[tvp] Fix description extraction, make thumbnail optional and fix tests

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

diff --git a/youtube_dl/extractor/tvp.py b/youtube_dl/extractor/tvp.py
index 06ea2b40a759158baa2c561498e5155011f418ec..f9bf600b05b2312673f013a38154a4f33e3e75de 100644 (file)
--- a/youtube_dl/extractor/tvp.py
+++ b/youtube_dl/extractor/tvp.py
@@ -15,16 +15,16 @@ from ..utils import (
 class TVPIE(InfoExtractor):
     IE_NAME = 'tvp'
     IE_DESC = 'Telewizja Polska'
-    _VALID_URL = r'https?://[^/]+\.tvp\.(?:pl|info)/(?:(?!\d+/)[^/]+/)*(?P<id>\d+)'
+    _VALID_URL = r'https?://[^/]+\.tvp\.(?:pl|info)/(?:video/(?:[^,\s]*,)*|(?:(?!\d+/)[^/]+/)*)(?P<id>\d+)'
 
     _TESTS = [{
-        'url': 'http://vod.tvp.pl/194536/i-seria-odc-13',
-        'md5': '8aa518c15e5cc32dfe8db400dc921fbb',
+        'url': 'https://vod.tvp.pl/video/czas-honoru,i-seria-odc-13,194536',
+        'md5': 'a21eb0aa862f25414430f15fdfb9e76c',
         'info_dict': {
             'id': '194536',
             'ext': 'mp4',
-            'title': 'Czas honoru, I seria – odc. 13',
-            'description': 'md5:76649d2014f65c99477be17f23a4dead',
+            'title': 'Czas honoru, odc. 13 – Władek',
+            'description': 'md5:437f48b93558370b031740546b696e24',
         },
     }, {
         'url': 'http://www.tvp.pl/there-can-be-anything-so-i-shortened-it/17916176',
@@ -37,13 +37,15 @@ class TVPIE(InfoExtractor):
         },
     }, {
         # page id is not the same as video id(#7799)
-        'url': 'http://vod.tvp.pl/22704887/08122015-1500',
-        'md5': 'cf6a4705dfd1489aef8deb168d6ba742',
+        'url': 'https://wiadomosci.tvp.pl/33908820/28092017-1930',
+        'md5': '84cd3c8aec4840046e5ab712416b73d0',
         'info_dict': {
-            'id': '22680786',
+            'id': '33908820',
             'ext': 'mp4',
-            'title': 'Wiadomości, 08.12.2015, 15:00',
+            'title': 'Wiadomości, 28.09.2017, 19:30',
+            'description': 'Wydanie główne codziennego serwisu informacyjnego.'
         },
+        'skip': 'HTTP Error 404: Not Found',
     }, {
         'url': 'http://vod.tvp.pl/seriale/obyczajowe/na-sygnale/sezon-2-27-/odc-39/17834272',
         'only_matching': True,
@@ -74,8 +76,10 @@ class TVPIE(InfoExtractor):
         return {
             '_type': 'url_transparent',
             'url': 'tvp:' + video_id,
-            'description': self._og_search_description(webpage, default=None),
-            'thumbnail': self._og_search_thumbnail(webpage),
+            'description': self._og_search_description(
+                webpage, default=None) or self._html_search_meta(
+                'description', webpage, default=None),
+            'thumbnail': self._og_search_thumbnail(webpage, default=None),
             'ie_key': 'TVPEmbed',
         }
 
@@ -86,6 +90,14 @@ class TVPEmbedIE(InfoExtractor):
     _VALID_URL = r'(?:tvp:|https?://[^/]+\.tvp\.(?:pl|info)/sess/tvplayer\.php\?.*?object_id=)(?P<id>\d+)'
 
     _TESTS = [{
+        'url': 'tvp:194536',
+        'md5': 'a21eb0aa862f25414430f15fdfb9e76c',
+        'info_dict': {
+            'id': '194536',
+            'ext': 'mp4',
+            'title': 'Czas honoru, odc. 13 – Władek',
+        },
+    }, {
         'url': 'http://www.tvp.pl/sess/tvplayer.php?object_id=22670268',
         'md5': '8c9cd59d16edabf39331f93bf8a766c7',
         'info_dict': {
@@ -150,8 +162,7 @@ class TVPEmbedIE(InfoExtractor):
                 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
             self._sort_formats(m3u8_formats)
             m3u8_formats = list(filter(
-                lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
-                m3u8_formats))
+                lambda f: f.get('vcodec') != 'none', m3u8_formats))
             formats.extend(m3u8_formats)
             for i, m3u8_format in enumerate(m3u8_formats, 2):
                 http_url = '%s-%d.mp4' % (video_url_base, i)