mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-16 01:38:00 +00:00
7a5c1cfe93
* All modules and binary names are changed * All documentation references changed * yt-dlp no longer loads youtube-dlc config files * All URLs changed to point to organization account Co-authored-by: Pccode66 Co-authored-by: pukkandan
80 lines
2.9 KiB
Python
80 lines
2.9 KiB
Python
# coding: utf-8
|
|
from __future__ import unicode_literals
|
|
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from .brightcove import BrightcoveNewIE
|
|
from ..utils import (
|
|
int_or_none,
|
|
parse_age_limit,
|
|
smuggle_url,
|
|
unescapeHTML,
|
|
)
|
|
|
|
|
|
class VrakIE(InfoExtractor):
|
|
_VALID_URL = r'https?://(?:www\.)?vrak\.tv/videos\?.*?\btarget=(?P<id>[\d.]+)'
|
|
_TEST = {
|
|
'url': 'http://www.vrak.tv/videos?target=1.2306782&filtre=emission&id=1.1806721',
|
|
'info_dict': {
|
|
'id': '5345661243001',
|
|
'ext': 'mp4',
|
|
'title': 'Obésité, film de hockey et Roseline Filion',
|
|
'timestamp': 1488492126,
|
|
'upload_date': '20170302',
|
|
'uploader_id': '2890187628001',
|
|
'creator': 'VRAK.TV',
|
|
'age_limit': 8,
|
|
'series': 'ALT (Actualité Légèrement Tordue)',
|
|
'episode': 'Obésité, film de hockey et Roseline Filion',
|
|
'tags': list,
|
|
},
|
|
'params': {
|
|
'skip_download': True,
|
|
},
|
|
}
|
|
BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/2890187628001/default_default/index.html?videoId=%s'
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
title = self._html_search_regex(
|
|
r'<h\d\b[^>]+\bclass=["\']videoTitle["\'][^>]*>([^<]+)',
|
|
webpage, 'title', default=None) or self._og_search_title(webpage)
|
|
|
|
content = self._parse_json(
|
|
self._search_regex(
|
|
r'data-player-options-content=(["\'])(?P<content>{.+?})\1',
|
|
webpage, 'content', default='{}', group='content'),
|
|
video_id, transform_source=unescapeHTML)
|
|
|
|
ref_id = content.get('refId') or self._search_regex(
|
|
r'refId":"([^&]+)"', webpage, 'ref id')
|
|
|
|
brightcove_id = self._search_regex(
|
|
r'''(?x)
|
|
java\.lang\.String\s+value\s*=\s*["']brightcove\.article\.\d+\.%s
|
|
[^>]*
|
|
java\.lang\.String\s+value\s*=\s*["'](\d+)
|
|
''' % re.escape(ref_id), webpage, 'brightcove id')
|
|
|
|
return {
|
|
'_type': 'url_transparent',
|
|
'ie_key': BrightcoveNewIE.ie_key(),
|
|
'url': smuggle_url(
|
|
self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
|
|
{'geo_countries': ['CA']}),
|
|
'id': brightcove_id,
|
|
'description': content.get('description'),
|
|
'creator': content.get('brand'),
|
|
'age_limit': parse_age_limit(content.get('rating')),
|
|
'series': content.get('showName') or content.get(
|
|
'episodeName'), # this is intentional
|
|
'season_number': int_or_none(content.get('seasonNumber')),
|
|
'episode': title,
|
|
'episode_number': int_or_none(content.get('episodeNumber')),
|
|
'tags': content.get('tags', []),
|
|
}
|