Ejemplo n.º 1
0
        public void SetsPropertiesCorrectly()
        {
            var definition = new TableSubsectionDefinition(5, 12);

            Assert.Equal(5, definition.FirstNumber);
            Assert.Equal(12, definition.Count);
        }
Ejemplo n.º 2
0
        public void SecondPartInvalidFormatFalse()
        {
            var input = StringBytesTestConverter.Convert("85 9t", false);

            var result = TableSubsectionDefinition.TryRead(log, input.Bytes, out var _);

            Assert.False(result);
        }
Ejemplo n.º 3
0
        public void TryReadIncorrectFormatMultiplePartsFalse()
        {
            var input = StringBytesTestConverter.Convert("76362 100 1000", false);

            var result = TableSubsectionDefinition.TryRead(log, input.Bytes, out var _);

            Assert.False(result);
        }
Ejemplo n.º 4
0
        public void SecondPartInvalidFormatFalse()
        {
            var bytes = OtherEncodings.StringAsLatin1Bytes("85 9t");

            var input = new RandomAccessBuffer(bytes);

            var result = TableSubsectionDefinition.TryRead(log, input, out var _);

            Assert.False(result);
        }
Ejemplo n.º 5
0
        public void TryReadIncorrectFormatMultiplePartsFalse()
        {
            var bytes = OtherEncodings.StringAsLatin1Bytes(@"76362 100 1000");

            var input = new RandomAccessBuffer(bytes);

            var result = TableSubsectionDefinition.TryRead(log, input, out var _);

            Assert.False(result);
        }
Ejemplo n.º 6
0
        public void ValidWithLongTrue()
        {
            var input = StringBytesTestConverter.Convert("214748364700 6", false);

            var result = TableSubsectionDefinition.TryRead(log, input.Bytes, out var definition);

            Assert.True(result);

            Assert.Equal(214748364700L, definition.FirstNumber);
            Assert.Equal(6, definition.Count);
        }
Ejemplo n.º 7
0
        public void ValidTrue()
        {
            var input = StringBytesTestConverter.Convert("12 32", false);

            var result = TableSubsectionDefinition.TryRead(log, input.Bytes, out var definition);

            Assert.True(result);

            Assert.Equal(12, definition.FirstNumber);
            Assert.Equal(32, definition.Count);
        }
Ejemplo n.º 8
0
        public void ValidTrue()
        {
            var bytes = OtherEncodings.StringAsLatin1Bytes("12 32");

            var input = new RandomAccessBuffer(bytes);

            var result = TableSubsectionDefinition.TryRead(log, input, out var definition);

            Assert.True(result);

            Assert.Equal(12, definition.FirstNumber);
            Assert.Equal(32, definition.Count);
        }
Ejemplo n.º 9
0
        public void ValidWithLongTrue()
        {
            var bytes = OtherEncodings.StringAsLatin1Bytes("214748364700 6");

            var input = new RandomAccessBuffer(bytes);

            var result = TableSubsectionDefinition.TryRead(log, input, out var definition);

            Assert.True(result);

            Assert.Equal(214748364700L, definition.FirstNumber);
            Assert.Equal(6, definition.Count);
        }
Ejemplo n.º 10
0
        public CrossReferenceTablePart Parse(ISeekableTokenScanner scanner, long offset, bool isLenientParsing)
        {
            var builder = new CrossReferenceTablePartBuilder
            {
                Offset   = offset,
                XRefType = CrossReferenceType.Table
            };

            if (scanner.CurrentPosition != offset)
            {
                scanner.Seek(offset);
            }

            scanner.MoveNext();

            if (scanner.CurrentToken is OperatorToken operatorToken)
            {
                if (operatorToken.Data == "xref")
                {
                    scanner.MoveNext();
                }
                else
                {
                    throw new PdfDocumentFormatException($"Unexpected operator in xref position: {operatorToken}.");
                }
            }

            if (scanner.CurrentToken is NumericToken firstObjectNumber)
            {
                if (!scanner.TryReadToken(out NumericToken objectCount))
                {
                    throw new PdfDocumentFormatException($"Unexpected token following xref and {firstObjectNumber}. We found: {scanner.CurrentToken}.");
                }

                var definition = new TableSubsectionDefinition(firstObjectNumber.Long, objectCount.Int);

                var tokenizer = new EndOfLineTokenizer();

                scanner.RegisterCustomTokenizer((byte)'\r', tokenizer);
                scanner.RegisterCustomTokenizer((byte)'\n', tokenizer);

                var readingLine = false;
                var tokens      = new List <IToken>();
                var count       = 0;
                while (scanner.MoveNext())
                {
                    if (scanner.CurrentToken is EndOfLineToken)
                    {
                        if (!readingLine)
                        {
                            continue;
                        }

                        readingLine = false;

                        count = ProcessTokens(tokens, scanner, builder, isLenientParsing, count, ref definition);

                        tokens.Clear();

                        continue;
                    }

                    if (scanner.CurrentToken is CommentToken)
                    {
                        continue;
                    }

                    var isLineOperator = scanner.CurrentToken is OperatorToken op && (op.Data == FreeEntry || op.Data == InUseEntry);

                    if (!(scanner.CurrentToken is NumericToken) && !isLineOperator)
                    {
                        break;
                    }

                    readingLine = true;
                    tokens.Add(scanner.CurrentToken);
                }

                if (tokens.Count > 0)
                {
                    ProcessTokens(tokens, scanner, builder, isLenientParsing, count, ref definition);
                }

                scanner.DeregisterCustomTokenizer(tokenizer);
            }

            builder.Dictionary = ParseTrailer(scanner, isLenientParsing);

            return(builder.Build());
        }
Ejemplo n.º 11
0
        private static int ProcessTokens(List <IToken> tokens, ISeekableTokenScanner scanner, CrossReferenceTablePartBuilder builder, bool isLenientParsing,
                                         int objectCount, ref TableSubsectionDefinition definition)
        {
            string GetErrorMessage()
            {
                var representation = "Invalid line format in xref table: [" + string.Join(", ", tokens.Select(x => x.ToString())) + "]";

                return(representation);
            }

            if (objectCount == definition.Count)
            {
                if (tokens.Count == 2)
                {
                    if (tokens[0] is NumericToken newFirstObjectToken && tokens[1] is NumericToken newObjectCountToken)
                    {
                        definition = new TableSubsectionDefinition(newFirstObjectToken.Long, newObjectCountToken.Int);

                        return(0);
                    }
                }

                throw new PdfDocumentFormatException($"Found a line with 2 unexpected entries in the cross reference table: {tokens[0]}, {tokens[1]}.");
            }

            if (tokens.Count <= 2)
            {
                if (!isLenientParsing)
                {
                    throw new PdfDocumentFormatException(GetErrorMessage());
                }

                return(objectCount);
            }

            var lastToken = tokens[tokens.Count - 1];

            if (lastToken is OperatorToken operatorToken)
            {
                if (operatorToken.Data == FreeEntry)
                {
                    return(objectCount + 1);
                }

                if (operatorToken.Data != InUseEntry)
                {
                    if (!isLenientParsing)
                    {
                        throw new PdfDocumentFormatException(GetErrorMessage());
                    }

                    return(objectCount);
                }

                if (tokens[0] is NumericToken offset && tokens[1] is NumericToken generationNumber)
                {
                    if (offset.Long >= builder.Offset && offset.Long <= scanner.CurrentPosition)
                    {
                        throw new PdfDocumentFormatException($"Object offset {offset} is within its own cross-reference table for object {definition.FirstNumber + objectCount}");
                    }

                    builder.Add(definition.FirstNumber + objectCount, generationNumber.Int, offset.Long);

                    return(objectCount + 1);
                }
            }
            else
            {
                if (!isLenientParsing)
                {
                    throw new PdfDocumentFormatException(GetErrorMessage());
                }
            }

            return(objectCount);
        }
Ejemplo n.º 12
0
            public bool TryParse(IRandomAccessRead source, long offset, bool isLenientParsing, CosObjectPool pool, out CrossReferenceTablePartBuilder builder)
            {
                builder = null;

                var tableStartOffset = source.GetPosition();

                if (source.Peek() != 'x')
                {
                    return(false);
                }

                var xref = ReadHelper.ReadString(source);

                if (!xref.Trim().Equals("xref"))
                {
                    return(false);
                }

                // check for trailer after xref
                var str = ReadHelper.ReadString(source);

                byte[] b = OtherEncodings.StringAsLatin1Bytes(str);

                source.Rewind(b.Length);

                if (str.StartsWith("trailer"))
                {
                    log.Warn("skipping empty xref table");
                    return(false);
                }

                builder = new CrossReferenceTablePartBuilder
                {
                    Offset   = offset,
                    XRefType = CrossReferenceType.Table
                };

                // Tables can have multiple sections. Each starts with a starting object id and a count.
                while (true)
                {
                    if (!TableSubsectionDefinition.TryRead(log, source, out var subsectionDefinition))
                    {
                        log.Warn($"Unexpected subsection definition in the cross-reference table at offset {offset}");

                        if (isLenientParsing)
                        {
                            break;
                        }

                        return(false);
                    }

                    var currentObjectId = subsectionDefinition.FirstNumber;

                    ReadHelper.SkipSpaces(source);
                    for (var i = 0; i < subsectionDefinition.Count; i++)
                    {
                        if (source.IsEof() || ReadHelper.IsEndOfName((char)source.Peek()))
                        {
                            break;
                        }

                        if (source.Peek() == 't')
                        {
                            break;
                        }

                        //Ignore table contents
                        var currentLine = ReadHelper.ReadLine(source);
                        var splitString = currentLine.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
                        if (splitString.Length < 3)
                        {
                            log.Warn("invalid xref line: " + currentLine);
                            break;
                        }

                        // This supports the corrupt table as reported in PDFBOX-474 (XXXX XXX XX n)
                        if (splitString[splitString.Length - 1].Equals(InUseEntry))
                        {
                            try
                            {
                                var objectOffset = long.Parse(splitString[0]);

                                if (objectOffset >= tableStartOffset && objectOffset <= source.GetPosition())
                                {
                                    // PDFBOX-3923: offset points inside this table - that can't be good
                                    throw new InvalidOperationException(
                                              $"Object offset {objectOffset} is within its own cross-reference table for object {currentObjectId}");
                                }

                                var generation = int.Parse(splitString[1]);
                                builder.Add(currentObjectId, generation, objectOffset);
                            }
                            catch (FormatException e)
                            {
                                throw new InvalidOperationException("Bad", e);
                            }
                        }
                        else if (!splitString[2].Equals(FreeEntry))
                        {
                            throw new InvalidOperationException(
                                      $"Corrupt cross-reference table entry for object {currentObjectId}. The indicator was not 'n' or 'f' but {splitString[2]}.");
                        }

                        currentObjectId++;

                        ReadHelper.SkipSpaces(source);
                    }

                    ReadHelper.SkipSpaces(source);
                    if (!ReadHelper.IsDigit(source))
                    {
                        break;
                    }
                }

                if (!TryParseTrailer(source, isLenientParsing, pool, out var trailer))
                {
                    throw new InvalidOperationException($"Something went wrong trying to read the XREF table at {offset}.");
                }

                builder.Dictionary = trailer;
                builder.Previous   = trailer.GetLongOrDefault(CosName.PREV);

                return(true);
            }
Ejemplo n.º 13
0
        public void ToStringRepresentsPdfForm()
        {
            var definition = new TableSubsectionDefinition(420, 69);

            Assert.Equal("420 69", definition.ToString());
        }