class FileBloomFilterDeduper(Deduper):
def __init__(self, working_dir, capacity, false_positive_rate=0.01):
super(FileBloomFilterDeduper, self).__init__(working_dir)
sync_file = os.path.join(self.working_dir,
BLOOM_FILETER_STATUS_FILENAME)
self.filter = FileBloomFilter(sync_file, capacity,
false_positive_rate=false_positive_rate)
self.is_shutdown = False
def exist(self, key):
return self.filter.verify(key)
def shutdown(self):
if self.is_shutdown is True:
return
self.is_shutdown = True
try:
self.filter.sync()
finally:
self.filter.close()
def __del__(self):
self.shutdown()
class FileBloomFilterDeduper(Deduper):
def __init__(self, sync_file, capacity):
self.filter = FileBloomFilter(sync_file, capacity)
def exist(self, key):
return self.filter.verify(key)
def __del__(self):
try:
self.filter.sync()
finally:
self.filter.close()
def __init__(self, working_dir, capacity, false_positive_rate=0.01):
super(FileBloomFilterDeduper, self).__init__(working_dir)
sync_file = os.path.join(self.working_dir,
BLOOM_FILETER_STATUS_FILENAME)
self.filter = FileBloomFilter(sync_file, capacity,
false_positive_rate=false_positive_rate)
self.is_shutdown = False
def _init_bloom_filter(self):
size = self.job.context.job.size
base = 1 if not self.job.is_bundle else 1000
bloom_filter_file = os.path.join(self.root, 'bloomfilter')
if not os.path.exists(bloom_filter_file):
if size > 0:
bloom_filter_size = size * 10 * base
else:
bloom_filter_size = UNLIMIT_BLOOM_FILTER_CAPACITY
else:
if size > 0:
bloom_filter_size = size * 2 * base
else:
bloom_filter_size = UNLIMIT_BLOOM_FILTER_CAPACITY
return FileBloomFilter(bloom_filter_file, bloom_filter_size)
def testPutGet(self):
num = str(12345)
self.assertEqual(self.node.put(num), num)
self.assertEqual(self.node.put(num), '')
num2 = str(67890)
nums = [num, num2]
self.assertEqual(self.node.put(nums), [num2])
self.node.shutdown()
self.assertGreater(
os.path.getsize(os.path.join(self.dir_, 'bloomfilter')), 0)
bloom_filter_hook = FileBloomFilter(
os.path.join(self.dir_, 'bloomfilter'), 5)
self.node = Node(self.node_dir, verify_exists_hook=bloom_filter_hook)
num3 = str(13579)
nums = [num, num2, num3]
self.assertEqual(self.node.put(nums), [num3])
def __init__(self, sync_file, capacity):
self.filter = FileBloomFilter(sync_file, capacity)
def __init__(self, sync_file, capacity):
self.filter = FileBloomFilter(sync_file, capacity)