OpenGraph on Ajax Based Website
No. Open Graph markup must be present on HTML pages which are GETable with pure HTTP.
This is because when a user interacts with an OG object (like, performs an action etc) Facebook will perform an HTTP GET on the OG URL, and expect to see OG tags returned in the markup.
The solution is to create canonical URLs for each of your objects. These URLs contain basic HTML markup including OG tags.
On requests to these URLs, if you see the incoming useragent string containing 'facebookexternalhit' then you render the HTML. If you don't, you serve a 302 which redirects to your ajax URL. On the ajax URLs, your like buttons and any OG actions you publish should point to the canonical URL object
Example:
As a user, I'm on http://yoursite.com/#!/artists/monet. I click a like button, or publish an action, but the like button's href parameter, or the URL of the object when you post the action should be a web hittable canonical URL for the object - in this case, perhaps http://yoursite.com/artists/monet
When a user using a browser hits http://yoursite.com/artists/monet you should redirect them to http://yoursite.com/#!/artists/monet, but if the incoming useragent says it is Facebook's scraper, you just return markup which represents the artist Monet.
For real world examples, see Deezer, Rdio and Mog who all use this design pattern.
A little bit more investigation lead to the following findings:
Let's say you made an application with a hash that looks like this:
http://yoursite.com/#/artists/monetThe Facebook scraper will call your url without the /#/artists/monet part. This is a problem because you have no way of knowing what information you have to parse into the meta og: tags.
Then try the same with the suggested url as Simon says:
http://yoursite.com/#!/artists/monetNow you'll notice that the Facebook scraper is respecting the google ajax specifications and it will convert the #! to ?_escaped_fragment_= so the URL looks like this:
http://yoursite.com/?_escaped_fragment_=/artists/monetYou can check this out for yourself with the facebook debugger: https://developers.facebook.com/tools/debug
- upload the php script to your server
- go to the facebook debugger
- enter the url with the
/#/part - click 'See exactly what our scraper sees for your URL' - no hash fragment
- enter the url again with
/#!/ - click 'See exactly what our scraper sees for your URL' - hash fragment has been turned to
?_escaped_fragment_=
The script
<html> <head> <title>Scraping</title> </head> <body> <? print_r($_SERVER); ?> </body></html>Or summarized: always use /#!/ (hashbang) deeplinks ;)
I ran a quick test that seems to work. Dependant on the fact the FB scraper doesn't run JavaScript.
As most of my sites are static Single Page Apps with no server logic, I can generate the static pages quickly with tools such as Grunt and Gulp.
If you Share http://wh-share-test.s3-website-eu-west-1.amazonaws.com/test
Facebook will scrape the test page meta tags, when a user clicks the link the JS redirects to /#/test for my single page app to react and present the correct view.
Seems hacky but works;
<html lang="en" xmlns="http://www.w3.org/1999/xhtml"><head> <meta charset="utf-8" /> <title>This is a shared item</title></head><body> <h1>This is a shared item page.</h1> <script> var path = window.location.pathname; window.location = '/#' + path; </script></body></html>