private void assertMatchEquals(List<Completion> res, String... expected) {
String[] result = new String[res.size()];
for (int i = 0; i < res.size(); i++) {
result[i] = res.get(i).toString();
}
if (!Arrays.equals(stripScore(expected), stripScore(result))) {
int colLen = Math.max(maxLen(expected), maxLen(result));
StringBuilder b = new StringBuilder();
String format = "%" + colLen + "s " + "%" + colLen + "s\n";
b.append(String.format(Locale.ROOT, format, "Expected", "Result"));
for (int i = 0; i < Math.max(result.length, expected.length); i++) {
b.append(
String.format(
Locale.ROOT,
format,
i < expected.length ? expected[i] : "--",
i < result.length ? result[i] : "--"));
}
System.err.println(b.toString());
fail("Expected different output:\n" + b.toString());
}
}
public void testThreeByte() throws Exception {
String key =
new String(new byte[] {(byte) 0xF0, (byte) 0xA4, (byte) 0xAD, (byte) 0xA2}, "UTF-8");
FSTCompletionBuilder builder = new FSTCompletionBuilder();
builder.add(new BytesRef(key), 0);
FSTCompletion lookup = builder.build();
List<Completion> result = lookup.lookup(stringToCharSequence(key), 1);
assertEquals(1, result.size());
}
public void testRandom() throws Exception {
List<TermFreq> freqs = new ArrayList<TermFreq>();
Random rnd = random();
for (int i = 0; i < 2500 + rnd.nextInt(2500); i++) {
int weight = rnd.nextInt(100);
freqs.add(new TermFreq("" + rnd.nextLong(), weight));
}
FSTCompletionLookup lookup = new FSTCompletionLookup();
lookup.build(new TermFreqArrayIterator(freqs.toArray(new TermFreq[freqs.size()])));
for (TermFreq tf : freqs) {
final String term = tf.term.utf8ToString();
for (int i = 1; i < term.length(); i++) {
String prefix = term.substring(0, i);
for (LookupResult lr : lookup.lookup(stringToCharSequence(prefix), true, 10)) {
assertTrue(lr.key.toString().startsWith(prefix));
}
}
}
}
public void testMultilingualInput() throws Exception {
List<TermFreq> input = LookupBenchmarkTest.readTop50KWiki();
FSTCompletionLookup lookup = new FSTCompletionLookup();
lookup.build(new TermFreqArrayIterator(input));
for (TermFreq tf : input) {
assertNotNull(
"Not found: " + tf.term.toString(),
lookup.get(_TestUtil.bytesToCharSequence(tf.term, random())));
assertEquals(
tf.term.utf8ToString(),
lookup
.lookup(_TestUtil.bytesToCharSequence(tf.term, random()), true, 1)
.get(0)
.key
.toString());
}
List<LookupResult> result = lookup.lookup(stringToCharSequence("wit"), true, 5);
assertEquals(5, result.size());
assertTrue(result.get(0).key.toString().equals("wit")); // exact match.
assertTrue(result.get(1).key.toString().equals("with")); // highest count.
}
public void testLargeInputConstantWeights() throws Exception {
FSTCompletionLookup lookup = new FSTCompletionLookup(10, true);
Random r = random();
List<TermFreq> keys = new ArrayList<TermFreq>();
for (int i = 0; i < 5000; i++) {
keys.add(new TermFreq(_TestUtil.randomSimpleString(r), -1));
}
lookup.build(new TermFreqArrayIterator(keys));
// All the weights were constant, so all returned buckets must be constant, whatever they
// are.
Long previous = null;
for (TermFreq tf : keys) {
Long current =
((Number) lookup.get(_TestUtil.bytesToCharSequence(tf.term, random()))).longValue();
if (previous != null) {
assertEquals(previous, current);
}
previous = current;
}
}