Fetch NOAA weather stations for IL from web.

get_noaa_stations.py

#!/bin/env python

import csv
import re
from urllib2 import urlopen

from BeautifulSoup import BeautifulSoup, BeautifulStoneSoup

NOAA_ROOT_URL = 'http://www.weather.gov/'

list_page = urlopen('http://www.weather.gov/xml/current_obs/seek.php?state=il')
list_page_soup = BeautifulSoup(list_page.read())

station_links = list_page_soup.findAll('a', href=re.compile('^/data/current_obs/.*\.xml$'))
station_urls = [l['href'] for l in station_links]
station_urls = list(set(station_urls))	# Remove duplicates

station_data = []

for url in station_urls:
    observation_xml = urlopen(NOAA_ROOT_URL + url)
    observation_soup = BeautifulStoneSoup(observation_xml.read())
	
    station_id = observation_soup.current_observation.station_id.string
    name = observation_soup.current_observation.location.string
    latitude = observation_soup.current_observation.latitude.string
    longitude = observation_soup.current_observation.longitude.string
	
    station_data.append((station_id, name, '%s %s' % (latitude, longitude), 'NOAA'))
    
    print 'Got ', station_id
	
output = csv.writer(open("noaa_station_data.csv", "wb"))
output.writerows([('station_id', 'name', 'location', 'source')])
output.writerows(station_data)