diff options
Diffstat (limited to 'youtube_dl/extractor/nbc.py')
-rw-r--r-- | youtube_dl/extractor/nbc.py | 11 |
1 files changed, 9 insertions, 2 deletions
diff --git a/youtube_dl/extractor/nbc.py b/youtube_dl/extractor/nbc.py index 6cbe03d0f..dc2091be0 100644 --- a/youtube_dl/extractor/nbc.py +++ b/youtube_dl/extractor/nbc.py @@ -10,6 +10,8 @@ from ..compat import ( from ..utils import ( ExtractorError, find_xpath_attr, + lowercase_escape, + unescapeHTML, ) @@ -46,18 +48,23 @@ class NBCIE(InfoExtractor): 'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442', }, 'skip': 'Only works from US', + }, + { + # This video has expired but with an escaped embedURL + 'url': 'http://www.nbc.com/parenthood/episode-guide/season-5/just-like-at-home/515', + 'skip': 'Expired' } ] def _real_extract(self, url): video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) - theplatform_url = self._search_regex( + theplatform_url = unescapeHTML(lowercase_escape(self._html_search_regex( [ r'(?:class="video-player video-player-full" data-mpx-url|class="player" src)="(.*?)"', r'"embedURL"\s*:\s*"([^"]+)"' ], - webpage, 'theplatform url').replace('_no_endcard', '') + webpage, 'theplatform url').replace('_no_endcard', '').replace('\\/', '/'))) if theplatform_url.startswith('//'): theplatform_url = 'http:' + theplatform_url return self.url_result(theplatform_url) |