You can clone with
HTTPS or Subversion.
So ok, we gonna hate webserver programmers.
Some of my feeds return content-type: text/html instead of of content-type: application/rss+atom for example.
Maybe it's not a good thing to believe webservers content type guessers and we have could try ourselves to detect if it's a xml (feed) file before trying a HTTP request to find meta tags ?
I was slightly (okay, not just slightly) scared that this might happen, but was too tipsy at that point to care.. It seems that feedparser actually provides feed version functionality, we should definetly use that instead of request headers.
Oh yeah, I haven't see it.
Feedparser use its internal bozo detection which is set to 0 when a feed is wellformed or 1 when not.
I think it's just our solution.
And it will save us one HTTP request if user provide the feed url.
Now fixed by feedparser.
Thanks @orva !