private static void CPPictclas()
{
if (!ICTCLAS_Init(null))
{
System.Console.WriteLine("Init ICTCLAS failed!");
System.Console.Read();
}
StreamReader sr = null;
StreamWriter sw = new StreamWriter(outDir+".txt",false,System.Text.Encoding.Default);
Stopwatch sp = new Stopwatch();
sp.Start();
int nWrdcnt;
//对fileList中的每个文件进行分词，结果写入文本文件
for (int i = 0; i < fileList.Count; i += 1)
{
sr = new StreamReader(fileList[i], System.Text.Encoding.Default);
string input = "";
input = sr.ReadLine();
result_t []result;
while (input != null)
{
if (input == "")
{
input = sr.ReadLine();
continue;
}
try
{
result = new result_t[input.Length];
nWrdcnt = ICTCLAS_ParagraphProcessAW(input, result, eCodeType.CODE_TYPE_GB, 1);
}
catch (Exception e)
{
Console.WriteLine(e.Message);
continue;
}
byte[] mybyte = System.Text.Encoding.Default.GetBytes(input);
byte[] byteWord = new byte[1] ;
for (int j = 0; j < nWrdcnt; ++j)
{
try
{
byteWord = new byte[result[j].length];
}
catch (Exception e)
{
Console.WriteLine(e.Message);
}
Array.Copy(mybyte, result[j].start, byteWord, 0, result[j].length);
string watch = System.Text.Encoding.Default.GetString(byteWord);
if (watch == " ")
continue;
sw.Write(System.Text.Encoding.Default.GetString(byteWord)+" ");
}
sw.WriteLine("");
input = sr.ReadLine();
}
sr.Close();
}
sw.Close();
sp.Stop();
Console.WriteLine("ICTCLAS:" + sp.ElapsedMilliseconds + "ms" + " Word Segmentation Speed: " + (float)totalFileSizeInByte / 1024 / sp.ElapsedMilliseconds * 1000 + " KB/s");
ICTCLAS_Exit();
}
private static void CPPictclas()
{
if (!ICTCLAS_Init(null))
{
System.Console.WriteLine("Init ICTCLAS failed!");
System.Console.Read();
}
StreamReader sr = null;
StreamWriter sw = new StreamWriter(outDir + ".txt", false, System.Text.Encoding.Default);
Stopwatch sp = new Stopwatch();
sp.Start();
int nWrdcnt;
//对fileList中的每个文件进行分词，结果写入文本文件
for (int i = 0; i < fileList.Count; i += 1)
{
sr = new StreamReader(fileList[i], System.Text.Encoding.Default);
string input = "";
input = sr.ReadLine();
result_t [] result;
while (input != null)
{
if (input == "")
{
input = sr.ReadLine();
continue;
}
try
{
result = new result_t[input.Length];
nWrdcnt = ICTCLAS_ParagraphProcessAW(input, result, eCodeType.CODE_TYPE_GB, 1);
}
catch (Exception e)
{
Console.WriteLine(e.Message);
continue;
}
byte[] mybyte = System.Text.Encoding.Default.GetBytes(input);
byte[] byteWord = new byte[1];
for (int j = 0; j < nWrdcnt; ++j)
{
try
{
byteWord = new byte[result[j].length];
}
catch (Exception e)
{
Console.WriteLine(e.Message);
}
Array.Copy(mybyte, result[j].start, byteWord, 0, result[j].length);
string watch = System.Text.Encoding.Default.GetString(byteWord);
if (watch == " ")
{
continue;
}
sw.Write(System.Text.Encoding.Default.GetString(byteWord) + " ");
}
sw.WriteLine("");
input = sr.ReadLine();
}
sr.Close();
}
sw.Close();
sp.Stop();
Console.WriteLine("ICTCLAS:" + sp.ElapsedMilliseconds + "ms" + " Word Segmentation Speed: " + (float)totalFileSizeInByte / 1024 / sp.ElapsedMilliseconds * 1000 + " KB/s");
ICTCLAS_Exit();
}
