[yt-dlp.git] / youtube_dl / extractor / thesun.py

from __future__ import unicode_literals

from .common import InfoExtractor
from .ooyala import OoyalaIE


class TheSunIE(InfoExtractor):
    _VALID_URL = r'https://(?:www\.)?thesun\.co\.uk/\w+/(?P<id>\d+)/[\w-]'
    _TEST = {
        'url': 'https://www.thesun.co.uk/tvandshowbiz/2261604/orlando-bloom-and-katy-perry-post-adorable-instagram-video-together-celebrating-thanksgiving-after-split-rumours/',
        'md5': '5667123b24f25f43f4c4f381ef34c5c2',
        'info_dict': {
            'id': 'h4OXN0NzE6rv6ObkEifKcNA-gYUw4xFf',
            'ext': 'mp4',
            'title': 'Katy Perry and Orlando Bloom shut down split rumours with cute Thanksgiving video',
            'description': 'Still going strong',
            'duration': 31.28,
        }
    }

    def _real_extract(self, url):
        video_id = self._match_id(url)

        webpage = self._download_webpage(url, video_id)
        ooyala_id = self._search_regex(r'id\s*=\s*"thesun-ooyala-player-([^"]+)"', webpage, 'ooyala id')

        return OoyalaIE._build_url_result(ooyala_id)
Commit	Line	Data
5f3e0b69 E	1	from __future__ import unicode_literals
	2
	3	from .common import InfoExtractor
	4	from .ooyala import OoyalaIE
	5
	6
	7	class TheSunIE(InfoExtractor):
	8	_VALID_URL = r'https://(?:www\.)?thesun\.co\.uk/\w+/(?P<id>\d+)/[\w-]'
	9	_TEST = {
	10	'url': 'https://www.thesun.co.uk/tvandshowbiz/2261604/orlando-bloom-and-katy-perry-post-adorable-instagram-video-together-celebrating-thanksgiving-after-split-rumours/',
	11	'md5': '5667123b24f25f43f4c4f381ef34c5c2',
	12	'info_dict': {
	13	'id': 'h4OXN0NzE6rv6ObkEifKcNA-gYUw4xFf',
	14	'ext': 'mp4',
	15	'title': 'Katy Perry and Orlando Bloom shut down split rumours with cute Thanksgiving video',
	16	'description': 'Still going strong',
	17	'duration': 31.28,
	18	}
	19	}
	20
	21	def _real_extract(self, url):
	22	video_id = self._match_id(url)
	23
	24	webpage = self._download_webpage(url, video_id)
	25	ooyala_id = self._search_regex(r'id\s=\s"thesun-ooyala-player-([^"]+)"', webpage, 'ooyala id')
	26
	27	return OoyalaIE._build_url_result(ooyala_id)