X-Git-Url: http://git.shiar.nl/barcat.git/blobdiff_plain/b04f42d23d88fce792cc0ad8db5a0c1e35477143..cdad4c9fc575cb94974ae27e451aed47fc072710:/barcat

diff --git a/barcat b/barcat
index 40f332d..4c3cf8a 100755
--- 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 {
@@ -47,6 +47,7 @@ GetOptions(\%opt,
 		);
 	},
 	'markers|m=s',
+	'stat|s!',
 	'unmodified|u!',
 	'width|w=i',
 	'usage|h' => sub { podexit() },
@@ -56,10 +57,12 @@ GetOptions(\%opt,
 $opt{width} ||= $ENV{COLUMNS} || 80;
 $opt{color} //= -t *STDOUT;  # enable on tty
 $opt{trim}   *= $opt{width} / 100 if $opt{trimpct};
-$opt{units}   = $opt{'human-readable'} && ['', qw( k M G T P E Z Y y z a f p n μ m )];
+$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 {
@@ -67,16 +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) {
@@ -95,10 +103,12 @@ sub show_lines {
 
 state $nr = $opt{hidemin} ? $opt{hidemin} - 1 : 0;
 @lines and @lines > $nr or return;
+@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 $lines[$_] }
@@ -130,19 +140,21 @@ if ($opt{markers} // 1 and $size > 0) {
 	}
 }
 
+@lines > $nr or return if $opt{hidemin};
+
 sub sival {
 	my $unit = int(log($_[0]) / log(1000) - ($_[0] < 1));
 	my $float = $_[0] !~ /^ (?: 0*\.)? [0-9]{1,3} $/x;
 	sprintf('%*.*f%*s',
 		$float ? 5 : 3, $float,  # length and tenths
 		$_[0] / 1000 ** $unit,   # number
-		$float ? 1 : 3,          # unit size
+		$float ? 0 : 3,          # unit size
 		$#{$opt{units}} >> 1 < abs $unit ? "e$unit" : $opt{units}->[$unit]
 	);
 }
 
 while ($nr <= $#lines) {
-	$nr >= $opt{hidemax} and last if $opt{hidemax};
+	$nr >= $opt{hidemax} and last if defined $opt{hidemax};
 	my $val = $values[$nr];
 	if (length $val) {
 		my $color = !$opt{color} ? 0 :
@@ -163,6 +175,20 @@ while ($nr <= $#lines) {
 }
 show_lines();
 
+if ($opt{stat}) {
+	if ($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 '';
+}
+
 __END__
 
 =head1 NAME
@@ -261,6 +287,10 @@ or 68% of all entries.
 
 =back
 
+=item -s, --stat
+
+Total statistics after all data.
+
 =item -u, --unmodified
 
 Do not strip leading whitespace.
@@ -332,9 +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: