Improve reading corrupted files

pdfcpu · Jun 15, 2024 · a467f3c · a467f3c
1 parent c342327
commit a467f3c
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 11 deletions.
diff --git a/pkg/pdfcpu/extract.go b/pkg/pdfcpu/extract.go
@@ -224,15 +224,25 @@ func imageStub(
 	thumb, imgMask bool,
 	objNr int) (*model.Image, error) {
 
-	w := sd.IntEntry("Width")
-	if w == nil {
+	obj, ok := sd.Find("Width")
+	if !ok {
 		return nil, errors.Errorf("pdfcpu: missing image width obj#%d", objNr)
 	}
+	i, err := ctx.DereferenceInteger(obj)
+	if err != nil {
+		return nil, err
+	}
+	w := i.Value()
 
-	h := sd.IntEntry("Height")
-	if h == nil {
+	obj, ok = sd.Find("Height")
+	if !ok {
 		return nil, errors.Errorf("pdfcpu: missing image height obj#%d", objNr)
 	}
+	i, err = ctx.DereferenceInteger(obj)
+	if err != nil {
+		return nil, err
+	}
+	h := i.Value()
 
 	cs, err := ColorSpaceString(ctx, sd)
 	if err != nil {
@@ -271,7 +281,7 @@ func imageStub(
 		interpol = true
 	}
 
-	i, err := StreamLength(ctx, sd)
+	size, err := StreamLength(ctx, sd)
 	if err != nil {
 		return nil, err
 	}
@@ -288,13 +298,13 @@ func imageStub(
 		IsImgMask:   imgMask,
 		HasImgMask:  mask,
 		HasSMask:    sMask,
-		Width:       *w,
-		Height:      *h,
+		Width:       w,
+		Height:      h,
 		Cs:          cs,
 		Comp:        comp,
 		Bpc:         bpc,
 		Interpol:    interpol,
-		Size:        i,
+		Size:        size,
 		Filter:      filters,
 		DecodeParms: s,
 	}

diff --git a/pkg/pdfcpu/read.go b/pkg/pdfcpu/read.go
@@ -1340,7 +1340,10 @@ func parseAndLoad(c context.Context, ctx *model.Context, line string, offset *in
 		Offset:     offset,
 		Generation: generation}
 
-	ctx.Table[*objNr] = &entry
+	if !ctx.XRefTable.Exists(*objNr) {
+		ctx.Table[*objNr] = &entry
+	}
+
 	o, err := ParseObjectWithContext(c, ctx, *entry.Offset, *objNr, *entry.Generation)
 	if err != nil {
 		return err

diff --git a/pkg/pdfcpu/writeImage.go b/pkg/pdfcpu/writeImage.go
@@ -91,8 +91,25 @@ func pdfImage(xRefTable *model.XRefTable, sd *types.StreamDict, thumb bool, objN
 
 	bpc := *sd.IntEntry("BitsPerComponent")
 
-	w := *sd.IntEntry("Width")
-	h := *sd.IntEntry("Height")
+	obj, ok := sd.Find("Width")
+	if !ok {
+		return nil, errors.Errorf("pdfcpu: missing image width obj#%d", objNr)
+	}
+	i, err := xRefTable.DereferenceInteger(obj)
+	if err != nil {
+		return nil, err
+	}
+	w := i.Value()
+
+	obj, ok = sd.Find("Height")
+	if !ok {
+		return nil, errors.Errorf("pdfcpu: missing image height obj#%d", objNr)
+	}
+	i, err = xRefTable.DereferenceInteger(obj)
+	if err != nil {
+		return nil, err
+	}
+	h := i.Value()
 
 	decode := decodeArr(sd.ArrayEntry("Decode"))