#!/usr/bin/perl
use strict;
use warnings;
use Data::Dumper;
use File::Path qw(make_path);

#Converts KMC usage into man pages. It's a bigger script than the man pages, but the plan is to have a future script that works for all packages
#The man pages are placed in the man folder in the main directory of the kmc package

createManPages();

sub createManPages {

  my $source= 'bin';
  my $destination= 'man';
  my $app_name = 'KMC';


  unless ( -d $destination ) {
    make_path( $destination );
  }

  my @files;

  push(@files,"$source/kmc");
  push(@files,"$source/kmc_dump");

  if ( scalar @files > 0 ) {

    print "Creating manpages\n";
    for my $file ( @files ) {
      $file =~ s/\n$//;

      my $filename = $file;
      $filename =~ s/$source\///;

      my $uc_filename = uc($filename);
      my $man_file = $filename;

      $man_file = $destination . '/' . $man_file . '.1';

      my $kmc_short_description = q{count kmers in genomic sequences};
      my $kmc_long_description = q{KMC—K-mer Counter is a utility designed for counting k-mers (sequences of consecutive k symbols) in a set of reads from genome sequencing projects.};

      my $kmc_dump_short_description = q{list k-mers from kmc-generated database};
      my $kmc_dump_long_description = q{KMC's end product is a database of k-mers stored in compact binary form. The kmc_dump program allows the production of a textual list of k-mers together with their counters for the analysis ran};


      my $cmd = "help2man -N -m $filename -n $filename --no-discard-stderr $file | sed 's/usage://gi'";
      my @output;
      push(@output, `$cmd`);

      for my $line (@output) {
	$line =~ s/\n$//;

      }

      my @lines_not_to_print;
      my @lines_to_print;
      if ($filename eq 'kmc') {
	  my $ss_param_seen = 0;
	  my $ss_param_seen2 = 1;
	  my $ip_tag_seen = 0;

	  for (my $i = 0; $i < scalar @output; $i++) {


	      my $output_line = $output[$i];

	      if ($output_line =~ m/^(\.TH\s)([a-zA-Z0-9-]+)(\s.*)/) {
		  $output_line = $1 . uc($filename) . $3;
	      }
	      $output_line =~ s/^(K\-Mer \\- )(kmc)/$2 - $kmc_short_description/;
	      if ($output_line =~ m/^K\\-Mer Counter/) {
		  $output_line = $kmc_long_description;
	      }

	      if ($output_line =~ m/^\.SS \"Parameters\:\"/ && ! $ss_param_seen) {
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);
		  push(@lines_not_to_print, $i + 2);
		  push(@lines_not_to_print, $i + 3);

		  $ss_param_seen = 1;
		  $ss_param_seen2 = 0;

	      }
	      elsif ($output_line =~ m/^\.IP/ && ! $ip_tag_seen) {
		  $output_line = '.SH SYNOPSIS';
		  if ($output[$i + 2] =~ /^kmc/) {
		      $output[$i +2] = ".PP\n" . $output[$i + 2];
		  }
		  $ip_tag_seen = 1;
	      }
	      elsif ($output_line =~ m/^\.SS ""/) {
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);
		  push(@lines_not_to_print, $i + 2);
		  push(@lines_not_to_print, $i + 3);
	      }
	      elsif ($output_line =~ m/^Example:/) {
		  $output_line = '.SH EXAMPLES';
		  $output[$i + 2] = ".PP\n" . $output[$i + 2];

	      }
	      elsif ($output_line =~ m/^\.SS \"Parameters\:\"/ && $ss_param_seen && ! $ss_param_seen2) {
		  last;

	      }

	      if ( ! grep( /^$i$/, @lines_not_to_print ) ){
		  #print "$output_line\n";
		  push(@lines_to_print, $output_line);
	      }


	  }
      }
      elsif ($filename eq 'kmc_dump') {

	  my $seen_kmc_start = 0;
	  my $seen_parameters = 0;
	  my $seen_options = 0;
	  my $synopsis_seen = 0;
	  my @parameters_lines;
	  my @options_lines;
	  for (my $i = 0; $i < scalar @output; $i++) {
	      my $output_line = $output[$i];
	      if($output_line =~ m/^(\.TH\s)([a-zA-Z0-9-]+)(\s.*)/ ) {
		  $output_line = $1 . uc($filename) . $3;
	      }
	      $output_line =~ s/^(KMC \\- )(kmc_dump)/$2 - $kmc_dump_short_description/;
	      if ($output_line =~ m/^KMC dump ver\./) {
		  $output_line = $kmc_dump_long_description;
		  $seen_kmc_start = 1;
	      }
	      if ($output_line =~ m/^kmc/ && $seen_kmc_start == 1 && ! $synopsis_seen) {
		  $output_line = "\.SH SYNOPSIS\n$output_line";
		  $synopsis_seen = 1;
	      }
	      if ($output_line =~ m/^kmc/ && $seen_kmc_start == 1 && $synopsis_seen) {
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);

	      }
	      if($output_line =~ m/Parameters:/ && ! $seen_parameters) {
		  push(@parameters_lines, '.PP');
		  push(@parameters_lines, $output_line);
		  push(@parameters_lines, '.PP');
		  push(@parameters_lines, $output[$i + 1]);
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);
		  $seen_parameters = 1;
	      }
	      if($output_line =~ m/Parameters:/ && $seen_parameters) {
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);
	      }
	      if($output_line =~ m/Options:/ && ! $seen_options) {
		  push(@options_lines, '.SH OPTIONS');
		  push(@options_lines, $output[$i + 1]);
		  push(@options_lines, '.PP');
		  push(@options_lines, $output[$i + 2]);
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);
		  push(@lines_not_to_print, $i + 2);
		  $seen_options = 1;
	      }
	      if($output_line =~ m/Options:/ && $seen_options) {
		  push(@lines_not_to_print, $i);
		  push(@lines_not_to_print, $i + 1);
		  push(@lines_not_to_print, $i + 2);
	      }

	      if ( ! grep( /^$i$/, @lines_not_to_print ) ){
		  push(@lines_to_print, $output_line);
	      }

	  }
	  for my $line(@options_lines) {
	      push(@lines_to_print, $line);
	  }
	  for my $line(@parameters_lines) {
	      push(@lines_to_print, $line);
	  }

      }
      open (my $man_fh, ">", $man_file);
      for my $line(@lines_to_print) {
	  print $man_fh "$line\n";
      }
      writeAuthorAndCopyright($man_fh,$filename);
      close($man_fh);
      print "Manpage creation complete\n";
    }
  }
}

sub writeAuthorAndCopyright {

  my ($man_fh,$filename) = @_;

  my $author_blurb = <<END_OF_AUTHOR_BLURB;
.SH "AUTHOR"
.sp
$filename was originally written by:
.PP
Sebastian Deorowicz (sebastian.deorowicz\@polsl.pl)
.PP
Marek Kokot
.PP
Szymon Grabowski
.PP
Agnieszka Debudaj-Grabysz
END_OF_AUTHOR_BLURB

  print $man_fh "$author_blurb\n";

  my $copyright_blurb = <<'END_OF_C_BLURB';
.SH "COPYING"
.sp
KMC is a free software distributed under GNU GPL3 licence for academic, research, and commercial use.
END_OF_C_BLURB

  print $man_fh "$copyright_blurb\n";

}
