Add purl.oclc.org namespace support (unidoc#265)

* Regenerate code to handle purl.oclc.org docs correctly
* Make purl.oclc.org docs be opened just as the openxml ones
* Modify opening routine to change strict docs to non-strict
- `CustomXMLType` and `CustomXMLTypeStrict` were added to properly replace
these in strict docs
- `Decode` of `zippkg` was modified to replace the namespace in
relationships
- `TestOpenStrict` was modified to open the strict file, save it as
non-strict, reopen non-strict and validate its correctness

.gitignore

@@ -14,3 +14,5 @@ coverage*
 **/.coverprofile
 gover.coverprofile
 spreadsheet/formula/y.output
+
+.idea

document/document.go

@@ -683,7 +683,7 @@ func (d *Document) onNewRelationship(decMap *zippkg.DecodeMap, target, typ strin
 	dt := gooxml.DocTypeDocument
 
 	switch typ {
-	case gooxml.OfficeDocumentType:
+	case gooxml.OfficeDocumentType, gooxml.OfficeDocumentTypeStrict:
 		d.x = wml.NewDocument()
 		decMap.AddTarget(target, d.x, typ, 0)
 		// look for the document relationships file as well
@@ -694,11 +694,11 @@ func (d *Document) onNewRelationship(decMap *zippkg.DecodeMap, target, typ strin
 		decMap.AddTarget(target, d.CoreProperties.X(), typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.ExtendedPropertiesType:
+	case gooxml.ExtendedPropertiesType, gooxml.ExtendedPropertiesTypeStrict:
 		decMap.AddTarget(target, d.AppProperties.X(), typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.ThumbnailType:
+	case gooxml.ThumbnailType, gooxml.ThumbnailTypeStrict:
 		// read our thumbnail
 		for i, f := range files {
 			if f == nil {
@@ -718,21 +718,21 @@ func (d *Document) onNewRelationship(decMap *zippkg.DecodeMap, target, typ strin
 			}
 		}
 
-	case gooxml.SettingsType:
+	case gooxml.SettingsType, gooxml.SettingsTypeStrict:
 		decMap.AddTarget(target, d.Settings.X(), typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.NumberingType:
+	case gooxml.NumberingType, gooxml.NumberingTypeStrict:
 		d.Numbering = NewNumbering()
 		decMap.AddTarget(target, d.Numbering.X(), typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.StylesType:
+	case gooxml.StylesType, gooxml.StylesTypeStrict:
 		d.Styles.Clear()
 		decMap.AddTarget(target, d.Styles.X(), typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.HeaderType:
+	case gooxml.HeaderType, gooxml.HeaderTypeStrict:
 		hdr := wml.NewHdr()
 		decMap.AddTarget(target, hdr, typ, uint32(len(d.headers)))
 		d.headers = append(d.headers, hdr)
@@ -743,7 +743,7 @@ func (d *Document) onNewRelationship(decMap *zippkg.DecodeMap, target, typ strin
 		decMap.AddTarget(zippkg.RelationsPathFor(target), hdrRel.X(), typ, 0)
 		d.hdrRels = append(d.hdrRels, hdrRel)
 
-	case gooxml.FooterType:
+	case gooxml.FooterType, gooxml.FooterTypeStrict:
 		ftr := wml.NewFtr()
 		decMap.AddTarget(target, ftr, typ, uint32(len(d.footers)))
 		d.footers = append(d.footers, ftr)
@@ -754,33 +754,33 @@ func (d *Document) onNewRelationship(decMap *zippkg.DecodeMap, target, typ strin
 		decMap.AddTarget(zippkg.RelationsPathFor(target), ftrRel.X(), typ, 0)
 		d.ftrRels = append(d.ftrRels, ftrRel)
 
-	case gooxml.ThemeType:
+	case gooxml.ThemeType, gooxml.ThemeTypeStrict:
 		thm := dml.NewTheme()
 		decMap.AddTarget(target, thm, typ, uint32(len(d.themes)))
 		d.themes = append(d.themes, thm)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, len(d.themes))
 
-	case gooxml.WebSettingsType:
+	case gooxml.WebSettingsType, gooxml.WebSettingsTypeStrict:
 		d.webSettings = wml.NewWebSettings()
 		decMap.AddTarget(target, d.webSettings, typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.FontTableType:
+	case gooxml.FontTableType, gooxml.FontTableTypeStrict:
 		d.fontTable = wml.NewFonts()
 		decMap.AddTarget(target, d.fontTable, typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.EndNotesType:
+	case gooxml.EndNotesType, gooxml.EndNotesTypeStrict:
 		d.endNotes = wml.NewEndnotes()
 		decMap.AddTarget(target, d.endNotes, typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.FootNotesType:
+	case gooxml.FootNotesType, gooxml.FootNotesTypeStrict:
 		d.footNotes = wml.NewFootnotes()
 		decMap.AddTarget(target, d.footNotes, typ, 0)
 		rel.TargetAttr = gooxml.RelativeFilename(dt, src.Typ, typ, 0)
 
-	case gooxml.ImageType:
+	case gooxml.ImageType, gooxml.ImageTypeStrict:
 		var iref common.ImageRef
 		for i, f := range files {
 			if f == nil {

document/document_test.go

@@ -9,6 +9,8 @@ package document_test
 
 import (
 	"bytes"
+	"io/ioutil"
+	"os"
 	"testing"
 
 	"baliance.com/gooxml/common"
@@ -44,6 +46,35 @@ func TestOpen(t *testing.T) {
 	testhelper.CompareZip(t, "simple-1.docx", got.Bytes(), true)
 }
 
+func TestOpenStrict(t *testing.T) {
+	strict, err := document.Open("testdata/strict.docx")
+	if err != nil {
+		t.Errorf("error opening document: %s", err)
+	}
+
+	gotStrict := bytes.Buffer{}
+	if err := strict.Validate(); err != nil {
+		t.Errorf("created an invalid document: %s", err)
+	}
+	strict.Save(&gotStrict)
+	ioutil.WriteFile("testdata/non-strict.docx", gotStrict.Bytes(), 0644)
+
+	// run test assuming that the doc is a valid non-strict doc
+	nonStrict, err := document.Open("testdata/non-strict.docx")
+	if err != nil {
+		t.Errorf("error opening document: %s", err)
+	}
+
+	gotNonStrict := bytes.Buffer{}
+	if err := nonStrict.Validate(); err != nil {
+		t.Errorf("created an invalid document: %s", err)
+	}
+	nonStrict.Save(&gotNonStrict)
+	testhelper.CompareZip(t, "non-strict.docx", gotNonStrict.Bytes(), true)
+
+	os.Remove("testdata/non-strict.docx")
+}
+
 func TestOpenHeaderFooter(t *testing.T) {
 	wb, err := document.Open("testdata/header-footer-multiple.docx")
 	if err != nil {

filenames.go

@@ -70,12 +70,12 @@ func AbsoluteFilename(dt DocType, typ string, index int) string {
 	switch typ {
 	case CorePropertiesType:
 		return "docProps/core.xml"
-	case ExtendedPropertiesType:
+	case ExtendedPropertiesType, ExtendedPropertiesTypeStrict:
 		return "docProps/app.xml"
-	case ThumbnailType:
+	case ThumbnailType, ThumbnailTypeStrict:
 		return "docProps/thumbnail.jpeg"
 
-	case OfficeDocumentType:
+	case OfficeDocumentType, OfficeDocumentTypeStrict:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return "xl/workbook.xml"
@@ -87,7 +87,7 @@ func AbsoluteFilename(dt DocType, typ string, index int) string {
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 
-	case ThemeType, ThemeContentType:
+	case ThemeType, ThemeTypeStrict, ThemeContentType:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return fmt.Sprintf("xl/theme/theme%d.xml", index)
@@ -99,7 +99,7 @@ func AbsoluteFilename(dt DocType, typ string, index int) string {
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 
-	case StylesType:
+	case StylesType, StylesTypeStrict:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return "xl/styles.xml"
@@ -111,41 +111,41 @@ func AbsoluteFilename(dt DocType, typ string, index int) string {
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 
-	case ChartType, ChartContentType:
+	case ChartType, ChartTypeStrict, ChartContentType:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return fmt.Sprintf("xl/charts/chart%d.xml", index)
 		default:
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
-	case TableType, TableContentType:
+	case TableType, TableTypeStrict, TableContentType:
 		return fmt.Sprintf("xl/tables/table%d.xml", index)
 
-	case DrawingType, DrawingContentType:
+	case DrawingType, DrawingTypeStrict, DrawingContentType:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return fmt.Sprintf("xl/drawings/drawing%d.xml", index)
 		default:
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 
-	case CommentsType, CommentsContentType:
+	case CommentsType, CommentsTypeStrict, CommentsContentType:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return fmt.Sprintf("xl/comments%d.xml", index)
 		default:
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 
-	case VMLDrawingType, VMLDrawingContentType:
+	case VMLDrawingType, VMLDrawingTypeStrict, VMLDrawingContentType:
 		switch dt {
 		case DocTypeSpreadsheet:
 			return fmt.Sprintf("xl/drawings/vmlDrawing%d.vml", index)
 		default:
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 
-	case ImageType:
+	case ImageType, ImageTypeStrict:
 		switch dt {
 		case DocTypeDocument:
 			return fmt.Sprintf("word/media/image%d.png", index)
@@ -157,31 +157,31 @@ func AbsoluteFilename(dt DocType, typ string, index int) string {
 			Log("unsupported type %s pair and %v", typ, dt)
 		}
 	// SML
-	case WorksheetType, WorksheetContentType:
+	case WorksheetType, WorksheetTypeStrict, WorksheetContentType:
 		return fmt.Sprintf("xl/worksheets/sheet%d.xml", index)
-	case SharedStingsType, SharedStringsContentType:
+	case SharedStingsType, SharedStingsTypeStrict, SharedStringsContentType:
 		return "xl/sharedStrings.xml"
 
 	// WML
-	case FontTableType:
+	case FontTableType, FontTableTypeStrict:
 		return "word/fontTable.xml"
-	case EndNotesType:
+	case EndNotesType, EndNotesTypeStrict:
 		return "word/endnotes.xml"
-	case FootNotesType:
+	case FootNotesType, FootNotesTypeStrict:
 		return "word/footnotes.xml"
-	case NumberingType:
+	case NumberingType, NumberingTypeStrict:
 		return "word/numbering.xml"
-	case WebSettingsType:
+	case WebSettingsType, WebSettingsTypeStrict:
 		return "word/webSettings.xml"
-	case SettingsType:
+	case SettingsType, SettingsTypeStrict:
 		return "word/settings.xml"
-	case HeaderType:
+	case HeaderType, HeaderTypeStrict:
 		return fmt.Sprintf("word/header%d.xml", index)
-	case FooterType:
+	case FooterType, FooterTypeStrict:
 		return fmt.Sprintf("word/footer%d.xml", index)
 
 	// PML
-	case SlideType:
+	case SlideType, SlideTypeStrict:
 		return fmt.Sprintf("ppt/slides/slide%d.xml", index)
 	case SlideLayoutType:
 		return fmt.Sprintf("ppt/slideLayouts/slideLayout%d.xml", index)