from .youtube import YoutubeIE
from ..compat import (
compat_urllib_parse,
+ compat_urllib_parse_unquote,
+ compat_urllib_request,
compat_urlparse,
compat_xml_parse_error,
)
from .nbc import NBCSportsVPlayerIE
from .ooyala import OoyalaIE
from .rutv import RUTVIE
+from .sportbox import SportBoxEmbedIE
from .smotri import SmotriIE
from .condenast import CondeNastIE
from .udn import UDNEmbedIE
+from .senateisvp import SenateISVPIE
+from .bliptv import BlipTVIE
+from .svt import SVTIE
class GenericIE(InfoExtractor):
'upload_date': '20100513',
}
},
+ # Direct link to a media delivered compressed (requires Accept-Encoding == *)
+ {
+ 'url': 'http://calimero.tk/muzik/FictionJunction-Parallel_Hearts.flac',
+ 'md5': '128c42e68b13950268b648275386fc74',
+ 'info_dict': {
+ 'id': 'FictionJunction-Parallel_Hearts',
+ 'ext': 'flac',
+ 'title': 'FictionJunction-Parallel_Hearts',
+ 'upload_date': '20140522',
+ },
+ 'expected_warnings': [
+ 'URL could be a direct video link, returning it as such.'
+ ]
+ },
# ooyala video
{
'url': 'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219',
'skip_download': True,
},
},
+ # SportBox embed
+ {
+ 'url': 'http://www.vestifinance.ru/articles/25753',
+ 'info_dict': {
+ 'id': '25753',
+ 'title': 'Вести Экономика ― Прямые трансляции с Форума-выставки "Госзаказ-2013"',
+ },
+ 'playlist': [{
+ 'info_dict': {
+ 'id': '370908',
+ 'title': 'Госзаказ. День 3',
+ 'ext': 'mp4',
+ }
+ }, {
+ 'info_dict': {
+ 'id': '370905',
+ 'title': 'Госзаказ. День 2',
+ 'ext': 'mp4',
+ }
+ }, {
+ 'info_dict': {
+ 'id': '370902',
+ 'title': 'Госзаказ. День 1',
+ 'ext': 'mp4',
+ }
+ }],
+ 'params': {
+ # m3u8 download
+ 'skip_download': True,
+ },
+ },
# Embedded TED video
{
'url': 'http://en.support.wordpress.com/videos/ted-talks/',
'info_dict': {
'id': '100183293',
'ext': 'mp4',
- 'title': 'Тайны перевала Дятлова • Тайна перевала Дятлова 1 серия 2 часть',
+ 'title': 'Тайны перевала Дятлова • 1 серия 2 часть',
'description': 'Документальный сериал-расследование одной из самых жутких тайн ХХ века',
'thumbnail': 're:^https?://.*\.jpg$',
'duration': 694,
'title': 'Facebook Creates "On This Day" | Crunch Report',
},
},
+ # SVT embed
+ {
+ 'url': 'http://www.svt.se/sport/ishockey/jagr-tacklar-giroux-under-intervjun',
+ 'info_dict': {
+ 'id': '2900353',
+ 'ext': 'flv',
+ 'title': 'Här trycker Jagr till Giroux (under SVT-intervjun)',
+ 'duration': 27,
+ 'age_limit': 0,
+ },
+ },
# RSS feed with enclosure
{
'url': 'http://podcastfeeds.nbcnews.com/audio/podcast/MSNBC-MADDOW-NETCAST-M4V.xml',
# m3u8 downloads
'skip_download': True,
}
+ },
+ # Contains a SMIL manifest
+ {
+ 'url': 'http://www.telewebion.com/fa/1263668/%D9%82%D8%B1%D8%B9%D9%87%E2%80%8C%DA%A9%D8%B4%DB%8C-%D9%84%DB%8C%DA%AF-%D9%82%D9%87%D8%B1%D9%85%D8%A7%D9%86%D8%A7%D9%86-%D8%A7%D8%B1%D9%88%D9%BE%D8%A7/%2B-%D9%81%D9%88%D8%AA%D8%A8%D8%A7%D9%84.html',
+ 'info_dict': {
+ 'id': 'file',
+ 'ext': 'flv',
+ 'title': '+ Football: Lottery Champions League Europe',
+ 'uploader': 'www.telewebion.com',
+ },
+ 'params': {
+ # rtmpe downloads
+ 'skip_download': True,
+ }
}
]
force_videoid = smuggled_data['force_videoid']
video_id = force_videoid
else:
- video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0]
+ video_id = compat_urllib_parse_unquote(os.path.splitext(url.rstrip('/').split('/')[-1])[0])
self.to_screen('%s: Requesting header' % video_id)
full_response = None
if head_response is False:
- full_response = self._request_webpage(url, video_id)
+ request = compat_urllib_request.Request(url)
+ request.add_header('Accept-Encoding', '*')
+ full_response = self._request_webpage(request, video_id)
head_response = full_response
# Check for direct link to a video
head_response.headers.get('Last-Modified'))
return {
'id': video_id,
- 'title': os.path.splitext(url_basename(url))[0],
+ 'title': compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0]),
'direct': True,
'formats': [{
'format_id': m.group('format_id'),
self._downloader.report_warning('Falling back on generic information extractor.')
if not full_response:
- full_response = self._request_webpage(url, video_id)
+ request = compat_urllib_request.Request(url)
+ # Some webservers may serve compressed content of rather big size (e.g. gzipped flac)
+ # making it impossible to download only chunk of the file (yet we need only 512kB to
+ # test whether it's HTML or not). According to youtube-dl default Accept-Encoding
+ # that will always result in downloading the whole file that is not desirable.
+ # Therefore for extraction pass we have to override Accept-Encoding to any in order
+ # to accept raw bytes and being able to download only a chunk.
+ # It may probably better to solve this by checking Content-Type for application/octet-stream
+ # after HEAD request finishes, but not sure if we can rely on this.
+ request.add_header('Accept-Encoding', '*')
+ full_response = self._request_webpage(request, video_id)
# Maybe it's a direct link to a video?
# Be careful not to download the whole thing!
head_response.headers.get('Last-Modified'))
return {
'id': video_id,
- 'title': os.path.splitext(url_basename(url))[0],
+ 'title': compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0]),
'direct': True,
'url': url,
'upload_date': upload_date,
}
# Look for embedded blip.tv player
- mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
- if mobj:
- return self.url_result('http://blip.tv/a/a-' + mobj.group(1), 'BlipTV')
- mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9_]+)', webpage)
- if mobj:
- return self.url_result(mobj.group(1), 'BlipTV')
+ bliptv_url = BlipTVIE._extract_url(webpage)
+ if bliptv_url:
+ return self.url_result(bliptv_url, 'BlipTV')
+
+ # Look for SVT player
+ svt_url = SVTIE._extract_url(webpage)
+ if svt_url:
+ return self.url_result(svt_url, 'SVT')
# Look for embedded condenast player
matches = re.findall(
if rutv_url:
return self.url_result(rutv_url, 'RUTV')
+ # Look for embedded SportBox player
+ sportbox_urls = SportBoxEmbedIE._extract_urls(webpage)
+ if sportbox_urls:
+ return _playlist_from_matches(sportbox_urls, ie='SportBoxEmbed')
+
# Look for embedded TED player
mobj = re.search(
r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed(?:-ssl)?\.ted\.com/.+?)\1', webpage)
mobj = re.search(
r'<iframe[^>]+?src=(["\'])(?P<url>https?://m(?:lb)?\.mlb\.com/shared/video/embed/embed\.html\?.+?)\1',
webpage)
+ if not mobj:
+ mobj = re.search(
+ r'data-video-link=["\'](?P<url>http://m.mlb.com/video/[^"\']+)',
+ webpage)
if mobj is not None:
return self.url_result(mobj.group('url'), 'MLB')
return self.url_result(
compat_urlparse.urljoin(url, mobj.group('url')), 'UDNEmbed')
+ # Look for Senate ISVP iframe
+ senate_isvp_url = SenateISVPIE._search_iframe_url(webpage)
+ if senate_isvp_url:
+ return self.url_result(senate_isvp_url, 'SenateISVP')
+
def check_video(vurl):
if YoutubeIE.suitable(vurl):
return True
if refresh_header:
found = re.search(REDIRECT_REGEX, refresh_header)
if found:
- new_url = found.group(1)
+ new_url = compat_urlparse.urljoin(url, found.group(1))
self.report_following_redirect(new_url)
return {
'_type': 'url',
# here's a fun little line of code for you:
video_id = os.path.splitext(video_id)[0]
- entries.append({
- 'id': video_id,
- 'url': video_url,
- 'uploader': video_uploader,
- 'title': video_title,
- 'age_limit': age_limit,
- })
+ if determine_ext(video_url) == 'smil':
+ entries.append({
+ 'id': video_id,
+ 'formats': self._extract_smil_formats(video_url, video_id),
+ 'uploader': video_uploader,
+ 'title': video_title,
+ 'age_limit': age_limit,
+ })
+ else:
+ entries.append({
+ 'id': video_id,
+ 'url': video_url,
+ 'uploader': video_uploader,
+ 'title': video_title,
+ 'age_limit': age_limit,
+ })
if len(entries) == 1:
return entries[0]