Add CoreTokenScanner.ClearPreReadByte() and further fix #1332

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

@@ -11,6 +11,20 @@
 
     public class GithubIssuesTests
     {
+        [Fact]
+        public void Issues1332_1()
+        {
+            var path = IntegrationHelpers.GetSpecificTestDocumentPath("0012156.pdf");
+            using (var document = PdfDocument.Open(path, new ParsingOptions() { UseLenientParsing = true }))
+            {
+                foreach (var page in document.GetPages())
+                {
+                    Assert.NotNull(page);
+                    Assert.NotEmpty(page.Paths);
+                }
+            }
+        }
+
         [Fact]
         public void Issues1332()
         {

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

@@ -104,6 +104,17 @@ public void Seek(long position)
             inputBytes.Seek(position);
         }
 
+        /// <summary>
+        /// Discards any byte that was read ahead of the current token (see <see cref="ITokenizer.ReadsNextByte"/>).
+        /// This must be called after seeking to an absolute position where the next <see cref="MoveNext"/> is
+        /// expected to read the byte at that position, otherwise the stale pre-read byte from the previous
+        /// location is reused and tokenization starts one byte too late.
+        /// </summary>
+        public void ClearPreReadByte()
+        {
+            hasBytePreRead = false;
+        }
+
         /// <inheritdoc />
         public bool MoveNext()
         {

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

@@ -785,6 +785,8 @@ public void DeregisterCustomTokenizer(ITokenizer tokenizer)
 
             Seek(offset.Value1);
 
+            coreTokenScanner.ClearPreReadByte();
+
             if (!MoveNext())
             {
                 TryBruteForceFileToFindReference(reference, out var bfObjectToken);
@@ -817,9 +819,11 @@ private bool TryBruteForceFileToFindReference(IndirectReference reference, [NotN
             {
                 // Brute force read the entire file
                 isBruteForcing = true;
-                
+
                 Seek(fileHeaderOffset.Value);
 
+                coreTokenScanner.ClearPreReadByte();
+
                 while (MoveNext())
                 {
                     objectLocationProvider.Cache((ObjectToken)CurrentToken!, true);