use v5.32;

use HTML::TableExtract;
use File::Slurp;
use Data::Dumper;
use Text::SimpleTable::AutoWidth;
use WWW::Mechanize ();
use List::AllUtils qw/ sum pairmap /;
use List::UtilsBy  qw/ nsort_by partition_by /;

use experimental qw/ signatures /;

unless (caller) {
    my $ts = get_mersenne_results();

    generate_output_table($ts)->draw;
}

sub get_mersenne_results() {

    # log in to Mersenne.org and get results for the last year
    # excluding everything but PHP and DD results

    my $mech = WWW::Mechanize->new;

    my $url =
      'https://www.mersenne.org/results/?extf=1&exp1=1&execm=1&excert=1&exp_lo=2&exp_hi=&limit=10000';

    $mech->get($url);

    $mech->submit_form(
        form_number => 1,
        fields      => {
            user_login    => $ENV{'MERSENNE_USER'},
            user_password => $ENV{'MERSENNE_PASSWORD'},
        } );

    # load the results into a table object
    my $html_string = $mech->content;

    return extract_first_table($html_string);
}

sub extract_first_table ($html) {
    $html =~ s/\n//g;

    my $te = HTML::TableExtract->new( depth => 0, count => 2 );

    $te->parse($html);

    return $te->first_table_found;

}

sub generate_output_table ($ts) {

    my $tbl = Text::SimpleTable::AutoWidth->new(
        captions => [qw/ Computer GHZDaysPerDay /] );

    # group GHZ Days results by computer, compute GHZ Days per Day (GHZ Days / Days)
    $tbl->row(@$_) for 
        reverse
        nsort_by { $_->[1] } 
        pairmap { [ $a, $b ] } # group for the sorting
        pairmap { $a => sprintf "%.0f", sum(@$b) / @$b } # take the mean
        pairmap { 
            # calculate our stats
            $a => [ map { $_->{ghz_days} / $_->{days} } @$b ] 
        } 
        partition_by { $_->{machine} } # group by machine
        grep { $_->{days} > 0 } # skip entries with no days
        map {  # get the info
            my $machine  = $_->[0];
            my $days     = $_->[4] =~ s/\s//gr;
            my $ghz_days = $_->[6] =~ s/\s//gr;

            +{ days => $days, ghz_days => $ghz_days, machine => $machine }
        } 
        $ts->rows;

    return $tbl;
}