def test_hashing_PyNaCl(self):
from PQencryption.hashing import sha_512_hashlib
from PQencryption.hashing import sha_512_PyNaCl
salt = "a" * 128
message = "This is a message. Hash me!"
hashed_PyNaCl = sha_512_PyNaCl.sha512_hash(salt, message)
self.assertEqual(
hashed_PyNaCl,
"ab90b1da9cd3a8625a75a0e0aaaa5c7a14ab9dde9006d23cacac665cc0edbc9309d8cc715aaf715cbcad61e9ddb32eac785881e880bff32c22108cb58cf6a8bf"
)
def example_hashing_PyNaCl():
from PQencryption.hashing import sha_512_PyNaCl
# In production the salt should come from a hardware random number generator
# and will be shared between parties.
# Salt must be 128 bytes in hex.
salt = "a" * 128
message = "This is a message. Hash me!"
print(message)
hashed = sha_512_PyNaCl.sha512_hash(salt, message)
print(hashed)
# make sure all memory is flushed after operations
del salt
del message
gc.collect()
# ["housenum", "zipcode", "date_of_birth", "sex"]
# act_data = act_data.drop(input['internal_id'], axis=1)
# 2.3 PI columns (personal identifier)
hashedPI = []
if df[PI].isnull().sum().sum() == 0:
for i in range(0, len(df)):
combine_PI = str()
for j in range(0, len(PI)):
id_feature = df.iloc[i][PI[j]]
combine_PI = combine_PI + str(id_feature)
# Remove space from strings #
combine_PI = combine_PI.replace(' ', '')
hashed = sha_512_PyNaCl.sha512_hash(salt, combine_PI.encode('UTF-8'))
hashedPI.append(hashed)
else:
print(
"Work cannot be done because missing values are in personal identifiers!"
)
print(df.isnull().sum())
# print(len(hashedPI))
hashedPI_df = pd.DataFrame(hashedPI, columns=['encString'])
hashedData_df = pd.concat([hashedPI_df, act_data], axis=1, join='inner')
print(len(hashedData_df))
hashedData_df.to_csv("/data/encrypted_%s.csv" % (input['party_name']),
index=None,