X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/15b1d10671b48df598afd70e17ba21e9e64ac766..2b4515201a54c38e34434ea2817d606be134a7c9:/youtube_dl/extractor/vrt.py?ds=inline diff --git a/youtube_dl/extractor/vrt.py b/youtube_dl/extractor/vrt.py index bbd3bbf..00c72e3 100644 --- a/youtube_dl/extractor/vrt.py +++ b/youtube_dl/extractor/vrt.py @@ -4,11 +4,13 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..utils import float_or_none +from ..utils import ( + float_or_none, +) class VRTIE(InfoExtractor): - _VALID_URL = r'https?://(?:deredactie|sporza|cobra)\.be/cm/(?:[^/]+/)+(?P[^/]+)/*' + _VALID_URL = r'https?://(?:deredactie|sporza|cobra(?:\.canvas)?)\.be/cm/(?:[^/]+/)+(?P[^/]+)/*' _TESTS = [ # deredactie.be { @@ -22,7 +24,8 @@ class VRTIE(InfoExtractor): 'timestamp': 1414271750.949, 'upload_date': '20141025', 'duration': 929, - } + }, + 'skip': 'HTTP Error 404: Not Found', }, # sporza.be { @@ -36,7 +39,8 @@ class VRTIE(InfoExtractor): 'timestamp': 1413835980.560, 'upload_date': '20141020', 'duration': 3238, - } + }, + 'skip': 'HTTP Error 404: Not Found', }, # cobra.be { @@ -50,8 +54,39 @@ class VRTIE(InfoExtractor): 'timestamp': 1413967500.494, 'upload_date': '20141022', 'duration': 661, - } + }, + 'skip': 'HTTP Error 404: Not Found', }, + { + # YouTube video + 'url': 'http://deredactie.be/cm/vrtnieuws/videozone/nieuws/cultuurenmedia/1.2622957', + 'md5': 'b8b93da1df1cea6c8556255a796b7d61', + 'info_dict': { + 'id': 'Wji-BZ0oCwg', + 'ext': 'mp4', + 'title': 'ROGUE ONE: A STAR WARS STORY Official Teaser Trailer', + 'description': 'md5:8e468944dce15567a786a67f74262583', + 'uploader': 'Star Wars', + 'uploader_id': 'starwars', + 'upload_date': '20160407', + }, + 'add_ie': ['Youtube'], + }, + { + 'url': 'http://cobra.canvas.be/cm/cobra/videozone/rubriek/film-videozone/1.2377055', + 'info_dict': { + 'id': '2377055', + 'ext': 'mp4', + 'title': 'Cafe Derby', + 'description': 'Lenny Van Wesemael debuteert met de langspeelfilm Café Derby. Een waar gebeurd maar ook verzonnen verhaal.', + 'upload_date': '20150626', + 'timestamp': 1435305240.769, + }, + 'params': { + # m3u8 download + 'skip_download': True, + } + } ] def _real_extract(self, url): @@ -62,18 +97,41 @@ class VRTIE(InfoExtractor): video_id = self._search_regex( r'data-video-id="([^"]+)_[^"]+"', webpage, 'video id', fatal=False) + src = self._search_regex( + r'data-video-src="([^"]+)"', webpage, 'video src', default=None) + + video_type = self._search_regex( + r'data-video-type="([^"]+)"', webpage, 'video type', default=None) + + if video_type == 'YouTubeVideo': + return self.url_result(src, 'Youtube') + formats = [] + mobj = re.search( r'data-video-iphone-server="(?P[^"]+)"\s+data-video-iphone-path="(?P[^"]+)"', webpage) if mobj: formats.extend(self._extract_m3u8_formats( '%s/%s' % (mobj.group('server'), mobj.group('path')), - video_id, 'mp4')) - mobj = re.search(r'data-video-src="(?P[^"]+)"', webpage) - if mobj: - formats.extend(self._extract_f4m_formats( - '%s/manifest.f4m' % mobj.group('src'), video_id)) + video_id, 'mp4', m3u8_id='hls', fatal=False)) + + if src: + formats = self._extract_wowza_formats(src, video_id) + if 'data-video-geoblocking="true"' not in webpage: + for f in formats: + if f['url'].startswith('rtsp://'): + http_format = f.copy() + http_format.update({ + 'url': f['url'].replace('rtsp://', 'http://').replace('vod.', 'download.').replace('/_definst_/', '/').replace('mp4:', ''), + 'format_id': f['format_id'].replace('rtsp', 'http'), + 'protocol': 'http', + }) + formats.append(http_format) + + if not formats and 'data-video-geoblocking="true"' in webpage: + self.raise_geo_restricted('This video is only available in Belgium') + self._sort_formats(formats) title = self._og_search_title(webpage)