Esempio n. 1
0
        private CosBase ParseObjectFromFile(long offset, IRandomAccessRead reader,
                                            CosObjectKey key,
                                            CosObjectPool pool,
                                            bool isLenientParsing)
        {
            reader.Seek(offset);

            var objectNumber     = ObjectHelper.ReadObjectNumber(reader);
            var objectGeneration = ObjectHelper.ReadGenerationNumber(reader);

            ReadHelper.ReadExpectedString(reader, "obj", true);

            if (objectNumber != key.Number || objectGeneration != key.Generation)
            {
                throw new InvalidOperationException($"Xref for {key} points to object {objectNumber} {objectGeneration} at {offset}");
            }

            ReadHelper.SkipSpaces(reader);

            var baseObject = baseParser.Parse(reader, pool);

            var endObjectKey = ReadHelper.ReadString(reader);

            var atStreamStart = string.Equals(endObjectKey, "stream");

            if (atStreamStart)
            {
                var streamStartBytes = OtherEncodings.StringAsLatin1Bytes(endObjectKey);

                reader.Rewind(streamStartBytes.Length);

                baseObject = ReadNormalObjectStream(reader, baseObject, offset, isLenientParsing, out endObjectKey);
            }

            if (!string.Equals(endObjectKey, "endobj"))
            {
                var message =
                    $"Object ({objectNumber}:{objectGeneration}) at offset {offset} does not end with \'endobj\' but with \'{endObjectKey}\'";

                if (isLenientParsing)
                {
                    log.Warn(message);
                }
                else
                {
                    throw new InvalidOperationException(message);
                }
            }

            return(baseObject);
        }
 public string ReadString()
 {
     return(mReadHelper.ReadString());
 }
            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);
            }