|
|
|
@ -7,6 +7,7 @@ from .common import InfoExtractor
|
|
|
|
|
from ..utils import (
|
|
|
|
|
ExtractorError,
|
|
|
|
|
int_or_none,
|
|
|
|
|
traverse_obj,
|
|
|
|
|
smuggle_url,
|
|
|
|
|
unsmuggle_url,
|
|
|
|
|
)
|
|
|
|
@ -55,9 +56,6 @@ class LiTVIE(InfoExtractor):
|
|
|
|
|
episode_title = program_info['title']
|
|
|
|
|
content_id = season_list['contentId']
|
|
|
|
|
|
|
|
|
|
if prompt:
|
|
|
|
|
self.to_screen('Downloading playlist %s - add --no-playlist to just download video %s' % (content_id, video_id))
|
|
|
|
|
|
|
|
|
|
all_episodes = [
|
|
|
|
|
self.url_result(smuggle_url(
|
|
|
|
|
self._URL_TEMPLATE % (program_info['contentType'], episode['contentId']),
|
|
|
|
@ -71,12 +69,6 @@ class LiTVIE(InfoExtractor):
|
|
|
|
|
|
|
|
|
|
video_id = self._match_id(url)
|
|
|
|
|
|
|
|
|
|
noplaylist = self.get_param('noplaylist')
|
|
|
|
|
noplaylist_prompt = True
|
|
|
|
|
if 'force_noplaylist' in data:
|
|
|
|
|
noplaylist = data['force_noplaylist']
|
|
|
|
|
noplaylist_prompt = False
|
|
|
|
|
|
|
|
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
|
|
|
|
|
|
program_info = self._parse_json(self._search_regex(
|
|
|
|
@ -84,14 +76,9 @@ class LiTVIE(InfoExtractor):
|
|
|
|
|
video_id)
|
|
|
|
|
|
|
|
|
|
season_list = list(program_info.get('seasonList', {}).values())
|
|
|
|
|
if season_list:
|
|
|
|
|
if not noplaylist:
|
|
|
|
|
return self._extract_playlist(
|
|
|
|
|
season_list[0], video_id, program_info,
|
|
|
|
|
prompt=noplaylist_prompt)
|
|
|
|
|
|
|
|
|
|
if noplaylist_prompt:
|
|
|
|
|
self.to_screen('Downloading just video %s because of --no-playlist' % video_id)
|
|
|
|
|
playlist_id = traverse_obj(season_list, 0, 'contentId')
|
|
|
|
|
if self._yes_playlist(playlist_id, video_id, smuggled_data):
|
|
|
|
|
return self._extract_playlist(season_list[0], video_id, program_info)
|
|
|
|
|
|
|
|
|
|
# In browsers `getMainUrl` request is always issued. Usually this
|
|
|
|
|
# endpoint gives the same result as the data embedded in the webpage.
|
|
|
|
|