Merge remote-tracking branch 'rg3/master' into universalmusicfrance

test/test_all_urls.py

@@ -99,7 +99,7 @@ class TestAllURLsMatching(unittest.TestCase):
     def test_keywords(self):
         self.assertMatch(':ytsubs', ['youtube:subscriptions'])
         self.assertMatch(':ytsubscriptions', ['youtube:subscriptions'])
-        self.assertMatch(':ythistory', ['youtube:history'])
+        # self.assertMatch(':ythistory', ['youtube:history'])
         self.assertMatch(':thedailyshow', ['ComedyCentralShows'])
         self.assertMatch(':tds', ['ComedyCentralShows'])
 

youtube_dl/extractor/__init__.py

@@ -809,6 +809,8 @@ from .youtube import (
     YoutubeIE,
     YoutubeChannelIE,
     YoutubeFavouritesIE,
+    # disabled because it can wipe the watch history (see #6893)
+    # remember to uncomment test in test/test_all_urls when it's fixed
     #YoutubeHistoryIE,
     YoutubePlaylistIE,
     YoutubeRecommendedIE,

youtube_dl/extractor/nfl.py

@@ -16,12 +16,53 @@ from ..utils import (
 
 class NFLIE(InfoExtractor):
     IE_NAME = 'nfl.com'
-    _VALID_URL = r'''(?x)https?://
+    _VALID_URL = r'''(?x)
-        (?P<host>(?:www\.)?(?:nfl\.com|.*?\.clubs\.nfl\.com))/
+                    https?://
+                        (?P<host>
+                            (?:www\.)?
+                            (?:
+                                (?:
+                                    nfl|
+                                    buffalobills|
+                                    miamidolphins|
+                                    patriots|
+                                    newyorkjets|
+                                    baltimoreravens|
+                                    bengals|
+                                    clevelandbrowns|
+                                    steelers|
+                                    houstontexans|
+                                    colts|
+                                    jaguars|
+                                    titansonline|
+                                    denverbroncos|
+                                    kcchiefs|
+                                    raiders|
+                                    chargers|
+                                    dallascowboys|
+                                    giants|
+                                    philadelphiaeagles|
+                                    redskins|
+                                    chicagobears|
+                                    detroitlions|
+                                    packers|
+                                    vikings|
+                                    atlantafalcons|
+                                    panthers|
+                                    neworleanssaints|
+                                    buccaneers|
+                                    azcardinals|
+                                    stlouisrams|
+                                    49ers|
+                                    seahawks
+                                )\.com|
+                                .+?\.clubs\.nfl\.com
+                            )
+                        )/
                         (?:.+?/)*
-        (?P<id>(?:[a-z0-9]{16}|\w{8}\-(?:\w{4}\-){3}\w{12}))'''
+                        (?P<id>[^/#?&]+)
-    _TESTS = [
+                    '''
-        {
+    _TESTS = [{
         'url': 'http://www.nfl.com/videos/nfl-game-highlights/0ap3000000398478/Week-3-Redskins-vs-Eagles-highlights',
         'md5': '394ef771ddcd1354f665b471d78ec4c6',
         'info_dict': {
@@ -33,8 +74,7 @@ class NFLIE(InfoExtractor):
             'timestamp': 1411337580,
             'thumbnail': 're:^https?://.*\.jpg$',
         }
-        },
+    }, {
-        {
         'url': 'http://prod.www.steelers.clubs.nfl.com/video-and-audio/videos/LIVE_Post_Game_vs_Browns/9d72f26a-9e2b-4718-84d3-09fb4046c266',
         'md5': 'cf85bdb4bc49f6e9d3816d130c78279c',
         'info_dict': {
@@ -46,8 +86,7 @@ class NFLIE(InfoExtractor):
             'timestamp': 1388354455,
             'thumbnail': 're:^https?://.*\.jpg$',
         }
-        },
+    }, {
-        {
         'url': 'http://www.nfl.com/news/story/0ap3000000467586/article/patriots-seahawks-involved-in-lategame-skirmish',
         'info_dict': {
             'id': '0ap3000000467607',
@@ -57,12 +96,24 @@ class NFLIE(InfoExtractor):
             'timestamp': 1422850320,
             'upload_date': '20150202',
         },
+    }, {
+        'url': 'http://www.patriots.com/video/2015/09/18/10-days-gillette',
+        'md5': '4c319e2f625ffd0b481b4382c6fc124c',
+        'info_dict': {
+            'id': 'n-238346',
+            'ext': 'mp4',
+            'title': '10 Days at Gillette',
+            'description': 'md5:8cd9cd48fac16de596eadc0b24add951',
+            'timestamp': 1442618809,
+            'upload_date': '20150918',
         },
-        {
+    }, {
         'url': 'http://www.nfl.com/videos/nfl-network-top-ten/09000d5d810a6bd4/Top-10-Gutsiest-Performances-Jack-Youngblood',
         'only_matching': True,
-        }
+    }, {
-    ]
+        'url': 'http://www.buffalobills.com/video/videos/Rex_Ryan_Show_World_Wide_Rex/b1dcfab2-3190-4bb1-bfc0-d6e603d6601a',
+        'only_matching': True,
+    }]
 
     @staticmethod
     def prepend_host(host, url):
@@ -95,13 +146,14 @@ class NFLIE(InfoExtractor):
         webpage = self._download_webpage(url, video_id)
 
         config_url = NFLIE.prepend_host(host, self._search_regex(
-            r'(?:config|configURL)\s*:\s*"([^"]+)"', webpage, 'config URL',
+            r'(?:(?:config|configURL)\s*:\s*|<nflcs:avplayer[^>]+data-config\s*=\s*)(["\'])(?P<config>.+?)\1',
-            default='static/content/static/config/video/config.json'))
+            webpage, 'config URL', default='static/content/static/config/video/config.json',
+            group='config'))
         # For articles, the id in the url is not the video id
         video_id = self._search_regex(
-            r'contentId\s*:\s*"([^"]+)"', webpage, 'video id', default=video_id)
+            r'(?:<nflcs:avplayer[^>]+data-contentId\s*=\s*|contentId\s*:\s*)(["\'])(?P<id>.+?)\1',
-        config = self._download_json(config_url, video_id,
+            webpage, 'video id', default=video_id, group='id')
-                                     note='Downloading player config')
+        config = self._download_json(config_url, video_id, 'Downloading player config')
         url_template = NFLIE.prepend_host(
             host, '{contentURLTemplate:}'.format(**config))
         video_data = self._download_json(

youtube_dl/utils.py

@@ -619,7 +619,7 @@ def _create_http_connection(ydl_handler, http_class, is_https, *args, **kwargs):
     # expected HTTP responses to meet HTTP/1.0 or later (see also
     # https://github.com/rg3/youtube-dl/issues/6727)
     if sys.version_info < (3, 0):
-        kwargs['strict'] = True
+        kwargs[b'strict'] = True
     hc = http_class(*args, **kwargs)
     source_address = ydl_handler._params.get('source_address')
     if source_address is not None: