PDFBOX-5929: remove orphan annotations from the structure tree if annotations were removed from the page

git-svn-id: https://svn.apache.org/repos/asf/pdfbox/trunk@1922671 13f79535-47bb-0310-9956-ffa450edef68

Author: THausherr

pdfbox/src/main/java/org/apache/pdfbox/multipdf/Splitter.java

@@ -454,6 +454,10 @@ private COSBase createDictionaryClone(COSBase src, COSBase dstParent, COSDiction
                     // replace annotation with clone
                     dstDict.setItem(COSName.OBJ, dstObj);
                 }
+                else
+                {
+                    removePossibleOrphanAnnotation(srcObj, srcDict, currentPageDict, dstDict);
+                }
             }
             else
             {
@@ -492,6 +496,35 @@ private COSBase createDictionaryClone(COSBase src, COSBase dstParent, COSDiction
             }
             return dstDict;
         }
+
+        private void removePossibleOrphanAnnotation(COSDictionary srcObj, COSDictionary srcDict,
+                COSDictionary currentPageDict, COSDictionary dstDict)
+        {
+            // PDFBOX-5929: Check whether this is an "orphan" annotation that isn't in the page
+            COSBase objType = srcObj.getDictionaryObject(COSName.TYPE);
+            COSBase objSubtype = srcObj.getDictionaryObject(COSName.SUBTYPE);
+            if (COSName.ANNOT.equals(objType) || COSName.LINK.equals(objSubtype))
+            {
+                COSDictionary srcPageDict = srcDict.getCOSDictionary(COSName.PG);
+                if (srcPageDict == null)
+                {
+                    // /Pg entry is not always on this level
+                    srcPageDict = currentPageDict;
+                }
+                if (srcPageDict != null)
+                {
+                    COSArray annotationArray = srcPageDict.getCOSArray(COSName.ANNOTS);
+                    if (annotationArray == null || annotationArray.indexOfObject(srcObj) == -1)
+                    {
+                        // Ideally the entire OBJR entry should be removed.
+                        // Removing the OBJ entry is done to avoid potential page orphans
+                        // from the annotation destination.
+                        LOG.warn("An annotation OBJ that isn't in the page has been removed from the structure tree");
+                        dstDict.removeItem(COSName.OBJ);
+                    }
+                }
+            }
+        }
     }
 
     // Look for /StructParent and /StructParents and add them to the destination tree

pdfbox/src/test/java/org/apache/pdfbox/multipdf/PDFMergerUtilityTest.java

@@ -924,6 +924,7 @@ else if (kdict.containsKey(COSName.NUMS))
                     PDPage page = annotation.getPage();
                     if (annotation instanceof PDAnnotationLink)
                     {
+                        // PDFBOX-5928: check whether the destination of a link annotation is an orphan
                         PDAnnotationLink link = (PDAnnotationLink) annotation;
                         PDDestination destination = link.getDestination();
                         if (destination == null)
@@ -1130,6 +1131,35 @@ void testSplitWithStructureTreeAndDestinations() throws IOException
         }
     }
 
+    /**
+     * PDFBOX-5929: Check that orphan annotations are removed from the structure tree if annotations
+     * were removed from the pages (don't do that!).
+     *
+     * @throws IOException
+     */
+    @Test
+    void testSplitWithStructureTreeAndDestinationsAndRemovedAnnotations() throws IOException
+    {
+        try (PDDocument doc = Loader.loadPDF(new File(SRCDIR,"PDFBOX-5762-722238.pdf")))
+        {
+            Splitter splitter = new Splitter();
+            for (PDPage page : doc.getPages())
+            {
+                page.setAnnotations(Collections.emptyList());
+            }
+            splitter.setStartPage(1);
+            splitter.setEndPage(2);
+            splitter.setSplitAtPage(2);
+            List<PDDocument> splitResult = splitter.split(doc);
+            assertEquals(1, splitResult.size());
+            try (PDDocument dstDoc = splitResult.get(0))
+            {
+                assertEquals(2, dstDoc.getNumberOfPages());
+                checkForPageOrphans(dstDoc);
+            }
+        }
+    }
+
     /**
      * Check for the bug that happened in PDFBOX-5792, where a destination was outside a target
      * document and hit an NPE in the next call of Splitter.fixDestinations().