Update upstream source from tag 'upstream/2019.09.01'

[youtubedl] / youtube_dl / extractor / dailymotion.py

diff --git a/youtube_dl/extractor/dailymotion.py b/youtube_dl/extractor/dailymotion.py
index 1816c559e7481b659b5d91fd3944a03eb18f0f9e..745971900b23f1cbf44ce286b9c4a4a7e527a3da 100644 (file)
--- a/youtube_dl/extractor/dailymotion.py
+++ b/youtube_dl/extractor/dailymotion.py
@@ -48,7 +48,14 @@ class DailymotionBaseInfoExtractor(InfoExtractor):
 
 
 class DailymotionIE(DailymotionBaseInfoExtractor):
 
 
 class DailymotionIE(DailymotionBaseInfoExtractor):

-    _VALID_URL = r'(?i)https?://(?:(www|touch)\.)?dailymotion\.[a-z]{2,3}/(?:(?:(?:embed|swf|#)/)?video|swf)/(?P<id>[^/?_]+)'
+    _VALID_URL = r'''(?ix)
+                    https?://
+                        (?:
+                            (?:(?:www|touch)\.)?dailymotion\.[a-z]{2,3}/(?:(?:(?:embed|swf|\#)/)?video|swf)|
+                            (?:www\.)?lequipe\.fr/video
+                        )
+                        /(?P<id>[^/?_]+)
+                    '''

     IE_NAME = 'dailymotion'
 
     _FORMATS = [
     IE_NAME = 'dailymotion'
 
     _FORMATS = [


@@ -133,14 +140,26 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
     }, {
         'url': 'http://www.dailymotion.com/swf/x3ss1m_funny-magic-trick-barry-and-stuart_fun',
         'only_matching': True,
     }, {
         'url': 'http://www.dailymotion.com/swf/x3ss1m_funny-magic-trick-barry-and-stuart_fun',
         'only_matching': True,

+    }, {
+        'url': 'https://www.lequipe.fr/video/x791mem',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.lequipe.fr/video/k7MtHciueyTcrFtFKA2',
+        'only_matching': True,

     }]
 
     @staticmethod
     def _extract_urls(webpage):
     }]
 
     @staticmethod
     def _extract_urls(webpage):

+        urls = []

         # Look for embedded Dailymotion player
         # Look for embedded Dailymotion player

-        matches = re.findall(
-            r'<(?:(?:embed|iframe)[^>]+?src=|input[^>]+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=)(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/(?:embed|swf)/video/.+?)\1', webpage)
-        return list(map(lambda m: unescapeHTML(m[1]), matches))
+        # https://developer.dailymotion.com/player#player-parameters
+        for mobj in re.finditer(
+                r'<(?:(?:embed|iframe)[^>]+?src=|input[^>]+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=)(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/(?:embed|swf)/video/.+?)\1', webpage):
+            urls.append(unescapeHTML(mobj.group('url')))
+        for mobj in re.finditer(
+                r'(?s)DM\.player\([^,]+,\s*{.*?video[\'"]?\s*:\s*["\']?(?P<id>[0-9a-zA-Z]+).+?}\s*\);', webpage):
+            urls.append('https://www.dailymotion.com/embed/video/' + mobj.group('id'))
+        return urls

 
     def _real_extract(self, url):
         video_id = self._match_id(url)
 
     def _real_extract(self, url):
         video_id = self._match_id(url)


@@ -166,11 +185,11 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
             webpage, 'comment count', default=None))
 
         player_v5 = self._search_regex(
             webpage, 'comment count', default=None))
 
         player_v5 = self._search_regex(

-            [r'buildPlayer\(({.+?})\);\n',  # See https://github.com/rg3/youtube-dl/issues/7826
+            [r'buildPlayer\(({.+?})\);\n',  # See https://github.com/ytdl-org/youtube-dl/issues/7826

              r'playerV5\s*=\s*dmp\.create\([^,]+?,\s*({.+?})\);',
              r'buildPlayer\(({.+?})\);',
              r'var\s+config\s*=\s*({.+?});',
              r'playerV5\s*=\s*dmp\.create\([^,]+?,\s*({.+?})\);',
              r'buildPlayer\(({.+?})\);',
              r'var\s+config\s*=\s*({.+?});',

-             # New layout regex (see https://github.com/rg3/youtube-dl/issues/13580)
+             # New layout regex (see https://github.com/ytdl-org/youtube-dl/issues/13580)

              r'__PLAYER_CONFIG__\s*=\s*({.+?});'],
             webpage, 'player v5', default=None)
         if player_v5:
              r'__PLAYER_CONFIG__\s*=\s*({.+?});'],
             webpage, 'player v5', default=None)
         if player_v5: