def __init__(self, config_file):
if not os.path.exists(config_file):
logging.warning('[config] config file "%s" does not exist, program exit...' % config_file)
sys.exit()
with open(config_file) as config_fp:
import json
config_data = json.load(config_fp, object_pairs_hook=collections.OrderedDict)
self.notifications = collections.OrderedDict() # key: notify_name, value: Notification
notifications = config_data[NOTIFY_KEY] if NOTIFY_KEY in config_data else None
if not notifications:
logging.warning("[config] no notification specified, program exit...")
sys.exit()
for notify_name in notifications:
from wbnt_path import adjust_path
adjust_path()
from util.setting.notify import Notification
self.notifications[notify_name] = Notification.create(notify_name, notifications[notify_name])
from util.setting.audience import Audience
audiences = config_data[AUDIENCE_KEY] if AUDIENCE_KEY in config_data else None
self.audiences = Audience.create(audiences)
def setUp(self):
from wbnt_path import adjust_path
adjust_path()
if __name__ == '__main__':
from argparse import ArgumentParser
arg_parser = ArgumentParser(description="Test utility for HTMLParser")
arg_parser.add_argument('htm_file')
arg_parser.add_argument("-s", "--strip", dest="strip", action="store_const",
const=True, help="strip (default: '%s')" % False)
arg_parser.add_argument("-c", "--combine_br", dest="combine_br", action="store_const",
const=True, help="combine <br> (default: '%s')" % False)
arg_parser.add_argument("-rmw", "--remove_ws", dest="remove_ws", action="store_const",
const=True, help="remove whitespace (default: '%s')" % False)
arg_parser.add_argument("-rmh", "--remove_hyperlink", dest="remove_hyperlink", action="store_const",
const=True, help="remove hyperlink tag (default: '%s')" % False)
args = arg_parser.parse_args()
if not args.htm_file or not os.path.exists(args.htm_file):
import sys
sys.exit()
with open(args.htm_file, encoding='utf8') as fd:
web_content = fd.read()
if args.remove_hyperlink:
web_content = HtmAnalyzer.remove_hyperlink_tag(web_content)
from wbnt_path import adjust_path
adjust_path()
from util.setting.parser import HtmParseSetting, STRIP_KEY, REMOVE_WS_KEY, REMOVE_LIST_KEY, COMBINE_BR_KEY
remove_list = None
setting = HtmParseSetting({STRIP_KEY: args.strip, REMOVE_WS_KEY: args.remove_ws,
REMOVE_LIST_KEY: remove_list, COMBINE_BR_KEY: args.combine_br})
content = HTMLParser().parse_and_retrieve(web_content, setting=setting)
for i, line in zip(range(len(content)), content):
print(i, "=>", line)
