zipscript.py

import xlrd
import time
from simplegeo import Client, json, APIError
#from multiprocessing.pool import ThreadPool as Pool
from multiprocessing import Pool

def f(z):
    z = str(z)  
    neighborhoods = []
    error = ""
    try:
        # z = str(10001)
        results = client.context.get_context_by_address(z)
        for b in results["features"]:
            if b["classifiers"][0]["category"] == 'Neighborhood':
                n = b["name"]
                #print n
                neighborhoods.append(n)
    except APIError, e:
        return (z, error)

    print "%s hoods=%s." % (z, len(neighborhoods))

    return (z, neighborhoods)

if __name__ == '__main__':  
    row = 1
    neighborhoods = []
    zipcodes = set()

    book = xlrd.open_workbook("zipcodes.xls")
    sheet = book.sheet_by_index(0)
    rows = sheet.nrows
    print rows
    client = Client('SbAc87cM7EK2qCvmjymQtXU66MdXGNPr','CNmJaCEGGY3V7Tq3NEJsvNJCdJZjvgfN')

    while row <= 65535:
        current_row = sheet.row(row)[0]
        zip = sheet.cell_value(row,0)   
        zip = int(zip)
        row += 1

        z = str(zip)
        z = z.zfill(5)
        zipcodes.add(z)
            
    print "Found %s codes" % len(zipcodes)
    
    zipcodes = set(list(zipcodes))
    print zipcodes

    start = time.time()
    pool = Pool(16))
    result = pool.map(f, zipcodes)
    from pprint import pprint
    pprint(result)
    print 'Duration %s' % (time.time() - start, )