- # Look for embed_in_page embeds [2]
- for video_id, account_id, player_id, embed in re.findall(
- # According to examples from [3] it's unclear whether video id
- # may be optional and what to do when it is
- # According to [4] data-video-id may be prefixed with ref:
- r'''(?sx)
- <video[^>]+
- data-video-id=["\'](\d+|ref:[^"\']+)["\'][^>]*>.*?
- </video>.*?
- <script[^>]+
- src=["\'](?:https?:)?//players\.brightcove\.net/
- (\d+)/([^/]+)_([^/]+)/index(?:\.min)?\.js
- ''', webpage):
- entries.append(
- 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s'
- % (account_id, player_id, embed, video_id))
+ # Look for <video> tags [2] and embed_in_page embeds [3]
+ # [2] looks like:
+ # <video data-video-id="5320421710001" data-account="245991542" data-player="SJWAiyYWg" data-embed="default" class="video-js" controls itemscope itemtype="http://schema.org/VideoObject">
+
+ for video, script_tag, account_id, player_id, embed in re.findall(
+ r'''(?isx)
+ (<video[^>]+>)
+ (?:.*?
+ (<script[^>]+
+ src=["\'](?:https?:)?//players\.brightcove\.net/
+ (\d+)/([^/]+)_([^/]+)/index(?:\.min)?\.js
+ )
+ )?
+ ''', webpage
+ ):
+ attrs = extract_attributes(video)
+
+ # According to examples from [4] it's unclear whether video id
+ # may be optional and what to do when it is
+ video_id = attrs.get('data-video-id')
+ # See PR#12099/bostonglobe.py for 'data-brightcove-video-id' variant
+
+ if not account_id:
+ account_id = attrs.get('data-account')
+ if not player_id:
+ player_id = attrs.get('data-player')
+ if not embed:
+ embed = attrs.get('data-embed')
+
+ if video_id and account_id and player_id and embed:
+ entries.append(
+ 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s'
+ % (account_id, player_id, embed, video_id))