Merge pull request #12909 from remitamine/raw-sub

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

diff --git a/youtube_dl/extractor/rtlnl.py b/youtube_dl/extractor/rtlnl.py
index 5e916c4ab2d1fe54cf520baf5a063d5bdb849493..3e22998c6d8384893ac801c98c4f6bde6988c2d1 100644 (file)
--- a/youtube_dl/extractor/rtlnl.py
+++ b/youtube_dl/extractor/rtlnl.py
@@ -14,22 +14,22 @@ class RtlNlIE(InfoExtractor):
     _VALID_URL = r'''(?x)
         https?://(?:www\.)?
         (?:
-            rtlxl\.nl/\#!/[^/]+/|
-            rtl\.nl/system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html\b.+?\buuid=
+            rtlxl\.nl/[^\#]*\#!/[^/]+/|
+            rtl\.nl/(?:system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html\b.+?\buuid=|video/)
         )
         (?P<id>[0-9a-f-]+)'''
 
     _TESTS = [{
-        'url': 'http://www.rtlxl.nl/#!/rtl-nieuws-132237/6e4203a6-0a5e-3596-8424-c599a59e0677',
-        'md5': 'cc16baa36a6c169391f0764fa6b16654',
+        'url': 'http://www.rtlxl.nl/#!/rtl-nieuws-132237/82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
+        'md5': '473d1946c1fdd050b2c0161a4b13c373',
         'info_dict': {
-            'id': '6e4203a6-0a5e-3596-8424-c599a59e0677',
+            'id': '82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
             'ext': 'mp4',
-            'title': 'RTL Nieuws - Laat',
-            'description': 'md5:6b61f66510c8889923b11f2778c72dc5',
-            'timestamp': 1408051800,
-            'upload_date': '20140814',
-            'duration': 576.880,
+            'title': 'RTL Nieuws',
+            'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
+            'timestamp': 1461951000,
+            'upload_date': '20160429',
+            'duration': 1167.96,
         },
     }, {
         # best format avaialble a3t
@@ -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 +67,12 @@ 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,
     }]
 
     def _real_extract(self, url):
@@ -125,10 +131,12 @@ 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.