trunk/data/parse_trivia.pm

use strict;

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

my ($all_years,$all_titles,$all_quotes);

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

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

sub fix_title {
        my $t = shift;
        $t =~ s/,\s+The\s*$//;
        return $t;
};

my $i = 0;

sub parse_trivia {
        my ($t, $call) = @_;

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

                if (/^#\s+(.*)\s*$/) {
                        $title = $1;
                        if ($title =~ m#^("*)(.*)\1\s*\((\d+)\)(:?\s*\(\w+\))*$#) {
                                ($title, $year) = (fix_title($2),$3);
                                $all_titles->{$title}++;
                                $all_years->{$3}++;
                        } 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] ";

                        $call->(
                                title => $title,
                                year => $year,
                                trivia => $trivia,
                                qv => [ @qv ],
                        );

                        $trivia = '';
                        @qv = ();
                        $in = 0;
                } else {
                        print "#$_\n";
                }

                #       last if ($i > 1000);    # XXX remove this!
        }
}

sub dump_data($$) {
        my ($name,$hash) = @_;

        open(my $fh, "> $name") || die "can't open $name: $!";

        foreach my $k (sort keys %{$hash}) {
                print $fh "$k\t",$hash->{$k},"\n";
        }

        close($fh);
}

#dump_data('titles.data', $all_titles);
#dump_data('quotes.data', $all_quotes);
#dump_data('years.data', $all_years);

1;
1	dpavlin	83	use strict;
2
3			my $header = 1;
4			my ($title, $year) = ('',undef);
5			my $in = 0;
6			my $trivia = '';
7			my @qv;
8
9			my ($all_years,$all_titles,$all_quotes);
10
11			sub qv {
12			my $t = shift \|\| return;
13
14			sub qv_print {
15			my $v = shift \|\| return '';
16			# $v =~ s/(.)\s,\s*(.+)/$2 $1/g;
17			push @qv, $v;
18			$all_quotes->{$v}++;
19			print "+ $v ";
20			return '';
21			}
22			$t =~ s#([_'"])([^_'"]+?)\1 \(qv\)#qv_print($2)#ge;
23			}
24
25			sub fix_title {
26			my $t = shift;
27			$t =~ s/,\s+The\s*$//;
28			return $t;
29			};
30
31			my $i = 0;
32
33			sub parse_trivia {
34			my ($t, $call) = @_;
35
36			while(<$t>) {
37			if ($header && /^=====+/) {
38			$header = 0;
39			next;
40			}
41			next if $header;
42
43			if (/^#\s+(.)\s$/) {
44			$title = $1;
45			if ($title =~ m#^(")(.)\1\s\((\d+)\)(:?\s\(\w+\))*$#) {
46			($title, $year) = (fix_title($2),$3);
47			$all_titles->{$title}++;
48			$all_years->{$3}++;
49			} else {
50			$year = undef;
51			}
52
53			print "# $title ", ( $year ? "[$year]" : "" ), "\n";
54			next;
55
56			} elsif (/^-\s(.)\s$/) {
57			$in = 1;
58			$trivia = "$1\n";
59			qv($1);
60			} elsif (/^\s\s(.)\s$/) {
61			$trivia .= "$1\n";
62			qv($1);
63			} elsif (/^$/ && $in) {
64			$i++;
65			print "[$i] ";
66
67			$call->(
68			title => $title,
69			year => $year,
70			trivia => $trivia,
71			qv => [ @qv ],
72			);
73
74			$trivia = '';
75			@qv = ();
76			$in = 0;
77			} else {
78			print "#$_\n";
79			}
80
81			# last if ($i > 1000); # XXX remove this!
82			}
83			}
84
85			sub dump_data($$) {
86			my ($name,$hash) = @_;
87
88			open(my $fh, "> $name") \|\| die "can't open $name: $!";
89
90			foreach my $k (sort keys %{$hash}) {
91			print $fh "$k\t",$hash->{$k},"\n";
92			}
93
94			close($fh);
95			}
96
97			#dump_data('titles.data', $all_titles);
98			#dump_data('quotes.data', $all_quotes);
99			#dump_data('years.data', $all_years);
100
101			1;