// Test get the Microsoft 365 cell images with unsupported charset rich data rich value
+	// Test get the cell images with unsupported charset rich data rich value
 	f.Pkg.Store(defaultXMLRichDataRichValueRel, MacintoshCyrillicCharset)
 	_, err = f.GetPictures("Sheet1", "A1")
 	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
-	// Test get the Microsoft 365 image cells without block of metadata records
+	// Test get the image cells without block of metadata records
 	cells, err = f.GetPictureCells("Sheet1")
 	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
 	assert.Empty(t, cells)
-	// Test get the Microsoft 365 cell images with rich data rich value relationships
+	// Test get the cell images with rich data rich value relationships
 	f.Pkg.Store(defaultXMLMetadata, []byte(`<metadata><valueMetadata count="1"><bk><rc t="1" v="0"/></bk></valueMetadata></metadata>`))
 	f.Pkg.Store(defaultXMLRichDataRichValueRel, []byte(`<richValueRels/>`))
 	pics, err = f.GetPictures("Sheet1", "A1")
 	assert.NoError(t, err)
 	assert.Empty(t, pics)
-	// Test get the Microsoft 365 cell images with unsupported charset meta data
+	// Test get the cell images with unsupported charset meta data
 	f.Pkg.Store(defaultXMLMetadata, MacintoshCyrillicCharset)
 	_, err = f.GetPictures("Sheet1", "A1")
 	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
-	// Test get the Microsoft 365 cell images without block of metadata records
+	// Test get the cell images without block of metadata records
 	f.Pkg.Store(defaultXMLMetadata, []byte(`<metadata><valueMetadata/></metadata>`))
 	pics, err = f.GetPictures("Sheet1", "A1")
 	assert.NoError(t, err)
 	assert.Empty(t, pics)
+
+	f = prepareWorkbook()
+	// Test get the cell images with empty image cell rich value
+	f.Pkg.Store(defaultXMLRichDataRichValue, []byte(`<rvData count="1"><rv s="0"><v></v><v>5</v></rv></rvData>`))
+	pics, err = f.GetPictures("Sheet1", "A1")
+	assert.EqualError(t, err, "strconv.Atoi: parsing \"\": invalid syntax")
+	assert.Empty(t, pics)
+	// Test get the cell images without image cell rich value
+	f.Pkg.Store(defaultXMLRichDataRichValue, []byte(`<rvData count="1"><rv s="0"><v>0</v><v>1</v></rv></rvData>`))
+	pics, err = f.GetPictures("Sheet1", "A1")
+	assert.NoError(t, err)
+	assert.Empty(t, pics)
+	// Test get the cell images with unsupported charset rich value
+	f.Pkg.Store(defaultXMLRichDataRichValue, MacintoshCyrillicCharset)
+	_, err = f.GetPictures("Sheet1", "A1")
+	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
+
+	f = prepareWorkbook()
+	// Test get the cell images with invalid rich value index
+	f.Pkg.Store(defaultXMLMetadata, []byte(`<metadata><valueMetadata count="1"><bk><rc t="1" v="1"/></bk></valueMetadata></metadata>`))
+	pics, err = f.GetPictures("Sheet1", "A1")
+	assert.NoError(t, err)
+	assert.Empty(t, pics)
 }
 
 func TestGetImageCells(t *testing.T) {

templates.go

@@ -280,6 +280,7 @@ const (
 	defaultXMLPathVolatileDeps         = "xl/volatileDependencies.xml"
 	defaultXMLPathWorkbook             = "xl/workbook.xml"
 	defaultXMLPathWorkbookRels         = "xl/_rels/workbook.xml.rels"
+	defaultXMLRichDataRichValue        = "xl/richData/rdrichvalue.xml"
 	defaultXMLRichDataRichValueRel     = "xl/richData/richValueRel.xml"
 	defaultXMLRichDataRichValueRelRels = "xl/richData/_rels/richValueRel.xml.rels"
 )

xmlMetaData.go

@@ -68,6 +68,23 @@ type xlsxMetadataRecord struct {
 	V int `xml:"v,attr"`
 }
 
+// xlsxRichValueData directly maps the rvData element that specifies rich value
+// data.
+type xlsxRichValueData struct {
+	XMLName xml.Name        `xml:"rvData"`
+	Count   int             `xml:"count,attr,omitempty"`
+	Rv      []xlsxRichValue `xml:"rv"`
+	ExtLst  *xlsxInnerXML   `xml:"extLst"`
+}
+
+// xlsxRichValue directly maps the rv element that specifies rich value data
+// information for a single rich value
+type xlsxRichValue struct {
+	S  int           `xml:"s,attr"`
+	V  []string      `xml:"v"`
+	Fb *xlsxInnerXML `xml:"fb"`
+}
+
 // xlsxRichValueRels directly maps the richValueRels element. This element that
 // specifies a list of rich value relationships.
 type xlsxRichValueRels struct {