prepare anchor match during option parsing

[barcat.git] / barcat

diff --git a/barcat b/barcat
index c4244d4e61d5e99d4c80d6432387782f27d4296d..e6b22b6872becc110a6fe3934077c51e61851b8a 100755 (executable)
--- a/barcat
+++ b/barcat
@@ -17,7 +17,10 @@ my %opt;
 GetOptions(\%opt,
        'color|c!',
        'C' => sub { $opt{color} = 0 },
-       'field|f=s',
+       'field|f=s' => sub {
+               $opt{anchor} = /^[0-9]+$/ ? qr/(?:\S*\h+){$_}\K/ : $_ for $_[1];
+       },
+       'human-readable|H!',
        'interval|t:i',
        'trim|length|l=s' => sub {
                my ($optname, $optval) = @_;
@@ -30,8 +33,15 @@ GetOptions(\%opt,
        },
        'value-length=i',
        'hidemin=i',
-       'hidemax|limit|L=i' => sub {
-               $opt{$_[1] < 0 ? 'hidemin' : 'hidemax'} = abs $_[1];
+       'hidemax=i',
+       'limit|L=s' => sub {
+               my ($optname, $optval) = @_;
+               $optval ||= 0;
+               ($opt{hidemin}, $opt{hidemax}) =
+               $optval =~ m/\A (?: ([0-9]+)? - )? ([0-9]+)? \z/x or die(
+                       "Value \"$optval\" invalid for option limit",
+                       " (range expected)\n"
+               );
        },
        'markers|m=s',
        'unmodified|u!',
@@ -43,6 +53,8 @@ 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{anchor} //= qr/\A/;
 
 if (defined $opt{interval}) {
        $opt{interval} ||= 1;
@@ -56,13 +68,11 @@ 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 $valmatch = qr/$opt{anchor} ( \h* -? [0-9]* \.? [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;
        if (defined $opt{trim}) {
                my $trimpos = abs $opt{trim};
                if ($trimpos <= 1) {
@@ -116,6 +126,12 @@ if ($opt{markers} // 1 and $size > 0) {
        }
 }
 
+sub sival {
+       my $unit = int(log($_[0]) / log(1000) - ($_[0] < 1));
+       sprintf "%3.1f%1s", $_[0] / 1000 ** $unit,
+               $#{$opt{units}} >> 1 < abs $unit ? "e$unit" : $opt{units}->[$unit];
+}
+
 while ($nr <= $#lines) {
        $nr >= $opt{hidemax} and last if $opt{hidemax};
        my $val = $values[$nr];
@@ -124,7 +140,7 @@ while ($nr <= $#lines) {
                        $val == $order[0] ? 32 : # max
                        $val == $order[-1] ? 31 : # min
                        90;
-               $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;
@@ -176,6 +192,11 @@ A string can indicate the starting position of a value
 or capture the numbers itself,
 for example I<-f'(\d+)'> for the first digits anywhere.
 
+=item -H, --human-readable
+
+Format values using SI unit prefixes,
+turning long numbers like I<12356789> into I<12.4M>.
+
 =item -t, --interval[=<seconds>]
 
 Interval time to output partial progress.
@@ -190,7 +211,7 @@ unless C<--length=0>.
 Prepend a dash (i.e. make negative) to enforce padding
 regardless of encountered contents.
 
-=item -L, --limit=[-]<count>
+=item -L, --limit=(<count>|<start>-[<end>])
 
 Stop output after a number of lines.
 All input is still counted and analyzed for statistics,
@@ -284,7 +305,7 @@ Total population history from the World Bank dataset (XML):
 
     curl http://api.worldbank.org/v2/country/1W/indicator/SP.POP.TOTL |
     xmllint --xpath '//*[local-name()="date" or local-name()="value"]' - |
-    sed -r 's,</wb:value>,\n,g; s,(<[^>]+>)+, ,g' | barcat -f1
+    sed -r 's,</wb:value>,\n,g; s,(<[^>]+>)+, ,g' | barcat -f1 -H
 
 Movies per year from prepared JSON data: