Chand,
I have rewritten the script to process multiple input files and allow fields specified in the rules file to be skipped if the field named in a rule does not appear as an input file's field header. (If this happens, a note will be included in the reject file as you requested stating that a rule is invalid.)

Making the script replace every occurrence of a search string rather than just the first occurrence was done by just changing a call to sub() to be a call to gsub(). I have expanded the shell portion of the script to support several options and provide a built-in man page. The in-line comments explaining what the script does have also been expanded in hopes that you will be able to make further enhancements yourself.

Note that the rule that you said should convert the input line:rc$$$21|rc#21|yyy|to:rc21|rc#21|yyy|, it won't do that because the rule in you rules file has two spaces in the replace field. Therefore, the result of applying that rule to every occurrence that matches will instead produce the output:

Code:

rc      21|rc#21|yyy|

Because the script is so large now, I have attached it rather than including it in-line here. The name of the script is substitute, but to upload it I had to use the name substitute.sh. It is written as a Korn shell script that calls awk. You should be able to use a Bourne shell, or bash, as well as a Korn shell if you just change the first line of the script from #!/bin/kshto the path to your shell. (However, it won't work with csh or any of its variants.)
If you are on a Solaris system, use nawk or /usr/xpg4/bin/awk instead of awk.

I hope this helps,
Don