def test_legacy_hash_ascii_data(self):
"""Test legacy hash function against known results."""
splitter = splitters.Splitter()
file_name = os.path.join(os.path.dirname(__file__), 'files', 'sample-data.jsonl')
with open(file_name, 'r') as flo:
lines = flo.read().split('\n')
for line in lines:
if line is None or line == '':
continue
seed, key, hashed, bucket = json.loads(line)
assert hashfns.legacy.legacy_hash(key, seed) == hashed
assert splitter.get_bucket(key, seed, splits.HashAlgorithm.LEGACY) == bucket
def test_murmur_more_ascii_data(self):
"""Test legacy hash function against known results."""
splitter = splitters.Splitter()
file_name = os.path.join(os.path.dirname(__file__), 'files', 'murmur3-custom-uuids.csv')
with open(file_name, 'r') as flo:
lines = flo.read().split('\n')
for line in lines:
if line is None or line == '':
continue
seed, key, hashed, bucket = line.split(',')
seed = int(seed)
bucket = int(bucket)
hashed = int(hashed)
assert hashfns._murmur_hash(key, seed) == hashed
assert splitter.get_bucket(key, seed, splits.HashAlgorithm.MURMUR) == bucket
def test_murmur_hash_non_ascii_data(self):
"""Test legacy hash function against known results."""
splitter = splitters.Splitter()
file_name = os.path.join(
os.path.dirname(__file__), 'files',
'murmur3-sample-data-non-alpha-numeric-v2.csv')
with io.open(file_name, 'r', encoding='utf-8') as flo:
lines = flo.read().split('\n')
for line in lines:
if line is None or line == '':
continue
seed, key, hashed, bucket = line.split(',')
seed = int(seed)
bucket = int(bucket)
hashed = int(hashed)
assert hashfns._murmur_hash(key, seed) == hashed
assert splitter.get_bucket(key, seed,
splits.HashAlgorithm.MURMUR) == bucket
