selenium snapshots

gistfile1.py

#!/usr/bin/env python2.7

from xml.dom import minidom
import urllib2
import os
import shutil
import random
import re
import sys

num_snapshots = 100

if len(sys.argv) > 1:
    num_snapshots = int(sys.argv[1])

out_dir = "selenium_snapshots"
host = "http://localhost.com:8080"
req = urllib2.urlopen(host + "/sitemap.xml")
dom = minidom.parse(req)

print "Taking %d snapshots from %s" % (num_snapshots, host)

shutil.rmtree(out_dir, True)
os.makedirs(out_dir)

def get_text(node):
    return node.childNodes[0].data

def url_filename(url):
    return re.sub(r'[^\w-]+', "_", url)

urls = [get_text(node) for node in dom.getElementsByTagName("loc")]
random.shuffle(urls)
urls = urls[:num_snapshots]

from selenium import webdriver
browser = webdriver.Firefox()

for url in urls:
    print "Running " + url
    browser.get(url)
    browser.save_screenshot(os.path.join(out_dir, url_filename(url)) + ".png")

browser.quit()