mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-02-24 08:39:59 -05:00
124 lines
4.6 KiB
Python
Executable File
124 lines
4.6 KiB
Python
Executable File
#!/usr/bin/env python
|
|
# vim: set sw=4 et:
|
|
|
|
import logging
|
|
import argparse
|
|
import time
|
|
import umbra
|
|
import sys
|
|
import signal
|
|
import os
|
|
import umbra
|
|
import signal
|
|
import threading
|
|
import traceback
|
|
import pprint
|
|
|
|
if __name__ == "__main__":
|
|
arg_parser = argparse.ArgumentParser(prog=os.path.basename(__file__),
|
|
description='umbra - browser automation tool communicating via AMQP',
|
|
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
|
|
arg_parser.add_argument('-e', '--executable', dest='chrome_exe', default='chromium-browser',
|
|
help='Executable to use to invoke chrome')
|
|
arg_parser.add_argument('-u', '--url', dest='amqp_url', default='amqp://guest:guest@localhost:5672/%2f',
|
|
help='URL identifying the amqp server to talk to')
|
|
arg_parser.add_argument('--exchange', dest='amqp_exchange', default='umbra',
|
|
help='AMQP exchange name')
|
|
arg_parser.add_argument('--queue', dest='amqp_queue', default='urls',
|
|
help='AMQP queue to consume urls from')
|
|
arg_parser.add_argument('-n', '--max-browsers', dest='max_browsers', default='1',
|
|
help='Max number of chrome instances simultaneously browsing pages')
|
|
arg_parser.add_argument('-v', '--verbose', dest='log_level',
|
|
action="store_const", default=logging.INFO, const=logging.DEBUG)
|
|
arg_parser.add_argument('--version', action='version',
|
|
version="umbra {}".format(umbra.version))
|
|
args = arg_parser.parse_args(args=sys.argv[1:])
|
|
|
|
logging.basicConfig(stream=sys.stdout, level=args.log_level,
|
|
format='%(asctime)s %(process)d %(levelname)s %(threadName)s %(name)s.%(funcName)s(%(filename)s:%(lineno)d) %(message)s')
|
|
|
|
logging.info("umbra {} starting up".format(umbra.version))
|
|
|
|
controller = umbra.Umbra(args.amqp_url, args.chrome_exe,
|
|
max_active_browsers=int(args.max_browsers),
|
|
exchange_name=args.amqp_exchange, queue_name=args.amqp_queue)
|
|
|
|
def browserdump_str(pp, b):
|
|
x = []
|
|
x.append(pp.pformat(b.__dict__))
|
|
if b._chrome_instance:
|
|
x.append("=> {} chrome instance:".format(b))
|
|
x.append(pp.pformat(b._chrome_instance.__dict__))
|
|
if b._behavior:
|
|
x.append("=> {} active behavior:".format(b))
|
|
x.append(pp.pformat(b._behavior.__dict__))
|
|
return "\n".join(x)
|
|
|
|
def dump_state(signum, frame):
|
|
pp = pprint.PrettyPrinter(indent=4)
|
|
state_strs = []
|
|
|
|
for th in threading.enumerate():
|
|
state_strs.append(str(th))
|
|
stack = traceback.format_stack(sys._current_frames()[th.ident])
|
|
state_strs.append("".join(stack))
|
|
|
|
state_strs.append("umbra controller:")
|
|
state_strs.append(pp.pformat(controller.__dict__))
|
|
state_strs.append("")
|
|
|
|
for b in controller._browser_pool._in_use:
|
|
state_strs.append("{} (in use):".format(b))
|
|
state_strs.append(browserdump_str(pp, b))
|
|
state_strs.append("")
|
|
for b in controller._browser_pool._available:
|
|
state_strs.append("{} (not in use):".format(b))
|
|
state_strs.append(browserdump_str(pp, b))
|
|
state_strs.append("")
|
|
|
|
logging.warn("dumping state (caught signal {})\n{}".format(signum, "\n".join(state_strs)))
|
|
|
|
|
|
class ShutdownRequested(Exception):
|
|
pass
|
|
|
|
def sigterm(signum, frame):
|
|
raise ShutdownRequested('shutdown requested (caught SIGTERM)')
|
|
def sigint(signum, frame):
|
|
raise ShutdownRequested('shutdown requested (caught SIGINT)')
|
|
|
|
signal.signal(signal.SIGQUIT, dump_state)
|
|
signal.signal(signal.SIGHUP, controller.reconnect)
|
|
signal.signal(signal.SIGTERM, sigterm)
|
|
signal.signal(signal.SIGINT, sigint)
|
|
|
|
controller.start()
|
|
|
|
try:
|
|
while True:
|
|
time.sleep(0.5)
|
|
except ShutdownRequested as e:
|
|
logging.info(e)
|
|
except BaseException as e:
|
|
logging.fatal(e, exc_info=sys.exc_info())
|
|
finally:
|
|
try:
|
|
controller.shutdown()
|
|
for th in threading.enumerate():
|
|
if th != threading.current_thread():
|
|
th.join()
|
|
except BaseException as e:
|
|
logging.warn("caught exception {}".format(e))
|
|
for i in range(6,0,-1):
|
|
controller.shutdown_now()
|
|
try:
|
|
for th in threading.enumerate():
|
|
if th != threading.current_thread():
|
|
th.join()
|
|
break # if we get here, we're done, all threads finished
|
|
except:
|
|
logging.warn("caught exception {}".format(e))
|
|
|
|
logging.info("all finished, exiting")
|
|
|