If the shell script that sets the variables is simple, then parsing that file is the best way to go. Substituting variable makes things a little harder but isn't too hard to accomplish in Perl. If it is more complicated than this, then it is best to let bash handle the shell scripts. Personally, I would have the config file be simple key=value format. It could be read by bash or perl, but would not be a proper shell script (no #!, no commands).
You can use bash to run the shell script, print the environment with printenv, and parse the results with Perl
#!bash
# myenvprint.sh
source myenv.sh
printenv
#!perl
my $myenv = `myenvprint.sh`
my %env = parse_env($myenv);
| [reply]
[d/l]
[select] |