#!/usr/bin/env python

"""
Put your key in ~/.calais and give entities.py a URL and it'll use 
semanticproxy to extract names, places, etc from it. 
"""

import rdflib
import sys
import os.path

def find_license():
    try:
        return file(os.path.expanduser('~/.calais')).read().strip()
    except:
        raise Exception("unable to find calais license key in ~/.calais")

g = rdflib.ConjunctiveGraph()

url = sys.argv[1]
key = find_license()
calais = "http://service.semanticproxy.com/processurl/%s/rdf/%s" % (key, url)
name = rdflib.URIRef('http://s.opencalais.com/1/pred/name')

g.load(calais)

e = 'http://s.opencalais.com/1/type/em/e/'
for s, o in g.subject_objects(predicate=rdflib.RDF.type):
    if o.startswith('http://s.opencalais.com/1/type/em/e/'):
        o_type = o.split('/')[-1]
        for o_name in g.objects(s, name):
            print 'a %s named %s' % (o_type, o_name)