private WordDictionaryFile LoadFromTextFile(String fileName)
{
WordDictionaryFile dictFile = new WordDictionaryFile();
dictFile.Dicts = new List <IDataNode>();
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();
POSType pos = (POSType)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);
}
public PositionLength(int position, int length, WordAttribute wordAttr)
{
this.Position = position;
this.Length = length;
this.WordAttr = wordAttr;
this.Level = 0;
}
public List <IDataNode> Load()
{
List <IDataNode> nodes = new List <IDataNode>();
using (StreamReader sr = new StreamReader(setting.Uri))
{
string line = sr.ReadLine();
while (line != null)
{
string[] parts = line.Split(new char[] { ' ' });
WordAttribute wa = new WordAttribute();
if (parts.Length == 3)
{
wa.Word = parts[0];
wa.Frequency = Double.Parse(parts[1]);
wa.POS = (BluePrint.SegmentFramework.POSType)Convert.ToInt32(parts[2]);
}
else
{
wa.Word = parts[0];
}
nodes.Add(wa);
line = sr.ReadLine();
}
}
return(nodes);
}
private WordDictionaryFile LoadFromBinFile(String fileName, out string verNumStr)
{
WordDictionaryFile dictFile = new WordDictionaryFile();
dictFile.Dicts = new List <IDataNode>();
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);
}
var matches = verRegex.Matches(ver);
if (matches.Count > 0)
{
verNumStr = matches[0].Value;
}
else
{
verNumStr = null;
}
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));
POSType pos = (POSType)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);
}
