public void TestSimpleRegex() { TokenAssertions.AssertTokens("a token, 100 23 1g 144kg other token", _filter, new Token("a", 0), new Token("token", 2), new Token("other", 25), new Token("token", 31), null ); }
public void TestWordFilterChaining() { string[] filteredWords = { "de", "com" }; string text = "Bolo dé Açafrão com Rúcula"; TokenAssertions.AssertTokens(text, new WordFilter(new SpecialCharactersFilter(), filteredWords), new Token("bolo", 0), new Token("acafrao", 8), new Token("rucula", 20), null ); }
public void TestFilter() { string[] filteredWords = { "de", "com" }; string text = "Bolo de chocolate com calda de morango"; TokenAssertions.AssertTokens(text, new WordFilter(filteredWords), new Token("Bolo", 0), new Token("chocolate", 8), new Token("calda", 22), new Token("morango", 31), null); }
public void TestReplacements() { string text = "áéíóú calção aviões aiouao"; TokenAssertions.AssertTokens(text, new SpecialCharactersFilter(), new Token("aeiou", 0), new Token("calcao", 6), new Token("avioes", 13), new Token("aiouao", 20), null ); }
public void TestSimpleStrings() { string text = "a foo Bar a�a\n45\n\n\n"; TokenAssertions.AssertTokens(new StringTokenizer(text), new Token("a", 0), new Token("foo", 2), new Token("Bar", 6), new Token("a�a", 10), new Token("45", 14), null ); TokenAssertions.AssertTokens(new StringTokenizer(""), (Token)null); TokenAssertions.AssertTokens(new StringTokenizer("\n\t "), (Token)null); TokenAssertions.AssertTokens(new StringTokenizer("\n\t a"), new Token("a", 4), null); }