#!/usr/bin/perl
# The Missing Textutils, Ondrej Bojar, obo@cuni.cz
# http://www.cuni.cz/~obo/textutils
#
# 'greplines' selects the lines identified by their position (order) in file.
#
# $Id: greplines,v 1.11 2012-11-22 18:05:44 bojar Exp $
#

use Getopt::Long;
use strict;

my $inverse = 0;
my $print_numbers = 0;
my $help = 0;
my $args = 0;
my $startnr = 1;
my $preserve_order = 0;
my $numdelim = ":";
GetOptions(
  "numbers"=>\$print_numbers,
  "delimiter=s" => \$numdelim,
  "inverse" => \$inverse,
  "args" => \$args,
  "start=i" => \$startnr,
  "preserve-order" => \$preserve_order,
) or exit 1;

$numdelim =~ s/\\n/\n/g; # I know this is a bad way
$numdelim =~ s/\\t/\t/g;

if ($help || 0 == scalar @ARGV) {
  print STDERR "usage: greplines file_with_lines
Prints out only the specified lines of stdin to stdout.
--inverse  .. to print lines except those listed in file_with_lines
--args ... read numbers to print from args, not from files listed as args
--start=N  ... lines are numbered from N
--preserve-order ... load all selected lines to memory, dump afterwards in
                     order of the numbers given
";
  exit 1;
}

my %wanted;
my $cnt = 0;
while (my $arg = shift @ARGV) {
  if ($args) {
    foreach my $lineno (split /\s*,\s*/, $arg) {
      $cnt++;
      $wanted{$lineno} = $cnt;
    }
  } else {
    my $hdl = my_open($arg);
    while (<$hdl>) {
      chomp;
      foreach my $lineno (split /\s+/) {
        $cnt++;
        $wanted{$lineno} = $cnt;
      }
    }
    close $hdl;
  }
}
print STDERR "$cnt lines are going to be ".($inverse?"skipped":"printed out").".\n";

my $nr = $startnr;
my %mem = ();
while (<>) {
  if ($wanted{$nr} xor $inverse) {
    my $out = "";
    $out .= "$nr$numdelim" if $print_numbers;
    $out .= "$_";
    if ($preserve_order) {
      $mem{$wanted{$nr}} = $out;
    } else {
      print $out;
    }
  }
  $nr++;
}

if ($preserve_order) {
  foreach my $k (sort {$a<=>$b} keys %mem) {
    print $mem{$k};
  }
}


sub my_open {
  my $f = shift;
  if ($f eq "-") {
    binmode(STDIN, ":utf8");
    return *STDIN;
  }

  die "Not found: $f" if ! -e $f;

  my $opn;
  my $hdl;
  my $ft = `file '$f'`;
  # file might not recognize some files!
  if ($f =~ /\.gz$/ || $ft =~ /gzip compressed data/) {
    $opn = "zcat '$f' |";
  } elsif ($f =~ /\.bz2$/ || $ft =~ /bzip2 compressed data/) {
    $opn = "bzcat '$f' |";
  } else {
    $opn = "$f";
  }
  open $hdl, $opn or die "Can't open '$opn': $!";
  binmode $hdl, ":utf8";
  return $hdl;
}