def test_record_deduper_dup(self):
record = factories.doctor_factory()
diff_record = factories.doctor_factory()
doctors = [models.Doctor(**diff_record)]
record_deduper = deduper.RecordDeduper(record, doctors,
deduper.DedupManager.DEFAULT_FIELD_WEIGHTINGS,
deduper.DedupManager.DEFAULT_MATCH_THRESHOLD,
deduper.DedupManager.DEFAULT_KILL_THRESHOLD,
MetricsRegistry()
)
new, doctor = record_deduper.run()
self.assertFalse(new)
self.assertEqual(type(doctor), models.Doctor)
self.assertNotEqual(doctor, doctors[0])
def test_dedup_manager(self, MockRecordDeduper):
dedupe_manager = deduper.DedupManager()
record_deduper_instance = MockRecordDeduper.return_value
record_deduper_instance.run.return_value = True, models.Doctor()
record_deduper_instance.run.assert_called_once()
record = factories.doctor_factory()
dedupe_manager.dedup(record)
def test_dup_factory(self):
original = factories.doctor_factory()
dup = factories.dup_doctor_factory(original, {'source': 'new source'})
self.assertEqual(dup['source'], 'new source')
self.assertTrue(dup['is_dup'])
self.assertEqual(dup['last_name'], original['last_name'])