#443 handle case where file version comment token included in string by tokenization

instead just brute force the raw content

src/UglyToad.PdfPig.Tests/Fonts/Parser/Parts/BaseFontRangeParserTests.cs

@@ -16,7 +16,7 @@
 
             var builder = new CharacterMapBuilder();
 
-            parser.Parse(new NumericToken(1), input, builder);
+            parser.Parse(new NumericToken(1), input.scanner, builder);
 
             Assert.Equal(2, builder.BaseFontCharacterMap.Count);
 
@@ -31,7 +31,7 @@
 
             var builder = new CharacterMapBuilder();
 
-            parser.Parse(new NumericToken(1), input, builder);
+            parser.Parse(new NumericToken(1), input.scanner, builder);
 
             Assert.Equal(7, builder.BaseFontCharacterMap.Count);
 
@@ -47,7 +47,7 @@
 
             var builder = new CharacterMapBuilder();
 
-            parser.Parse(new NumericToken(2), input, builder);
+            parser.Parse(new NumericToken(2), input.scanner, builder);
 
             Assert.Equal(6, builder.BaseFontCharacterMap.Count);
 

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

@@ -5,6 +5,8 @@
     using PdfPig.Core;
     using PdfPig.Parser.FileStructure;
     using PdfPig.Tokenization.Scanner;
+    using PdfPig.Tokens;
+    using System.Linq;
     using Xunit;
 
     public class FileHeaderParserTests
@@ -13,7 +15,7 @@
         [Fact]
         public void NullScannerThrows()
         {
-            Action action = () => FileHeaderParser.Parse(null, false, log);
+            Action action = () => FileHeaderParser.Parse(null, null, false, log);
 
             Assert.Throws<ArgumentNullException>(action);
         }
@@ -31,7 +33,7 @@
 
             var scanner = StringBytesTestConverter.Scanner(input);
 
-            var result = FileHeaderParser.Parse(scanner, false, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
 
             Assert.Equal(format, result.VersionString);
             Assert.Equal(0, result.OffsetInFile);
@@ -46,7 +48,7 @@
 
             var scanner = StringBytesTestConverter.Scanner(input);
 
-            var result = FileHeaderParser.Parse(scanner, false, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
 
             Assert.Equal(1.2m, result.Version);
             Assert.Equal(TestEnvironment.IsUnixPlatform ? 7 : 9, result.OffsetInFile);
@@ -57,7 +59,7 @@
         {
             var scanner = StringBytesTestConverter.Scanner(string.Empty);
             
-            Action action = () => FileHeaderParser.Parse(scanner, false, log);
+            Action action = () => FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
 
             Assert.Throws<PdfDocumentFormatException>(action);
         }
@@ -68,7 +70,7 @@
             var scanner = StringBytesTestConverter.Scanner(@"one    
     %PDF-1.2");
 
-            var result = FileHeaderParser.Parse(scanner, false, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
 
             Assert.Equal(1.2m, result.Version);
             Assert.Equal(TestEnvironment.IsUnixPlatform ? 12 : 13, result.OffsetInFile);
@@ -80,7 +82,7 @@
             var scanner = StringBytesTestConverter.Scanner(@"one    
     %PDF-1.7");
 
-            var result = FileHeaderParser.Parse(scanner, true, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, true, log);
 
             Assert.Equal(1.7m, result.Version);
             Assert.Equal(TestEnvironment.IsUnixPlatform ? 12 : 13, result.OffsetInFile);
@@ -92,7 +94,7 @@
             var scanner = StringBytesTestConverter.Scanner(@"one two
 three %PDF-1.6");
 
-            var result = FileHeaderParser.Parse(scanner, true, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, true, log);
 
             Assert.Equal(1.6m, result.Version);
             Assert.Equal(TestEnvironment.IsUnixPlatform ? 14 : 15, result.OffsetInFile);
@@ -103,7 +105,7 @@ three %PDF-1.6");
         {
             var scanner = StringBytesTestConverter.Scanner(@"one two");
 
-            Action action = () => FileHeaderParser.Parse(scanner, true, log);
+            Action action = () => FileHeaderParser.Parse(scanner.scanner, scanner.bytes, true, log);
 
             Assert.Throws<PdfDocumentFormatException>(action);
         }
@@ -113,7 +115,7 @@ three %PDF-1.6");
         {
             var scanner = StringBytesTestConverter.Scanner("%Pdeef-1.69");
 
-            Action action = () => FileHeaderParser.Parse(scanner, false, log);
+            Action action = () => FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
 
             Assert.Throws<PdfDocumentFormatException>(action);
         }
@@ -123,7 +125,7 @@ three %PDF-1.6");
         {
             var scanner = StringBytesTestConverter.Scanner("%Pdeef-1.69");
 
-            var result = FileHeaderParser.Parse(scanner, true, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, true, log);
 
             Assert.Equal(1.4m, result.Version);
         }
@@ -133,9 +135,9 @@ three %PDF-1.6");
         {
             var scanner = StringBytesTestConverter.Scanner(@"%FDF-1.6");
 
-            var result = FileHeaderParser.Parse(scanner, false, log);
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
 
-            Assert.Equal(0, scanner.CurrentPosition);
+            Assert.Equal(0, scanner.scanner.CurrentPosition);
             Assert.Equal(0, result.OffsetInFile);
         }
 
@@ -144,11 +146,33 @@ three %PDF-1.6");
         {
             var input = OtherEncodings.StringAsLatin1Bytes("%PDF-1.7\r\n%âãÏÓ\r\n1 0 obj\r\n<</Lang(en-US)>>\r\nendobj");
 
-            var scanner = new CoreTokenScanner(new ByteArrayInputBytes(input), ScannerScope.None);
+            var bytes = new ByteArrayInputBytes(input);
 
-            var result = FileHeaderParser.Parse(scanner, false, log);
+            var scanner = new CoreTokenScanner(bytes, ScannerScope.None);
+
+            var result = FileHeaderParser.Parse(scanner, bytes, false, log);
 
             Assert.Equal(1.7m, result.Version);
         }
+
+        [Fact]
+        public void Issue443()
+        {
+            const string hex =
+                @"00 0F 4A 43 42 31 33 36 36 31 32 32 37 2E 70 64 66 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 50 44 46 20 43 41 52 4F 01 00 FF FF FF FF 00 00 00 00 00 04 DF 28 00 00 00 00 AF 51 7E 82 AF 52 D7 09 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 81 81 03 0D 00 00 25 50 44 46 2D 31 2E 31 0A 25 E2 E3 CF D3 0D 0A 31 20 30 20 6F 62 6A";
+
+            var bytes = hex.Split(' ', StringSplitOptions.RemoveEmptyEntries).Select(x => HexToken.Convert(x[0], x[1]));
+
+            var str = OtherEncodings.BytesAsLatin1String(bytes.ToArray());
+
+            var scanner = StringBytesTestConverter.Scanner(str);
+
+            var result = FileHeaderParser.Parse(scanner.scanner, scanner.bytes, false, log);
+
+            Assert.Equal(0, scanner.scanner.CurrentPosition);
+            Assert.Equal(129, result.OffsetInFile);
+            Assert.Equal(1.1m, result.Version);
+            Assert.Equal("PDF-1.1", result.VersionString);
+        }
     }
 }

src/UglyToad.PdfPig.Tests/Parser/Parts/FileStructure/CrossReferenceTableParserTests.cs

@@ -25,7 +25,7 @@
 trailer
 << >>");
 
-            var result = CrossReferenceTableParser.Parse(input, 4, false);
+            var result = CrossReferenceTableParser.Parse(input.scanner, 4, false);
 
             Assert.Equal(4, result.ObjectOffsets.Count);
         }
@@ -288,7 +288,7 @@ trailer
 
         private static CoreTokenScanner GetReader(string input)
         {
-            return StringBytesTestConverter.Scanner(input);
+            return StringBytesTestConverter.Scanner(input).scanner;
         }
     }
 }

src/UglyToad.PdfPig.Tests/StringBytesTestConverter.cs

@@ -31,11 +31,12 @@
             public IInputBytes Bytes { get; set; }
         }
 
-        internal static CoreTokenScanner Scanner(string s)
+        internal static (CoreTokenScanner scanner, IInputBytes bytes) Scanner(string s)
         {
-            var result = new CoreTokenScanner(new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(s)));
+            var inputBytes = new ByteArrayInputBytes(OtherEncodings.StringAsLatin1Bytes(s));
+            var result = new CoreTokenScanner(inputBytes);
 
-            return result;
+            return (result, inputBytes);
         }
     }
 }

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

@@ -29,40 +29,41 @@
     internal static class FileHeaderParser
     {
         [NotNull]
-        public static HeaderVersion Parse([NotNull]ISeekableTokenScanner scanner, bool isLenientParsing, ILog log)
+        public static HeaderVersion Parse([NotNull] ISeekableTokenScanner scanner, IInputBytes inputBytes, bool isLenientParsing, ILog log)
         {
             if (scanner == null)
             {
                 throw new ArgumentNullException(nameof(scanner));
             }
 
-            // Read the first token
+            var startPosition = scanner.CurrentPosition;
-            if (!scanner.MoveNext())
-            {
-                throw new PdfDocumentFormatException($"Could not read the first token in the document at position {scanner.CurrentPosition}.");
-            }
 
-            var comment = scanner.CurrentToken as CommentToken;
+            const int junkTokensTolerance = 30;
-
-            const int junkTokensTolerance = 25;
             var attempts = 0;
-            while (comment == null)
+            CommentToken comment;
+            do
             {
-                if (attempts == junkTokensTolerance)
+                if (attempts == junkTokensTolerance || !scanner.MoveNext())
                 {
-                    throw new PdfDocumentFormatException("Could not find the version header comment at the start of the document.");
+                    if (!TryBruteForceVersionLocation(startPosition, inputBytes, out var version))
-                }
+                    {
+                        throw new PdfDocumentFormatException("Could not find the version header comment at the start of the document.");
+                    }
 
-                if (!scanner.MoveNext())
+                    scanner.Seek(startPosition);
-                {
+                    return version;
-                    throw new PdfDocumentFormatException("Could not find the version header comment at the start of the document.");
                 }
 
                 comment = scanner.CurrentToken as CommentToken;
 
                 attempts++;
-            }
+            } while (comment == null);
 
+            return GetHeaderVersionAndResetScanner(comment, scanner, isLenientParsing, log);
+        }
+
+        private static HeaderVersion GetHeaderVersionAndResetScanner(CommentToken comment, ISeekableTokenScanner scanner, bool isLenientParsing, ILog log)
+        {
             if (comment.Data.IndexOf("PDF-1.", StringComparison.OrdinalIgnoreCase) != 0 && comment.Data.IndexOf("FDF-1.", StringComparison.OrdinalIgnoreCase) != 0)
             {
                 return HandleMissingVersion(comment, isLenientParsing, log);
@@ -70,7 +71,7 @@
 
             const int toDecimalStartLength = 4;
 
-            if (!decimal.TryParse(comment.Data.Substring(toDecimalStartLength), 
+            if (!decimal.TryParse(comment.Data.Substring(toDecimalStartLength),
                 NumberStyles.Number,
                 CultureInfo.InvariantCulture,
                 out var version))
@@ -90,6 +91,61 @@
             return result;
         }
 
+        private static bool TryBruteForceVersionLocation(long startPosition, IInputBytes inputBytes, out HeaderVersion headerVersion)
+        {
+            headerVersion = null;
+
+            inputBytes.Seek(startPosition);
+
+            // %PDF-x.y or %FDF-x.y
+            const int versionLength = 8;
+            const int bufferLength = 64;
+
+            // Slide a window of bufferLength bytes across the file allowing for the fact the version could get split by
+            // the window (so always ensure an overlap of versionLength bytes between the end of the previous and start of the next buffer).
+            var buffer = new byte[bufferLength];
+
+            var currentOffset = startPosition;
+            int readLength;
+            do
+            {
+                readLength = inputBytes.Read(buffer, bufferLength);
+
+                var content = OtherEncodings.BytesAsLatin1String(buffer);
+
+                var pdfIndex = content.IndexOf("%PDF-", StringComparison.OrdinalIgnoreCase);
+                var fdfIndex = content.IndexOf("%FDF-", StringComparison.OrdinalIgnoreCase);
+                var actualIndex = pdfIndex >= 0 ? pdfIndex : fdfIndex;
+
+                if (actualIndex >= 0 && content.Length - actualIndex >= versionLength)
+                {
+                    var numberPart = content.Substring(actualIndex + 5, 3);
+                    if (decimal.TryParse(
+                            numberPart,
+                            NumberStyles.Number,
+                            CultureInfo.InvariantCulture,
+                            out var version))
+                    {
+                        var afterCommentSymbolIndex = actualIndex + 1;
+
+                        headerVersion = new HeaderVersion(
+                            version,
+                            content.Substring(afterCommentSymbolIndex, versionLength - 1),
+                            currentOffset + afterCommentSymbolIndex);
+
+                        inputBytes.Seek(startPosition);
+
+                        return true;
+                    }
+                }
+
+                currentOffset += readLength - versionLength;
+                inputBytes.Seek(currentOffset);
+            } while (readLength == bufferLength);
+
+            return false;
+        }
+
         private static HeaderVersion HandleMissingVersion(CommentToken comment, bool isLenientParsing, ILog log)
         {
             if (isLenientParsing)

src/UglyToad.PdfPig/Parser/PdfDocumentFactory.cs

@@ -97,7 +97,7 @@
             var crossReferenceStreamParser = new CrossReferenceStreamParser(filterProvider);
             var crossReferenceParser = new CrossReferenceParser(log, xrefValidator, crossReferenceStreamParser);
 
-            var version = FileHeaderParser.Parse(scanner, isLenientParsing, log);
+            var version = FileHeaderParser.Parse(scanner, inputBytes, isLenientParsing, log);
 
             var crossReferenceOffset = FileTrailerParser.GetFirstCrossReferenceOffset(inputBytes, scanner,
                 isLenientParsing) + version.OffsetInFile;