Register new version in the changelog.

[youtubedl] / youtube_dl / extractor / rtlnl.py

diff --git a/youtube_dl/extractor/rtlnl.py b/youtube_dl/extractor/rtlnl.py
index cfce4550ada568cfe13fae859a2bb745671074b5..543d94417f6d52f9cb5f4bd4508d5bcf3d4b984e 100644 (file)
--- a/youtube_dl/extractor/rtlnl.py
+++ b/youtube_dl/extractor/rtlnl.py
@@ -12,10 +12,10 @@ class RtlNlIE(InfoExtractor):
     IE_NAME = 'rtl.nl'
     IE_DESC = 'rtl.nl and rtlxl.nl'
     _VALID_URL = r'''(?x)
     IE_NAME = 'rtl.nl'
     IE_DESC = 'rtl.nl and rtlxl.nl'
     _VALID_URL = r'''(?x)

-        https?://(www\.)?
+        https?://(?:www\.)?

         (?:
             rtlxl\.nl/\#!/[^/]+/|
         (?:
             rtlxl\.nl/\#!/[^/]+/|

-            rtl\.nl/system/videoplayer/[^?#]+?/video_embed\.html\#uuid=
+            rtl\.nl/system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html\b.+?\buuid=

         )
         (?P<id>[0-9a-f-]+)'''
 
         )
         (?P<id>[0-9a-f-]+)'''
 


@@ -43,26 +43,60 @@ class RtlNlIE(InfoExtractor):
             'upload_date': '20150215',
             'description': 'Er zijn nieuwe beelden vrijgegeven die vlak na de aanslag in Kopenhagen zijn gemaakt. Op de video is goed te zien hoe omstanders zich bekommeren om één van de slachtoffers, terwijl de eerste agenten ter plaatse komen.',
         }
             'upload_date': '20150215',
             'description': 'Er zijn nieuwe beelden vrijgegeven die vlak na de aanslag in Kopenhagen zijn gemaakt. Op de video is goed te zien hoe omstanders zich bekommeren om één van de slachtoffers, terwijl de eerste agenten ter plaatse komen.',
         }

+    }, {
+        # empty synopsis and missing episodes (see https://github.com/rg3/youtube-dl/issues/6275)
+        'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a/autoplay=false',
+        'info_dict': {
+            'id': 'f536aac0-1dc3-4314-920e-3bd1c5b3811a',
+            'ext': 'mp4',
+            'title': 'RTL Nieuws - Meer beelden van overval juwelier',
+            'thumbnail': 're:^https?://screenshots\.rtl\.nl/system/thumb/sz=[0-9]+x[0-9]+/uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a$',
+            'timestamp': 1437233400,
+            'upload_date': '20150718',
+            'duration': 30.474,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        # encrypted m3u8 streams, georestricted
+        'url': 'http://www.rtlxl.nl/#!/afl-2-257632/52a74543-c504-4cde-8aa8-ec66fe8d68a7',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.rtl.nl/system/videoplayer/derden/embed.html#!/uuid=bb0353b0-d6a4-1dad-90e9-18fe75b8d1f0',
+        'only_matching': True,

     }]
 
     def _real_extract(self, url):
         uuid = self._match_id(url)
         info = self._download_json(
     }]
 
     def _real_extract(self, url):
         uuid = self._match_id(url)
         info = self._download_json(

-            'http://www.rtl.nl/system/s4m/vfd/version=2/uuid=%s/fmt=flash/' % uuid,
+            'http://www.rtl.nl/system/s4m/vfd/version=2/uuid=%s/fmt=adaptive/' % uuid,

             uuid)
 
         material = info['material'][0]
             uuid)
 
         material = info['material'][0]

-        progname = info['abstracts'][0]['name']
-        subtitle = material['title'] or info['episodes'][0]['name']
-        description = material.get('synopsis') or info['episodes'][0]['synopsis']
+        title = info['abstracts'][0]['name']
+        subtitle = material.get('title')
+        if subtitle:
+            title += ' - %s' % subtitle
+        description = material.get('synopsis')
+
+        meta = info.get('meta', {})

 
 

-        # Use unencrypted m3u8 streams (See https://github.com/rg3/youtube-dl/issues/4118)
-        videopath = material['videopath'].replace('.f4m', '.m3u8')
-        m3u8_url = 'http://manifest.us.rtl.nl' + videopath
+        # m3u8 streams are encrypted and may not be handled properly by older ffmpeg/avconv.
+        # To workaround this previously adaptive -> flash trick was used to obtain
+        # unencrypted m3u8 streams (see https://github.com/rg3/youtube-dl/issues/4118)
+        # and bypass georestrictions as well.
+        # Currently, unencrypted m3u8 playlists are (intentionally?) invalid and therefore
+        # unusable albeit can be fixed by simple string replacement (see
+        # https://github.com/rg3/youtube-dl/pull/6337)
+        # Since recent ffmpeg and avconv handle encrypted streams just fine encrypted
+        # streams are used now.
+        videopath = material['videopath']
+        m3u8_url = meta.get('videohost', 'http://manifest.us.rtl.nl') + videopath

 
         formats = self._extract_m3u8_formats(m3u8_url, uuid, ext='mp4')
 
 
         formats = self._extract_m3u8_formats(m3u8_url, uuid, ext='mp4')
 

-        video_urlpart = videopath.split('/flash/')[1][:-5]
+        video_urlpart = videopath.split('/adaptive/')[1][:-5]

         PG_URL_TEMPLATE = 'http://pg.us.rtl.nl/rtlxl/network/%s/progressive/%s.mp4'
 
         formats.extend([
         PG_URL_TEMPLATE = 'http://pg.us.rtl.nl/rtlxl/network/%s/progressive/%s.mp4'
 
         formats.extend([


@@ -79,7 +113,7 @@ class RtlNlIE(InfoExtractor):
         self._sort_formats(formats)
 
         thumbnails = []
         self._sort_formats(formats)
 
         thumbnails = []

-        meta = info.get('meta', {})
+

         for p in ('poster_base_url', '"thumb_base_url"'):
             if not meta.get(p):
                 continue
         for p in ('poster_base_url', '"thumb_base_url"'):
             if not meta.get(p):
                 continue


@@ -95,7 +129,7 @@ class RtlNlIE(InfoExtractor):
 
         return {
             'id': uuid,
 
         return {
             'id': uuid,

-            'title': '%s - %s' % (progname, subtitle),
+            'title': title,

             'formats': formats,
             'timestamp': material['original_date'],
             'description': description,
             'formats': formats,
             'timestamp': material['original_date'],
             'description': description,