2. I'm guessing here, but you probably want more than one line out of the file, right? Try something like:

open IN,"<",$fileName or die "Can't open $filename: $!";
while (<IN>) { # read line from IN handle to $_
   if ($. == $some_number || $. == $some_other_number) { # $. is the c
+urrent line number
       # do stuff with the line in $_
   }
   last if $. > $last_interesting_line_number;
}
close IN;
[download]

You might want to take a look at Tie::File.

I now need to find a specific line in the file not only the first and last like the example shows(I know the suffix) is there a better way that to run on the (loop) array and find it ?

In the meantime, here's one way you can do it by hand. The assumption is that the original data text file is not changed often.

First, here's how to create the index file:

open IN, "<", $textfile or die "Can't open textfile: $!";
binmode IN;
open IDX, ">", "$textfile.idx" or die "Can't create index file: $!";
binmode IDX;
print IDX pack "N", 0;
while(<IN>) {
    print IDX, pack "N", tell IN;
}
[download]

Next, here's how to look up a line by number (in $lineno):

open TEXT, "<", $textfile or die "Can't open textfile: $!";
open IDX, "<", "$textfile.idx" or die "Can't open index file: $!";
binmode IDX;
seek IDX, 4*$lineno, 0;
read IDX, my($buf), 4;
seek TEXT, unpack("N", $buf), 0;
$line = <TEXT>;
[download]

As you may have guessed, the pack/unpack serves to make the integer index a fixed length (4 bytes).

Or you can use Tie::File, but where's the fun in that? :)

my $first_line = qx/head -1 $filename/;
print $first_line;

my $last_line = qx/tail -1 $filename/;
print $last_line;

my $n = 100;
my $nth_line = qx/tail +$n $filename | head -1/;
print $nth_line;
[download]

Hello and thank you all!
A few answers:
1. yes, I'm running on linux !!
2. this is a reporter file, it gets appended all the time, every minute or so
3. the last line is the most updated(since the report row /line is added every minute), that's why I'm interested in it, and I used to read the first line too since it holds the columns name(I get the data by the column name) BUT
now I found a bug- if a change is made, and a column is added,then a new row is added (similar to the first row holding the names) but this row also holds the name of the new column, BUT it doesn't know about it since it reads the column name from the first row in the file
So, I need to change this and search the file for this row and if it's not found-I'll end up with the first one

It look something like this:


#name: time  input   output   users   connected  ...
      12:23   23      34       780      560
      12:24    21      40       780      570  
....
[download]

ok-
in keeping with the *nix commands theme how about:


#
#  find last row that defines the columns ( i.e. starts with #name: )
#
my $column_names = qx/grep ^\#name $filename | tail -1/;
print $column_names;
[download]

-hth

If you're going to look for multiple lines in the file, perhaps you should read the file as one huge string, then use regular expressions to find the lines of interest. For example:

#!/usr/bin/perl -w
use strict;
use warnings;

# Slurp entire file into $file
my $file;
open FH, $0 or die $!;
{
        # localize scope as described in perlvar
        local $/;
        $file = <FH>;
}

my $line = '---not found---';
$line=$1 if $file=~/\n(ope[^\n]*)/;

print "Line='", $line, "'";
[download]

Line='open FH, $0 or die $!;'
[download]

roboticus,
Thanks, but as I've shown in my 1st posting, what I'm doing now, is to read the file into an array and then I can read from the array line-by-line, I don't see how your sugestion is different.
The main problem is that I read this file a lot(I actually have various reporter files), and I need to find a specific line ( the "header" ) and the last line(the updated "data" )
my question is: is there a simple way to do so, and is there a way to do so without "saving" the whole file into an array or string or such.
Thanks

Sounds to me like Joost had the right idea: Try File::Tail. That would allow you to read the whole file in at startup, scan it for the last header line, and then just pick up the new lines (data or header) as they're added instead of repeatedly reading and processing the entire file.

just dave:

I was thinking that by using a single string rather than an array, you can avoid the explicit looping, making your code clearer. Since you also mentioned having a memory leak, I thought that a single string would fragment memory much less than a bazillion smaller strings. That might make it easier for perl to reuse the memory and/or recognize what can be freed

Other than that, it's not much different at all.

Of course, since you're rereading the same file repeatedly, I think that Joost's suggestion is my favorite so far. Then, while looking up File::Tail, I saw File::Tail::App, which is pretty cool because it does much of the grunt work for you.

--roboticus

Hi, [id://just dave]. This isn't in reply to your question, but I would suggest that you get into the habit of use'ing strict and warnings. Also, it's a very good idea to check every file operation after you perform it, to ensure that you've actually accomplished what you thought you did. Doing so can save you endless hours of painful debugging. Ask me how I know ;-)

chomp($record = <DATA>);
$keys{$_} = $c++ for split / /, $record;

## Get last record and put it in $record
@record = split / /, $record;
print $record[$keys{'timestamp'}];
[download]