Handle the case where a stream-length is reported incorrectly in a dictionary.

Author: packdat

PdfSharpCore/Pdf.IO/Lexer.cs

@@ -34,6 +34,8 @@
 using System.IO;
 using PdfSharpCore.Internal;
 using PdfSharpCore.Pdf.Internal;
+using System.Collections.Generic;
+using System.Linq;
 
 namespace PdfSharpCore.Pdf.IO
 {
@@ -170,6 +172,19 @@ public Symbol ScanNextToken()
         /// Reads the raw content of a stream.
         /// </summary>
         public byte[] ReadStream(int length)
+        {
+            var pos = MoveToStartOfStream();
+            _pdfSteam.Position = pos;
+            byte[] bytes = new byte[length];
+            int read = _pdfSteam.Read(bytes, 0, length);
+            Debug.Assert(read == length);
+
+            // Synchronize idxChar etc.
+            Position = pos + length;
+            return bytes;
+        }
+
+        internal long MoveToStartOfStream()
         {
             long pos;
 
@@ -187,15 +202,45 @@ public byte[] ReadStream(int length)
             }
             else
                 pos = _idxChar + 1;
+            return pos;
+        }
 
-            _pdfSteam.Position = pos;
-            byte[] bytes = new byte[length];
-            int read = _pdfSteam.Read(bytes, 0, length);
-            Debug.Assert(read == length);
+        /// <summary>
+        /// Scans the input stream for the specified marker.<br></br>
+        /// Returns the bytes from the current position up to the start of the marker or the end of the stream.<br></br>
+        /// The position of the input-stream is the byte right after the marker (if found) or the end of the stream.
+        /// </summary>
+        /// <param name="marker">The marker to scan for</param>
+        /// <param name="markerFound">Receives a boolean that indicates whether the marker was found</param>
+        /// <returns></returns>
+        internal byte[] ScanUntilMarker(byte[] marker, out bool markerFound)
+        {
+            markerFound = false;
+            var result = new List<byte>();
+            while (true)
+            {
+                var markerIndex = 0;
+                while (_currChar != Chars.EOF && _currChar != marker[markerIndex])
+                {
+                    result.Add((byte)_currChar);
+                    ScanNextChar(false);
+                }
+                while (_currChar != Chars.EOF && markerIndex < marker.Length && _currChar == marker[markerIndex])
+                {
+                    markerIndex++;
+                    ScanNextChar(false);
+                }
+                if (_currChar == Chars.EOF || markerIndex == marker.Length)
+                {
+                    if (markerIndex == marker.Length)
+                        markerFound = true;
+                    break;
+                }
+                // only part of the marker was found, add to result and continue
+                result.AddRange(marker.Take(markerIndex));
+            }
 
-            // Synchronize idxChar etc.
-            Position = pos + length;
-            return bytes;
+            return result.ToArray();
         }
 
         /// <summary>
@@ -722,7 +767,6 @@ public char MoveToNonWhiteSpace()
             }
             return _currChar;
         }
-
 // #if DEBUG
 //         public string SurroundingsOfCurrentPosition(bool hex)
 //         {

PdfSharpCore/Pdf.IO/Parser.cs

@@ -35,6 +35,7 @@
 using PdfSharpCore.Exceptions;
 using PdfSharpCore.Internal;
 using PdfSharpCore.Pdf.Advanced;
+using PdfSharpCore.Pdf.Internal;
 using PdfSharpCore.Pdf.IO.enums;
 
 namespace PdfSharpCore.Pdf.IO
@@ -272,6 +273,7 @@ public PdfObject ReadObject(PdfObject pdfObject, PdfObjectID objectID, bool incl
 #if true_
                 ReadStream(dict);
 #else
+                var startOfStream = _lexer.Position;
                 int length = GetStreamLength(dict);
                 byte[] bytes = _lexer.ReadStream(length);
 #if true_
@@ -301,7 +303,21 @@ public PdfObject ReadObject(PdfObject pdfObject, PdfObjectID objectID, bool incl
 #endif
                 PdfDictionary.PdfStream stream = new PdfDictionary.PdfStream(bytes, dict);
                 dict.Stream = stream;
-                ReadSymbol(Symbol.EndStream);
+                try
+                {
+                    ReadSymbol(Symbol.EndStream);
+                }
+                catch (PdfReaderException)
+                {
+                    // stream length may be incorrect, scan byte by byte up to the "endstream" keyword
+                    _lexer.Position = startOfStream;
+                    _lexer.Position = _lexer.MoveToStartOfStream();
+                    bytes = _lexer.ScanUntilMarker(PdfEncoders.RawEncoding.GetBytes("\nendstream"), out var markerFound);
+                    if (!markerFound)
+                        throw;
+                    stream = new PdfDictionary.PdfStream(bytes, dict);
+                    dict.Stream = stream;
+                }
                 symbol = ScanNextToken();
 #endif
                 if (symbol == Symbol.Eof)