I have a php script in a folder (I call it the root folder). The script can basically list all files in subfolders of this root folder. The user can specify which subfolder should be displayed by using GET-parameters.
script.php?foo
would display the content of
<root folder>/foo/
and
script.php?.bar
would display the content of
<root folder>/.bar/
However, users could also "cheat" and use commands like /.. to display the content of folders they souldn't be able to see.
For example with
script.php?/../..
the users could get very high in the folder hierarchy.
Do you have an idea how to prevent users of doing "cheats" like this.
For reason of simplicity, let's say the GET-parameter is stored in $searchStatement.
You could use realpath to resolve the relative path to an absolute one and then check if that path begins with your "root" folder's path:
$absolutePath = realpath(__DIR__ . '/' . trim($searchStatement, '/'));
if (strpos($absolutePath, __DIR__ .'/') !== 0) {
die('Access denied.');
}
You just should validate the input before you use it.
For example you might want to only allow the characters a-z and / to allow subdirectories. Probably you want to allow the . as well. If you make this subset small, it's easy to validate if the input is allowed or not by the allowed characters already.
At the moment you allow ., as you have noticed, you have the problem that relative paths could be created like /../../ which could be used for directory traversal attacks.
To validate if a string contains only characters of a specific range, you can validate this with a regular expression or the filter functions. If your website does not need to allow any relative path parts you can look if they exist in the path to validate the input:
$valid = !array_intersect(array('', '.', '..'), explode('/', $path));
Valid will be FALSE if there is any // or /./ or /../ part inside the path.
If you need to allow relative paths, realpath has already been suggested, so to query the input against your directory structure first. I would only use it as last resort as it is relatively expensive, but it's good to know about.
However you can resolve the string your own as well with some simple function like the following one:
/**
* resolve path to itself
*
* #param string $path
* #return string resolved path
*/
function resolvePath($path)
{
$path = trim($path, '/');
$segmentsIn = explode('/', $path);
$segmentsOut = array();
foreach ($segmentsIn as $in)
{
switch ($in)
{
case '':
$segmentsOut = array();
break;
case '.':
break;
case '..';
array_pop($segmentsOut);
break;
default:
$segmentsOut[] = $in;
}
}
return implode('/', $segmentsOut);
}
Usage:
$tests = array(
'hello',
'world/.',
'../minka',
'../../42',
'../.bar',
'../hello/path/./to/../../world',
);
foreach($tests as $path)
{
printf("%s -> %s\n", $path, resolvePath($path));
}
Output:
hello -> hello
world/. -> world
../minka -> minka
../../42 -> 42
../.bar -> .bar
../hello/path/./to/../../world -> hello/world
I can only suggest you first validate the input based on it's own data before letting touch it the filesystem, even through realpath.
Have a look at the chroot function:
bool chroot ( string $directory )
Changes the root directory of the current process to directory, and changes the current working directory to "/".
A call to that method prevents further access to files outside of the current directory.
Note however that requires root privileges.
Have you tried something with realpath, it should resolve all the /.. in your path. By testing the realpath of the arguments against your current path like:
substr($realpath, 0, strlen('/basepath/cant/go/above')) === '/basepath/cant/go/above'
you make sure that any /.. havent escaped from where you want.
Related
I have a base path /whatever/foo/
and
$_GET['path'] should be relative to it.
However how do I accomplish this (reading the directory), without allowing directory traversal?
eg.
/\.\.|\.\./
Will not filter properly.
Well, one option would be to compare the real paths:
$basepath = '/foo/bar/baz/';
$realBase = realpath($basepath);
$userpath = $basepath . $_GET['path'];
$realUserPath = realpath($userpath);
if ($realUserPath === false || strpos($realUserPath, $realBase) !== 0) {
//Directory Traversal!
} else {
//Good path!
}
Basically, realpath() will resolve the provided path to an actual hard physical path (resolving symlinks, .., ., /, //, etc)... So if the real user path does not start with the real base path, it is trying to do a traversal. Note that the output of realpath will not have any "virtual directories" such as . or .....
ircmaxell's answer wasn't fully correct. I've seen that solution in several snippets but it has a bug which is related to the output of realpath(). The realpath() function removes the trailing directory separator, so imagine two contiguous directories such as:
/foo/bar/baz/
/foo/bar/baz_baz/
As realpath() would remove the last directory separator, your method would return "good path" if $_GET['path'] was equal to "../baz_baz" as it would be something like
strpos("/foo/bar/baz_baz", "/foo/bar/baz")
Maybe:
$basepath = '/foo/bar/baz/';
$realBase = realpath($basepath);
$userpath = $basepath . $_GET['path'];
$realUserPath = realpath($userpath);
if ($realUserPath === false || strcmp($realUserPath, $realBase) !== 0 || strpos($realUserPath, $realBase . DIRECTORY_SEPARATOR) !== 0) {
//Directory Traversal!
} else {
//Good path!
}
It is not sufficient to check for patterns like ../ or the likes. Take "../" for instance which URI encodes to "%2e%2e%2f". If your pattern check happens before a decode, you would miss this traversal attempt. There are some other tricks hackers can do to circumvent a pattern checker especially when using encoded strings.
I've had the most success stopping these by canonicalizing any path string to its absolute path using something like realpath() as ircmaxwell suggests. Only then do I begin checking for traversal attacks by matching them against a base path I've predefined.
You may be tempted to try and use regex to remove all ../s but there are some nice functions built into PHP that will do a much better job:
$page = basename(realpath($_GET));
basename - strips out all directory information from the path e.g. ../pages/about.php would become about.php
realpath - returns a full path to the file e.g. about.php would become /home/www/pages/about.php, but only if the file exists.
Combined they return just the files name but only if the file exists.
When looking into the creation of new files or folders, I've figured I can use a two stage approach:
First check for traversal attempts using a custom implementation of a realpath() like function, which however works for arbitrary paths, not just existing files. There's a good starting point here. Extend it with urldecode() and whatever else you think may worth checking.
Now using this crude method you can filter out some traversal attempts, but it may be possible that you miss some hackish combination of special characters, symlinks, escaping sequences etc. But since you know for sure the target file does not exist (check using file_exists) noone can overwrite anything. The worst case scenario would be that someone can get your code creating a file or folder somewhere, which may be an acceptable risk in most cases, provided your code does not allow them to write into that file/folder straight away.
Finally so the path now points to an existing location, therefore you can now do the proper check using the methods suggested above utilising realpath(). If at this point it turns out a traversal has happened, you are still safe more or less, as long as you make sure to prevent any attempts writing into the target path. Also right now you can delete the target file/dir and say it was a traversal attempt.
I'm not saying it cannot be hacked, since after all still it may allow illegitimate changes to be done to the FS, but still better than only doing custom checks, that cannot utilise realpath(), and the window for abuse left open by making a temporary and empty file or folder somewhere is lower, than allowing them to make it permanent and even write into it, as it would happen with only a custom check that may miss some edge cases.
Also correct me if I'm wrong pls!
I have written a function to check for traversal:
function isTraversal($basePath, $fileName)
{
if (strpos(urldecode($fileName), '..') !== false)
return true;
$realBase = realpath($basePath);
$userPath = $basePath.$fileName;
$realUserPath = realpath($userPath);
while ($realUserPath === false)
{
$userPath = dirname($userPath);
$realUserPath = realpath($userPath);
}
return strpos($realUserPath, $realBase) !== 0;
}
This line alone if (strpos(urldecode($fileName), '..') !== false) should be enough to prevent traversal, however, there are many different ways hackers can traverse directories so its better to make sure the user starts with the real base path.
Just checking the user starts with the real base path is not enough because a hacker could traverse to the current directory and discover the directory structure.
The while allows the code to work when $fileName does not exist.
1
put a null index.htm for -Index block
2
filter sQS on start
// Path Traversal Attack
if( strpos($_SERVER["QUERY_STRING"], "../") ){
exit("P.T.A. B-(");
}
I have a base path /whatever/foo/
and
$_GET['path'] should be relative to it.
However how do I accomplish this (reading the directory), without allowing directory traversal?
eg.
/\.\.|\.\./
Will not filter properly.
Well, one option would be to compare the real paths:
$basepath = '/foo/bar/baz/';
$realBase = realpath($basepath);
$userpath = $basepath . $_GET['path'];
$realUserPath = realpath($userpath);
if ($realUserPath === false || strpos($realUserPath, $realBase) !== 0) {
//Directory Traversal!
} else {
//Good path!
}
Basically, realpath() will resolve the provided path to an actual hard physical path (resolving symlinks, .., ., /, //, etc)... So if the real user path does not start with the real base path, it is trying to do a traversal. Note that the output of realpath will not have any "virtual directories" such as . or .....
ircmaxell's answer wasn't fully correct. I've seen that solution in several snippets but it has a bug which is related to the output of realpath(). The realpath() function removes the trailing directory separator, so imagine two contiguous directories such as:
/foo/bar/baz/
/foo/bar/baz_baz/
As realpath() would remove the last directory separator, your method would return "good path" if $_GET['path'] was equal to "../baz_baz" as it would be something like
strpos("/foo/bar/baz_baz", "/foo/bar/baz")
Maybe:
$basepath = '/foo/bar/baz/';
$realBase = realpath($basepath);
$userpath = $basepath . $_GET['path'];
$realUserPath = realpath($userpath);
if ($realUserPath === false || strcmp($realUserPath, $realBase) !== 0 || strpos($realUserPath, $realBase . DIRECTORY_SEPARATOR) !== 0) {
//Directory Traversal!
} else {
//Good path!
}
It is not sufficient to check for patterns like ../ or the likes. Take "../" for instance which URI encodes to "%2e%2e%2f". If your pattern check happens before a decode, you would miss this traversal attempt. There are some other tricks hackers can do to circumvent a pattern checker especially when using encoded strings.
I've had the most success stopping these by canonicalizing any path string to its absolute path using something like realpath() as ircmaxwell suggests. Only then do I begin checking for traversal attacks by matching them against a base path I've predefined.
You may be tempted to try and use regex to remove all ../s but there are some nice functions built into PHP that will do a much better job:
$page = basename(realpath($_GET));
basename - strips out all directory information from the path e.g. ../pages/about.php would become about.php
realpath - returns a full path to the file e.g. about.php would become /home/www/pages/about.php, but only if the file exists.
Combined they return just the files name but only if the file exists.
When looking into the creation of new files or folders, I've figured I can use a two stage approach:
First check for traversal attempts using a custom implementation of a realpath() like function, which however works for arbitrary paths, not just existing files. There's a good starting point here. Extend it with urldecode() and whatever else you think may worth checking.
Now using this crude method you can filter out some traversal attempts, but it may be possible that you miss some hackish combination of special characters, symlinks, escaping sequences etc. But since you know for sure the target file does not exist (check using file_exists) noone can overwrite anything. The worst case scenario would be that someone can get your code creating a file or folder somewhere, which may be an acceptable risk in most cases, provided your code does not allow them to write into that file/folder straight away.
Finally so the path now points to an existing location, therefore you can now do the proper check using the methods suggested above utilising realpath(). If at this point it turns out a traversal has happened, you are still safe more or less, as long as you make sure to prevent any attempts writing into the target path. Also right now you can delete the target file/dir and say it was a traversal attempt.
I'm not saying it cannot be hacked, since after all still it may allow illegitimate changes to be done to the FS, but still better than only doing custom checks, that cannot utilise realpath(), and the window for abuse left open by making a temporary and empty file or folder somewhere is lower, than allowing them to make it permanent and even write into it, as it would happen with only a custom check that may miss some edge cases.
Also correct me if I'm wrong pls!
I have written a function to check for traversal:
function isTraversal($basePath, $fileName)
{
if (strpos(urldecode($fileName), '..') !== false)
return true;
$realBase = realpath($basePath);
$userPath = $basePath.$fileName;
$realUserPath = realpath($userPath);
while ($realUserPath === false)
{
$userPath = dirname($userPath);
$realUserPath = realpath($userPath);
}
return strpos($realUserPath, $realBase) !== 0;
}
This line alone if (strpos(urldecode($fileName), '..') !== false) should be enough to prevent traversal, however, there are many different ways hackers can traverse directories so its better to make sure the user starts with the real base path.
Just checking the user starts with the real base path is not enough because a hacker could traverse to the current directory and discover the directory structure.
The while allows the code to work when $fileName does not exist.
1
put a null index.htm for -Index block
2
filter sQS on start
// Path Traversal Attack
if( strpos($_SERVER["QUERY_STRING"], "../") ){
exit("P.T.A. B-(");
}
I have a directory and a dynamic file name. Are there any functions I can use to make sure it is impossible to break out from the specified directory?
for example:
secure_path('/tmp/', 'file.txt') -> /tmp/file.txt
secure_path('/tmp/', '../file.txt') -> /tmp/file.txt
If you're only working in the single directory and not subdirectories below you could do
$file = "/tmp/".basename($input);
Which should give you the filename at the end of any path given in input, and append it to the directory you wanted.
Use realpath() to get a canonic form of the path, and eliminate any ../-s. Then verify that it contains the full path of the directory you want to contain it to.
Such as:
$path = "/tmp/file.txt";
$directory = "{full path of your files directory}";
if (strpos(realpath($path), $directory) === 0) {
// path OK
}
else {
// path not ok
}
I don't really know how ../file.txt is supposed to turn into /tmp/file.txt in your example, but to check whether a path is a subpath of another, use realpath with a simple comparison:
$path = '../foo';
$allowedPath = '/tmp/';
$path = realpath($path);
if (substr($path, 0, strlen($allowedPath)) !== $allowedPath) {
// path is not within allowed path!
}
I have a base path /whatever/foo/
and
$_GET['path'] should be relative to it.
However how do I accomplish this (reading the directory), without allowing directory traversal?
eg.
/\.\.|\.\./
Will not filter properly.
Well, one option would be to compare the real paths:
$basepath = '/foo/bar/baz/';
$realBase = realpath($basepath);
$userpath = $basepath . $_GET['path'];
$realUserPath = realpath($userpath);
if ($realUserPath === false || strpos($realUserPath, $realBase) !== 0) {
//Directory Traversal!
} else {
//Good path!
}
Basically, realpath() will resolve the provided path to an actual hard physical path (resolving symlinks, .., ., /, //, etc)... So if the real user path does not start with the real base path, it is trying to do a traversal. Note that the output of realpath will not have any "virtual directories" such as . or .....
ircmaxell's answer wasn't fully correct. I've seen that solution in several snippets but it has a bug which is related to the output of realpath(). The realpath() function removes the trailing directory separator, so imagine two contiguous directories such as:
/foo/bar/baz/
/foo/bar/baz_baz/
As realpath() would remove the last directory separator, your method would return "good path" if $_GET['path'] was equal to "../baz_baz" as it would be something like
strpos("/foo/bar/baz_baz", "/foo/bar/baz")
Maybe:
$basepath = '/foo/bar/baz/';
$realBase = realpath($basepath);
$userpath = $basepath . $_GET['path'];
$realUserPath = realpath($userpath);
if ($realUserPath === false || strcmp($realUserPath, $realBase) !== 0 || strpos($realUserPath, $realBase . DIRECTORY_SEPARATOR) !== 0) {
//Directory Traversal!
} else {
//Good path!
}
It is not sufficient to check for patterns like ../ or the likes. Take "../" for instance which URI encodes to "%2e%2e%2f". If your pattern check happens before a decode, you would miss this traversal attempt. There are some other tricks hackers can do to circumvent a pattern checker especially when using encoded strings.
I've had the most success stopping these by canonicalizing any path string to its absolute path using something like realpath() as ircmaxwell suggests. Only then do I begin checking for traversal attacks by matching them against a base path I've predefined.
You may be tempted to try and use regex to remove all ../s but there are some nice functions built into PHP that will do a much better job:
$page = basename(realpath($_GET));
basename - strips out all directory information from the path e.g. ../pages/about.php would become about.php
realpath - returns a full path to the file e.g. about.php would become /home/www/pages/about.php, but only if the file exists.
Combined they return just the files name but only if the file exists.
When looking into the creation of new files or folders, I've figured I can use a two stage approach:
First check for traversal attempts using a custom implementation of a realpath() like function, which however works for arbitrary paths, not just existing files. There's a good starting point here. Extend it with urldecode() and whatever else you think may worth checking.
Now using this crude method you can filter out some traversal attempts, but it may be possible that you miss some hackish combination of special characters, symlinks, escaping sequences etc. But since you know for sure the target file does not exist (check using file_exists) noone can overwrite anything. The worst case scenario would be that someone can get your code creating a file or folder somewhere, which may be an acceptable risk in most cases, provided your code does not allow them to write into that file/folder straight away.
Finally so the path now points to an existing location, therefore you can now do the proper check using the methods suggested above utilising realpath(). If at this point it turns out a traversal has happened, you are still safe more or less, as long as you make sure to prevent any attempts writing into the target path. Also right now you can delete the target file/dir and say it was a traversal attempt.
I'm not saying it cannot be hacked, since after all still it may allow illegitimate changes to be done to the FS, but still better than only doing custom checks, that cannot utilise realpath(), and the window for abuse left open by making a temporary and empty file or folder somewhere is lower, than allowing them to make it permanent and even write into it, as it would happen with only a custom check that may miss some edge cases.
Also correct me if I'm wrong pls!
I have written a function to check for traversal:
function isTraversal($basePath, $fileName)
{
if (strpos(urldecode($fileName), '..') !== false)
return true;
$realBase = realpath($basePath);
$userPath = $basePath.$fileName;
$realUserPath = realpath($userPath);
while ($realUserPath === false)
{
$userPath = dirname($userPath);
$realUserPath = realpath($userPath);
}
return strpos($realUserPath, $realBase) !== 0;
}
This line alone if (strpos(urldecode($fileName), '..') !== false) should be enough to prevent traversal, however, there are many different ways hackers can traverse directories so its better to make sure the user starts with the real base path.
Just checking the user starts with the real base path is not enough because a hacker could traverse to the current directory and discover the directory structure.
The while allows the code to work when $fileName does not exist.
1
put a null index.htm for -Index block
2
filter sQS on start
// Path Traversal Attack
if( strpos($_SERVER["QUERY_STRING"], "../") ){
exit("P.T.A. B-(");
}
I want to make sure a file path set via query string does not go outside of the desired subdirectory. Right now, I am checking that:
The path does not start with "/", to prevent the user from giving an absolute path.
The path does not contain "..", to prevent the user from giving a path that is outside of the desired subdirectory.
The path does not contain ":", to prevent the use of a url (i.e. "http://", "ftp://", etc.). Should I ever run this script on a Windows server (not likely), this will also prevent absolute paths beginning with a drive specifier (i.e. "C:\"). Note: I'm aware that a colon is a valid character in a Unix filenames, but I will never be using it in a filename.
The path does not start with "\". Just in case I change my mind about running on a Windows server, this prevents Windows network paths from being specified (i.e. "\\someserver\someshare"). Again, I'm aware that a backslash is a valid Unix filename character, but I also won't be using it in any filenames.
Are these checks sufficient?
Background
I have a PHP script that takes (via query string) the path to a sample source file to be shown to a user. So I might give them a link like "view_sample.php?path=accounting_app/report_view.php" or "view_sample.php?path=ajax_demo/get_info.js".
The script looks basically like this:
$path = $_GET['path'];
if(path_is_valid($path) && is_file("sample/$path"))
{
header('Content-Type: text/plain');
readfile("sample/$path");
}
My concern is that a malicious user would see the url and try to do something like "view_sample.php?path=../../database/connection_info.php" and gain access to a file which is not in the "sample" directory.
Are the four checks I defined above (which would be implemented in the path_is_valid() function) sufficient to lock out a malicious user? (Also, I think checks 1, 3, and 4 are basically irrelevant since I am prepending a relative path, but if I didn't do this would the checks be sufficient?)
Call
$path = realpath("sample/$path");
Then check that the resulting path starts with the directory you're expecting.
<?php
// Current path information
$path = $_GET['path'];
$vroot = "sample";
// Validate that the $path is a subfolder of $vroot
$vroot = realpath($vroot);
if(substr(realpath($path), 0, strlen($vroot)) != $vroot or !is_dir($path)) {lid!
exit("Invalid path");
} else {
echo "Ah, everything is alright!";
}
?>
The use of realpath should not change the path, so I use it in the following way:
function checkPath($pathToCheck) {
global $basepath;
$fullpath = $basepath.'/'.$pathToCheck;
if ($fullpath==realpath($fullpath) && is_dir($fullpath)) {
return $fullpath;
} else {
error_die('path not allowed: '.htmlentities($pathToCheck));
}
}