public unsafe void Open(BinaryReader reader)
{
uint magic = reader.ReadUInt32();
//CanSeekの時のみストリーム長のチェック
if (reader.BaseStream.CanSeek && reader.BaseStream.Length != (magic ^ DictionaryMagicID))
{
throw new MeCabInvalidFileException("dictionary file is broken", this.FileName);
}
this.Version = reader.ReadUInt32();
if (this.Version != DicVersion)
{
throw new MeCabInvalidFileException("incompatible version", this.FileName);
}
this.Type = (DictionaryType)reader.ReadUInt32();
this.LexSize = reader.ReadUInt32();
this.LSize = reader.ReadUInt32();
this.RSize = reader.ReadUInt32();
uint dSize = reader.ReadUInt32();
uint tSize = reader.ReadUInt32();
uint fSize = reader.ReadUInt32();
reader.ReadUInt32(); //dummy
string charSet = StrUtils.GetString(reader.ReadBytes(32), Encoding.ASCII);
this.encoding = StrUtils.GetEncoding(charSet);
this.da.Open(reader, dSize);
this.tokens = new Token[tSize / sizeof(Token)];
for (int i = 0; i < this.tokens.Length; i++)
{
this.tokens[i] = Token.Create(reader);
}
this.features = reader.ReadBytes((int)fSize);
if (reader.BaseStream.ReadByte() != -1)
{
throw new MeCabInvalidFileException("dictionary file is broken", this.FileName);
}
}
public void Open(MemoryMappedFile mmf, string filePath = null)
{
this.FileName = filePath;
using (MemoryMappedViewStream stream = mmf.CreateViewStream(
0L, 0L, MemoryMappedFileAccess.Read))
using (BinaryReader reader = new BinaryReader(stream))
{
uint magic = reader.ReadUInt32();
if (stream.CanSeek && stream.Length < (magic ^ DictionaryMagicID)) //正確なサイズ取得ができないので不等号で代用
{
throw new MeCabInvalidFileException("dictionary file is broken", filePath);
}
this.Version = reader.ReadUInt32();
if (this.Version != DicVersion)
{
throw new MeCabInvalidFileException("incompatible version", filePath);
}
this.Type = (DictionaryType)reader.ReadUInt32();
this.LexSize = reader.ReadUInt32();
this.LSize = reader.ReadUInt32();
this.RSize = reader.ReadUInt32();
uint dSize = reader.ReadUInt32();
uint tSize = reader.ReadUInt32();
uint fSize = reader.ReadUInt32();
reader.ReadUInt32(); //dummy
string charSet = StrUtils.GetString(reader.ReadBytes(32), Encoding.ASCII);
this.encoding = StrUtils.GetEncoding(charSet);
long offset = stream.Position;
this.da.Open(mmf, offset, dSize);
offset += dSize;
this.tokens = mmf.CreateViewAccessor(offset, tSize, MemoryMappedFileAccess.Read);
offset += tSize;
this.features = mmf.CreateViewAccessor(offset, fSize, MemoryMappedFileAccess.Read);
}
}
public void Open(BinaryReader reader, string fileName = null)
{
uint cSize = reader.ReadUInt32();
if (reader.BaseStream.CanSeek)
{
long fSize = sizeof(uint) + 32 * cSize + sizeof(uint) * charInfoList.Length;
if (reader.BaseStream.Length != fSize)
{
throw new MeCabInvalidFileException("invalid file size", fileName);
}
}
this.cList = new string[cSize];
for (int i = 0; i < this.cList.Length; i++)
{
this.cList[i] = StrUtils.GetString(reader.ReadBytes(32), Encoding.ASCII);
}
for (int i = 0; i < this.charInfoList.Length; i++)
{
this.charInfoList[i] = new CharInfo(reader.ReadUInt32());
}
}
/// <summary>
/// バイト配列の中から終端が\0で表された文字列を取り出す。
/// </summary>
/// <remarks>
/// バイト配列の長さはInt32.MaxValueを超えていても良い。
/// </remarks>
/// <param name="bytes">バイト配列</param>
/// <param name="offset">オフセット位置</param>
/// <param name="enc">文字エンコーディング</param>
/// <returns>文字列（\0は含まない）</returns>
public unsafe static string GetString(byte[] bytes, long offset, Encoding enc)
{
fixed(byte *pBytes = bytes)
return(StrUtils.GetString(pBytes + offset, enc));
}
/// <summary>
/// バイト配列の中から終端が\0で表された文字列を取り出す。
/// </summary>
/// <remarks>
/// バイト配列の長さはInt32.MaxValueを超えていても良い。
/// </remarks>
/// <param name="bytes">バイト配列</param>
/// <param name="enc">文字エンコーディング</param>
/// <returns>文字列（\0は含まない）</returns>
public static string GetString(byte[] bytes, Encoding enc)
{
return(StrUtils.GetString(bytes, 0L, enc));
}
public string GetFeature(uint featurePos)
{
return(StrUtils.GetString(this.features, (long)featurePos, this.encoding));
}