def analyze_perf(field_name, entity, runtime_threshold_micros):
fieldType = common_pb2.FieldTypes()
fieldType.name = field_name
types = [fieldType]
start_time = datetime.datetime.now()
results = match.analyze_text(context + entity, types)
analyze_time = datetime.datetime.now() - start_time
print('--- analyze_time[{}]: {}.{} seconds'.format(
types[0].name, analyze_time.seconds, analyze_time.microseconds))
assert analyze_time.seconds < 1
assert analyze_time.microseconds < runtime_threshold_micros
from analyzer import matcher, common_pb2
from tests import *
fieldType = common_pb2.FieldTypes()
fieldType.name = common_pb2.FieldTypesEnum.Name(common_pb2.US_SSN)
types = [fieldType]
def test_valid_us_ssn_very_weak_match():
num1 = '078-051120'
num2 = '07805-1120'
results = match.analyze_text('{} {}'.format(num1, num2), types)
assert len(results) == 2
assert results[0].text == num1
assert results[0].score > 0.01 and results[0].score < 0.31
assert results[1].text == num2
assert results[1].score > 0.01 and results[0].score < 0.31
def test_valid_us_ssn_weak_match():
num = '078051120'
results = match.analyze_text(num, types)
assert len(results) == 1
assert results[0].text == num
assert results[0].score > 0.29 and results[0].score < 0.41
def test_valid_us_ssn_medium_match():
num = '078-05-1120'