feat(归并文件): 归并根包文件

2026-10-29 08:46:55 +08:00 · 2026-10-19 10:17:59 +08:00
1 parent ec24b6a17d
commit 1a8de5a3f1
@@ -20,6 +20,7 @@ import (
 	"encoding/xml"
 	"fmt"
 	"io"
+	"maps"
 	"path"
 	"slices"
 	"strings"
@@ -74,6 +75,14 @@ type resourcePageFrame struct {
 	link   bool
 }

+// ResourceLocation 保存资源引用的页面实例范围，Page从1开始，Bounds单位为毫米
+// Object为承载引用的图元标识；隐藏或完全裁剪的图元不产生定位结果
+type ResourceLocation struct {
+	Page   int    `json:"page"`
+	Object string `json:"object"`
+	Bounds Box    `json:"bounds"`
+}
+
 // Resources 列出当前文档及所选版本的资源，标记缺失文件，不读取二进制内容或执行媒体动作
 // 重复查询复用已完成的扫描，返回值及引用位置与阅读器相互独立
 // 入参: ctx 取消上下文
@@ -485,3 +494,186 @@ func findResource(items []ResourceInfo, kind, id string) (ResourceInfo, error) {
 	}
 	return ResourceInfo{}, fmt.Errorf("resource not found: %s/%s", kind, id)
 }
+
+// ResourceLocations 定位当前文档及版本内的一处资源引用，返回各页面上的可见使用范围
+// 模板、图案和复合资源沿引用关系定位承载图元，不将资源声明位置当作页面位置
+// 入参: ctx 取消上下文, kind 资源类型, id 资源标识, reference 为Resources返回的引用索引
+// 返回: []ResourceLocation 页面位置，无可见图元时为空, error 引用、度量或取消错误
+func (r *Renderer) ResourceLocations(ctx context.Context, kind, id string, reference int) ([]ResourceLocation, error) {
+	catalog, err := r.Reader.resourceCatalog(ctx)
+	if err != nil {
+		return nil, err
+	}
+	item, err := findResource(catalog.items, kind, id)
+	if err != nil {
+		return nil, err
+	}
+	if reference < 0 || reference >= len(item.References) {
+		return nil, fmt.Errorf("resource reference not found")
+	}
+	ref := item.References[reference]
+	doc, err := r.Reader.DocContext(ctx)
+	if err != nil {
+		return nil, err
+	}
+	wanted := make(map[string]bool)
+	users := make(map[string][]string)
+	for _, file := range catalog.scan.files {
+		root, err := parseEditorXMLContext(ctx, file.data)
+		if err != nil {
+			return nil, err
+		}
+		var visit func(*editorXML, string) error
+		visit = func(node *editorXML, object string) error {
+			if err := ctx.Err(); err != nil {
+				return err
+			}
+			if !node.matchesOFD(node.name.Local) || node.matchesOFD("Attachments") {
+				return nil
+			}
+			if slices.Contains([]string{"TextObject", "PathObject", "ImageObject", "CompositeObject"}, node.name.Local) {
+				object = node.attr("ID")
+			}
+			owner := resourceReferenceOwner(node)
+			if owner != "" && resourceColorNode(node) && node.attr("ColorSpace") == "" && doc.CommonData.DefaultCS != 0 {
+				dependency := fmt.Sprint(doc.CommonData.DefaultCS)
+				users[dependency] = append(users[dependency], owner)
+				if ref.Role == "DefaultCS" && sameResourceID(dependency, item.ID) {
+					wanted[owner] = true
+				}
+			}
+			add := func(role, value string) {
+				if owner == "" {
+					return
+				}
+				if role != "AttachID" {
+					if dependency := editorResourceID(value); dependency != "" && dependency != owner {
+						users[dependency] = append(users[dependency], owner)
+					}
+				}
+				match := value == item.ID
+				if kind != "Attachment" {
+					match = sameResourceID(value, item.ID)
+				}
+				if file.name == ref.File && object == ref.Object && role == ref.Role && match {
+					wanted[owner] = true
+				}
+			}
+			for _, attr := range node.attrs {
+				if attr.Name.Space == "" && (editorResourceReferenceKind(attr.Name.Local, root.matchesOFD("Res")) == editorReferenceID || attr.Name.Local == "AttachID") {
+					add(attr.Name.Local, attr.Value)
+				}
+			}
+			if slices.Contains([]string{"Thumbnail", "Substitution"}, node.name.Local) {
+				add(node.name.Local, strings.TrimSpace(editorImportText(file.data, node)))
+			}
+			parent := node.parent
+			for parent != nil && parent.matchesOFD("PageBlock") {
+				parent = parent.parent
+			}
+			if owner != "" && parent != nil && parent.matchesOFD("Layer") {
+				if layer := editorResourceID(parent.attr("ID")); layer != "" && layer != owner {
+					users[layer] = append(users[layer], owner)
+				}
+			}
+			for _, child := range node.children {
+				if err := visit(child, object); err != nil {
+					return err
+				}
+			}
+			return nil
+		}
+		if err := visit(root, ""); err != nil {
+			return nil, err
+		}
+	}
+	queue := slices.Collect(maps.Keys(wanted))
+	for len(queue) != 0 {
+		if err := ctx.Err(); err != nil {
+			return nil, err
+		}
+		id := queue[0]
+		queue = queue[1:]
+		for _, user := range users[id] {
+			if !wanted[user] {
+				wanted[user] = true
+				queue = append(queue, user)
+			}
+		}
+	}
+	view := *r
+	view.RenderAnnotations, view.pageContext = true, ctx
+	var result []ResourceLocation
+	for page := range doc.Pages.Page {
+		if ref.Page >= 0 && ref.Page != page || len(wanted) == 0 {
+			continue
+		}
+		content, err := r.Reader.PageContentByIndexContext(ctx, page)
+		if err != nil {
+			return nil, err
+		}
+		targets := make(map[string]*StructureTarget, len(wanted))
+		for id := range wanted {
+			targets[id] = &StructureTarget{ID: id}
+		}
+		visitor := &structureVisitor{ctx: ctx, renderer: &view, targets: targets, boundsOnly: true}
+		if err := view.WalkPage(content, visitor); err != nil {
+			return nil, err
+		}
+		for _, id := range slices.Sorted(maps.Keys(targets)) {
+			box := targets[id].Bounds
+			if box.W > 0 && box.H > 0 {
+				result = append(result, ResourceLocation{Page: page + 1, Object: id, Bounds: box})
+			}
+		}
+	}
+	if err := ctx.Err(); err != nil {
+		return nil, err
+	}
+	return result, nil
+}
+
+// resourceReferenceOwner 查找承载引用的资源定义或最外层图元，图层只承接其自身的绘制参数
+// 入参: node 引用节点
+// 返回: string 图元或资源标识，无法映射时为空
+func resourceReferenceOwner(node *editorXML) string {
+	id := ""
+	for current := node; current != nil; current = current.parent {
+		if !current.matchesOFD(current.name.Local) {
+			return ""
+		}
+		if slices.Contains([]string{"TextObject", "PathObject", "ImageObject", "CompositeObject", "CompositeGraphicUnit"}, current.name.Local) || current.parent != nil && current.name.Local == editorResourceElement(current.parent.name.Local) {
+			id = editorResourceID(current.attr("ID"))
+		}
+		if id == "" && current.matchesOFD("Layer") {
+			id = editorResourceID(current.attr("ID"))
+		}
+	}
+	return id
+}
+
+// Media 获取已加载的音视频资源，页面资源须先读取所属页面
+// 不加载注解或签名
+// 入参: id 资源标识
+// 返回: MultiMedia 资源信息, error 资源缺失或类型错误
+func (r *Reader) Media(id string) (MultiMedia, error) {
+	if _, err := r.docStructure(); err != nil {
+		return MultiMedia{}, err
+	}
+	media, ok := resourceValue(r.mediaCache, id)
+	if !ok || media.Type != "Audio" && media.Type != "Video" {
+		return MultiMedia{}, fmt.Errorf("audio or video resource %q is unavailable", id)
+	}
+	return media, nil
+}
+
+// OpenMedia 打开文档内音视频原始数据，不解码或执行播放
+// 入参: id 资源标识
+// 返回: io.ReadCloser 数据流，由调用方关闭, error 资源或文件读取错误
+func (r *Reader) OpenMedia(id string) (io.ReadCloser, error) {
+	media, err := r.Media(id)
+	if err != nil {
+		return nil, err
+	}
+	return r.openFile(r.ResPath(media.MediaFile))
+}