This closes #2157, add support parsing rdrichvaluestructure.xml

Author: shcabin

excelize.go

@@ -672,3 +672,14 @@ func (f *File) getRichValueWebImageRelationships(rID string) *xlsxRelationship {
 	}
 	return nil
 }
+
+// richValueStructureReader provides a function to get the pointer to the structure after
+// deserialization of xl/richData/rdrichvaluestructure.xml.
+func (f *File) richValueStructureReader() (*xlsxRichValueStructuresData, error) {
+	var richValueStruct xlsxRichValueStructuresData
+	if err := f.xmlNewDecoder(bytes.NewReader(namespaceStrictToTransitional(f.readXML(defaultXMLRdRichValueStructurePart)))).
+		Decode(&richValueStruct); err != nil && err != io.EOF {
+		return &richValueStruct, err
+	}
+	return &richValueStruct, nil
+}

picture.go

@@ -31,10 +31,24 @@ type PictureInsertType byte
 const (
 	PictureInsertTypePlaceOverCells PictureInsertType = iota
 	PictureInsertTypePlaceInCell
-	PictureInsertTypeIMAGE
+	PictureInsertTypeIMAGE // created directly by formula (ex, =IMAGE), A type of PlaceInCell
 	PictureInsertTypeDISPIMG
 )
 
+// CalcOrigin: indicates how the rich value was created.
+const (
+	CalcOriginNone    = string(iota + '0')
+	CalcOriginFormula // RichValue created directly by formula (ex, =IMAGE)
+	CalcOriginComplexFormula
+	CalcOriginDotNotation
+	CalcOriginReference
+	CalcOriginStandalone           // Standalone RichValue directly stored in a cell without formula dependency (copy/paste as value or LocalImageValue)
+	CalcOriginStandaloneDecorative // Standalone RichValue created from the alt text pane after selecting "decorative"
+	CalcOriginNested
+	CalcOriginJSApi
+	CalcOriginPythonResult
+)
+
 // parseGraphicOptions provides a function to parse the format settings of
 // the picture with default value.
 func parseGraphicOptions(opts *GraphicOptions) *GraphicOptions {
@@ -943,6 +957,33 @@ func (f *File) getImageCellRel(c *xlsxC, pic *Picture) (*xlsxRelationship, error
 		return r, err
 	}
 	rv := richValue.Rv[richValueIdx].V
+	rs := richValue.Rv[richValueIdx].S
+
+	richValueStructure, err := f.richValueStructureReader()
+	if err != nil {
+		return r, err
+	}
+	if rs < len(richValueStructure.S) {
+		pic.InsertType = PictureInsertTypePlaceInCell
+		for idx, key := range richValueStructure.S[rs].K {
+			if idx >= len(rv) {
+				break // invalid key
+			}
+			switch key.N {
+			case "_rvRel:LocalImageIdentifier", "WebImageIdentifier":
+				r, err = f.getRichDataRichValueRel(rv[idx])
+			case "Text":
+				pic.Format.AltText = rv[idx]
+			case "CalcOrigin":
+				// cell image inserted by IMAGE formula function
+				if rv[idx] == CalcOriginFormula {
+					pic.InsertType = PictureInsertTypeIMAGE
+				}
+			}
+		}
+		return r, err
+	}
+	// fallback, there is no valid struct definition
 	if len(rv) == 2 && rv[1] == "5" {
 		pic.InsertType = PictureInsertTypePlaceInCell
 		return f.getRichDataRichValueRel(rv[0])

picture_test.go

@@ -500,6 +500,11 @@ func TestGetCellImages(t *testing.T) {
 		})
 		return f
 	}
+	addStructure := func(f *File) *File {
+		f.Pkg.Store(defaultXMLRdRichValueStructurePart, []byte(`<rvStructures xmlns="http://schemas.microsoft.com/office/spreadsheetml/2017/richdata" count="1">
+    		<s t="_localImage"><k n="_rvRel:LocalImageIdentifier" t="i"/><k n="CalcOrigin" t="i"/><k n="Text" t="s"/></s></rvStructures>`))
+		return f
+	}
 	f = prepareWorkbook()
 	pics, err := f.GetPictures("Sheet1", "A1")
 	assert.NoError(t, err)
@@ -509,6 +514,17 @@ func TestGetCellImages(t *testing.T) {
 	assert.NoError(t, err)
 	assert.Equal(t, []string{"A1"}, cells)
 
+	f = addStructure(prepareWorkbook())
+	// Test get the cell images with rich value struct
+	pics, err = f.GetPictures("Sheet1", "A1")
+	assert.NoError(t, err)
+	assert.Equal(t, 1, len(pics))
+	assert.Equal(t, PictureInsertTypePlaceInCell, pics[0].InsertType)
+	cells, err = f.GetPictureCells("Sheet1")
+	assert.NoError(t, err)
+	assert.Equal(t, []string{"A1"}, cells)
+
+	f = prepareWorkbook()
 	// Test get the cell images without image relationships parts
 	f.Relationships.Delete(defaultXMLRdRichValueRelRels)
 	f.Pkg.Store(defaultXMLRdRichValueRelRels, []byte(fmt.Sprintf(`<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="%s" Target="../media/image1.png"/></Relationships>`, SourceRelationshipHyperLink)))
@@ -605,6 +621,18 @@ func TestGetCellImages(t *testing.T) {
 	f.Pkg.Store(defaultXMLRdRichValuePart, []byte(`<rvData count="1"><rv s="1"><v></v><v>1</v><v>0</v><v>0</v></rv></rvData>`))
 	_, err = f.GetPictures("Sheet1", "A1")
 	assert.EqualError(t, err, "strconv.Atoi: parsing \"\": invalid syntax")
+
+	f = addStructure(prepareWorkbook())
+	// Test get the cell images with no valid definition and fallback old-style
+	f.Pkg.Store(defaultXMLRdRichValueStructurePart, []byte(`<rvStructures xmlns="http://schemas.microsoft.com/office/spreadsheetml/2017/richdata" count="1"></rvStructures>`))
+	pics, err = f.GetPictures("Sheet1", "A1")
+	assert.NoError(t, err)
+	assert.Equal(t, 1, len(pics))
+
+	// Test get the cell images with unsupported charset rich value
+	f.Pkg.Store(defaultXMLRdRichValueStructurePart, MacintoshCyrillicCharset)
+	_, err = f.GetPictures("Sheet1", "A1")
+	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
 }
 
 func TestGetImageCells(t *testing.T) {
@@ -615,3 +643,33 @@ func TestGetImageCells(t *testing.T) {
 	assert.EqualError(t, err, "XML syntax error on line 1: invalid UTF-8")
 	assert.NoError(t, f.Close())
 }
+
+func TestGetCellImagesAndAltText(t *testing.T) {
+	f, err := OpenFile(filepath.Join("test", "CellImage.xlsx"))
+	assert.NoError(t, err)
+	type imageType struct {
+		cell       string
+		insertType PictureInsertType
+		altText    string
+	}
+	want := []imageType{
+		{"B1", PictureInsertTypePlaceInCell, "Smiling alarm clock face"},
+		{"B2", PictureInsertTypePlaceInCell, ""},
+		{"B3", PictureInsertTypePlaceInCell, "Bullseye outline"},
+		{"B4", PictureInsertTypeIMAGE, ""},
+		{"B5", PictureInsertTypeIMAGE, "other alt_text"},
+
+		{"D1", PictureInsertTypePlaceInCell, "Smiling alarm clock face"},
+		{"D2", PictureInsertTypePlaceInCell, ""},
+		{"D3", PictureInsertTypePlaceInCell, "Bullseye outline"},
+		{"D4", PictureInsertTypePlaceInCell, ""},
+		{"D5", PictureInsertTypePlaceInCell, "other alt_text"},
+	}
+	for _, c := range want {
+		p, err := f.GetPictures("Sheet1", c.cell)
+		assert.NoError(t, err)
+		assert.Equal(t, 1, len(p))
+		assert.Equal(t, c.insertType, p[0].InsertType, c.cell)
+	}
+	assert.NoError(t, f.Close())
+}

templates.go

@@ -309,6 +309,7 @@ const (
 	defaultXMLRdRichValueRelRels          = "xl/richData/_rels/richValueRel.xml.rels"
 	defaultXMLRdRichValueWebImagePart     = "xl/richData/rdRichValueWebImage.xml"
 	defaultXMLRdRichValueWebImagePartRels = "xl/richData/_rels/rdRichValueWebImage.xml.rels"
+	defaultXMLRdRichValueStructurePart    = "xl/richData/rdrichvaluestructure.xml"
 )
 
 // IndexedColorMapping is the table of default mappings from indexed color value

test/CellImage.xlsx

@@ -115,3 +115,25 @@ type xlsxWebImageSupportingRichData struct {
 	MoreImagesAddress xlsxExternalReference `xml:"moreImagesAddress"`
 	Blip              xlsxExternalReference `xml:"blip"`
 }
+
+// xlsxRichValueStructuresData directly maps the rvStructures element that specifies
+// rich value data.
+type xlsxRichValueStructuresData struct {
+	XMLName xml.Name                 `xml:"rvStructures"`
+	Count   string                   `xml:"count,attr"`
+	S       []xlsxRichValueStructure `xml:"s"`
+}
+
+// xlsxRichValueStructure directly maps the RichValueStructure element that specifies rich
+// value data.
+type xlsxRichValueStructure struct {
+	T string             `xml:"t,attr"`
+	K []xlsxRichValueKey `xml:"k"`
+}
+
+// xlsxRichValueKey directly maps the rich value key element that specifies rich value
+// data.
+type xlsxRichValueKey struct {
+	N string `xml:"n,attr"`
+	T string `xml:"t,attr"`
+}