The regex used non-greedy match, but alas it failed on input like this:
<object class="...> ... class="BrightcoveExperience"
It captured two objects and the intervening HTML. This commit fixes this by
not allowing a ">" to appear before BrightcoveExperience.
Video in question: http://www.harpercollinschildrens.com/feature/petethecat/
Now the IEs can set a formats field in the info_dict, with the formats ordered from worst to best quality. It's a list of dicts with the following fields:
* Mandatory: url and ext
* Optional: format and format_id
The format_id is used for choosing which formats have to be downloaded.
Now a video result is processed by the method process_video_result.
Extract the info from http://videoplayer.vevo.com/VideoService/AuthenticateVideo?isrc={id}
Some videos don't have an smil manifest, extract the video urls directly from the json and use the last version of the video.
Extract all the available formats and set the 'formats' field of the result