brozzler/umbra/umbra.py

118 lines
5.1 KiB
Python
Raw Normal View History

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
import uuid
import logging
import threading
2014-01-22 07:38:03 +00:00
from kombu import Connection, Exchange, Queue
logging.basicConfig(level=logging.INFO)
2014-01-21 06:41:46 +00:00
class Umbra:
2014-01-22 07:38:03 +00:00
def __init__(self, port, amqpurl):
self.cmd_id = 0
self.chrome_debug_port = port
2014-01-22 07:38:03 +00:00
self.producer = None
self.amqpurl = amqpurl
self.launch_tab_socket = self.get_websocket(self.on_open)
threading.Thread(target=self.launch_tab_socket.run_forever).start()
2014-01-21 06:41:46 +00:00
def get_websocket(self, on_open, url=None):
2014-01-22 01:32:41 +00:00
def fetch_debugging_json():
return loads(urllib.request.urlopen("http://localhost:%s/json" % self.chrome_debug_port).read().decode('utf-8').replace("\\n",""))
time.sleep(0.5)
debug_info = []
while not debug_info:
debug_info = fetch_debugging_json()
if url:
debug_info = [x for x in debug_info if x['url'] == url]
return_socket = websocket.WebSocketApp(debug_info[0]['webSocketDebuggerUrl'], on_message = self.handle_message, on_error = print )
return_socket.on_open = on_open
return return_socket
def handle_message(self, ws, message):
message = loads(message)
if "method" in message.keys() and message["method"] == "Network.requestWillBeSent":
request_queue = Queue('requests', routing_key='request', exchange=self.umbra_exchange)
self.producer.publish(message['params']['request'],routing_key='request', exchange=self.umbra_exchange, declare=[request_queue])
2014-01-22 07:38:03 +00:00
2014-01-22 01:32:41 +00:00
2014-01-22 07:38:03 +00:00
def start_amqp(self):
self.umbra_exchange = Exchange('umbra', 'direct', durable=True)
url_queue = Queue('urls', routing_key='url', exchange=self.umbra_exchange)
with Connection(self.amqpurl) as conn:
self.producer = conn.Producer(serializer='json')
with conn.Consumer(url_queue, callbacks=[self.fetch_url]) as consumer:
while True:
conn.drain_events()
def on_open(self, ws):
2014-01-22 07:38:03 +00:00
threading.Thread(target=self.start_amqp).start()
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))
2014-01-22 07:38:03 +00:00
def fetch_url(self, body, message):
url = body['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):
ws.on_message=self.handle_message
self.send_command(tab=ws, method="Network.enable")
self.send_command(tab=ws, method="Runtime.evaluate", params={"expression":"document.location = '%s';" % url})
def do_close():
time.sleep(10)
self.send_command(tab=ws, method="Runtime.evaluate", params={"expression":"window.open('', '_self', ''); window.close(); "})
#threading.Thread(target=do_close).start()
socket = self.get_websocket(on_open, new_page)
2014-01-22 07:38:03 +00:00
message.ack()
threading.Thread(target=socket.run_forever).start()
2014-01-21 06:41:46 +00:00
class Chrome():
def __init__(self, port, executable, browser_wait):
2014-01-21 06:41:46 +00:00
self.port = port
self.executable = executable
self.browser_wait=browser_wait
2014-01-21 06:41:46 +00:00
def __enter__(self):
import psutil, subprocess
self.chrome_process = subprocess.Popen([self.executable, "--disable-web-sockets", "--temp-profile", "--remote-debugging-port=%s" % self.port])
2014-01-21 06:41:46 +00:00
start = time.time()
import socket
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
while sock.connect_ex(('127.0.0.1',int(self.port))) != 0 and (time.time() - start) < float(self.browser_wait):
time.sleep(0.1)
2014-01-21 06:41:46 +00:00
def __exit__(self, *args):
self.chrome_process.kill()
def main():
2014-01-21 06:41:46 +00:00
arg_parser = argparse.ArgumentParser(prog=os.path.basename(sys.argv[0]),
description='umbra - Browser automation tool',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
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='chromium-browser',
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')
2014-01-22 07:38:03 +00:00
arg_parser.add_argument('-u', '--url', dest='amqpurl', default='amqp://guest:guest@localhost:5672//',
help='URL identifying the amqp server to talk to')
2014-01-21 06:41:46 +00:00
args = arg_parser.parse_args(args=sys.argv[1:])
with Chrome(args.port, args.executable, args.browser_wait):
2014-01-22 07:38:03 +00:00
Umbra(args.port, args.amqpurl)
while True:
time.sleep(1)
if __name__ == "__main__":
main()