public List <string> GetLabels(string sentence)
{
List <string> labels = new List <string>();
bool bInitDict = HLSplitInit(@"C:\dll\");
if (!bInitDict)
{
Console.WriteLine("初始化分词字典失败!", "错误");
return(null);
}
IntPtr hHandle = HLOpenSplit(); //创建分词句柄
if (hHandle == IntPtr.Zero)
{
//创建分词句柄失败
Console.WriteLine("创建分词句柄失败!", "错误");
HLFreeSplit();//卸载分词字典
return(null);
}
short iExtraCalcFlag = 0; //附加计算标志，不进行附加计算
//获得附加计算标识
//if (this.chkPos.Checked)
iExtraCalcFlag |= HL_CAL_OPT_POS;//
//if (this.chkKeyword.Checked)
iExtraCalcFlag |= HL_CAL_OPT_KEYWORD;
//if (this.chkSeach.Checked)
//iExtraCalcFlag |= HL_CAL_OPT_SEARCH;
//if (this.chkFinger.Checked)
iExtraCalcFlag |= HL_CAL_OPT_FINGER;
DateTime bgdt = DateTime.Now;
bool bSuccess = HLSplitWord(hHandle, sentence, iExtraCalcFlag);
System.TimeSpan ts = DateTime.Now - bgdt;
Console.WriteLine("get labels---->" + ts);
//this.txtMsg.Text = string.Format("用时{0}分{1}秒{2}毫秒", ts.Minutes, ts.Seconds, ts.Milliseconds);
if (bSuccess)
{
//分词成功
int nResultCnt = HLGetWordCnt(hHandle);//取得分词个数
for (int i = 0; i < nResultCnt; i++)
{
//取得分词结果
IntPtr h = HLGetWordAt(hHandle, i);
//取得一个分词结果
SHLSegWord pWord = (SHLSegWord)Marshal.PtrToStructure(h, typeof(SHLSegWord));
if ((GetNatureString(pWord.s_dwPOS).Equals(".n") || //名词
GetNatureString(pWord.s_dwPOS).Equals(".v") || //动词
GetNatureString(pWord.s_dwPOS).Equals(".ns") || //地名
GetNatureString(pWord.s_dwPOS).Equals(".m") || //数词
GetNatureString(pWord.s_dwPOS).Equals(".mq") || //数量词
GetNatureString(pWord.s_dwPOS).Equals(".nr")) &&
pWord.s_szWord.Length > 1)
{
labels.Add(pWord.s_szWord);
}
}
HLCloseSplit(hHandle);//关闭分词句柄
}
else
{
//分词失败
Console.WriteLine("分词失败!", "错误");
HLCloseSplit(hHandle); //关闭分词句柄
HLFreeSplit(); //卸载分词字典
labels = null;
}
HLFreeSplit(); //卸载分词词典
return(labels);
}
/// <summary>
/// 判断是否含有标点
/// </summary>
/// <param name="str"></param>
/// <returns></returns>
public bool IsHasPunctuation(string str)
{
if (str == null)
{
return(false);
}
bool hasPunctuation = false;
bool bInitDict = HLSplitInit(@"C:\dll\");
if (!bInitDict)
{
Console.WriteLine("初始化分词字典失败!", "错误");
}
IntPtr hHandle = HLOpenSplit(); //创建分词句柄
if (hHandle == IntPtr.Zero)
{
//创建分词句柄失败
Console.WriteLine("创建分词句柄失败!", "错误");
HLFreeSplit();//卸载分词字典
}
short iExtraCalcFlag = 0; //附加计算标志，不进行附加计算
//获得附加计算标识
//if (this.chkPos.Checked)
iExtraCalcFlag |= HL_CAL_OPT_POS;//
//if (this.chkKeyword.Checked)
iExtraCalcFlag |= HL_CAL_OPT_KEYWORD;
//if (this.chkSeach.Checked)
//iExtraCalcFlag |= HL_CAL_OPT_SEARCH;
//if (this.chkFinger.Checked)
iExtraCalcFlag |= HL_CAL_OPT_FINGER;
DateTime bgdt = DateTime.Now;
bool bSuccess = HLSplitWord(hHandle, str, iExtraCalcFlag);
System.TimeSpan ts = DateTime.Now - bgdt;
Console.WriteLine("get names----->" + ts);
//this.txtMsg.Text = string.Format("用时{0}分{1}秒{2}毫秒", ts.Minutes, ts.Seconds, ts.Milliseconds);
if (bSuccess)
{
//分词成功
int nResultCnt = HLGetWordCnt(hHandle);//取得分词个数
for (short i = 0; i < nResultCnt; i++)
{
//取得分词结果
IntPtr h = HLGetWordAt(hHandle, i);
//取得一个分词结果
SHLSegWord pWord = (SHLSegWord)Marshal.PtrToStructure(h, typeof(SHLSegWord));
if (GetNatureString(pWord.s_dwPOS).Equals(".w"))
{
hasPunctuation = true;
}
}
HLCloseSplit(hHandle);//关闭分词句柄
}
else
{
//分词失败
Console.WriteLine("分词失败!", "错误");
HLCloseSplit(hHandle); //关闭分词句柄
HLFreeSplit(); //卸载分词字典
}
HLFreeSplit(); //卸载分词词典
return(hasPunctuation);
}
