def test_run_export_with_target_bucket(self):
# validation/main.py INTEGRATION TEST
folder_prefix = 'dummy-prefix-2018-03-24/'
bucket_nyc = gcs_utils.get_hpo_bucket('nyc')
test_util.get_synpuf_results_files()
test_util.populate_achilles(self.hpo_bucket, hpo_id=None)
main.run_export(folder_prefix=folder_prefix, target_bucket=bucket_nyc)
bucket_objects = gcs_utils.list_bucket(bucket_nyc)
actual_object_names = [obj['name'] for obj in bucket_objects]
for report in common.ALL_REPORT_FILES:
expected_object_name = folder_prefix + common.ACHILLES_EXPORT_PREFIX_STRING + 'default' + '/' + report
self.assertIn(expected_object_name, actual_object_names)
datasources_json_path = folder_prefix + common.ACHILLES_EXPORT_DATASOURCES_JSON
self.assertIn(datasources_json_path, actual_object_names)
datasources_json = gcs_utils.get_object(bucket_nyc,
datasources_json_path)
datasources_actual = json.loads(datasources_json)
datasources_expected = {
'datasources': [{
'name': 'default',
'folder': 'default',
'cdmVersion': 5
}]
}
self.assertDictEqual(datasources_expected, datasources_actual)
def setUpClass(cls):
print(
'\n**************************************************************')
print(cls.__name__)
print('**************************************************************')
dataset_id = bq_utils.get_dataset_id()
test_util.delete_all_tables(dataset_id)
test_util.populate_achilles()
def setUpClass(cls):
print(
'\n**************************************************************')
print(cls.__name__)
print('**************************************************************')
fake_bucket = gcs_utils.get_hpo_bucket(test_util.FAKE_HPO_ID)
dataset_id = bq_utils.get_dataset_id()
test_util.delete_all_tables(dataset_id)
test_util.get_synpuf_results_files()
test_util.populate_achilles(fake_bucket)
def test_heel_analyses(self, mock_hpo_bucket):
# Long-running test
mock_hpo_bucket.return_value = self.get_mock_hpo_bucket()
# create randomized tables to bypass BQ rate limits
random_string = str(randint(10000, 99999))
randomized_hpo_id = FAKE_HPO_ID + '_' + random_string
# prepare
self._load_dataset(randomized_hpo_id)
test_util.populate_achilles(hpo_id=randomized_hpo_id,
include_heel=False)
# define tables
achilles_heel_results = randomized_hpo_id + '_' + achilles_heel.ACHILLES_HEEL_RESULTS
achilles_results_derived = randomized_hpo_id + '_' + achilles_heel.ACHILLES_RESULTS_DERIVED
# run achilles heel
achilles_heel.create_tables(randomized_hpo_id, True)
achilles_heel.run_heel(hpo_id=randomized_hpo_id)
# validate
query = sql_wrangle.qualify_tables(
'SELECT COUNT(1) as num_rows FROM %s' % achilles_heel_results)
response = bq_utils.query(query)
rows = bq_utils.response2rows(response)
self.assertEqual(ACHILLES_HEEL_RESULTS_COUNT, rows[0]['num_rows'])
query = sql_wrangle.qualify_tables(
'SELECT COUNT(1) as num_rows FROM %s' % achilles_results_derived)
response = bq_utils.query(query)
rows = bq_utils.response2rows(response)
self.assertEqual(ACHILLES_RESULTS_DERIVED_COUNT, rows[0]['num_rows'])
# test new heel re-categorization
errors = [
2, 4, 5, 101, 200, 206, 207, 209, 400, 405, 406, 409, 411, 413,
500, 505, 506, 509, 600, 605, 606, 609, 613, 700, 705, 706, 709,
711, 713, 715, 716, 717, 800, 805, 806, 809, 813, 814, 906, 1006,
1609, 1805
]
query = sql_wrangle.qualify_tables(
"""SELECT analysis_id FROM {table_id}
WHERE achilles_heel_warning LIKE 'ERROR:%'
GROUP BY analysis_id""".format(table_id=achilles_heel_results))
response = bq_utils.query(query)
rows = bq_utils.response2rows(response)
actual_result = [row["analysis_id"] for row in rows]
for analysis_id in actual_result:
self.assertIn(analysis_id, errors)
warnings = [
4, 5, 7, 8, 9, 200, 210, 302, 400, 402, 412, 420, 500, 511, 512,
513, 514, 515, 602, 612, 620, 702, 712, 720, 802, 812, 820
]
query = sql_wrangle.qualify_tables(
"""SELECT analysis_id FROM {table_id}
WHERE achilles_heel_warning LIKE 'WARNING:%'
GROUP BY analysis_id""".format(table_id=achilles_heel_results))
response = bq_utils.query(query)
rows = bq_utils.response2rows(response)
actual_result = [row["analysis_id"] for row in rows]
for analysis_id in actual_result:
self.assertIn(analysis_id, warnings)
notifications = [
101, 103, 105, 114, 115, 118, 208, 301, 410, 610, 710, 810, 900,
907, 1000, 1800, 1807
]
query = sql_wrangle.qualify_tables(
"""SELECT analysis_id FROM {table_id}
WHERE achilles_heel_warning LIKE 'NOTIFICATION:%' and analysis_id is not null
GROUP BY analysis_id""".format(table_id=achilles_heel_results))
response = bq_utils.query(query)
rows = bq_utils.response2rows(response)
actual_result = [row["analysis_id"] for row in rows]
for analysis_id in actual_result:
self.assertIn(analysis_id, notifications)