[rtlnl] add support for another type of embeds

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

diff --git a/youtube_dl/extractor/rtlnl.py b/youtube_dl/extractor/rtlnl.py
index 8598b5840a7e2c2af2449da12ef8b2445b03830e..bba25a233e3fef88ae8e06e504e4524e64c3bd1c 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)
-        https?://(?:www\.)?
+        https?://(?:(?:www|static)\.)?
         (?:
-            rtlxl\.nl/\#!/[^/]+/|
-            rtl\.nl/system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html\b.+?\buuid=
+            rtlxl\.nl/[^\#]*\#!/[^/]+/|
+            rtl\.nl/(?:(?:system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html|embed)\b.+?\buuid=|video/)
         )
         (?P<id>[0-9a-f-]+)'''
 
@@ -40,7 +40,7 @@ class RtlNlIE(InfoExtractor):
             'ext': 'mp4',
             'timestamp': 1424039400,
             'title': 'RTL Nieuws - Nieuwe beelden Kopenhagen: chaos direct na aanslag',
-            'thumbnail': 're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed$',
+            'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed$',
             '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.',
         }
@@ -52,7 +52,7 @@ class RtlNlIE(InfoExtractor):
             'id': 'f536aac0-1dc3-4314-920e-3bd1c5b3811a',
             'ext': 'mp4',
             'title': 'RTL Nieuws - Meer beelden van overval juwelier',
-            'thumbnail': 're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a$',
+            'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a$',
             'timestamp': 1437233400,
             'upload_date': '20150718',
             'duration': 30.474,
@@ -67,6 +67,15 @@ class RtlNlIE(InfoExtractor):
     }, {
         'url': 'http://www.rtl.nl/system/videoplayer/derden/embed.html#!/uuid=bb0353b0-d6a4-1dad-90e9-18fe75b8d1f0',
         'only_matching': True,
+    }, {
+        'url': 'http://rtlxl.nl/?_ga=1.204735956.572365465.1466978370#!/rtl-nieuws-132237/3c487912-023b-49ac-903e-2c5d79f8410f',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.rtl.nl/video/c603c9c2-601d-4b5e-8175-64f1e942dc7d/',
+        'only_matching': True,
+    }, {
+        'url': 'https://static.rtl.nl/embed/?uuid=1a2970fc-5c0b-43ff-9fdc-927e39e6d1bc&autoplay=false&publicatiepunt=rtlnieuwsnl',
+        'only_matching': True,
     }]
 
     def _real_extract(self, url):
@@ -125,10 +134,12 @@ class RtlNlIE(InfoExtractor):
                 try:
                     # Find hls format with the same width and height corresponding
                     # to progressive format and copy metadata from it.
-                    f = next(f for f in formats
-                             if f.get('width') == width and f.get('height') == height).copy()
-                    f.update(pg_format(format_id, width, height))
-                    pg_formats.append(f)
+                    f = next(f for f in formats if f.get('height') == height)
+                    # hls formats may have invalid width
+                    f['width'] = width
+                    f_copy = f.copy()
+                    f_copy.update(pg_format(format_id, width, height))
+                    pg_formats.append(f_copy)
                 except StopIteration:
                     # Missing hls format does mean that no progressive format with
                     # such width and height exists either.