From 9e5751b9fe72f7425e4cb3f22a56b6a95b59e41d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Fri, 19 Aug 2016 01:13:45 +0700 Subject: [PATCH] [globo:article] Relax _VALID_URL and video id regex (Closes #10379) --- youtube_dl/extractor/globo.py | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/globo.py b/youtube_dl/extractor/globo.py index 3de8356f6..dbacbfc61 100644 --- a/youtube_dl/extractor/globo.py +++ b/youtube_dl/extractor/globo.py @@ -396,12 +396,12 @@ def _real_extract(self, url): class GloboArticleIE(InfoExtractor): - _VALID_URL = 'https?://.+?\.globo\.com/(?:[^/]+/)*(?P[^/]+)\.html' + _VALID_URL = 'https?://.+?\.globo\.com/(?:[^/]+/)*(?P[^/]+)(?:\.html)?' _VIDEOID_REGEXES = [ r'\bdata-video-id=["\'](\d{7,})', r'\bdata-player-videosids=["\'](\d{7,})', - r'\bvideosIDs\s*:\s*["\'](\d{7,})', + r'\bvideosIDs\s*:\s*["\']?(\d{7,})', r'\bdata-id=["\'](\d{7,})', r']+\bid=["\'](\d{7,})', ] @@ -423,6 +423,9 @@ class GloboArticleIE(InfoExtractor): }, { 'url': 'http://gshow.globo.com/programas/tv-xuxa/O-Programa/noticia/2014/01/xuxa-e-junno-namoram-muuuito-em-luau-de-zeze-di-camargo-e-luciano.html', 'only_matching': True, + }, { + 'url': 'http://oglobo.globo.com/rio/a-amizade-entre-um-entregador-de-farmacia-um-piano-19946271', + 'only_matching': True, }] @classmethod