Bug 40748: Check for HTTP URLs and skip shell when using system()

This change checks for HTTP(S) URLs and skips the shell when using
system() so that shell metacharacters aren't interpreted.

Test plan:
0. Apply the patch
1. Insert the following into the Babeltheque_url_update system preference:
$(whoami > /tmp/rce.log)
2. koha-shell kohadev
3. perl misc/cronjobs/social_data/update_social_data.pl
4. Note the script says the following:
$(whoami > /tmp/rce.log) is not a HTTP URL. Aborting.
5. Note also that /tmp/rce.log is not created
6. Change Babeltheque_url_update syspref to something like the following:
http://localhost:8080
7. perl misc/cronjobs/social_data/update_social_data.pl
8. Note that the wget succeeds but the bunzip2 fails (as expected) since
we haven't fetched a bzipped file
9. Try to create a payload that passes the HTTP URL test with shell
metacharacters and note that they're not interpetted and instead
just used as part of a URL string.

Signed-off-by: Jake Deery <jake.deery@openfifth.co.uk>

Signed-off-by: Jonathan Druart <jonathan.druart@bugs.koha-community.org>
Signed-off-by: Paul Derscheid <paul.derscheid@lmscloud.de>
This commit is contained in:
David Cook 2025-09-04 01:20:38 +00:00 committed by Paul Derscheid
parent 19174500e9
commit 193268edbb
No known key found for this signature in database
GPG Key ID: C25F9FD44CED595B
1 changed files with 17 additions and 8 deletions

View File

@ -5,13 +5,22 @@ use Modern::Perl;
use Koha::Script -cron;
use C4::Context;
use C4::SocialData;
use URI;
my $url = quotemeta( C4::Context->preference("Babeltheque_url_update") );
my $output_dir = qq{/tmp};
my $output_filepath = qq{$output_dir/social_data.csv};
system(qq{/bin/rm -f $output_filepath});
system(qq{/bin/rm -f $output_dir/social_data.csv.bz2});
system(qq{/usr/bin/wget $url -O $output_dir/social_data.csv.bz2 }) == 0 or die "Can't get bz2 file from url $url ($?)";
system(qq{/bin/bunzip2 $output_dir/social_data.csv.bz2 }) == 0 or die "Can't extract bz2 file ($?)";
my $syspref_value = C4::Context->preference("Babeltheque_url_update");
my $url = URI->new($syspref_value);
C4::SocialData::update_data $output_filepath;
#NOTE: Both HTTP and HTTPS URLs are instances of the URI::http class
if ( $url && ref $url && $url->isa('URI::http') ) {
my $output_dir = qq{/tmp};
my $output_filepath = qq{$output_dir/social_data.csv};
system(qq{/bin/rm -f $output_filepath});
system(qq{/bin/rm -f $output_dir/social_data.csv.bz2});
system( '/usr/bin/wget', $url, '-O', "$output_dir/social_data.csv.bz2" ) == 0
or die "Can't get bz2 file from url $url ($?)";
system(qq{/bin/bunzip2 $output_dir/social_data.csv.bz2 }) == 0 or die "Can't extract bz2 file ($?)";
C4::SocialData::update_data $output_filepath;
} else {
print "$syspref_value is not a HTTP URL. Aborting.\n";
}