source: nawszelkiwypadek/tools/gue_dic/count_prefs.pl @ 635ee52

#! /usr/bin/perl

use locale;
use strict;

my @prefs;

sub addPref {

    my $pref = shift;
    my $desc = shift;
    my $i;
    for ($i=0; $i< @prefs; ++$i) {
        my @tab = @{$prefs[$i]};
        if (${@{$prefs[$i]}}[0] =~ /^$pref/) {
            ${@{$prefs[$i]}}[1]{$desc}++;
            return;
        }
    }
    my @new;
    my %hash;

    $hash{$desc}++;
    push(@new, $pref);
    push(@new, \%hash);

    push(@prefs, \@new);
}

sub printPrefs {

    my $i;
    for $i (@prefs) {
        my @tab = @$i;
#       print $tab[0]."\t";
        my $pref = $tab[0];
        my %hash = %{$tab[1]};
        my @keys = keys(%hash);
#       print(@keys."\n");
        my $sum = 0;
        my $key;
        for $key (@keys) {
            $sum += $hash{$key};
        }
        for $key (@keys) {
            print $pref."\t";
            print $key."\t";
            print $hash{$key}."\t";
            print $sum."\n";
        }
    }
}

if (@ARGV < 2) {
    print "USAGE: count_prefs.pl MIN_PREF_LEN MAX_PREF_LEN\n";
    exit;
}

my $MIN = shift;
my $MAX = shift;
my $PART = shift;

if ($MIN > $MAX) {
    print "MIN_PREF_LEN > MAX_PREF_LEN! ($MIN > $MAX)\n";
    exit;
}
my $begin = "";
while (<>) {
    my $len = $MIN;
    $_ =~ /(\w+);(.*)$/;
    my $pref = $1;
    my $desc = $2;
    if ($begin eq "")  {
        $begin = substr($pref, 0, $MIN);
    }
    if ($pref !~ /^$begin.*/) {
        printPrefs();
        undef(@prefs);
        $begin = "";
    }
    while ($len <= $MAX) {
        addPref(substr($pref, 0, $len++), $desc);
    }
}

printPrefs();