def test_nsrl_saved(self):
bq = NSRLBloomQuery()
self._test_random_entries(bq)
for has in self.NOT_IN_NSRL:
self.assertFalse(bq.md5_exists(has),
"Hash in NSRL bloom when its not supposed to be {0}".format(self.NOT_IN_NSRL.index(has)))
def nsrl_unknowns_q(self):
# the file same list should be smaller then
bq = NSRLBloomQuery()
nsrl_members = [
pk for pk, md5 in filter(lambda x: bq.md5_exists(x[1]),
self.values_list('pk', 'md5'))
]
return ~Q(pk__in=nsrl_members)
def test_file_process_bloom(self):
bq = NSRLBloomQuery()
bq.save_file = tempfile.mktemp()
bq.nsrl_dir = NSRL_TEST_DATA_DIR
bq.generate_new_bloom()
# then do some queries
self._test_random_entries(bq)
bq.save_bloom()
print "File size {0}".format(os.path.getsize(bq.save_file))
def nsrl_unknowns_q(self):
# the file same list should be smaller then
bq = NSRLBloomQuery()
nsrl_members = [pk for pk, md5 in filter(lambda x: bq.md5_exists(x[1]), self.values_list('pk', 'md5'))]
return ~Q(pk__in=nsrl_members)
