2 |
|
|
3 |
use strict; |
use strict; |
4 |
use Search::Estraier 0.06; |
use Search::Estraier 0.06; |
5 |
|
use parse_trivia; |
6 |
|
|
7 |
# score for words in title |
# score for words in title |
8 |
my $title_rank = 3; |
my $title_rank = 3; |
17 |
create => 1, |
create => 1, |
18 |
); |
); |
19 |
|
|
20 |
|
my $nr = 1; |
21 |
|
|
22 |
my $header = 1; |
parse_trivia($t, sub { |
|
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}++; |
|
|
return ''; |
|
|
} |
|
|
$t =~ s#([_'"])([^_'"]+?)\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#^("*)(.*)\1\s*\((\d+)\)(:?\s*\(\w+\))*$#) { |
|
|
($title, $year) = ($2,$3); |
|
|
$all_titles->{$title}++; |
|
|
$all_years->{$3}++; |
|
|
} else { |
|
|
$year = undef; |
|
|
} |
|
|
|
|
|
print "# $title ", ( $year ? "[$year]" : "" ), "\n"; |
|
|
next; |
|
23 |
|
|
24 |
} elsif (/^-\s(.*)\s*$/) { |
my $a = {@_}; |
|
$in = 1; |
|
|
$trivia = "$1\n"; |
|
|
qv($1); |
|
|
} elsif (/^\s\s(.*)\s*$/) { |
|
|
$trivia .= "$1\n"; |
|
|
qv($1); |
|
|
} elsif (/^$/ && $in) { |
|
|
$i++; |
|
|
print "$i "; |
|
25 |
|
|
26 |
# create a document object |
# create a document object |
27 |
my $doc = Search::Estraier::Document->new; |
my $doc = Search::Estraier::Document->new; |
28 |
|
|
29 |
# add attributes to the document object |
# add attributes to the document object |
30 |
$doc->add_attr('@uri', "file://localhost/trivia/$i"); |
$doc->add_attr('@uri', "file://localhost/trivia/$nr"); |
31 |
|
|
32 |
$doc->add_attr('@title', $title); |
$doc->add_attr('@title', $a->{title}); |
33 |
$doc->add_hidden_text( |
$doc->add_hidden_text( |
34 |
($title x $title_rank) |
(($a->{title} . ' ') x $title_rank) |
35 |
); |
); |
36 |
|
|
37 |
$doc->add_attr('@size', length($trivia)); |
$doc->add_attr('@size', length($a->{trivia})); |
38 |
|
|
39 |
$doc->add_attr('year', $year) if ($year); |
$doc->add_attr('year', $a->{year}) if ($a->{year}); |
40 |
foreach my $q (@qv) { |
foreach my $q ( $a->{qv} ) { |
41 |
$doc->add_attr('quote', $q); |
$doc->add_attr('quote', $q); |
42 |
$doc->add_hidden_text($q); |
$doc->add_hidden_text($q); |
43 |
} |
} |
44 |
|
|
45 |
# add the body text to the document object |
# add the body text to the document object |
46 |
$doc->add_text($trivia); |
$doc->add_text($a->{trivia}); |
47 |
|
|
48 |
|
|
49 |
# register the document object to the database |
# register the document object to the database |
50 |
$node->put_doc($doc); |
$node->put_doc($doc); |
51 |
|
|
52 |
$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); |
|
|
|
|