Fix #930

pdfcpu · Aug 20, 2024 · b9c28ae · b9c28ae
1 parent 3aff1b0
commit b9c28ae
Showing 1 changed file with 79 additions and 8 deletions.
diff --git a/pkg/pdfcpu/optimize.go b/pkg/pdfcpu/optimize.go
@@ -63,22 +63,93 @@ func optimizeContentStreamUsage(ctx *model.Context, sd *types.StreamDict, objNr
 	return nil, nil
 }
 
+func removeEmptyContentStreams(ctx *model.Context, pageDict types.Dict, obj types.Object, pageObjNumber int) error {
+	var contentArr types.Array
+
+	if ir, ok := obj.(types.IndirectRef); ok {
+
+		objNr := ir.ObjectNumber.Value()
+		entry, found := ctx.FindTableEntry(objNr, ir.GenerationNumber.Value())
+		if !found {
+			return errors.Errorf("removeEmptyContentStreams: obj#:%d illegal indRef for Contents\n", pageObjNumber)
+		}
+
+		contentStreamDict, ok := entry.Object.(types.StreamDict)
+		if ok {
+			if err := contentStreamDict.Decode(); err != nil {
+				return err
+			}
+			if len(contentStreamDict.Content) == 0 {
+				pageDict.Delete("Contents")
+			}
+			return nil
+		}
+
+		contentArr, ok = entry.Object.(types.Array)
+		if !ok {
+			return errors.Errorf("removeEmptyContentStreams: obj#:%d page content entry neither stream dict nor array.\n", pageObjNumber)
+		}
+
+	} else if contentArr, ok = obj.(types.Array); !ok {
+		return errors.Errorf("removeEmptyContentStreams: obj#:%d corrupt page content array\n", pageObjNumber)
+	}
+
+	var newContentArr types.Array
+
+	for _, c := range contentArr {
+
+		ir, ok := c.(types.IndirectRef)
+		if !ok {
+			return errors.Errorf("removeEmptyContentStreams: obj#:%d corrupt page content array entry\n", pageObjNumber)
+		}
+
+		objNr := ir.ObjectNumber.Value()
+		entry, found := ctx.FindTableEntry(objNr, ir.GenerationNumber.Value())
+		if !found {
+			return errors.Errorf("removeEmptyContentStreams: obj#:%d illegal indRef for Contents\n", pageObjNumber)
+		}
+
+		contentStreamDict, ok := entry.Object.(types.StreamDict)
+		if !ok {
+			return errors.Errorf("identifyPageContent: obj#:%d page content entry is no stream dict\n", pageObjNumber)
+		}
+
+		if err := contentStreamDict.Decode(); err != nil {
+			return err
+		}
+		if len(contentStreamDict.Content) > 0 {
+			newContentArr = append(newContentArr, c)
+		}
+	}
+
+	pageDict["Contents"] = newContentArr
+
+	return nil
+}
+
 func optimizePageContent(ctx *model.Context, pageDict types.Dict, pageObjNumber int) error {
-	if !ctx.OptimizeDuplicateContentStreams {
+	o, found := pageDict.Find("Contents")
+	if !found {
 		return nil
 	}
-	if log.OptimizeEnabled() {
-		log.Optimize.Println("identifyPageContent begin")
+
+	if err := removeEmptyContentStreams(ctx, pageDict, o, pageObjNumber); err != nil {
+		return err
 	}
 
-	o, found := pageDict.Find("Contents")
+	o, found = pageDict.Find("Contents")
 	if !found {
-		if log.OptimizeEnabled() {
-			log.Optimize.Println("identifyPageContent end: no \"Contents\"")
-		}
 		return nil
 	}
 
+	if !ctx.OptimizeDuplicateContentStreams {
+		return nil
+	}
+
+	if log.OptimizeEnabled() {
+		log.Optimize.Println("identifyPageContent begin")
+	}
+
 	var contentArr types.Array
 
 	if ir, ok := o.(types.IndirectRef); ok {
@@ -685,7 +756,7 @@ func parseResourcesDict(ctx *model.Context, pageDict types.Dict, pageNumber, pag
 	return nil
 }
 
-// Iterate over all pages and optimize resources.
+// Iterate over all pages and optimize content & resources.
 func parsePagesDict(ctx *model.Context, pagesDict types.Dict, pageNumber int) (int, error) {
 	// TODO Integrate resource consolidation based on content stream requirements.