def _compare_name_fields(
project,
rdr_dataset,
pii_dataset,
hpo,
concept_id,
pii_field
):
"""
For an hpo, compare all first, middle, and last name fields to omop settings.
This compares a site's name field values found in their uploaded PII
tables with the values in the OMOP observation table.
:param project: project to search for the datasets
:param rdr_dataset: contains datasets from the rdr group
:param pii_dataset: dataset created from submitted hpo sites. the pii tables
:param hpo: string identifier of hpo
:param concept_id: integer value of concept id for concept in the rdr_dataset
:param pii_field: string value of field name with data matching the
concept_id. used to extract the correct values from the pii tables
:return: a match_values dictionary.
"""
match_values = {}
rdr_names = readers.get_rdr_match_values(
project, rdr_dataset, consts.ID_MATCH_TABLE, concept_id
)
pii_names = readers.get_pii_values(
project,
pii_dataset,
hpo,
consts.PII_NAME_TABLE,
pii_field
)
for person_id, pii_name in pii_names:
rdr_name = rdr_names.get(person_id)
if rdr_name is None or pii_name is None:
match_str = consts.MISSING
else:
pii_name = normalizer.normalize_name(pii_name)
rdr_name = normalizer.normalize_name(rdr_name)
match_str = consts.MATCH if rdr_name == pii_name else consts.MISMATCH
match_values[person_id] = match_str
return match_values
def _compare_name_fields(project, rdr_dataset, pii_dataset, hpo, concept_id,
pii_field, pii_tables):
"""
For an hpo, compare all first, middle, and last name fields to omop settings.
This compares a site's name field values found in their uploaded PII
tables with the values in the OMOP observation table.
:param project: project to search for the datasets
:param rdr_dataset: contains datasets from the rdr group
:param pii_dataset: dataset created from submitted hpo sites. the pii tables
:param hpo: string identifier of hpo
:param concept_id: integer value of concept id for concept in the rdr_dataset
:param pii_field: string value of field name with data matching the
concept_id. used to extract the correct values from the pii tables
:return: a match_values dictionary.
"""
match_values = {}
table_name = hpo + consts.PII_NAME_TABLE
if table_name in pii_tables:
rdr_names = readers.get_rdr_match_values(project, rdr_dataset,
consts.ID_MATCH_TABLE,
concept_id)
try:
pii_names = readers.get_pii_values(project, pii_dataset, hpo,
consts.PII_NAME_TABLE,
pii_field)
except (oauth2client.client.HttpAccessTokenRefreshError,
googleapiclient.errors.HttpError):
LOGGER.exception(
f"Unable to read PII for: {hpo}\tdata field:\t{pii_field}")
raise
for person_id, pii_name in pii_names:
rdr_name = rdr_names.get(person_id)
if rdr_name is None or pii_name is None:
match_str = consts.MISSING
else:
pii_name = normalizer.normalize_name(pii_name)
rdr_name = normalizer.normalize_name(rdr_name)
match_str = consts.MATCH if rdr_name == pii_name else consts.MISMATCH
match_values[person_id] = match_str
else:
raise RuntimeError('Table {} doesnt exist.'.format(table_name))
return match_values
def test_normalize_spaced_name(self):
# test
actual = normalizer.normalize_name('Jo Anne')
# post condition
expected = 'joanne'
self.assertEqual(actual, expected)
def test_normalize_hyphenated_name(self):
# test
actual = normalizer.normalize_name('Foo-Bar')
# post condition
expected = 'foobar'
self.assertEqual(actual, expected)
def test_normalize_name_mixed_case(self):
# test
actual = normalizer.normalize_name('O\'Keefe')
# post condition
expected = 'okeefe'
self.assertEqual(actual, expected)
def test_normalize_non_string_name(self):
# test
actual = normalizer.normalize_name(11)
# post-condition
expected = ''
self.assertEqual(actual, expected)
def test_normalize_None_name(self):
# test
actual = normalizer.normalize_name(None)
# post condition
expected = ''
self.assertEqual(actual, expected)
