dosage/scripts/webtoons.py

40 lines
1.3 KiB
Python
Raw Normal View History

2019-06-12 07:42:06 +00:00
#!/usr/bin/env python
# SPDX-License-Identifier: MIT
2019-06-12 07:42:06 +00:00
# Copyright (C) 2019-2020 Tobias Gruetzmacher
# Copyright (C) 2019-2020 Daniel Ring
"""
Script to get a list of WebToons comics and save the info in a
JSON file for further processing.
"""
from urllib.parse import urlsplit, parse_qs
2019-06-12 07:42:06 +00:00
from scriptutil import ComicListUpdater
from dosagelib.util import check_robotstxt
class WebToonsUpdater(ComicListUpdater):
def collect_results(self):
# Parse the comic list page
data = self.get_url('https://www.webtoons.com/en/dailySchedule')
for comiclink in data.xpath('//a[contains(@class, "daily_card_item")]'):
comicurl = comiclink.attrib['href']
name = comiclink.xpath('.//div[@class="info"]/p[@class="subj"]')[0].text
try:
check_robotstxt(comicurl, self.session)
except IOError as e:
print('[%s] INFO: robots.txt denied: %s' % (name, e))
continue
self.add_comic(name, comicurl)
def get_entry(self, name, url):
shortName = name.replace(' ', '')
titleNum = int(parse_qs(urlsplit(url).query)['title_no'][0])
url = url.rsplit('/', 1)[0].replace('https://www.webtoons.com/en/', '')
return u"cls('%s', '%s', %d)," % (shortName, url, titleNum)
if __name__ == '__main__':
WebToonsUpdater(__file__).run()