]> jfr.im git - yt-dlp.git/blobdiff - yt_dlp/extractor/facebook.py
[extractors] Use new framework for existing embeds (#4307)
[yt-dlp.git] / yt_dlp / extractor / facebook.py
index 5b34f3bff3f55c2d423ffb5b59c4f21769336648..d434b359aec594b154724029ed2fcb7f866f65e6 100644 (file)
@@ -57,6 +57,13 @@ class FacebookIE(InfoExtractor):
                 )
                 (?P<id>[0-9]+)
                 '''
+    _EMBED_REGEX = [
+        r'<iframe[^>]+?src=(["\'])(?P<url>https?://www\.facebook\.com/(?:video/embed|plugins/video\.php).+?)\1',
+        # Facebook API embed https://developers.facebook.com/docs/plugins/embedded-video-player
+        r'''(?x)<div[^>]+
+                class=(?P<q1>[\'"])[^\'"]*\bfb-(?:video|post)\b[^\'"]*(?P=q1)[^>]+
+                data-href=(?P<q2>[\'"])(?P<url>(?:https?:)?//(?:www\.)?facebook.com/.+?)(?P=q2)''',
+    ]
     _LOGIN_URL = 'https://www.facebook.com/login.php?next=http%3A%2F%2Ffacebook.com%2Fhome.php&login_attempt=1'
     _CHECKPOINT_URL = 'https://www.facebook.com/checkpoint/?next=http%3A%2F%2Ffacebook.com%2Fhome.php&_fb_noscript=1'
     _NETRC_MACHINE = 'facebook'
@@ -311,21 +318,6 @@ class FacebookIE(InfoExtractor):
         'graphURI': '/api/graphql/'
     }
 
-    @staticmethod
-    def _extract_urls(webpage):
-        urls = []
-        for mobj in re.finditer(
-                r'<iframe[^>]+?src=(["\'])(?P<url>https?://www\.facebook\.com/(?:video/embed|plugins/video\.php).+?)\1',
-                webpage):
-            urls.append(mobj.group('url'))
-        # Facebook API embed
-        # see https://developers.facebook.com/docs/plugins/embedded-video-player
-        for mobj in re.finditer(r'''(?x)<div[^>]+
-                class=(?P<q1>[\'"])[^\'"]*\bfb-(?:video|post)\b[^\'"]*(?P=q1)[^>]+
-                data-href=(?P<q2>[\'"])(?P<url>(?:https?:)?//(?:www\.)?facebook.com/.+?)(?P=q2)''', webpage):
-            urls.append(mobj.group('url'))
-        return urls
-
     def _perform_login(self, username, password):
         login_page_req = sanitized_Request(self._LOGIN_URL)
         self._set_cookie('facebook.com', 'locale', 'en_US')