private void checkSanity(BloomFilter<Object> bf) {
assertFalse(bf.mightContain(new Object()));
assertFalse(bf.apply(new Object()));
for (int i = 0; i < 100; i++) {
Object o = new Object();
bf.put(o);
assertTrue(bf.mightContain(o));
assertTrue(bf.apply(o));
}
}
public void testCreateAndCheckMitz32BloomFilterWithKnownFalsePositives() {
int numInsertions = 1000000;
BloomFilter<String> bf =
BloomFilter.create(
Funnels.unencodedCharsFunnel(),
numInsertions,
0.03,
BloomFilterStrategies.MURMUR128_MITZ_32);
// Insert "numInsertions" even numbers into the BF.
for (int i = 0; i < numInsertions * 2; i += 2) {
bf.put(Integer.toString(i));
}
// Assert that the BF "might" have all of the even numbers.
for (int i = 0; i < numInsertions * 2; i += 2) {
assertTrue(bf.mightContain(Integer.toString(i)));
}
// Now we check for known false positives using a set of known false positives.
// (These are all of the false positives under 900.)
ImmutableSet<Integer> falsePositives =
ImmutableSet.of(
49, 51, 59, 163, 199, 321, 325, 363, 367, 469, 545, 561, 727, 769, 773, 781);
for (int i = 1; i < 900; i += 2) {
if (!falsePositives.contains(i)) {
assertFalse("BF should not contain " + i, bf.mightContain(Integer.toString(i)));
}
}
// Check that there are exactly 29824 false positives for this BF.
int knownNumberOfFalsePositives = 29824;
int numFpp = 0;
for (int i = 1; i < numInsertions * 2; i += 2) {
if (bf.mightContain(Integer.toString(i))) {
numFpp++;
}
}
assertEquals(knownNumberOfFalsePositives, numFpp);
double actualFpp = (double) knownNumberOfFalsePositives / numInsertions;
double expectedFpp = bf.expectedFpp();
// The normal order of (expected, actual) is reversed here on purpose.
assertEquals(actualFpp, expectedFpp, 0.00015);
}
public void testCreateAndCheckBloomFilterWithKnownUtf8FalsePositives64() {
int numInsertions = 1000000;
BloomFilter<String> bf =
BloomFilter.create(
Funnels.stringFunnel(UTF_8),
numInsertions,
0.03,
BloomFilterStrategies.MURMUR128_MITZ_64);
// Insert "numInsertions" even numbers into the BF.
for (int i = 0; i < numInsertions * 2; i += 2) {
bf.put(Integer.toString(i));
}
// Assert that the BF "might" have all of the even numbers.
for (int i = 0; i < numInsertions * 2; i += 2) {
assertTrue(bf.mightContain(Integer.toString(i)));
}
// Now we check for known false positives using a set of known false positives.
// (These are all of the false positives under 900.)
ImmutableSet<Integer> falsePositives = ImmutableSet.of(129, 471, 723, 89, 751, 835, 871);
for (int i = 1; i < 900; i += 2) {
if (!falsePositives.contains(i)) {
assertFalse("BF should not contain " + i, bf.mightContain(Integer.toString(i)));
}
}
// Check that there are exactly 29763 false positives for this BF.
int knownNumberOfFalsePositives = 29763;
int numFpp = 0;
for (int i = 1; i < numInsertions * 2; i += 2) {
if (bf.mightContain(Integer.toString(i))) {
numFpp++;
}
}
assertEquals(knownNumberOfFalsePositives, numFpp);
double actualFpp = (double) knownNumberOfFalsePositives / numInsertions;
double expectedFpp = bf.expectedFpp();
// The normal order of (expected, actual) is reversed here on purpose.
assertEquals(actualFpp, expectedFpp, 0.00033);
}
public void testPutReturnValue() {
for (int i = 0; i < 10; i++) {
BloomFilter<String> bf = BloomFilter.create(Funnels.unencodedCharsFunnel(), 100);
for (int j = 0; j < 10; j++) {
String value = new Object().toString();
boolean mightContain = bf.mightContain(value);
boolean put = bf.put(value);
assertTrue(mightContain != put);
}
}
}
public void testJavaSerialization() {
BloomFilter<byte[]> bf = BloomFilter.create(Funnels.byteArrayFunnel(), 100);
for (int i = 0; i < 10; i++) {
bf.put(Ints.toByteArray(i));
}
BloomFilter<byte[]> copy = SerializableTester.reserialize(bf);
for (int i = 0; i < 10; i++) {
assertTrue(copy.mightContain(Ints.toByteArray(i)));
}
assertEquals(bf.expectedFpp(), copy.expectedFpp());
SerializableTester.reserializeAndAssert(bf);
}
public void testBloom() throws Exception {
int numInsertions = 1000000;
double fpp = 0.03D;
Random random = new Random(1L);
BloomFilter<Long> filter = BloomFilter.create(Funnels.longFunnel(), numInsertions, fpp);
for (int l = 0; l < numInsertions; l++) {
filter.put(random.nextLong());
}
random = new Random(1L);
for (int l = 0; l < numInsertions; l++) {
assertTrue(filter.mightContain(random.nextLong()));
}
}
protected String redirect(String title) {
if (useBloomFilter && !redirectFilter.mightContain(title)) return title;
String to = redirects.get(title);
int count = 0;
while (to != null && !to.equals(title) && count++ <= 50) {
title = to;
to = redirects.get(title);
}
if (count >= 50) {
System.out.println(
"Fixed point reached with title : "
+ title
+ " ; stopping to loop in redirects at this moment");
}
return title;
}
public void testPutAll() {
int element1 = 1;
int element2 = 2;
BloomFilter<Integer> bf1 = BloomFilter.create(Funnels.integerFunnel(), 100);
bf1.put(element1);
assertTrue(bf1.mightContain(element1));
assertFalse(bf1.mightContain(element2));
BloomFilter<Integer> bf2 = BloomFilter.create(Funnels.integerFunnel(), 100);
bf2.put(element2);
assertFalse(bf2.mightContain(element1));
assertTrue(bf2.mightContain(element2));
assertTrue(bf1.isCompatible(bf2));
bf1.putAll(bf2);
assertTrue(bf1.mightContain(element1));
assertTrue(bf1.mightContain(element2));
assertFalse(bf2.mightContain(element1));
assertTrue(bf2.mightContain(element2));
}
@Override
public boolean hasSeen(LookupKey key) {
return filter.mightContain(key);
}