estimate_analysis_time

#!/usr/bin/env perl

=head1 Description

Wanna know how long your hive analysis will take, but don't like to math?
Pass some basic values in here and let the script guess then

=cut

use strict;
use warnings;
use Bio::EnsEMBL::Compara::DBSQL::DBAdaptor;
use Getopt::Long;

my ( $ready, $time, $capacity, $help );

GetOptions(
    "r|ready=i"    => \$ready,
    "t|time=s"     => \$time,
    "c|capacity=s" => \$capacity,
    "h|help"       => \$help,
  );

die_with_usage() if $help;
unless ( $ready && $time && $capacity ) {
	print "-r, -t and -c options must be defined\n\n";
	die_with_usage(); 
}

unless ( $time =~ m/^([\d\.]+)([smh]$)/ ) {
	print STDERR "Unrecognised time format. Please use e.g '1.5s', '10m' etc.\n\n";	
	die_with_usage();
}

my ( $time_value, $time_unit ) = ( $1, $2 );

my $total_walltime = $ready * $time_value;
my $conc_walltime = $total_walltime / $capacity;

my ( $sec, $min, $hr );
if ( $time_unit eq 's' ) { # seconds
	$sec = $conc_walltime;
	$min = $sec/60;
	$hr  = $min/60;
} elsif ( $time_unit eq 'm' ) { # minutes
	$min = $conc_walltime;
	$sec = $min * 60;
	$hr  = $min / 60;
} elsif ( $time_unit eq 'h' ) { #hour
	$hr  = $conc_walltime;
	$min = $hr * 60;
	$sec = $min * 60;
}

if ( $min < 1 ){
	print "only seconds to go : $sec\n";
} elsif ( $hr < 1 ) {
	my ( $m_str, $s_str ) = numeric_to_time($min);
	print "just minutes left : " . $m_str . "m " . $s_str . "s\n";
} else {
	my ( $h_str, $m_str ) = numeric_to_time($hr);
	print "gonna be a few hours : " . $h_str . 'h ' . $m_str . "m\n";
}

sub numeric_to_time {
	my $num = shift;

	my $int   = int($num);
	my $float = $num - $int;

	my $sec = 0;
	$sec = 60 * $float if defined $float;
	my $ssec = sprintf("%.1f", $sec);

	return ( $int, $ssec );
}

sub die_with_usage {
	print STDERR <<EOF;
estimate_analysis_time.pl <options>

	r|ready		number of ready jobs
	t|time 		avg time per job (unit denoted by s,m or h)
	c|capacity	number of concurrent jobs allowed by hive/analysis capacity

EOF
	exit(0);
}