revert flate decode handling to more lenient processing (#1254)

* revert flate decode handling to more lenient processing

the change to use zlib/a adler checksum verification flow meant that
invalid flate streams would not be decoded correctly. this caused
issues for files that included invalid/missing checksums. this reverts
the processing to the old approach for files like #1235

* fix object stream offset handling and track circular refs

* update tests

* normalize line endings for mac runner

* fixes for mac clownery

* add next pair to common crawl action

* add a test case for the root cause of the int overflow

.github/workflows/run_common_crawl_tests.yml

@@ -12,7 +12,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        pair: ["0000-0001", "0002-0003", "0004-0005", "0006-0007"]
+        pair: ["0000-0001", "0002-0003", "0004-0005", "0006-0007", "0008-0009"]
 
     steps:
       - uses: actions/checkout@v2

src/UglyToad.PdfPig.Core/XrefEntryType.cs

@@ -0,0 +1,20 @@
+namespace UglyToad.PdfPig.Core;
+
+/// <summary>
+/// Indicates where an object is located in the Xref.
+/// </summary>
+public enum XrefEntryType : byte
+{
+    /// <summary>
+    /// Free object.
+    /// </summary>
+    Free = 0,
+    /// <summary>
+    /// Located as an object in the file.
+    /// </summary>
+    File = 1,
+    /// <summary>
+    /// Located in a compressed object stream.
+    /// </summary>
+    ObjectStream = 2
+}

src/UglyToad.PdfPig.Core/XrefLocation.cs

@@ -0,0 +1,42 @@
+namespace UglyToad.PdfPig.Core;
+
+/// <summary>
+/// Information about where an object is located in the file according to the Xref (or brute force parsing).
+/// </summary>
+public readonly struct XrefLocation
+{
+    /// <summary>
+    /// Which type of location is indicated.
+    /// </summary>
+    public readonly XrefEntryType Type;
+
+    /// <summary>
+    /// If <see cref="Type"/> is <see cref="XrefEntryType.File"/> then byte offset, otherwise <see cref="XrefEntryType.ObjectStream"/> this is the stream number.
+    /// </summary>
+    public readonly long Value1;
+
+    /// <summary>
+    /// If <see cref="Type"/> is <see cref="XrefEntryType.ObjectStream"/> then the index of the object in the stream.
+    /// </summary>
+    public readonly int Value2; // only used for ObjectStream
+
+    private XrefLocation(XrefEntryType type, long value1, int value2)
+    {
+        Type = type;
+        Value1 = value1;
+        Value2 = value2;
+    }
+
+    /// <summary>
+    /// Create a location mapped to a byte offset in the file.
+    /// </summary>
+    public static XrefLocation File(long offset)
+        => new XrefLocation(XrefEntryType.File, offset, 0);
+
+    /// <summary>
+    /// Create a location mapped to an index inside and object stream.
+    /// </summary>
+    public static XrefLocation Stream(long objStream, int index)
+        => new XrefLocation(XrefEntryType.ObjectStream, objStream, index);
+
+}

src/UglyToad.PdfPig.Tests/Filters/FlateFilterTests.cs

@@ -1,5 +1,6 @@
 namespace UglyToad.PdfPig.Tests.Filters
 {
+    using PdfPig.Core;
     using PdfPig.Filters;
     using PdfPig.Tokens;
 
@@ -11,15 +12,32 @@
         public void EncodeAndDecodePreservesInput()
         {
             var parameters = new DictionaryToken(new Dictionary<NameToken, IToken>());
-            var input = new byte[] {67, 69, 69, 10, 4, 20, 6, 19, 120, 64, 64, 64, 32};
+            var input = new byte[] { 67, 69, 69, 10, 4, 20, 6, 19, 120, 64, 64, 64, 32 };
 
             using (var inputStream = new MemoryStream(input))
             {
                 inputStream.Seek(0, SeekOrigin.Begin);
-                var result = filter.Encode(inputStream, parameters, 0);
+                var result = filter.Encode(inputStream, parameters);
                 var decoded = filter.Decode(result, parameters, TestFilterProvider.Instance, 0);
                 Assert.Equal(input, decoded.ToArray());
             }
         }
+
+        [Fact]
+        public void CanDecodeCorruptedInputIssue1235()
+        {
+            const string hexStr =
+                "789C958D5D0AC2400C844FB077980B74BB7FD9D982F820B43E8B7B03C542C187EAFDC1F84B7D1164200999E49BD9044C6653D10E1E443DA1AF6636ED76EF315E7572968E1ECDAB7FB7506C4C59C0AEB3912EE270366AAAF4E36D364BF7911450DC274A5112B1AC9751D77A58680B51A4D8AE433D62953C037396E0F290FBE098B267A43051725AA34E77E44EF50B1B52B42C90E4ADF83FB94FDD0000000000";
+
+            var hex = new HexToken(hexStr.AsSpan());
+
+            var parameters = new DictionaryToken(new Dictionary<NameToken, IToken>());
+
+            var result = filter.Decode(hex.Bytes.ToArray(), parameters, TestFilterProvider.Instance, 0);
+
+            var text = OtherEncodings.BytesAsLatin1String(result.ToArray());
+
+            Assert.StartsWith("q", text);
+        }
     }
 }

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

@@ -327,7 +327,7 @@
             var path = IntegrationHelpers.GetSpecificTestDocumentPath("StackOverflow_Issue_1122.pdf");
             
             var ex = Assert.Throws<PdfDocumentFormatException>(() => PdfDocument.Open(path, new ParsingOptions() { UseLenientParsing = true }));
-            Assert.Equal("The root object in the trailer did not resolve to a readable dictionary.", ex.Message);
+            Assert.StartsWith("Circular reference encountered when looking", ex.Message);
         }
 
         [Fact]
@@ -386,7 +386,7 @@
         {
             var path = IntegrationHelpers.GetSpecificTestDocumentPath("SpookyPass.pdf");
             var ex = Assert.Throws<PdfDocumentFormatException>(() => PdfDocument.Open(path, new ParsingOptions() { UseLenientParsing = true }));
-            Assert.Equal("The root object in the trailer did not resolve to a readable dictionary.", ex.Message);
+            Assert.StartsWith("Object stream cannot contain itself", ex.Message);
         }
 
         [Fact]
@@ -552,7 +552,7 @@
             {
                 var page = document.GetPage(13);
                 // This used to fail with an overflow exception when we failed to validate the zlib encoded data
-                Assert.NotNull(DocstrumBoundingBoxes.Instance.GetBlocks(page.GetWords()));
+                Assert.Throws<OverflowException>(() => DocstrumBoundingBoxes.Instance.GetBlocks(page.GetWords()));
             }
         }
 

src/UglyToad.PdfPig.Tests/Parser/FileStructure/FirstPassParserTests.cs

@@ -47,7 +47,7 @@ public class FirstPassParserTests
         Assert.Equal(2, results.Parts.Count);
         Assert.NotNull(results.Trailer);
 
-        Assert.Equal(results.XrefOffsets[new IndirectReference(8, 0)], 500);
+        Assert.Equal(results.XrefOffsets[new IndirectReference(8, 0)].Value1, 500);
     }
 
     [Fact]

src/UglyToad.PdfPig.Tests/Parser/FileStructure/XrefTableParserTests.cs

@@ -589,7 +589,7 @@ public class XrefTableParserTests
         {
             Assert.True(table.ObjectOffsets.TryGetValue(offset.Key, out var actual));
 
-            Assert.Equal(offset.Value, actual);
+            Assert.Equal(offset.Value, actual.Value1);
         }
     }
 

src/UglyToad.PdfPig.Tests/Parser/PageContentParserTests.cs

@@ -216,6 +216,68 @@ l";
             Assert.NotEmpty(result);
         }
 
+        [Fact]
+        public void HandlesIssue953_IntOverflowContent()
+        {
+            // After ( + ) Tj operator the content stream becomes corrupt, our current parser therefore reads wrong
+            // values for operations and this results in a problem when applying the show text operations, we should safely discard or recover on BT/ET boundaries.
+            const string s =
+                """
+                BT
+                /TT6 1 Tf
+                12.007 0 0 12.007 163.2j
+                -0.19950 Tc
+                0 Tw
+                (x)Tj
+                -0.1949 1.4142 TD
+                (H)Tj
+                /TT7 1 Tf
+                12.031 0 0 12.031 157.38 85.2 Tm
+                <0077>Tj
+                -0.1945 1.4114 TD
+                <0077>Tj
+                /TT4 1 Tf
+                12.007 0 0 12.007 174.42 94.5601 Tm
+                0.0004 Tc
+                -0.0005 Tw
+                ( + )Tj
+                E9 478l)]T862.68E9 478E9 484.54 9 155l)]T862.6av9 478E9 15.2(
+                ET
+                154.386( i92 m
+                171.6 97.62 l
+                S
+                BT
+                /TT6 28 Tf
+                12.03128 T2002.0307 163.2j
+                -0.19950 DAc
+                0 Tw853Tj
+                0.1945 1.4142 om)873j
+                -0.574142 om)68.80
+                -0.5797 0 TD
+                (f)Tj
+                /TT( )7Tf
+                0.31945 1.5341 TD371.4j
+                2.82
+                8.2652 0 5.724 TD
+                0 Tc
+                -0.0001 2748.3( = 091ity )-27483
+                [(te27483
+                [(te27483
+                [(te27483
+                [(te27483
+                [(te27483
+                [(Eq.)52   \(2.1
+                ( 
+                """;
+
+            var input = StringBytesTestConverter.Convert(s, false);
+
+            var lenientParser = new PageContentParser(ReflectionGraphicsStateOperationFactory.Instance, new StackDepthGuard(256), true);
+            var result = lenientParser.Parse(1, input.Bytes, log);
+
+            Assert.NotEmpty(result);
+        }
+
         private static string LineEndingsToWhiteSpace(string str)
         {
             return str.Replace("\r\n", " ").Replace('\n', ' ').Replace('\r', ' ');

src/UglyToad.PdfPig.Tests/Parser/Parts/BruteForceSearcherTests.cs

@@ -59,7 +59,7 @@ startxref
 
             Assert.Equal(4, locations.Count);
 
-            Assert.Equal(TestDataOffsets, locations.Values);
+            Assert.Equal(TestDataOffsets, locations.Values.Select(x => x.Value1));
         }
 
         [Fact]
@@ -111,7 +111,7 @@ endobj
                 s.IndexOf("11 0 obj", StringComparison.OrdinalIgnoreCase)
             };
 
-            Assert.Equal(expectedLocations, locations.Values);
+            Assert.Equal(expectedLocations, locations.Values.Select(x => x.Value1));
         }
 
         [Fact]
@@ -142,7 +142,7 @@ endobj";
                 s.IndexOf("5 0 obj", StringComparison.OrdinalIgnoreCase)
             };
 
-            Assert.Equal(expectedLocations, locations.Values);
+            Assert.Equal(expectedLocations, locations.Values.Select(x => x.Value1));
         }
 
         [Fact]
@@ -156,17 +156,17 @@ endobj";
 
                 Assert.Equal(13, locations.Count);
 
-                Assert.Equal(6183, locations[new IndirectReference(1, 0)]);
-                Assert.Equal(244, locations[new IndirectReference(2, 0)]);
-                Assert.Equal(15, locations[new IndirectReference(3, 0)]);
-                Assert.Equal(222, locations[new IndirectReference(4, 0)]);
-                Assert.Equal(5766, locations[new IndirectReference(5, 0)]);
-                Assert.Equal(353, locations[new IndirectReference(6, 0)]);
-                Assert.Equal(581, locations[new IndirectReference(7, 0)]);
-                Assert.Equal(5068, locations[new IndirectReference(8, 0)]);
-                Assert.Equal(5091, locations[new IndirectReference(9, 0)]);
+                Assert.Equal(6183, locations[new IndirectReference(1, 0)].Value1);
+                Assert.Equal(244, locations[new IndirectReference(2, 0)].Value1);
+                Assert.Equal(15, locations[new IndirectReference(3, 0)].Value1);
+                Assert.Equal(222, locations[new IndirectReference(4, 0)].Value1);
+                Assert.Equal(5766, locations[new IndirectReference(5, 0)].Value1);
+                Assert.Equal(353, locations[new IndirectReference(6, 0)].Value1);
+                Assert.Equal(581, locations[new IndirectReference(7, 0)].Value1);
+                Assert.Equal(5068, locations[new IndirectReference(8, 0)].Value1);
+                Assert.Equal(5091, locations[new IndirectReference(9, 0)].Value1);
 
-                var s = GetStringAt(bytes, locations[new IndirectReference(3, 0)]);
+                var s = GetStringAt(bytes, locations[new IndirectReference(3, 0)].Value1);
                 Assert.StartsWith("3 0 obj", s);
             }
         }
@@ -180,17 +180,17 @@ endobj";
 
             Assert.Equal(13, locations.Count);
 
-            Assert.Equal(6183, locations[new IndirectReference(1, 0)]);
-            Assert.Equal(244, locations[new IndirectReference(2, 0)]);
-            Assert.Equal(15, locations[new IndirectReference(3, 0)]);
-            Assert.Equal(222, locations[new IndirectReference(4, 0)]);
-            Assert.Equal(5766, locations[new IndirectReference(5, 0)]);
-            Assert.Equal(353, locations[new IndirectReference(6, 0)]);
-            Assert.Equal(581, locations[new IndirectReference(7, 0)]);
-            Assert.Equal(5068, locations[new IndirectReference(8, 0)]);
-            Assert.Equal(5091, locations[new IndirectReference(9, 0)]);
+            Assert.Equal(6183, locations[new IndirectReference(1, 0)].Value1);
+            Assert.Equal(244, locations[new IndirectReference(2, 0)].Value1);
+            Assert.Equal(15, locations[new IndirectReference(3, 0)].Value1);
+            Assert.Equal(222, locations[new IndirectReference(4, 0)].Value1);
+            Assert.Equal(5766, locations[new IndirectReference(5, 0)].Value1);
+            Assert.Equal(353, locations[new IndirectReference(6, 0)].Value1);
+            Assert.Equal(581, locations[new IndirectReference(7, 0)].Value1);
+            Assert.Equal(5068, locations[new IndirectReference(8, 0)].Value1);
+            Assert.Equal(5091, locations[new IndirectReference(9, 0)].Value1);
 
-            var s = GetStringAt(bytes, locations[new IndirectReference(3, 0)]);
+            var s = GetStringAt(bytes, locations[new IndirectReference(3, 0)].Value1);
             Assert.StartsWith("3 0 obj", s);
         }
 
@@ -203,21 +203,21 @@ endobj";
 
             Assert.Equal(13, locations.Count);
 
-            Assert.Equal(17, locations[new IndirectReference(1, 0)]);
-            Assert.Equal(249, locations[new IndirectReference(2, 0)]);
-            Assert.Equal(14291, locations[new IndirectReference(3, 0)]);
-            Assert.Equal(275, locations[new IndirectReference(4, 0)]);
-            Assert.Equal(382, locations[new IndirectReference(5, 0)]);
-            Assert.Equal(13283, locations[new IndirectReference(6, 0)]);
-            Assert.Equal(13309, locations[new IndirectReference(7, 0)]);
-            Assert.Equal(13556, locations[new IndirectReference(8, 0)]);
-            Assert.Equal(13926, locations[new IndirectReference(9, 0)]);
-            Assert.Equal(14183, locations[new IndirectReference(10, 0)]);
-            Assert.Equal(14224, locations[new IndirectReference(11, 0)]);
-            Assert.Equal(14428, locations[new IndirectReference(12, 0)]);
-            Assert.Equal(14488, locations[new IndirectReference(13, 0)]);
+            Assert.Equal(17, locations[new IndirectReference(1, 0)].Value1);
+            Assert.Equal(249, locations[new IndirectReference(2, 0)].Value1);
+            Assert.Equal(14291, locations[new IndirectReference(3, 0)].Value1);
+            Assert.Equal(275, locations[new IndirectReference(4, 0)].Value1);
+            Assert.Equal(382, locations[new IndirectReference(5, 0)].Value1);
+            Assert.Equal(13283, locations[new IndirectReference(6, 0)].Value1);
+            Assert.Equal(13309, locations[new IndirectReference(7, 0)].Value1);
+            Assert.Equal(13556, locations[new IndirectReference(8, 0)].Value1);
+            Assert.Equal(13926, locations[new IndirectReference(9, 0)].Value1);
+            Assert.Equal(14183, locations[new IndirectReference(10, 0)].Value1);
+            Assert.Equal(14224, locations[new IndirectReference(11, 0)].Value1);
+            Assert.Equal(14428, locations[new IndirectReference(12, 0)].Value1);
+            Assert.Equal(14488, locations[new IndirectReference(13, 0)].Value1);
 
-            var s = GetStringAt(bytes, locations[new IndirectReference(12, 0)]);
+            var s = GetStringAt(bytes, locations[new IndirectReference(12, 0)].Value1);
             Assert.StartsWith("12 0 obj", s);
         }
 
@@ -230,7 +230,7 @@ endobj";
 
             var locations = BruteForceSearcher.GetObjectLocations(input);
 
-            Assert.Equal(TestDataOffsets, locations.Values);
+            Assert.Equal(TestDataOffsets, locations.Values.Select(x => x.Value1));
         }
 
         [Fact]
@@ -265,7 +265,7 @@ endobj
                 s.IndexOf("11 0 obj", StringComparison.OrdinalIgnoreCase)
             };
 
-            Assert.Equal(expectedLocations, locations.Values);
+            Assert.Equal(expectedLocations, locations.Values.Select(x => x.Value1));
         }
 
         private static string GetStringAt(IInputBytes bytes, long location)

src/UglyToad.PdfPig.Tests/Parser/Parts/DirectObjectFinderTests.cs

@@ -15,8 +15,8 @@
             var reference1 = new IndirectReference(7, 0);
             var reference2 = new IndirectReference(9, 0);
 
-            scanner.Objects[reference1] = new ObjectToken(10, reference1, new IndirectReferenceToken(reference2));
-            scanner.Objects[reference2] = new ObjectToken(12, reference2, new NumericToken(69));
+            scanner.Objects[reference1] = new ObjectToken(XrefLocation.File(10), reference1, new IndirectReferenceToken(reference2));
+            scanner.Objects[reference2] = new ObjectToken(XrefLocation.File(12), reference2, new NumericToken(69));
 
             Assert.True(DirectObjectFinder.TryGet(new IndirectReferenceToken(reference1), scanner, out NumericToken result));
 
@@ -29,8 +29,8 @@
             var reference1 = new IndirectReference(7, 0);
             var reference2 = new IndirectReference(9, 0);
 
-            scanner.Objects[reference1] = new ObjectToken(10, reference1, new IndirectReferenceToken(reference2));
-            scanner.Objects[reference2] = new ObjectToken(12, reference2, new NumericToken(69));
+            scanner.Objects[reference1] = new ObjectToken(XrefLocation.File(10), reference1, new IndirectReferenceToken(reference2));
+            scanner.Objects[reference2] = new ObjectToken(XrefLocation.File(12), reference2, new NumericToken(69));
 
             var result = DirectObjectFinder.Get<NumericToken>(reference1, scanner);
 
@@ -43,8 +43,8 @@
             var reference1 = new IndirectReference(7, 0);
             var reference2 = new IndirectReference(9, 0);
 
-            scanner.Objects[reference1] = new ObjectToken(10, reference1, new IndirectReferenceToken(reference2));
-            scanner.Objects[reference2] = new ObjectToken(12, reference2, new NumericToken(69));
+            scanner.Objects[reference1] = new ObjectToken(XrefLocation.File(10), reference1, new IndirectReferenceToken(reference2));
+            scanner.Objects[reference2] = new ObjectToken(XrefLocation.File(12), reference2, new NumericToken(69));
 
             var result = DirectObjectFinder.Get<NumericToken>(new IndirectReferenceToken(reference1), scanner);
 
@@ -57,7 +57,7 @@
             var reference = new IndirectReference(10, 0);
 
             const string expected = "Goopy";
-            scanner.Objects[reference] = new ObjectToken(10, reference, new ArrayToken(new []
+            scanner.Objects[reference] = new ObjectToken(XrefLocation.File(10), reference, new ArrayToken(new []
             {
                 new StringToken(expected)
             }));
@@ -74,12 +74,12 @@
             var reference2 = new IndirectReference(69, 0);
 
             const string expected = "Goopy";
-            scanner.Objects[reference] = new ObjectToken(10, reference, new ArrayToken(new[]
+            scanner.Objects[reference] = new ObjectToken(XrefLocation.File(10), reference, new ArrayToken(new[]
             {
                 new IndirectReferenceToken(reference2) 
             }));
 
-            scanner.Objects[reference2] = new ObjectToken(69, reference2, new StringToken(expected));
+            scanner.Objects[reference2] = new ObjectToken(XrefLocation.File(69), reference2, new StringToken(expected));
 
             var result = DirectObjectFinder.Get<StringToken>(reference, scanner);
 
@@ -91,7 +91,7 @@
         {
             var reference = new IndirectReference(10, 0);
 
-            scanner.Objects[reference] = new ObjectToken(10, reference, new ArrayToken(new[]
+            scanner.Objects[reference] = new ObjectToken(XrefLocation.File(10), reference, new ArrayToken(new[]
             {
                 new NumericToken(5), new NumericToken(6), new NumericToken(0)   
             }));

src/UglyToad.PdfPig.Tests/TestObjectLocationProvider.cs

@@ -6,14 +6,14 @@
 
     internal class TestObjectLocationProvider : IObjectLocationProvider
     {
-        public Dictionary<IndirectReference, long> Offsets { get; } = new Dictionary<IndirectReference, long>();
+        public Dictionary<IndirectReference, XrefLocation> Offsets { get; } = new Dictionary<IndirectReference, XrefLocation>();
 
-        public bool TryGetOffset(IndirectReference reference, out long offset)
+        public bool TryGetOffset(IndirectReference reference, out XrefLocation offset)
         {
             return Offsets.TryGetValue(reference, out offset);
         }
 
-        public void UpdateOffset(IndirectReference reference, long offset)
+        public void UpdateOffset(IndirectReference reference, XrefLocation offset)
         {
             Offsets[reference] = offset;
         }

src/UglyToad.PdfPig.Tokens/ObjectToken.cs

@@ -12,7 +12,7 @@
         /// <summary>
         /// The offset to the start of the object number from the start of the file in bytes.
         /// </summary>
-        public long Position { get; }
+        public XrefLocation Position { get; }
 
         /// <summary>
         /// The object and generation number of the object.
@@ -30,7 +30,7 @@
         /// <param name="position">The offset in bytes from the start of the file for this object.</param>
         /// <param name="number">The identifier for this object.</param>
         /// <param name="data">The data contained in this object.</param>
-        public ObjectToken(long position, IndirectReference number, IToken data)
+        public ObjectToken(XrefLocation position, IndirectReference number, IToken data)
         {
             Position = position;
             Number = number;

src/UglyToad.PdfPig/AcroForms/AcroFormFactory.cs

@@ -28,12 +28,12 @@
 
         private readonly IPdfTokenScanner tokenScanner;
         private readonly ILookupFilterProvider filterProvider;
-        private readonly IReadOnlyDictionary<IndirectReference, long> objectOffsets;
+        private readonly IReadOnlyDictionary<IndirectReference, XrefLocation> objectOffsets;
 
         public AcroFormFactory(
             IPdfTokenScanner tokenScanner,
             ILookupFilterProvider filterProvider,
-            IReadOnlyDictionary<IndirectReference, long> objectOffsets)
+            IReadOnlyDictionary<IndirectReference, XrefLocation> objectOffsets)
         {
             this.tokenScanner = tokenScanner ?? throw new ArgumentNullException(nameof(tokenScanner));
             this.filterProvider = filterProvider ?? throw new ArgumentNullException(nameof(filterProvider));

src/UglyToad.PdfPig/Filters/FlateFilter.cs

@@ -2,11 +2,10 @@
 {
     using Fonts;
     using System;
-    using System.Buffers.Binary;
     using System.IO;
     using System.IO.Compression;
     using Tokens;
-    using UglyToad.PdfPig.Core;
+    using Core;
     using Util;
 
     /// <summary>
@@ -55,89 +54,41 @@
             return input;
         }
 
-        private static Memory<byte> Decompress(Memory<byte> input, int predictor, int colors, int bitsPerComponent, int columns)
+        private static Memory<byte> Decompress(Memory<byte> input,
+            int predictor,
+            int colors,
+            int bitsPerComponent,
+            int columns)
         {
-#if NET
             using var memoryStream = MemoryHelper.AsReadOnlyMemoryStream(input);
+            // The first 2 bytes are the header which DeflateStream does not support.
+            memoryStream.ReadByte();
+            memoryStream.ReadByte();
+
             try
             {
-                using (var zlib = new ZLibStream(memoryStream, CompressionMode.Decompress))
-                using (var output = new MemoryStream((int)(input.Length * 1.5)))
-                using (var f = PngPredictor.WrapPredictor(output, predictor, colors, bitsPerComponent, columns))
-                {
-                    zlib.CopyTo(f);
-                    f.Flush();
+                using var deflate = new DeflateStream(memoryStream, CompressionMode.Decompress);
+                using var output = new MemoryStream((int)(input.Length * 1.5));
+                using var f = PngPredictor.WrapPredictor(output, predictor, colors, bitsPerComponent, columns);
+                
+                deflate.CopyTo(f);
+                f.Flush();
 
-                    return output.AsMemory();
-                }
+                return output.AsMemory();
             }
             catch (InvalidDataException ex)
             {
                 throw new CorruptCompressedDataException("Invalid Flate compressed stream encountered", ex);
             }
-#else
-            // Ideally we would like to use the ZLibStream class but that is only available in .NET 5+.
-            // We look at the raw data now
-            // *  First we have 2 bytes, specifying the type of compression
-            // * Then we have the deflated data
-            // * Then we have a 4 byte checksum (Adler32)
-
-            // Would be so nice to have zlib do the framing here... but the deflate stream already reads data from the stream that we need.
-
-            using var memoryStream = MemoryHelper.AsReadOnlyMemoryStream(input.Slice(2, input.Length - 2 /* Header */ - 4 /* Checksum */));
-            // The first 2 bytes are the header which DeflateStream can't handle. After the s
-            var adlerBytes = input.Slice(input.Length - 4, 4).Span;
-            uint expected = BinaryPrimitives.ReadUInt32BigEndian(adlerBytes);
-            uint altExpected = expected;
-
-            // Sometimes the data ends with "\r\n", "\r" or "\n" and we don't know if it is part of the zlib
-            // Ideally this would have been removed by the caller from the provided length...
-            if (adlerBytes[3] == '\n' || adlerBytes[3] == '\r')
-            {
-                if (adlerBytes[3] == '\n' && adlerBytes[2] == '\r')
-                {
-                    // Now we don't know which value is the good one. The value could be ok, or padding.
-                    // Lets allow both values for now. Allowing two out of 2^32 is much better than allowing everything
-                    adlerBytes = input.Slice(input.Length - 6, 4).Span;
-                }
-                else
-                {
-                    // Same but now for just '\n' or '\r' instead of '\r\n'
-                    adlerBytes = input.Slice(input.Length - 5, 4).Span;
-                }
-
-                altExpected = BinaryPrimitives.ReadUInt32BigEndian(adlerBytes);
-            }
-
-
-            try
-            {
-                using (var deflate = new DeflateStream(memoryStream, CompressionMode.Decompress))
-                using (var adlerStream = new Adler32ChecksumStream(deflate))
-                using (var output = new MemoryStream((int)(input.Length * 1.5)))
-                using (var f = PngPredictor.WrapPredictor(output, predictor, colors, bitsPerComponent, columns))
-                {
-                    adlerStream.CopyTo(f);
-                    f.Flush();
-
-                    uint actual = adlerStream.Checksum;
-                    if (expected != actual && altExpected != actual)
-                    {
-                        throw new CorruptCompressedDataException("Flate stream has invalid checksum");
-                    }
-
-                    return output.AsMemory();
-                }
-            }
-            catch (InvalidDataException ex)
-            {
-                throw new CorruptCompressedDataException("Invalid Flate compressed stream encountered", ex);
-            }
-#endif
         }
 
-        /// <inheritdoc />
-        public byte[] Encode(Stream input, DictionaryToken streamDictionary, int index)
+        /// <summary>
+        /// Convert a decoded data stream back to the encoded version.
+        /// </summary>
+        /// <param name="input">The decoded data.</param>
+        /// <param name="streamDictionary">The stream dictionary with the parameters to use.</param>
+        /// <returns>The Flate encoded data.</returns>
+        public byte[] Encode(Stream input, DictionaryToken streamDictionary)
         {
             const int headerLength = 2;
             const int checksumLength = 4;

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

@@ -16,7 +16,7 @@ internal static partial class FirstPassParser
     {
         log ??= new NoOpLog();
 
-        IReadOnlyDictionary<IndirectReference, long>? bruteForceOffsets = null;
+        IReadOnlyDictionary<IndirectReference, XrefLocation>? bruteForceOffsets = null;
         var didBruteForce = false;
         DictionaryToken? bruteForceTrailer = null;
 
@@ -92,7 +92,7 @@ internal static partial class FirstPassParser
         }
 
         DictionaryToken? lastTrailer = null;
-        var flattenedOffsets = new Dictionary<IndirectReference, long>();
+        var flattenedOffsets = new Dictionary<IndirectReference, XrefLocation>();
         foreach (var xrefPart in orderedXrefs)
         {
             if (xrefPart.Dictionary != null)
@@ -230,12 +230,12 @@ internal class FirstPassResults
     /// <summary>
     /// All offsets found if a brute-force search was applied.
     /// </summary>
-    public IReadOnlyDictionary<IndirectReference, long>? BruteForceOffsets { get; }
+    public IReadOnlyDictionary<IndirectReference, XrefLocation>? BruteForceOffsets { get; }
 
     /// <summary>
     /// All offsets found from the leaf xref.
     /// </summary>
-    public IReadOnlyDictionary<IndirectReference, long> XrefOffsets { get; }
+    public IReadOnlyDictionary<IndirectReference, XrefLocation> XrefOffsets { get; }
 
     /// <summary>
     /// The trailer dictionary of the leaf xref if we found any.
@@ -244,8 +244,8 @@ internal class FirstPassResults
 
     public FirstPassResults(
         IReadOnlyList<IXrefSection> parts,
-        IReadOnlyDictionary<IndirectReference, long>? bruteForceOffsets,
-        IReadOnlyDictionary<IndirectReference, long> xrefOffsets,
+        IReadOnlyDictionary<IndirectReference, XrefLocation>? bruteForceOffsets,
+        IReadOnlyDictionary<IndirectReference, XrefLocation> xrefOffsets,
         DictionaryToken? trailer)
     {
         Parts = parts;

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

@@ -14,7 +14,7 @@ internal interface IXrefSection
     /// <summary>
     /// The bytes offsets of the objects in this xref.
     /// </summary>
-    public IReadOnlyDictionary<IndirectReference, long> ObjectOffsets { get; }
+    public IReadOnlyDictionary<IndirectReference, XrefLocation> ObjectOffsets { get; }
 
     /// <summary>
     /// The dictionary for this xref, for the trailer xref this is the trailer dictionary, for streams the stream dictionary.

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

@@ -19,7 +19,7 @@ internal static class XrefBruteForcer
         // Guard against circular references; only read xref at each offset once
         var xrefOffsetSeen = new HashSet<long>();
 
-        var bruteForceObjPositions = new Dictionary<IndirectReference, long>();
+        var bruteForceObjPositions = new Dictionary<IndirectReference, XrefLocation>();
 
         DictionaryToken? trailer = null;
 
@@ -123,7 +123,7 @@ internal static class XrefBruteForcer
 
             if (buffer.EndsWith(" obj") && numericsQueue[0] > 0)
             {
-                bruteForceObjPositions[new IndirectReference(numericsQueue[0], (int)numericsQueue[1])] = positionsQueue[0];
+                bruteForceObjPositions[new IndirectReference(numericsQueue[0], (int)numericsQueue[1])] = XrefLocation.File(positionsQueue[0]);
 
                 lastObjPosition = positionsQueue[0];
 
@@ -208,12 +208,12 @@ internal static class XrefBruteForcer
 
     public class Result(
         IReadOnlyList<IXrefSection> xRefParts,
-        IReadOnlyDictionary<IndirectReference, long> objectOffsets,
+        IReadOnlyDictionary<IndirectReference, XrefLocation> objectOffsets,
         DictionaryToken? lastTrailer)
     {
         public IReadOnlyList<IXrefSection> XRefParts { get; } = xRefParts;
 
-        public IReadOnlyDictionary<IndirectReference, long> ObjectOffsets { get; } = objectOffsets;
+        public IReadOnlyDictionary<IndirectReference, XrefLocation> ObjectOffsets { get; } = objectOffsets;
 
         public DictionaryToken? LastTrailer { get; } = lastTrailer;
     }

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

@@ -10,7 +10,7 @@ internal sealed class XrefStream : IXrefSection
     /// <summary>
     /// The corresponding byte offset for each keyed object in this document.
     /// </summary>
-    public IReadOnlyDictionary<IndirectReference, long> ObjectOffsets { get; }
+    public IReadOnlyDictionary<IndirectReference, XrefLocation> ObjectOffsets { get; }
 
     public DictionaryToken Dictionary { get; }
 
@@ -20,7 +20,7 @@ internal sealed class XrefStream : IXrefSection
 
     public XrefStream(
         long offset,
-        IReadOnlyDictionary<IndirectReference, long> objectOffsets,
+        IReadOnlyDictionary<IndirectReference, XrefLocation> objectOffsets,
         DictionaryToken streamDictionary,
         XrefOffsetCorrection correctionType,
         long offsetCorrection)

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

@@ -98,7 +98,7 @@ internal static class XrefStreamParser
                 ? stackalloc byte[fieldSizes.LineLength]
                 : new byte[fieldSizes.LineLength];
 
-            var numbers = new List<(long obj, int gen, int off)>();
+            var numbers = new List<(long obj, int gen, XrefLocation location)>();
 
             foreach (var objectNumber in objectNumbers)
             {
@@ -136,7 +136,7 @@ internal static class XrefStreamParser
 
             return new XrefStream(
                 xrefOffset,
-                numbers.ToDictionary(x => new IndirectReference(x.obj, x.gen), x => (long)x.off),
+                numbers.ToDictionary(x => new IndirectReference(x.obj, x.gen), x => x.location),
                 dictToken,
                 offsetCorrectionType,
                 offsetCorrection);
@@ -175,7 +175,7 @@ internal static class XrefStreamParser
         int type,
         long objectNumber,
         XrefFieldSize fieldSizes,
-        List<(long, int, int)> results,
+        List<(long, int, XrefLocation)> results,
         ReadOnlySpan<byte> lineBuffer)
     {
         switch (type)
@@ -184,19 +184,23 @@ internal static class XrefStreamParser
                 // Ignore free objects.
                 break;
             case 1:
-                // Non object stream entries.
-                var offset = 0;
-                for (var i = 0; i < fieldSizes.Field2Size; i++)
-                {
-                    offset += (lineBuffer[i + fieldSizes.Field1Size] & 0x00ff) << ((fieldSizes.Field2Size - i - 1) * 8);
-                }
-                var genNum = 0;
-                for (var i = 0; i < fieldSizes.Field3Size; i++)
-                {
-                    genNum += (lineBuffer[i + fieldSizes.Field1Size + fieldSizes.Field2Size] & 0x00ff) << ((fieldSizes.Field3Size - i - 1) * 8);
+                var offset = ReadUnsigned(
+                    lineBuffer,
+                    fieldSizes.Field1Size,
+                    fieldSizes.Field2Size);
+
+                var genNum = ReadUnsigned(
+                    lineBuffer,
+                    fieldSizes.Field1Size + fieldSizes.Field2Size,
+                    fieldSizes.Field3Size);
+
+                if (offset < 0)
+                { 
+                    throw new PdfDocumentFormatException(
+                        $"Location with negative offset {offset} found for object {objectNumber}");
                 }
 
-                results.Add((objectNumber, genNum, offset));
+                results.Add((objectNumber, (int)genNum, XrefLocation.File(offset)));
 
                 break;
             case 2:
@@ -205,28 +209,49 @@ internal static class XrefStreamParser
                  * 2nd argument is object number of object stream
                  * 3rd argument is index of object within object stream
                  * 
-                 * For sequential PDFParser we do not need this information
-                 * because
-                 * These objects are handled by the dereferenceObjects() method
-                 * since they're only pointing to object numbers
-                 * 
-                 * However for XRef aware parsers we have to know which objects contain
-                 * object streams. We will store this information in normal xref mapping
-                 * table but add object stream number with minus sign in order to
-                 * distinguish from file offsets
                  */
-                var objstmObjNr = 0;
-                for (var i = 0; i < fieldSizes.Field2Size; i++)
+            
+                var objectStreamNumber = ReadUnsigned(
+                    lineBuffer,
+                    fieldSizes.Field1Size,
+                    fieldSizes.Field2Size);
+
+                var streamIndex = ReadUnsigned(
+                    lineBuffer,
+                    fieldSizes.Field1Size + fieldSizes.Field2Size,
+                    fieldSizes.Field3Size);
+
+                if (objectStreamNumber < 0)
                 {
-                    objstmObjNr += (lineBuffer[i + fieldSizes.Field1Size] & 0x00ff) << ((fieldSizes.Field2Size - i - 1) * 8);
+                    throw new PdfDocumentFormatException(
+                        $"Location with negative or zero object stream number {objectStreamNumber} found for object {objectNumber}");
                 }
 
-                results.Add((objectNumber, 0, -objstmObjNr));
+                if (streamIndex < 0)
+                {
+                    throw new PdfDocumentFormatException(
+                        $"Location with negative stream index {streamIndex} found for object {objectNumber} in stream {objectStreamNumber}");
+                }
+
+                results.Add((objectNumber, 0, XrefLocation.Stream(objectStreamNumber, (int)streamIndex)));
 
                 break;
         }
     }
 
+    private static long ReadUnsigned(ReadOnlySpan<byte> buffer, int start, int width)
+    {
+        long value = 0;
+
+        for (int i = 0; i < width; i++)
+        {
+            value <<= 8;
+            value |= buffer[start + i];
+        }
+
+        return value;
+    }
+
     private static (long from, long? to) ReadStreamTolerant(IInputBytes bytes)
     {
         var buffer = new CircularByteBuffer("endstream ".Length);

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

@@ -13,7 +13,7 @@ internal sealed class XrefTable : IXrefSection
     /// <summary>
     /// The corresponding byte offset for each keyed object in this document.
     /// </summary>
-    public IReadOnlyDictionary<IndirectReference, long> ObjectOffsets { get; }
+    public IReadOnlyDictionary<IndirectReference, XrefLocation> ObjectOffsets { get; }
 
     public DictionaryToken? Dictionary { get; }
 
@@ -23,7 +23,7 @@ internal sealed class XrefTable : IXrefSection
 
     public XrefTable(
         long offset,
-        IReadOnlyDictionary<IndirectReference, long> objectOffsets,
+        IReadOnlyDictionary<IndirectReference, XrefLocation> objectOffsets,
         DictionaryToken? trailer,
         XrefOffsetCorrection correctionType,
         long offsetCorrection)

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

@@ -152,7 +152,7 @@ internal static class XrefTableParser
             }
         }
 
-        var offsets = new Dictionary<IndirectReference, long>();
+        var offsets = new Dictionary<IndirectReference, XrefLocation>();
         if (readNums.Count == 0)
         {
             if (trailer != null)
@@ -233,7 +233,7 @@ internal static class XrefTableParser
                 if (type == occupiedSentinel)
                 {
                     var indirectRef = new IndirectReference(objNum, (int)gen);
-                    offsets[indirectRef] = objOffset;
+                    offsets[indirectRef] = XrefLocation.File(objOffset);
                 }
 
                 objNum++;

src/UglyToad.PdfPig/Parser/Parts/BruteForceSearcher.cs

@@ -18,7 +18,7 @@
         /// </summary>
         /// <param name="bytes">The bytes of the document.</param>
         /// <returns>The object keys and offsets for the objects in this document.</returns>
-        public static IReadOnlyDictionary<IndirectReference, long> GetObjectLocations(IInputBytes bytes)
+        public static IReadOnlyDictionary<IndirectReference, XrefLocation> GetObjectLocations(IInputBytes bytes)
         {
             if (bytes is null)
             {
@@ -29,7 +29,7 @@
 
             var lastEndOfFile = GetLastEndOfFileMarker(bytes);
 
-            var results = new Dictionary<IndirectReference, long>();
+            var results = new Dictionary<IndirectReference, XrefLocation>();
 
             var generationBytes = new StringBuilder();
             var objectNumberBytes = new StringBuilder();
@@ -174,7 +174,7 @@
                 var obj = long.Parse(objectNumberBytes.ToString(), CultureInfo.InvariantCulture);
                 var generation = int.Parse(generationBytes.ToString(), CultureInfo.InvariantCulture);
 
-                results[new IndirectReference(obj, generation)] = bytes.CurrentOffset;
+                results[new IndirectReference(obj, generation)] = XrefLocation.File(bytes.CurrentOffset);
 
                 generationBytes.Clear();
                 objectNumberBytes.Clear();

src/UglyToad.PdfPig/Tokenization/Scanner/IObjectLocationProvider.cs

@@ -6,9 +6,9 @@
 
     internal interface IObjectLocationProvider
     {
-        bool TryGetOffset(IndirectReference reference, out long offset);
+        bool TryGetOffset(IndirectReference reference, out XrefLocation offset);
 
-        void UpdateOffset(IndirectReference reference, long offset);
+        void UpdateOffset(IndirectReference reference, XrefLocation offset);
 
         bool TryGetCached(IndirectReference reference, [NotNullWhen(true)] out ObjectToken? objectToken);
 

src/UglyToad.PdfPig/Tokenization/Scanner/ObjectLocationProvider.cs

@@ -13,16 +13,16 @@
 
         private readonly IInputBytes bytes;
 
-        private IReadOnlyDictionary<IndirectReference, long>? bruteForcedOffsets;
+        private IReadOnlyDictionary<IndirectReference, XrefLocation>? bruteForcedOffsets;
 
-        private readonly Dictionary<IndirectReference, long> offsets;
+        private readonly Dictionary<IndirectReference, XrefLocation> offsets;
 
         public ObjectLocationProvider(
-            IReadOnlyDictionary<IndirectReference, long> xrefOffsets,
-            IReadOnlyDictionary<IndirectReference, long>? bruteForcedOffsets,
+            IReadOnlyDictionary<IndirectReference, XrefLocation> xrefOffsets,
+            IReadOnlyDictionary<IndirectReference, XrefLocation>? bruteForcedOffsets,
             IInputBytes bytes)
         {
-            offsets = new Dictionary<IndirectReference, long>();
+            offsets = new Dictionary<IndirectReference, XrefLocation>();
             foreach (var xrefOffset in xrefOffsets)
             {
                 offsets[xrefOffset.Key] = xrefOffset.Value;
@@ -32,7 +32,7 @@
             this.bytes = bytes;
         }
 
-        public bool TryGetOffset(IndirectReference reference, out long offset)
+        public bool TryGetOffset(IndirectReference reference, out XrefLocation offset)
         {
             if (bruteForcedOffsets != null && bruteForcedOffsets.TryGetValue(reference, out var bfOffset))
             {
@@ -42,16 +42,6 @@
 
             if (offsets.TryGetValue(reference, out offset))
             {
-                if (offset + reference.ObjectNumber == 0)
-                {
-                    // We have a case where 'offset' and
-                    // 'reference.ObjectNumber' have the same value
-                    // and opposite signs.
-                    // This results in an infinite recursion in
-                    // PdfTokenScanner.GetObjectFromStream() where 
-                    // `var streamObjectNumber = offset * -1;`
-                    throw new PdfDocumentFormatException("Avoiding infinite recursion in ObjectLocationProvider.TryGetOffset() as 'offset' and 'reference.ObjectNumber' have the same value and opposite signs.");
-                }
                 return true;
             }
 
@@ -63,7 +53,7 @@
             return bruteForcedOffsets.TryGetValue(reference, out offset);
         }
 
-        public void UpdateOffset(IndirectReference reference, long offset)
+        public void UpdateOffset(IndirectReference reference, XrefLocation offset)
         {
             offsets[reference] = offset;
         }
@@ -81,8 +71,9 @@
             }
 
             // Don't cache incorrect locations.
-            if (!force && offsets.TryGetValue(objectToken.Number, out var expected)
-                && objectToken.Position != expected)
+            if (!force
+                && offsets.TryGetValue(objectToken.Number, out var expected)
+                && (objectToken.Position.Type != expected.Type || objectToken.Position.Value1 != expected.Value1))
             {
                 return;
             }

src/UglyToad.PdfPig/Tokenization/Scanner/PdfTokenScanner.cs

@@ -1,5 +1,8 @@
 namespace UglyToad.PdfPig.Tokenization.Scanner
 {
+    using Core;
+    using Encryption;
+    using Filters;
     using System;
     using System.Collections.Generic;
     using System.Diagnostics;
@@ -7,9 +10,6 @@
     using System.Globalization;
     using System.Linq;
     using System.Text.RegularExpressions;
-    using Core;
-    using Encryption;
-    using Filters;
     using Tokens;
     using UglyToad.PdfPig.Parser.FileStructure;
 
@@ -164,7 +164,7 @@
                         var actualReference = new IndirectReference(objectNumber.Int, generation.Int);
                         var actualToken = encryptionHandler.Decrypt(actualReference, readTokens[0]);
 
-                        CurrentToken = new ObjectToken(startPosition, actualReference, actualToken);
+                        CurrentToken = new ObjectToken(XrefLocation.File(startPosition), actualReference, actualToken);
 
                         readTokens.Clear();
                         coreTokenScanner.Seek(previousTokenPositions[0]);
@@ -191,7 +191,7 @@
                         var actualReference = new IndirectReference(objectNumber.Int, generation.Int);
                         var actualToken = encryptionHandler.Decrypt(actualReference, readTokens[0]);
                 
-                        CurrentToken = new ObjectToken(startPosition, actualReference, actualToken);
+                        CurrentToken = new ObjectToken(XrefLocation.File(startPosition), actualReference, actualToken);
                         readTokens.Clear();
                         coreTokenScanner.Seek(previousTokenPositions[2]);
                 
@@ -291,9 +291,9 @@
 
             token = encryptionHandler.Decrypt(reference, token);
 
-            CurrentToken = new ObjectToken(startPosition, reference, token);
+            CurrentToken = new ObjectToken(XrefLocation.File(startPosition), reference, token);
 
-            objectLocationProvider.UpdateOffset(reference, startPosition);
+            objectLocationProvider.UpdateOffset(reference, XrefLocation.File(startPosition));
 
             readTokens.Clear();
             return true;
@@ -626,10 +626,10 @@
                 // We can only find it if we know where it is.
                 if (objectLocationProvider.TryGetOffset(lengthReference.Data, out var offset))
                 {
-                    if (offset < 0)
+                    if (offset.Type == XrefEntryType.ObjectStream)
                     {
-                        ushort searchDepth = 0;
-                        var result = GetObjectFromStream(lengthReference.Data, offset, ref searchDepth);
+                        Span<int> stack = stackalloc int[7];
+                        var result = GetObjectFromStream(lengthReference.Data, offset, stack, 0);
 
                         if (!(result.Data is NumericToken streamLengthToken))
                         {
@@ -639,8 +639,9 @@
 
                         return streamLengthToken.Long;
                     }
+
                     // Move to the length object and read it.
-                    Seek(offset);
+                    Seek(offset.Value1);
 
                     // Keep a copy of the read tokens here since this list must be empty prior to move next.
                     var oldData = new List<IToken>(readTokens);
@@ -721,19 +722,31 @@
 
         public ObjectToken? Get(IndirectReference reference)
         {
-            ushort searchDepth = 0;
-            return Get(reference, ref searchDepth);
+            Span<int> stack = stackalloc int[7];
+            return Get(reference, stack, 0);
         }
 
-        private ObjectToken? Get(IndirectReference reference, ref ushort searchDepth)
+        private ObjectToken? Get(IndirectReference reference, Span<int> navSet, byte depth)
         {
-            if (searchDepth > 100)
+            if (depth >= navSet.Length)
             {
-                throw new PdfDocumentFormatException("Reached maximum search depth while getting indirect reference.");
+                var chain = string.Join(", ", navSet.ToArray());
+                throw new PdfDocumentFormatException($"Deep object chain detected when looking for {reference}: {chain}.");
             }
 
-            searchDepth++;
+            // Cycle detection (linear scan, but depth is tiny)
+            for (var i = 0; i < depth; i++)
+            {
+                if (navSet[i] == reference.ObjectNumber)
+                {
+                    var chain = string.Join(", ", navSet.ToArray());
+                    throw new PdfDocumentFormatException(
+                        $"Circular reference encountered when looking for object {reference}. Involved objects were: {chain}");
+                }
+            }
 
+            navSet[depth] = (int)reference.ObjectNumber;
+            depth++;
 
             if (isDisposed)
             {
@@ -756,20 +769,20 @@
             }
 
             // Negative offsets refer to a stream with that number.
-            if (offset < 0)
+            if (offset.Type == XrefEntryType.ObjectStream)
             {
-                var result = GetObjectFromStream(reference, offset, ref searchDepth);
+                if (offset.Value1 == reference.ObjectNumber)
+                {
+                    throw new PdfDocumentFormatException(
+                        $"Object stream cannot contain itself, looking for object {reference} in {offset.Value1}");
+                }
+
+                var result = GetObjectFromStream(reference, offset, navSet, depth);
 
                 return result;
             }
 
-            if (offset == 0 && reference.Generation > ushort.MaxValue)
-            {
-                // TODO - To remove as should not happen anymore
-                return new ObjectToken(offset, reference, NullToken.Instance);
-            }
-
-            Seek(offset);
+            Seek(offset.Value1);
 
             if (!MoveNext())
             {
@@ -793,7 +806,7 @@
         {
             // Using 0 position as it isn't written to stream and this value doesn't
             // seem to be used by any callers. In future may need to revisit this.
-            overwrittenTokens[reference] = new ObjectToken(0, reference, token);
+            overwrittenTokens[reference] = new ObjectToken(XrefLocation.File(0), reference, token);
         }
 
         private bool TryBruteForceFileToFindReference(IndirectReference reference, [NotNullWhen(true)] out ObjectToken? result)
@@ -826,11 +839,11 @@
             }
         }
 
-        private ObjectToken GetObjectFromStream(IndirectReference reference, long offset, ref ushort searchDepth)
+        private ObjectToken GetObjectFromStream(IndirectReference reference, XrefLocation offset, Span<int> navSet, byte depth)
         {
-            var streamObjectNumber = offset * -1;
+            var streamObjectNumber = offset.Value1;
 
-            var streamObject = Get(new IndirectReference(streamObjectNumber, 0), ref searchDepth);
+            var streamObject = Get(new IndirectReference(streamObjectNumber, 0), navSet, depth);
 
             if (!(streamObject?.Data is StreamToken stream))
             {
@@ -853,7 +866,7 @@
             return result;
         }
 
-        private IReadOnlyList<ObjectToken> ParseObjectStream(StreamToken stream, long offset)
+        private IReadOnlyList<ObjectToken> ParseObjectStream(StreamToken stream, XrefLocation offset)
         {
             if (!stream.StreamDictionary.TryGet(NameToken.N, out var numberToken)
             || !(numberToken is NumericToken numberOfObjects))

src/UglyToad.PdfPig/Writer/Colors/OutputIntentsFactory.cs

@@ -15,7 +15,7 @@
 
             var profileBytes = ProfileStreamReader.GetSRgb2014();
 
-            var compressedBytes = DataCompresser.CompressBytes(profileBytes);
+            var compressedBytes = DataCompressor.CompressBytes(profileBytes);
 
             var profileStreamDictionary = new Dictionary<NameToken, IToken>
             {

src/UglyToad.PdfPig/Writer/DataCompressor.cs

@@ -6,7 +6,7 @@
     using Filters;
     using Tokens;
 
-    internal static class DataCompresser
+    internal static class DataCompressor
     {
         public static byte[] CompressBytes(IReadOnlyList<byte> bytes) => CompressBytes(bytes.ToArray());
         public static byte[] CompressBytes(byte[] bytes)
@@ -15,7 +15,7 @@
             {
                 var parameters = new DictionaryToken(new Dictionary<NameToken, IToken>());
                 var flater = new FlateFilter();
-                var result = flater.Encode(memoryStream, parameters, 0);
+                var result = flater.Encode(memoryStream, parameters);
                 return result;
             }
         }

src/UglyToad.PdfPig/Writer/Fonts/TrueTypeWritingFont.cs

@@ -51,7 +51,7 @@
             var newEncoding = new TrueTypeSubsetEncoding(characterMapping.Keys.ToList());
             var subsetBytes = TrueTypeSubsetter.Subset(fontFileBytes.ToArray(), newEncoding);
 
-            var embeddedFile = DataCompresser.CompressToStream(subsetBytes);
+            var embeddedFile = DataCompressor.CompressToStream(subsetBytes);
 
             var fileRef = writer.WriteToken(embeddedFile);
 
@@ -110,7 +110,7 @@
             var descriptor = writer.WriteToken(new DictionaryToken(descriptorDictionary));
 
             var toUnicodeCMap = ToUnicodeCMapBuilder.ConvertToCMapStream(characterMapping);
-            var toUnicodeStream = DataCompresser.CompressToStream(toUnicodeCMap);
+            var toUnicodeStream = DataCompressor.CompressToStream(toUnicodeCMap);
             var toUnicode = writer.WriteToken(toUnicodeStream);
 
             var dictionary = new Dictionary<NameToken, IToken>

src/UglyToad.PdfPig/Writer/NoTextTokenWriter.cs

@@ -107,7 +107,7 @@ namespace UglyToad.PdfPig.Writer
                 }
                 outputStreamT.Seek(0, SeekOrigin.Begin);
 
-                var compressedBytes = DataCompresser.CompressBytes(outputStreamT.ToArray());
+                var compressedBytes = DataCompressor.CompressBytes(outputStreamT.ToArray());
                 var outputStreamDictionary = new Dictionary<NameToken, IToken>()
                 {
                     { NameToken.Length, new NumericToken(compressedBytes.Length) },

src/UglyToad.PdfPig/Writer/PdfPageBuilder.cs

@@ -767,7 +767,7 @@
                     }
                 }
 
-                var compressedSmask = DataCompresser.CompressBytes(smaskData);
+                var compressedSmask = DataCompressor.CompressBytes(smaskData);
 
                 // Create a soft-mask.
                 var smaskDictionary = new Dictionary<NameToken, IToken>
@@ -786,7 +786,7 @@
                 smaskReference = documentBuilder.AddImage(new DictionaryToken(smaskDictionary), compressedSmask);
             }
 
-            var compressed = DataCompresser.CompressBytes(data);
+            var compressed = DataCompressor.CompressBytes(data);
 
             var imgDictionary = new Dictionary<NameToken, IToken>
             {
@@ -1218,7 +1218,7 @@
 
                     var bytes = memoryStream.ToArray();
 
-                    var stream = DataCompresser.CompressToStream(bytes);
+                    var stream = DataCompressor.CompressToStream(bytes);
 
                     return writer.WriteToken(stream);
                 }

src/UglyToad.PdfPig/Writer/PdfStreamWriter.cs

@@ -58,7 +58,7 @@
 
             var ir = ReserveObjectNumber();
             offsets.Add(ir.Data, Stream.Position);
-            var obj = new ObjectToken(Stream.Position, ir.Data, token);
+            var obj = new ObjectToken(XrefLocation.File(Stream.Position), ir.Data, token);
             TokenWriter.WriteToken(obj, Stream);
             return ir;
         }
@@ -71,7 +71,7 @@
             }
 
             offsets.Add(indirectReference.Data, Stream.Position);
-            var obj = new ObjectToken(Stream.Position, indirectReference.Data, token);
+            var obj = new ObjectToken(XrefLocation.File(Stream.Position), indirectReference.Data, token);
             TokenWriter.WriteToken(obj, Stream);
             return indirectReference;
         }
@@ -98,7 +98,6 @@
             TokenWriter.WriteCrossReferenceTable(offsets, catalogReference.Data, Stream, documentInformationReference?.Data);
         }
 
-
         public void Dispose()
         {
             if (DisposeStream)