mirror of
https://github.com/xiaoqidun/ofdgo.git
synced 2026-10-29 17:27:08 +08:00
xiaoqidun committed
1 parent
84808cda61
commit
ec24b6a17d
17 files changed
+1908
-2026
No files matched your search
+239
-66
@@ -21,6 +21,7 @@ import (
|
||||
"fmt"
|
||||
"image/color"
|
||||
"math"
|
||||
"reflect"
|
||||
"strconv"
|
||||
|
||||
basegeom "github.com/xiaoqidun/ofdgo/internal/base/geom"
|
||||
@@ -34,6 +35,13 @@ type pdfPointBounds struct {
|
||||
found bool
|
||||
}
|
||||
|
||||
// pdfClipPath 保存导入器只读的裁剪几何，不缓存对象位移或文字裁剪
|
||||
type pdfClipPath struct {
|
||||
box Box
|
||||
data string
|
||||
path *[1]PathObject
|
||||
}
|
||||
|
||||
// pdfGeometryPath 转换PDF路径并衔接OFD几何类型
|
||||
// 入参: ctx 取消上下文, source PDF路径, matrix 坐标变换
|
||||
// 返回: GeometryPath 独立几何路径, error 指令、坐标或取消错误
|
||||
@@ -286,70 +294,6 @@ func (b *pdfPointBounds) box() Box {
|
||||
return Box{X: b.minX, Y: b.minY, W: math.Max(b.maxX-b.minX, 1e-6), H: math.Max(b.maxY-b.minY, 1e-6)}
|
||||
}
|
||||
|
||||
// pathBoundsContext 逐段计算变换后的控制点边界，不创建临时点集合
|
||||
// 入参: ctx 取消上下文, path PDF路径
|
||||
// 返回: Box 控制点边界，空路径返回零值, error 取消错误
|
||||
func (p *pdfImporter) pathBoundsContext(ctx context.Context, path pdfgo.Path) (Box, error) {
|
||||
if err := ctx.Err(); err != nil {
|
||||
return Box{}, err
|
||||
}
|
||||
minX, minY, maxX, maxY := math.Inf(1), math.Inf(1), math.Inf(-1), math.Inf(-1)
|
||||
found := false
|
||||
for _, segment := range path.Segments {
|
||||
if err := ctx.Err(); err != nil {
|
||||
return Box{}, err
|
||||
}
|
||||
for index, point := range segment.Points {
|
||||
if index&255 == 0 {
|
||||
if err := ctx.Err(); err != nil {
|
||||
return Box{}, err
|
||||
}
|
||||
}
|
||||
point = p.matrix.Apply(point)
|
||||
minX, minY = math.Min(minX, point.X), math.Min(minY, point.Y)
|
||||
maxX, maxY = math.Max(maxX, point.X), math.Max(maxY, point.Y)
|
||||
found = true
|
||||
}
|
||||
}
|
||||
if err := ctx.Err(); err != nil {
|
||||
return Box{}, err
|
||||
}
|
||||
if !found {
|
||||
return Box{}, nil
|
||||
}
|
||||
return Box{X: minX, Y: minY, W: math.Max(maxX-minX, 1e-6), H: math.Max(maxY-minY, 1e-6)}, nil
|
||||
}
|
||||
|
||||
// pathDataContext 转换路径并在分段处理时检查取消
|
||||
// 入参: ctx 取消上下文, path PDF路径, origin 对象边界
|
||||
// 返回: string OFD路径数据, error 取消错误
|
||||
func (p *pdfImporter) pathDataContext(ctx context.Context, path pdfgo.Path, origin Box) (string, error) {
|
||||
if err := ctx.Err(); err != nil {
|
||||
return "", err
|
||||
}
|
||||
data := make([]byte, 0, len(path.Segments)*48)
|
||||
for index, segment := range path.Segments {
|
||||
if err := ctx.Err(); err != nil {
|
||||
return "", err
|
||||
}
|
||||
if index != 0 {
|
||||
data = append(data, ' ')
|
||||
}
|
||||
data = append(data, segment.Operator...)
|
||||
for _, point := range segment.Points {
|
||||
point = p.matrix.Apply(point)
|
||||
data = append(data, ' ')
|
||||
data = strconv.AppendFloat(data, point.X-origin.X, 'g', -1, 64)
|
||||
data = append(data, ' ')
|
||||
data = strconv.AppendFloat(data, point.Y-origin.Y, 'g', -1, 64)
|
||||
}
|
||||
}
|
||||
if err := ctx.Err(); err != nil {
|
||||
return "", err
|
||||
}
|
||||
return string(data), nil
|
||||
}
|
||||
|
||||
// path 保留每次路径绘制操作，重叠描边不合并为单次绘制
|
||||
// 入参: mark PDF路径绘制信息
|
||||
// 返回: error 错误信息
|
||||
@@ -383,7 +327,7 @@ func (p *pdfImporter) appendPath(mark pdfgo.PathMark) error {
|
||||
}
|
||||
return p.gradientPath(mark)
|
||||
}
|
||||
box, err := p.pathBoundsContext(p.ctx, mark.Path)
|
||||
box, err := codecpdf.PathBounds(p.ctx, mark.Path, p.matrix)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
@@ -447,7 +391,7 @@ func (p *pdfImporter) appendPath(mark pdfgo.PathMark) error {
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
data, err := drawing.pathDataContext(p.ctx, mark.Path, origin)
|
||||
data, err := codecpdf.PathData(p.ctx, mark.Path, drawing.matrix, origin)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
@@ -465,3 +409,232 @@ func (p *pdfImporter) appendPath(mark pdfgo.PathMark) error {
|
||||
p.report.PathObjects++
|
||||
return nil
|
||||
}
|
||||
|
||||
// clipPath 复用页面内重复路径编码，按估算字节预算淘汰且不缓存错误
|
||||
// 入参: path 裁剪路径
|
||||
// 返回: pdfClipPath 变换后的边界及编码, error 取消错误
|
||||
func (p *pdfImporter) clipPath(path pdfgo.Path) (pdfClipPath, error) {
|
||||
key, err := codecpdf.ClipPathKey(p.ctx, path, p.matrix)
|
||||
if err != nil {
|
||||
return pdfClipPath{}, err
|
||||
}
|
||||
if p.clipPaths == nil {
|
||||
p.clipPaths = &renderCache[[32]byte, pdfClipPath]{limit: 8 << 20}
|
||||
}
|
||||
if cached, ok := p.clipPaths.get(key); ok {
|
||||
return cached, nil
|
||||
}
|
||||
box, err := codecpdf.PathBounds(p.ctx, path, p.matrix)
|
||||
if err != nil {
|
||||
return pdfClipPath{}, err
|
||||
}
|
||||
data, err := codecpdf.PathData(p.ctx, path, p.matrix, box)
|
||||
if err != nil {
|
||||
return pdfClipPath{}, err
|
||||
}
|
||||
rule := "NonZero"
|
||||
if path.EvenOdd {
|
||||
rule = "Even-Odd"
|
||||
}
|
||||
boundary := pdfBoundary(box)
|
||||
fill, stroke := true, false
|
||||
result := pdfClipPath{box: box, data: data, path: &[1]PathObject{{Boundary: boundary, AbbreviatedData: data, Rule: rule, Fill: &fill, Stroke: &stroke}}}
|
||||
p.clipPaths.put(key, result, len(data)+len(boundary)+512)
|
||||
return result, nil
|
||||
}
|
||||
|
||||
// sharedClipObjects 将连续重复的大裁剪区提取为复合对象，同步缓冲区及来源区段
|
||||
// 入参: objects 本库生成且尚未分配标识的对象
|
||||
// 返回: []GraphicObject 提取后的对象, error 资源创建或取消错误
|
||||
func (p *pdfImporter) sharedClipObjects(objects []GraphicObject) ([]GraphicObject, error) {
|
||||
var output []GraphicObject
|
||||
var mapped pdfObjectBuffer
|
||||
for start := 0; start < len(objects); {
|
||||
if err := p.ctx.Err(); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
clips := pdfPageClip(objects[start])
|
||||
end := start + 1
|
||||
if clips != nil {
|
||||
for end < len(objects) && p.objects.origin(start).equal(p.objects.origin(end)) && pdfMatchesPageClip(objects[end], clips) {
|
||||
if err := p.ctx.Err(); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
end++
|
||||
}
|
||||
}
|
||||
if end-start < 2 {
|
||||
if output != nil {
|
||||
mapped.recordOrigin(len(output), p.objects.origin(start))
|
||||
output = append(output, objects[start])
|
||||
}
|
||||
start = end
|
||||
continue
|
||||
}
|
||||
if output == nil {
|
||||
output = append(make([]GraphicObject, 0, len(objects)), objects[:start]...)
|
||||
if p.objects != nil {
|
||||
for _, run := range p.objects.sequences {
|
||||
if run.start >= start {
|
||||
break
|
||||
}
|
||||
mapped.sequences = append(mapped.sequences, run)
|
||||
}
|
||||
}
|
||||
}
|
||||
members := objects[start:end:end]
|
||||
for index := range members {
|
||||
*editorObjectClips(&members[index]) = nil
|
||||
}
|
||||
id, err := p.objectValidation().addOwnedVector(p.ctx, members, p.pageWidth, p.pageHeight)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
mapped.recordOrigin(len(output), p.objects.origin(start))
|
||||
output = append(output, GraphicObject{Type: "CompositeObject", CompositeGraphicUnit: CompositeGraphicUnit{
|
||||
Boundary: pdfBoundary(Box{W: p.pageWidth, H: p.pageHeight}), ResourceID: id, Clips: clips,
|
||||
}})
|
||||
start = end
|
||||
}
|
||||
if output == nil {
|
||||
return objects, nil
|
||||
}
|
||||
if p.objects != nil {
|
||||
mapped.parts = [][]GraphicObject{output}
|
||||
mapped.size = len(output)
|
||||
*p.objects = mapped
|
||||
}
|
||||
return output, nil
|
||||
}
|
||||
|
||||
// pdfPageClip 将导入裁剪区还原到页面坐标，仅提取包含大量重复路径数据的裁剪
|
||||
// 入参: object 导入对象
|
||||
// 返回: *Clips 页面坐标裁剪，不满足提取条件时为空
|
||||
func pdfPageClip(object GraphicObject) *Clips {
|
||||
if object.Type != "TextObject" && object.Type != "PathObject" && object.Type != "ImageObject" {
|
||||
return nil
|
||||
}
|
||||
clips := *editorObjectClips(&object)
|
||||
if clips == nil || clips.usesObjectMatrix() || len(clips.Clip) == 0 {
|
||||
return nil
|
||||
}
|
||||
boundary, _ := editorGeometry(object)
|
||||
box, err := ParseBox(boundary)
|
||||
if err != nil {
|
||||
return nil
|
||||
}
|
||||
translation := ""
|
||||
if box.X != 0 || box.Y != 0 {
|
||||
translation = pdfNumbers(1, 0, 0, 1, -box.X, -box.Y)
|
||||
}
|
||||
bytes := 0
|
||||
result := &Clips{Clip: make([]Clip, len(clips.Clip))}
|
||||
for index, clip := range clips.Clip {
|
||||
if len(clip.Area) != 1 || clip.Area[0].CTM != translation {
|
||||
return nil
|
||||
}
|
||||
area := clip.Area[0]
|
||||
area.CTM = ""
|
||||
result.Clip[index].Area = []ClipArea{area}
|
||||
for _, path := range area.Path {
|
||||
bytes += len(path.AbbreviatedData)
|
||||
}
|
||||
}
|
||||
if bytes < 2048 {
|
||||
return nil
|
||||
}
|
||||
return result
|
||||
}
|
||||
|
||||
// pdfMatchesPageClip 比较页面坐标裁剪，不将相同外接矩形视为相同路径
|
||||
// 入参: object 导入对象, pageClip 已还原的裁剪区
|
||||
// 返回: bool 是否可共享裁剪区
|
||||
func pdfMatchesPageClip(object GraphicObject, pageClip *Clips) bool {
|
||||
if object.Type != "TextObject" && object.Type != "PathObject" && object.Type != "ImageObject" {
|
||||
return false
|
||||
}
|
||||
clips := *editorObjectClips(&object)
|
||||
if clips == nil || clips.usesObjectMatrix() || len(clips.Clip) != len(pageClip.Clip) {
|
||||
return false
|
||||
}
|
||||
boundary, _ := editorGeometry(object)
|
||||
box, err := ParseBox(boundary)
|
||||
if err != nil {
|
||||
return false
|
||||
}
|
||||
translation := ""
|
||||
if box.X != 0 || box.Y != 0 {
|
||||
translation = pdfNumbers(1, 0, 0, 1, -box.X, -box.Y)
|
||||
}
|
||||
for index, clip := range clips.Clip {
|
||||
if len(clip.Area) != 1 || clip.Area[0].CTM != translation {
|
||||
return false
|
||||
}
|
||||
area, expected := clip.Area[0], pageClip.Clip[index].Area[0]
|
||||
if area.DrawParam != expected.DrawParam || !reflect.DeepEqual(area.Path, expected.Path) || !reflect.DeepEqual(area.Text, expected.Text) {
|
||||
return false
|
||||
}
|
||||
}
|
||||
return true
|
||||
}
|
||||
|
||||
// clips 省略覆盖对象边界的矩形裁剪，复用其余几何并保留位移及路径交集
|
||||
// 入参: paths 裁剪路径, origin 对象边界
|
||||
// 返回: *Clips OFD裁剪区域，无裁剪时为空, error 未解析的裁剪字形
|
||||
func (p *pdfImporter) clips(paths []pdfgo.Path, origin Box) (*Clips, error) {
|
||||
if len(paths) == 0 {
|
||||
return nil, nil
|
||||
}
|
||||
var clips *Clips
|
||||
var areas []ClipArea
|
||||
translation := ""
|
||||
for _, path := range paths {
|
||||
if err := p.ctx.Err(); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
if codecpdf.ClipContainsBoundary(path, p.matrix, origin) {
|
||||
continue
|
||||
}
|
||||
if clips == nil {
|
||||
clips = &Clips{Clip: make([]Clip, 0, len(paths))}
|
||||
areas = make([]ClipArea, len(paths))
|
||||
if origin.X != 0 || origin.Y != 0 {
|
||||
translation = pdfNumbers(1, 0, 0, 1, -origin.X, -origin.Y)
|
||||
}
|
||||
}
|
||||
index := len(clips.Clip)
|
||||
area := &areas[index]
|
||||
area.CTM = translation
|
||||
if len(path.Segments) != 0 {
|
||||
encoded, err := p.clipPath(path)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
area.Path = encoded.path[:1:1]
|
||||
}
|
||||
for _, mark := range path.Text {
|
||||
objects, ok := p.clipTexts[mark]
|
||||
if !ok {
|
||||
count := p.objects.len()
|
||||
textCount := p.report.TextObjects
|
||||
if err := p.text(pdfgo.TextMark{Font: mark.Font, Glyphs: mark.Glyphs, Positions: mark.Positions, Matrix: mark.Matrix, Size: mark.Size, HorizontalScale: mark.HorizontalScale, Mode: 7, Clip: mark, Style: pdfgo.Style{Fill: pdfgo.Paint{Alpha: 1}}}); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
p.objects.truncate(count)
|
||||
p.report.TextObjects = textCount
|
||||
objects = p.clipTexts[mark]
|
||||
}
|
||||
if len(path.Text) == 1 {
|
||||
area.Text = objects[:len(objects):len(objects)]
|
||||
} else {
|
||||
area.Text = append(area.Text, objects...)
|
||||
}
|
||||
}
|
||||
if len(area.Path)+len(area.Text) == 0 {
|
||||
fill, stroke := true, false
|
||||
area.Path = []PathObject{{Boundary: "0 0 1 1", AbbreviatedData: "M 0 0 C", Fill: &fill, Stroke: &stroke}}
|
||||
}
|
||||
clips.Clip = append(clips.Clip, Clip{Area: areas[index : index+1 : index+1]})
|
||||
}
|
||||
return clips, nil
|
||||
}
|
||||
Reference in new issue
Block a user