def test_intersection_capacity_fail(self):
bloom_one = BloomFilter(1000, 0.001)
bloom_two = BloomFilter(100, 0.001)
def _run():
_ = bloom_one.intersection(bloom_two)
self.assertRaises(ValueError, _run)
def test_union_k_fail(self):
bloom_one = BloomFilter(100, 0.01)
bloom_two = BloomFilter(100, 0.001)
def _run():
_ = bloom_one.union(bloom_two)
self.assertRaises(ValueError, _run)
def test_union_size(self):
fpr = 0.001
# False positive rate with small numbers is high, therefore let's test with bigger sets
bloom_one = BloomFilter(100000, fpr)
bloom_two = BloomFilter(100000, fpr)
listA = [str(random.getrandbits(8)) for i in range(10000)]
listB = [str(random.getrandbits(8)) for i in range(10000)]
for char in listA:
bloom_one.add(char)
for char in listB:
bloom_two.add(char)
merged_bloom = bloom_one.union(bloom_two)
bloom_one_count = bloom_one.count
bloom_two_count = bloom_two.count
listA_uniq_count = len(set(listA))
listB_uniq_count = len(set(listB))
merged_bloom_count = merged_bloom.count
listAB_uniq_count = len(set(listA).union(set(listB)))
assert bloom_one_count == listA_uniq_count
assert bloom_two_count == listB_uniq_count
assert (listAB_uniq_count * (1 - fpr) <= merged_bloom_count <= listAB_uniq_count * (1 + fpr))
def test_intersection_size(self):
fpr = 0.001
# False positive rate with small numbers is high, therefore let's test with bigger sets
bloom_one = BloomFilter(100000, fpr)
bloom_two = BloomFilter(100000, fpr)
listA = [str(random.getrandbits(14)) for i in range(71000)]
listB = [str(random.getrandbits(12)) for i in range(69000)]
for char in listA:
bloom_one.add(char)
for char in listB:
bloom_two.add(char)
merged_bloom = bloom_one.intersection(bloom_two)
bloom_one_count = bloom_one.count
bloom_two_count = bloom_two.count
listA_uniq_count = len(set(listA))
listB_uniq_count = len(set(listB))
merged_bloom_count = merged_bloom.count
listAB_uniq_count = len(set(listA).intersection(set(listB)))
assert bloom_one_count == listA_uniq_count
assert bloom_two_count == listB_uniq_count
# Intersection guarantees to have all elements of the intersection but the false positive rate might be slightly higher than that of the pure intersection:
assert (listAB_uniq_count * (1 - 2 * fpr) <= merged_bloom_count <= listAB_uniq_count * (1 + 2 * fpr))
def initBloomFilter(stringUnique):
'''
keyHashMat is (hahsLenth * keyNum)
:param stringUnique:
:return:
'''
print('---initialize Bloom Filter---')
BF = BloomFilter(capacity=int(len(stringUnique) * 1.2), error_rate=0.30)
keyHashMat = []
for i in stringUnique:
BF.add(i)
keyHashMat.append([int(x) for x in BF.keyhash(i)])
keyHashMat = np.mat(keyHashMat).T
return BF, BF.num_bits, BF.bitarray, keyHashMat
def test_union(self):
bloom_one = BloomFilter(100, 0.001)
bloom_two = BloomFilter(100, 0.001)
chars = [chr(i) for i in range(97, 123)]
for char in chars[int(len(chars) / 2):]:
bloom_one.add(char)
for char in chars[:int(len(chars) / 2)]:
bloom_two.add(char)
new_bloom = bloom_one.union(bloom_two)
for char in chars:
self.assertTrue(char in new_bloom)
def test_union(self):
bloom_one = BloomFilter(100, 0.001)
bloom_two = BloomFilter(100, 0.001)
chars = [chr(i) for i in range_fn(97, 123)]
for char in chars[int(len(chars)/2):]:
bloom_one.add(char)
for char in chars[:int(len(chars)/2)]:
bloom_two.add(char)
new_bloom = bloom_one.union(bloom_two)
for char in chars:
self.assertTrue(char in new_bloom)
from pybloom.pybloom import BloomFilter
import bitarray
import numpy as np
from sklearn import linear_model
from scipy.linalg import solve
f = BloomFilter(capacity=60, error_rate=0.30)
[f.add(x) for x in range(50)]
# print(9 in f)
# print('num of key: '+str(len(f)))
print(f.bitarray) #the final array
# f.add('10')
# print(f.bitarray)
# print(len(f.bitarray))
print(f.num_bits) #bloomFilter's size(length)
#
# a = bitarray.bitarray('0'*10)
# print(a)
print(f.keyhash(5))
#
#
#
# a = np.mat(np.array([[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 1, ]]).T)
# y = np.mat(np.array([1, 1, 2, 2])).T
#
# num = np.mat(np.array([1, 1, 2])).T
#
filter = klass(*args)
for item in self.EXPECTED:
filter.add(item)
f = tempfile.TemporaryFile()
filter.tofile(f)
stringio = StringIO()
filter.tofile(stringio)
streams_to_test = [f, stringio]
if not running_python_3:
cstringio = cStringIO.StringIO()
filter.tofile(cstringio)
streams_to_test.append(cstringio)
del filter
for stream in streams_to_test:
stream.seek(0)
filter = klass.fromfile(stream)
for item in self.EXPECTED:
self.assertTrue(item in filter)
del(filter)
stream.close()
if __name__ == '__main__':
# unittest.main()
f = BloomFilter(capacity=10000, error_rate=0.001)
for i in range_fn(0, f.capacity):
f.add(i)
print (0 in f)
import tempfile
import time
from pybloom.pybloom import BloomFilter
NS = 10**9
for _p in xrange(1, 3):
p = 10 ** _p
for e in xrange(9):
X = int(1000 * 10 ** (e / 2.0))
print X, p,
bloomfilter = BloomFilter(X + 1, 1.0/p)
t = time.time()
for x in xrange(X):
bloomfilter.add(x)
print (time.time() - t) / X * NS,
t = time.time()
for x in xrange(X):
x in bloomfilter
print (time.time() - t) / X * NS,
t = time.time()
for x in xrange(X, 2*X):
x in bloomfilter
print (time.time() - t ) / X * NS
import redis
import csv
from pybloom.pybloom import BloomFilter
redis_client = redis.StrictRedis(host='192.168.192.12', port=6381, db=2)
filter_key = 'test'
redis_client.delete(filter_key)
bf = BloomFilter(redis_client, filter_key, 100000000, 0.0000001)
id = '1000'
if not bf.contain(id):
bf.add(id)
else:
print("already in")
import tempfile
import time
from pybloom.pybloom import BloomFilter
NS = 10**9
for _p in xrange(1, 3):
p = 10**_p
for e in xrange(9):
X = int(1000 * 10**(e / 2.0))
print X, p,
bloomfilter = BloomFilter(X + 1, 1.0 / p)
t = time.time()
for x in xrange(X):
bloomfilter.add(x)
print(time.time() - t) / X * NS,
t = time.time()
for x in xrange(X):
x in bloomfilter
print(time.time() - t) / X * NS,
t = time.time()
for x in xrange(X, 2 * X):
x in bloomfilter
print(time.time() - t) / X * NS
def get_filter(shingles):
f = BloomFilter(capacity=10000, error_rate=0.001)
for sg in shingles:
f.add(" ".join(sg))
return f
import redis
import csv
from pybloom.pybloom import BloomFilter
redis_client = redis.StrictRedis(host='192.168.192.12', port=6381, db=2)
filter_key = 'test'
redis_client.delete(filter_key)
bf = BloomFilter(redis_client, filter_key, 100000000, 0.0000001)
id = '1000'
if not bf.contain(id):
bf.add(id)
else:
print("already in")
def test_nstar(self):
bloom = BloomFilter(1000, 0.001)
chars = [chr(i) for i in range_fn(0,200)]
for char in chars:
bloom.add(char)
self.assertTrue(bloom.nstar() > len(chars)-10 and bloom.nstar() < len(chars)+10)
def test_nstar_intersection_2(self):
bloom_one = BloomFilter(200, 0.001)
bloom_two = BloomFilter(200, 0.001)
chars = [chr(i) for i in range_fn(0, 200)]
for char in chars[int(len(chars)/2):]:
bloom_one.add(char)
for char in chars[:int(len(chars)/2)]:
bloom_two.add(char)
new_bloom = bloom_one.intersection(bloom_two)
self.assertTrue(bloom_one.nstar() > len(chars)/2-10 and bloom_one.nstar() < len(chars)/2+10)
self.assertTrue(bloom_two.nstar() > len(chars)/2-10 and bloom_two.nstar() < len(chars)/2+10)
#The nstar operator will fail on the intersection of the filters..
self.assertTrue(new_bloom.nstar() > 10)
self.assertTrue(bloom_one.nstar_intersection(bloom_two) < 10)
def test_nstar_union(self):
bloom_one = BloomFilter(200, 0.001)
bloom_two = BloomFilter(200, 0.001)
chars = [chr(i) for i in range_fn(0, 200)]
for char in chars[:int(len(chars)/2)]:
bloom_one.add(char)
for char in chars[int(len(chars)/2):]:
bloom_two.add(char)
new_bloom = bloom_one.union(bloom_two)
self.assertTrue(bloom_one.nstar() > len(chars)/2-10 and bloom_one.nstar() < len(chars)/2+10)
self.assertTrue(bloom_two.nstar() > len(chars)/2-10 and bloom_two.nstar() < len(chars)/2+10)
self.assertTrue(new_bloom.nstar() > len(chars)-10 and new_bloom.nstar() < len(chars)+10)
