import os
import re
-import xml.etree.ElementTree
from .common import InfoExtractor
from .youtube import YoutubeIE
ExtractorError,
HEADRequest,
+ parse_xml,
smuggle_url,
unescapeHTML,
unified_strdate,
)
from .brightcove import BrightcoveIE
from .ooyala import OoyalaIE
+from .rutv import RUTVIE
class GenericIE(InfoExtractor):
'info_dict': {
'id': '9ODmcdjQcHQ',
'ext': 'mp4',
+ 'title': 'Tested: Grinding Coffee at 2000 Frames Per Second',
+ 'upload_date': '20140225',
+ 'description': 'md5:06a40fbf30b220468f1e0957c0f558ff',
+ 'uploader': 'Tested',
+ 'uploader_id': 'testedcom',
},
# No need to test YoutubeIE here
'params': {
'skip_download': True,
},
},
+ # funnyordie embed
+ {
+ 'url': 'http://www.theguardian.com/world/2014/mar/11/obama-zach-galifianakis-between-two-ferns',
+ 'md5': '7cf780be104d40fea7bae52eed4a470e',
+ 'info_dict': {
+ 'id': '18e820ec3f',
+ 'ext': 'mp4',
+ 'title': 'Between Two Ferns with Zach Galifianakis: President Barack Obama',
+ 'description': 'Episode 18: President Barack Obama sits down with Zach Galifianakis for his most memorable interview yet.',
+ },
+ },
+ # RUTV embed
+ {
+ 'url': 'http://www.rg.ru/2014/03/15/reg-dfo/anklav-anons.html',
+ 'info_dict': {
+ 'id': '776940',
+ 'ext': 'mp4',
+ 'title': 'Охотское море стало целиком российским',
+ 'description': 'md5:5ed62483b14663e2a95ebbe115eb8f43',
+ },
+ 'params': {
+ # m3u8 download
+ 'skip_download': True,
+ },
+ }
]
def report_download_webpage(self, video_id):
newurl = newurl.replace(' ', '%20')
newheaders = dict((k,v) for k,v in req.headers.items()
if k.lower() not in ("content-length", "content-type"))
+ try:
+ # This function was deprecated in python 3.3 and removed in 3.4
+ origin_req_host = req.get_origin_req_host()
+ except AttributeError:
+ origin_req_host = req.origin_req_host
return HEADRequest(newurl,
headers=newheaders,
- origin_req_host=req.get_origin_req_host(),
+ origin_req_host=origin_req_host,
unverifiable=True)
else:
raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
# Is it an RSS feed?
try:
- doc = xml.etree.ElementTree.fromstring(webpage.encode('utf-8'))
+ doc = parse_xml(webpage)
if doc.tag == 'rss':
return self._extract_rss(url, video_id, doc)
except compat_xml_parse_error:
# Look for embedded (iframe) Vimeo player
mobj = re.search(
- r'<iframe[^>]+?src="((?:https?:)?//player\.vimeo\.com/video/.+?)"', webpage)
+ r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//player\.vimeo\.com/video/.+?)\1', webpage)
if mobj:
- player_url = unescapeHTML(mobj.group(1))
+ player_url = unescapeHTML(mobj.group('url'))
surl = smuggle_url(player_url, {'Referer': url})
return self.url_result(surl, 'Vimeo')
if mobj is not None:
return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
+ # Look for funnyordie embed
+ matches = re.findall(r'<iframe[^>]+?src="(https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"', webpage)
+ if matches:
+ urlrs = [self.url_result(unescapeHTML(eurl), 'FunnyOrDie')
+ for eurl in matches]
+ return self.playlist_result(
+ urlrs, playlist_id=video_id, playlist_title=video_title)
+
+ # Look for embedded RUTV player
+ rutv_url = RUTVIE._extract_url(webpage)
+ if rutv_url:
+ return self.url_result(rutv_url, 'RUTV')
+
# Start with something easy: JW Player in SWFObject
mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
if mobj is None: