/// <summary> Returns an enumerable sentence sequence created with the specified culture. </summary>
/// <param name="text">String to decompose</param>
/// <param name="cultureNfo">Culture to use for decomposing</param>
/// <returns>Sequence of sentences</returns>
public static IEnumerable <Sentence> Sentences(this string text, CultureInfo cultureNfo)
{
using (SentenceIterator itr = GetSentenceIterator(text, cultureNfo)) {
while (itr.MoveNext())
{
yield return(itr.Current);
}
}
}
/// <summary> Returns a set of detected sentences for the specified <paramref name="input"/>. </summary>
/// <param name="input">Text to process</param>
/// <returns>Set of sentences</returns>
public IEnumerable <Sentence> Detect(string input)
{
if (string.IsNullOrEmpty(input))
{
yield break;
}
Dictionary <string, string> tokens = new Dictionary <string, string>(100);
TokenFactory tokenFactory = new TokenFactory(tokens);
string text = input;
// First, we strip time expression
text = _lang.StripTimeExpression(text, tokenFactory);
// Next, we strip date expressions
text = _lang.StripDateExpression(text, tokenFactory);
// Then we strip abbreviations
text = _lang.StripAbbreviations(text, tokenFactory);
// Detect sentences and reverse any replacement
int shift = 0;
using (SentenceIterator itr = new SentenceIterator(text, _lang)) {
while (itr.MoveNext())
{
Sentence sentence = itr.Current;
string sentTxt = sentence.Text;
int sentLgth = sentTxt.Length;
int sentBgn = sentence.Begin;
string rplTxt = sentTxt;
using (Dictionary <string, string> .Enumerator tokItr = tokens.GetEnumerator()) {
while (tokItr.MoveNext())
{
var entry = tokItr.Current;
string token = entry.Key;
string expr = entry.Value;
rplTxt = rplTxt.Replace(token, expr);
}
}
int rplLgth = rplTxt.Length;
int rplDelta = rplLgth - sentLgth;
int rplBgn = sentBgn + shift;
yield return(new Sentence(rplTxt, rplBgn));
shift += rplDelta;
}
}
}