def test_given_a_unsupported_language_for_supported_entities_then_expect_an_error(
):
language_query_parameter = "language=he"
response_status, response_content = analyzer_supported_entities(
language_query_parameter)
expected_response = """
{"error": "No matching recognizers were found to serve the request."}
"""
assert response_status == 500
assert equal_json_strings(expected_response, response_content)
def test_given_an_illegal_input_for_supported_entities_then_igonre_and_proceed(
):
language_query_parameter = "uknown=input"
response_status, response_content = analyzer_supported_entities(
language_query_parameter)
expected_response = """
["PHONE_NUMBER", "US_DRIVER_LICENSE", "US_PASSPORT", "SG_NRIC_FIN", "LOCATION", "CREDIT_CARD",
"CRYPTO", "UK_NHS", "US_SSN", "US_BANK_NUMBER", "EMAIL_ADDRESS", "DATE_TIME", "IP_ADDRESS",
"PERSON", "IBAN_CODE", "NRP", "US_ITIN", "DOMAIN_NAME"]
"""
assert response_status == 200
assert equal_json_strings(expected_response, response_content)
def test_given_a_correct_analyze_input_with_high_threshold_and_unmatched_entities_then_anonymize_partially(
):
language_query_parameter = "language=en"
response_status, response_content = analyzer_supported_entities(
language_query_parameter)
assert response_status == 200
suppotred_entities = json.loads(response_content)
analyzer_request = {
"text": "John Smith drivers license is AC432223",
"language": "en",
"score_threshold": 0.7,
"entities": suppotred_entities,
}
expected_response = """
[
{"entity_type": "PERSON", "start": 0, "end": 10, "score": 0.85, "analysis_explanation": null}
]
"""
analyzer_data = analyze_and_assert(analyzer_request, expected_response)
anonymizer_request = {
"text": analyzer_request["text"],
"anonymizers": {
"DEFAULT": {
"type": "replace",
"new_value": "ANONYMIZED"
},
"US_DRIVER_LICENSE": {
"type": "mask",
"masking_char": "*",
"chars_to_mask": 4,
"from_end": True,
},
"PERSON": {
"type": "replace",
"new_value": "<PERSON>"
},
},
"analyzer_results": analyzer_data,
}
expected_response = """{"text": "<PERSON> drivers license is AC432223", "items": [{"operator": "replace", "entity_type": "PERSON", "start": 0, "end": 8, "text": "<PERSON>"}]}"""
anonymize_and_assert(anonymizer_request, expected_response)