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

from __future__ import unicode_literals

import re

from .common import InfoExtractor
from .ooyala import OoyalaIE


class TheSunIE(InfoExtractor):
    _VALID_URL = r'https://(?:www\.)?thesun\.co\.uk/[^/]+/(?P<id>\d+)'
    _TEST = {
        'url': 'https://www.thesun.co.uk/tvandshowbiz/2261604/orlando-bloom-and-katy-perry-post-adorable-instagram-video-together-celebrating-thanksgiving-after-split-rumours/',
        'info_dict': {
            'id': '2261604',
            'title': 'md5:cba22f48bad9218b64d5bbe0e16afddf',
        },
        'playlist_count': 2,
    }

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

        webpage = self._download_webpage(url, article_id)

        entries = []
        for ooyala_id in re.findall(
                r'<[^>]+\b(?:id\s*=\s*"thesun-ooyala-player-|data-content-id\s*=\s*")([^"]+)',
                webpage):
            entries.append(OoyalaIE._build_url_result(ooyala_id))

        return self.playlist_result(
            entries, article_id, self._og_search_title(webpage, fatal=False))
Commit	Line	Data
5f3e0b69 E	1	from __future__ import unicode_literals
5f3e0b69 E	2
90e3f18f S	3	import re
90e3f18f S	4
5f3e0b69 E	5	from .common import InfoExtractor
	6	from .ooyala import OoyalaIE
	7
	8
	9	class TheSunIE(InfoExtractor):
90e3f18f	10	_VALID_URL = r'https://(?:www\.)?thesun\.co\.uk/[^/]+/(?P<id>\d+)'
5f3e0b69 E	11	_TEST = {
5f3e0b69 E	12	'url': 'https://www.thesun.co.uk/tvandshowbiz/2261604/orlando-bloom-and-katy-perry-post-adorable-instagram-video-together-celebrating-thanksgiving-after-split-rumours/',
5f3e0b69	13	'info_dict': {
90e3f18f S	14	'id': '2261604',
	15	'title': 'md5:cba22f48bad9218b64d5bbe0e16afddf',
	16	},
	17	'playlist_count': 2,
5f3e0b69 E	18	}
	19
	20	def _real_extract(self, url):
90e3f18f S	21	article_id = self._match_id(url)
	22
	23	webpage = self._download_webpage(url, article_id)
5f3e0b69	24
90e3f18f S	25	entries = []
	26	for ooyala_id in re.findall(
	27	r'<[^>]+\b(?:id\s=\s"thesun-ooyala-player-\|data-content-id\s=\s")([^"]+)',
	28	webpage):
	29	entries.append(OoyalaIE._build_url_result(ooyala_id))
5f3e0b69	30
90e3f18f S	31	return self.playlist_result(
90e3f18f S	32	entries, article_id, self._og_search_title(webpage, fatal=False))