public void testSimpleMatching() throws Exception {
DefaultTokenizer tokenizer = new DefaultTokenizer(configuration);
List<Token> tokens;
tokens = tokenizer.tokenize("hello world");
TokenPatternImpl tp1 = new TokenPatternImpl("hello world", tokens, configuration);
assertEquals("aaaaa aaaaa", tp1.toSymbolicString());
tokens = tokenizer.tokenize("hello pinnochio");
assertTrue(tp1.match(tokens));
assertEquals("aaaaa aaaaaaaaa", tp1.toSymbolicString());
tokens = tokenizer.tokenize("hello you");
assertTrue(tp1.match(tokens));
assertEquals("aaaaa aaaaaaaaa", tp1.toSymbolicString());
tokens = tokenizer.tokenize("hello Mr. FanDango");
assertFalse(tp1.match(tokens));
assertEquals("aaaaa aaaaaaaaa", tp1.toSymbolicString());
configuration.setDiscriminateTextCase(true);
tokens = tokenizer.tokenize("hello Mr. FanDango");
TokenPatternImpl tp2 = new TokenPatternImpl("hello Mr. FanDango", tokens, configuration);
assertEquals("aaaaa Aa. AaaAaaaa", tp2.toSymbolicString());
}
public void testNoneExpandableMatching() throws Exception {
DefaultTokenizer tokenizer = new DefaultTokenizer(configuration);
TokenPatternImpl tp;
// both cases non-expandable
configuration.setLowerCaseExpandable(false);
configuration.setUpperCaseExpandable(false);
tp = new TokenPatternImpl("Hello", tokenizer.tokenize("Hello"), configuration);
assertTrue(tp.match(tokenizer.tokenize("Wooop")));
assertFalse(tp.match(tokenizer.tokenize("Greetings")));
assertFalse(tp.match(tokenizer.tokenize("Hi")));
// both cases expandable
configuration.setLowerCaseExpandable(true);
configuration.setUpperCaseExpandable(true);
tp = new TokenPatternImpl("Hello", tokenizer.tokenize("Hello"), configuration);
assertTrue(tp.match(tokenizer.tokenize("Wooop")));
assertTrue(tp.match(tokenizer.tokenize("Greetings")));
assertTrue(tp.match(tokenizer.tokenize("Hi")));
assertTrue(tp.match(tokenizer.tokenize("HHi")));
// only lower case expandable
configuration.setLowerCaseExpandable(true);
configuration.setUpperCaseExpandable(false);
tp = new TokenPatternImpl("Hello", tokenizer.tokenize("Hello"), configuration);
assertTrue(tp.match(tokenizer.tokenize("Wooop")));
assertTrue(tp.match(tokenizer.tokenize("Greetings")));
assertTrue(tp.match(tokenizer.tokenize("Hi")));
assertFalse(tp.match(tokenizer.tokenize("HHi")));
// only upper case expandable
configuration.setLowerCaseExpandable(false);
configuration.setUpperCaseExpandable(true);
tp = new TokenPatternImpl("Hello", tokenizer.tokenize("Hello"), configuration);
assertTrue(tp.match(tokenizer.tokenize("Wooop")));
assertFalse(tp.match(tokenizer.tokenize("Greetings")));
assertFalse(tp.match(tokenizer.tokenize("Hi")));
assertTrue(tp.match(tokenizer.tokenize("HHiiii")));
}