X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fbloomberg.py;h=2fbfad1ba04adf915c41441173a6464c6b96b100;hb=HEAD;hp=bd538be50bc4a650d000eaffb5e292a4e8e76cbe;hpb=4315f74fa8e97ca1fdd1fe919f777b3942da2028;p=youtube-dl diff --git a/youtube_dl/extractor/bloomberg.py b/youtube_dl/extractor/bloomberg.py index bd538be50..2fbfad1ba 100644 --- a/youtube_dl/extractor/bloomberg.py +++ b/youtube_dl/extractor/bloomberg.py @@ -1,3 +1,4 @@ +# coding: utf-8 from __future__ import unicode_literals import re @@ -20,6 +21,22 @@ class BloombergIE(InfoExtractor): 'params': { 'format': 'best[format_id^=hds]', }, + }, { + # video ID in BPlayer(...) + 'url': 'http://www.bloomberg.com/features/2016-hello-world-new-zealand/', + 'info_dict': { + 'id': '938c7e72-3f25-4ddb-8b85-a9be731baa74', + 'ext': 'flv', + 'title': 'Meet the Real-Life Tech Wizards of Middle Earth', + 'description': 'Hello World, Episode 1: New Zealand’s freaky AI babies, robot exoskeletons, and a virtual you.', + }, + 'params': { + 'format': 'best[format_id^=hds]', + }, + }, { + # data-bmmrid= + 'url': 'https://www.bloomberg.com/politics/articles/2017-02-08/le-pen-aide-briefed-french-central-banker-on-plan-to-print-money', + 'only_matching': True, }, { 'url': 'http://www.bloomberg.com/news/articles/2015-11-12/five-strange-things-that-have-been-happening-in-financial-markets', 'only_matching': True, @@ -32,8 +49,14 @@ class BloombergIE(InfoExtractor): name = self._match_id(url) webpage = self._download_webpage(url, name) video_id = self._search_regex( - r'["\']bmmrId["\']\s*:\s*(["\'])(?P.+?)\1', - webpage, 'id', group='url') + (r'["\']bmmrId["\']\s*:\s*(["\'])(?P(?:(?!\1).)+)\1', + r'videoId\s*:\s*(["\'])(?P(?:(?!\1).)+)\1', + r'data-bmmrid=(["\'])(?P(?:(?!\1).)+)\1'), + webpage, 'id', group='id', default=None) + if not video_id: + bplayer_data = self._parse_json(self._search_regex( + r'BPlayer\(null,\s*({[^;]+})\);', webpage, 'id'), name) + video_id = bplayer_data['id'] title = re.sub(': Video$', '', self._og_search_title(webpage)) embed_info = self._download_json(