Add support for get the Microsoft 365 cell images (#1857)

- Update unit tests
2024-03-26 15:19:23 +00:00 · 2024-03-26 15:19:23 +00:00 · 838232fd27
parent 703b73779c
commit 838232fd27
5 changed files with 259 additions and 22 deletions
--- a/excelize.go
+++ b/excelize.go
@ -589,3 +589,41 @@ func (f *File) setContentTypePartProjectExtensions(contentType string) error {
 	}
 	return err
 }
+
+// metadataReader provides a function to get the pointer to the structure
+// after deserialization of xl/metadata.xml.
+func (f *File) metadataReader() (*xlsxMetadata, error) {
+	var mataData xlsxMetadata
+	if err := f.xmlNewDecoder(bytes.NewReader(namespaceStrictToTransitional(f.readXML(defaultXMLMetadata)))).
+		Decode(&mataData); err != nil && err != io.EOF {
+		return &mataData, err
+	}
+	return &mataData, 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) {
+	var richValueRels xlsxRichValueRels
+	if err := f.xmlNewDecoder(bytes.NewReader(namespaceStrictToTransitional(f.readXML(defaultXMLRichDataRichValueRel)))).
+		Decode(&richValueRels); err != nil && err != io.EOF {
+		return &richValueRels, err
+	}
+	return &richValueRels, nil
+}
+
+// getRichDataRichValueRelRelationships provides a function to get drawing
+// relationships from xl/richData/_rels/richValueRel.xml.rels by given
+// relationship ID.
+func (f *File) getRichDataRichValueRelRelationships(rID string) *xlsxRelationship {
+	if rels, _ := f.relsReader(defaultXMLRichDataRichValueRelRels); rels != nil {
+		rels.mu.Lock()
+		defer rels.mu.Unlock()
+		for _, v := range rels.Relationships {
+			if v.ID == rID {
+				return &v
+			}
+		}
+	}
+	return nil
+}
--- a/picture.go
+++ b/picture.go
@ -497,13 +497,13 @@ func (f *File) GetPictureCells(sheet string) ([]string, error) {
 	}
 	f.mu.Unlock()
 	if ws.Drawing == nil {
-		return f.getEmbeddedImageCells(sheet)
+		return f.getImageCells(sheet)
 	}
 	target := f.getSheetRelationshipsTargetByID(sheet, ws.Drawing.RID)
 	drawingXML := strings.TrimPrefix(strings.ReplaceAll(target, "..", "xl"), "/")
 	drawingRelationships := strings.ReplaceAll(
 		strings.ReplaceAll(target, "../drawings", "xl/drawings/_rels"), ".xml", ".xml.rels")
-	embeddedImageCells, err := f.getEmbeddedImageCells(sheet)
+	embeddedImageCells, err := f.getImageCells(sheet)
 	if err != nil {
 		return nil, err
 	}
@ -771,9 +771,9 @@ func (f *File) cellImagesReader() (*decodeCellImages, error) {
 	return f.DecodeCellImages, nil
 }

-// getEmbeddedImageCells returns all the Kingsoft WPS Office embedded image
-// cells reference by given worksheet name.
-func (f *File) getEmbeddedImageCells(sheet string) ([]string, error) {
+// getImageCells returns all the Microsoft 365 cell images and the Kingsoft WPS
+// Office embedded image cells reference by given worksheet name.
+func (f *File) getImageCells(sheet string) ([]string, error) {
 	var (
 		err   error
 		cells []string
@ -791,14 +791,73 @@ func (f *File) getEmbeddedImageCells(sheet string) ([]string, error) {
 				}
 				cells = append(cells, c.R)
 			}
+			r, err := f.getImageCellRel(&c)
+			if err != nil {
+				return cells, err
+			}
+			if r != nil {
+				cells = append(cells, c.R)
+			}
+
 		}
 	}
 	return cells, err
 }

-// getCellImages provides a function to get the Kingsoft WPS Office embedded
-// cell images by given worksheet name and cell reference.
+// getImageCellRel returns the Microsoft 365 cell image relationship.
+func (f *File) getImageCellRel(c *xlsxC) (*xlsxRelationship, error) {
+	var r *xlsxRelationship
+	if c.Vm == nil || c.V != formulaErrorVALUE {
+		return r, nil
+	}
+	metaData, err := f.metadataReader()
+	if err != nil {
+		return r, err
+	}
+	vmd := metaData.ValueMetadata
+	if vmd == nil || int(*c.Vm) > len(vmd.Bk) || len(vmd.Bk[*c.Vm-1].Rc) == 0 {
+		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) {
+		return r, err
+	}
+	rID := richValueRel.Rels[vmd.Bk[*c.Vm-1].Rc[0].V].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
+// the Kingsoft WPS Office embedded cell images by given worksheet name and cell
+// reference.
 func (f *File) getCellImages(sheet, cell string) ([]Picture, error) {
+	pics, err := f.getDispImages(sheet, cell)
+	if err != nil {
+		return pics, err
+	}
+	_, err = f.getCellStringFunc(sheet, cell, func(x *xlsxWorksheet, c *xlsxC) (string, bool, error) {
+		r, err := f.getImageCellRel(c)
+		if err != nil || r == nil {
+			return "", true, err
+		}
+		pic := Picture{Extension: filepath.Ext(r.Target), Format: &GraphicOptions{}}
+		if buffer, _ := f.Pkg.Load(strings.TrimPrefix(strings.ReplaceAll(r.Target, "..", "xl"), "/")); buffer != nil {
+			pic.File = buffer.([]byte)
+			pics = append(pics, pic)
+		}
+		return "", true, nil
+	})
+	return pics, err
+}
+
+// getDispImages provides a function to get the Kingsoft WPS Office embedded
+// cell images by given worksheet name and cell reference.
+func (f *File) getDispImages(sheet, cell string) ([]Picture, error) {
 	formula, err := f.GetCellFormula(sheet, cell)
 	if err != nil {
 		return nil, err
--- a/picture_test.go
+++ b/picture_test.go
@ -448,13 +448,67 @@ func TestGetCellImages(t *testing.T) {
 	_, err := f.getCellImages("Sheet1", "A1")
 	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)}}},
+		}},
+	})
+	pics, err := f.GetPictures("Sheet1", "A1")
+	assert.NoError(t, err)
+	assert.Equal(t, 1, len(pics))
+	cells, err := f.GetPictureCells("Sheet1")
+	assert.NoError(t, err)
+	assert.Equal(t, []string{"A1"}, cells)
+
+	// Test get the Microsoft 365 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
+	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
+	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
+	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
+	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
+	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
+	f.Pkg.Store(defaultXMLMetadata, []byte(`<metadata><valueMetadata/></metadata>`))
+	pics, err = f.GetPictures("Sheet1", "A1")
+	assert.NoError(t, err)
+	assert.Empty(t, pics)
 }

-func TestGetEmbeddedImageCells(t *testing.T) {
+func TestGetImageCells(t *testing.T) {
 	f := NewFile()
 	f.Sheet.Delete("xl/worksheets/sheet1.xml")
 	f.Pkg.Store("xl/worksheets/sheet1.xml", MacintoshCyrillicCharset)
-	_, err := f.getEmbeddedImageCells("Sheet1")
+	_, err := f.getImageCells("Sheet1")
 	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
 	assert.NoError(t, f.Close())
 }
--- a/templates.go
+++ b/templates.go
@ -266,19 +266,22 @@ var supportedChartDataLabelsPosition = map[ChartType][]ChartDataLabelPositionTyp
 }

 const (
-	defaultTempFileSST           = "sharedStrings"
-	defaultXMLPathCalcChain      = "xl/calcChain.xml"
-	defaultXMLPathCellImages     = "xl/cellimages.xml"
-	defaultXMLPathCellImagesRels = "xl/_rels/cellimages.xml.rels"
-	defaultXMLPathContentTypes   = "[Content_Types].xml"
-	defaultXMLPathDocPropsApp    = "docProps/app.xml"
-	defaultXMLPathDocPropsCore   = "docProps/core.xml"
-	defaultXMLPathSharedStrings  = "xl/sharedStrings.xml"
-	defaultXMLPathStyles         = "xl/styles.xml"
-	defaultXMLPathTheme          = "xl/theme/theme1.xml"
-	defaultXMLPathVolatileDeps   = "xl/volatileDependencies.xml"
-	defaultXMLPathWorkbook       = "xl/workbook.xml"
-	defaultXMLPathWorkbookRels   = "xl/_rels/workbook.xml.rels"
+	defaultTempFileSST                 = "sharedStrings"
+	defaultXMLMetadata                 = "xl/metadata.xml"
+	defaultXMLPathCalcChain            = "xl/calcChain.xml"
+	defaultXMLPathCellImages           = "xl/cellimages.xml"
+	defaultXMLPathCellImagesRels       = "xl/_rels/cellimages.xml.rels"
+	defaultXMLPathContentTypes         = "[Content_Types].xml"
+	defaultXMLPathDocPropsApp          = "docProps/app.xml"
+	defaultXMLPathDocPropsCore         = "docProps/core.xml"
+	defaultXMLPathSharedStrings        = "xl/sharedStrings.xml"
+	defaultXMLPathStyles               = "xl/styles.xml"
+	defaultXMLPathTheme                = "xl/theme/theme1.xml"
+	defaultXMLPathVolatileDeps         = "xl/volatileDependencies.xml"
+	defaultXMLPathWorkbook             = "xl/workbook.xml"
+	defaultXMLPathWorkbookRels         = "xl/_rels/workbook.xml.rels"
+	defaultXMLRichDataRichValueRel     = "xl/richData/richValueRel.xml"
+	defaultXMLRichDataRichValueRelRels = "xl/richData/_rels/richValueRel.xml.rels"
 )

 // IndexedColorMapping is the table of default mappings from indexed color value
--- a/xmlMetaData.go
+++ b/xmlMetaData.go
@ -0,0 +1,83 @@
+// Copyright 2016 - 2024 The excelize Authors. All rights reserved. Use of
+// this source code is governed by a BSD-style license that can be found in
+// the LICENSE file.
+//
+// Package excelize providing a set of functions that allow you to write to and
+// read from XLAM / XLSM / XLSX / XLTM / XLTX files. Supports reading and
+// writing spreadsheet documents generated by Microsoft Excel™ 2007 and later.
+// Supports complex components by high compatibility, and provided streaming
+// API for generating or reading data from a worksheet with huge amounts of
+// data. This library needs Go version 1.18 or later.
+
+package excelize
+
+import "encoding/xml"
+
+// xlsxMetadata directly maps the metadata element. A cell in a spreadsheet
+// application can have metadata associated with it. Metadata is just a set of
+// additional properties about the particular cell, and this metadata is stored
+// in the metadata xml part. There are two types of metadata: cell metadata and
+// value metadata. Cell metadata contains information about the cell itself,
+// and this metadata can be carried along with the cell as it moves
+// (insert, shift, copy/paste, merge, unmerge, etc). Value metadata is
+// information about the value of a particular cell. Value metadata properties
+// can be propagated along with the value as it is referenced in formulas.
+type xlsxMetadata struct {
+	XMLName         xml.Name             `xml:"metadata"`
+	MetadataTypes   *xlsxInnerXML        `xml:"metadataTypes"`
+	MetadataStrings *xlsxInnerXML        `xml:"metadataStrings"`
+	MdxMetadata     *xlsxInnerXML        `xml:"mdxMetadata"`
+	FutureMetadata  []xlsxFutureMetadata `xml:"futureMetadata"`
+	CellMetadata    *xlsxMetadataBlocks  `xml:"cellMetadata"`
+	ValueMetadata   *xlsxMetadataBlocks  `xml:"valueMetadata"`
+	ExtLst          *xlsxInnerXML        `xml:"extLst"`
+}
+
+// xlsxFutureMetadata directly maps the futureMetadata element. This element
+// represents future metadata information.
+type xlsxFutureMetadata struct {
+	Bk     []xlsxFutureMetadataBlock `xml:"bk"`
+	ExtLst *xlsxInnerXML             `xml:"extLst"`
+}
+
+// xlsxFutureMetadataBlock directly maps the kb element. This element represents
+// a block of future metadata information. This is a location for storing
+// feature extension information.
+type xlsxFutureMetadataBlock struct {
+	ExtLst *xlsxInnerXML `xml:"extLst"`
+}
+
+// xlsxMetadataBlocks directly maps the metadata element. This element
+// represents cell metadata information. Cell metadata is information metadata
+// about a specific cell, and it stays tied to that cell position.
+type xlsxMetadataBlocks struct {
+	Count int                 `xml:"count,attr,omitempty"`
+	Bk    []xlsxMetadataBlock `xml:"bk"`
+}
+
+// xlsxMetadataBlock directly maps the bk element. This element represents a
+// block of metadata records.
+type xlsxMetadataBlock struct {
+	Rc []xlsxMetadataRecord `xml:"rc"`
+}
+
+// xlsxMetadataRecord directly maps the rc element. This element represents a
+// reference to a specific metadata record.
+type xlsxMetadataRecord struct {
+	T int `xml:"t,attr"`
+	V int `xml:"v,attr"`
+}
+
+// xlsxRichValueRels directly maps the richValueRels element. This element that
+// specifies a list of rich value relationships.
+type xlsxRichValueRels struct {
+	XMLName xml.Name                       `xml:"richValueRels"`
+	Rels    []xlsxRichValueRelRelationship `xml:"rel"`
+	ExtLst  *xlsxInnerXML                  `xml:"extLst"`
+}
+
+// xlsxRichValueRelRelationship directly maps the rel element. This element
+// specifies a relationship for a rich value property.
+type xlsxRichValueRelRelationship struct {
+	ID string `xml:"id,attr"`
+}