Adrian Schneider
08-05-2007, 10:00 PM
In this post we'll go over some basic modification vulnerabilities and how to prevent them.
Preparing Data
Whenever you accept input from a user, you must understand that you have no control over what is entered, period. The short version is that anything client-side can be manipulated. Even if you use Javascript or HTML limitations, they can all easily be avoided. This input includes but is not limited to:
Entered form input
Hidden form input (!)
URL query string
User AgentIn your application it is your responsibility to ensure the validity of the data they enter.
Use the vBulletin Input Cleaner!
I'm going to assume you know how to use it, even though sometimes you choose not to use it. If not, dig up a tutorial on it.
A common example is fetching a row from a table using an ID in the query string. I often see this: $row = $db->query_first("
SELECT *
FROM " . TABLE_PREFIX . "table
WHERE id = '$_GET[id]'
");Not good.
Here is what you should use... $id = $vbulletin->input->clean_gpc('g', 'id', TYPE_UINT);
$row = $db->query_first("
SELECT *
FROM " . TABLE_PREFIX . "table
WHERE id = $id
"); The TYPE_UINT ensures that the variable is an integer, and is >= 0. You can also use $id in the rest of your application, more confidently. But wait... if they don't enter a valid integer here, it will just be 0! We can prevent that query from running in this case. Though no harm would be done, it is good to do as little processing as possible to keep things running fast.
$id = $vbulletin->input->clean_gpc('g', 'id', TYPE_UINT);
if (!$id)
{
standard_error('Please enter a valid product ID to view.');
}
$row = $db->query_first("
SELECT *
FROM " . TABLE_PREFIX . "table
WHERE id = $id
");That's better. The same concept applies to a lot of different types of data. If you expect them to enter an email address, then you can use the is_valid_email() function. If you expect them to enter one of a few specific answers, have those answers in an array and ensure that it matches. Example,<select name="meal">
<option value="1">Breakfast</option>
<option value="2">Lunch</option>
<option value="3">Dinner</option>
</select>$meal = $vbulletin->input->clean_gpc('p', 'meal', TYPE_UINT);
if (!in_array($meal, array(1, 2, 3))
{
standard_error('Invalid meal specified.');
}If you expect them to enter data in a certain pattern - for example, a md5 hash, then you can use a regular expression to match it. $hash = $vbulletin->input->clean_gpc('p', 'hash', TYPE_STR);
if (!preg_match('/^([a-f0-9]{32})$/', $hash))
{
standard_error('Please enter a valid md5 hash.');
}Not a great example, but you should really learn to use regular expressions to match the validity of data. An EXCELLENT tool to have. If you use the vBulletin datamanager for your custom application, be sure to use the error checking functionality in it. Very flexible.
Point: ensure the data you expect is what you expect. If it's not, do not let them continue. False data can do a number of things of varying damage:
Display a PHP error. Not a big deal, but it may disclose your script names and path. The less information they know about it the better.
Display SQL error. Injection, anyone?
Get into DB . It looks horrible to visitors when they see random junk on your website. Ohter side effects include: breaking search capabilities, false statistics, even DB errors down the road! Injection round two! SQL Injection
What is SQL injection? It's when your script does not properly clean user input, and they can actually manipulate the query using the input value.
Example,
A script to update a user's email address. Here is the query: UPDATE user SET email = '$value' WHERE userid = 1;You'd expect the user to enter an email address, but what if they don't? If the value they enter contains a single quote, it will end the string email value in the query, and it will be parsed as SQL. This is injection. What could happen from this? It really depends on the query, but it can be devastating. If I enter my@email.com', username = 'HAHA', password = ''#The final query is: UPDATE user SET email = 'my@email.com', username = 'HAHA', password = ''#' WHERE userid = 1;Since a # is a MySQL comment starter (one of the few ways), here is the query: UPDATE user SET email = 'my@email.com', username = 'HAHA', password = '';This affects the query in two ways:
Adds two new columns and values into the query
Removes the WHERE clauseIn English - wiped out your entire user table.
Preventing Injection
It is very easy to prevent injection. Generally, users either input text (string) or a number (int/float). For the latter, they should already be safe assuming you used the input cleaner as suggested. For strings, however, you must use the $db->escape_string() function. What does this do? In a standard MySQL setup, it's just a wrapper for mysql_real_escape_string (http://php.net/mysql_real_escape_string) which in simpliest terms, prepends a backslash to any characters which may alter query. Remember to use $db->escape_string() instead of the actual function, because not every board runs MySQL!
Automation
I highly recommend using a function to generate queries. This is how the datamanager works, and also many of my own private hacks. How does this make it more secure? - by not relying on the coder to escape it. If you do so, you can wrap each value in the $db->sql_prepare() function, which makes it SQL safe.
Here is what the function does to...
Strings, returns it escaped, and wrapped in single quotes
Numbers, returns them as is
Booleans, returns them in their integer form (1 and 0)
Anything else, returns it escaped, and wrapped in single quotesVery useful tool.
XSS (Cross Site Scripting)
Cross site scripting is when a user injects HTML into any text that he posts, and in that HTML uses Javascript to communicate with his own server. What can this do? He can steal your cookie information and take over your online identity. Imagine someone doing this to an administrative account. Uh oh!
Like SQL injection, XSS is actually very easy to prevent. You have two options:
1) Use TYPE_NOHTML instead of TYPE_STR when accepting string input.This has a few (minor) downsides. If you want to allow searching, or for any other reason want to retain the exact information the user enters. The alternative, is
2) Clean it using htmlspecialchars_uni() when displaying the text. This is where it gets tricky. If you display it all over the place, it can be difficult to remember to clean it every time.
So what does this do exactly? Any HTML characters, such as "<", ">", and "&" are parsed as HTML, so if you convert them to their entity form ("<", ">", "&", the browser knows to display that character, and not parse it as HTML. Now it is impossible for any hackers to inject HTML into your site.
Summary
Only accept input that is what you expect.
Escape strings when using them inside queries.
Convert all HTML characters to entities before displaying them.
If you have any further questions, you may contact me at vBlogetin.com (http://www.vblogetin.com/), or at SirAdrian.com (http://www.siradrian.com)
You may not copy this article without my permission.
Preparing Data
Whenever you accept input from a user, you must understand that you have no control over what is entered, period. The short version is that anything client-side can be manipulated. Even if you use Javascript or HTML limitations, they can all easily be avoided. This input includes but is not limited to:
Entered form input
Hidden form input (!)
URL query string
User AgentIn your application it is your responsibility to ensure the validity of the data they enter.
Use the vBulletin Input Cleaner!
I'm going to assume you know how to use it, even though sometimes you choose not to use it. If not, dig up a tutorial on it.
A common example is fetching a row from a table using an ID in the query string. I often see this: $row = $db->query_first("
SELECT *
FROM " . TABLE_PREFIX . "table
WHERE id = '$_GET[id]'
");Not good.
Here is what you should use... $id = $vbulletin->input->clean_gpc('g', 'id', TYPE_UINT);
$row = $db->query_first("
SELECT *
FROM " . TABLE_PREFIX . "table
WHERE id = $id
"); The TYPE_UINT ensures that the variable is an integer, and is >= 0. You can also use $id in the rest of your application, more confidently. But wait... if they don't enter a valid integer here, it will just be 0! We can prevent that query from running in this case. Though no harm would be done, it is good to do as little processing as possible to keep things running fast.
$id = $vbulletin->input->clean_gpc('g', 'id', TYPE_UINT);
if (!$id)
{
standard_error('Please enter a valid product ID to view.');
}
$row = $db->query_first("
SELECT *
FROM " . TABLE_PREFIX . "table
WHERE id = $id
");That's better. The same concept applies to a lot of different types of data. If you expect them to enter an email address, then you can use the is_valid_email() function. If you expect them to enter one of a few specific answers, have those answers in an array and ensure that it matches. Example,<select name="meal">
<option value="1">Breakfast</option>
<option value="2">Lunch</option>
<option value="3">Dinner</option>
</select>$meal = $vbulletin->input->clean_gpc('p', 'meal', TYPE_UINT);
if (!in_array($meal, array(1, 2, 3))
{
standard_error('Invalid meal specified.');
}If you expect them to enter data in a certain pattern - for example, a md5 hash, then you can use a regular expression to match it. $hash = $vbulletin->input->clean_gpc('p', 'hash', TYPE_STR);
if (!preg_match('/^([a-f0-9]{32})$/', $hash))
{
standard_error('Please enter a valid md5 hash.');
}Not a great example, but you should really learn to use regular expressions to match the validity of data. An EXCELLENT tool to have. If you use the vBulletin datamanager for your custom application, be sure to use the error checking functionality in it. Very flexible.
Point: ensure the data you expect is what you expect. If it's not, do not let them continue. False data can do a number of things of varying damage:
Display a PHP error. Not a big deal, but it may disclose your script names and path. The less information they know about it the better.
Display SQL error. Injection, anyone?
Get into DB . It looks horrible to visitors when they see random junk on your website. Ohter side effects include: breaking search capabilities, false statistics, even DB errors down the road! Injection round two! SQL Injection
What is SQL injection? It's when your script does not properly clean user input, and they can actually manipulate the query using the input value.
Example,
A script to update a user's email address. Here is the query: UPDATE user SET email = '$value' WHERE userid = 1;You'd expect the user to enter an email address, but what if they don't? If the value they enter contains a single quote, it will end the string email value in the query, and it will be parsed as SQL. This is injection. What could happen from this? It really depends on the query, but it can be devastating. If I enter my@email.com', username = 'HAHA', password = ''#The final query is: UPDATE user SET email = 'my@email.com', username = 'HAHA', password = ''#' WHERE userid = 1;Since a # is a MySQL comment starter (one of the few ways), here is the query: UPDATE user SET email = 'my@email.com', username = 'HAHA', password = '';This affects the query in two ways:
Adds two new columns and values into the query
Removes the WHERE clauseIn English - wiped out your entire user table.
Preventing Injection
It is very easy to prevent injection. Generally, users either input text (string) or a number (int/float). For the latter, they should already be safe assuming you used the input cleaner as suggested. For strings, however, you must use the $db->escape_string() function. What does this do? In a standard MySQL setup, it's just a wrapper for mysql_real_escape_string (http://php.net/mysql_real_escape_string) which in simpliest terms, prepends a backslash to any characters which may alter query. Remember to use $db->escape_string() instead of the actual function, because not every board runs MySQL!
Automation
I highly recommend using a function to generate queries. This is how the datamanager works, and also many of my own private hacks. How does this make it more secure? - by not relying on the coder to escape it. If you do so, you can wrap each value in the $db->sql_prepare() function, which makes it SQL safe.
Here is what the function does to...
Strings, returns it escaped, and wrapped in single quotes
Numbers, returns them as is
Booleans, returns them in their integer form (1 and 0)
Anything else, returns it escaped, and wrapped in single quotesVery useful tool.
XSS (Cross Site Scripting)
Cross site scripting is when a user injects HTML into any text that he posts, and in that HTML uses Javascript to communicate with his own server. What can this do? He can steal your cookie information and take over your online identity. Imagine someone doing this to an administrative account. Uh oh!
Like SQL injection, XSS is actually very easy to prevent. You have two options:
1) Use TYPE_NOHTML instead of TYPE_STR when accepting string input.This has a few (minor) downsides. If you want to allow searching, or for any other reason want to retain the exact information the user enters. The alternative, is
2) Clean it using htmlspecialchars_uni() when displaying the text. This is where it gets tricky. If you display it all over the place, it can be difficult to remember to clean it every time.
So what does this do exactly? Any HTML characters, such as "<", ">", and "&" are parsed as HTML, so if you convert them to their entity form ("<", ">", "&", the browser knows to display that character, and not parse it as HTML. Now it is impossible for any hackers to inject HTML into your site.
Summary
Only accept input that is what you expect.
Escape strings when using them inside queries.
Convert all HTML characters to entities before displaying them.
If you have any further questions, you may contact me at vBlogetin.com (http://www.vblogetin.com/), or at SirAdrian.com (http://www.siradrian.com)
You may not copy this article without my permission.