However, if there are several consecutive matches it will print duplicated data.

This one doesn't produce duplicates:

    perl -ne 'BEGIN{$re=qr/foo/}if(/$re/){print$p if$p!~/$re/;print$_}
+;$p=$_' file
[download]

If you have a grep that supports it you can also do this:

    grep -B 1 foobar file

And finally, this is a more generalised method of printing a matched line and the n-1 previous lines, with and without duplicates:

    perl -ne '$i=$.% 2; $a[$i]=$_; print @a[$i+1..$#a,0..$i] if /foo/'
+ file
    perl -ne '$i=$.% 2; $a[$i]=$_; print @a[$i+1..$#a,0..$i] and @a=()
+ if /foo/' file
[download]

--
John.

Thanks works great!

Keep the previous line in a variable or at least keep its length or position (via tell()) in a variable to seek() back if you need:

my $p;
while(<FILE>){
   #du stuff
   if( prevous line needed and defined $p ){
      seek FILE, -(length()+$p), 1;
      my $prevous_line = <FILE>;
      # next <FILE> will return the current line again,
      # so call it again before going on!
  }
} continue { $p = length() }
[download]

Here's a way to do it using Tie::File, which also gives you the option of grabbing the previous n lines, not just the line immediately previous (maybe more useful if you need context in a log file.)

#! /usr/bin/perl

use strict ;
use warnings ;
$|++ ;

use Tie::File ;

tie my @file, 'Tie::File', 'test_h.dat' or die "Can't tie file: $!" ;

my $backtrack_lines = 1 ;
my $match = '' ;
for my $i ( 0 .. $#file )
{
    if ( $file[$i] =~ /baz/ )
    {
        $match = join "\n", @file[ ( $i - $backtrack_lines ) .. $i ] ;
    }
}

print $match ;

__END__
[download]

Those who know that they are profound strive for clarity. Those who
would like to seem profound to the crowd strive for obscurity.
            --Friedrich Nietzsche