def load_predefined_recognizers(self):
# TODO: Change the code to dynamic loading -
# Task #598: Support loading of the pre-defined recognizers
# from the given path.
# Currently this is not integrated into the init method to speed up
# loading time if these are not actually needed (SpaCy for example) is
# time consuming to load
self.recognizers.extend([
CreditCardRecognizer(),
DomainRecognizer(),
EmailRecognizer(),
IbanRecognizer(),
IpRecognizer(),
NhsRecognizer(),
UsBankRecognizer(),
UsLicenseRecognizer(),
UsItinRecognizer(),
UsPassportRecognizer(),
UsPhoneRecognizer(),
UsSsnRecognizer()
])
# Okera addition
if 'PRESIDIO_DISABLE_ML' in os.environ and \
os.environ['PRESIDIO_DISABLE_ML'] == 'true':
logging.info("Disabling ML recognizer.")
else:
logging.info("Enabling ML recognizer.")
self.recognizers.extend([SpacyRecognizer()])
def load_recognizers(self, path):
# TODO: Change the code to dynamic loading -
# Task #598: Support loading of the pre-defined recognizers
# from the given path.
self.recognizers.extend(
[CreditCardRecognizer(),
UsPhoneRecognizer(),
DomainRecognizer()])
def load_predefined_recognizers(self):
# TODO: Change the code to dynamic loading -
# Task #598: Support loading of the pre-defined recognizers
# from the given path.
# Currently this is not integrated into the init method to speed up
# loading time if these are not actually needed (SpaCy for example) is
# time consuming to load
self.recognizers.extend([
CreditCardRecognizer(),
CryptoRecognizer(),
DomainRecognizer(),
EmailRecognizer(),
IbanRecognizer(),
IpRecognizer(),
NhsRecognizer(),
UsBankRecognizer(),
UsLicenseRecognizer(),
UsItinRecognizer(),
UsPassportRecognizer(),
UsPhoneRecognizer(),
UsSsnRecognizer(),
SpacyRecognizer()
])
from unittest import TestCase
from assertions import assert_result
from analyzer.predefined_recognizers import DomainRecognizer
from analyzer.entity_recognizer import EntityRecognizer
domain_recognizer = DomainRecognizer()
entities = ["DOMAIN_NAME"]
class TestDomainRecognizer(TestCase):
def test_invalid_domain(self):
domain = 'microsoft.'
results = domain_recognizer.analyze(domain, entities)
assert len(results) == 0
def test_invalid_domain_with_exact_context(self):
domain = 'microsoft.'
context = 'my domain is '
results = domain_recognizer.analyze(context + domain, entities)
assert len(results) == 0
def test_valid_domain(self):
domain = 'microsoft.com'
results = domain_recognizer.analyze(domain, entities)
assert len(results) == 1
assert_result(results[0], entities[0], 0, 13,
EntityRecognizer.MAX_SCORE)