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;
}
}
