RE: http-methods & http-trace NSE Script Enhancement Ideas

[King Thorin <kingthorin () hotmail com>]

Date: Wed, 23 May 2012 20:03:30 +0200
Subject: Re: http-methods & http-trace NSE Script Enhancement Ideas
From: patrik () cqure net
To: kingthorin () hotmail com
CC: toni.ruottu () iki fi; paulino () calderonpale com; nmap-dev () insecure org



On Wed, May 23, 2012 at 7:33 PM, King Thorin <kingthorin () hotmail com> wrote:



I just had a quick look at http-cors. It does not appear to follow redirects or check status codes at all, only setting 
and getting header values. I'm not sure if those header values would or wouldn't be present in a redirect.




I still need someone (or a bunch of people) to confirm if I'm correct in my experiences with allow and public being 
lacking on redirect responses. Also I still need to know how to provide updates for the scripts in question.




I'd propose a script parameter such as:

redirect_count



So code for htt-trace.nse could look like ( I threw this together quickly, it's not necessarily perfect or useable in 
this form):



--- Validates the HTTP response and returns header list

--@param response The HTTP response

--@param response_headers The HTTP response headers

local validate = function(response, response_headers, followed_redirects)

  local output_lines = {}



  if not(response:match("HTTP/1.[01] 200") or response:match("TRACE / HTTP/1.[01]")) then

    return

  else

    output_lines[ #output_lines+1 ] = "TRACE is enabled"

    if followed-redirects > 0

      output_lines[ #output_lines+1 ] = "Followed " .. followed_redirects .. " redirects." -- We followed some 
redirects, tell the user

  end

  if nmap.verbosity() >= 2 then

    output_lines[ #output_lines+1 ]= "Headers:"

    for _, value in pairs(response_headers) do

      output_lines [ #output_lines+1 ] = value

    end

  end

  if #output_lines > 0 then

    return stdnse.strjoin("\n", output_lines)

  end

end



---

--MAIN

---

action = function(host, port)

  local path = stdnse.get_script_args("http-trace.path") or "/"

  local num_redirects = stdnse.get_script_args("http-trace.redirect_count") or 2 -- Set default low [2] and let user 
make it bigger if needed

  local followed_redirects = 0



  local req = http.generic_request(host, port, "TRACE", path) -- Request zero

  while (req.status == 301 or req.status == 302) and req.header["location"] and followed_redirects < num_redirects do 
-- Follow 2 or redirect_count redirects

    req = http.generic_request(host, port, "TRACE", req.header["location"])

    followed_redirects = followed_redirects + 1

  end -- Hopefully when we finish looping we received a HTTP 200 OK after following some redirects (at least we tried)



  return validate(req.body, req.rawheader, followed_redirects)

end



PS > The thread has now had two top and one bottom reply, what's the actual preference on this list?





Date: Wed, 23 May 2012 18:41:06 +0300

Subject: Re: http-methods & http-trace NSE Script Enhancement Ideas

From: toni.ruottu () iki fi

To: paulino () calderonpale com

CC: kingthorin () hotmail com; nmap-dev () insecure org



Does this affect http-cors too?



On Wednesday, 23 May 2012, Paulino Calderon  wrote:

On 23/05/2012 07:17 a.m., King Thorin wrote:







I was just looking through some online docs and some nmap results. I've



never seen a server that includes public or allow header(s) on a



redirect response [maybe my experience is limited?]. It seems to me that the http-methods NSE should follow



redirects (HTTP 301, 302, 303) in order to perform the necessary OPTIONS



  request on a page/resource that's providing a HTTP 200.











Perhaps similar to the http-trace script:



http://nmap.org/svn/scripts/http-trace.nse



Though



  even that only follows one 301 or 302 redirect.







Further, maybe both scripts should follow a configurable



  # of redirects (default 2, 3, 4 and configurable further) looking for a



  HTTP 200&  handle 301, 302, and 303 redirect codes.











Reference:



http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html







I've emailed the devs of both scripts without any luck.















I'd be glad to provide the necessary changes, if someone can simply fill me in as to how they should be submitted.











_______________________________________________



Sent through the nmap-dev mailing list



http://cgi.insecure.org/mailman/listinfo/nmap-dev



Archived at http://seclists.org/nmap-dev/





I think adding a configuration value for redirects will work better in some cases. I would say most of the libraries 
follow 2-3 redirects but no more than that. In your experience, what would be a good default?








--



Paulino Calderón Pale



Website: http://calderonpale.com



Twitter: http://twitter.com/calderpwn







_______________________________________________



Sent through the nmap-dev mailing list



http://cgi.insecure.org/mailman/listinfo/nmap-dev



Archived at http://seclists.org/nmap-dev/





_______________________________________________

Sent through the nmap-dev mailing list

http://cgi.insecure.org/mailman/listinfo/nmap-dev

Archived at http://seclists.org/nmap-dev/


The http library does support http redirects for get and head requests.While redirection may seem trivial to implement 
at first there are a actually a few things to consider.
Therefore, why not make trace a function in the http library wrapping the generic_request method and adding redirect 
support in the same way as has already been done for get, head, post and put requests?

Cheers,Patrik
-- 
Patrik Karlssonhttp://www.cqure.nethttp://twitter.com/nevdull77


Hi Patrik, I like that idea at face value. However I'm not a developer 
and looking at http://nmap.org/nsedoc/lib/http.html does not clarify redirect_ok or MAX_REDIRECT_COUNT for me at all.

  
I'd either need some better examples/direction or someone else to pickup the idea and modify the NSEs in question.

  

Rick

                                          
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/