Re: Fw: On e-mail sent with header: Content-Transfer-Encoding:
John D. Hardin
jhardin at impsec.org
Tue Feb 3 06:14:26 PST 2004
On Tue, 3 Feb 2004, Min-Soo Kim wrote:
> As it looks like you're blocking my server's IP, I try to reach
> Feb 3 15:34:51 jikji-home sendmail: i136YXbF014664:
> to=<esd-l at spconnect.com>, ctladdr=<minsukim at jikji.org> (1000/0),
> delay=00:00:15, xdelay=00:00:15, mailer=esmtp, pri=34112,
> relay=merlin2.spconnect.com. [188.8.131.52], dsn=4.0.0,
> stat=Deferred: 452 4.2.1 mail i136YpP28780 from 184.108.40.206
> temporary greylist embargoed
You'll have to talk to the list owner. I don't administer the mailing
I'll CC the list so that he sees this.
> ----- Original Message -----
> From: "Min-Soo Kim" <minsukim at jikji.org>
> Yes, I'm a Korean, but I really hate spams all over the world, and
> I'm using your Sanitizer and spambnc quite successfully to block
> spam e-mails among other tools I've been used.
I maintain the sanitizer, and it's not intended to be an antispam
tool. What is "spambnc"? I don't recognize it.
> allocated by international organization, but also because we
> cannot block UTF-7/UTF-8 encoding from each other for better
> communication in near future. It's not a matter of whether one
> country is notorious for sending / relaying spams, but how we're
> going to prevent myself / my kids / my friend / my neighbor being
> exposed to that 4-letter spammers and threir e-mails is what
> matters and is more important.
You should be able to check for specific text strings in procmail
regardless of their character set. You just need to be careful if the
character maps to an ASCII character that is syntactically meaningful
Scanning for text in a base-64 encoded body is more difficult,
> I found that Sanitizer is quite useful tool to make the goal I had
> set more realistic.
> Here is my most recent problem. (I'm not a programmer, but an
> end-user) With this mail header below, images - entire HTML
> contents - can be seen and not filtered, and the continuing
> problem is I have to use Korean encoding for my daily life and
> should accept this transfer-endocing according to
> http://support.microsoft.com/default.aspx?scid=kb;EN-US;323489 .
> God feeling is this is not a proper e-mail, but I do not know how
> to block them.
> > E-mail header is like this:
> MIME-Version: 1.0
> X-Security: MIME headers sanitized on jikji-home.jikji.org
> See http://www.impsec.org/email-tools/sanitizer-intro.html
> for details. $Revision: 1.139 $Date: 2003-09-07 10:14:23-07
> Content-Type: multipart/mixed;boundary= "----=_NextPart_000_0058_18DCD15F.9AF77D72"
> X-Priority: 3
> X-MSMail-Priority: Normal
> X-Mailer: Microsoft Outlook Express 6.00.2462.0000
> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2462.0000
> X-SpamBouncer: 1.8 (1/26/04)
> X-SBPass: No Freemail Filtering
> X-SBScore: 0 (Spam Threshold: 5) (Block Threshold: 2)
> X-SBClass: OK
> X-Folder: Default
> X-UIDL: )n7!!3h`!!dY`"!<0T!!
> Content-Type: text/html; charset= "ks_c_5601-1987"
> Content-Transfer-Encoding: base64
Yeah, base64-encoded plain text and HTML text is an annoyance. Add to
that a non-english character set...
I think your problem is going to be resolved within spambnc. I would
wager that it is capable of decoding the message body and scanning it
for spam content (SpamAssassin is). The problem is it probably doesn't
know how to recognize spam in Korean! You may be able to help out here
if you can work with the spambnc people to develop recognition strings
for Korean-language spams. Unfortunately this may require some
programming skills in addition to being literate in Korean...
> I've searched to get some hint, and what I found is as follows,
> 1. http://news.spamcop.net/pipermail/spamcop-list/2003-February/031732.html
> "The messages _do_ contain valid HTML, and when I remove the header line,
> the message is correctly parsed."
Which header line are they referring to?
> 2. http://wpbl.pc9.org/procmailrc
> "# The following will catch Mimail.Q, MiMail.R, Mydoom, Novarg, Shimg
> # and automatically drop them after recording the sender for WPBL
> # Last updated 2004-01-27 21:30 CST"
...which is looking for specific strings in the base64 body,
essentially a very lightweight antivirus tool.
> While I'm not sure whether this procmailrc is the one that blocks
> the spam mail I'm getting, I put this recipe into my .procmailrc
> with a hope, and now I'm wondering if you could make this thing
> working with existing local-rules.procmail, as I said before, I'm
> not goot at hacking.
As far as email worms ar concerned I'm not sure it's needed. The
reason the sanitizer doesn't block certain worms is that they aren't
using bare executable file attachments, but instead are wrapping the
attack in a .ZIP file. Since historically this has been the proper
method to get a legitimate file past an email security system, we now
have a problem: do we start considering all .ZIP files as suspicious,
or do we attempt to identify specific evil .ZIP attachments?
The local-rules is an attempt to do the latter. If at some point
worms' usage of .ZIP files becomes unmanageable through local rules,
then I will add .ZIP to the list of extensions to mangle. You can also
add "ZIP" to your local list of mangled extensions if you feel that is
an appropriate security policy for your site.
I hope to be able to add .ZIP attachment unwrapping per a suggestion
made earlier, which may help as well.
But, note: The sanitizer is not an anti-spam tool and not an antivirus
tool, and I will resist attempts to make it either.
> It's been a long mail, but the point is simple.
My response was longer... :)
> 1. Thank you so much Mr. John Hardin for the Sanitizer. You
> saved a lot of my energy for better use.
You're very welcome.
> 2. Appreciate if you could add Mimail.Q MiMail.R Mydoom, Shimg
> into your distribution. I could send you some e-mails if you need
The .ZIP variants of those attacks are the subject of the discussion
above. Perhaps it should be discussed specifically on-list.
As for spam detection on base-64-encoded text, there are several ways
1) write a tool that decodes base64-encoded text body parts. I has
considered doing this as a "sanitizing" step in the next generation of
the sanitizer, but I have not carefully explored the side effects.
There may be situations where base64-encoding text is *critical* to
reliable delivery of that text. Anybody have and comments or examples
Anyway, if you did this, then non-base64-aware text scanners (e.g.
procmail rules) would be able to see the message body.
2) decode the body in a separate file, scan it for spamminess, and
assign the result back to the original message. (Note how I leave
"scan it for spamminess" as a black box :)
3) have procmail scan for the encoded spammy strings in the message
body. This will be less obvious (what word of phrase does that string
of gibberish represent?) and less efficient (one phrase will generate
3 (I think) search strings based on the byte alignment of the
encoding). It's also more difficult to generate the search
Free project idea for the list: write a tool that will take a simple
text string (or list thereof) and generate procmail search expressions
that will detect that string in an encoded text body part. Extra
credit: support simple regular expressions. :)
The all you'd need to do would be to type all your nasty words into a
text file using your character set of choice, run the tool over it,
and put the result into your /etc/procmail/antispam.procmail file.
> With regards, Min-Soo Kim.
Regards. Sorry that you're suffering so disproportionately from the
John Hardin KA7OHZ ICQ#15735746 http://www.impsec.org/~jhardin/
jhardin at impsec.org pgpk -a jhardin at impsec.org
key: 0xB8732E79 - 2D8C 34F4 6411 F507 136C AF76 D822 E6E6 B873 2E79
"Bother," said Pooh as he struggled with /etc/sendmail.cf, "it never
does quite what I want. I wish Christopher Robin was here."
-- Peter da Silva in a.s.r
61 days until the Slovakian Presidential Election
More information about the esd-l