Re: [exim] Tainted arg 2 for mailman_transport transport com…

On 21/07/2022 07:27, Thomas Krichel via Exim-users wrote:
> 2022-07-21 06:19:30 1oEPWy-002t7O-0x == nep-test@??? R=mailman_router
> T=mailman_transport defer (0): Expansion of
> "${sg{sg{$/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}"
> from command "/var/lib/mailman/mail/mailman '${if
> def:local_part_suffix
> {${sg{$local_part_suffix}{-(\\w+)(\\+.*)?}{\$1}}} {post}}'
> ${sg{sg{$/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}"
> in mailman_transport transport failed: $ not followed by letter,
> digit, or {

Here's how to get more info on this expansion problem.

I'm using "noutf8" only in case this message messes up UTF8 characters.
I manually replaced "$local_part" with "nep-test".

# exim -d-all+expand+noutf8 -be
[ignore irrelevant output here, before the '>' prompt]
> MM_HOME=/var/lib/mailman
Defined macro 'MM_HOME'
> MM_WRAP=MM_HOME/mail/mailman
Defined macro 'MM_WRAP'
> MM_LISTCHK=MM_HOME/lists/${lc:nep-test}/config.pck
Defined macro 'MM_LISTCHK'
> ${sg{sg{$MM_LISTCHK}{\/config.pck$}{}}{.*\/}{}}
/considering: ${sg{sg{$/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}
/considering: sg{$/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}
|-------text: sg{
|considering: $/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}
|failed to expand: sg{$/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}
\___error message: $ not followed by letter, digit, or {
|failed to expand: ${sg{sg{$/var/lib/mailman/lists/${lc::$local_part}/config.pck}{\/config.pck$}{}}{.*\/}{}}
\___error message: $ not followed by letter, digit, or {
Failed: $ not followed by letter, digit, or {
> ^D



So, there's a spare '$' before MM_LISTCHK. Eyeballing that, I suspect it should be
associated with the previous "sg" that looked like plain text. And the braces '{ ]'
don't look right...

Needing to use two ${sg} 's on something makes me wonder if the approach was right....
--
Cheers,
Jeremy