class MongoTest(BaseTest, unittest.TestCase):
'''Test the Mongo client'''
def setUp(self):
self.client = Client('mongo', 'testing', 6, 3, host='localhost',
weeks=3)
def tearDown(self):
self.client.delete()
def find_duplicates(bithash_corpus, table_name, ntables, nmatchbits):
try:
redis_client = Client('redis', table_name, ntables, nmatchbits)
redis_client.delete()
except ConnectionError:
print 'Connection error. Is redis running? Execute:\n>> redis-server'
return -1
# insert the first 20k documents
redis_client.insert(bithash_corpus[:20000])
# Find duplicates of the remaining ~2k documents
dup_lists = redis_client.find_all(bithash_corpus[20001:])
idxs_and_simhashs = [(20001 + idx, dup_lists[idx])
for idx, val in enumerate(dup_lists) if len(val) > 0]
return idxs_and_simhashs