how does facebook/digg get all of the iamges for a webpage?

DO they 开发者_如何学运维use a php page to analyze the link, and return all of the images as josn? Is there a way to do this with just javascript, so you dont have to go to the server to analyze the page?

I don't now how they do it. I'd implement a small service for that purpose. Given an URL return some relevant image (or generate a screenshot). This service could also cache results for better performance. But still, the page needs to be accessed in order to grab the <img src=... or to take the photograph.

Facebook calls back to the server. If you use Firebug (or, as I did, the Web Inspector in Safari), you can inspect the ajax calls. Facebook calls back to a script at /ajax/composer/attachment.php - in there is some JavaScript which contains HTML that gets inserted into the page. Here is what it looks if I point the Facebook attach link dialogue to the BBC News homepage in Safari Web Inspector:

Facebook JavaScript response when you attach a link in Safari Web Inspector http://tommorris.org/files/Facebook-20100529-181745.jpg

I put up the full JavaScript response on Gist (it is all one-line and minified originally, so I just flung it through TextMate to wrap it).

I'm not sure if you could do it on the client-side - because of browser protections on cross-site scripting - and even if you could, you probably ought not to because of this potential security problem: imagine if someone puts in a URL that points to a page which only they have access to. You don't necessarily want to put what's on someone else's customised or private page up on your Facebook/Digg type site. Imagine if it was something like Flickr and there were private pics - or worse, a porno site. No, better to proxy it back to your server and then grab the images. Plus, it'll probably be faster. No need to tax your end user's potentially slow connection downloading a page when your server will probably be able to do it quicker...