@Override
public void apply(TokenStream stream) throws TokenizerException {
// TODO Auto-generated method stub
if (stream != null) {
String token;
Stemmer s;
while (stream.hasNext()) {
token = stream.next();
if (token != null) {
token = token.toLowerCase();
if (isLettersOnly(token)) {
s = new Stemmer();
for (char c : token.toCharArray()) {
s.add(c);
}
s.stem();
stream.set(s.toString());
}
}
}
stream.reset();
}
}
@Override
public void apply(TokenStream stream) throws TokenizerException {
if (stream != null) {
String token;
while (stream.hasNext()) {
token = stream.next();
stream.previous();
if (token != null) {
token = replaceDate(token);
if (token.isEmpty()) stream.remove();
else {
stream.set(token);
stream.next();
}
}
}
}
stream.reset();
}
/**
* Test method for {@link
* edu.buffalo.cse.ir.wikiindexer.tokenizer.TokenStream#set(java.lang.String[])}.
*/
@Test
public void testSet() {
// set on null and empty streams
TokenStream stream = new TokenStream((String) null);
stream.set("invalid");
assertNull(stream.getAllTokens());
stream = null;
stream = new TokenStream("");
stream.set("invalid");
assertNull(stream.getAllTokens());
stream = null;
// valid posiiton, null or empty tokens
stream = new TokenStream("this");
stream.append("is", "a", "test", "stream");
stream.set((String) null);
assertEquals("this", stream.next());
stream.previous();
stream.set("");
assertEquals("this", stream.next());
stream = null;
// valid new token, invalid position
stream = new TokenStream("this");
stream.append("is", "a", "test", "stream");
stream.seekEnd();
stream.set("valid");
assertEquals("stream", stream.previous());
stream = null;
// correct set, single token
stream = new TokenStream("this");
stream.append("is", "a", "test", "stream");
stream.set("that");
assertEquals(5, stream.getAllTokens().size());
assertEquals("that", stream.next());
stream = null;
// correct set, multiple tokens at the end
stream = new TokenStream("this");
stream.append("is", "a", "test", "stream");
stream.seekEnd();
stream.previous();
stream.set("of", "the", "set", "method");
assertEquals(8, stream.getAllTokens().size());
assertEquals("method", stream.next());
stream = null;
// correct set, multiple tokens at the start
stream = new TokenStream("this");
stream.append("is", "a", "test", "stream");
stream.set("you", "think", "styx");
assertEquals(7, stream.getAllTokens().size());
assertEquals("styx", stream.next());
stream = null;
// correct set, multiple tokens in the middle
stream = new TokenStream("this");
stream.append("is", "a", "test", "stream");
stream.seekEnd();
stream.previous();
stream.previous();
stream.set("really", "interesting");
assertEquals(6, stream.getAllTokens().size());
assertEquals("interesting", stream.next());
assertEquals("stream", stream.next());
assertFalse(stream.hasNext());
stream = null;
}