# -*- coding:cp1251
import sys
import thread
import pycurl
import StringIO
import re
import urllib
import pickle
import simplejson

fi = open("yande.txt", 'r')
items = simplejson.load(fi)
fi.close()
itemss = list(items.keys())

def some_function():
    global items, count_thread, d
    
    while itemss:
        key = itemss.pop()
        params = urllib.urlencode({'text': key.encode('utf8')})
        
        url = "http://wordstat.yandex.ru/?%s" % params

        data = StringIO.StringIO()
        curl = pycurl.Curl()
        curl.setopt(pycurl.FOLLOWLOCATION, 0)
        curl.setopt(pycurl.CONNECTTIMEOUT, 30)
        curl.setopt(pycurl.URL, url)
        curl.setopt(pycurl.WRITEFUNCTION, data.write)

        try:
            curl.perform()
        except:
            pass
        curl.close()

        rez = data.getvalue()
        pattern = re.compile(r'<td>\s+<a href="\?page[^"]+">(.*?)</a>\s+</td>\s+<td align="right">([\d.]+)</td>', re.S)
        group = pattern.findall(rez)

        d = dict((a, b) for a, b in group if int(b) > 300)

    count_thread -= 1

count_thread = 0

for i in xrange(0,100):
    count_thread += 1
    thread.start_new_thread(some_function,())

while(count_thread>0):
    pass

f = open("yandex-result.txt", 'a+')
f.write(simplejson.dumps(d))
f.close()