mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-02-24 08:39:59 -05:00
32 lines
1.2 KiB
Python
32 lines
1.2 KiB
Python
#!/usr/bin/env python
|
|
# vim: set sw=4 et:
|
|
|
|
import argparse
|
|
import os
|
|
import sys
|
|
import logging
|
|
import umbra
|
|
import kombu
|
|
|
|
arg_parser = argparse.ArgumentParser(prog=os.path.basename(__file__),
|
|
description="brozzler-add-site - register site to crawl with brozzler-hq",
|
|
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
|
|
arg_parser.add_argument('urls', metavar='URL', nargs='+', help='seed URL(s) of sites to crawl')
|
|
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("-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, os.path.basename(__file__)))
|
|
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")
|
|
|
|
with kombu.Connection(args.amqp_url) as conn:
|
|
q = conn.SimpleQueue("brozzler.sites.new")
|
|
for url in args.urls:
|
|
q.put({"seed":url})
|
|
|
|
|