Sedona Public Library's Brand New CSV County Getter

sed_counties.py

#! /usr/bin/python2                                                                                                                                           \
                                                                                                                                                               

''' specify a csv file as a command line argument.                                                                                                             
                                                                                                                                                               
creates a new csv with addresses replaced with counties.                                                                                                       
                                                                                                                                                               
skips users who throw errors and outputs their details to ./errors                                                                                             
'''

import sys
import re
import urllib, urllib2
import cookielib
from os.path import isfile
from csv import DictReader, DictWriter

addresses = DictReader(open(sys.argv[1]), ['patron barcode', 'patron address'])
outfile = DictWriter(open('./counties.csv', 'w'), ['patron barcode', 'patron county'])
errorfile = open('./errors', 'w')

# we need to handle cookies                                                                                                                                    
cookieFile = './cookies.lwp'
cookieJar = cookielib.LWPCookieJar()
if isfile(cookieFile):
    cookieJar.load(cookieFile)

opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookieJar))
urllib2.install_opener(opener)

def melissa(address):
    queryurl = "http://www.melissadata.com/lookups/AddressCheck.asp?" + urllib.urlencode({"InData": address})

    html = urllib2.urlopen(queryurl).read()

    if "Address Verified" in html:
        cindex = html.index('County')
        return re.search('<b>(.*)</b>', html[cindex:cindex + 50]).group(1).capitalize() # not the prettiest, but the html we're getting isn't marked up very well.                                                                                                                                                           
    elif "Cookies must be enabled" in html:
        raise Exception("Melissa wants to eat your cookies.")
    else:
        raise Exception("Melissa returned some kind of error.")

for address in addresses:
    try:
        county = {'patron barcode': address['patron barcode'],
              'patron county': melissa(address['patron address'])}

        outfile.writerow(county)
    except Exception as e:
        errorfile.write(str(e) +'\n')
        errorfile.write(str(address) + '\n')
        errorfile.write('\n')

cookieJar.save(cookieFile)
errorfile.close()