Change defaults for PdfMerger; add flags to PdfSplitter to disable tags and outlines copying

Now by default tags and outlines in PdfMerger are copied even if destination document is not tagged and has no outlines.

Author: yulian-gaponenko

kernel/src/main/java/com/itextpdf/kernel/pdf/PdfCatalog.java

@@ -149,15 +149,6 @@ public PdfCatalog setAdditionalAction(PdfName key, PdfAction action) {
         return this;
     }
 
-    /**
-     * This flag determines if Outline tree of the document has been built via calling getOutlines method. If this flag is false all outline operations will be ignored
-     *
-     * @return state of outline mode.
-     */
-    public boolean isOutlineMode() {
-        return outlineMode;
-    }
-
     /**
      * This method sets a page mode of the document.
      * </p>
@@ -387,6 +378,15 @@ boolean hasOutlines() {
         return getPdfObject().containsKey(PdfName.Outlines);
     }
 
+    /**
+     * This flag determines if Outline tree of the document has been built via calling getOutlines method. If this flag is false all outline operations will be ignored
+     *
+     * @return state of outline mode.
+     */
+    boolean isOutlineMode() {
+        return outlineMode;
+    }
+
     /**
      * This method removes all outlines associated with a given page
      *

kernel/src/main/java/com/itextpdf/kernel/pdf/PdfDocument.java

@@ -1023,7 +1023,7 @@ public List<PdfPage> copyPagesTo(List<Integer> pagesToCopy, PdfDocument toDocume
                 toDocument.addPage(newPage);
             }
             pageInsertIndex++;
-            if (toDocument.getCatalog().isOutlineMode()) {
+            if (toDocument.hasOutlines()) {
                 List<PdfOutline> pageOutlines = page.getOutlines(false);
                 if (pageOutlines != null)
                     outlinesToCopy.addAll(pageOutlines);

kernel/src/main/java/com/itextpdf/kernel/utils/PdfMerger.java

@@ -54,18 +54,38 @@ public class PdfMerger {
 
     private PdfDocument pdfDocument;
     private boolean closeSrcDocuments;
+    private boolean mergeTags;
+    private boolean mergeOutlines;
 
     /**
-     * This class is used to merge a number of existing documents into one;
-     * @param pdfDocument - the document into which source documents will be merged.
+     * This class is used to merge a number of existing documents into one. By default, if source document
+     * contains tags and outlines, they will be also copied to the destination document.
+     * @param pdfDocument the document into which source documents will be merged.
      */
     public PdfMerger(PdfDocument pdfDocument) {
+        this(pdfDocument, true, true);
+    }
+
+    /**
+     * This class is used to merge a number of existing documents into one.
+     * @param pdfDocument the document into which source documents will be merged.
+     * @param mergeTags if true, then tags from the source document are copied even if destination document is not set as
+     *                  tagged. Note, that if false, tag structure is still could be copied if the destination document
+     *                  is explicitly marked as tagged with {@link PdfDocument#setTagged()}.
+     * @param mergeOutlines if true, then outlines from the source document are copied even if in destination document
+     *                      outlines are not initialized. Note, that if false, outlines are still could be copied if the
+     *                      destination document outlines were explicitly initialized with {@link PdfDocument#initializeOutlines()}.
+     */
+    public PdfMerger(PdfDocument pdfDocument, boolean mergeTags, boolean mergeOutlines) {
         this.pdfDocument = pdfDocument;
+        this.mergeTags = mergeTags;
+        this.mergeOutlines = mergeOutlines;
     }
 
     /**
-     * If set to <i>true</i> then passed to the <i>{@code PdfMerger#merge}</i> method source documents will be closed immediately after merging
-     * specified pages into current document. If <i>false</i> - PdfDocuments are left open. Default value - <i>false</i>.
+     * If set to <i>true</i> then passed to the <i>{@code PdfMerger#merge}</i> method source documents will be closed
+     * immediately after merging specified pages into current document. If <i>false</i> - PdfDocuments are left open.
+     * Default value - <i>false</i>.
      * @param closeSourceDocuments should be true to close pdf documents in merge method.
      * @return this {@code PdfMerger} instance.
      */
@@ -102,6 +122,13 @@ public PdfMerger merge(PdfDocument from, int fromPage, int toPage) {
      * @return this {@code PdfMerger} instance.
      */
     public PdfMerger merge(PdfDocument from, List<Integer> pages) {
+        if (mergeTags && from.isTagged()) {
+            pdfDocument.setTagged();
+        }
+        if (mergeOutlines && from.hasOutlines()) {
+            pdfDocument.initializeOutlines();
+        }
+
         from.copyPagesTo(pages, pdfDocument);
         if (closeSrcDocuments) {
             from.close();

kernel/src/main/java/com/itextpdf/kernel/utils/PdfSplitter.java

@@ -57,13 +57,12 @@ This file is part of the iText (R) project.
 import java.util.Collections;
 import java.util.List;
 import java.util.Map;
-import java.util.TreeSet;
-import java.util.regex.Matcher;
-import java.util.regex.Pattern;
 
 public class PdfSplitter {
 
     private PdfDocument pdfDocument;
+    private boolean preserveTagged;
+    private boolean preserveOutlines;
 
     /**
      * Creates a new instance of PdfSplitter class.
@@ -75,6 +74,26 @@ public PdfSplitter(PdfDocument pdfDocument) {
             throw new PdfException(PdfException.CannotSplitDocumentThatIsBeingWritten);
         }
         this.pdfDocument = pdfDocument;
+        this.preserveTagged = true;
+        this.preserveOutlines = true;
+    }
+
+    /**
+     * If original document is tagged, then by default all resultant document will also be tagged.
+     * This could be changed with this flag - if set to false, resultant documents will be not tagged, even if
+     * original document is tagged.
+     */
+    public void setPreserveTagged(boolean preserveTagged) {
+        this.preserveTagged = preserveTagged;
+    }
+
+    /**
+     * If original document has outlines, then by default all resultant document will also have outlines.
+     * This could be changed with this flag - if set to false, resultant documents won't contain outlines, even if
+     * original document had them.
+     */
+    public void setPreserveOutlines(boolean preserveOutlines) {
+        this.preserveOutlines = preserveOutlines;
     }
 
     /**
@@ -234,9 +253,9 @@ protected PdfWriter getNextPdfWriter(PageRange documentPageRange) {
 
     private PdfDocument createPdfDocument(PageRange currentPageRange) {
         PdfDocument newDocument = new PdfDocument(getNextPdfWriter(currentPageRange));
-        if (pdfDocument.isTagged())
+        if (pdfDocument.isTagged() && preserveTagged)
             newDocument.setTagged();
-        if (pdfDocument.getCatalog().isOutlineMode())
+        if (pdfDocument.hasOutlines() && preserveOutlines)
             newDocument.initializeOutlines();
         return newDocument;
     }

kernel/src/test/java/com/itextpdf/kernel/utils/PdfSplitterTest.java

@@ -39,7 +39,6 @@ public static void beforeClass() {
     public void splitDocumentTest01() throws IOException, InterruptedException {
         String inputFileName =  sourceFolder + "iphone_user_guide.pdf";
         PdfDocument inputPdfDoc = new PdfDocument(new PdfReader(inputFileName));
-        inputPdfDoc.initializeOutlines();
 
         List<Integer> pageNumbers = Arrays.asList(30, 100);
 
@@ -72,7 +71,6 @@ protected PdfWriter getNextPdfWriter(PageRange documentPageRange) {
     public void splitDocumentTest02() throws IOException, InterruptedException {
         String inputFileName =  sourceFolder + "iphone_user_guide.pdf";
         PdfDocument inputPdfDoc = new PdfDocument(new PdfReader(inputFileName));
-        inputPdfDoc.initializeOutlines();
 
         new PdfSplitter(inputPdfDoc) {
             int partNumber = 1;
@@ -113,7 +111,6 @@ public void documentReady(PdfDocument pdfDocument, PageRange pageRange) {
     public void splitDocumentTest03() throws IOException, InterruptedException {
         String inputFileName =  sourceFolder + "iphone_user_guide.pdf";
         PdfDocument inputPdfDoc = new PdfDocument(new PdfReader(inputFileName));
-        inputPdfDoc.initializeOutlines();
 
         PageRange pageRange1 = new PageRange().addPageSequence(4, 15).addSinglePage(18).addPageSequence(1, 2);
         PageRange pageRange2 = new PageRange().addSinglePage(99).addSinglePage(98).addPageSequence(70, 99);
@@ -149,7 +146,6 @@ public void splitDocumentByOutlineTest() throws IOException, InterruptedExceptio
 
         String inputFileName =  sourceFolder + "iphone_user_guide.pdf";
         PdfDocument inputPdfDoc = new PdfDocument(new PdfReader(inputFileName));
-        inputPdfDoc.initializeOutlines();
         PdfSplitter splitter = new PdfSplitter(inputPdfDoc);
         List listTitles = new ArrayList();
         listTitles.add("Syncing iPod Content from Your iTunes Library");
@@ -165,7 +161,6 @@ public void splitDocumentByOutlineTest() throws IOException, InterruptedExceptio
     public void splitDocumentBySize() throws IOException, InterruptedException {
         String inputFileName = sourceFolder + "splitBySize.pdf";
         PdfDocument inputPdfDoc = new PdfDocument(new PdfReader(inputFileName));
-        inputPdfDoc.initializeOutlines();
         PdfSplitter splitter = new PdfSplitter(inputPdfDoc) {
 
             int partNumber = 1;