import os
import re
+import xml.etree.ElementTree
from .common import InfoExtractor
from .youtube import YoutubeIE
'title': 'R\u00e9gis plante sa Jeep',
}
},
- # embedded vimeo video
- {
- 'add_ie': ['Vimeo'],
- 'url': 'http://skillsmatter.com/podcast/home/move-semanticsperfect-forwarding-and-rvalue-references',
- 'file': '22444065.mp4',
- 'md5': '2903896e23df39722c33f015af0666e2',
- 'info_dict': {
- 'title': 'ACCU 2011: Move Semantics,Perfect Forwarding, and Rvalue references- Scott Meyers- 13/04/2011',
- 'uploader_id': 'skillsmatter',
- 'uploader': 'Skills Matter',
- }
- },
# bandcamp page with custom domain
{
'add_ie': ['Bandcamp'],
raise ExtractorError('Invalid URL protocol')
return response
+ def _extract_rss(self, url, video_id, doc):
+ playlist_title = doc.find('./channel/title').text
+ playlist_desc_el = doc.find('./channel/description')
+ playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
+
+ entries = [{
+ '_type': 'url',
+ 'url': e.find('link').text,
+ 'title': e.find('title').text,
+ } for e in doc.findall('./channel/item')]
+
+ return {
+ '_type': 'playlist',
+ 'id': url,
+ 'title': playlist_title,
+ 'description': playlist_desc,
+ 'entries': entries,
+ }
+
def _real_extract(self, url):
parsed_url = compat_urlparse.urlparse(url)
if not parsed_url.scheme:
self.report_extraction(video_id)
+ # Is it an RSS feed?
+ try:
+ doc = xml.etree.ElementTree.fromstring(webpage)
+ if doc.tag == 'rss':
+ return self._extract_rss(url, video_id, doc)
+ except xml.etree.ElementTree.ParseError:
+ pass
+
# it's tempting to parse this further, but you would
# have to take into account all the variations like
# Video Title - Site Name
r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
# Look for BrightCove:
- bc_url = BrightcoveIE._extract_brightcove_url(webpage)
- if bc_url is not None:
+ bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
+ if bc_urls:
self.to_screen('Brightcove video detected.')
- surl = smuggle_url(bc_url, {'Referer': url})
- return self.url_result(surl, 'Brightcove')
+ entries = [{
+ '_type': 'url',
+ 'url': smuggle_url(bc_url, {'Referer': url}),
+ 'ie_key': 'Brightcove'
+ } for bc_url in bc_urls]
+
+ return {
+ '_type': 'playlist',
+ 'title': video_title,
+ 'id': video_id,
+ 'entries': entries,
+ }
# Look for embedded (iframe) Vimeo player
mobj = re.search(