private WordDictionaryFile LoadFromTextFile(String fileName)
{
WordDictionaryFile dictFile = new WordDictionaryFile();
dictFile.Dicts = new List<WordAttribute>();
using (StreamReader sr = new StreamReader(fileName, Encoding.UTF8))
{
while (!sr.EndOfStream)
{
string line = sr.ReadLine();
string[] strs = line.Split(new char[] { '|' });
if (strs.Length == 3)
{
string word = strs[0].Trim();
POS pos = (POS)int.Parse(strs[1].Substring(2, strs[1].Length - 2), System.Globalization.NumberStyles.HexNumber);
double frequency = double.Parse(strs[2]);
WordAttribute dict = new WordAttribute(word, pos, frequency);
dictFile.Dicts.Add(dict);
}
}
}
return dictFile;
}
private WordDictionaryFile LoadFromBinFile(String fileName, out string verNumStr)
{
WordDictionaryFile dictFile = new WordDictionaryFile();
dictFile.Dicts = new List<WordAttribute>();
FileStream fs = new FileStream(fileName, FileMode.Open, FileAccess.Read);
byte[] version = new byte[32];
fs.Read(version, 0, version.Length);
String ver = Encoding.UTF8.GetString(version, 0, version.Length);
int zeroPosition = ver.IndexOf('\0');
if (zeroPosition >= 0)
{
ver = ver.Substring(0, zeroPosition);
}
verNumStr = Framework.Regex.GetMatch(ver, "Pan Gu Segment V(.+)", true);
while (fs.Position < fs.Length)
{
byte[] buf = new byte[sizeof(int)];
fs.Read(buf, 0, buf.Length);
int length = BitConverter.ToInt32(buf, 0);
buf = new byte[length];
fs.Read(buf, 0, buf.Length);
string word = Encoding.UTF8.GetString(buf, 0, length - sizeof(int) - sizeof(double));
POS pos = (POS)BitConverter.ToInt32(buf, length - sizeof(int) - sizeof(double));
double frequency = BitConverter.ToDouble(buf, length - sizeof(double));
WordAttribute dict = new WordAttribute(word, pos, frequency);
string.Intern(dict.Word);
dictFile.Dicts.Add(dict);
}
fs.Close();
return dictFile;
}