Support trailer info to be a dictionary instead of an indirect reference

src/UglyToad.PdfPig.Tests/Integration/DocumentInformationTests.cs

@@ -96,5 +96,23 @@
                 Assert.Equal("D:20190306232856Z00'00'", information.CreationDate);
             }
         }
+
+        [Fact]
+        public void CanReadDocumentInfromationDirectory()
+        {
+            // Issue 884
+            var path = IntegrationHelpers.GetSpecificTestDocumentPath("info_dictionary.pdf");
+
+            // Lenient Parsing On -> can process
+            using (var document = PdfDocument.Open(path))
+            {
+                var information = document.Information;
+                Assert.Equal("SumatraPDF 3.2", information.Producer);
+            }
+
+            // Lenient Parsing Off -> throws
+            var ex = Assert.Throws<PdfDocumentFormatException>(() => PdfDocument.Open(path, ParsingOptions.LenientParsingOff));
+            Assert.Equal("The info token in the trailer dictionary should only contain indirect references, instead got: <Producer, (SumatraPDF 3.2)>.", ex.Message);
+        }
     }
 }

src/UglyToad.PdfPig.Tests/Integration/SpecificTestDocuments/info_dictionary.pdf

@@ -0,0 +1,42 @@
+%PDF-1.0
+%µ¶
+
+1 0 obj
+<</Type/Catalog/Pages 2 0 R>>
+endobj
+
+2 0 obj
+<</Kids[3 0 R]/Count 1/Type/Pages/MediaBox[0 0 595 792]>>
+endobj
+
+3 0 obj
+<</Type/Page/Parent 2 0 R/Contents 4 0 R/Resources<<>>>>
+endobj
+
+4 0 obj
+<</Length 58>>
+stream
+q
+BT
+/ 96 Tf
+1 0 0 1 36 684 Tm
+(Hello World!) Tj
+ET
+Q
+
+endstream
+endobj
+
+xref
+0 5
+0000000000 65536 f 
+0000000016 00000 n 
+0000000062 00000 n 
+0000000136 00000 n 
+0000000209 00000 n 
+
+trailer
+<</Size 5/Root 1 0 R/Info <</Producer(SumatraPDF 3.2)>>>>
+startxref
+316
+%%EOF

src/UglyToad.PdfPig/CrossReference/CrossReferenceTableBuilder.cs

@@ -28,7 +28,7 @@
             parts.Add(part);
         }
 
-        public CrossReferenceTable Build(long firstCrossReferenceOffset, long offsetCorrection, ILog log)
+        public CrossReferenceTable Build(long firstCrossReferenceOffset, long offsetCorrection, bool isLenientParsing, ILog log)
         {
             CrossReferenceType type = CrossReferenceType.Table;
             DictionaryToken trailerDictionary = new DictionaryToken(new Dictionary<NameToken, IToken>());
@@ -118,7 +118,7 @@
                 }
             }
 
-            return new CrossReferenceTable(type, objectOffsets, new TrailerDictionary(trailerDictionary), 
+            return new CrossReferenceTable(type, objectOffsets, new TrailerDictionary(trailerDictionary, isLenientParsing), 
                 parts.Select(x =>
                 {
                     var prev = x.GetPreviousOffset();

src/UglyToad.PdfPig/CrossReference/TrailerDictionary.cs

@@ -35,7 +35,7 @@
         /// <summary>
         /// The object reference for the document's information dictionary if it contains one.
         /// </summary>
-        public IndirectReference? Info { get; }
+        public IToken? Info { get; }
 
         /// <summary>
         /// A list containing two-byte string tokens which act as file identifiers.
@@ -51,7 +51,8 @@
         /// Create a new <see cref="TrailerDictionary"/>.
         /// </summary>
         /// <param name="dictionary">The parsed dictionary from the document.</param>
-        internal TrailerDictionary(DictionaryToken dictionary)
+        /// <param name="isLenientParsing">Indicates if the parsing is in lenient mode</param>
+        internal TrailerDictionary(DictionaryToken dictionary, bool isLenientParsing)
         {
             if (dictionary is null)
             {
@@ -68,9 +69,13 @@
 
             Root = rootReference.Data;
 
-            if (dictionary.TryGet(NameToken.Info, out IndirectReferenceToken reference))
+            if (dictionary.TryGet(NameToken.Info, out var infoToken))
             {
-                Info = reference.Data;
+                if (!isLenientParsing && infoToken is not IndirectReferenceToken)
+                {
+                    throw new PdfDocumentFormatException($"The info token in the trailer dictionary should only contain indirect references, instead got: {infoToken}.");
+                }
+                Info = infoToken;
             }
 
             if (dictionary.TryGet(NameToken.Id, out ArrayToken arr))

src/UglyToad.PdfPig/Parser/DocumentInformationFactory.cs

@@ -17,12 +17,17 @@
         /// </summary>
         public static DocumentInformation Create(IPdfTokenScanner pdfTokenScanner, TrailerDictionary trailer, bool isLenientParsing)
         {
-            if (!trailer.Info.HasValue)
+            var token = trailer.Info;
+            if (token is IndirectReferenceToken reference)
+            {
+                token = DirectObjectFinder.Get<IToken>(reference.Data, pdfTokenScanner);
+            }
+
+            if (token == null)
             {
                 return DocumentInformation.Default;
             }
-
-            var token = DirectObjectFinder.Get<IToken>(trailer.Info.Value, pdfTokenScanner);
+            
             if (token is DictionaryToken infoParsed)
             {
                 var title = GetEntryOrDefault(infoParsed, NameToken.Title, pdfTokenScanner);

src/UglyToad.PdfPig/Parser/FileStructure/CrossReferenceParser.cs

@@ -238,7 +238,7 @@
                 throw new PdfDocumentFormatException("The cross reference was not found.");
             }
 
-            var resolved = table.Build(crossReferenceLocation, offsetCorrection, log);
+            var resolved = table.Build(crossReferenceLocation, offsetCorrection, isLenientParsing, log);
             
             // check the offsets of all referenced objects
             if (!CrossReferenceObjectOffsetValidator.ValidateCrossReferenceOffsets(bytes, resolved, log, out var actualOffsets))