X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/33cd347759d6d999325ebf3c69b7ed5692c343b2..88ab9dae6d22f29ff7ac32f25c2d5c26d732edf1:/youtube_dl/extractor/breakcom.py diff --git a/youtube_dl/extractor/breakcom.py b/youtube_dl/extractor/breakcom.py index 53a898d..1bfc9f3 100644 --- a/youtube_dl/extractor/breakcom.py +++ b/youtube_dl/extractor/breakcom.py @@ -1,18 +1,20 @@ +from __future__ import unicode_literals + import re import json from .common import InfoExtractor -from ..utils import determine_ext class BreakIE(InfoExtractor): - _VALID_URL = r'(?:http://)?(?:www\.)?break\.com/video/([^/]+)' + _VALID_URL = r'http://(?:www\.)?break\.com/video/([^/]+)' _TEST = { - u'url': u'http://www.break.com/video/when-girls-act-like-guys-2468056', - u'file': u'2468056.mp4', - u'md5': u'a3513fb1547fba4fb6cfac1bffc6c46b', - u'info_dict': { - u"title": u"When Girls Act Like D-Bags" + 'url': 'http://www.break.com/video/when-girls-act-like-guys-2468056', + 'md5': 'a3513fb1547fba4fb6cfac1bffc6c46b', + 'info_dict': { + 'id': '2468056', + 'ext': 'mp4', + 'title': 'When Girls Act Like D-Bags', } } @@ -21,18 +23,18 @@ class BreakIE(InfoExtractor): video_id = mobj.group(1).split("-")[-1] embed_url = 'http://www.break.com/embed/%s' % video_id webpage = self._download_webpage(embed_url, video_id) - info_json = self._search_regex(r'var embedVars = ({.*?});', webpage, - u'info json', flags=re.DOTALL) + info_json = self._search_regex(r'var embedVars = ({.*})\s*?', + webpage, 'info json', flags=re.DOTALL) info = json.loads(info_json) video_url = info['videoUri'] - m_youtube = re.search(r'(https?://www\.youtube\.com/watch\?v=.*)', video_url) - if m_youtube is not None: - return self.url_result(m_youtube.group(1), 'Youtube') + youtube_id = info.get('youtubeId') + if youtube_id: + return self.url_result(youtube_id, 'Youtube') + final_url = video_url + '?' + info['AuthToken'] - return [{ - 'id': video_id, - 'url': final_url, - 'ext': determine_ext(final_url), - 'title': info['contentName'], + return { + 'id': video_id, + 'url': final_url, + 'title': info['contentName'], 'thumbnail': info['thumbUri'], - }] + }