def test_fuzzy_match(example_IRS_service_data,
example_IRS_search_object_with_spelled_out_saint,
mock_config_object):
client = get_mongo_client()
if 'pytest_fuzzy_test' in client.list_collection_names():
client.drop_collection('pytest_fuzzy_test')
client.create_collection('pytest_fuzzy_test')
insert_services(example_IRS_service_data, client, 'pytest_fuzzy_test')
refresh_ngrams(client, 'pytest_fuzzy_test')
name = example_IRS_search_object_with_spelled_out_saint['name']
zip_code = example_IRS_search_object_with_spelled_out_saint['zip']
dc = locate_potential_duplicate(name, zip_code, client,
'pytest_fuzzy_test')
client.drop_collection('pytest_fuzzy_test')
assert dc == 'ST FERIOLE ISLAND PARK'
scraper = SummerMealSitesScraper(
source=data_source_name,
data_url=data_url,
data_page_url=data_url,
data_format="DF",
extract_usecols=None,
drop_duplicates_columns=[
'siteName', 'siteAddress', 'siteZip', 'siteCity', 'siteState'
],
rename_columns={
'siteName': 'name',
'siteStatus': 'notes',
'siteAddress': 'address1',
'siteCity': 'city',
'siteState': 'state',
'siteZip': 'zip',
'sitePhone': 'phone',
'Country': 'country'
},
service_summary="Food Bank",
check_collection="services",
dump_collection="tmpSummerMealSites",
dupe_collection="tmpSummerMealSitesDuplicates",
data_source_collection_name=data_source_name,
collection_dupe_field='name')
if __name__ == '__main__':
client = get_mongo_client()
scraper.main_scraper(client)