[npo] Add support for 2doc.nl (Closes #10842)

[youtube-dl] / youtube_dl / extractor / npo.py

diff --git a/youtube_dl/extractor/npo.py b/youtube_dl/extractor/npo.py
index f95867d589feddf5df3f34366d37b76f859efa41..c3915ec6e642f6952cd6e478db6b27b1f684c003 100644 (file)
--- a/youtube_dl/extractor/npo.py
+++ b/youtube_dl/extractor/npo.py
@@ -5,6 +5,7 @@ import re
 from .common import InfoExtractor
 from ..utils import (
     fix_xml_ampersands,
+    orderedSet,
     parse_duration,
     qualities,
     strip_jsonp,
@@ -446,7 +447,7 @@ class NPOPlaylistBaseIE(NPOIE):
 
         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(
@@ -458,8 +459,9 @@ class NPOPlaylistBaseIE(NPOIE):
 
 class VPROIE(NPOPlaylistBaseIE):
     IE_NAME = 'vpro'
-    _VALID_URL = r'https?://(?:www\.)?(?:tegenlicht\.)?vpro\.nl/(?:[^/]+/){2,}(?P<id>[^/]+)\.html'
-    _PLAYLIST_TITLE_RE = r'<title>\s*([^>]+?)\s*-\s*Teledoc\s*-\s*VPRO\s*</title>'
+    _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 = [
@@ -473,12 +475,13 @@ class VPROIE(NPOPlaylistBaseIE):
                 'description': 'md5:52cf4eefbc96fffcbdc06d024147abea',
                 'upload_date': '20130225',
             },
+            'skip': 'Video gone',
         },
         {
             'url': 'http://www.vpro.nl/programmas/2doc/2015/sergio-herman.html',
             'info_dict': {
                 'id': 'sergio-herman',
-                'title': 'Sergio Herman: Fucking perfect',
+                'title': 'sergio herman: fucking perfect',
             },
             'playlist_count': 2,
         },
@@ -487,14 +490,36 @@ class VPROIE(NPOPlaylistBaseIE):
             'url': 'http://www.vpro.nl/programmas/2doc/2015/education-education.html',
             'info_dict': {
                 'id': 'education-education',
-                'title': '2Doc',
+                '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+'
@@ -507,3 +532,19 @@ class WNLIE(NPOPlaylistBaseIE):
         },
         '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,
+    }]