private MobipocketDocument(PalmDatabaseFormat pdb)
{
if (pdb == null)
{
throw new ArgumentNullException("pdb");
}
if (pdb.Records.Count < 1)
{
throw new Exception("palm database must contain at least one record");
}
//set pdb
this._pdb = pdb;
//parse header
this.Header = MobipocketHeader.FromPalmDatabase(_pdb);
//parse records
_records.AddRange(ParseMobipocketRecords());
}
/// <summary>
/// Creates new instance of mobipocket header from provided bytes
/// </summary>
/// <param name="bytes"></param>
/// <returns></returns>
public static MobipocketHeader FromBinary(ArraySegment<byte> bytes)
{
if (bytes.Array == null)
throw new ArgumentException("bytes array cannot be null");
var header = new MobipocketHeader();
try
{
using (var stream = new MemoryStream(bytes.Array, bytes.Offset, bytes.Count))
using (var reader = new BinaryReader(stream))
{
if (bytes.Count < 0x80)
throw new Exception("header is to small");
//palmdoc header start
//Get compression
header.Compression = (MobipocketCompression)Enum.ToObject(typeof(MobipocketCompression), reader.ReadUInt16(Endian.BigEndian));
//usnused zeroes
reader.BaseStream.Seek(2, SeekOrigin.Current);
//length of text
header.TotalTextLength = reader.ReadUInt32(Endian.BigEndian);
//count of dpb records that contain text
header.TextRecordCount = reader.ReadUInt16(Endian.BigEndian);
//maximum record size, always 4096
header.MaxRecordSize = reader.ReadUInt16(Endian.BigEndian);
//read mobipocket encryption
header.Encryption = (MobipocketEncryption)Enum.ToObject(typeof(MobipocketEncryption), reader.ReadUInt16(Endian.BigEndian));
//skip two zeroes
reader.BaseStream.Seek(2, SeekOrigin.Current);
//palmdoc header end
//mobi header start
//read identifier "MOBI"
var identifier = reader.ReadUInt32(Endian.BigEndian);
if (MobiIdentifier != identifier)
throw new Exception("unsupported identifier: " + identifier);
//read byte length of header
var length = reader.ReadUInt32(Endian.BigEndian);
//read the type of document
header.Type = (MobipocketType)Enum.ToObject(typeof(MobipocketType), reader.ReadUInt32(Endian.BigEndian));
//read encoding of document
header.Encoding = (MobipocketEncoding)Enum.ToObject(typeof(MobipocketEncoding), reader.ReadUInt32(Endian.BigEndian));
//read uid
header.UniqueId = reader.ReadUInt32(Endian.BigEndian);
//read version
header.Version = reader.ReadUInt32(Endian.BigEndian);
//40 0xff bytes
reader.BaseStream.Seek(40, SeekOrigin.Current);
//first no book record
header.FirstNonTextRecord = reader.ReadUInt32(Endian.BigEndian);
//offset in header of full name data
header._fullNameOffset = reader.ReadUInt32(Endian.BigEndian);
//length of full name data
header._fullNameLength = reader.ReadUInt32(Endian.BigEndian);
//locale of mobipocket file
header.Locale = reader.ReadUInt32(Endian.BigEndian);
//dictionary input language
header.DictionaryInfo.InputLanguage = reader.ReadUInt32(Endian.BigEndian);
//dictionary output language
header.DictionaryInfo.OutputLanguage = reader.ReadUInt32(Endian.BigEndian);
//minimal mobi pocket reader version
header.MinReaderVersion = reader.ReadUInt32(Endian.BigEndian);
//index of first image in pdb database
header.FirtImageRecord = reader.ReadUInt32(Endian.BigEndian);
//offset of huffman data
header.HuffmanInfo.RecordOffset = reader.ReadUInt32(Endian.BigEndian);
//count of huffman data
header.HuffmanInfo.RecordCount = reader.ReadUInt32(Endian.BigEndian);
//Skip 8 bytes, often zeroes
reader.BaseStream.Seek(8, SeekOrigin.Current);
//Check if we have exth record
var exthBitField = reader.ReadUInt32(Endian.BigEndian);
header._exthEnabled = ((exthBitField & 0x40) == 0x40);
if (length > 0x80)
{
byte[] data = new byte[32];
reader.Read(data, 0, 32);
// DRM Offset
var drmOffset = reader.ReadUInt32(Endian.BigEndian);
// DRM Count
var drmCount = reader.ReadUInt32(Endian.BigEndian);
// DRM Size
var drmSize = reader.ReadUInt32(Endian.BigEndian);
// DRM Flags
var drmFlags = reader.ReadUInt32(Endian.BigEndian);
header.Drm = new MobipocketDrm(drmOffset, drmCount, drmSize, drmFlags);
header.Drm.Parse(bytes);
}
//read trailing byte flags if present
if (length > 228)
{
reader.BaseStream.Seek(0xf2, SeekOrigin.Begin);
header.TrailingEntries = reader.ReadUInt16(Endian.BigEndian);
}
//read full name
reader.BaseStream.Seek(header._fullNameOffset, SeekOrigin.Begin);
var nameBuffer = new byte[header._fullNameLength];
checked { reader.Read(nameBuffer, 0, (int)header._fullNameLength); }
header.FullName = header.GetString(nameBuffer);
//mobi header end
if (header._exthEnabled)
{
//exth header start
//header.Data, length + 16
checked
{
Exth = ExthHeader.FromBinary(new ArraySegment<byte>(bytes.Array,
bytes.Offset + (int)length + 16,
bytes.Count - (int)length - 16));
}
//exth headet end
}
}
return header;
}
catch (Exception e)
{
throw new Exception("cannot parse mobipocket header from binary", e);
}
}
/// <summary>
/// Creates new instance of mobipocket header from provided bytes
/// </summary>
/// <param name="bytes"></param>
/// <returns></returns>
public static MobipocketHeader FromBinary(ArraySegment <byte> bytes)
{
if (bytes.Array == null)
{
throw new ArgumentException("bytes array cannot be null");
}
var header = new MobipocketHeader();
try
{
using (var stream = new MemoryStream(bytes.Array, bytes.Offset, bytes.Count))
using (var reader = new BinaryReader(stream))
{
if (bytes.Count < 0x80)
{
throw new Exception("header is to small");
}
//palmdoc header start
//Get compression
header.Compression = (MobipocketCompression)Enum.ToObject(typeof(MobipocketCompression), reader.ReadUInt16(Endian.BigEndian));
//usnused zeroes
reader.BaseStream.Seek(2, SeekOrigin.Current);
//length of text
header.TotalTextLength = reader.ReadUInt32(Endian.BigEndian);
//count of dpb records that contain text
header.TextRecordCount = reader.ReadUInt16(Endian.BigEndian);
//maximum record size, always 4096
header.MaxRecordSize = reader.ReadUInt16(Endian.BigEndian);
//read mobipocket encryption
header.Encryption = (MobipocketEncryption)Enum.ToObject(typeof(MobipocketEncryption), reader.ReadUInt16(Endian.BigEndian));
//skip two zeroes
reader.BaseStream.Seek(2, SeekOrigin.Current);
//palmdoc header end
//mobi header start
//read identifier "MOBI"
var identifier = reader.ReadUInt32(Endian.BigEndian);
if (MobiIdentifier != identifier)
{
throw new Exception("unsupported identifier: " + identifier);
}
//read byte length of header
var length = reader.ReadUInt32(Endian.BigEndian);
//read the type of document
header.Type = (MobipocketType)Enum.ToObject(typeof(MobipocketType), reader.ReadUInt32(Endian.BigEndian));
//read encoding of document
header.Encoding = (MobipocketEncoding)Enum.ToObject(typeof(MobipocketEncoding), reader.ReadUInt32(Endian.BigEndian));
//read uid
header.UniqueId = reader.ReadUInt32(Endian.BigEndian);
//read version
header.Version = reader.ReadUInt32(Endian.BigEndian);
//40 0xff bytes
reader.BaseStream.Seek(40, SeekOrigin.Current);
//first no book record
header.FirstNonTextRecord = reader.ReadUInt32(Endian.BigEndian);
//offset in header of full name data
header._fullNameOffset = reader.ReadUInt32(Endian.BigEndian);
//length of full name data
header._fullNameLength = reader.ReadUInt32(Endian.BigEndian);
//locale of mobipocket file
header.Locale = reader.ReadUInt32(Endian.BigEndian);
//dictionary input language
header.DictionaryInfo.InputLanguage = reader.ReadUInt32(Endian.BigEndian);
//dictionary output language
header.DictionaryInfo.OutputLanguage = reader.ReadUInt32(Endian.BigEndian);
//minimal mobi pocket reader version
header.MinReaderVersion = reader.ReadUInt32(Endian.BigEndian);
//index of first image in pdb database
header.FirtImageRecord = reader.ReadUInt32(Endian.BigEndian);
//offset of huffman data
header.HuffmanInfo.RecordOffset = reader.ReadUInt32(Endian.BigEndian);
//count of huffman data
header.HuffmanInfo.RecordCount = reader.ReadUInt32(Endian.BigEndian);
//Skip 8 bytes, often zeroes
reader.BaseStream.Seek(8, SeekOrigin.Current);
//Check if we have exth record
var exthBitField = reader.ReadUInt32(Endian.BigEndian);
header._exthEnabled = ((exthBitField & 0x40) == 0x40);
if (length > 0x80)
{
byte[] data = new byte[32];
reader.Read(data, 0, 32);
// DRM Offset
var drmOffset = reader.ReadUInt32(Endian.BigEndian);
// DRM Count
var drmCount = reader.ReadUInt32(Endian.BigEndian);
// DRM Size
var drmSize = reader.ReadUInt32(Endian.BigEndian);
// DRM Flags
var drmFlags = reader.ReadUInt32(Endian.BigEndian);
header.Drm = new MobipocketDrm(drmOffset, drmCount, drmSize, drmFlags);
header.Drm.Parse(bytes);
}
//read trailing byte flags if present
if (length > 228)
{
reader.BaseStream.Seek(0xf2, SeekOrigin.Begin);
header.TrailingEntries = reader.ReadUInt16(Endian.BigEndian);
}
//read full name
reader.BaseStream.Seek(header._fullNameOffset, SeekOrigin.Begin);
var nameBuffer = new byte[header._fullNameLength];
checked { reader.Read(nameBuffer, 0, (int)header._fullNameLength); }
header.FullName = header.GetString(nameBuffer);
//mobi header end
if (header._exthEnabled)
{
//exth header start
//header.Data, length + 16
checked
{
Exth = ExthHeader.FromBinary(new ArraySegment <byte>(bytes.Array,
bytes.Offset + (int)length + 16,
bytes.Count - (int)length - 16));
}
//exth headet end
}
}
return(header);
}
catch (Exception e)
{
throw new Exception("cannot parse mobipocket header from binary", e);
}
}
