3 # © Copyright 2022, Scott Gasch
5 """Wrapper around US Census address geocoder API described here:
6 https://www2.census.gov/geo/pdfs/maps-data/data/Census_Geocoder_User_Guide.pdf
7 https://geocoding.geo.census.gov/geocoder/Geocoding_Services_API.pdf
11 from typing import Any, Dict, List, Optional
14 from requests.utils import requote_uri
18 logger = logging.getLogger(__name__)
21 def geocode_address(address: str) -> Optional[Dict[str, Any]]:
22 """Send a single address to the US Census geocoding API. The response
23 is a parsed JSON chunk of data with N addressMatches in the result
24 section and the details of each match within it. Returns None on error.
26 >>> json = geocode_address('4600 Silver Hill Rd,, 20233')
27 >>> json['result']['addressMatches'][0]['matchedAddress']
28 '4600 SILVER HILL RD, WASHINGTON, DC, 20233'
30 >>> json['result']['addressMatches'][0]['coordinates']
31 {'x': -76.92743, 'y': 38.84599}
34 url = 'https://geocoding.geo.census.gov/geocoder/geographies/onelineaddress'
35 url += f'?address={address}'
36 url += '&returntype=geographies&layers=all&benchmark=4&vintage=4&format=json'
37 url = requote_uri(url)
38 logger.debug('GET: %s', url)
41 except Exception as e:
45 if r.status_code != 200:
46 logger.error('Unexpected response code %d, wanted 200. Fail.', r.status_code)
48 # print(json.dumps(r.json(), indent=4, sort_keys=True))
52 def batch_geocode_addresses(addresses: List[str]):
53 """Send up to addresses for batch geocoding. Each line of the input
54 list should be a single address of the form: STREET ADDRESS, CITY,
55 STATE, ZIP. Components may be omitted but the commas may not be.
56 Result is an array of the same size as the input array with one
57 answer record per line. Returns None on error.
59 This code will deal with requests >10k addresses by chunking them
60 internally because the census website disallows requests > 10k lines.
62 >>> batch_geocode_addresses(
64 ... '4600 Silver Hill Rd, Washington, DC, 20233',
65 ... '935 Pennsylvania Avenue, NW, Washington, DC, 20535-0001',
66 ... '1600 Pennsylvania Avenue NW, Washington, DC, 20500',
67 ... '700 Pennsylvania Avenue NW, Washington, DC, 20408',
70 ['"1"," 4600 Silver Hill Rd, Washington, DC, 20233","Match","Exact","4600 SILVER HILL RD, WASHINGTON, DC, 20233","-76.92743,38.84599","76355984","L","24","033","802405","2004"', '"2"," 935 Pennsylvania Avenue, NW, Washington, DC","No_Match"', '"3"," 1600 Pennsylvania Avenue NW, Washington, DC, 20500","Match","Exact","1600 PENNSYLVANIA AVE NW, WASHINGTON, DC, 20500","-77.03534,38.898754","76225813","L","11","001","980000","1034"', '"4"," 700 Pennsylvania Avenue NW, Washington, DC, 20408","Match","Exact","700 PENNSYLVANIA AVE NW, WASHINGTON, DC, 20408","-77.02304,38.89362","76226346","L","11","001","980000","1025"']
74 url = 'https://geocoding.geo.census.gov/geocoder/geographies/addressbatch'
75 payload = {'benchmark': '4', 'vintage': '4'}
77 for chunk in list_utils.shard(addresses, 9999):
80 raw_file += f'{n}, {address}\n'
82 files = {'addressFile': ('input.csv', raw_file)}
83 logger.debug('POST: %s', url)
85 r = requests.post(url, files=files, data=payload)
86 except Exception as e:
89 if r.status_code != 200:
91 logger.error('Unexpected response code %d, wanted 200. Fail.', r.status_code)
93 for line in r.text.split('\n'):
100 if __name__ == '__main__':