@Test
public void testBloomFilterPredicateValuesExisting() throws Exception {
BloomFilter bloomFilter = new BloomFilter(TEST_VALUES.size() * 10, 0.01);
for (Object o : TEST_VALUES.keySet()) {
if (o instanceof Long) {
bloomFilter.addLong((Long) o);
} else if (o instanceof Integer) {
bloomFilter.addLong((Integer) o);
} else if (o instanceof String) {
bloomFilter.addString((String) o);
} else if (o instanceof BigDecimal) {
bloomFilter.addString(o.toString());
} else if (o instanceof Slice) {
bloomFilter.addString(((Slice) o).toStringUtf8());
} else if (o instanceof Timestamp) {
bloomFilter.addLong(((Timestamp) o).getTime());
} else if (o instanceof Double) {
bloomFilter.addDouble((Double) o);
} else {
fail("Unsupported type " + o.getClass());
}
}
for (Map.Entry<Object, Type> testValue : TEST_VALUES.entrySet()) {
boolean matched = checkInBloomFilter(bloomFilter, testValue.getKey(), testValue.getValue());
assertTrue(matched, "type " + testValue.getClass());
}
// test unsupported type: can be supported by ORC but is not implemented yet
assertTrue(
checkInBloomFilter(bloomFilter, new Date(), DATE),
"unsupported type DATE should always return true");
}
@Test
public void testHiveBloomFilterSerde() throws Exception {
BloomFilter bloomFilter = new BloomFilter(1_000_000L, 0.05);
// String
bloomFilter.addString(TEST_STRING);
assertTrue(bloomFilter.testString(TEST_STRING));
assertFalse(bloomFilter.testString(TEST_STRING_NOT_WRITTEN));
// Integer
bloomFilter.addLong(TEST_INTEGER);
assertTrue(bloomFilter.testLong(TEST_INTEGER));
assertFalse(bloomFilter.testLong(TEST_INTEGER + 1));
// Re-construct
HiveBloomFilter hiveBloomFilter =
new HiveBloomFilter(
ImmutableList.copyOf(Longs.asList(bloomFilter.getBitSet())),
bloomFilter.getBitSize(),
bloomFilter.getNumHashFunctions());
// String
assertTrue(hiveBloomFilter.testString(TEST_STRING));
assertFalse(hiveBloomFilter.testString(TEST_STRING_NOT_WRITTEN));
// Integer
assertTrue(hiveBloomFilter.testLong(TEST_INTEGER));
assertFalse(hiveBloomFilter.testLong(TEST_INTEGER + 1));
}