geocode.py

   1 #!/usr/bin/env python3
   2
   3 # © Copyright 2021-2022, Scott Gasch
   4
   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
   8 import logging
   9 import re
  10 from typing import Dict, Optional
  11
  12 import requests
  13 from bs4 import BeautifulSoup
  14 from requests.utils import requote_uri
  15
  16 import string_utils
  17
  18 logger = logging.getLogger(__name__)
  19
  20
  21 def geocode_address(address: str) -> Optional[Dict[str, str]]:
  22     """Send a single address to the US Census geocoding API.
  23
  24     >>> out = geocode_address('4600 Silver Hill Rd,, 20233')
  25     >>> out['Matched Address']
  26     '4600 SILVER HILL RD, WASHINGTON, DC, 20233'
  27     >>> out['Interpolated Longitude (X) Coordinates']
  28     -76.92743
  29     >>> out['Interpolated Latitude (Y) Coordinates']
  30     38.84599
  31
  32     """
  33     url = 'https://geocoding.geo.census.gov/geocoder/geographies/onelineaddress'
  34     url += f'?address={address}'
  35     url += '&layers=all&benchmark=4&vintage=4'
  36     url = requote_uri(url)
  37     logger.debug('GET: %s', url)
  38     r = requests.get(url)
  39     if r.status_code != 200:
  40         logger.error('Unexpected response code %d, wanted 200.  Fail.', r.status_code)
  41         return None
  42     else:
  43         soup = BeautifulSoup(r.text, 'html.parser')
  44         result = soup.find('div', id='pl_gov_census_geo_geocoder_domain_AddressResult')
  45         logger.debug('Unhelpful result blurb: "%s"', result)
  46         output = result.get_text('\n')
  47         label = None
  48         out = {}
  49         for line in output.split('\n'):
  50             if re.match(r'.*: *$', line):
  51                 line = line.strip()
  52                 label = line[:-1]
  53                 logger.debug('Label is: "%s"', label)
  54             else:
  55                 if label:
  56                     value = line.strip()
  57                     if string_utils.is_integer_number(value):
  58                         value = int(value)
  59                     elif string_utils.is_number(value):
  60                         value = float(value)
  61                     logger.debug('Value is: "%s"', value)
  62                     out[label] = value
  63     return out
  64
  65
  66 if __name__ == '__main__':
  67     import doctest
  68
  69     doctest.testmod()