Merge pull request #225 from a5i/v3-fix-combine-objects

Fix combine objects optimizations

pdf/creator/creator_test.go

@@ -2415,37 +2415,40 @@ func TestCombineDuplicateDirectObjects(t *testing.T) {
 			"mollit anim id est laborum.")
 		p.SetTextAlignment(TextAlignmentJustify)
 		p.SetMargins(0, 0, 5, 0)
-		for j := 0; j < 7; j++ {
+
+		for j := 0; j < 2; j++ {
 			subchap1.Add(p)
 		}
 
 		subchap2 := c.NewSubchapter(ch1, "Mechanism")
 		subchap2.SetMargins(0, 0, 5, 0)
-		for j := 0; j < 15; j++ {
+		for j := 0; j < 3; j++ {
 			subchap2.Add(p)
 		}
 
 		subchap3 := c.NewSubchapter(ch1, "Discussion")
 		subchap3.SetMargins(0, 0, 5, 0)
-		for j := 0; j < 19; j++ {
+		for j := 0; j < 4; j++ {
 			subchap3.Add(p)
 		}
 
 		subchap4 := c.NewSubchapter(ch1, "Conclusion")
 		subchap4.SetMargins(0, 0, 5, 0)
-		for j := 0; j < 23; j++ {
+		for j := 0; j < 3; j++ {
 			subchap4.Add(p)
 		}
-
 		c.Draw(ch1)
 
-		for i := 0; i < 50; i++ {
+		for i := 0; i < 5; i++ {
 			ch2 := c.NewChapter("References")
+			ch2.SetMargins(1, 1, 1, 1)
 			for j := 0; j < 13; j++ {
 				ch2.Add(p)
 			}
-
+			metadata := core.MakeDict()
+			metadata.Set(core.PdfObjectName("TEST"), core.MakeString("---------------- ## ----------------"))
 			c.Draw(ch2)
+			c.getActivePage().Metadata = metadata
 		}
 
 		// Set a function to create the front Page.

pdf/model/optimize/combine_duplicate_direct_objects.go

@@ -18,17 +18,16 @@ type CombineDuplicateDirectObjects struct {
 
 // Optimize optimizes PDF objects to decrease PDF size.
 func (dup *CombineDuplicateDirectObjects) Optimize(objects []core.PdfObject) (optimizedObjects []core.PdfObject, err error) {
-
+	updateObjectNumbers(objects)
 	dictsByHash := make(map[string][]*core.PdfObjectDictionary)
 	var processDict func(pDict *core.PdfObjectDictionary)
-	processDict = func(pDict *core.PdfObjectDictionary) {
 
+	processDict = func(pDict *core.PdfObjectDictionary) {
 		for _, key := range pDict.Keys() {
 			obj := pDict.Get(key)
 			if dict, isDictObj := obj.(*core.PdfObjectDictionary); isDictObj {
 				hasher := md5.New()
 				hasher.Write([]byte(dict.DefaultWriteString()))
-
 				hash := string(hasher.Sum(nil))
 				dictsByHash[hash] = append(dictsByHash[hash], dict)
 				processDict(dict)

pdf/model/optimize/combine_identical_indirect_objects.go

@@ -18,6 +18,7 @@ type CombineIdenticalIndirectObjects struct {
 
 // Optimize optimizes PDF objects to decrease PDF size.
 func (c *CombineIdenticalIndirectObjects) Optimize(objects []core.PdfObject) (optimizedObjects []core.PdfObject, err error) {
+	updateObjectNumbers(objects)
 	replaceTable := make(map[core.PdfObject]core.PdfObject)
 	toDelete := make(map[core.PdfObject]struct{})
 

pdf/model/optimize/optimizer.go

@@ -82,3 +82,21 @@ func replaceObjectsInPlace(objects []core.PdfObject, objTo map[core.PdfObject]co
 		}
 	}
 }
+
+// Update all the object numbers prior to get hash of objects.
+func updateObjectNumbers(objects []core.PdfObject) {
+	// Update numbers
+	for idx, obj := range objects {
+		switch o := obj.(type) {
+		case *core.PdfIndirectObject:
+			o.ObjectNumber = int64(idx + 1)
+			o.GenerationNumber = 0
+		case *core.PdfObjectStream:
+			o.ObjectNumber = int64(idx + 1)
+			o.GenerationNumber = 0
+		case *core.PdfObjectStreams:
+			o.ObjectNumber = int64(idx + 1)
+			o.GenerationNumber = 0
+		}
+	}
+}

pdf/model/writer.go

@@ -636,17 +636,16 @@ func (this *PdfWriter) writeObject(num int, obj PdfObject) {
 func (this *PdfWriter) updateObjectNumbers() {
 	// Update numbers
 	for idx, obj := range this.objects {
-		if io, isIndirect := obj.(*PdfIndirectObject); isIndirect {
-			io.ObjectNumber = int64(idx + 1)
-			io.GenerationNumber = 0
-		}
-		if so, isStream := obj.(*PdfObjectStream); isStream {
-			so.ObjectNumber = int64(idx + 1)
-			so.GenerationNumber = 0
-		}
-		if so, isObjectStreams := obj.(*PdfObjectStreams); isObjectStreams {
-			so.ObjectNumber = int64(idx + 1)
-			so.GenerationNumber = 0
+		switch o := obj.(type) {
+		case *PdfIndirectObject:
+			o.ObjectNumber = int64(idx + 1)
+			o.GenerationNumber = 0
+		case *PdfObjectStream:
+			o.ObjectNumber = int64(idx + 1)
+			o.GenerationNumber = 0
+		case *PdfObjectStreams:
+			o.ObjectNumber = int64(idx + 1)
+			o.GenerationNumber = 0
 		}
 	}
 }