public IEnumerable <MecabWordInfo> MecabWordEnumerable(string sentence)
{
foreach (var node in tagger.ParseToNodes(sentence))
{
if (node.CharType > 0)
{
var features = node.Feature.Split(',');
#region 填充 MecabWordInfo 各项 Property
MecabWordInfo word = new MecabWordInfo
{
Word = node.Surface,
PartOfSpeech = features[0],
Description = features[1],
Feature = node.Feature,
Kana = " "
};
// 加这一步是为了防止乱码进入分词导致无法读取假名
if (features.Length >= 8)
{
word.Kana = features[7];
}
if (word.PartOfSpeech == "記号" ||
WanaKana.IsHiragana(node.Surface) ||
WanaKana.IsKatakana(node.Surface))
{
word.Kana = " ";
}
#endregion
yield return(word);
}
}
}
/// <summary>
/// Progress sentence
/// </summary>
/// <param name="sentence"></param>
/// <returns></returns>
public List <MecabWordInfo> SentenceHandle(string sentence)
{
List <MecabWordInfo> ret = new List <MecabWordInfo>();
foreach (var node in tagger.ParseToNodes(sentence))
{
if (node.CharType > 0)
{
var features = node.Feature.Split(',');
#region 填充 MecabWordInfo 各项 Property
MecabWordInfo word = new MecabWordInfo
{
Word = node.Surface,
PartOfSpeech = features[0],
Description = features[1],
Feature = node.Feature
};
// 加这一步是为了防止乱码进入分词导致无法读取假名
if (features.Length >= 8)
{
word.Kana = features[7];
}
// 清理不需要的假名
if (word.PartOfSpeech == "記号")
{
word.Kana = "";
}
if (WanaKana.IsHiragana(node.Surface))
{
word.Kana = "";
}
if (WanaKana.IsKatakana(node.Surface))
{
word.Kana = "";
}
#endregion
ret.Add(word);
}
}
return(ret);
}
public bool IsKatkana(string input) => WanaKana.IsKatakana(input);
public Boolean IsKatkana(String input) => WanaKana.IsKatakana(input);
public void IsKatakanaResponsesMatch(string input, bool expectedResult)
{
Assert.Equal(expectedResult, WanaKana.IsKatakana(input));
}
public void EmptyReturnsFalse()
{
Assert.False(WanaKana.IsKatakana(string.Empty));
}
