Merge branch 'pr-democracynow' of https://github.com/atomicdryad/youtube-dl into...

[youtube-dl] / youtube_dl / extractor / onionstudios.py

diff --git a/youtube_dl/extractor/onionstudios.py b/youtube_dl/extractor/onionstudios.py
index d5d03fd443e2c68d7f71b41239638110bea4379a..0f1f448fe3126670932b371498b73b0bf0a7924e 100644 (file)
--- a/youtube_dl/extractor/onionstudios.py
+++ b/youtube_dl/extractor/onionstudios.py
@@ -27,6 +27,13 @@ class OnionStudiosIE(InfoExtractor):
         'only_matching': True,
     }]
 
+    @staticmethod
+    def _extract_url(webpage):
+        mobj = re.search(
+            r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?onionstudios\.com/embed.+?)\1', webpage)
+        if mobj:
+            return mobj.group('url')
+
     def _real_extract(self, url):
         video_id = self._match_id(url)
 
@@ -42,19 +49,21 @@ class OnionStudiosIE(InfoExtractor):
         self._sort_formats(formats)
 
         title = self._search_regex(
-            r'share_title\s*=\s*"([^"]+)"', webpage, 'title')
+            r'share_title\s*=\s*(["\'])(?P<title>[^\1]+?)\1',
+            webpage, 'title', group='title')
         description = self._search_regex(
-            r'share_description\s*=\s*"([^"]+)"', webpage,
-            'description', default=None)
+            r'share_description\s*=\s*(["\'])(?P<description>[^\1]+?)\1',
+            webpage, 'description', default=None, group='description')
         thumbnail = self._search_regex(
-            r'poster="([^"]+)"', webpage, 'thumbnail', default=False)
+            r'poster\s*=\s*(["\'])(?P<thumbnail>[^\1]+?)\1',
+            webpage, 'thumbnail', default=False, group='thumbnail')
 
         uploader_id = self._search_regex(
-            r'twitter_handle\s*=\s*"([^"]+)"',
-            webpage, 'uploader id', fatal=False)
+            r'twitter_handle\s*=\s*(["\'])(?P<uploader_id>[^\1]+?)\1',
+            webpage, 'uploader id', fatal=False, group='uploader_id')
         uploader = self._search_regex(
-            r'window\.channelName\s*=\s*"Embedded:([^"]+)"',
-            webpage, 'uploader', default=False)
+            r'window\.channelName\s*=\s*(["\'])Embedded:(?P<uploader>[^\1]+?)\1',
+            webpage, 'uploader', default=False, group='uploader')
 
         return {
             'id': video_id,