public void SingleWordTest()
{
var str = "Hello";
var boundaries = WordBoundaryAlgorithm.FindWordBoundaries(new Utf32Buffer(str).AsSlice()).ToList();
Assert.Single(boundaries);
Assert.Equal(0, boundaries[0]);
}
public void CombinedLettersAndDigits()
{
var str = "Hello99 World";
var boundaries = WordBoundaryAlgorithm.FindWordBoundaries(new Utf32Buffer(str).AsSlice()).ToList();
Assert.Equal(2, boundaries.Count);
Assert.Equal(0, boundaries[0]);
Assert.Equal(8, boundaries[1]);
}
public void TwoWordTest()
{
var str = "Hello World";
var boundaries = WordBoundaryAlgorithm.FindWordBoundaries(new Utf32Buffer(str).AsSlice()).ToList();
Assert.Equal(2, boundaries.Count);
Assert.Equal(0, boundaries[0]);
Assert.Equal(6, boundaries[1]);
}
public void Punctuation2()
{
var str = "Hello () World";
var boundaries = WordBoundaryAlgorithm.FindWordBoundaries(new Utf32Buffer(str).AsSlice()).ToList();
Assert.Equal(3, boundaries.Count);
Assert.Equal(0, boundaries[0]);
Assert.Equal(6, boundaries[1]);
Assert.Equal(9, boundaries[2]);
}
public void TestIsWordBoundary()
{
var str = new Utf32Buffer("Hello () World").AsSlice();
// Get the boundaries (assuming FindWordBoundaries is correct)
var boundaries = WordBoundaryAlgorithm.FindWordBoundaries(str).ToList();
var boundaries2 = new List <int>();
for (int i = 0; i < str.Length; i++)
{
if (WordBoundaryAlgorithm.IsWordBoundary(str.SubSlice(0, i), str.SubSlice(i, str.Length - i)))
{
boundaries2.Add(i);
}
}
Assert.Equal(boundaries, boundaries2);
}