#!/usr/bin/python

import urllib
import simplejson
import sys
import os

APIKEY = '### YOU KEY HERE ###'

def get_suggest(text):
  gateway = 'http://api.zemanta.com/services/rest/0.0/'
  args = {'method': 'zemanta.suggest',
          'api_key': APIKEY,
          'text': text,
          'format': 'json'}
  
  args_enc = urllib.urlencode(args)
  s = urllib.urlopen(gateway, args_enc).read()
  
  s = simplejson.loads(s)
  return s
  
def main():
  #f = open('textfile.txt', 'r')
  outname = os.path.splitext(os.path.basename(sys.argv[1]))[0]
  f = open(sys.argv[1])
  output_combined = open(outname+'-combined.txt', 'w')
  output_tags = open(outname+'-tags.txt', 'w')
  output_links = open(outname+'-links.txt', 'w')
  s = f.read()
  
  spl = s.split()
  
  txt = ""
  c = 0
  
  for i in spl:
    txt += i
    c += 1
    if (c % 340) == 0:
      suggest = get_suggest(txt)
      keywords= suggest['keywords']
      links = suggest['markup']['links']
      
      for k in keywords:
        output = k["name"].replace(" ", "~") + '\n'        
        try:
          output_combined.write(output)
          output_tags.write(output)
        except:
          pass
        
        
      for link in links:
        output = link['anchor'].replace(" ", "~") + '\n'
        try:
          output_combined.write(output)s
          output_links.write(output)
        except:
          pass
        
        
      txt = ""
      print c
      
      #breaking at 30k words for this experiment
      if c > 30000:
        break
  
if __name__ == '__main__':
  main()