public void GetBoundaries_Word()
{
var parts = BreakIterator.GetBoundaries(BreakIterator.UBreakIteratorType.WORD, new Locale("en-US"), WordBoundaryTestData.Text);
Assert.That(parts.Count(), Is.EqualTo(WordBoundaryTestData.ExpectedOnlyWords.Length));
Assert.That(parts.ToArray(), Is.EquivalentTo(WordBoundaryTestData.ExpectedOnlyWords));
}
public void GetBoundaries_LocaleDependentBreakPoints_Line()
{
var text = "論‼";
var parts = BreakIterator.GetBoundaries(BreakIterator.UBreakIteratorType.LINE, new Locale("ja@lb=loose"), text);
// we only get 2 boundaries if the ja locale is passed correctly
Assert.That(parts.Count(), Is.EqualTo(2));
}
public void GetBoundaries_Sentence()
{
var text = "Aa bb. Ccdef 3.5 x? Y?x! Z";
var expected = new[] {
new Boundary(0, 7), new Boundary(7, 20), new Boundary(20, 22), new Boundary(22, 25), new Boundary(25, 26)
};
var parts = BreakIterator.GetBoundaries(BreakIterator.UBreakIteratorType.SENTENCE, new Locale("en-US"), text);
Assert.That(parts.Count(), Is.EqualTo(expected.Length));
Assert.That(parts.ToArray(), Is.EquivalentTo(expected));
}
public void GetBoundaries_Character()
{
var text = "abc? 1";
var expected = new[] {
new Boundary(0, 1), new Boundary(1, 2), new Boundary(2, 3), new Boundary(3, 4), new Boundary(4, 5), new Boundary(5, 6)
};
var parts = BreakIterator.GetBoundaries(BreakIterator.UBreakIteratorType.CHARACTER, new Locale("en-US"), text);
Assert.That(parts.Count(), Is.EqualTo(expected.Length));
Assert.That(parts.ToArray(), Is.EquivalentTo(expected));
}
public void GetWordAndLineBoundariesWithHyphenatedText()
{
var text = "Good-day, kind sir !";
var expectedWords = new[] {
new Boundary(0, 4), new Boundary(5, 8), new Boundary(10, 14), new Boundary(15, 18)
};
var expectedLines = new[] {
new Boundary(0, 5), new Boundary(5, 10), new Boundary(10, 15), new Boundary(15, 20)
};
var wordBoundaries = BreakIterator.GetBoundaries(BreakIterator.UBreakIteratorType.WORD, new Locale("en-US"), text);
var lineBoundaries = BreakIterator.GetBoundaries(BreakIterator.UBreakIteratorType.LINE, new Locale("en-US"), text);
Assert.That(wordBoundaries.Count(), Is.EqualTo(expectedWords.Length));
Assert.That(wordBoundaries.ToArray(), Is.EquivalentTo(expectedWords));
Assert.That(lineBoundaries.Count(), Is.EqualTo(expectedLines.Length));
Assert.That(lineBoundaries.ToArray(), Is.EquivalentTo(expectedLines));
}
