from .common import InfoExtractor
from ..utils import (
fix_xml_ampersands,
+ orderedSet,
parse_duration,
qualities,
strip_jsonp,
entries = [
self.url_result('npo:%s' % video_id if not video_id.startswith('http') else video_id)
- for video_id in re.findall(self._PLAYLIST_ENTRY_RE, webpage)
+ for video_id in orderedSet(re.findall(self._PLAYLIST_ENTRY_RE, webpage))
]
playlist_title = self._html_search_regex(
class VPROIE(NPOPlaylistBaseIE):
IE_NAME = 'vpro'
- _VALID_URL = r'https?://(?:www\.)?(?:tegenlicht\.)?vpro\.nl/(?:[^/]+/){2,}(?P<id>[^/]+)\.html'
- _PLAYLIST_TITLE_RE = r'<h1[^>]+class=["\'].*?\bmedia-platform-title\b.*?["\'][^>]*>([^<]+)'
+ _VALID_URL = r'https?://(?:www\.)?(?:(?:tegenlicht\.)?vpro|2doc)\.nl/(?:[^/]+/)*(?P<id>[^/]+)\.html'
+ _PLAYLIST_TITLE_RE = (r'<h1[^>]+class=["\'].*?\bmedia-platform-title\b.*?["\'][^>]*>([^<]+)',
+ r'<h5[^>]+class=["\'].*?\bmedia-platform-subtitle\b.*?["\'][^>]*>([^<]+)')
_PLAYLIST_ENTRY_RE = r'data-media-id="([^"]+)"'
_TESTS = [
'title': 'education education',
},
'playlist_count': 2,
+ },
+ {
+ 'url': 'http://www.2doc.nl/documentaires/series/2doc/2015/oktober/de-tegenprestatie.html',
+ 'info_dict': {
+ 'id': 'de-tegenprestatie',
+ 'title': 'De Tegenprestatie',
+ },
+ 'playlist_count': 2,
+ }, {
+ 'url': 'http://www.2doc.nl/speel~VARA_101375237~mh17-het-verdriet-van-nederland~.html',
+ 'info_dict': {
+ 'id': 'VARA_101375237',
+ 'ext': 'm4v',
+ 'title': 'MH17: Het verdriet van Nederland',
+ 'description': 'md5:09e1a37c1fdb144621e22479691a9f18',
+ 'upload_date': '20150716',
+ },
+ 'params': {
+ # Skip because of m3u8 download
+ 'skip_download': True
+ },
}
]
class WNLIE(NPOPlaylistBaseIE):
+ IE_NAME = 'wnl'
_VALID_URL = r'https?://(?:www\.)?omroepwnl\.nl/video/detail/(?P<id>[^/]+)__\d+'
_PLAYLIST_TITLE_RE = r'(?s)<h1[^>]+class="subject"[^>]*>(.+?)</h1>'
_PLAYLIST_ENTRY_RE = r'<a[^>]+href="([^"]+)"[^>]+class="js-mid"[^>]*>Deel \d+'
},
'playlist_count': 4,
}]
+
+
+class AndereTijdenIE(NPOPlaylistBaseIE):
+ IE_NAME = 'anderetijden'
+ _VALID_URL = r'https?://(?:www\.)?anderetijden\.nl/programma/(?:[^/]+/)+(?P<id>[^/?#&]+)'
+ _PLAYLIST_TITLE_RE = r'(?s)<h1[^>]+class=["\'].*?\bpage-title\b.*?["\'][^>]*>(.+?)</h1>'
+ _PLAYLIST_ENTRY_RE = r'<figure[^>]+class=["\']episode-container episode-page["\'][^>]+data-prid=["\'](.+?)["\']'
+
+ _TESTS = [{
+ 'url': 'http://anderetijden.nl/programma/1/Andere-Tijden/aflevering/676/Duitse-soldaten-over-de-Slag-bij-Arnhem',
+ 'info_dict': {
+ 'id': 'Duitse-soldaten-over-de-Slag-bij-Arnhem',
+ 'title': 'Duitse soldaten over de Slag bij Arnhem',
+ },
+ 'playlist_count': 3,
+ }]