private void assertAutomaton(Automaton automaton) throws Exception {
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
final AutomatonTestUtil.RandomAcceptedStrings ras =
new AutomatonTestUtil.RandomAcceptedStrings(automaton);
int num = 1000 * RANDOM_MULTIPLIER;
for (int i = 0; i < num; i++) {
final String string;
if (random.nextBoolean()) {
// likely not accepted
string = _TestUtil.randomUnicodeString(random);
} else {
// will be accepted
int[] codepoints = ras.getRandomAcceptedString(random);
try {
string = UnicodeUtil.newString(codepoints, 0, codepoints.length);
} catch (Exception e) {
System.out.println(codepoints.length + " codepoints:");
for (int j = 0; j < codepoints.length; j++) {
System.out.println(" " + Integer.toHexString(codepoints[j]));
}
throw e;
}
}
byte bytes[] = string.getBytes("UTF-8");
assertEquals(cra.run(string), bra.run(bytes, 0, bytes.length));
}
}
/** Simple smoke test for regular expression. */
public void testSmoke() {
RegExp r = new RegExp("a(b+|c+)d");
Automaton a = r.toAutomaton();
assertTrue(a.isDeterministic());
CharacterRunAutomaton run = new CharacterRunAutomaton(a);
assertTrue(run.run("abbbbbd"));
assertTrue(run.run("acd"));
assertFalse(run.run("ad"));
}
public void testSpecialCase2() throws Exception {
RegExp re = new RegExp(".+\u0775");
String input = "\ufadc\ufffd\ub80b\uda5a\udc68\uf234\u0056\uda5b\udcc1\ufffd\ufffd\u0775";
Automaton automaton = re.toAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
assertTrue(cra.run(input));
byte[] bytes = input.getBytes("UTF-8");
assertTrue(bra.run(bytes, 0, bytes.length)); // this one fails!
}
public void testSpecialCase3() throws Exception {
RegExp re = new RegExp("(\\鯺)*(.)*\\Ӕ");
String input =
"\u5cfd\ufffd\ub2f7\u0033\ue304\u51d7\u3692\udb50\udfb3\u0576\udae2\udc62\u0053\u0449\u04d4";
Automaton automaton = re.toAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
assertTrue(cra.run(input));
byte[] bytes = input.getBytes("UTF-8");
assertTrue(bra.run(bytes, 0, bytes.length));
}
public void testSpecialCase() {
RegExp re = new RegExp(".?");
Automaton automaton = re.toAutomaton();
CharacterRunAutomaton cra = new CharacterRunAutomaton(automaton);
ByteRunAutomaton bra = new ByteRunAutomaton(automaton);
// make sure character dfa accepts empty string
assertTrue(cra.isAccept(cra.getInitialState()));
assertTrue(cra.run(""));
assertTrue(cra.run(new char[0], 0, 0));
// make sure byte dfa accepts empty string
assertTrue(bra.isAccept(bra.getInitialState()));
assertTrue(bra.run(new byte[0], 0, 0));
}
