#!/usr/bin/perl                                                                                               
#
# pm_sample_histo_test.pl <FNames>
#
# Make a brief histogram to see if there's any obvious bias in the sampler.
# Assumes you're sampling from a file containing a million lines, each of
# which contains just the line number.
#
use 5.10.1;
use strict;
use warnings;
 
my %H;
 
for my $FName (@ARGV) {
    print "-- $FName --\n";
    open my $FH, '<', $FName or die "can't open $FName\n";
    while (<$FH>) {
        next if /^random/;
        my $I = int $_/10000;
        $H{$I}++;
    }
}
 
my $max=0;
for (keys %H) { $max = $H{$_} if $H{$_} > $max; }
 
for my $I (0 .. 99) {
    my $bar = substr('*' x int(50 * ($H{$I}/$max)) . ' 'x50, 0, 50);
    if ($I > 1 and $I < 98) {
        my $avg;
        my $sum = 0;
        $sum += $H{$_} for $I-2 .. $I+2;
        $avg = $sum/5;
        substr($bar,int(50 * ($avg/$max)),1)='+';
    }
    printf "% 3u: (% 3u) %s\n", $I, $H{$I} // 0, $bar;
}