Hi,

New problem, or challenge as they prefer in the US.

I need to insert some lines in a file before certain other lines.
To make it more clear:

Original file:

aaaa
bbbbb
ccccc
ddddd
bbbbb
fffff
ggggg

aaaa
bbbbb
ccccc
ddddd
NEW_NEW_NEW
bbbbb
fffff
ggggg

$file_to_update = "test.txt";
open (FILE_TO_UPDATE, "<$file_to_update");
my @lines = <FILE_TO_UPDATE>;
close (FILE_TO_UPDATE);

open (FILE_TO_UPDATE, ">$file_to_update");
for (@lines) {
  if ($_ =~ m/fffff/) {
  print FILE_TO_UPDATE "NEW_NEW_NEW";
  }
  print FILE_TO_UPDATE;
}
close (FILE_TO_UPDATE);

aaaa
bbbbb
ccccc
ddddd
bbbbb
NEW_NEW_NEW
fffff
ggggg

perl -0pe'  
  s/bbbbb(?!.*bbbbb)/NEW_NEW_NEW\n$&/s
  ' infile

Thanks for the one-liner.
For this example it works, but you understand it is part of a bigger universe :-)
And I am having trouble incorporating it in the example code.
Not to mention that I do not understand the wizardry in your one-liner.
I understand it is a conditional substitute, but what is the $ and the & doing for example ?

But let me make the problem a bit more real life.

Here is the test.txt data file (edited for the forum).
It is a ksh script and this is how it comes after an installation.
I need to modify it and I would like to have this modification done by a script.

if [ -f $topload_inprogress ]
then
		blahlblah
		rm $inprogress
		exit 1
fi 

if [ ! -f $params ]
then
	blahlblah
	rm $inprogress
	exit 1
fi 

if(($?==1))
then 
	blahlblah
	rm $inprogress
	exit 1
fi

if [[ -f $ftp_fin ]]
then
	rm $ftp_fin
else
	blahlblah
	rm $inprogress
	exit 1
fi

files=`ls $CADATA/srcdata/$source/custom/BULK/after`
for name in $files
do
	blahlblah
done

rm $inprogress
$CAROOT/prog/ca_log_message TOPOLOGY_STAT "Finished the bulk ftp topology load of $source data source."
exit 0

The file comes like this, I have no control over it.
And if I have to "prepair" it manually, then I can make the text insert manually too.
But I don't want to.

The last "rm $inprogress" is at the end of the file, so it is the last occurrence.

The text-block needs to be between the last "done" and the last "rm $inprogress", that's a given.

I use Perl, because I need to make many more modifications in other files and I need to start things, stop things, etc.
Perl is perfect for this.
This is one of the many sub-routines in the script.
In fact I need to make 2 text-block insertions in this file and the second one has this problem I described.

If necessary I could call a sed command from within Perl, right ?
But I do not see why it could not be done with Perl.
Okay, I can't, but I am not the brightest Perl programmer around
Not yet.

Now I am thinking about counting the "done"s, printing all lines including the last "done". Then print my text-block and then print the last lines.
Or perhaps read 2 lines, test the 2nd line for "Finished the bulk ftp".
Print all lines until the test is true, print my test-block and print the last lines.
Something should be possible.

E.J.

OK,
you can use something like this (you should add exception handling, of course):

#!/usr/bin/perl

use warnings;
use strict;

my ( $shellscript, $pattern ) = ( shift, 'rm \$inprogress' );

my $newcode = <<_NEWCODE_;
line1
line2
...
line6
_NEWCODE_

{
    local ( $^I, @ARGV, $/ ) = ( '', $shellscript );
    while (<>) {
        s/$pattern(?!.*$pattern)/$newcode\n$&/s;
        print;
    }
}

I use Perl, because I need to make many more modifications in other files and I need to start things, stop things, etc. Perl is perfect for this. This is one of the many sub-routines in the script. In fact I need to make 2 text-block insertions in this file and the second one has this problem I described. If necessary I could call a sed command from within Perl, right ? But I do not see why it could not be done with Perl.