mmontagna/geonames_countries

#!/usr/bin/env python
import urllib2, json, gzip

geonames_country_url = 'http://download.geonames.org/export/dump/countryInfo.txt'

def find_header(lines):
  #Assume header is the commented line with the > # of tabs.
  lines = filter(lambda line: line.startswith('#'), lines)
  sorted_lines = sorted(lines, key=lambda x: x.split("\t"), reverse=True)
  return sorted_lines[0].upper().strip('#').split("\t")
response = urllib2.urlopen(geonames_country_url)

response_lines = map(lambda x: x.strip(), response.read().split("\n"))

content_lines = filter(lambda x: not x.startswith('#'), response_lines)

num_fields = len(content_lines[0].split("\t"))

packaged_data = {'version' : 0}
packaged_data['header'] = find_header(response_lines)
packaged_data['countries'] = []

for line in content_lines:
  packaged_data['countries'].append(line.split("\t"))

with gzip.open('geonames_countries/geonames_countries.json.gz', 'w') as f:
  json.dump(packaged_data, f)