going berserk on the manipulation pipeline

This commit is contained in:
Yanick Champoux 2023-08-26 12:36:25 -04:00
parent 1200d98323
commit 89cc65421c

View File

@ -6,11 +6,11 @@ use Data::Dumper;
use Text::SimpleTable::AutoWidth; use Text::SimpleTable::AutoWidth;
use WWW::Mechanize (); use WWW::Mechanize ();
use List::AllUtils qw/ sum pairmap /; use List::AllUtils qw/ sum pairmap /;
use List::UtilsBy qw/ nsort_by /; use List::UtilsBy qw/ nsort_by partition_by /;
use experimental qw/ signatures /; use experimental qw/ signatures /;
unless(caller) { unless (caller) {
my $ts = get_mersenne_results(); my $ts = get_mersenne_results();
generate_output_table($ts)->draw; generate_output_table($ts)->draw;
@ -41,7 +41,7 @@ sub get_mersenne_results() {
return extract_first_table($html_string); return extract_first_table($html_string);
} }
sub extract_first_table($html) { sub extract_first_table ($html) {
$html =~ s/\n//g; $html =~ s/\n//g;
my $te = HTML::TableExtract->new( depth => 0, count => 2 ); my $te = HTML::TableExtract->new( depth => 0, count => 2 );
@ -54,36 +54,29 @@ sub extract_first_table($html) {
sub generate_output_table ($ts) { sub generate_output_table ($ts) {
# group GHZ Days results by computer, compute GHZ Days per Day (GHZ Days / Days)
my $list;
foreach my $row ( $ts->rows ) {
foreach my $cell ($row) {
my $machine = $cell->[0];
my $days = $cell->[4] =~ s/\s//gr;
my $ghz_days = $cell->[6] =~ s/\s//gr;
if ( $days > 0 ) {
my $perf = $ghz_days / $days;
push $list->{$machine}->@*, $perf;
}
}
}
# create hash with each computer and its average GHz Days per day
my %ranks = pairmap {
$a => sprintf "%.0f", sum(@$b)/@$b;
} %$list;
# sort hash by the average and print to screen
my $tbl = Text::SimpleTable::AutoWidth->new( my $tbl = Text::SimpleTable::AutoWidth->new(
captions => [qw/ Computer GHZDaysPerDay /] ); captions => [qw/ Computer GHZDaysPerDay /] );
$tbl->row( @$_ ) for # group GHZ Days results by computer, compute GHZ Days per Day (GHZ Days / Days)
$tbl->row(@$_) for
reverse reverse
nsort_by { $_->[1] } nsort_by { $_->[1] }
pairmap { [$a,$b] } pairmap { [ $a, $b ] } # group for the sorting
%ranks; 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; return $tbl;
} }