void RemoveUnnassesaryWords(TajikWord word, TajikSentence sentence)
{
var shouldBeRemoved = ShouldBeRemoved(word.Value);
if (shouldBeRemoved)
{
sentence.Words.Remove(word);
}
}
/// <summary>
/// RU: Соберет слова из предложения
/// TJ: Гирифтани калимаҳо аз ҷумла
/// </summary>
public IEnumerable <TajikWord> SplitWordsFromSentences(TajikSentence sentence)
{
List <TajikWord> wordsInstances = new List <TajikWord>();
var words = sentence.Content.Replace(" - ", "-").Split(' ').GroupBy(w => w).Select(s => s.FirstOrDefault());
foreach (var word in words)
{
wordsInstances.Add(new TajikWord(Regex.Replace(word, Statics.SentenceEndSymbols, string.Empty)));
}
return(wordsInstances);
}
/// <summary>
/// RU: Нормализация текста
/// TJ: Нормаликунонии матн
/// </summary>
public void NormalizeWords(IEnumerable <TajikWord> words, TajikSentence sentence)
{
foreach (var word in words.ToList())
{
ShakliJam(word);
Ishorakuni(word);
BandakiI(word);
BandakiE(word);
BandakiU(word);
RemoveUnnassesaryWords(word, sentence);
}
}
