This didn't make it work completely. I have found the way to do it. First I remove all the returns and make one file of it. After that I build in again returns looking for a certain pattern. If somebody can make it <CR> shorter and knows how I can skip to make intermediate file and create straigt away the target file it is welcome. Pieter Here is the code :

#!perl -w 

use strict;
use FileHandle;

my $file = "c:/perl/training/test/common.grb";
#source file 
my $file1 = "c:/perl/training/test/test5.txt"; #intermediate file, jus
+t contain one long line 
my @users = readFile($file);


for(@users) 
{
chomp;
s/\s+/ /g;
}

writeFile("c:/perl/training/test/test5.txt", @users);
my @newUsers = readFile($file1);

for (@newUsers) 
{
s/(.)G57/$1\nG57/g; 
} 

writeFile("c:/perl/training/test/test6.txt", @newUsers);>  #target fil
+e, each line start with G57 


sub readFile 
{ 
my ($fileName) = @_; 
sysopen(F, $fileName, O_RDONLY) or die "\nCannot open #<CR> $file: $!\
+n"; 
my @fileContents = <F>; 
close(F);
return @fileContents;
}

sub writeFile
{
my ($file, @contents) = @_;

if(-e $file)
{ 
unlink $file or die "\nunable to remove $file: $!\n";
} 

sysopen(F, $file, O_RDWR | O_EXCL | O_CREAT, 0666) or die "\nCannot cr
+eate $file: $!\n"; 
print F @contents or die "\nCannot write to $file: $!\n"; 
close(F) or die "\nCannot close filehandle - $file: $!\n"; 
}
[download]

Edit 2001-03-21 by mirod: added code tags