trunk/data/convert.pl

#!/usr/bin/perl -w

use strict;

# score for words in title
my $title_rank = 3;

open(my $t, "gzip -cd trivia.list.gz |") || die "can't open trivia.list.gz: $!";


my $header = 1;
my ($title, $year) = ('',undef);
my $in = 0;
my $trivia = '';
my @qv;

sub qv {
        my $t = shift || return;

        sub qv_print {
                my $v = shift || return '';
                $v =~ s/(.*)\s*,\s*(.+)/$2 $1/g;
                push @qv, $v;
                return '';
        }
        $t =~ s#([_'])([^\1]+?)\1 \(qv\)#qv_print($2)#ge;
}

my $i = 0;

while(<$t>) {
        if ($header && /^=====+/) {
                $header = 0;
                next;
        }
        next if $header;

        if (/^#\s+(.*)\s*$/) {
                $title = $1;
                if ($title =~ m#\s*\((\d+)\)\s*$#) {
                        $year = $1;
                } else {
                        $year = undef;
                }

                print "# $title ", ( $year ? "[$year]" : "" ), "\n";
                next;

        } elsif (/^-\s(.*)\s*$/) {
                $in = 1;
                $trivia = "$1\n";
                qv($1);
        } elsif (/^\s\s(.*)\s*$/) {
                $trivia .= "$1\n";
                qv($1);
        } elsif (/^$/ && $in) {
                $i++;
                print "$i ";
                my $out = '@title=' . $title . "\n" .
                        '@size=' . length($trivia) . "\n";
                $out .= "year=$year\n" if ($year);
                $out .= 'quote=' . join(",", @qv) . "\n" if (@qv);
                $out .= "\n$trivia\n" .
                        ("\t$title\n" x $title_rank) .
                        "\t" . join("\n\t", @qv) . "\n";

                open(my $t, "> trivia/$i.est") || die "can't open $i.est: $!";
                print $t $out;
                close($t);
                $trivia = '';
                @qv = ();
                $in = 0;
        } else {
                print "#$_\n";
        }

}

1	dpavlin	6	#!/usr/bin/perl -w
2
3			use strict;
4
5	dpavlin	10	# score for words in title
6			my $title_rank = 3;
7
8	dpavlin	6	open(my $t, "gzip -cd trivia.list.gz \|") \|\| die "can't open trivia.list.gz: $!";
9
10
11			my $header = 1;
12	dpavlin	17	my ($title, $year) = ('',undef);
13	dpavlin	6	my $in = 0;
14			my $trivia = '';
15			my @qv;
16
17			sub qv {
18			my $t = shift \|\| return;
19
20			sub qv_print {
21			my $v = shift \|\| return '';
22			$v =~ s/(.)\s,\s*(.+)/$2 $1/g;
23			push @qv, $v;
24			return '';
25			}
26			$t =~ s#([_'])([^\1]+?)\1 \(qv\)#qv_print($2)#ge;
27			}
28
29			my $i = 0;
30
31			while(<$t>) {
32	dpavlin	17	if ($header && /^=====+/) {
33	dpavlin	6	$header = 0;
34			next;
35			}
36			next if $header;
37
38			if (/^#\s+(.)\s$/) {
39			$title = $1;
40	dpavlin	17	if ($title =~ m#\s\((\d+)\)\s$#) {
41			$year = $1;
42			} else {
43			$year = undef;
44			}
45
46			print "# $title ", ( $year ? "[$year]" : "" ), "\n";
47	dpavlin	6	next;
48	dpavlin	17
49	dpavlin	6	} elsif (/^-\s(.)\s$/) {
50			$in = 1;
51			$trivia = "$1\n";
52			qv($1);
53			} elsif (/^\s\s(.)\s$/) {
54			$trivia .= "$1\n";
55			qv($1);
56			} elsif (/^$/ && $in) {
57			$i++;
58			print "$i ";
59	dpavlin	17	my $out = '@title=' . $title . "\n" .
60			'@size=' . length($trivia) . "\n";
61			$out .= "year=$year\n" if ($year);
62			$out .= 'quote=' . join(",", @qv) . "\n" if (@qv);
63			$out .= "\n$trivia\n" .
64			("\t$title\n" x $title_rank) .
65			"\t" . join("\n\t", @qv) . "\n";
66
67	dpavlin	6	open(my $t, "> trivia/$i.est") \|\| die "can't open $i.est: $!";
68	dpavlin	17	print $t $out;
69	dpavlin	6	close($t);
70			$trivia = '';
71			@qv = ();
72			$in = 0;
73			} else {
74			print "#$_\n";
75			}
76
77			}
78