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
16 logger = logging.getLogger(__name__)
19 def geocode_address(address: str) -> Optional[Dict[str, Any]]:
20 """Send a single address to the US Census geocoding API. The response
21 is a parsed JSON chunk of data with N addressMatches in the result
22 section and the details of each match within it. Returns None on error.
24 >>> json = geocode_address('4600 Silver Hill Rd,, 20233')
25 >>> json['result']['addressMatches'][0]['matchedAddress']
26 '4600 SILVER HILL RD, WASHINGTON, DC, 20233'
28 >>> json['result']['addressMatches'][0]['coordinates']
29 {'x': -76.92743, 'y': 38.84599}
32 url = 'https://geocoding.geo.census.gov/geocoder/geographies/onelineaddress'
33 url += f'?address={address}'
34 url += '&returntype=geographies&layers=all&benchmark=4&vintage=4&format=json'
35 url = requote_uri(url)
36 logger.debug('GET: %s', url)
39 except Exception as e:
43 if r.status_code != 200:
44 logger.error('Unexpected response code %d, wanted 200. Fail.', r.status_code)
46 # print(json.dumps(r.json(), indent=4, sort_keys=True))
50 def batch_geocode_addresses(addresses: List[str]):
51 """Send up to addresses for batch geocoding. Each line of the input
52 list should be a single address of the form: STREET ADDRESS, CITY,
53 STATE, ZIP. Components may be omitted but the commas may not be.
54 Result is an array of the same size as the input array with one
55 answer record per line. Returns None on error.
57 >>> batch_geocode_addresses(
59 ... '4600 Silver Hill Rd, Washington, DC, 20233',
60 ... '935 Pennsylvania Avenue, NW, Washington, DC, 20535-0001',
61 ... '1600 Pennsylvania Avenue NW, Washington, DC, 20500',
62 ... '700 Pennsylvania Avenue NW, Washington, DC, 20408',
65 ['"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"']
68 # TODO: use list_utils.shard to break up the input if it's >10k records
69 # b/c the census website has a hard limit at 10k.
71 url = 'https://geocoding.geo.census.gov/geocoder/geographies/addressbatch'
72 payload = {'benchmark': '4', 'vintage': '4'}
74 for n, address in enumerate(addresses):
75 raw_file += f'{n+1}, {address}\n'
76 files = {'addressFile': ('input.csv', raw_file)}
77 logger.debug('POST: %s', url)
79 r = requests.post(url, files=files, data=payload)
80 except Exception as e:
83 if r.status_code != 200:
85 logger.error('Unexpected response code %d, wanted 200. Fail.', r.status_code)
89 for line in r.text.split('\n'):
96 if __name__ == '__main__':