public static void getSentenceStemmsAndSubStemm(string lang, IEnumerable <string> sentence, Action <string, List <WordStemms> > onOK)
{
var stemmer = Creators.createStemmer(lang);
if (stemmer == null)
{
return;
}
foreach (var word in sentence)
{
var all = new List <WordStemms>();
var first = new WordStemms()
{
word = word
};
all.Add(first);
stemm(stemmer, word, first.stemms);
foreach (var subw in first.stemms.Where(w => w != word))
{
var other = new WordStemms()
{
word = subw
};
stemm(stemmer, subw, other.stemms);
all.Add(other);
}
onOK(word, all);
}
}
static List <string> getWordStemms(string lang, string word)
{
var stemmer = Creators.createStemmer(lang);
var res = new List <string>();
stemm(stemmer, word, res);
return(res);
}
public static void getSentenceStemms(string lang, IEnumerable <string> sentence, Action <string, List <string> > onWordOK)
{
var stemmer = Creators.createStemmer(lang);
if (stemmer == null)
{
return;
}
var list = new List <string>();
foreach (var word in sentence)
{
stemm(stemmer, word, list);
onWordOK(word, list);
list.Clear();
}
}
public static List <TPosLen>[] wordBreak(string lang, IList <string> texts)
{
var breaker = Creators.createBreaker(lang);
var res = new List <TPosLen> [texts.Count];
for (var i = 0; i < texts.Count; i++)
{
var pl = res[i] = new List <TPosLen>();
var word = texts[i];
wordBreak(word, breaker, (type, pos, len) => {
if (type != PutTypes.put)
{
return; // && type != PutTypes.alt) return;
}
pl.Add(new TPosLen {
Pos = pos, Len = len
});
});
}
return(res);
}
public static IEnumerable <string> wordBreak(string lang, IEnumerable <string> sentences)
{
var breaker = Creators.createBreaker(lang);
return(sentences.SelectMany(par => wordBreak(breaker, par)));
}