keep ordered values, only resort on interval

[barcat.git] / barcat

diff --git a/barcat b/barcat
index 5ac55bab9cfbe5987aa818c2a5ef512644d40e40..1114656f74c0cab9a0a69eb91f3599ca78ffea26 100755 (executable)
--- a/barcat
+++ b/barcat
@@ -17,7 +17,12 @@ my %opt;
 GetOptions(\%opt,
        'color|c!',
        'C' => sub { $opt{color} = 0 },
-       'field|f=s',
+       'field|f=s' => sub {
+               eval {
+                       local $_ = $_[1];
+                       $opt{anchor} = /^[0-9]+$/ ? qr/(?:\S*\h+){$_}\K/ : qr/$_/;
+               } or die $@ =~ s/(?: at .+)?$/ for option $_[0]/r;
+       },
        'human-readable|H!',
        'interval|t:i',
        'trim|length|l=s' => sub {
@@ -42,6 +47,7 @@ GetOptions(\%opt,
                );
        },
        'markers|m=s',
+       'stat|s!',
        'unmodified|u!',
        'width|w=i',
        'usage|h' => sub { podexit() },
@@ -51,7 +57,9 @@ GetOptions(\%opt,
 $opt{width} ||= $ENV{COLUMNS} || 80;
 $opt{color} //= -t *STDOUT;  # enable on tty
 $opt{trim}   *= $opt{width} / 100 if $opt{trimpct};
-$opt{units}   = $opt{'human-readable'} && ['', qw( k M G T <> n μ m )];
+$opt{units}   = [split //, ' kMGTPEZYyzafpnμm'] if $opt{'human-readable'};
+$opt{anchor} //= qr/\A/;
+$opt{'value-length'} = 6 if $opt{units};
 
 if (defined $opt{interval}) {
        $opt{interval} ||= 1;
@@ -64,14 +72,13 @@ if (defined $opt{interval}) {
 
 $SIG{INT} = 'IGNORE';  # continue after assumed eof
 
-my (@lines, @values);
-my $anchor = !defined $opt{field} ? qr/\A/ :
-       $opt{field} =~ /^[0-9]+$/ ? qr/(?:\S*\h+){$opt{field}}\K/ :
-       $opt{field};
+my (@lines, @values, @order);
+my $valmatch = qr/$opt{anchor} ( \h* -? [0-9]* \.? [0-9]+ (?: e[+-]?[0-9]+ )? |)/x;
 while (readline) {
        s/\r?\n\z//;
        s/^\h*// unless $opt{unmodified};
-       push @values, s/$anchor ( \h* -? [0-9]* \.? [0-9]+ |)/\n/x && $1;
+       push @values, s/$valmatch/\n/ && $1;
+       push @order, $1 if length $1;
        if (defined $opt{trim}) {
                my $trimpos = abs $opt{trim};
                if ($trimpos <= 1) {
@@ -90,10 +97,12 @@ 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;
+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 $lines[$_] }
@@ -125,17 +134,28 @@ 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 ? 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 :
                        $val == $order[0] ? 32 : # max
                        $val == $order[-1] ? 31 : # min
                        90;
-               $val = sprintf "%3.1f%1s", $val / 1000**$_, $opt{units}->[$_]
-                       for $opt{units} ? int(log($val) / log(1000) - ($val < 1)) : ();
-               $val = sprintf "%*s", $lenval, $val;
+               $val = $opt{units} ? sival($val) : sprintf "%*s", $lenval, $val;
                $val = "\e[${color}m$val\e[0m" if $color;
        }
        my $line = $lines[$nr] =~ s/\n/$val/r;
@@ -149,6 +169,20 @@ while ($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
@@ -191,6 +225,8 @@ for example I<-f'(\d+)'> for the first digits anywhere.
 
 Format values using SI unit prefixes,
 turning long numbers like I<12356789> into I<12.4M>.
+Also changes an exponent I<1.602176634e-19> to I<160.2z>.
+Short integers are aligned but kept without decimal point.
 
 =item -t, --interval[=<seconds>]
 
@@ -245,6 +281,10 @@ or 68% of all entries.
 
 =back
 
+=item -s, --stat
+
+Total statistics after all data.
+
 =item -u, --unmodified
 
 Do not strip leading whitespace.
@@ -316,9 +356,9 @@ 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: