2014-01-21 06:41:46 +00:00
|
|
|
#!/usr/bin/env python
|
|
|
|
|
from json import dumps, loads
|
2014-01-21 22:53:28 +00:00
|
|
|
import os,sys,argparse, urllib.request, urllib.error, urllib.parse
|
2014-01-21 06:41:46 +00:00
|
|
|
import websocket
|
|
|
|
|
import time
|
2014-01-21 21:32:33 +00:00
|
|
|
import uuid
|
|
|
|
|
import logging
|
|
|
|
|
import threading
|
|
|
|
|
logging.basicConfig(level=logging.DEBUG)
|
2014-01-21 06:41:46 +00:00
|
|
|
|
2014-01-21 21:32:33 +00:00
|
|
|
class Umbra:
|
|
|
|
|
def __init__(self, port):
|
|
|
|
|
self.cmd_id = 0
|
|
|
|
|
self.chrome_debug_port = port
|
|
|
|
|
self.launch_tab_socket = self.get_websocket(self.on_open)
|
|
|
|
|
self.launch_tab_socket.run_forever()
|
2014-01-21 06:41:46 +00:00
|
|
|
|
2014-01-21 21:32:33 +00:00
|
|
|
def get_websocket(self, on_open, url=None):
|
2014-01-21 22:53:28 +00:00
|
|
|
debug_info = loads(urllib.request.urlopen("http://localhost:%s/json" % self.chrome_debug_port).read())
|
2014-01-21 21:32:33 +00:00
|
|
|
if url: #Polling for the data url we used to initialize the window
|
2014-01-21 22:53:28 +00:00
|
|
|
while not [x for x in debug_info if x['url'] == url]:
|
|
|
|
|
debug_info = loads(urllib.request.urlopen("http://localhost:%s/json" % self.chrome_debug_port).read())
|
2014-01-21 21:32:33 +00:00
|
|
|
time.sleep(0.5)
|
2014-01-21 22:53:28 +00:00
|
|
|
debug_info = [x for x in debug_info if x['url'] == url]
|
2014-01-21 21:32:33 +00:00
|
|
|
return_socket = websocket.WebSocketApp(debug_info[0]['webSocketDebuggerUrl'], on_message = self.on_message)
|
|
|
|
|
return_socket.on_open = on_open
|
|
|
|
|
return return_socket
|
|
|
|
|
|
|
|
|
|
def on_message(self, ws, message):
|
|
|
|
|
message = loads(message)
|
2014-01-21 22:53:28 +00:00
|
|
|
if "method" in list(message.keys()) and message["method"] == "Network.requestWillBeSent":
|
|
|
|
|
print(message)
|
2014-01-21 21:32:33 +00:00
|
|
|
|
|
|
|
|
def on_open(self, ws):
|
|
|
|
|
self.fetch_url("http://archive.org")
|
2014-01-21 22:47:49 +00:00
|
|
|
self.fetch_url("http://facebook.com")
|
|
|
|
|
self.fetch_url("http://flickr.com")
|
2014-01-21 21:32:33 +00:00
|
|
|
time.sleep(10)
|
|
|
|
|
sys.exit(0)
|
|
|
|
|
|
|
|
|
|
def send_command(self,tab=None, **kwargs):
|
|
|
|
|
if not tab:
|
|
|
|
|
tab = self.launch_tab_socket
|
|
|
|
|
command = {}
|
|
|
|
|
command.update(kwargs)
|
|
|
|
|
self.cmd_id += 1
|
|
|
|
|
command['id'] = self.cmd_id
|
|
|
|
|
tab.send(dumps(command))
|
|
|
|
|
|
|
|
|
|
def fetch_url(self, url):
|
|
|
|
|
new_page = 'data:text/html;charset=utf-8,<html><body>%s</body></html>' % str(uuid.uuid4())
|
|
|
|
|
self.send_command(method="Runtime.evaluate", params={"expression":"window.open('%s');" % new_page})
|
|
|
|
|
def on_open(ws):
|
|
|
|
|
self.send_command(tab=ws, method="Network.enable")
|
|
|
|
|
self.send_command(tab=ws, method="Runtime.evaluate", params={"expression":"document.location = '%s';" % url})
|
|
|
|
|
socket = self.get_websocket(on_open, new_page)
|
|
|
|
|
threading.Thread(target=socket.run_forever).start()
|
2014-01-21 06:41:46 +00:00
|
|
|
|
|
|
|
|
class Chrome():
|
2014-01-21 22:47:49 +00:00
|
|
|
def __init__(self, port, executable, browser_wait):
|
2014-01-21 06:41:46 +00:00
|
|
|
self.port = port
|
2014-01-21 22:47:49 +00:00
|
|
|
self.executable = executable
|
|
|
|
|
self.browser_wait=browser_wait
|
2014-01-21 06:41:46 +00:00
|
|
|
|
|
|
|
|
def __enter__(self):
|
|
|
|
|
import psutil, subprocess
|
2014-01-21 22:47:49 +00:00
|
|
|
self.chrome_process = subprocess.Popen([self.executable, "--temp-profile", "--remote-debugging-port=%s" % self.port])
|
2014-01-21 06:41:46 +00:00
|
|
|
start = time.time()
|
|
|
|
|
open_debug_port = lambda conn: conn.laddr[1] == int(self.port)
|
|
|
|
|
chrome_ps_wrapper = psutil.Process(self.chrome_process.pid)
|
2014-01-21 22:53:28 +00:00
|
|
|
while time.time() - start < self.browser_wait and len(list(filter(open_debug_port, chrome_ps_wrapper.get_connections()))) == 0:
|
2014-01-21 06:41:46 +00:00
|
|
|
time.sleep(1)
|
2014-01-21 22:53:28 +00:00
|
|
|
if len(list(filter(open_debug_port, chrome_ps_wrapper.get_connections()))) == 0:
|
2014-01-21 06:41:46 +00:00
|
|
|
self.chrome_process.kill()
|
|
|
|
|
raise Exception("Chrome failed to listen on the debug port in time!")
|
|
|
|
|
|
|
|
|
|
def __exit__(self, *args):
|
2014-01-21 22:53:28 +00:00
|
|
|
print("Killing")
|
2014-01-21 06:41:46 +00:00
|
|
|
self.chrome_process.kill()
|
|
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
|
arg_parser = argparse.ArgumentParser(prog=os.path.basename(sys.argv[0]),
|
|
|
|
|
description='umbra - Browser automation tool',
|
|
|
|
|
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
|
2014-01-21 22:47:49 +00:00
|
|
|
arg_parser.add_argument('-w', '--browser-wait', dest='browser_wait', default='10',
|
|
|
|
|
help='Seconds to wait for browser initialization')
|
|
|
|
|
arg_parser.add_argument('-e', '--executable', dest='executable', default='google-chrome',
|
|
|
|
|
help='Executable to use to invoke chrome')
|
2014-01-21 06:41:46 +00:00
|
|
|
arg_parser.add_argument('-p', '--port', dest='port', default='9222',
|
|
|
|
|
help='Port to have invoked chrome listen on for debugging connections')
|
|
|
|
|
args = arg_parser.parse_args(args=sys.argv[1:])
|
2014-01-21 22:47:49 +00:00
|
|
|
with Chrome(args.port, args.executable, args.browser_wait):
|
2014-01-21 21:32:33 +00:00
|
|
|
Umbra(args.port)
|
