@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));
}
@Test
public void testOrcHiveBloomFilterSerde() throws Exception {
BloomFilter bloomFilterWrite = new BloomFilter(1000L, 0.05);
bloomFilterWrite.addString(TEST_STRING);
assertTrue(bloomFilterWrite.testString(TEST_STRING));
OrcProto.BloomFilter.Builder bloomFilterBuilder = OrcProto.BloomFilter.newBuilder();
bloomFilterBuilder.addAllBitset(Longs.asList(bloomFilterWrite.getBitSet()));
bloomFilterBuilder.setNumHashFunctions(bloomFilterWrite.getNumHashFunctions());
OrcProto.BloomFilter bloomFilter = bloomFilterBuilder.build();
OrcProto.BloomFilterIndex bloomFilterIndex = OrcProto.BloomFilterIndex.getDefaultInstance();
byte[] bytes = serializeBloomFilterToIndex(bloomFilter, bloomFilterIndex);
// Read through method
InputStream inputStream = new ByteArrayInputStream(bytes);
OrcMetadataReader metadataReader = new OrcMetadataReader();
List<HiveBloomFilter> bloomFilters = metadataReader.readBloomFilterIndexes(inputStream);
assertEquals(bloomFilters.size(), 1);
assertTrue(bloomFilters.get(0).testString(TEST_STRING));
assertFalse(bloomFilters.get(0).testString(TEST_STRING_NOT_WRITTEN));
assertEquals(bloomFilterWrite.getBitSize(), bloomFilters.get(0).getBitSize());
assertEquals(bloomFilterWrite.getNumHashFunctions(), bloomFilters.get(0).getNumHashFunctions());
// Validate bit set
assertTrue(Arrays.equals(bloomFilters.get(0).getBitSet(), bloomFilterWrite.getBitSet()));
// Read directly: allows better inspection of the bit sets (helped to fix a lot of bugs)
CodedInputStream input = CodedInputStream.newInstance(bytes);
OrcProto.BloomFilterIndex deserializedBloomFilterIndex =
OrcProto.BloomFilterIndex.parseFrom(input);
List<OrcProto.BloomFilter> bloomFilterList = deserializedBloomFilterIndex.getBloomFilterList();
assertEquals(bloomFilterList.size(), 1);
OrcProto.BloomFilter bloomFilterRead = bloomFilterList.get(0);
// Validate contents of ORC bloom filter bit set
assertTrue(
Arrays.equals(
Longs.toArray(bloomFilterRead.getBitsetList()), bloomFilterWrite.getBitSet()));
// hash functions
assertEquals(bloomFilterWrite.getNumHashFunctions(), bloomFilterRead.getNumHashFunctions());
// bit size
assertEquals(bloomFilterWrite.getBitSet().length, bloomFilterRead.getBitsetCount());
}