Merge pull request #259 from plaisted/master

PdfMerger: Fix indirect object lookup dict

src/UglyToad.PdfPig.Tests/Writer/PdfMergerTests.cs

@@ -3,6 +3,7 @@
     using Integration;
     using PdfPig.Writer;
     using System;
+    using System.Collections.Generic;
     using System.IO;
     using Xunit;
 
@@ -101,6 +102,21 @@
             }
         }
 
+        [Fact]
+        public void DedupsObjectsFromSameDoc()
+        {
+            var one = IntegrationHelpers.GetDocumentPath("Multiple Page - from Mortality Statistics.pdf");
+
+            var result = PdfMerger.Merge(new List<byte[]> { File.ReadAllBytes(one) }, new List<IReadOnlyList<int>> { new List<int> { 1, 2}  });
+
+            using (var document = PdfDocument.Open(result, ParsingOptions.LenientParsingOff))
+            {
+                Assert.Equal(2, document.NumberOfPages);
+                Assert.True(document.Structure.CrossReferenceTable.ObjectOffsets.Count <= 29,
+                    "Expected object count to be lower than 30"); // 45 objects with duplicates, 29 with correct re-use
+            }
+        }
+
         [Fact]
         public void CanMergeWithObjectStream()
         {

src/UglyToad.PdfPig/Writer/PdfMerger.cs

@@ -242,7 +242,7 @@
 
                 currentVersion = Math.Max(version, currentVersion);
 
-                var referencesFromDocument = new Dictionary<IndirectReferenceToken, IndirectReferenceToken>();
+                var referencesFromDocument = new Dictionary<IndirectReference, IndirectReferenceToken>();
 
                 var currentNodeReference = context.ReserveNumberToken();
                 var pagesReferences = new List<IndirectReferenceToken>();
@@ -387,7 +387,7 @@
             }
 
             private IndirectReferenceToken CopyPageNode(PageTreeNode pageNode, IndirectReferenceToken parentPagesObject, IPdfTokenScanner tokenScanner, 
-                IDictionary<IndirectReferenceToken, IndirectReferenceToken> referencesFromDocument)
+                IDictionary<IndirectReference, IndirectReferenceToken> referencesFromDocument)
             {
                 Debug.Assert(pageNode.IsPage);
 
@@ -421,7 +421,7 @@
             /// <param name="tokenScanner">scanner get the content from the original document</param>
             /// <param name="referencesFromDocument">Map of previously copied</param>
             /// <returns>A reference of the token that was copied. With all the reference updated</returns>
-            private IToken CopyToken(IToken tokenToCopy, IPdfTokenScanner tokenScanner, IDictionary<IndirectReferenceToken, IndirectReferenceToken> referencesFromDocument)
+            private IToken CopyToken(IToken tokenToCopy, IPdfTokenScanner tokenScanner, IDictionary<IndirectReference, IndirectReferenceToken> referencesFromDocument)
             {
                 // This token need to be deep copied, because they could contain reference. So we have to update them.
                 switch (tokenToCopy)
@@ -450,14 +450,14 @@
                         }
                     case IndirectReferenceToken referenceToken:
                         {
-                            if (referencesFromDocument.TryGetValue(referenceToken, out var newReferenceToken))
+                            if (referencesFromDocument.TryGetValue(referenceToken.Data, out var newReferenceToken))
                             {
                                 return newReferenceToken;
                             }
 
                             //we add the token to referencesFromDocument to prevent stackoverflow on references cycles 
                             newReferenceToken = context.ReserveNumberToken();
-                            referencesFromDocument.Add(referenceToken, newReferenceToken);
+                            referencesFromDocument.Add(referenceToken.Data, newReferenceToken);
 
                             var tokenObject = DirectObjectFinder.Get<IToken>(referenceToken.Data, tokenScanner);
                             Debug.Assert(!(tokenObject is IndirectReferenceToken));