public void ļcien()
{
LatvianTokenizer tokenizer = new LatvianTokenizer(compile: false);
tokenizer.Add<LuMiiAbbreviationToken>();
tokenizer.Compile();
string text = "Ļ.cien. kungs!";
Token[] tokens = tokenizer.Tokenize(text).ToArray();
Assert.AreEqual(3, tokens.Length);
Assert.AreEqual("Ļ.cien.", tokens[0].Text);
}
public void kā_arī()
{
LatvianTokenizer tokenizer = new LatvianTokenizer(compile: false);
tokenizer.Add<LuMiiCollocationToken>();
tokenizer.Compile();
string text = "Es skrienu kā arī lecu.";
Token[] tokens = tokenizer.Tokenize(text).ToArray();
Assert.AreEqual(5, tokens.Length);
Assert.AreEqual("kā arī", tokens[2].Text);
}
public void kaut_jel_milj_kgs()
{
LatvianTokenizer tokenizer = new LatvianTokenizer(compile: false);
tokenizer.Add<LuMiiAbbreviationToken>();
tokenizer.Add<LuMiiCollocationToken>();
tokenizer.Compile();
string text = "kaut JEL 2 milj. mans k-gs";
Token[] tokens = tokenizer.Tokenize(text).ToArray();
Assert.AreEqual(5, tokens.Length);
Assert.AreEqual("kaut JEL", tokens[0].Text);
Assert.AreEqual("milj.", tokens[2].Text);
Assert.AreEqual("k-gs", tokens[4].Text);
}
public void InitialsRemoved()
{
LatvianTokenizer tokenizer = new LatvianTokenizer(compile: false);
tokenizer.Remove<InitialsToken>();
tokenizer.Compile(); // optional
Token[] tokens = tokenizer.Tokenize("A.Bērziņš").ToArray();
Assert.AreEqual("A", tokens[0].Text);
Assert.AreEqual(".", tokens[1].Text);
Assert.AreEqual("Bērziņš", tokens[2].Text);
}