From 065472936a81fcad263dc2f1b04fdfe4a221eeb9 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Wed, 6 Nov 2013 17:37:39 +0100 Subject: [PATCH] Add an extractor for space.com (fixes #1718) It uses Brightcove, but requires some special process for getting a url with the playerKey field in some videos --- youtube_dl/extractor/__init__.py | 1 + youtube_dl/extractor/space.py | 35 ++++++++++++++++++++++++++++++++ 2 files changed, 36 insertions(+) create mode 100644 youtube_dl/extractor/space.py diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index 888a91cce..d8626fb39 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -115,6 +115,7 @@ from .slideshare import SlideshareIE from .sohu import SohuIE from .soundcloud import SoundcloudIE, SoundcloudSetIE, SoundcloudUserIE from .southparkstudios import SouthParkStudiosIE +from .space import SpaceIE from .spankwire import SpankwireIE from .spiegel import SpiegelIE from .stanfordoc import StanfordOpenClassroomIE diff --git a/youtube_dl/extractor/space.py b/youtube_dl/extractor/space.py new file mode 100644 index 000000000..0d32a0688 --- /dev/null +++ b/youtube_dl/extractor/space.py @@ -0,0 +1,35 @@ +import re + +from .common import InfoExtractor +from .brightcove import BrightcoveIE +from ..utils import RegexNotFoundError, ExtractorError + + +class SpaceIE(InfoExtractor): + _VALID_URL = r'https?://www\.space\.com/\d+-(?P[^/\.\?]*?)-video.html' + _TEST = { + u'add_ie': ['Brightcove'], + u'url': u'http://www.space.com/23373-huge-martian-landforms-detail-revealed-by-european-probe-video.html', + u'info_dict': { + u'id': u'2780937028001', + u'ext': u'mp4', + u'title': u'Huge Martian Landforms\' Detail Revealed By European Probe | Video', + u'description': u'md5:db81cf7f3122f95ed234b631a6ea1e61', + u'uploader': u'TechMedia Networks', + }, + } + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + title = mobj.group('title') + webpage = self._download_webpage(url, title) + try: + # Some videos require the playerKey field, which isn't define in + # the BrightcoveExperience object + brightcove_url = self._og_search_video_url(webpage) + except RegexNotFoundError: + # Other videos works fine with the info from the object + brightcove_url = BrightcoveIE._extract_brightcove_url(webpage) + if brightcove_url is None: + raise ExtractorError(u'The webpage does not contain a video', expected=True) + return self.url_result(brightcove_url, BrightcoveIE.ie_key()) -- 2.40.0