about summary refs log tree commit diff
path: root/youtube_dl/extractor/rtlnl.py
diff options
context:
space:
mode:
authorPhilipp Hagemeister <phihag@phihag.de>2015-02-16 15:45:01 +0100
committerPhilipp Hagemeister <phihag@phihag.de>2015-02-16 15:45:45 +0100
commit59b8ab583491410f4f365b1cb268a4360d7be877 (patch)
tree3f03d472c215225f1c0b589e3fedb73e95f943d4 /youtube_dl/extractor/rtlnl.py
parenta5681804415a9b8d011cd238b7d61e6eb4ec7c4b (diff)
downloadyoutube-dl-59b8ab583491410f4f365b1cb268a4360d7be877.tar.gz
youtube-dl-59b8ab583491410f4f365b1cb268a4360d7be877.tar.xz
youtube-dl-59b8ab583491410f4f365b1cb268a4360d7be877.zip
[rtlnl|generic] Add support for rtl.nl embeds (Fixes #4959)
Diffstat (limited to 'youtube_dl/extractor/rtlnl.py')
-rw-r--r--youtube_dl/extractor/rtlnl.py63
1 files changed, 48 insertions, 15 deletions
diff --git a/youtube_dl/extractor/rtlnl.py b/youtube_dl/extractor/rtlnl.py
index a3ca79f2c..cfce4550a 100644
--- a/youtube_dl/extractor/rtlnl.py
+++ b/youtube_dl/extractor/rtlnl.py
@@ -1,16 +1,25 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
-import re
-
 from .common import InfoExtractor
-from ..utils import parse_duration
+from ..utils import (
+    int_or_none,
+    parse_duration,
+)
 
 
-class RtlXlIE(InfoExtractor):
-    IE_NAME = 'rtlxl.nl'
-    _VALID_URL = r'https?://(www\.)?rtlxl\.nl/#!/[^/]+/(?P<uuid>[^/?]+)'
+class RtlNlIE(InfoExtractor):
+    IE_NAME = 'rtl.nl'
+    IE_DESC = 'rtl.nl and rtlxl.nl'
+    _VALID_URL = r'''(?x)
+        https?://(www\.)?
+        (?:
+            rtlxl\.nl/\#!/[^/]+/|
+            rtl\.nl/system/videoplayer/[^?#]+?/video_embed\.html\#uuid=
+        )
+        (?P<id>[0-9a-f-]+)'''
 
-    _TEST = {
+    _TESTS = [{
         'url': 'http://www.rtlxl.nl/#!/rtl-nieuws-132237/6e4203a6-0a5e-3596-8424-c599a59e0677',
         'md5': 'cc16baa36a6c169391f0764fa6b16654',
         'info_dict': {
@@ -22,21 +31,30 @@ class RtlXlIE(InfoExtractor):
             'upload_date': '20140814',
             'duration': 576.880,
         },
-    }
+    }, {
+        'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed/autoplay=false',
+        'md5': 'dea7474214af1271d91ef332fb8be7ea',
+        'info_dict': {
+            'id': '84ae5571-ac25-4225-ae0c-ef8d9efb2aed',
+            'ext': 'mp4',
+            'timestamp': 1424039400,
+            'title': 'RTL Nieuws - Nieuwe beelden Kopenhagen: chaos direct na aanslag',
+            'thumbnail': 're:^https?://screenshots\.rtl\.nl/system/thumb/sz=[0-9]+x[0-9]+/uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed$',
+            '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.',
+        }
+    }]
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        uuid = mobj.group('uuid')
-
+        uuid = self._match_id(url)
         info = self._download_json(
             'http://www.rtl.nl/system/s4m/vfd/version=2/uuid=%s/fmt=flash/' % uuid,
             uuid)
 
         material = info['material'][0]
-        episode_info = info['episodes'][0]
-
         progname = info['abstracts'][0]['name']
         subtitle = material['title'] or info['episodes'][0]['name']
+        description = material.get('synopsis') or info['episodes'][0]['synopsis']
 
         # Use unencrypted m3u8 streams (See https://github.com/rg3/youtube-dl/issues/4118)
         videopath = material['videopath'].replace('.f4m', '.m3u8')
@@ -58,14 +76,29 @@ class RtlXlIE(InfoExtractor):
                 'quality': 0,
             }
         ])
-
         self._sort_formats(formats)
 
+        thumbnails = []
+        meta = info.get('meta', {})
+        for p in ('poster_base_url', '"thumb_base_url"'):
+            if not meta.get(p):
+                continue
+
+            thumbnails.append({
+                'url': self._proto_relative_url(meta[p] + uuid),
+                'width': int_or_none(self._search_regex(
+                    r'/sz=([0-9]+)', meta[p], 'thumbnail width', fatal=False)),
+                'height': int_or_none(self._search_regex(
+                    r'/sz=[0-9]+x([0-9]+)',
+                    meta[p], 'thumbnail height', fatal=False))
+            })
+
         return {
             'id': uuid,
             'title': '%s - %s' % (progname, subtitle),
             'formats': formats,
             'timestamp': material['original_date'],
-            'description': episode_info['synopsis'],
+            'description': description,
             'duration': parse_duration(material.get('duration')),
+            'thumbnails': thumbnails,
         }