Update upstream source from tag 'upstream/2019.09.28'

[youtubedl] / youtube_dl / extractor / rmcdecouverte.py

diff --git a/youtube_dl/extractor/rmcdecouverte.py b/youtube_dl/extractor/rmcdecouverte.py
index 2340dae535b3ed719ffd363283a1c542a2da80b1..c3623edcc7adc3fad26d769d79c7088fc98c2593 100644 (file)
--- a/youtube_dl/extractor/rmcdecouverte.py
+++ b/youtube_dl/extractor/rmcdecouverte.py
@@ -1,39 +1,55 @@
 # coding: utf-8
 from __future__ import unicode_literals
 
 # coding: utf-8
 from __future__ import unicode_literals
 

+import re
+

 from .common import InfoExtractor
 from .brightcove import BrightcoveLegacyIE
 from ..compat import (
     compat_parse_qs,
     compat_urlparse,
 )
 from .common import InfoExtractor
 from .brightcove import BrightcoveLegacyIE
 from ..compat import (
     compat_parse_qs,
     compat_urlparse,
 )

+from ..utils import smuggle_url

 
 
 class RMCDecouverteIE(InfoExtractor):
 
 
 class RMCDecouverteIE(InfoExtractor):

-    _VALID_URL = r'https?://rmcdecouverte\.bfmtv\.com/mediaplayer-replay.*?\bid=(?P<id>\d+)'
+    _VALID_URL = r'https?://rmcdecouverte\.bfmtv\.com/(?:(?:[^/]+/)*program_(?P<id>\d+)|(?P<live_id>mediaplayer-direct))'

 
 

-    _TEST = {
-        'url': 'http://rmcdecouverte.bfmtv.com/mediaplayer-replay/?id=1430&title=LES%20HEROS%20DU%2088e%20ETAGE',
+    _TESTS = [{
+        'url': 'https://rmcdecouverte.bfmtv.com/wheeler-dealers-occasions-a-saisir/program_2566/',

         'info_dict': {
         'info_dict': {

-            'id': '5111223049001',
+            'id': '5983675500001',

             'ext': 'mp4',
             'ext': 'mp4',

-            'title': ': LES HEROS DU 88e ETAGE',
-            'description': 'Découvrez comment la bravoure de deux hommes dans la Tour Nord du World Trade Center a sauvé  la vie d\'innombrables personnes le 11 septembre 2001.',
+            'title': 'CORVETTE',
+            'description': 'md5:c1e8295521e45ffebf635d6a7658f506',

             'uploader_id': '1969646226001',
             'uploader_id': '1969646226001',

-            'upload_date': '20160904',
-            'timestamp': 1472951103,
+            'upload_date': '20181226',
+            'timestamp': 1545861635,

         },
         'params': {
         },
         'params': {

-            # rtmp download

             'skip_download': True,
         },
             'skip_download': True,
         },

-        'skip': 'Only works from France',
-    }
+        'skip': 'only available for a week',
+    }, {
+        # live, geo restricted, bypassable
+        'url': 'https://rmcdecouverte.bfmtv.com/mediaplayer-direct/',
+        'only_matching': True,
+    }]

     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1969646226001/default_default/index.html?videoId=%s'
 
     def _real_extract(self, url):
     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1969646226001/default_default/index.html?videoId=%s'
 
     def _real_extract(self, url):

-        video_id = self._match_id(url)
-        webpage = self._download_webpage(url, video_id)
+        mobj = re.match(self._VALID_URL, url)
+        display_id = mobj.group('id') or mobj.group('live_id')
+        webpage = self._download_webpage(url, display_id)

         brightcove_legacy_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)
         brightcove_legacy_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)

-        brightcove_id = compat_parse_qs(compat_urlparse.urlparse(brightcove_legacy_url).query)['@videoPlayer'][0]
-        return self.url_result(self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id, 'BrightcoveNew', brightcove_id)
+        if brightcove_legacy_url:
+            brightcove_id = compat_parse_qs(compat_urlparse.urlparse(
+                brightcove_legacy_url).query)['@videoPlayer'][0]
+        else:
+            brightcove_id = self._search_regex(
+                r'data-video-id=["\'](\d+)', webpage, 'brightcove id')
+        return self.url_result(
+            smuggle_url(
+                self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
+                {'geo_countries': ['FR']}),
+            'BrightcoveNew', brightcove_id)