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
12 from typing import Any, Dict, List, Optional
15 from requests.utils import requote_uri
19 logger = logging.getLogger(__name__)
22 def geocode_address(address: str) -> Optional[Dict[str, Any]]:
23 """Send a single address to the US Census geocoding API. The response
24 is a parsed JSON chunk of data with N addressMatches in the result
25 section and the details of each match within it. Returns None on error.
27 >>> json = geocode_address('4600 Silver Hill Rd,, 20233')
28 >>> json['result']['addressMatches'][0]['matchedAddress']
29 '4600 SILVER HILL RD, WASHINGTON, DC, 20233'
31 >>> json['result']['addressMatches'][0]['coordinates']
32 {'x': -76.92743, 'y': 38.84599}
35 url = 'https://geocoding.geo.census.gov/geocoder/geographies/onelineaddress'
36 url += f'?address={address}'
37 url += '&returntype=geographies&layers=all&benchmark=4&vintage=4&format=json'
38 url = requote_uri(url)
39 logger.debug('GET: %s', url)
42 except Exception as e:
46 if r.status_code != 200:
48 logger.error('Unexpected response code %d, wanted 200. Fail.', r.status_code)
50 logger.debug('Response: %s', json.dumps(r.json(), indent=4, sort_keys=True))
54 def batch_geocode_addresses(addresses: List[str]):
55 """Send up to addresses for batch geocoding. Each line of the input
56 list should be a single address of the form: STREET ADDRESS, CITY,
57 STATE, ZIP. Components may be omitted but the commas may not be.
58 Result is an array of the same size as the input array with one
59 answer record per line. Returns None on error.
61 This code will deal with requests >10k addresses by chunking them
62 internally because the census website disallows requests > 10k lines.
64 >>> batch_geocode_addresses(
66 ... '4600 Silver Hill Rd, Washington, DC, 20233',
67 ... '935 Pennsylvania Avenue, NW, Washington, DC, 20535-0001',
68 ... '1600 Pennsylvania Avenue NW, Washington, DC, 20500',
69 ... '700 Pennsylvania Avenue NW, Washington, DC, 20408',
72 ['"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"']
76 url = 'https://geocoding.geo.census.gov/geocoder/geographies/addressbatch'
77 payload = {'benchmark': '4', 'vintage': '4'}
79 for chunk in list_utils.shard(addresses, 9999):
82 raw_file += f'{n}, {address}\n'
84 files = {'addressFile': ('input.csv', raw_file)}
85 logger.debug('POST: %s', url)
87 r = requests.post(url, files=files, data=payload)
88 except Exception as e:
92 if r.status_code != 200:
94 logger.error('Unexpected response code %d, wanted 200. Fail.', r.status_code)
96 logger.debug('Response: %s', r.text)
97 for line in r.text.split('\n'):
104 if __name__ == '__main__':