Cracking PDF Hashes with hashcat
Do you have a PDF document lying around somewhere, but it's encrypted and you've forgotten the password to it?
hashcat is a great open-source hash cracker with GPU acceleration. It also comes with features such as masking, dictionary attacks and even statistical methods of password guessing. It also supports 300+ hash types (e.g. SHA-1, MD5, WPA, Django...) out of the box.
Don't use this for unlawful purposes!
Extracting the hash from the PDF
First, we need to extract the password hash from the PDF. There's an online tool for this, but for now let's use this python script (pdf2john.py).
Run it with python pdf2john.py <your-pdf-file>.
The output should resemble something like this:
We're only interested in the second part (beginning with $pdf$4... up till ...69d6). So, let's copy that out:
This is the hash we'll supply to hashcat later on.
Starting brute force with hashcat
First, grab the latest copy of hashcat from here.
This command runs a brute force attack on the hash (up till the maximum number of characters):
hashcat -a 3 -m 10500 '<hash>' (note: the hash must be in quotes, or else some OSes might interpret the $ as a variable)
Let's break it down.
-a specifies the attack mode. In this case, 3 indicates brute force.
-m specifies the type of hash. hashcat can actually autodetect the hash type, but for this purpose, we'll specify it as as 10500, which is PDF 1.4 - 1.6 (Acrobat 5 - 8).
hashcat also supports masking options. So if you wanted to try all lowercase alphanumeric passwords (a-z, 0-9) up to 10 characters, you could do
.\hashcat -a 3 -1 ?l?d -i -m 10500 '<hash>' ?1?1?1?1?1?1?1?1?1?1?1?1
-1 specifies the character set in the first position (hashcat supports multiple character sets). In this case, ?l refers to the set abcde...xyz, and ?d refers to all digits 0-9.
-i specifies that we want to progressively try the mask, starting from --increment-min (default 1). What this does is it will try only 1 character of the mask (e.g. a), then 2, up till --increment-max (default being the length of the mask, e.g. zzzzzzzzzz).
?1?1?1?1?1?1?1?1?1?1?1?1 is the mask itself. ?1 refers to the character set in the first position, that we specified above with -1.
And that's it! Password cracking is exponentially slower with regards to the length of the password1, so any clues as to the content of the password will speed it up greatly.
-
Theoretically, a 51-character alphanumeric, case-sensitive password would be uncrackable. ↩