john-dev - Re: gpg2john (was: Work needed before Jumbo-8)

Follow @Openwall on Twitter for new release announcements and other news

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <062c7c2d001d6fde2ffa9d469e98c2e9@smtp.hushmail.com>
Date: Tue, 12 Mar 2013 21:01:23 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: gpg2john (was: Work needed before Jumbo-8)

On 12 Mar, 2013, at 4:01 , Lukas Odzioba <lukas.odzioba@...il.com> wrote:
> 2013/3/5 magnum <john.magnum@...hmail.com>:
>> We need someone to fix gpg2john (c++ != magnum) so it can process multiple infiles and more importantly, it must process all keys in each input file. The latter is a must IMO - at least it must emit a warning that more keys are present. Any volunteer?
> 
> Patch attached. Changes I made are pretty trivial, before me someone
> else merged and added some code to make gpg2john.c

Thanks! It works fine with my multiple keys file too. I'll commit this and I think it is barely Good Enough[tm] for Jumbo 8 now, but some things can be improved:

1. Supplying a bad file (not a key file) results in no output at all.
2. Supplying a public key file, same lack of output.
3. The full path is used for the login field.

For 1 and 2, we might want to add some warning or other output to stderr. For 3 we should definitely use the basename and ideally also strip the .gpg or .pgp extension (are .pgp files supported at all?).

> Sometimes encoded file may contain additional info about user (packet
> type 13), it looks like that:
> 
> Old: User ID Packet(tag 13)(47 bytes)
>        User ID - Random User (Just for test) <random@...dom.com>
> 
> We have got user name, comment and an email. When gpg file contains
> multiple keys it can have multiple identities.
> Currently gpg2john formats output that way:
> 	printf("%s:$gpg$*%d*%d*%d*", filename, key.m_algorithm,
> key.m_datalen, key.bits());
> In my opinion we should add an email too.

This is not a blocker for Jumbo-8 but it would be excellent if someone can fix it: We should definitely try to find this UserID field and put in the GECOS field of our output (as is - it's stored as one string and it's not trivial to split it canonically).

> Should we somehow modify copyrights?

I believe you could just add lines after the existing Copyright line:

 /*
  * pgpry - PGP private key recovery
  * Copyright (C) 2010 Jonas Gehring
+ * Modified for John the Ripper:
+ * Copyright (C) 2012 Dhiru (...)
+ * Copyright (C) 2013 Lukasz (...)
  *
  * This program is free software: you can redistribute it and/or modify	 
  * it under the terms of the GNU General Public License as published by

magnum

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.