You appear to be parsing SWIFT messages. The easiest way (as SWIFT messages are fairly short) would be to read the whole message into an array and then do array manipulation by iterating over the array indices and then inserting elements at the appropriate places. Of course you will need to find the right item after which to insert your new item, which will need some understanding of the sequence in which the items are allowed to appear, but I guess you have that already.

my $seen_16r_secmove = 0;
my $seen_17b         = 0;
my @buf;
while (<>) {
   if (/^:17B:/) {
      $seen_17b = 1;
      print @buf;
      @buf = $_;
   } elsif ($seen_16r_secmove && $seen_17b && /^:35B:/) {
      print;
   } else {
      $seen_16r_secmove = 1 if /^:16R:SECMOVE$/;
      push @buf, $_;
   }
}

print @buf;
[download]

Or a version that doesn't start buffering before it needs to:

my $seen_16r_secmove = 0;
my $seen_17b         = 0;
my @buf;
while (<>) {
   if (/^:17B:/) {
      $seen_17b = 1;
      print @buf;
      @buf = $_;
      next;
   }

   $active = 1 if /^:16R:SECMOVE$/;

   if (!$seen_17b || $seen_16r_secmove && /^:35B:/) {
      print;
   } else {
      push @buf, $_;
   }
}

print @buf;
[download]

Both handle all of these according to the given specs:

:16R:SECMOVE
:17B:
:35B:

:16R:SECMOVE
:17B:
:35B:
:35B:

:16R:SECMOVE
:35B:
:17B:
:35B:

:17B:
:16R:SECMOVE
:35B:
[download]

Untested.

Update: Simplified.

 
#!/usr/bin/perl

use strict;

my $data_pos = tell DATA;
my $record;

while (my $line = <DATA>) {
    chomp( $line );
    $record = $line if ( $line =~ /^:16R:USECU/ );
}

seek DATA, $data_pos, 0;

while (my $line = <DATA>) {
    chomp( $line );
    print "$line\n" if ( $record && $line =~/^:16S:GENL/ );
}

__DATA__
:16S:LINK
:16S:GENL
:16R:USECU
[download]

Thanks for all the advice and input! it looks like i will definitely have to loop thru the file twice.

Print everything till you hit :17B:, then you consider this the start of a record to watch for. Feed the lines that follow to an array till the first :35B: record, at which point you are ready to process this "block" between 17B and 35B. If there was a :16R:SECMOVE in between you print the first record of array, then the last, and then the rest, minus the last one, then go back to printing rows till you hit :17B: again. Of course spit out the array if there was no :16R:SECMOVE in there. Don't forget to empty the array before reuse.

Obviously if :16R: can occur before :17B: and still need to process it that way, just modify accordingly. Guess it would help if I knew anything about these records and if they have any kinda order or whatnot. You could probably do this as an AWK oneliner too.

Building on what Corion said above, you might be interested in the Tie::File module. It allows you to access a file as a Perl array.