Please read Writeup Formatting Tips, as it is difficult to understand your intention w/o seeing your input file properly formatted (see the section on <code> tags).

What have you tried? Why do you want to do this in Perl, and do you have any experience with it? This is, after all a help resource and not a code writing service.

For some introductory guidance, you'll probably want to take a look at I/O Operators to see how to read/write files and using substr (or pack for the more adventurous) to reformat your strings. You can likely perform all this in a one-liner using the -pi switches (see perlrun), but I'm guessing that's outside your current experience level. If you post some code, or even pseudo-code, I'll be happy to provide pointers and debug assistance (see How (Not) To Ask A Question).

Based on your sparse spec. this might get you started.

The two things to note are:

1. pack & unpack for dealing with fixed width data.
2. Storing the data from each line until you've looked at the next line to decide whether it's a full line or an extension.

#! perl -sw
use 5.010;
use strict;

my( $lastID, $lastF1, $lastF2 ) = 0;

while( <DATA> ) {
    chomp;
    my( $id, $f1, $f2 ) = unpack 'a5 a10 a10', $_;
    if( $id && $id == $lastID ) {
        if( substr( $f1, 0, 1 ) eq ' ' ) { ## second field extension
            $lastF2 .= substr( $f1, 1 );
        }
        else {
            $lastF1 .= $f1;
        }
    }
    else {
        say pack 'A5 A20 A20', $lastID, $lastF1, $lastF2 if $lastID;
        ( $lastID, $lastF1, $lastF2 ) = ( $id, $f1, $f2 );
    }
}

__DATA__
12345aaaaaaaaaabbbbbbbbbb
12365sssssssssscccccccccc
12365sssss
12367ddddddddddvvvvvvvvvv
12367 vvvvv
[download]

...if the second or third column exceeds the length...

Or, what am I missing? :-)