Fix #353

pdfcpu · Jul 13, 2021 · 281745f · 281745f
1 parent 63d5ab6
commit 281745f
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 90 deletions.
diff --git a/pkg/api/extract.go b/pkg/api/extract.go
@@ -32,6 +32,7 @@ import (
 )
 
 // ExtractImagesRaw returns []pdfcpu.Image containing io.Readers for images contained in selectedPages.
+// Beware of memory intensive returned slice.
 func ExtractImagesRaw(rs io.ReadSeeker, selectedPages []string, conf *pdfcpu.Configuration) ([]pdfcpu.Image, error) {
 	if rs == nil {
 		return nil, errors.New("pdfcpu: ExtractImages: Please provide rs")
@@ -40,10 +41,7 @@ func ExtractImagesRaw(rs io.ReadSeeker, selectedPages []string, conf *pdfcpu.Con
 		conf = pdfcpu.NewDefaultConfiguration()
 	}
 
-	var images []pdfcpu.Image
-
-	fromStart := time.Now()
-	ctx, durRead, durVal, durOpt, err := readValidateAndOptimize(rs, conf, fromStart)
+	ctx, _, _, _, err := readValidateAndOptimize(rs, conf, time.Now())
 	if err != nil {
 		return nil, err
 	}
@@ -52,12 +50,12 @@ func ExtractImagesRaw(rs io.ReadSeeker, selectedPages []string, conf *pdfcpu.Con
 		return nil, err
 	}
 
-	fromWrite := time.Now()
 	pages, err := PagesForPageSelection(ctx.PageCount, selectedPages, true)
 	if err != nil {
 		return nil, err
 	}
 
+	var images []pdfcpu.Image
 	for i, v := range pages {
 		if !v {
 			continue
@@ -69,11 +67,6 @@ func ExtractImagesRaw(rs io.ReadSeeker, selectedPages []string, conf *pdfcpu.Con
 		images = append(images, ii...)
 	}
 
-	durWrite := time.Since(fromWrite).Seconds()
-	durTotal := time.Since(fromStart).Seconds()
-	log.Stats.Printf("XRefTable:\n%s\n", ctx)
-	pdfcpu.TimingStats("write images", durRead, durVal, durOpt, durWrite, durTotal)
-
 	return images, nil
 }
 

diff --git a/pkg/pdfcpu/extract.go b/pkg/pdfcpu/extract.go
@@ -52,7 +52,7 @@ func (ctx *Context) StreamLength(sd *StreamDict) (int64, error) {
 		return 0, nil
 	}
 
-	i, err := ctx.DereferenceInteger(indRef)
+	i, err := ctx.DereferenceInteger(*indRef)
 	if err != nil || i == nil {
 		return 0, err
 	}
@@ -100,13 +100,13 @@ func colorSpaceNameComponents(cs Name) int {
 }
 
 // ColorSpaceComponents returns the corresponding number of used color components for sd's colorspace.
-func (ctx *Context) ColorSpaceComponents(sd *StreamDict) (int, error) {
+func (xRefTable *XRefTable) ColorSpaceComponents(sd *StreamDict) (int, error) {
 	o, found := sd.Find("ColorSpace")
 	if !found {
 		return 0, nil
 	}
 
-	o, err := ctx.Dereference(o)
+	o, err := xRefTable.Dereference(o)
 	if err != nil {
 		return 0, err
 	}
@@ -123,7 +123,7 @@ func (ctx *Context) ColorSpaceComponents(sd *StreamDict) (int, error) {
 			return 3, nil
 
 		case ICCBasedCS:
-			iccProfileStream, _, err := ctx.DereferenceStreamDict(cs[1])
+			iccProfileStream, _, err := xRefTable.DereferenceStreamDict(cs[1])
 			if err != nil {
 				return 0, err
 			}
@@ -136,7 +136,7 @@ func (ctx *Context) ColorSpaceComponents(sd *StreamDict) (int, error) {
 
 		case IndexedCS:
 
-			baseCS, err := ctx.Dereference(cs[1])
+			baseCS, err := xRefTable.Dereference(cs[1])
 			if err != nil {
 				return 0, err
 			}
@@ -146,7 +146,7 @@ func (ctx *Context) ColorSpaceComponents(sd *StreamDict) (int, error) {
 				return colorSpaceNameComponents(cs), nil
 
 			case Array:
-				iccProfileStream, _, err := ctx.DereferenceStreamDict(cs[1])
+				iccProfileStream, _, err := xRefTable.DereferenceStreamDict(cs[1])
 				if err != nil {
 					return 0, err
 				}