Handle incorrect/missing length of stream-objects. Fixes empira#73

src/foundation/src/PDFsharp/src/PdfSharp/Pdf.IO/Lexer.cs

@@ -152,6 +152,7 @@ public Symbol ScanNextToken()
         /// </summary>
         public byte[] ReadStream(int length)
         {
+            // TODO: use "MoveToStartOfStream()"
             int pos;
 
             // Skip illegal blanks behind «stream».
@@ -166,8 +167,10 @@ public byte[] ReadStream(int length)
                 else
                     pos = _idxChar + 1;
             }
-            else
+            else if (_currChar == Chars.LF)
                 pos = _idxChar + 1;
+            else
+                pos = _idxChar;
 
             _pdfStream.Position = pos;
             byte[] bytes = new byte[length];
@@ -184,6 +187,102 @@ public byte[] ReadStream(int length)
             return bytes;
         }
 
+        /// <summary>
+        /// Moves the current position to the first byte of a stream-object.<br></br>
+        /// The current position is expected be be located right after the "stream" keyword.
+        /// </summary>
+        /// <returns>The position of the first byte of a stream</returns>
+        private int MoveToStartOfStream()
+        {
+            int pos;
+
+            // Skip illegal blanks behind «stream».
+            while (_currChar == Chars.SP)
+                ScanNextChar(true);
+
+            // Skip new line behind «stream».
+            if (_currChar == Chars.CR)
+            {
+                if (_nextChar == Chars.LF)
+                    pos = _idxChar + 2;
+                else
+                    pos = _idxChar + 1;
+            }
+            else if (_currChar == Chars.LF)
+                pos = _idxChar + 1;
+            else
+                pos = _idxChar;
+            return pos;
+        }
+
+        /// <summary>
+        /// Searches for the end of a stream, starting at the current position.<br></br>
+        /// The current position is expected be be located right after the "stream" keyword.<br></br>
+        /// After the search, the position in the input-stream is the first byte of the stream (if found) or the end of the input-stream.
+        /// </summary>
+        /// <param name="streamLength">Receives the length of the stream, if the end of the stream was found</param>
+        /// <returns><b>true</b> if the end of a stream was found, otherwise <b>false</b></returns>
+        internal bool SearchEndOfStream(out int streamLength)
+        {
+            // locate first byte of the stream
+            Position = MoveToStartOfStream();
+
+            streamLength = 0;
+            var streamStart = Position;
+            // "endstream" shall be preceded by an end-of-line marker according to the spec
+            var marker = PdfEncoders.RawEncoding.GetBytes("\nendstream");
+            if (!SearchForMarker(marker, out var markerPosition))
+                return false;
+
+            // the "endstream" keyword may be preceded by either LF or CR+LF
+            // check the byte right before the marker for CR (LF is already included in the marker)
+            var length = markerPosition - streamStart;
+            _pdfStream.Position = markerPosition - 1;
+            var prefix = _pdfStream.ReadByte();
+            if (prefix == 13)
+                length--;           // exclude CR from stream-data
+            streamLength = length;
+            // reset position to start of stream
+            Position = streamStart;
+            return true;
+        }
+
+        /// <summary>
+        /// Scans the input stream for the specified marker, starting from the current position.<br></br>
+        /// After the search, the position in the input-stream is the byte right after the marker (if found) or the end of the input-stream.
+        /// </summary>
+        /// <param name="marker">The marker to scan for</param>
+        /// <param name="markerPosition">Receives the position of the marker in the input stream (if found)</param>
+        /// <returns><b>true</b> if the marker was found, otherwise <b>false</b></returns>
+        internal bool SearchForMarker(byte[] marker, out int markerPosition)
+        {
+            while (true)
+            {
+                var markerIndex = 0;
+                while (_currChar != Chars.EOF && _currChar != marker[markerIndex])
+                {
+                    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)
+                    {
+                        markerPosition = Position - marker.Length;
+                        return true;
+                    }
+                    break;
+                }
+                // only part of the marker was found, continue
+            }
+            markerPosition = 0;
+            return false;
+        }
+
         /// <summary>
         /// Reads a string in raw encoding.
         /// </summary>

src/foundation/src/PDFsharp/src/PdfSharp/Pdf.IO/Parser.cs

@@ -54,10 +54,11 @@ public Parser(PdfDocument document)
         /// </summary>
         public int MoveToObject(PdfObjectID objectID)
         {
-            int position = _document.IrefTable[objectID]?.Position ?? throw new AggregateException("Invalid object ID.");
+            int position = _document.IrefTable[objectID]?.Position 
+                ?? throw new PdfReaderException($"Invalid object ID. The object ({objectID}) could not be located in the reference-table.");
             if (position < 0)
             {
-                throw new AggregateException($"Invalid position {position} for object ID {objectID}.");
+                throw new PdfReaderException($"Invalid position {position} for object ID {objectID}.");
             }
             return _lexer.Position = position;
         }
@@ -274,8 +275,11 @@ public PdfObject ReadObject(PdfObject? pdfObject, PdfObjectID objectID, bool inc
 #if true_
                 ReadStream(dict);
 #else
-                int length = GetStreamLength(dict!); // NRT HACK
-                byte[] bytes = _lexer.ReadStream(length);
+                var startOfStream = _lexer.Position;
+                try
+                {
+                    int length = GetStreamLength(dict!); // NRT HACK
+                    byte[] bytes = _lexer.ReadStream(length);
 #if true_
                 if (dict.Elements.GetString("/Filter") == "/FlateDecode")
                 {
@@ -301,9 +305,26 @@ public PdfObject ReadObject(PdfObject? pdfObject, PdfObjectID objectID, bool inc
                 End: ;
                 }
 #endif
-                var stream = new PdfDictionary.PdfStream(bytes, dict ?? NRT.ThrowOnNull<PdfDictionary>());
-                dict.Stream = stream;
-                ReadSymbol(Symbol.EndStream);
+                    var stream = new PdfStream(bytes, dict ?? NRT.ThrowOnNull<PdfDictionary>());
+                    dict.Stream = stream;
+                    ReadSymbol(Symbol.EndStream);
+                }
+                catch (NotImplementedException)
+                {
+                    throw;  // File-streams
+                }
+                catch
+                {
+                    // most likely there is a problem with the stream-length.
+                    // it may be incorrect or completely missing.
+                    // scan manually for the end of the stream
+                    _lexer.Position = startOfStream;
+                    if (!_lexer.SearchEndOfStream(out var streamLength))
+                        throw;
+                    var bytes = _lexer.ReadStream(streamLength);
+                    dict!.Stream = new PdfStream(bytes, dict ?? NRT.ThrowOnNull<PdfDictionary>());
+                    ReadSymbol(Symbol.EndStream);
+                }
                 symbol = ScanNextToken();
 #endif
             }