2014-02-19 00:04:24 +00:00
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
2013-11-06 16:37:39 +00:00
|
|
|
import re
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
2015-11-13 23:54:16 +00:00
|
|
|
from .brightcove import BrightcoveLegacyIE
|
2013-11-06 16:37:39 +00:00
|
|
|
from ..utils import RegexNotFoundError, ExtractorError
|
|
|
|
|
|
|
|
|
|
|
|
class SpaceIE(InfoExtractor):
|
2014-01-22 10:25:32 +00:00
|
|
|
_VALID_URL = r'https?://(?:(?:www|m)\.)?space\.com/\d+-(?P<title>[^/\.\?]*?)-video\.html'
|
2013-11-06 16:37:39 +00:00
|
|
|
_TEST = {
|
2015-11-14 00:05:46 +00:00
|
|
|
'add_ie': ['BrightcoveLegacy'],
|
2014-02-19 00:04:24 +00:00
|
|
|
'url': 'http://www.space.com/23373-huge-martian-landforms-detail-revealed-by-european-probe-video.html',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '2780937028001',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'Huge Martian Landforms\' Detail Revealed By European Probe | Video',
|
|
|
|
'description': 'md5:db81cf7f3122f95ed234b631a6ea1e61',
|
|
|
|
'uploader': 'TechMedia Networks',
|
2013-11-06 16:37:39 +00:00
|
|
|
},
|
|
|
|
}
|
|
|
|
|
|
|
|
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
|
2015-11-13 23:54:16 +00:00
|
|
|
brightcove_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)
|
2013-11-06 16:37:39 +00:00
|
|
|
if brightcove_url is None:
|
2014-11-26 12:06:02 +00:00
|
|
|
raise ExtractorError(
|
|
|
|
'The webpage does not contain a video', expected=True)
|
2015-11-13 23:54:16 +00:00
|
|
|
return self.url_result(brightcove_url, BrightcoveLegacyIE.ie_key())
|