public virtual string Stem(string word)
{
try
{
lexer.Yyreset(new StringReader(word));
lexer.Yybegin(Morpha.any);
string wordRes = lexer.Next();
return(wordRes);
}
catch (IOException)
{
log.Warning("Morphology.stem() had error on word " + word);
return(word);
}
}
/// <summary>
/// Lemmatize the word, being sensitive to the tag, using the
/// passed in lexer.
/// </summary>
/// <param name="lowercase">
/// If this is true, words other than proper nouns will
/// be changed to all lowercase.
/// </param>
private static string Lemmatize(string word, string tag, Morpha lexer, bool lowercase)
{
bool wordHasForbiddenChar = word.IndexOf('_') >= 0 || word.IndexOf(' ') >= 0 || word.IndexOf('\n') >= 0;
string quotedWord = word;
if (wordHasForbiddenChar)
{
// choose something unlikely. Classical Vedic!
quotedWord = quotedWord.ReplaceAll("_", "\u1CF0");
quotedWord = quotedWord.ReplaceAll(" ", "\u1CF1");
quotedWord = quotedWord.ReplaceAll("\n", "\u1CF2");
}
string wordtag = quotedWord + '_' + tag;
try
{
lexer.SetOption(1, lowercase);
lexer.Yyreset(new StringReader(wordtag));
lexer.Yybegin(Morpha.scan);
string wordRes = lexer.Next();
lexer.Next();
// go past tag
if (wordHasForbiddenChar)
{
wordRes = wordRes.ReplaceAll("\u1CF0", "_");
wordRes = wordRes.ReplaceAll("\u1CF1", " ");
wordRes = wordRes.ReplaceAll("\u1CF2", "\n");
}
return(wordRes);
}
catch (IOException)
{
log.Warning("Morphology.stem() had error on word " + word + '/' + tag);
return(word);
}
}
