order incoming values before interval

[barcat.git] / barcat

diff --git a/barcat b/barcat
index aec811bc97dca30ae8052a07d9fa0ee29854e744..4c3cf8a763210ae3eb31fb7ecb41a25665da6ac0 100755 (executable)
--- a/barcat
+++ b/barcat
@@ -6,7 +6,7 @@ use List::Util qw( min max sum );
 use open qw( :std :utf8 );
 use experimental qw( lexical_subs );
 
-our $VERSION = '1.03';
+our $VERSION = '1.04';
 
 use Getopt::Long '2.33', qw( :config gnu_getopt );
 sub podexit {
@@ -61,6 +61,8 @@ $opt{units}   = [split //, ' kMGTPEZYyzafpnμm'] if $opt{'human-readable'};
 $opt{anchor} //= qr/\A/;
 $opt{'value-length'} = 6 if $opt{units};
 
+my (@lines, @values, @order);
+
 if (defined $opt{interval}) {
        $opt{interval} ||= 1;
        $SIG{ALRM} = sub {
@@ -68,16 +70,21 @@ if (defined $opt{interval}) {
                alarm $opt{interval};
        };
        alarm $opt{interval};
+
+       eval {
+               require Tie::Array::Sorted;
+               tie @order, 'Tie::Array::Sorted', sub { $_[1] <=> $_[0] };
+       } or warn $@, "Expect slowdown with large datasets!\n";
 }
 
 $SIG{INT} = 'IGNORE';  # continue after assumed eof
 
-my (@lines, @values);
 my $valmatch = qr/$opt{anchor} ( \h* -? [0-9]* \.? [0-9]+ (?: e[+-]?[0-9]+ )? |)/x;
 while (readline) {
        s/\r?\n\z//;
        s/^\h*// unless $opt{unmodified};
        push @values, s/$valmatch/\n/ && $1;
+       push @order, $1 if length $1;
        if (defined $opt{trim}) {
                my $trimpos = abs $opt{trim};
                if ($trimpos <= 1) {
@@ -99,9 +106,9 @@ state $nr = $opt{hidemin} ? $opt{hidemin} - 1 : 0;
 @lines or return;
 @lines > $nr or return unless $opt{hidemin};
 
-my @order  = sort { $b <=> $a } grep { length } @values;
-my $maxval = $opt{hidemax} ? max @values[0 .. $opt{hidemax} - 1] : $order[0];
-my $minval = min $order[-1], 0;
+@order = sort { $b <=> $a } @order unless tied @order;
+my $maxval = ($opt{hidemax} ? max grep { length } @values[0 .. $opt{hidemax} - 1] : $order[0]) // 0;
+my $minval = min $order[-1] // (), 0;
 my $lenval = $opt{'value-length'} // max map { length } @order;
 my $len    = defined $opt{trim} && $opt{trim} <= 0 ? -$opt{trim} + 1 :
        max map { length $values[$_] && length $lines[$_] }
@@ -169,7 +176,16 @@ while ($nr <= $#lines) {
 show_lines();
 
 if ($opt{stat}) {
-       printf '%d values', scalar @values;
+       if ($opt{hidemin} or $opt{hidemax}) {
+               $opt{hidemin} ||= 1;
+               $opt{hidemax} ||= @lines;
+               printf '%s of ', sum(@values[$opt{hidemin} - 1 .. $opt{hidemax} - 1]) // 0;
+       }
+       my $total = sum @order;
+       printf '%s total', $total;
+       printf ' in %d values', scalar @values;
+       printf ' (%s min, %*.*f avg, %s max)',
+               $order[-1], 0, 2, $total / @order, $order[0];
        say '';
 }
 
@@ -346,9 +362,9 @@ Git statistics, such commit count by year:
 
     git log --pretty=%ci | cut -b-4 | uniq -c | barcat
 
-Or the most frequent authors:
+Or the top 3 most frequent authors with statistics over all:
 
-    git shortlog -sn | barcat -L3
+    git shortlog -sn | barcat -L3 -s
 
 Latency history: