Merge pull request #7209 from lalinsky/vimeo-new-page
authorSergey M. <dstftw@gmail.com>
Sat, 17 Oct 2015 16:39:17 +0000 (22:39 +0600)
committerSergey M. <dstftw@gmail.com>
Sat, 17 Oct 2015 16:39:17 +0000 (22:39 +0600)
Extract config URL from (new?) React-based Vimeo's page

youtube_dl/extractor/vimeo.py

index fa1b2204946a25f9cd2fb80ee4c7e2cf90779048..88e462a4d02befd492f324709969a5791b4f31a2 100644 (file)
@@ -286,7 +286,14 @@ class VimeoIE(VimeoBaseInfoExtractor):
         try:
             try:
                 config_url = self._html_search_regex(
-                    r' data-config-url="(.+?)"', webpage, 'config URL')
+                    r' data-config-url="(.+?)"', webpage,
+                    'config URL', default=None)
+                if not config_url:
+                    # New react-based page
+                    vimeo_clip_page_config = self._search_regex(
+                        r'vimeo\.clip_page_config\s*=\s*({.+?});', webpage,
+                        'vimeo clip page config')
+                    config_url = self._parse_json(vimeo_clip_page_config, video_id)['player']['config_url']
                 config_json = self._download_webpage(config_url, video_id)
                 config = json.loads(config_json)
             except RegexNotFoundError: