from .common import InfoExtractor
from ..utils import (
compat_urllib_parse,
+ compat_urllib_request,
ExtractorError,
+ find_xpath_attr,
fix_xml_ampersands,
+ HEADRequest,
+ unescapeHTML,
url_basename,
RegexNotFoundError,
)
+
def _media_xml_tag(tag):
return '{http://search.yahoo.com/mrss/}%s' % tag
class MTVServicesInfoExtractor(InfoExtractor):
+ _MOBILE_TEMPLATE = None
@staticmethod
def _id_from_uri(uri):
return uri.split(':')[-1]
else:
return thumb_node.attrib['url']
- def _extract_video_formats(self, mdoc):
- if re.match(r'.*/error_country_block\.swf$', mdoc.find('.//src').text) is not None:
- raise ExtractorError('This video is not available from your country.', expected=True)
+ def _extract_mobile_video_formats(self, mtvn_id):
+ webpage_url = self._MOBILE_TEMPLATE % mtvn_id
+ req = compat_urllib_request.Request(webpage_url)
+ # Otherwise we get a webpage that would execute some javascript
+ req.add_header('Youtubedl-user-agent', 'curl/7')
+ webpage = self._download_webpage(req, mtvn_id,
+ 'Downloading mobile page')
+ metrics_url = unescapeHTML(self._search_regex(r'<a href="(http://metrics.+?)"', webpage, 'url'))
+ req = HEADRequest(metrics_url)
+ response = self._request_webpage(req, mtvn_id, 'Resolving url')
+ url = response.geturl()
+ # Transform the url to get the best quality:
+ url = re.sub(r'.+pxE=mp4', 'http://mtvnmobile.vo.llnwd.net/kip0/_pxn=0+_pxK=18639+_pxE=mp4', url, 1)
+ return [{'url': url,'ext': 'mp4'}]
+
+ def _extract_video_formats(self, mdoc, mtvn_id):
+ if re.match(r'.*/(error_country_block\.swf|geoblock\.mp4)$', mdoc.find('.//src').text) is not None:
+ if mtvn_id is not None and self._MOBILE_TEMPLATE is not None:
+ self.to_screen('The normal version is not available from your '
+ 'country, trying with the mobile version')
+ return self._extract_mobile_video_formats(mtvn_id)
+ raise ExtractorError('This video is not available from your country.',
+ expected=True)
formats = []
for rendition in mdoc.findall('.//rendition'):
mediagen_url = re.sub(r'&[^=]*?={.*?}(?=(&|$))', '', mediagen_url)
if 'acceptMethods' not in mediagen_url:
mediagen_url += '&acceptMethods=fms'
+
mediagen_doc = self._download_xml(mediagen_url, video_id,
'Downloading video urls')
else:
description = None
+ title_el = None
+ if title_el is None:
+ title_el = find_xpath_attr(
+ itemdoc, './/{http://search.yahoo.com/mrss/}category',
+ 'scheme', 'urn:mtvn:video_title')
+ if title_el is None:
+ title_el = itemdoc.find('.//{http://search.yahoo.com/mrss/}title')
+ if title_el is None:
+ title_el = itemdoc.find('.//title')
+ if title_el.text is None:
+ title_el = None
+
+ title = title_el.text
+ if title is None:
+ raise ExtractorError('Could not find video title')
+ title = title.strip()
+
+ # This a short id that's used in the webpage urls
+ mtvn_id = None
+ mtvn_id_node = find_xpath_attr(itemdoc, './/{http://search.yahoo.com/mrss/}category',
+ 'scheme', 'urn:mtvn:id')
+ if mtvn_id_node is not None:
+ mtvn_id = mtvn_id_node.text
+
return {
- 'title': itemdoc.find('title').text,
- 'formats': self._extract_video_formats(mediagen_doc),
+ 'title': title,
+ 'formats': self._extract_video_formats(mediagen_doc, mtvn_id),
'id': video_id,
'thumbnail': self._get_thumbnail_url(uri, itemdoc),
'description': description,
title = url_basename(url)
webpage = self._download_webpage(url, title)
try:
- # the url is in the format http://media.mtvnservices.com/fb/{mgid}.swf
- fb_url = self._og_search_video_url(webpage)
- mgid = url_basename(fb_url).rpartition('.')[0]
+ # the url can be http://media.mtvnservices.com/fb/{mgid}.swf
+ # or http://media.mtvnservices.com/{mgid}
+ og_url = self._og_search_video_url(webpage)
+ mgid = url_basename(og_url)
+ if mgid.endswith('.swf'):
+ mgid = mgid[:-4]
except RegexNotFoundError:
- mgid = self._search_regex(r'data-mgid="(.*?)"', webpage, u'mgid')
+ mgid = self._search_regex(
+ [r'data-mgid="(.*?)"', r'swfobject.embedSWF\(".*?(mgid:.*?)"'],
+ webpage, u'mgid')
return self._get_videos_info(mgid)
'info_dict': {
'id': '984696',
'ext': 'mp4',
- 'title': 'Short',
+ 'title': 'Arcade Fire: Behind the Scenes at the Biggest Music Experiment Yet',
}
}
_FEED_URL = 'http://all.mtvworldverticals.com/feed-xml/'