rudimentary link extraction and crawling

bin/crawl-url

@@ -0,0 +1,39 @@
+#!/usr/bin/env python
+# vim: set sw=4 et:
+
+import argparse
+import os
+import sys
+import logging
+import umbra
+
+arg_parser = argparse.ArgumentParser(prog=os.path.basename(__file__),
+        description='browse-url - open urls in chrome/chromium and run behaviors',
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+arg_parser.add_argument('urls', metavar='URL', nargs='+', help='URL(s) to browse')
+arg_parser.add_argument('-w', '--browser-wait', dest='browser_wait', default='60',
+        help='seconds to wait for browser initialization')
+arg_parser.add_argument('-e', '--executable', dest='chrome_exe', default='chromium-browser',
+        help='executable to use to invoke chrome')
+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')
+
+frontier = list(args.urls)
+
+def add_to_frontier(urls):
+    logging.info("adding {} urls to frontier".format(len(urls)))
+    frontier.extend(urls)
+
+with umbra.Browser(chrome_exe=args.chrome_exe) as browser:
+    try:
+        while True:
+            browser.browse_page(frontier.pop(), on_outlinks=add_to_frontier)
+    except IndexError:
+        logging.info("finished, frontier is empty")
+