Say I have an AJAX application with this URL: http://www.foo.com/bar#!a=1&b=2&c=3
What will the crawable AJAX request from GoogleBot look like?
I assumed this:
A) http://www.foo.com/bar?_escaped_fragment_=a%3D1%26b%3D2%26c%3D3
But it looks like it's really this: (i.e. the =s are not url encoded)
B) http://www.foo.com/bar?_escaped_fragment_=a=1%26b=2%26c=3
Is it correct that B) is the actual request I'll receive?
On close examination of the spec it looks like B) is correct, but I still find it a bit surprising since many canned query string parsers will probably not give you the result you want with this input.
I'll answer this myself.
The correct answer seems be be neither A) or B) in the above question. If the "clean" AJAX URL is http://www.foo.com/bar#!a=1&b=2&c=3, the "ugly" _escaped_fragment_ version is http://www.foo.com/bar?_escaped_fragment_=a=1%26b=2%26c=3. So the = are not escaped, but the & are. This is covered in the spec.