CORS errors in pouchdb - javascript

I'm getting CORS errors in Firefox and Chrome, but not in cURL. Here's my cURL:
curl -H "Origin: http://mymachine:8080" https://wamoyo.cloudant.com/simpsons -v
That's my command, here's the output:
* Hostname was NOT found in DNS cache
* Trying 184.173.163.133...
* Connected to wamoyo.cloudant.com (184.173.163.133) port 443 (#0)
* successfully set certificate verify locations:
* CAfile: none
CApath: /etc/ssl/certs
* SSLv3, TLS handshake, Client hello (1):
* SSLv3, TLS handshake, Server hello (2):
* SSLv3, TLS handshake, CERT (11):
* SSLv3, TLS handshake, Server finished (14):
* SSLv3, TLS handshake, Client key exchange (16):
* SSLv3, TLS change cipher, Client hello (1):
* SSLv3, TLS handshake, Finished (20):
* SSLv3, TLS change cipher, Client hello (1):
* SSLv3, TLS handshake, Finished (20):
* SSL connection using AES256-SHA
* Server certificate:
* subject: C=US; ST=Massachusetts; L=Boston; O=Cloudant, Inc.; OU=Engineering; CN=*.cloudant.com
* start date: 2013-01-29 00:00:00 GMT
* expire date: 2016-02-19 12:00:00 GMT
* subjectAltName: wamoyo.cloudant.com matched
* issuer: C=US; O=DigiCert Inc; OU=www.digicert.com; CN=DigiCert High Assurance CA-3
* SSL certificate verify ok.
> GET /simpsons HTTP/1.1
> User-Agent: curl/7.35.0
> Host: wamoyo.cloudant.com
> Accept: */*
> Origin: http://mymachine:8080
>
< HTTP/1.1 200 OK
< X-Couch-Request-ID: 1efb92f7dd
* Server CouchDB/1.0.2 (Erlang OTP/R14B) is not blacklisted
< Server: CouchDB/1.0.2 (Erlang OTP/R14B)
< Date: Wed, 09 Jul 2014 18:25:37 GMT
< Content-Type: text/plain;charset=utf-8
< Content-Length: 362
< Cache-Control: must-revalidate
< Access-Control-Expose-Headers: content-type, accept-ranges, etag, server, x-couch-request-id, x-couch-update-newrev
Here's the important bit:
< Access-Control-Allow-Origin: http://mymachine:8080
< Access-Control-Allow-Credentials: true
<
{"update_seq":"34-g1AAAADreJzLYWBgYMlgTmFQTElKzi9KdUhJMtPLzc_PK87IzEvVS87JL01JzCvRy0styQEqZUpkSLL___9_ViI_qiZjfJqSHIBkUj1YH5plRvj05bEASYYGIAXUuj8rkQtVrylhvQcgeoH2smYBAApoT3A","db_name":"simpsons","purge_seq":0,"other":{"data_size":593},"doc_del_count":0,"doc_count":6,"disk_size":750276,"disk_format_version":5,"compact_running":false,"instance_start_time":"0"}
* Connection #0 to host wamoyo.cloudant.com left intact
Okay, now the browser, still returns this error:
Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at https://wamoyo.cloudant.com/simpsons/_changes?timeout=25000&style=all_docs&feed=longpoll&since=35-g1AAAAEjeJzLYWBgYMlgTmFQTElKzi9KdUhJMtPLzc_PK87IzEvVS87JL01JzCvRy0styQEqZUpkSLL___9_VgZTIn8uUIA9zcQ81cgkCdUIY3xGJDkAyaR6qCmsEFOMkxMTUy1RTTHCZ0oeC5BkaABSQIP2g0ziApuUYpJmZmZqjmqSKWGTDkBMQnKTuVFymoFxUhYAK3pbEA&limit=25&_nonce=7YzIfmsBHKTHaGPq. This can be fixed by moving the resource to the same domain or enabling CORS.
When I run PouchDB's sync or replicate functions.
PouchDB.sync('https://wamoyo.cloudant.com/simpsons/', 'simpsons', {live: true})
.on('change', onChange)
.on('complete', onComplete)
.on('error', onError);
function onChange (info) {
alert('onChange running');
}
function onComplete (info) {
alert('onComplete running');
}
function onError (err) {
alert('onError ' + err);
}

CORS is applicable only in a browser context. So, for it to work with PouchDB, you have to set the CouchDB CORS Headers to allow whatever domain you are accessing it from.
CORS is a security feature meant only for browsers. Browsers try to protect the users from the website which otherwise may make AJAX requests to other domains.
For example: when you are on stackoverflow.com and if it tries to make a AJAX request to mail.google.com, then your browser has a reason to believe that this may not be allowable by mail.google.com. Hence, it asks mail.google.com through a OPTIONS request to tell if stackoverflow.com is white-listed to make that particular request. If it does, then browser allows the actual request. Otherwise, it will block it as an error.
Now, as far as curl or any other non-browser request tool is concerned, they work differently. They are a representative of you and hence it is assumed that you would not be doing anything wrong to yourself.

Related

Screen scraping a webserver using its IP address instead of its domain name

Is this possible? It works when baseUrl = "http://mashable.com" but won't work when I give it an IP address.
<script src='https://raw.github.com/padolsey/jQuery-Plugins/master/cross-domain-ajax/jquery.xdomainajax.js'></script>
<script>$(document).ready(function () {
baseUrl = "https://12.34.56.78:8000/";
$.ajax({
url: baseUrl,
type: "get",
dataType: "",
success: function (data) {
alert("Yeah we are om jere");
});
});
That's going to be difficult, as many websites may be hosted on the same server, thus sharing the same IP. It works with the domain name because your client sends it in the Host header along with the GET request.
See this curl output for Stack Overflow:
C:\Users\Yeah>curl --head -i -v stackoverflow.com/
* Hostname was NOT found in DNS cache
* Trying 198.252.206.140...
* Connected to stackoverflow.com (198.252.206.140) port 80 (#0)
> HEAD / HTTP/1.1
> User-Agent: curl/7.38.0
> Host: stackoverflow.com
> Accept: */*
>
< HTTP/1.1 200 OK
HTTP/1.1 200 OK
< [...]
You can see the domain name is being passed as a header.
If I, instead, try a query with the IP address found above it results in a 404 error:
C:\Users\Yeah>curl --head -i -v 198.252.206.140/
* Hostname was NOT found in DNS cache
* Trying 198.252.206.140...
* Connected to 198.252.206.140 (198.252.206.140) port 80 (#0)
> HEAD / HTTP/1.1
> User-Agent: curl/7.38.0
> Host: 198.252.206.140
> Accept: */*
>
< HTTP/1.1 404 Not Found
HTTP/1.1 404 Not Found
< [...]
As a counter-example, though, here's what I get if I try to do something similar with the Facebook website:
C:\Users\Yeah>curl --head -i -v --insecure -L https://www.facebook.com/
* Hostname was NOT found in DNS cache
* Trying 31.13.93.3...
* Connected to www.facebook.com (31.13.93.3) port 443 (#0)
* [SSL stuff ...]
> HEAD / HTTP/1.1
> User-Agent: curl/7.38.0
> Host: www.facebook.com
> Accept: */*
>
< HTTP/1.1 200 OK
HTTP/1.1 200 OK
< [...]
And if I then try with the IP address above:
C:\Users\Yeah>curl --head -i -v --insecure -L https://31.13.93.3/
* Hostname was NOT found in DNS cache
* Trying 31.13.93.3...
* Connected to 31.13.93.3 (31.13.93.3) port 443 (#0)
* [SSL stuff ...]
> HEAD / HTTP/1.1
> User-Agent: curl/7.38.0
> Host: 31.13.93.3
> Accept: */*
>
< HTTP/1.1 301 Moved Permanently
HTTP/1.1 301 Moved Permanently
< Location: http://www.facebook.com/
Location: http://www.facebook.com/
< [...]
<
* Connection #0 to host 31.13.93.3 left intact
* Issue another request to this URL: 'http://www.facebook.com/'
* Hostname was NOT found in DNS cache
* Trying 31.13.93.3...
* Connected to www.facebook.com (31.13.93.3) port 80 (#1)
> HEAD / HTTP/1.1
> User-Agent: curl/7.38.0
> Host: www.facebook.com
> Accept: */*
>
< HTTP/1.1 301 Moved Permanently
HTTP/1.1 301 Moved Permanently
< [...]
<
* Connection #1 to host www.facebook.com left intact
* Issue another request to this URL: 'https://www.facebook.com/'
* Found bundle for host www.facebook.com: 0x6097814fe0
* Hostname was NOT found in DNS cache
* Trying 31.13.93.3...
* Connected to www.facebook.com (31.13.93.3) port 443 (#2)
* [SSL stuff ...]
> HEAD / HTTP/1.1
> User-Agent: curl/7.38.0
> Host: www.facebook.com
> Accept: */*
>
< HTTP/1.1 200 OK
HTTP/1.1 200 OK
< [...]
Here -L (follow redirects) and --insecure (accept any certificate) are needed to make cUrl ultimately connect to the Facebook website, but these are usual client (i.e. browser) operations.
So that really depends on the particular website and server configuration you want to screen scrap.

WebSocket handshake with Ruby and EM::WebSocket::Server

I am trying to create a simple WebSocket connection in JavaScript against my Rails app. I get the following:
WebSocket connection to 'ws://localhost:4000/' failed: Error during WebSocket handshake: 'Sec-WebSocket-Accept' header is missing
What am I doing wrong? Here is my code:
JavaScript:
var socket = new WebSocket('ws://localhost:4000');
socket.onopen = function() {
var handshake =
"GET / HTTP/1.1\n" +
"Host: localhost\n" +
"Upgrade: websocket\n" +
"Connection: Upgrade\n" +
"Sec-WebSocket-Key: x3JJHMbDL1EzLkh9GBhXDw==\n" +
"Sec-WebSocket-Protocol: quote\n" +
"Sec-WebSocket-Version: 13\n" +
"Origin: http://localhost\n";
socket.send(handshake);
};
socket.onmessage = function(data) {
console.log(data);
};
Ruby:
require 'rubygems'
require 'em-websocket-server'
module QuoteService
class WebSocket < EventMachine::WebSocket::Server
def on_connect
handshake_response = "HTTP/1.1 101 Switching Protocols\n"
handshake_response << "Upgrade: websocket\n"
handshake_response << "Connection: Upgrade\n"
handshake_response << "Sec-WebSocket-Accept: HSmrc0sMlYUkAGmm5OPpG2HaGWk=\n"
handshake_response << "Sec-WebSocket-Protocol: quote\n"
send_message(handshake_response)
end
def on_receive(data)
puts 'RECEIVED: ' + data
end
end
end
EventMachine.run do
print 'Starting WebSocket server...'
EventMachine.start_server '0.0.0.0', 4000, QuoteService::WebSocket
puts 'running'
end
The handshake headers are per Wikipedia.
I think that once the connection is open the request and response have already occurred, so sending headers at that point is too late. In addition, headers have to end with a blank line, which you omitted.
According to the demos, you don't even have to set headers in the client or the server--the ruby module automatically takes care of the headers on the server side, and html5 automatically takes care of the headers on the client side. I think this should work:
require "em-websocket-server"
class EchoServer < EM::WebSocket::Server
def on_connect
EM::WebSocket::Log.debug "Connected"
puts "I felt a connection."
end
def on_receive msg
puts "RECEIVED: #{msg}"
send_message msg
end
end
EM.run do
myhost = "0.0.0.0"
myport = 8000
puts "Starting WebSocket server. Listening on port #{myport}..."
EM.start_server myhost, myport, EchoServer
end
html file:
<!DOCTYPE html> <html> <head><title>Test</title>
<script type="text/javascript">
var myWebSocket = new WebSocket("ws://localhost:8000");
myWebSocket.onopen = function(evt) {
console.log("Connection open. Sending message...");
myWebSocket.send("Hello WebSockets!"); };
myWebSocket.onmessage = function(evt) {
console.log(evt.data);
myWebSocket.close(); };
myWebSocket.onclose = function(evt) {
console.log("Connection closed."); };
myWebSocket.onerror = function(err) {
alert(err.name + " => " + err.message); } </script>
</head> <body> <div>Hello</div> </body> </html>
And it does work in Safari 5.1.9 (which is an older browser): I see the expected output on both the server and the client. However, the code does not work in Firefox 21: I get the error message...
Firefox can't establish a connection to the server at ws://localhost:8000/.
var myWebSocket = new WebSocket("ws://localhost:8000");
I notice that in both Firebug and Safari Developer Tools, the server does not send a Sec-WebSocket-Accept header:
Response Headers
Connection Upgrade
Upgrade WebSocket
WebSocket-Location ws://localhost:8000/
WebSocket-Origin null
Request Headers
Accept text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Encoding gzip, deflate
Accept-Language en-US,en;q=0.5
Cache-Control no-cache
Connection keep-alive, Upgrade
DNT 1
Host localhost:8000
Origin null
Pragma no-cache
Sec-WebSocket-Key r9xT+ywe533EHF09wxelkg==
Sec-WebSocket-Version 13
Upgrade websocket
User-Agent Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:21.0) Gecko/20100101 Firefox/21.0
Nothing I tried would make the code work in Firefox 21.0. To check whether Firefox 21.0 even supports websockets, I went to:
http://www.websocket.org/echo.html
and it said my browser does support websockets.
Is there any reason you have to use the em-websocket-server module? The last modification for that module on github was three years ago. And whenever you see require rubygems in ruby code, that should alert you that the code is old. I tried the newer em-websocket module, and I was able to successfully transfer data back and forth using websockets on both Firefox 21.0 and Safari 5.1.9:
require 'em-websocket'
myhost = "0.0.0.0"
myport = 8000
EM.run {
puts "Listening on port #{myport}..."
EM::WebSocket.run(:host => myhost, :port => myport, :debug => false) do |ws|
ws.onopen do |handshake|
path = handshake.path
query_str = handshake.query
origin = handshake.origin
puts "WebSocket opened:"
puts "\t path \t\t -> #{path}"
puts "\t query_str \t -> #{query_str}"
puts "\t origin \t -> #{origin}"
end
ws.onmessage { |msg|
ws.send "Pong: #{msg}"
}
ws.onclose {
puts "WebSocket closed"
}
ws.onerror { |e|
puts "Error: #{e.message}"
}
end
}
Same client side code. Now the response headers include Sec-WebSocket-Accept:
Response Headers
Connection Upgrade
Sec-WebSocket-Accept LyIm6d+kAAqkcTR744tVK9HMepY=
Upgrade websocket
Request Headers
Accept text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Encoding gzip, deflate
Accept-Language en-US,en;q=0.5
Cache-Control no-cache
Connection keep-alive, Upgrade
DNT 1
Host localhost:8000
Origin null
Pragma no-cache
Sec-WebSocket-Key pbK8lFHQAF+arl9tFvHn/Q==
Sec-WebSocket-Version 13
Upgrade websocket
User-Agent Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:21.0) Gecko/20100101 Firefox/21.0
In your code, I don't think you are setting any headers. Instead, you are just sending messages back and forth that happen to contain characters that look like headers. Apparently, your browser requires the Sec-WebSocket-Accept header in the response before it will allow the connection, and when the em-websocket-server module fails to set that header in the response, your browser refuses the connection.
The relevant source code for em-websockets-server looks like this:
module EM
module WebSocket
module Protocol
module Version76
# generate protocol 76 compatible response headers
def response
response = "HTTP/1.1 101 Web Socket Protocol Handshake\r\n"
response << "Upgrade: WebSocket\r\n"
response << "Connection: Upgrade\r\n"
response << "Sec-WebSocket-Origin: #{origin}\r\n"
response << "Sec-WebSocket-Location: #{scheme}://#{host}#{path}\r\n"
if protocol
response << "Sec-WebSocket-Protocol: #{protocol}\r\n"
end
response << "\r\n"
response << Digest::MD5.digest(keyset)
response
end
As you can see, it doesn't set the Sec-WebSocket-Accept header. That code is in a module called Version76, and searching google for websockets version 76 yields an obsolete protocol(which contains an example of a request and response):
https://datatracker.ietf.org/doc/html/draft-hixie-thewebsocketprotocol-76
Here is the current websockets protocol(which also contains an example of a request and response):
https://www.rfc-editor.org/rfc/rfc6455
Conclusion: em-websockets-server is obsolete.

JavaScript WebSocket closes connection straight away

I'm writing my own script to connect to a websocket server with JavaScript using the WebSocket API. I'm having problems with the connection closing straight away.
Here's the client side script:
var host = 'ws://localhost:8080';
try
{
debug.add('Connection request submitted for ' + host);
socket = new WebSocket(host);
debug.readyStateListener();
debug.add('Socket request started');
socket.onopen = function()
{
debug.add('Connection opened');
}
socket.onmessage = function(message)
{
debug.add('data received ' + message.data);
}
socket.onclose = function()
{
debug.add('Connection closed');
}
}
catch(e)
{
debug.add('WebSockets error ' + e.toString() );
}
This is the debug I receive:
Connection request submitted for ws://localhost:8080
socket readyState change to 0
Socket request started
socket readyState change to 3
Connection closed
The debug.readyStateListener() polls socket.readyState for changes. What's happening is it changes to 0 meaning the connection is opening, then straight away changes to 3 that the connection has been closed.
The server receives the connection fine but the connection is then closed straight away by the client.
I've tried it in Opera 11 with WebSockets enabled and in the latest version of Chrome. Both time's I get the same result.
I can communicate perfectly with the server through a raw connection, or simply by visiting http://localhost:8080/ in my browser this is the result:
GET / HTTP/1.1
User-Agent: Opera/9.80 (Windows NT 6.1; U; IBM EVV/3.0/EAK01AG9/LE; en) Presto/2.9.168 Version/11.51
Host: localhost:8080
Accept: application/xhtml+voice+xml;version=1.2, application/x-xhtml+voice+xml;version=1.2, text/html, application/xml;q=0.9, application/xhtml+xml, image/png, image/webp, image/jpeg, image/gif, image/x-xbitmap, */*;q=0.1
Accept-Language: nl-NL,nl;q=0.9,en;q=0.8
Accept-Encoding: gzip, deflate
Connection: Keep-Alive
The connection also stays alive without any flaws by sending a request through http until I end it.
Going through the JavaScript WebSocket API the server receives this request:
GET / HTTP/1.1
Host: localhost:8080
Origin: http://localhost
Upgrade: WebSocket
Sec-WebSocket-key1: L58(b Q]'9 4 9\ 0 *+ 6 a4
Connection: Upgrade
Sec-WebSocket-Key2: \+ 1 5d/9541840N*4
My last guess would be Connection: upgrade or Upgrade: WebSocket not being supported properly by the client. To me it would be more logical to receive Connection: keep-alive but I have no idea how to reslove this.

Reading incoming HTTP headers with node.js

Now as example, I'm getting an response which has partially the key/values as an javascript object:
status: '200 OK',
'content-encoding': 'gzip'
I can easily read out and log the status message by: headers.status but when I try to log the content-encoding (which I need in this particular situation) it errors on:
headers.'content-encoding' <- obviously the quotes it doesn't like
headers.content-encoding <- obviously the '-' it doesn't like
How am I suppose to get/read/log it's content-encoding value?
Greets,
m0rph3v5
Javascript also supports square bracket notation for referring to properties so if headers is an appropriate object, you can use headers['content-encoding'].
JavaScript properties have names as you know. When the name is a legal identifier and you know the literal name you want when you're writing the code, you can use it with dotted notation.
var foo = headers.foo;
When the name isn't a legal identifier, or if you want to determine the name you're looking up at runtime, you can use a string:
var encoding = headers['content-encoding'];
or
var name = 'content-encoding';
var encoding = headers[name];
or even
var x = 'encoding';
var encoding = headers['content-' + x];
As you can see, it doesn't have to be a literal string. This is very handy for general-purpose functions that have to accept a property name as a function argument or similar.
Note that property names are case sensitive.
I think you should install the very good express framework. I really simplifies node.js webdevelopment.
You could install it using npm
npm install express
This snippet shows you how to set headers and read headers
var express = require('express');
var app = express.createServer();
app.get('/', function(req, res){
console.log(req.header('a'));
res.header('time', 12345);
res.send('Hello World');
});
app.listen(3000);
Curl from command line
$curl http://localhost:3000/ -H "a:3434" -v
* About to connect() to localhost port 3000 (#0)
* Trying ::1... Connection refused
* Trying 127.0.0.1... connected
* Connected to localhost (127.0.0.1) port 3000 (#0)
> GET / HTTP/1.1
> User-Agent: curl/7.21.2 (i686-pc-linux-gnu) libcurl/7.21.2 OpenSSL/0.9.8o zlib/1.2.3.4 libidn/1.18
> Host: localhost:3000
> Accept: */*
> a:3434
>
< HTTP/1.1 200 OK
< X-Powered-By: Express
< time: 12345
< Content-Type: text/html; charset=utf-8
< Content-Length: 11
< Date: Tue, 28 Dec 2010 13:58:41 GMT
< X-Response-Time: 1ms
< Connection: keep-alive
<
* Connection #0 to host localhost left intact
* Closing connection #0
Hello World
The log outputting the header send via curl to node server:
$ node mo.js
3434

Please help test a CORS issue in Firefox jQuery ajax when 401

this is driving me nutters.
jQuery 1.4.2, windows XP sp3
Here is my test.
Load firefox 3.5+
http://plungjan.name/test/testcors.html
works
Save the file to harddisk and run from there
From my office the external works and the internal does not
What is also interesting is that I cannot run both in one go.
Background:
I do a GET to an internal web service that uses CORS.
Please do NOT post any answers about FF not handling cross domain request when it does since v3.5 as detailed here and here
It works in IE8 and FF3.6.6 from one server to the other and now almost from file system (file:///) to service.
Only from file system and only when FF 3.6.6 needs to negotiate (the user is already logged in, authorised and sends the credentials!) do I not get the data after negotiation. jQuery xhr returns status 0 and no data/responseText or whatever
Seems to me, jQuery reacts and saves the xhr from the 401 rather than from the 200 OK later
Here is the result I get at the end of the communication when I alert the XHR object:
Status:success
Data:[]
XHR:
some native functions,
readyState:4
status:0
responseXML:null
responseText:
withCredentials:true
if I make a call to the same server but without needing credentials, the data is returned just fine cross domain
So the communication is as follows:
GET /restapplicationusingcors/authenticationneeded-internal/someid
Accept: application/json
Accept-Language: en
.
.
Origin: null
Cookie: LtpaToken=...
the return is
HTTP/1.1 401 Unauthorized
Server: Apache
Pragma: No-cache
Cache-Control: no-cache
Expires: Thu, 01 Jan 1970 01:00:00 CET
WWW-Authenticate: Negotiate
Connection: close
Transfer-Encoding: chunked
Content-Type: text/html
Then FF sends
GET /restapplicationusingcors/authenticationneeded-internal/someid HTTP/1.1
Host: myhost.myintranet.bla
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2.6) Gecko/20100625 Firefox/3.6.6
Accept: application/json
Accept-Language: en
Accept-Encoding: gzip,deflate
Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7
Keep-Alive: 115
Connection: keep-alive
Origin: null
Cookie: LtpaToken=....
Authorization: Negotiate ....
and is rewarded with the file I need, but cannot get at in FF:
HTTP/1.1 200 OK
Date: Tue, 20 Jul 2010 12:08:39 GMT
Pragma: No-cache
Cache-Control: no-cache, max-age=600, s-maxage=3600
Expires: Thu, 01 Jan 1970 01:00:00 CET
X-Powered-By: ...
Content-Disposition: inline;filename=nnnnnn.json
Content-Language: en
Access-Control-Allow-Origin: ...
Keep-Alive: timeout=6, max=70
Connection: Keep-Alive
Transfer-Encoding: chunked
Content-Type: application/json;charset=UTF-8
THE DATA SENT FROM THE SERVER IS NOT IN THE XHR OBJECT
Here is my code
function getJSON(url,func,lang) {
accept = 'application/json';
lang=lang?lang:"*";
// gruesome hack to handle that APPENDS the mime header to */* !!!
// NOW HANDLED by first setting Accept to "" !!!
// if ($.browser.msie && url.indexOf('serveAsMime')==-1) {
// url+= '?serveAsMime='+accept;
// }
if (currentRequest != null) currentRequest.abort();
var requestObjectJSON = {
url : url,
// dataType: "json",
method : 'get',
beforeSend: function(xhr){
xhr.setRequestHeader('Accept', ""); // IE hack
xhr.setRequestHeader('Accept', accept);
xhr.setRequestHeader('Accept-Language', lang);
if (url.indexOf('-internal') !=-1) {
try {
xhr.withCredentials = true;
alert('set credentials')
}
catch(e) {
alert('cannot set xhr with credentials')
}
}
},
success: function(data,status,xhr) {
var responseText = xhr.responseText;
var responseJSON = xhr.responseJSON;
var t = "";
try{
for (var o in xhr) t += '\n'+o+':'+xhr[o];
}
catch(e) {
if (e.message.indexOf('.channel')==-1)alert(e.message);
}
alert('Status:'+status+'\nData:['+data+']\nXHR:'+t);
func(responseText);
},
}
currentRequest = $.ajax(requestObjectJSON);
}
This is a stab in the dark since I don't fully understand your problem, but I think you might be having a problem with file: URLs, which are not treated as having any origin. I'm not sure it's even possible to authorize CORS from a file URL.
So you need to set an ajax prefilter in your model/collection in order to use CORS. Otherwise it doesn't send the cookie.
$.ajaxPrefilter( function( options, originalOptions, jqXHR ) {
options.xhrFields = {
withCredentials: true
};
});
I put this in my Model/Collection initialize function.
These are the conditions to be met to make CORS working with secured services:
Service response should contain header Access-Control-Allow-Credentials: true (see Requests with credentials and Cannot use wildcard in Access-Control-Allow-Origin when credentials flag is true).
Service response header Access-Control-Allow-Origin should not be *. The idea is to return the value passed by client in header Origin (see examples in this post).
According to specification, OPTIONS method should return HTTP code 200, thus it cannot be secured (see The CORS).
For methods PUT/POST that need to pass certain request headers to service (like Content-Type or Accept), these headers need to be listed in Access-Control-Allow-Headers (see jQuery AJAX fails to work when headers are specified)
JavaScript should set this XMLHttpRequest property: xhr.withCredentials = true; (as answered by Kirby)
Altogether configuration for Apache:
# Static content:
SetEnvIf Request_URI ".*" no-jk
# RESTful service:
SetEnvIf Request_URI "^/backend/" !no-jk
SetEnvIf Request_Method "OPTIONS" no-jk
# Fallback value:
SetEnv http_origin "*"
SetEnvIf Origin "^https?://(localhost|.*\.myconpany\.org)(:[0-9]+)?$" http_origin=$0
Header set Access-Control-Allow-Credentials "true"
Header set Access-Control-Allow-Origin "%{http_origin}e"
Header set Access-Control-Allow-Methods "GET,POST,PUT,DELETE"
Header set Access-Control-Allow-Headers "Content-Type, Accept"
JkMount /* loadbalancer
CORS with file://
If you have problems by allowing origins from the file:// protocol, according to The Web Origin Concept it should be done the same way as any other origins. I could not find information about the browser support, but I think every browser which is supporting CORS does support this one either.
The Web Origin Concept tells us the following about the file URI scheme:
4. If uri-scheme is "file", the implementation MAY return an
implementation-defined value.
NOTE: Historically, user agents have granted content from the
file scheme a tremendous amount of privilege. However,
granting all local files such wide privileges can lead to
privilege escalation attacks. Some user agents have had
success granting local files directory-based privileges, but
this approach has not been widely adopted. Other user agents
use globally unique identifiers for each file URI, which is
the most secure option.
According to wikipedia the domain by the file URI scheme is localhost. It is omittable by the address bar, but I don't think it is omittable in the allow origin headers. So if your browser implementation allows origin with a file URI scheme, then you should add file://localhost to your allowed origins, and everything should work properly after that.
This was how it should work, now meet reality:
I tested with current firefox 29.0.1, and it did not work. However the file:// protocol is transformed into null origin by this implementation. So by firefox the null works. I tried with a wider domain list, but I did not manage to allow multiple domains. It seems like firefox does not support a list with multiple domains currently.
I tested with chrome 35.0.1916, it works the same way as firefox did.
I tested with msie 11.0.9600. By request from the file protocol it always shows an allow blocked content button, even by not allowing the null origin. By other domains it works the same way as the previous browsers.
HTTP basic auth:
The credentials part I tried out with PHP and HTTP basic auth.
http://test.loc
Displays :-) when logged in and :-( when unauthorized.
<?php
function authorized()
{
if (empty($_SERVER['PHP_AUTH_USER']) || empty($_SERVER['PHP_AUTH_PW']))
return false;
return ($_SERVER['PHP_AUTH_USER'] == 'username' && $_SERVER['PHP_AUTH_PW'] == 'password');
}
function unauthorized()
{
header('HTTP/1.1 401 Unauthorized');
header('WWW-Authenticate: Basic realm="Restricted Area"');
echo ':-(';
}
if (!isset($_GET['logout']) && authorized()) {
echo ':-)';
} else
unauthorized();
So this code changes the location by login and logout.
Cross domain CORS with HTTP basic auth
http://todo.loc
Gets the content of http://test.loc with cross domain XHR and displays it.
cross domain ajax<br />
<script>
var xhr = new XMLHttpRequest();
xhr.open('GET', "http://test.loc", true);
xhr.withCredentials = true;
xhr.onreadystatechange = function (){
if (xhr.readyState==4) {
document.body.innerHTML += xhr.responseText;
}
};
xhr.send();
</script>
Requires headers by http://test.loc:
Access-Control-Allow-Origin: http://todo.loc
Access-Control-Allow-Credentials: true
Cross scheme CORS with HTTP basic auth
file:///path/x.html
Gets the content of http://test.loc with cross scheme XHR and displays it.
cross scheme ajax<br />
<script>
var xhr = new XMLHttpRequest();
xhr.open('GET', "http://test.loc", true);
xhr.withCredentials = true;
xhr.onreadystatechange = function (){
if (xhr.readyState==4) {
document.body.innerHTML += xhr.responseText;
}
};
xhr.send();
</script>
Requires headers by http://test.loc:
Access-Control-Allow-Origin: null
Access-Control-Allow-Credentials: true
Conclusion:
I tested cross-sheme CORS with credentials called from file:// and it works pretty well in firefox, chrome and msie.

Categories