feat(参照标准): 修复资源依赖，优化压缩分析

ofdgo_compression.go

@@ -180,8 +180,9 @@ func (e *Editor) outputSnapshot(ctx context.Context, options WriteOptions) (*Edi
 func (r *Reader) compressionImageSafety(ctx context.Context, images []ImageInfo) (map[string]bool, bool, error) {
 	names := make(map[string]bool)
 	safe := true
-	if len(r.OFD.DocBody) != 0 {
+	structured := len(r.OFD.DocBody) != 0
-		dependencies, err := r.documentFiles(ctx, r.DocumentIndex())
+	if structured {
+		dependencies, structures, err := r.documentDependencies(ctx, r.DocumentIndex())
 		if err != nil {
 			if ctx.Err() != nil {
 				return nil, false, ctx.Err()
@@ -193,8 +194,8 @@ func (r *Reader) compressionImageSafety(ctx context.Context, images []ImageInfo)
 				safe = false
 				break
 			}
-			names[name] = true
 		}
+		names = structures
 	} else {
 		for name := range r.fileIndex {
 			names[name] = true
@@ -222,8 +223,9 @@ func (r *Reader) compressionImageSafety(ctx context.Context, images []ImageInfo)
 		buffer := bufio.NewReader(imageInput{ReadCloser: input, context: ctx})
 		header, _ := buffer.Peek(512)
 		header = bytes.TrimSpace(bytes.TrimPrefix(header, []byte{0xef, 0xbb, 0xbf}))
-		if len(header) > 0 && header[0] == '<' {
+		if structured || len(header) > 0 && header[0] == '<' {
 			decoder := xml.NewDecoder(buffer)
+			var stack []xml.Name
 			for {
 				token, err := decoder.Token()
 				if err == io.EOF {
@@ -233,16 +235,27 @@ func (r *Reader) compressionImageSafety(ctx context.Context, images []ImageInfo)
 					safe = false
 					break
 				}
-				if element, ok := token.(xml.StartElement); ok {
+				switch element := token.(type) {
-					if classifyOFDNamespace(element.Name.Space) == ofdXMLUnknown {
+				case xml.StartElement:
+					if len(stack) != 0 && classifyOFDNamespace(element.Name.Space) == ofdXMLUnknown && element.Name.Space != stack[len(stack)-1].Space {
 						safe = false
 						break
 					}
+					if len(stack) != 0 && stack[len(stack)-1].Local == "Extension" && element.Name.Local == "Data" {
+						if err := decoder.Skip(); err != nil {
+							safe = false
+						}
+						break
+					}
+					stack = append(stack, element.Name)
 					if element.Name.Local == "Pattern" {
 						geometry = false
 					}
 					resource, masked := "", false
 					for _, attr := range element.Attr {
+						if attr.Name.Space != "" {
+							continue
+						}
 						if attr.Name.Local == "Visible" && (attr.Value == "false" || attr.Value == "0") {
 							geometry = false
 						}
@@ -257,6 +270,11 @@ func (r *Reader) compressionImageSafety(ctx context.Context, images []ImageInfo)
 					if masked {
 						masks[resource] = true
 					}
+				case xml.EndElement:
+					stack = stack[:len(stack)-1]
+				}
+				if !safe {
+					break
 				}
 			}
 		}

ofdgo_reader_dependencies.go

@@ -15,8 +15,6 @@
 package ofdgo
 
 import (
-	"bufio"
-	"bytes"
 	"context"
 	"encoding/xml"
 	"fmt"
@@ -24,124 +22,264 @@ import (
 	"strings"
 )
 
-// documentFiles 按标准路径遍历独立文档依赖，签名摘要引用不作为内容依赖
+// 包内引用分为无引用、原始数据及OFD结构文件
+const (
+	packageReferenceNone = iota
+	packageReferenceData
+	packageReferenceXML
+)
+
+// packageReference 根据节点位置识别文件引用，不将签名摘要引用或扩展数据解释为结构依赖
+// 入参: root 文件根节点, parent 父节点, name 当前节点, attribute 属性名称，空值表示节点文本
+// 返回: int 引用类别
+func packageReference(root, parent, name, attribute string) int {
+	if attribute != "" {
+		switch attribute {
+		case "BaseLoc":
+			switch {
+			case root == "OFD" && parent == "Versions" && name == "Version",
+				root == "Document" && parent == "Pages" && name == "Page",
+				root == "Document" && parent == "CommonData" && name == "TemplatePage",
+				root == "Signatures" && parent == "Signatures" && name == "Signature",
+				root == "Res" && parent == "CompositeGraphicUnits" && name == "CompositeGraphicUnit":
+				return packageReferenceXML
+			}
+		case "FileLoc":
+			if root == "Annotations" && name == "PageAnnot" {
+				return packageReferenceXML
+			}
+		case "Link":
+			if root == "Res" && name == "DrawParam" {
+				return packageReferenceXML
+			}
+		case "Profile":
+			if root == "Res" && name == "ColorSpace" {
+				return packageReferenceData
+			}
+		}
+		return packageReferenceNone
+	}
+	switch name {
+	case "DocRoot":
+		if root == "OFD" && parent == "DocBody" || root == "DocVersion" && parent == "DocVersion" {
+			return packageReferenceXML
+		}
+	case "Signatures":
+		if root == "OFD" && parent == "DocBody" {
+			return packageReferenceXML
+		}
+	case "Cover":
+		if root == "OFD" && parent == "DocInfo" {
+			return packageReferenceData
+		}
+	case "PublicRes", "DocumentRes":
+		if root == "Document" && parent == "CommonData" {
+			return packageReferenceXML
+		}
+	case "PageRes":
+		if root == "Page" && parent == "Page" {
+			return packageReferenceXML
+		}
+	case "Annotations", "Attachments", "CustomTags", "Extensions":
+		if root == "Document" && parent == "Document" {
+			return packageReferenceXML
+		}
+	case "FileLoc":
+		switch {
+		case root == "Annotations" && parent == "Page":
+			return packageReferenceXML
+		case root == "Attachments" && parent == "Attachment", root == "CustomTags" && parent == "CustomTag":
+			return packageReferenceData
+		}
+	case "File":
+		if root == "DocVersion" && parent == "FileList" {
+			return packageReferenceData
+		}
+	case "FontFile":
+		if root == "Res" && parent == "Font" {
+			return packageReferenceData
+		}
+	case "MediaFile":
+		if root == "Res" && parent == "MultiMedia" {
+			return packageReferenceData
+		}
+	case "Profile":
+		if root == "Res" && parent == "ColorSpace" {
+			return packageReferenceData
+		}
+	case "SignedValue":
+		if root == "Signature" && parent == "Signature" {
+			return packageReferenceData
+		}
+	case "BaseLoc":
+		if root == "Signature" && parent == "Seal" {
+			return packageReferenceData
+		}
+	case "SchemaLoc":
+		if root == "CustomTags" && parent == "CustomTag" {
+			return packageReferenceData
+		}
+	case "ExtendData":
+		if parent == "Extension" {
+			return packageReferenceData
+		}
+	}
+	return packageReferenceNone
+}
+
+// documentFiles 遍历独立文档及版本依赖，只解析结构文件，不读取附件等原始数据
 // 入参: ctx 取消上下文, index 文档索引
 // 返回: map[string]bool 包内文件路径, error 读取或取消错误
 func (r *Reader) documentFiles(ctx context.Context, index int) (map[string]bool, error) {
+	files, _, err := r.documentDependencies(ctx, index)
+	return files, err
+}
+
+// documentDependencies 区分文档全部依赖与需要解析的结构文件，避免重复读取原始数据
+// 入参: ctx 取消上下文, index 文档索引
+// 返回: map[string]bool 全部依赖, map[string]bool 结构文件, error 读取或取消错误
+func (r *Reader) documentDependencies(ctx context.Context, index int) (map[string]bool, map[string]bool, error) {
 	if err := ctx.Err(); err != nil {
-		return nil, err
+		return nil, nil, err
 	}
 	if index < 0 || index >= r.DocumentCount() {
-		return nil, fmt.Errorf("document index out of range: %d", index)
+		return nil, nil, fmt.Errorf("document index out of range: %d", index)
+	}
+	var queue []string
+	seen := make(map[string]bool)
+	queued := make(map[string]bool)
+	add := func(name string, kind int) {
+		seen[name] = true
+		if kind == packageReferenceXML && !queued[name] {
+			queued[name] = true
+			queue = append(queue, name)
+		}
 	}
-	queue := []string{cleanPackagePath(r.OFD.DocBody[index].DocRoot)}
 	data, err := r.readFile("OFD.xml")
 	if err != nil {
-		return nil, err
+		return nil, nil, err
 	}
 	root, err := parseEditorXML(data)
 	if err != nil {
-		return nil, err
+		return nil, nil, err
 	}
 	body := root.childAt("DocBody", index)
 	if body == nil {
-		return nil, fmt.Errorf("document body not found: %d", index)
+		return nil, nil, fmt.Errorf("document body not found: %d", index)
 	}
 	var collect func(*editorXML)
 	collect = func(node *editorXML) {
-		if node.name.Local == "Cover" || node.name.Local == "Signatures" || node.name.Local == "DocRoot" {
+		if !node.matchesOFD(node.name.Local) {
+			return
+		}
+		parent := node.parent.name.Local
+		if kind := packageReference("OFD", parent, node.name.Local, ""); kind != packageReferenceNone {
 			if value := strings.TrimSpace(editorImportText(data, node)); value != "" {
-				queue = append(queue, resolveResourcePath("OFD.xml", "", value))
+				add(resolveResourcePath("OFD.xml", "", value), kind)
 			}
 		}
-		if node.name.Local == "Version" && node.attr("BaseLoc") != "" {
+		for _, attr := range node.attrs {
-			queue = append(queue, resolveResourcePath("OFD.xml", "", node.attr("BaseLoc")))
+			if attr.Name.Space != "" || strings.TrimSpace(attr.Value) == "" {
+				continue
+			}
+			if kind := packageReference("OFD", parent, node.name.Local, attr.Name.Local); kind != packageReferenceNone {
+				add(resolveResourcePath("OFD.xml", "", attr.Value), kind)
+			}
 		}
 		for _, child := range node.children {
 			collect(child)
 		}
 	}
 	collect(body)
-	seen := make(map[string]bool)
 	for len(queue) > 0 {
 		if err := ctx.Err(); err != nil {
-			return nil, err
+			return nil, nil, err
 		}
 		actual := queue[0]
 		queue = queue[1:]
-		if seen[actual] {
-			continue
-		}
-		seen[actual] = true
 		if !r.fileNames[actual] {
 			continue
 		}
-		input, err := r.openFile(actual)
+		if err := r.documentFileReferences(ctx, actual, add); err != nil {
+			return nil, nil, err
+		}
+	}
+	return seen, queued, ctx.Err()
+}
+
+// documentFileReferences 流式扫描结构文件，支持同一作用域内的未知命名空间
+// 入参: ctx 取消上下文, name 文件路径, add 接收引用路径及类别
+// 返回: error 读取或结构错误
+func (r *Reader) documentFileReferences(ctx context.Context, name string, add func(string, int)) error {
+	input, err := r.openFile(name)
+	if err != nil {
+		return err
+	}
+	defer input.Close()
+	decoder := xml.NewDecoder(imageInput{ReadCloser: input, context: ctx})
+	var stack []xml.Name
+	root, base := "", ""
+	var text strings.Builder
+	for {
+		if err := ctx.Err(); err != nil {
+			return err
+		}
+		token, err := decoder.Token()
+		if err == io.EOF {
+			if root == "" {
+				return fmt.Errorf("empty document dependency: %s", name)
+			}
+			return nil
+		}
 		if err != nil {
-			return nil, err
+			return fmt.Errorf("document dependency %s: %w", name, err)
+		}
+		switch node := token.(type) {
+		case xml.StartElement:
+			parent := ""
+			if len(stack) == 0 {
+				if root != "" {
+					return fmt.Errorf("multiple XML roots in %s", name)
+				}
+				root = node.Name.Local
+			} else {
+				owner := stack[len(stack)-1]
+				parent = owner.Local
+				if classifyOFDNamespace(node.Name.Space) == ofdXMLUnknown && node.Name.Space != owner.Space {
+					return fmt.Errorf("cannot determine document dependencies in %s", name)
+				}
+			}
+			if parent == "Extension" && node.Name.Local == "Data" {
+				if err := decoder.Skip(); err != nil {
+					return err
+				}
+				continue
+			}
+			for _, attr := range node.Attr {
+				if attr.Name.Space != "" || strings.TrimSpace(attr.Value) == "" {
+					continue
+				}
+				if len(stack) == 0 && root == "Res" && attr.Name.Local == "BaseLoc" {
+					base = attr.Value
+				} else if kind := packageReference(root, parent, node.Name.Local, attr.Name.Local); kind != packageReferenceNone {
+					add(resolveResourcePath(name, base, attr.Value), kind)
+				}
+			}
+			stack = append(stack, node.Name)
+			text.Reset()
+		case xml.CharData:
+			if len(stack) >= 2 && packageReference(root, stack[len(stack)-2].Local, stack[len(stack)-1].Local, "") != packageReferenceNone {
+				text.Write(node)
+			}
+		case xml.EndElement:
+			if len(stack) >= 2 {
+				kind := packageReference(root, stack[len(stack)-2].Local, node.Name.Local, "")
+				if value := strings.TrimSpace(text.String()); kind != packageReferenceNone && value != "" {
+					add(resolveResourcePath(name, base, value), kind)
+				}
+			}
+			stack = stack[:len(stack)-1]
+			text.Reset()
 		}
-		buffer := bufio.NewReader(imageInput{ReadCloser: input, context: ctx})
-		if prefix, _ := buffer.Peek(3); bytes.Equal(prefix, []byte{0xef, 0xbb, 0xbf}) {
-			buffer.Discard(3)
-		}
-		header, _ := buffer.Peek(128)
-		for len(header) > 0 && len(bytes.TrimSpace(header)) == 0 {
-			buffer.Discard(len(header))
-			header, _ = buffer.Peek(128)
-		}
-		if !bytes.HasPrefix(bytes.TrimSpace(header), []byte("<")) {
-			input.Close()
-			continue
-		}
-		decoder := xml.NewDecoder(buffer)
-		base, resource, depth := "", false, 0
-		var text strings.Builder
-		for {
-			token, err := decoder.Token()
-			if err == io.EOF {
-				break
-			}
-			if err != nil {
-				input.Close()
-				return nil, err
-			}
-			switch node := token.(type) {
-			case xml.StartElement:
-				if depth == 0 && classifyOFDNamespace(node.Name.Space) == ofdXMLUnknown {
-					input.Close()
-					goto nextFile
-				}
-				if classifyOFDNamespace(node.Name.Space) == ofdXMLUnknown {
-					input.Close()
-					return nil, fmt.Errorf("cannot determine document dependencies in %s", actual)
-				}
-				if depth == 0 {
-					resource = node.Name.Local == "Res"
-				}
-				for _, attr := range node.Attr {
-					if resource && depth == 0 && attr.Name.Local == "BaseLoc" {
-						base = attr.Value
-					} else if attr.Name.Local == "BaseLoc" || attr.Name.Local == "FileLoc" || node.Name.Local == "DrawParam" && attr.Name.Local == "Link" || node.Name.Local == "ColorSpace" && attr.Name.Local == "Profile" {
-						queue = append(queue, resolveResourcePath(actual, base, attr.Value))
-					}
-				}
-				depth++
-				text.Reset()
-			case xml.CharData:
-				text.Write(node)
-			case xml.EndElement:
-				switch node.Name.Local {
-				case "DocRoot", "Signatures", "Cover", "PublicRes", "DocumentRes", "PageRes", "Annotations", "Attachments", "FileLoc", "Profile", "SignedValue", "SchemaLoc", "ExtendData", "File", "FontFile", "MediaFile", "CustomTags", "Extensions":
-					if value := strings.TrimSpace(text.String()); value != "" {
-						queue = append(queue, resolveResourcePath(actual, base, value))
-					}
-				}
-				depth--
-				text.Reset()
-			}
-		}
-		input.Close()
-	nextFile:
 	}
-	return seen, ctx.Err()
 }

ofdgo_version_edit.go

@@ -83,34 +83,34 @@ func (r *Reader) versionEditGraph(ctx context.Context, entry string, parts map[s
 			if node != root && !node.matchesOFD(node.name.Local) {
 				return
 			}
-			add := func(attribute, value string, structure bool) {
+			parent := ""
+			if node.parent != nil {
+				parent = node.parent.name.Local
+			}
+			if parent == "Extension" && node.name.Local == "Data" {
+				return
+			}
+			add := func(attribute, value string, kind int) {
 				if strings.TrimSpace(value) == "" {
 					return
 				}
 				target := resolveResourcePath(name, base, value)
 				file.links = append(file.links, versionEditLink{node: node, attribute: attribute, target: target})
 				graph.used[target] = true
-				if structure {
+				if kind == packageReferenceXML {
 					queue = append(queue, target)
 				}
 			}
-			switch node.name.Local {
+			if kind := packageReference(root.name.Local, parent, node.name.Local, ""); kind != packageReferenceNone {
-			case "PublicRes", "DocumentRes", "PageRes", "Annotations", "Attachments", "Signatures", "CustomTags", "Extensions":
+				add("", editorImportText(data, node), kind)
-				if len(node.children) == 0 {
+			}
-					add("", editorImportText(data, node), true)
+			for _, attr := range node.attrs {
+				if attr.Name.Space != "" {
+					continue
+				}
+				if kind := packageReference(root.name.Local, parent, node.name.Local, attr.Name.Local); kind != packageReferenceNone {
+					add(attr.Name.Local, attr.Value, kind)
 				}
-			case "Page", "TemplatePage", "CompositeGraphicUnit", "Signature":
-				add("BaseLoc", node.attr("BaseLoc"), true)
-			case "PageAnnot":
-				add("FileLoc", node.attr("FileLoc"), true)
-			case "DrawParam":
-				add("Link", node.attr("Link"), true)
-			case "ColorSpace":
-				add("Profile", node.attr("Profile"), false)
-			case "FontFile", "MediaFile", "Profile", "SignedValue", "SchemaLoc", "ExtendData":
-				add("", editorImportText(data, node), false)
-			case "FileLoc":
-				add("", editorImportText(data, node), root.name.Local == "Annotations")
 			}
 			for _, child := range node.children {
 				visit(child)