feat(内存占用): 降低复杂文档阅读的内存占用

assets/webui/ofdgo.js

@@ -56,6 +56,7 @@ const state = {
 	renderAnnotations: true,
 	pageCache: new Map(),
 	svgFonts: new Map(),
+	svgImages: new Map(),
 	selectedPages: new Set(),
 	documentSelection: null,
 	pageInFlight: new Map(),
@@ -1710,6 +1711,7 @@ function resetPageFlow() {
 		document.fonts.delete(font);
 	}
 	state.svgFonts.clear();
+	state.svgImages.clear();
 	state.pageCache.clear();
 	state.selectedPages.clear();
 	state.documentSelection = null;
@@ -1862,6 +1864,12 @@ async function processPageRenderQueue() {
 				const page = await callWASM("ofdgoRenderPage", task.index);
 				await loadSVGFonts(page.fonts, task.openSeq);
 				if (task.openSeq === state.openSeq) {
+					for (const image of page.images) {
+						if (!state.svgImages.has(image.name)) {
+							state.svgImages.set(image.name, image.url);
+						}
+					}
+					delete page.images;
 					page.text = JSON.parse(page.text);
 					state.pageCache.set(task.index, page);
 				}
@@ -2229,7 +2237,7 @@ function layoutPageShell(shell, page) {
 
 function parseSVG(svgText, prefix = "") {
 	const template = document.createElement("template");
-	template.innerHTML = svgText;
+	template.innerHTML = svgText.replace(/xlink:href="(ofdgo-image-[a-f0-9]{64})"/g, (_, name) => `xlink:href="${state.svgImages.get(name)}"`);
 	const svg = document.adoptNode(template.content.firstElementChild);
 	prefixSVGIds(svg, prefix);
 	return svg;

internal/webui/wasm.go

@@ -19,6 +19,7 @@ package webui
 import (
 	"bufio"
 	"context"
+	"encoding/base64"
 	"encoding/json"
 	"fmt"
 	"io"
@@ -282,6 +283,10 @@ func renderPage(args []js.Value) (any, error) {
 	for i, font := range page.Fonts {
 		fonts[i] = map[string]any{"name": font.Name, "weight": font.Weight, "style": font.Style}
 	}
+	images := make([]any, len(page.Images))
+	for i, image := range page.Images {
+		images[i] = map[string]any{"name": image.Name, "url": "data:" + image.MIME + ";base64," + base64.StdEncoding.EncodeToString(image.Data)}
+	}
 	return successResult(map[string]any{
 		"index":  page.Index,
 		"number": page.Number,
@@ -291,6 +296,7 @@ func renderPage(args []js.Value) (any, error) {
 		"svg":    page.SVG,
 		"links":  links,
 		"fonts":  fonts,
+		"images": images,
 		"text":   string(textData),
 	}), nil
 }

internal/webui/webui.go

@@ -163,6 +163,7 @@ type PageSVG struct {
 	SVG    string           `json:"svg"`
 	Links  []ofdgo.PageLink `json:"-"`
 	Fonts  []ofdgo.SVGFont  `json:"-"`
+	Images []ofdgo.SVGImage `json:"-"`
 }
 
 // ExportFormat 导出格式
@@ -402,19 +403,17 @@ func (s *Session) Info() DocumentInfo {
 		info.CreationDate = docInfo.CreationDate
 		info.ModDate = docInfo.ModDate
 	}
-	pages := make([]*ofdgo.PageContent, 0, len(s.doc.Pages.Page))
 	for index, pageRef := range s.doc.Pages.Page {
 		pageInfo := PageInfo{Index: index, ID: pageRef.ID}
-		if page, err := s.pageContent(index); err == nil {
+		if area, err := s.Reader.PageArea(pageRef); err == nil {
-			pages = append(pages, page)
+			if box, err := s.pageBox(index, &ofdgo.PageContent{Area: area}); err == nil {
-			if box, err := s.pageBox(index, page); err == nil {
 				pageInfo.Width = box.W
 				pageInfo.Height = box.H
 			}
 		}
 		info.Pages = append(info.Pages, pageInfo)
 	}
-	if fonts, err := s.Renderer.FontInfosFromPages(pages); err == nil {
+	if fonts, err := s.Renderer.FontInfos(); err == nil {
 		info.Fonts = fonts
 	}
 	info.FontCount = len(info.Fonts)
@@ -440,7 +439,7 @@ func (s *Session) RenderPageSVG(index int) (PageSVG, error) {
 			s.textCache[index] = text
 		}
 	}
-	fonts, err := renderer.RenderToSVGWithFonts(page, &buf)
+	resources, err := renderer.RenderToSVGWithResources(page, &buf)
 	if err != nil {
 		return PageSVG{}, err
 	}
@@ -448,11 +447,11 @@ func (s *Session) RenderPageSVG(index int) (PageSVG, error) {
 	if err != nil {
 		return PageSVG{}, err
 	}
-	for i, font := range fonts {
+	for i, font := range resources.Fonts {
 		s.svgFonts[font.Name] = font.Data
-		fonts[i].Data = nil
+		resources.Fonts[i].Data = nil
 	}
-	return PageSVG{Index: index, Number: index + 1, ID: page.ID, Width: box.W, Height: box.H, SVG: buf.String(), Links: links, Fonts: fonts}, nil
+	return PageSVG{Index: index, Number: index + 1, ID: page.ID, Width: box.W, Height: box.H, SVG: buf.String(), Links: links, Fonts: resources.Fonts, Images: resources.Images}, nil
 }
 
 // SVGFontData 获取已渲染页面引用的字体数据

ofdgo_font_info.go

@@ -15,6 +15,7 @@
 package ofdgo
 
 import (
+	"encoding/xml"
 	"fmt"
 	"io"
 	"path"
@@ -214,8 +215,10 @@ func (r *Renderer) fontUsage(doc *Document, pages []*PageContent) map[string]int
 	usage := make(map[string]int)
 	if pages == nil {
 		for _, pageRef := range doc.Pages.Page {
-			if page, err := r.Reader.PageContent(pageRef); err == nil {
+			if counts, err := r.readPageFontUsage(pageRef); err == nil {
-				r.countPageFonts(page, usage)
+				for id, count := range counts {
+					usage[id] += count
+				}
 			}
 		}
 	} else {
@@ -233,14 +236,104 @@ func (r *Renderer) fontUsage(doc *Document, pages []*PageContent) map[string]int
 	return usage
 }
 
+// fontUsagePage 逐图元统计字体的页面解码器
+type fontUsagePage struct {
+	renderer *Renderer
+	usage    map[string]int
+}
+
+// readPageFontUsage 读取页面字体用量，不保留页面图元
+// 入参: page 页面引用
+// 返回: map[string]int 字体使用次数, error 错误信息
+func (r *Renderer) readPageFontUsage(page Page) (map[string]int, error) {
+	r.Reader.loadPageResources(page)
+	f, err := r.Reader.openFile(r.Reader.ResPath(page.BaseLoc))
+	if err != nil {
+		return nil, err
+	}
+	defer f.Close()
+	content := fontUsagePage{renderer: r, usage: make(map[string]int)}
+	if err := xml.NewDecoder(f).Decode(&content); err != nil {
+		return nil, err
+	}
+	if _, err := io.Copy(io.Discard, f); err != nil {
+		return nil, err
+	}
+	r.countAnnotationFonts(page.ID, content.usage)
+	return content.usage, nil
+}
+
+// UnmarshalXML 解析页面内容中的图层字体
+// 入参: d XML解码器, start 页面起始节点
+// 返回: error 错误信息
+func (p *fontUsagePage) UnmarshalXML(d *xml.Decoder, start xml.StartElement) error {
+	if start.Name.Local != "Page" {
+		return fmt.Errorf("expected element type <Page> but have <%s>", start.Name.Local)
+	}
+	return decodeObjectContainer(d, start, func(d *xml.Decoder, node xml.StartElement) error {
+		if node.Name.Local != "Content" {
+			return d.Skip()
+		}
+		return decodeObjectContainer(d, node, func(d *xml.Decoder, layer xml.StartElement) error {
+			if layer.Name.Local != "Layer" {
+				return d.Skip()
+			}
+			return p.decodeObjects(d, layer, p.renderer.drawParamDefaults(attrValue(layer, "DrawParam"), nil))
+		})
+	})
+}
+
+// decodeObjects 逐个读取图元并复用字体统计规则
+// 入参: d XML解码器, start 容器起始节点, defaults 默认绘制参数
+// 返回: error 错误信息
+func (p *fontUsagePage) decodeObjects(d *xml.Decoder, start xml.StartElement, defaults *DrawParam) error {
+	return decodeObjectContainer(d, start, func(d *xml.Decoder, node xml.StartElement) error {
+		switch node.Name.Local {
+		case "PageBlock":
+			return p.decodeObjects(d, node, defaults)
+		case "PathObject":
+			var obj struct {
+				PathObject
+				AbbreviatedData struct{} `xml:"AbbreviatedData"`
+			}
+			if err := d.DecodeElement(&obj, &node); err != nil {
+				return err
+			}
+			p.renderer.countObjectFonts(GraphicObject{Type: "PathObject", PathObject: obj.PathObject}, p.usage, defaults, nil)
+			return nil
+		case "ImageObject":
+			var obj ImageObject
+			if err := d.DecodeElement(&obj, &node); err != nil {
+				return err
+			}
+			p.renderer.countObjectFonts(GraphicObject{Type: "ImageObject", ImageObject: obj}, p.usage, defaults, nil)
+			return nil
+		}
+		var layer Layer
+		if err := layer.decodeObject(d, node); err != nil {
+			return err
+		}
+		for _, obj := range layer.Objects {
+			p.renderer.countObjectFonts(obj, p.usage, defaults, nil)
+		}
+		return nil
+	})
+}
+
 // countPageFonts 统计页面字体使用次数
 // 入参: page 页面内容, usage 字体使用次数
 func (r *Renderer) countPageFonts(page *PageContent, usage map[string]int) {
 	for _, layer := range page.Content.Layer {
 		r.countLayerFonts(layer, usage)
 	}
+	r.countAnnotationFonts(page.ID, usage)
+}
+
+// countAnnotationFonts 统计页面注释中的字体使用次数
+// 入参: pageID 页面标识, usage 字体使用次数
+func (r *Renderer) countAnnotationFonts(pageID string, usage map[string]int) {
 	if r.RenderAnnotations {
-		for _, annot := range r.Reader.Annots[page.ID] {
+		for _, annot := range r.Reader.Annots[pageID] {
 			if annot.Visible != nil && !*annot.Visible {
 				continue
 			}

ofdgo_renderer_svg.go

@@ -15,12 +15,17 @@
 package ofdgo
 
 import (
+	"bufio"
+	"bytes"
 	"crypto/sha256"
 	"fmt"
+	"image"
+	"image/png"
 	"io"
 	"strings"
 
 	"github.com/tdewolff/canvas"
+	canvasimage "github.com/tdewolff/canvas/image"
 	"github.com/tdewolff/canvas/renderers/svg"
 )
 
@@ -32,38 +37,114 @@ type SVGFont struct {
 	Data   []byte
 }
 
-// svgFontRenderer 分离字体资源的SVG渲染器
+// SVGImage SVG引用的图片资源，Name为图片地址，Data为只读图片数据
-type svgFontRenderer struct {
+type SVGImage struct {
+	Name string
+	MIME string
+	Data []byte
+}
+
+// SVGResources SVG引用的字体和图片资源
+type SVGResources struct {
+	Fonts  []SVGFont
+	Images []SVGImage
+}
+
+// svgResourceRenderer 分离外部资源的SVG渲染器
+type svgResourceRenderer struct {
 	*svg.SVG
-	renderer *Renderer
+	renderer   *Renderer
-	fonts    []SVGFont
+	writer     io.Writer
-	seen     map[string]bool
+	fonts      []SVGFont
-	styles   strings.Builder
+	images     []SVGImage
+	imageNames map[image.Image]string
+	seen       map[string]bool
+	styles     strings.Builder
+	err        error
 }
 
 // RenderToSVGWithFonts 渲染为SVG并返回页面引用的字体资源，不在SVG中嵌入字体
 // 入参: page 页面内容, writer 输出流
 // 返回: []SVGFont 字体资源，可按Name复用并按Weight、Style注册，error 错误信息
 func (r *Renderer) RenderToSVGWithFonts(page *PageContent, writer io.Writer) ([]SVGFont, error) {
+	resources, err := r.renderSVGResources(page, writer, false)
+	return resources.Fonts, err
+}
+
+// RenderToSVGWithResources 渲染为SVG并分离字体和图片，调用方需注册字体并将图片Name映射为可访问地址
+// 入参: page 页面内容, writer 输出流
+// 返回: SVGResources 可跨页面复用的资源, error 错误信息
+func (r *Renderer) RenderToSVGWithResources(page *PageContent, writer io.Writer) (SVGResources, error) {
+	return r.renderSVGResources(page, writer, true)
+}
+
+// renderSVGResources 渲染SVG并收集外部资源
+// 入参: page 页面内容, writer 输出流, images 是否分离图片
+// 返回: SVGResources 外部资源, error 错误信息
+func (r *Renderer) renderSVGResources(page *PageContent, writer io.Writer, images bool) (SVGResources, error) {
 	c, err := r.renderPage(page)
 	if err != nil {
-		return nil, err
+		return SVGResources{}, err
 	}
 	options := svg.DefaultOptions
 	options.EmbedFonts = false
-	s := &svgFontRenderer{
+	buffer := bufio.NewWriter(writer)
-		SVG:      svg.New(writer, c.W, c.H, &options),
+	s := &svgResourceRenderer{
+		SVG:      svg.New(buffer, c.W, c.H, &options),
 		renderer: r,
+		writer:   buffer,
 		seen:     make(map[string]bool),
 	}
+	if images {
+		s.imageNames = make(map[image.Image]string)
+	}
 	c.RenderTo(s)
+	if s.err != nil {
+		return SVGResources{}, s.err
+	}
 	s.SetCustomStyle(s.styles.String())
-	return s.fonts, s.Close()
+	if err := s.Close(); err != nil {
+		return SVGResources{}, err
+	}
+	return SVGResources{Fonts: s.fonts, Images: s.images}, buffer.Flush()
+}
+
+// RenderImage 绘制图片，分离资源时保持原始编码、尺寸和变换
+// 入参: img 图片对象, m 变换矩阵
+func (s *svgResourceRenderer) RenderImage(img image.Image, m canvas.Matrix) {
+	if s.imageNames == nil {
+		s.SVG.RenderImage(img, m)
+		return
+	}
+	name, ok := s.imageNames[img]
+	if !ok {
+		resource := SVGImage{MIME: "image/png"}
+		if encoded, ok := img.(*canvasimage.Image); ok {
+			resource.MIME, resource.Data = encoded.Mimetype, encoded.Bytes
+		} else {
+			var buffer bytes.Buffer
+			if err := png.Encode(&buffer, img); err != nil {
+				s.err = err
+				return
+			}
+			resource.Data = buffer.Bytes()
+		}
+		name = fmt.Sprintf("ofdgo-image-%x", sha256.Sum256(resource.Data))
+		resource.Name = name
+		s.imageNames[img] = name
+		if !s.seen[name] {
+			s.seen[name] = true
+			s.images = append(s.images, resource)
+		}
+	}
+	_, height := s.Size()
+	size := img.Bounds().Size()
+	fmt.Fprintf(s.writer, `<image transform="%s" width="%d" height="%d" xlink:href="%s"/>`, m.Translate(0, float64(size.Y)).ToSVG(height), size.X, size.Y, name)
 }
 
 // RenderText 绘制文字并记录实际使用的字体资源
 // 入参: text 文字对象, m 变换矩阵
-func (s *svgFontRenderer) RenderText(text *canvas.Text, m canvas.Matrix) {
+func (s *svgResourceRenderer) RenderText(text *canvas.Text, m canvas.Matrix) {
 	if text.Empty() {
 		return
 	}