def test_hash_doc_ids(self):
N = 1001
doc_ids = [str(i) for i in range(N)]
hashes = ShardAccessor.hash_doc_ids_sql_for_testing(doc_ids)
self.assertEqual(len(hashes), N)
self.assertTrue(
all(isinstance(hash_, int) for hash_ in hashes.values()))
def test_hash_in_python(self):
# test that python hashing matches with SQL hashing
N = 2048
doc_ids = [str(i) for i in range(N)]
sql_hashes = ShardAccessor.hash_doc_ids_sql_for_testing(doc_ids)
csiphash_hashes = ShardAccessor.hash_doc_ids_python(doc_ids)
self.assertEquals(len(csiphash_hashes), N)
self.assertTrue(all(isinstance(hash_, six.integer_types) for hash_ in csiphash_hashes.values()))
N_shards = 1024
part_mask = N_shards - 1
sql_shards = {doc_id: hash_ & part_mask for doc_id, hash_ in sql_hashes.items()}
python_shards = {doc_id: hash_ & part_mask for doc_id, hash_ in sql_hashes.items()}
self.assertEqual(python_shards, sql_shards)
def test_hash_doc_ids(self):
N = 1001
doc_ids = [str(i) for i in range(N)]
hashes = ShardAccessor.hash_doc_ids_sql_for_testing(doc_ids)
self.assertEquals(len(hashes), N)
self.assertTrue(all(isinstance(hash_, int) for hash_ in hashes.values()))
