use warnings;
use strict;

while (<DATA>) {
    if (/[A-Z]/) {
        print 'good ';
    }
    else {
        print 'bad  ';
    }
    print;
}

__DATA__
*
*,*
*.*
ABC
ABC,DEF,GHI_JKL
[download]

Outputs:

bad  *
bad  *,*
bad  *.*
good ABC
good ABC,DEF,GHI_JKL
[download]

I'm not sure whether the OP wants *.*,*;A to pass. I'd use

chomp;
print /[^A-Z_,]/ ? 'bad' : 'good'
[download]

Or, if the format is important, too:

print /^(?:[A-Z_]+,?)+(?<!,)$/ ? 'good' : 'bad';
[download]

i.e. at least one uppercase letter or underscore, followed by a comma, this repeating several times, and no trailing comma.

($q=q:Sq=~/;[c](.)(.)/;chr(-||-|5+lengthSq)`"S|oS2"`map{chr |+ord
}map{substrSq`S_+|`|}3E|-|`7**2-3:)=~y+S|`+$1,++print+eval$q,q,a,
[download]

 print /^(?:[A-Z_]+,?)+(?<!,)$/ ? 'good' : 'bad';

Hi I tried your suggestion as below

while (<DATA>) {
if (/^(?:[A-Z_]+,?)+(?<!,)$/) {print "good";}
else {print "bad";}
print;
}
[download]

__DATA__
*
*,*
*.*
ABC
ABC*
*ABC
ABC_TO
ABC_TO,DEF,GHI_JKL,GHI_JK1_TO
[download]

Good string format are ABC,ABC_TO and ABC_TO,DEF,GHI_JKL,GHI_JK1_TO but somehow I am getting this as BAD ABC_TO,DEF,GHI_JKL,GHI_JK1_TO

Strongly suggest you visit (and study) the Tutorials section devoted to regexen. You wouldn't have needed to post your question had you made an effort to learn. Downvoted.

I want to allow in text box can be "ABC" or comma separated list as "ABC,DEF,GHI_JKL". Don't want user to enter * or *,* or *.* to be entered in test box

Seems like your acceptable characters are A through Z and _ and ,

So, I would test for anything matches the inverse of that character set:

if ([^A-Z_,]) {
    # bad user input
} else {
    # good user input
}
[download]

Disclaimer: Not tested. YMMV

You can use bog standard HTML to do this - all 'modern' browsers support it:

<input type="text" ... ... maxlength="15" pattern="[a-zA-Z]{1,15}"
[download]

will only allow UPPER/lower case a to z up to a maximum length of 15 characters.

Nick

This is interesting and worth mentioning but needs a clear caveat that the backend has no idea whether the data came from an HTML5 browser or a hacker/hackbot that bypassed the filter completely. Client side validation is for courtesy, speed, saving unnecessary server load. All data must be validated on the backend/server regardless of what the front end does.

I also think it borders or a poor practice because the best code would use the same regex on the front and back end to avoid silly bugs and the backend has much more power to offer so will be hobbled by the subset the front end can do.

  print /^(?:[A-Z_0-9]+,?)+(?<!,)$/ ? 'good' : 'bad';

Thankyou everyone.