Monday, April 25, 2022

Right-to-Left Override RTLO Removal Tool for filenames #Windows #malware

Malware writers can trick you in 2 ways into thinking your file is a "PDF looking" file using the Right-to-Left Override (RTLO) technique.

“Right-to-Left Override” RTLO example

Firstly, maliciously constructed “.exe” can be built to display an PDF icon, so it looks like PDF default reader will open this file. If the filename is really long then, you can't see the extension (see image below). 

2ndly and may not be so obvious, malicious PDF filename is constructed as with a right-to-left override character is such a way that the file ends ".pdf" extension, but really is an ".exe".  


So in example below, the 2nd file looks like a ".txt" file, but is really a ".docx" file (the 1st file). The 1st file has been cleansed of the RTL Unicode character, and ends in ".docx". 

The PDF file is actually an ".exe" file, but looks like it will open with default PDF reader. 


Download RTLExample.7z ( it includes the above files with PDF ".exe" example. The files contain no viruses or malware. The PDF is safe ".exe", and just opens this page in Chrome). The "PDF" is safe ".exe", and just opens this page in Chrome. But GDrive marks these examples "Sorry, this file is infected with a virus", which good because they are detecting the RTL character and exe. But it a false positive, since there is no virus in the files. You can create you own examples by inserting the RTL character into the filename, see this video https://youtu.be/n2kV3Q2eTCY). 


Here's the same files as viewed from the command (cmd.exe) line. The box character represents the RTL character.


Note: Detection of malicious file is never done by a filename alone, so a good antivirus will flag the contents of this file, for known signatures. BUT you can remove the annoying RTL character with the free tool below! 

How is RTLO being abused by malware writers?

In apps that support Unicode like Window Explorer, the right-to-left override malware method uses  a RTL Unicode character, that will reverse the order of the characters that follow it. It's used mainly for Middle Eastern/Asian languages that you read right-to-left.

RTLO can be used to spoof fake extensions. To do this we need a hidden RTL Unicode character in the file name.

What is “Right-to-Left Override” RTLO?
The RTLO method is used to hide the true type of a file, so it might trick you into open text file (.txt) which really is a Word file (.docx) with malicious malware. More recently this file could hide a .wav file. Audio files such .wav file are being embedded with malware, is on the forefront of malware maliciousness. Read about that on my post here.
The method exploits a feature built into Windows Explorer. Since Microsoft Windows does a great job of supporting different languages from around the world, some of those languages that are written from right-to-left (RTL). 
Let’s say you want to use a right-to-left written language, like Hebrew or Arabic, on a site combined with a left-to-right written language like English or French. In this case, you would want bidirectional script support.
Bidirectional script support is the capability of a computer system to correctly display bi-directional text. In HTML we can use Unicode right-to-left marks and left-to-right marks to override the HTML bidirectional algorithm when it produces undesirable results:
left-to-right mark: ‎ (U+200E) Unicode character
right-to-left mark: ‏ (U+200F) Unicode character
How do you fix files that have the RTLO or other bad characters ? 

Here's a tool I built to clean up Right-to-Left Mark (and many others) and Unicode Control Characters from your files. It's super fast, small and written in native C++.

Updated Thu 21-Apr-22 - new build, fixed many recursive issues




Download
 touchRTL.7z (you need https://www.7-zip.org/ to unpack). For personal use only. Copy into c:\windows to use from cmd.exe.


License : 

touchRTL.7z personal use only, for commercial use buy touchLTRPRO. Contact as validated today available for license request. 

touchRTLPRO.7z, has flags to remove Unicode spaces and punctuations (math symbols, currency, open closing braces, and accent marks).  

Just run this command and it will recursively rename filenames to remove those characters under the specified directory name. If directory name, contains spaces you need quotes.


touchRTL -v -R -l -y "directory name"

where


Usage: touchRTL [-aclmpRuvxy] [-r REFFILE | -d DATETIME] PATH...

UNIX touch mimic, updates files access, modification and creation times of file(s) in PATH to the current time,
If PATH argument does not exist, creates corresponding new empty file or directory (using -y), unless -c or --n
Supports directory recursion and time stamping!
Supports Right-to-Left (RTL) character removal for files.
PATH argument can represent a filename(s) or directory. Double quote if it contains spaces. eg "c:\as is.txt"

  -a, --access-time        change only the file access time
  -c, --no-create          do not create any new files - If the file exists, touch will update the access time,
  -l, --RTL                remove Unicode control & format characters (esp. infamous right-to-left) from filena
  -m, --modif-time         change only the file modification time
  -p, --pause-exit         pause on exit (non-GNU extra)
  -R, --recursive          recursively touch files in specified directory and all subdirectories (non-GNU extra
  -u, --unicntrl           remove Unicode control characters only - https://www.fileformat.info/info/unicode/ca
  -v, --verbose            output the result of every file processed (non-GNU extra)
  -x, --creation-time      change only the file creation time (non-GNU extra)
  -y, --directory          specify directory, instead of default file
  -r, --reference REFFILE  use this file's times instead of current time
  -s, --spaces (PRO edtn)  remove Unicode spaces from filename
  -!, --puncs  (PRO edtn)  remove Unicode punctuations & symbols (math & modifiers) from filename

  -d, --date DATETIME      use YYYY-MM-DDThh:mm:ss[.ms] instead of current time (non-GNU, does not parse string
                           accepted "2033-04-01T07:07:07", "2033-04-01 07:07:07"

  -h, --help               Display this help and exit.

      --version            Display version information and license information.

For personal use only. Commercial license required for business use and removes page open. See --version for al
Copyright © 2019-2022 M. Pahulje <metadataconsult@gmail.com> - https://http://metadataconsulting.blogspot.com/

No comments:

Post a Comment