[escapist] Add support for og:video:url (Fixes #3557)

[youtube-dl.git] / youtube_dl / extractor / common.py

diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py
index 88f12797c6c351ab8eae943d183c81bc90028e8e..2e6eeac08e403fcf46efe674ed892cec4a7b9a84 100644 (file)
--- a/youtube_dl/extractor/common.py
+++ b/youtube_dl/extractor/common.py
@@ -18,6 +18,7 @@ from ..utils import (
     clean_html,
     compiled_regex_type,
     ExtractorError,
+    int_or_none,
     RegexNotFoundError,
     sanitize_filename,
     unescapeHTML,
@@ -373,7 +374,8 @@ class InfoExtractor(object):
         else:
             for p in pattern:
                 mobj = re.search(p, string, flags)
-                if mobj: break
+                if mobj:
+                    break
 
         if os.name != 'nt' and sys.stderr.isatty():
             _name = u'\033[0;34m%s\033[0m' % name
@@ -461,8 +463,9 @@ class InfoExtractor(object):
         return self._og_search_property('title', html, **kargs)
 
     def _og_search_video_url(self, html, name='video url', secure=True, **kargs):
-        regexes = self._og_regexes('video')
-        if secure: regexes = self._og_regexes('video:secure_url') + regexes
+        regexes = self._og_regexes('video') + self._og_regexes('video:url')
+        if secure:
+            regexes = self._og_regexes('video:secure_url') + regexes
         return self._html_search_regex(regexes, html, name, **kargs)
 
     def _og_search_url(self, html, **kargs):
@@ -589,6 +592,24 @@ class InfoExtractor(object):
         self.to_screen(msg)
         time.sleep(timeout)
 
+    def _extract_f4m_formats(self, manifest_url, video_id):
+        manifest = self._download_xml(
+            manifest_url, video_id, 'Downloading f4m manifest',
+            'Unable to download f4m manifest')
+
+        formats = []
+        for media_el in manifest.findall('{http://ns.adobe.com/f4m/1.0}media'):
+            formats.append({
+                'url': manifest_url,
+                'ext': 'flv',
+                'tbr': int_or_none(media_el.attrib.get('bitrate')),
+                'width': int_or_none(media_el.attrib.get('width')),
+                'height': int_or_none(media_el.attrib.get('height')),
+            })
+        self._sort_formats(formats)
+
+        return formats
+
 
 class SearchInfoExtractor(InfoExtractor):
     """