@Before
public void setup() {
stream =
new EmoticonTokenCombiner(new LatinTokenizer.Builder().setKeepPunctuation(true).build());
termAttr = stream.getAttribute(CharSequenceTermAttribute.class);
typeAttr = stream.getAttribute(TokenTypeAttribute.class);
}
private void verify(List<String> tokens, TokenType... types) {
for (int i = 0; i < tokens.size(); i++) {
assertTrue(stream.incrementToken());
assertEquals(tokens.get(i), termAttr.getTermString());
assertEquals(types[i], typeAttr.getType());
}
assertFalse(stream.incrementToken());
}
@Test
public void test() {
stream.reset("this is a smiley :)");
verify(
ImmutableList.of("this", "is", "a", "smiley", ":)"),
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.EMOTICON);
stream.reset("sad smiley :-(");
verify(
ImmutableList.of("sad", "smiley", ":-("),
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.EMOTICON);
stream.reset("smiley with space : )");
verify(
ImmutableList.of("smiley", "with", "space", ": )"),
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.EMOTICON);
stream.reset("First smiley. :p Second smiley :D False smiley :((");
verify(
ImmutableList.of(
"First", "smiley", ".", ":p", "Second", "smiley", ":D", "False", "smiley", ":", "(",
"("),
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.PUNCTUATION,
TokenType.EMOTICON,
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.EMOTICON,
TokenType.TOKEN,
TokenType.TOKEN,
TokenType.PUNCTUATION,
TokenType.PUNCTUATION,
TokenType.PUNCTUATION);
}
