def _process_row_series(row, rule, pattern, required_attr, config):
"""
Compute the sha256 string for one rule.
.. seealso::
:`_process_frame`
:param config: dictionary with run-time parameters
:rtype: string
:return sha_string:
"""
patient = NormalizedPatient(
patid=row[COL_PATID],
pat_first_name=row[COL_FIRST],
pat_last_name=row[COL_LAST],
pat_birth_date=HashGenerator.format_date(row[COL_DOB]),
pat_sex=row[COL_SEX],
pat_race=HashGenerator.format_race(row[COL_RACE]))
if not patient.has_all_data(required_attr):
# print("Skip hashing patient [{}] due to missing data for rule [{}]".format(patient.patid, rule)) # noqa
return ''
raw = pattern.format(patient) + config['SALT']
sha_string = utils.apply_sha256(raw)
# print("For patient [{}] (rule {}): {}, hash_string= {}".format(patient.patid, rule, raw, sha_string)) # noqa
return sha_string
def _process_row_series(cls, ser, rule, pattern, required_attr, config):
"""
Compute the sha string for one rule.
:param config: dictionary with run-time parameters
:rtype: string
:return sha_string:
"""
patient = NormalizedPatient(patid=ser['patid'],
pat_first_name=ser['first'],
pat_last_name=ser['last'],
pat_birth_date=ser['dob'],
pat_sex=ser['sex'],
pat_race=ser['race'])
# cls.log.info("Patient: {}".format(patient))
if not patient.has_all_data(required_attr):
cls.log.debug("Skip hashing patient [{}] due to missing data"
"for rule [{}]".format(patient.patid, rule))
return ''
raw = pattern.format(patient) + config['SALT']
sha_string = utils.apply_sha256(raw)
cls.log.debug("For patient [{}] (rule {}): {}, hash_string= {}".format(
patient.patid, rule, raw, sha_string)) # noqa
return sha_string
def main():
"""
Configure the logger object and read the command line arguments
for invoking the generator.
.. seealso::
:meth:`HashGenerator.generate`
"""
HashGenerator.configure_logger(logger)
NormalizedPatient.configure_logger(logger)
parser = argparse.ArgumentParser()
parser.add_argument("-v",
"--version",
default=False,
action='store_true',
help="Show the version number")
parser.add_argument("-c",
"--config",
default=DEFAULT_SETTINGS_FILE,
help="Application config file")
parser.add_argument(
'-i',
'--inputdir',
# required=True,
default='.',
help='input directory name')
parser.add_argument(
'-o',
'--outputdir',
# required=True,
default='.',
help='output directory name')
args = parser.parse_args()
if args.version:
import sys
print("deduper, version {}".format(__version__))
sys.exit()
config = Config(root_path=ROOT_PATH, defaults={})
config.from_pyfile(args.config)
start = time.monotonic()
success = HashGenerator.generate(config, args.inputdir, args.outputdir)
end = time.monotonic()
elapsed = (end - start)
if success:
logger.info("Done. Process duration: {}".format(
str(timedelta(seconds=elapsed))))
else:
logger.error("Failed!")
def setUp(self):
super(TestHashGenerator, self).setUp()
HashGenerator.configure_logger(logger)
NormalizedPatient.configure_logger(logger)