public void testRetrievingContent()
{
const string sentence = "A test";
var sample = new TokenSample(sentence, new[] { new Span(0, 1), new Span(2, 6) });
Assert.AreEqual("A test", sample.Text);
Assert.AreEqual(new Span(0, 1), sample.TokenSpans[0]);
Assert.AreEqual(new Span(2, 6), sample.TokenSpans[1]);
}
public void testCreationWithDetokenizer()
{
var detokenizer = DictionaryDetokenizerTest.CreateLatinDetokenizer();
var tokens = new[] {
"start",
"(", // move right
")", // move left
"end",
".", // move left
"hyphen",
"-", // move both
"string",
"."
};
var a = new TokenSample(detokenizer, tokens);
Assert.AreEqual("start () end. hyphen-string.", a.Text);
// 0123456789012345678901234567
Assert.AreEqual("start (" + TokenSample.DefaultSeparator + ") end" + TokenSample.DefaultSeparator + "."
+ " hyphen" + TokenSample.DefaultSeparator + "-" + TokenSample.DefaultSeparator + "string" +
TokenSample.DefaultSeparator + ".", a.ToString());
Assert.AreEqual(9, a.TokenSpans.Length);
Assert.AreEqual(new Span(0, 5), a.TokenSpans[0]);
Assert.AreEqual(new Span(6, 7), a.TokenSpans[1]);
Assert.AreEqual(new Span(7, 8), a.TokenSpans[2]);
Assert.AreEqual(new Span(9, 12), a.TokenSpans[3]);
Assert.AreEqual(new Span(12, 13), a.TokenSpans[4]);
Assert.AreEqual(new Span(14, 20), a.TokenSpans[5]);
Assert.AreEqual(new Span(20, 21), a.TokenSpans[6]);
Assert.AreEqual(new Span(21, 27), a.TokenSpans[7]);
Assert.AreEqual(new Span(27, 28), a.TokenSpans[8]);
}
