#!/usr/bin/env python2
# -*- coding: utf-8 -*-

import unicodecsv
import lxml.html
import urllib
import re
import datetime
import codecs
import time

DIGITS_ONLY = re.compile(r"[^0-9]")
TODAY = datetime.datetime.today().strftime("%m/%d/%Y")

from selenium import webdriver
from selenium.webdriver.common.keys import Keys

QUERIES = []
RESULTS = []

f = open("data.csv", "rb")
reader = unicodecsv.reader(f, encoding='utf-8')
for query in reader:
    if query[0].strip() != "":
        QUERIES.append(query[0].strip())
        RESULTS.append(query)


driver = webdriver.Firefox()

for index, query in enumerate(QUERIES):
    if index == 0:
        RESULTS[index].append(TODAY)
        continue

    driver.get("https://www.google.com/search?q=%s" % urllib.quote(query))
    source = driver.page_source
    while "CaptchaRedirect" in source:
        print "Enter captcha..."
        source = driver.page_source
        time.sleep(1)
    data = lxml.html.fromstring(source)
    nb_results = data.cssselect("#resultStats")[0].text
    nb_results = DIGITS_ONLY.sub('', nb_results)
    print "%s => %s" % (query, nb_results)
    RESULTS[index].append(nb_results)

driver.quit()


f = codecs.open("data.csv", "wb")
f.write(codecs.BOM_UTF8)
w = unicodecsv.writer(f, encoding='utf-8')
for row in RESULTS:
    w.writerow(row)
f.close()

print "data.csv updated."