Understanding XMLHttpRequest over CORS (responseText)

For a project I'm looking at various HTML5 and Javascript elements and security around them and I'm trying to get my head around CORS just now.

Based on my testing, if I remove..

<?php
 header("Access-Control-Allow-Origin: *"); 
 header('Access-Control-Allow-Methods: GET, POST, OPTIONS');
 ?>

..from the page that is trying to be accessed I see the following in the console log on Chrome:

XMLHttpRequest cannot load http://www.bla.com/index.php. Origin http://bla2.com is not allowed by Access-Control-Allow-Origin.

I understand this to be correct, however Wireshark shows HTTP/1.1 200 OK in the return and in the data shows the source of the page being requested. So is it just the browser and Javascript that is blocking responseText from being used in any substantial way even though it's actually transferred?

The code is just as below:

  function makeXMLRequest() {
  xmlhttp=new XMLHttpRequest();
xmlhttp.onreadystatechange = function() {
    if (xmlhttp.readyState==4) {
        alert(xmlhttp.responseText);
    }
}
xmlhttp.open("GET","http://www.bla.com/index.php",true);
xmlhttp.send();
}

Thanks in advance.

For a "simple" HTTP verb like GET or POST, yes, the entire page is fetched, and then the browser decides whether JavaScript gets to use the contents or not. The server doesn't need to know where the requests comes from; it is the browser's job to inspect the reply from the server and determine if JS is permitted to see the contents.

For a "non-simple" HTTP verb like PUT or DELETE, the browser issues a "preflight request" using an OPTIONS request. In that case, the browser first checks to see if the domain and the verb are supported, by checking for Access-Control-Allow-Origin and Access-Control-Allow-Methods, respectively. (See the "Handling a Not-So-Simple Request" on the CORS page of HTML5 Rocks for more information.) The preflight response also lists permissible non-simple headers, included in Access-Control-Allow-Headers.

This is because allowing a client to send a DELETE request to the server could be very bad, even if JavaScript never gets to see the cross-domain result -- again, remember that the server is generally not under any obligation to verify that the request is coming from a legitimate domain (although it may do so using the Origin header from the request).

So is it just the browser and Javascript that is blocking responseText from being used in any substantial way even though it's actually transferred?

Yes. You can make any request you like with JS.

It is access to the data that the same origin policy prevents.

Requests which do malicious things (such as "POST http://bank.example/give/money?to=attacker" or "POST http://forum.example.com/post?message=spamspamspamspam") are called CSRF attacks and have to be defended against by the server.

Understanding XMLHttpRequest over CORS (responseText)

Related

Recent Posts