The thing that has always prevented anyone from writing such a program is that it's very hard! It's easy to do a really bad job, but to do a good job, you have to be very clever. For example, suppose you see this:

    set bar=`echo $foo | awk '{print $5}' | cut -c1-4`
[download]

    $bar=`echo $foo | awk '{print \$5}' | cut -c1-4`
[download]

    $bar = substr((split($foo))[4], 0, 4);
[download]

A good learning process it to write a 'bad' or incomplete translator. Just going through and

TMP="/tmp/cas.$$.out"
to:
$tmp = "/tmp/cas.$$.out"
[download]

is useful and trying to map if ... ] then to if ( ... ) [ (sorry I'm a ksh fool) can be quite useful both in learning and in moving some of your stuff. The hard parts will still need to be done by hand.

a

The only known shell -> Perl translator on the internet is Tom Christiansen.

If you look at the history of porting, it has always been easier to write an emulator that emulates the source machine than to port the program to the target machine. The 8086, Java, C# and Inferno/Dis all follow this concept.

If you consider the possible pitfalls of writing a general csh to Perl converter, like obscure ways to define arrays, even more obscure ways to access variables and the source (I mean eval) statement, you will see that an emulator is the only way to go ...

The only known shell -> Perl translator on the internet is Tom Christiansen.

The only one known to you perhaps, but the statement is false.

As evidence, look at http://www.cpan.org/authors/id/MERLYN/sh2perl-0.03.tar.gz.

-- Randal L. Schwartz, Perl hacker