A quick & dirty script to get taxonomy data from Alchemy API for a list of URL's

get_alchemy_taxonomies.py

#!/usr/bin/env python

import urllib2
import urllib
import logging
import sys
import json

root = logging.getLogger()
root.setLevel(logging.DEBUG)

ch = logging.StreamHandler(sys.stdout)
ch.setLevel(logging.DEBUG)
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
ch.setFormatter(formatter)
root.addHandler(ch)

api_key = '****************************************'

def get_taxonomies(url):
    try:
        baseUrl = "http://access.alchemyapi.com/calls/url/URLGetRankedTaxonomy?"
        urlArgs = {'apikey':api_key, 'outputMode':'json', \
                   'extract': 'taxonomy', 'url': url}
        apiCallUrl = baseUrl + urllib.urlencode(urlArgs)
        response = urllib2.urlopen(apiCallUrl).read()
        if eval(response)['status'] == u'OK':
            logging.info("Reponse OK getting taxonomy for {0}".format(url))
            return(json.dumps(eval(response)))
        else:
            logging.error("Response status {0} getting taxonomy for {1}".format(eval(response)['status'], url))
    except:
        logging.error("Error {0} processing url {1}".format(sys.exc_info()[0], url))

if __name__ == "__main__":
    with open('urls.txt', 'r') as infile, open('alchemy_url_taxonomy.json', 'w') as outfile:
        for line in infile.readlines()[:2]:
            taxonomy_json = get_taxonomies(line.strip())
            if taxonomy_json:
                outfile.write(taxonomy_json + "\n")