Shell Programming and Scripting

Hi all,
I can’t resolve an array element access issue on (Linux/pdksh) .. So I’m positing for advice.By the way - a friend demonstrated to me - same script works as expected under Solaris.

I have been working on a documentation project where many *.jpg screen shots are used in the documentation. The process that we go through - has us inserting additional images into the sequence of JPG files (where required). Eventually the sequence of images is “frozen” before they go into the next phase of the documentation process.


We need to have these images renumbered before the sequence is frozen. So for example we might start with:

Image_1
…
Image_34

With the additional insertions (and deletions too) – the directory structure morphs into something ugly like:

Image_1
Image_1a
Image_1b
Image_9
Image_9c
Image_9f
Image_34
Image_34h

We need to have a script process the directory structure and sort / rename these files into something like Image_0000, Image_0005, image_00010, image_0015, etc. (incremented by 5’s).



Having 5 spaces (between files) allows us to insert images into the sequence – if we should need to in a last minute emergency effort during later stages of the documentation process.


The problem:

The script is giving us problems (on linux). I can load the array in a loop – but I can NOT access the array elements outside of the loop and I am looking for advice on how to resolve this issue on the Linux box.



Please note - The same script works on Solaris as expected – array elements are accessible outside of the loop!!! (We cant use the Solaris box for this process).


Thank you for your time and input on this matter - Dave

Code:

#!/bin/ksh

touch /tmp/1.jpg /tmp/23.jpg /tmp/65.jpg /tmp/88.jpg

i=0
ls -alF /tmp/*.[jJ][pP][gG] | awk '{print $8}' | while read filein
do
   array[$i]=$filein
   echo "in loop: i=$i, Array($i)='s: ${array[$i]}"
   i=$(($i + 1))
done

echo "out of loop - 2nd element: ${array[2]}"
echo "out of loop - all elements: ${array[@]}"

rm /tmp/1.jpg /tmp/23.jpg /tmp/65.jpg /tmp/88.jpg

This is a useless use of ls *:

Code:

ls -alF /tmp/*.[jJ][pP][gG] | awk '{print $8}' | while read filein

You can strip out a ls, a read, and an awk by just doing this instead:
Since you're globbing anyway, why not just do:

Code:

for filein in /tmp/*.[jJ][pP][gG]

The pipe chain was causing your array problems because, in many shells, that entire loop behind the pipes will be executed inside a separate subshell.

isn't there a setting in ksh, similar to bash's "shopt -s nocaseglob" to enable case insensitive globbing?

also pdksh is not maintained anymore.
you might like to try mksh
MirOS: mksh ? the MirBSD Korn Shell

my tip:
when you start resorting to arrays in shell scripts maybe you are using the wrong tool.

Corona688:

thank you - your advice worked nicely... That web page was rather informative (bookmarked that one).

Kurumi:

RE: that entire loop behind the pipes will be executed inside a separate subshell.

yes that's what I was thinking when I saw the array data in the loop - but not external to the loop.

Bigearsbilly:

I was not aware of the status of pdksh (I'll be moving the ubuntu systems over to mksh I guess). I installed mksh on my ubuntu 11.04 (beta) machine and tested again - same error. I still would have changed / optimized the line - but wanted to see if mksh would handle the issue differently.

RE: when you start resorting to arrays in shell scripts maybe you are using the wrong tool.

Yeah - I generally avoid arrays in ksh, and was thinking about moving over to perl... But:

• I knew I was doing something blatantly wrong - and a simple fix was probably not too far away.
• The use of the array was the right way to handle this simple task.
• Others in the group have agreed to use/support ksh in the environment. So using perl for this one simple task, seemed a bit excessive for what we wanted to do.

All: Thanks for your time on this matter!