def api_connection(n):
'''
Connect to API and download datafile and attachments
Parameters:
n(int): a whole number
Returns:
DataFrame: Pandas Dataframe with downloaded data from API
Outputs:
attachment: attachments from API
'''
from sodapy import Socrata
client = Socrata("data.cms.gov", None)
results = client.get('efgi-jnkv',
select='hcpcs_cd,'
'psps_submitted_charge_amt,'
'hcpcs_initial_modifier_cd,'
'hcpcs_second_modifier_cd,'
'carrier_num,'
'pricing_locality_cd,'
'type_of_service_cd,'
'place_of_service_cd,'
'provider_spec_cd,'
'psps_submitted_service_cnt,'
'psps_denied_services_cnt,'
'hcpcs_betos_cd',
where='psps_submitted_charge_amt > 5',
limit=n)
results_sample = pd.DataFrame.from_records(results)
#download attachments
client.download_attachments(
"efgi-jnkv",
download_dir='C:/Users/amcgrat/Desktop/UCD PROGRAM/Project/HPCPS')
client.close()
return results_sample
def test_get_metadata_and_attachments():
mock_adapter = {}
mock_adapter["prefix"] = PREFIX
adapter = requests_mock.Adapter()
mock_adapter["adapter"] = adapter
client = Socrata(DOMAIN, APPTOKEN, session_adapter=mock_adapter)
response_data = "get_song_metadata.txt"
setup_old_api_mock(adapter, "GET", response_data, 200)
response = client.get_metadata(DATASET_IDENTIFIER)
assert isinstance(response, dict)
assert "newBackend" in response
assert "attachments" in response["metadata"]
response = client.download_attachments(DATASET_IDENTIFIER)
assert isinstance(response, list)
assert len(response) == 0
client.close()
'psps_submitted_charge_amt,'
'hcpcs_initial_modifier_cd,'
'hcpcs_second_modifier_cd,'
'carrier_num,'
'pricing_locality_cd,'
'type_of_service_cd,'
'place_of_service_cd,'
'provider_spec_cd,'
'psps_submitted_service_cnt,'
'psps_denied_services_cnt,'
'psps_hcpcs_asc_ind_cd,'
'hcpcs_betos_cd',
where='psps_submitted_charge_amt > 5',
limit=3400000)
results_df_sample = pd.DataFrame.from_records(results)
client.download_attachments("efgi-jnkv", download_dir="~/Desktop")
client.close()
results_df_sample = results_df_sample.sample(frac=0.25, random_state=123)
results_df_sample['hcpcs_cd'] = results_df_sample.hcpcs_cd.str.pad(
5, side='left', fillchar='0')
#join dataframe to code categories dataframe.
results_df_sample = pd.merge(results_df_sample,
hcpcs_cs_cat,
on=['hcpcs_cd'],
how='left')
#Replace nulls in denied services count with zero.
results_df_sample['psps_denied_services_cnt'].replace(np.NaN,
