Avoid doing a true file seek for simple peeking the next char in the token parser

Author: rhuijben

src/UglyToad.PdfPig.Core/StreamInputBytes.cs

@@ -11,11 +11,12 @@ public sealed class StreamInputBytes : IInputBytes
     {
         private readonly Stream stream;
         private readonly bool shouldDispose;
+        private byte? peekByte;
 
         private bool isAtEnd;
 
         /// <inheritdoc />
-        public long CurrentOffset => stream.Position;
+        public long CurrentOffset => peekByte.HasValue ? stream.Position - 1 : stream.Position;
 
         /// <inheritdoc />
         public byte CurrentByte { get; private set; }
@@ -52,7 +53,8 @@ public StreamInputBytes(Stream stream, bool shouldDispose = true)
         /// <inheritdoc />
         public bool MoveNext()
         {
-            var b = stream.ReadByte();
+            var b = peekByte ?? stream.ReadByte();
+            peekByte = null;
 
             if (b == -1)
             {
@@ -68,18 +70,21 @@ public bool MoveNext()
         /// <inheritdoc />
         public byte? Peek()
         {
-            var current = CurrentOffset;
-
-            var b = stream.ReadByte();
-
-            stream.Seek(current, SeekOrigin.Begin);
-
-            if (b == -1)
+            if (!peekByte.HasValue)
             {
-                return null;
+                var v = stream.ReadByte();
+
+                if (v >= 0)
+                {
+                    peekByte = (byte)v;
+                }
+                else
+                {
+                    return null;
+                }
             }
 
-            return (byte)b;
+            return peekByte;
         }
 
         /// <inheritdoc />
@@ -92,6 +97,7 @@ public bool IsAtEnd()
         public void Seek(long position)
         {
             isAtEnd = false;
+            peekByte = null;
 
             if (position == 0)
             {
@@ -112,9 +118,15 @@ public int Read(Span<byte> buffer)
             {
                 return 0;
             }
+            else if (peekByte.HasValue)
+            {
+                buffer[0] = peekByte.Value;
+                peekByte = null;
 
-            int read = stream.Read(buffer);
+                return Read(buffer.Slice(1)) + 1;
+            }
 
+            int read = stream.Read(buffer);
             if (read > 0)
             {
                 CurrentByte = buffer[read - 1];

src/UglyToad.PdfPig/Parser/PdfDocumentFactory.cs

@@ -25,9 +25,9 @@
 
     internal static class PdfDocumentFactory
     {
-        public static PdfDocument Open(byte[] fileBytes, ParsingOptions? options = null)
+        public static PdfDocument Open(ReadOnlyMemory<byte> memory, ParsingOptions? options = null)
         {
-            var inputBytes = new MemoryInputBytes(fileBytes);
+            var inputBytes = new MemoryInputBytes(memory);
 
             return Open(inputBytes, options);
         }

src/UglyToad.PdfPig/PdfDocument.cs

@@ -102,6 +102,14 @@ internal PdfDocument(
         /// <returns>A <see cref="PdfDocument"/> providing access to the file contents.</returns>
         public static PdfDocument Open(byte[] fileBytes, ParsingOptions? options = null) => PdfDocumentFactory.Open(fileBytes, options);
 
+        /// <summary>
+        /// Creates a <see cref="PdfDocument"/> for reading from the provided file bytes.
+        /// </summary>
+        /// <param name="memory">The bytes of the PDF file.</param>
+        /// <param name="options">Optional parameters controlling parsing.</param>
+        /// <returns>A <see cref="PdfDocument"/> providing access to the file contents.</returns>
+        public static PdfDocument Open(ReadOnlyMemory<byte> memory, ParsingOptions? options = null) => PdfDocumentFactory.Open(memory, options);
+
         /// <summary>
         /// Opens a file and creates a <see cref="PdfDocument"/> for reading from the provided file path.
         /// </summary>

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

@@ -342,14 +342,9 @@ private bool TryReadStream(long startStreamTokenOffset, bool getLength, [NotNull
 
                 if ((char)inputBytes.CurrentByte == '\r')
                 {
-                    if (!inputBytes.MoveNext())
+                    if (inputBytes.Peek() == '\n')
                     {
-                        return false;
-                    }
-
-                    if ((char)inputBytes.CurrentByte != '\n')
-                    {
-                        inputBytes.Seek(inputBytes.CurrentOffset - 1);
+                        inputBytes.MoveNext();
                     }
                     break;
                 }