class BloomFilterTestCase(unittest.TestCase):
def setUp(self):
self.bloom_filter = BloomFilter(247, 5)
self.words = ['hola', 'como', 'estan', 'holis']
self.bloom_filter.insert_list(self.words)
def test_initial_conditions(self):
for word in self.words:
self.assertEqual(self.bloom_filter.check(word), 1)
def test_word_i_know_not_in(self):
self.assertEqual(self.bloom_filter.check("xd"), 0)
def test_add_item_multiple_items(self):
'''
Tests that multiple items can be added to the Bloom Filter.
'''
bf = BloomFilter(4, 0.05)
bf.add('christian')
bf.add('daniel')
bf.add('debra')
bf.add('charles-adrian')
self.assertTrue(bf.check('christian'))
self.assertTrue(bf.check('daniel'))
self.assertTrue(bf.check('debra'))
self.assertTrue(bf.check('charles-adrian'))
def test_check_gives_response_regardless_of_key(self):
'''
Tests that the Bloom Filter will give a response
no matter the key. We cannot test any further, as we don't
know with certainty whether we'll get a false positive.
'''
bf = BloomFilter(1, 0.05)
bf.add('christian')
self.assertTrue(bf.check('christian'))
r1 = bf.check('daniel') in (True, False)
r2 = bf.check('charles') in (True, False)
self.assertTrue(r1)
self.assertTrue(r2)
def test_add_item_one_item(self):
'''
Tests that an item can be added to the Bloom Filter.
'''
bf = BloomFilter(1, 0.05)
bf.add('christian')
self.assertTrue(bf.check('christian'))
def test_add_item_different_probability_one_item(self):
'''
Tests that an item can be added to the Bloom Filter with
different false positive probabilities.
'''
bf = BloomFilter(1, 0.10)
bf.add('christian')
self.assertTrue(bf.check('christian'))
bf = BloomFilter(1, 0.5)
bf.add('christian')
self.assertTrue(bf.check('christian'))
bf = BloomFilter(1, 0.9)
bf.add('christian')
self.assertTrue(bf.check('christian'))
for i in range(ITER):
bloom_filter = BloomFilter(m, k)
initial_size_bf.append(get_deep_size(bloom_filter))
bloom_filter, L_file, universe_file = generate_files_and_insert_to_bloom_filter(
bloom_filter, big_n, small_n)
size_after_insertion_bf.append(get_deep_size(bloom_filter))
current_time_bf = []
current_time_fl = []
with open('experimentos/files/universe_file.txt',
'r') as universe_file:
total_usernames_list = [
w.strip('\n') for w in universe_file.readlines()
]
for username_query in total_usernames_list:
ti_bf = time.time()
username_might_be_in_file = bloom_filter.check(
username_query)
tf_bf = time.time()
dt_bf = tf_bf - ti_bf
current_time_bf.append(dt_bf)
if username_might_be_in_file == 1:
disk_access[i] += 1
ti_fl = time.time()
username_in_file = search_username_in_file(
username_query, 'experimentos/files/L_file.txt')
tf_fl = time.time()
dt_fl = (tf_fl - ti_fl) * 19
current_time_fl.append(dt_fl)
if username_in_file is None:
false_positives[i] += 1
universe_file.close()
search_times_fl.append(
def test_add_item_different_probability_multiple_items(self):
'''
Tests that multiple items can be added to the Bloom Filter with
different false positive probabilities.
'''
bf = BloomFilter(4, 0.10)
bf.add('christian')
bf.add('daniel')
bf.add('debra')
bf.add('charles-adrian')
self.assertTrue(bf.check('christian'))
self.assertTrue(bf.check('daniel'))
self.assertTrue(bf.check('debra'))
self.assertTrue(bf.check('charles-adrian'))
bf = BloomFilter(4, 0.50)
bf.add('christian')
bf.add('daniel')
bf.add('debra')
bf.add('charles-adrian')
self.assertTrue(bf.check('christian'))
self.assertTrue(bf.check('daniel'))
self.assertTrue(bf.check('debra'))
self.assertTrue(bf.check('charles-adrian'))
bf = BloomFilter(4, 0.9)
bf.add('christian')
bf.add('daniel')
bf.add('debra')
bf.add('charles-adrian')
self.assertTrue(bf.check('christian'))
self.assertTrue(bf.check('daniel'))
self.assertTrue(bf.check('debra'))
self.assertTrue(bf.check('charles-adrian'))