#!/usr/bin/perl -w
use strict;
#
#  OHlottery.pl By Scott Cain
#
#  Uses CPAN modules to contact the Ohio Lottery web site search page
#  and extracts the winning numbers for all super lotto drawings for which
#  there is data, which starts at the beginning of 1998.  The frequency of
#  each number's occurrence is calculated and compared to the expected
#  value, i.e., 1/47 or 1/49 depending on when the drawing was held.
#
#  Things that could be done to make this better, but that I probably
#  won't do:
#    - have the results saved to a flat file so the script doesn't have
#      to hit the web server for every past result.
#    - make it more general to allow analysis of other games.
#

use LWP::Simple qw(get);
use HTML::TableExtract;
use Date::Calc qw(Day_of_Week
  Add_Delta_Days
  Today
  Date_to_Days
  Decode_Day_of_Week);

my @start_date = qw/1998 1 3/;    # a Saturday, the first for which data is available.
my @current_date = @start_date;
my $current_dow  = Day_of_Week(@current_date);
my $delta_days;

my %lotto;
my %lottoplus;
my $lotto_count     = 0;
my $lottoplus_count = 0;

#
# properly initialize hashes
#
for my $i ( 1 .. 49 ) {
    $lotto{$i}     = 0;
    $lottoplus{$i} = 0;
}

while ( Date_to_Days(@current_date) < Date_to_Days( Today() ) ) {

    my $URL =
      "http://www.ohiolottery.com/numbers/searchresults_bydate.asp?FromMonth="
      . $current_date[1] . "&FromDay=" . $current_date[2] . "&FromYear="
      . $current_date[0];

    my $page = get($URL);
    if ( $page ) {

        my $te = new HTML::TableExtract( depth => 1 );
        $te->parse($page);
        foreach my $ts ( $te->table_states ) {
            foreach my $row ( $ts->rows ) {
                my @game_str = grep ( /lotto/i, @$row );
                if ( $game_str[0] and
                     $game_str[0] =~ /plus/i and #differentiates old from new lotto
                     $$row[3] =~ /(\d+)-(\d+)-(\d+)-(\d+)-(\d+)-(\d+)/ )
                {
                    $lottoplus{$1}++;
                    $lottoplus{$2}++;
                    $lottoplus{$3}++;
                    $lottoplus{$4}++;
                    $lottoplus{$5}++;
                    $lottoplus{$6}++;
                    $lottoplus_count++;
                }
                elsif ( $game_str[0] and
                        $$row[3] =~ /(\d+)-(\d+)-(\d+)-(\d+)-(\d+)-(\d+)/)
                {
                    $lotto{$1}++;
                    $lotto{$2}++;
                    $lotto{$3}++;
                    $lotto{$4}++;
                    $lotto{$5}++;
                    $lotto{$6}++;
                    $lotto_count++;
                }
            }
        }
    }
    else {    # the LWP get didn't work
        print "Date_to_Text(@current_date) failed\n";
    }

    #
    # figure out the next day to use.
    #  (drawings are only held on Saturdays and Wednesdays)
    #
    $current_dow = Day_of_Week(@current_date);
    if ( $current_dow == Decode_Day_of_Week("Saturday") ) {
        $delta_days = 4;
    }
    else {
        $delta_days = 3;
    }
    @current_date = Add_Delta_Days( @current_date, $delta_days );
} # closes while(date) loop

#
# now do some simple statistics.
#
my $num_lottoplus_balls = 6 * $lottoplus_count;
my $num_lotto_balls     = 6 * $lotto_count;

for my $i ( 1 .. 49 ) {
    $lottoplus{$i} = $lottoplus{$i} / $num_lottoplus_balls;
    $lotto{$i}     = $lotto{$i} / $num_lotto_balls;
}

my @toptobottom = sort { $lottoplus{$b} <=> $lottoplus{$a} } keys %lottoplus;

#
# print fairly pretty results--
# probably could have used a sub here, but cut & paste is so convienent.
#
my $expect = 1.0 / 49.0;
print 'Percentages are % deviation from expected value';
print "\nTop 8 superlotto plus balls\n";
print "for $lottoplus_count drawings\n";

for my $i ( 0 .. 7 ) {
    my $deviation = 100 * ( $lottoplus{ $toptobottom[$i] } - $expect ) / $expect;
    printf( "%.2d -> %+.1f%%\n", $toptobottom[$i], $deviation );
}

print "\nBottom 8 superlotto plus balls\n";
for my $i ( 41 .. 48 ) {
    my $deviation = 100 * ( $lottoplus{ $toptobottom[$i] } - $expect ) / $expect;
    printf( "%.2d -> %+.1f%%\n", $toptobottom[$i], $deviation );
}

@toptobottom = sort { $lotto{$b} <=> $lotto{$a} } keys %lotto;

$expect = 1.0 / 47.0;
print "\n\nOld Super Lotto results included for historical comparison.\n";
print "\nTop 8 superlotto balls\n";
print "for $lotto_count drawings\n";
for my $i ( 0 .. 7 ) {
    my $deviation = 100 * ( $lotto{ $toptobottom[$i] } - $expect ) / $expect;
    printf( "%.2d -> %+.1f%%\n", $toptobottom[$i], $deviation );
}

print "\nBottom 8 superlotto balls\n";
for my $i ( 39 .. 46 ) {
    my $deviation = 100 * ( $lotto{ $toptobottom[$i] } - $expect ) / $expect;
    printf( "%.2d -> %+.1f%%\n", $toptobottom[$i], $deviation );
}