[brightcove] fix format sorting

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

diff --git a/youtube_dl/extractor/brightcove.py b/youtube_dl/extractor/brightcove.py
index 304fb89e3a854931a2fefc25b6bdef08a09787b8..a8919001db8acb83bec50956eb289a66211cf664 100644 (file)
--- a/youtube_dl/extractor/brightcove.py
+++ b/youtube_dl/extractor/brightcove.py
@@ -9,7 +9,6 @@ from ..compat import (
     compat_etree_fromstring,
     compat_parse_qs,
     compat_str,
-    compat_urllib_parse,
     compat_urllib_parse_urlparse,
     compat_urlparse,
     compat_xml_parse_error,
@@ -137,13 +136,16 @@ class BrightcoveLegacyIE(InfoExtractor):
         else:
             flashvars = {}
 
+        data_url = object_doc.attrib.get('data', '')
+        data_url_params = compat_parse_qs(compat_urllib_parse_urlparse(data_url).query)
+
         def find_param(name):
             if name in flashvars:
                 return flashvars[name]
             node = find_xpath_attr(object_doc, './param', 'name', name)
             if node is not None:
                 return node.attrib['value']
-            return None
+            return data_url_params.get(name)
 
         params = {}
 
@@ -295,7 +297,7 @@ class BrightcoveLegacyIE(InfoExtractor):
             'uploader': video_info.get('publisherName'),
         }
 
-        renditions = video_info.get('renditions')
+        renditions = video_info.get('renditions', []) + video_info.get('IOSRenditions', [])
         if renditions:
             formats = []
             for rend in renditions:
@@ -317,13 +319,23 @@ class BrightcoveLegacyIE(InfoExtractor):
                 if ext is None:
                     ext = determine_ext(url)
                 size = rend.get('size')
-                formats.append({
+                a_format = {
                     'url': url,
                     'ext': ext,
                     'height': rend.get('frameHeight'),
                     'width': rend.get('frameWidth'),
                     'filesize': size if size != 0 else None,
-                })
+                }
+
+                # m3u8 manifests with remote == false are media playlists
+                # Not calling _extract_m3u8_formats here to save network traffic
+                if ext == 'm3u8':
+                    a_format.update({
+                        'ext': 'mp4',
+                        'protocol': 'm3u8',
+                    })
+
+                formats.append(a_format)
             self._sort_formats(formats)
             info['formats'] = formats
         elif video_info.get('FLVFullLengthURL') is not None:
@@ -384,6 +396,7 @@ class BrightcoveNewIE(InfoExtractor):
             'formats': 'mincount:41',
         },
         'params': {
+            # m3u8 download
             'skip_download': True,
         }
     }, {
@@ -413,8 +426,8 @@ class BrightcoveNewIE(InfoExtractor):
 
         # Look for iframe embeds [1]
         for _, url in re.findall(
-                r'<iframe[^>]+src=(["\'])((?:https?:)//players\.brightcove\.net/\d+/[^/]+/index\.html.+?)\1', webpage):
-            entries.append(url)
+                r'<iframe[^>]+src=(["\'])((?:https?:)?//players\.brightcove\.net/\d+/[^/]+/index\.html.+?)\1', webpage):
+            entries.append(url if url.startswith('http') else 'http:' + url)
 
         # Look for embed_in_page embeds [2]
         for video_id, account_id, player_id, embed in re.findall(
@@ -427,7 +440,7 @@ class BrightcoveNewIE(InfoExtractor):
                     </video>.*?
                     <script[^>]+
                         src=["\'](?:https?:)?//players\.brightcove\.net/
-                        (\d+)/([\da-f-]+)_([^/]+)/index(?:\.min)?\.js
+                        (\d+)/([^/]+)_([^/]+)/index(?:\.min)?\.js
                 ''', webpage):
             entries.append(
                 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s'
@@ -521,7 +534,7 @@ class BrightcoveNewIE(InfoExtractor):
                     f.update({
                         'url': src or streaming_src,
                         'format_id': build_format_id('http' if src else 'http-streaming'),
-                        'preference': 2 if src else 1,
+                        'source_preference': 0 if src else -1,
                     })
                 else:
                     f.update({