protected internal virtual void copy_from(SpanishStemmer other)
{
I_p2 = other.I_p2;
I_p1 = other.I_p1;
I_pV = other.I_pV;
base.copy_from(other);
}
private void copy_from(SpanishStemmer other)
{
I_p2 = other.I_p2;
I_p1 = other.I_p1;
I_pV = other.I_pV;
base.CopyFrom(other);
}
public void Spanish_BaseTest()
{
var stemmer = new SpanishStemmer();
Assert.AreEqual("acerqu", stemmer.Stem("acerquen"));
Assert.AreEqual("Vision", stemmer.Stem("Visionado"));
Assert.AreEqual("algu", stemmer.Stem("algue"));
}
private TokenStream SpanishSteammer(TokenStream tokenStream)
{
//Obtener en una cadena cada token y aplicar el lematizador a cada término
string term = string.Empty;
IStemmer stemmer = new SpanishStemmer();
TokenStream tokenStreamtemp;
var termAttr = tokenStream.GetAttribute <ITermAttribute>();
while (tokenStream.IncrementToken())
{
term = term + " " + stemmer.Stem(termAttr.Term);
}
tokenStreamtemp = new StandardTokenizer(Version.LUCENE_30, new System.IO.StringReader(term));
return(tokenStreamtemp);
//
}
public TokenStream SpanishSteammer(TokenStream tokenStream)
{
//Obtener en una cadena cada token y aplicar el lematizador a cada término
string term = string.Empty;
IStemmer stemmer = new SpanishStemmer();
TokenStream tokenStreamtemp;
//var termAttr = tokenStream.GetAttribute<ITermAttribute>();
string cadena = "";
string[] token;
Lucene.Net.Analysis.Token current;
while ((current = tokenStream.Next()) != null)
{
cadena = current.ToString();
token = cadena.Split(',');
cadena = cadena.Substring(1, token[0].Length - 1);
term = term + " " + stemmer.Stem(cadena);
}
tokenStreamtemp = new StandardTokenizer(new System.IO.StringReader(term));
return(tokenStreamtemp);
//
}
private void copy_from(SpanishStemmer other)
{
I_p2 = other.I_p2;
I_p1 = other.I_p1;
I_pV = other.I_pV;
base.copy_from(other);
}
/// <summary>
/// 获取词干对象(线程不安全)
/// </summary>
/// <param name="language"></param>
/// <returns></returns>
//public static SnowballProgram GetSnowball(string language)
//{
// if (_dictSnowball.ContainsKey(language))
// return _dictSnowball[language];
// return null;
//}
/// <summary>
/// 获取词干对象(线程安全)
/// </summary>
/// <param name="language"></param>
/// <returns></returns>
public static SnowballProgram GetSnowball(string language)
{
SnowballProgram result = null;
switch (language)
{
case "DA":
result = new DanishStemmer();
break; //丹麦语
case "NL":
result = new DutchStemmer();
break; //荷兰语
case "EN":
result = new EnglishStemmer();
break; //英语
case "FI":
result = new FinnishStemmer();
break; //芬兰语
case "FR":
result = new FrenchStemmer();
break; //法语
case "DE2":
result = new German2Stemmer();
break; //德语2
case "DE":
result = new GermanStemmer();
break; //德语
case "HU":
result = new HungarianStemmer();
break;
case "IT":
result = new ItalianStemmer();
break;
case "文斯语":
result = new LovinsStemmer();
break;
case "NO":
result = new NorwegianStemmer();
break;
case "波特语":
result = new PorterStemmer();
break; //英语的
case "PT":
result = new PortugueseStemmer();
break; //葡萄牙语
case "RO":
result = new RomanianStemmer();
break;
case "RU":
result = new RussianStemmer();
break; //俄语
case "ES":
result = new SpanishStemmer();
break; //西班牙语
case "SV":
result = new SwedishStemmer();
break;
case "TR":
result = new TurkishStemmer();
break; //土耳其语
}
return(result);
}
} // End of the GetCurrentDomain method
/// <summary>
/// Get the stemmer based on the language
/// </summary>
/// <param name="language">A reference to the language</param>
/// <returns>A reference to a Stemmer</returns>
public static Stemmer GetStemmer(Language language)
{
// Create a default stemmer
Stemmer stemmer = new DefaultStemmer();
// Get the language code in lower case
string language_code = language.language_code.ToLower();
// Get a stemmer depending on the language
if (language_code == "da")
{
stemmer = new DanishStemmer();
}
else if (language_code == "nl")
{
stemmer = new DutchStemmer();
}
else if (language_code == "en")
{
stemmer = new EnglishStemmer();
}
else if (language_code == "fi")
{
stemmer = new FinnishStemmer();
}
else if (language_code == "fr")
{
stemmer = new FrenchStemmer();
}
else if (language_code == "de")
{
stemmer = new GermanStemmer();
}
else if (language_code == "it")
{
stemmer = new ItalianStemmer();
}
else if (language_code == "no")
{
stemmer = new NorwegianStemmer();
}
else if (language_code == "pt")
{
stemmer = new PortugueseStemmer();
}
else if (language_code == "ro")
{
stemmer = new RomanianStemmer();
}
else if (language_code == "es")
{
stemmer = new SpanishStemmer();
}
else if (language_code == "sv")
{
stemmer = new SwedishStemmer();
}
// Return the stemmer
return stemmer;
} // End of the GetStemmer method
