feat(文件拆分): 拆分PDF格式适配模块

ofdgo_pdf_image_cache.go

@@ -15,12 +15,10 @@
 package ofdgo
 
 import (
-	"bytes"
-	"crypto/sha256"
 	"image"
 	"image/color"
-	"reflect"
 
+	codecpdf "github.com/xiaoqidun/ofdgo/internal/codec/pdf"
 	"github.com/xiaoqidun/pdfgo"
 )
 
@@ -39,85 +37,20 @@ type pdfImageContent struct {
 	id     string
 }
 
-// pdfImageValueBudget 限制图像描述的检查深度和保留字节数
-type pdfImageValueBudget struct {
-	nodes int
-	bytes int
-}
-
 // imageContentKey 对无间接依赖的小图像生成内容键，超大图像仍沿用流缓存
 // 入参: source 图像描述, key 流缓存键
 // 返回: pdfImageContentKey 内容键, bool 是否可复用, error 取消错误
 func (p *pdfImporter) imageContentKey(source *pdfgo.Image, key pdfImageKey) (pdfImageContentKey, bool, error) {
-	if source == nil || source.Stream == nil || len(source.Stream.Data) > 1<<20 {
-		return pdfImageContentKey{}, false, nil
+	digest, reusable, err := codecpdf.ImageDigest(p.ctx, source)
+	if !reusable {
+		return pdfImageContentKey{}, reusable, err
 	}
-	budget := pdfImageValueBudget{nodes: 256, bytes: 65536}
-	for _, value := range []pdfgo.Object{source.Stream.Dictionary, source.ColorSpace, source.Decode, source.Mask, source.SoftMask} {
-		if !pdfImageDirectValue(value, &budget) {
-			return pdfImageContentKey{}, false, p.ctx.Err()
-		}
-	}
-	hash := sha256.New()
-	data := source.Stream.Data
-	for len(data) != 0 {
-		if err := p.ctx.Err(); err != nil {
-			return pdfImageContentKey{}, false, err
-		}
-		n := min(len(data), 65536)
-		hash.Write(data[:n])
-		data = data[n:]
-	}
-	result := pdfImageContentKey{intent: key.intent, process: key.process, tint: key.tint, size: key.size}
-	hash.Sum(result.data[:0])
-	return result, true, p.ctx.Err()
-}
-
-// pdfImageDirectValue 检查图像描述是否完全独立，拒绝间接引用和嵌套流
-// 入参: value 描述值, budget 描述检查预算
-// 返回: bool 是否不依赖阅读器资源
-func pdfImageDirectValue(value pdfgo.Object, budget *pdfImageValueBudget) bool {
-	if budget.nodes <= 0 || budget.bytes < 64 {
-		return false
-	}
-	budget.nodes--
-	budget.bytes -= 64
-	switch v := value.(type) {
-	case nil, pdfgo.Boolean, pdfgo.Integer, pdfgo.Real:
-		return true
-	case pdfgo.Name:
-		budget.bytes -= len(v)
-		return budget.bytes >= 0
-	case pdfgo.String:
-		budget.bytes -= len(v)
-		return budget.bytes >= 0
-	case pdfgo.Array:
-		for _, child := range v {
-			if !pdfImageDirectValue(child, budget) {
-				return false
-			}
-		}
-		return true
-	case pdfgo.Dictionary:
-		for key, child := range v {
-			budget.bytes -= len(key)
-			if !pdfImageDirectValue(child, budget) {
-				return false
-			}
-		}
-		return true
-	}
-	return false
+	return pdfImageContentKey{data: digest, intent: key.intent, process: key.process, tint: key.tint, size: key.size}, true, err
 }
 
 // matches 核对完整编码与有效图像属性，不以摘要相同代替内容相同
 // 入参: source 待复用图像
 // 返回: bool 是否可使用同一图像资源
 func (c pdfImageContent) matches(source *pdfgo.Image) bool {
-	s := c.source
-	return s.Width == source.Width && s.Height == source.Height && s.BitsPerComponent == source.BitsPerComponent &&
-		s.ImageMask == source.ImageMask && s.Interpolate == source.Interpolate &&
-		bytes.Equal(s.Stream.Data, source.Stream.Data) && reflect.DeepEqual(s.Stream.Dictionary, source.Stream.Dictionary) &&
-		reflect.DeepEqual(s.ColorSpace, source.ColorSpace) && reflect.DeepEqual(s.Decode, source.Decode) &&
-		reflect.DeepEqual(s.Mask, source.Mask) && reflect.DeepEqual(s.SoftMask, source.SoftMask)
+	return codecpdf.SameImage(c.source, source)
 }