1 |
dpavlin |
29 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
dpavlin |
52 |
use Search::Estraier 0.06; |
5 |
dpavlin |
29 |
|
6 |
|
|
# score for words in title |
7 |
|
|
my $title_rank = 3; |
8 |
|
|
|
9 |
|
|
open(my $t, "gzip -cd trivia.list.gz |") || die "can't open trivia.list.gz: $!"; |
10 |
|
|
|
11 |
dpavlin |
41 |
# open node |
12 |
dpavlin |
50 |
my $node = Search::Estraier::Node->new( |
13 |
|
|
url => 'http://localhost:1978/node/trivia', |
14 |
|
|
user => 'admin', |
15 |
|
|
passwd => 'admin', |
16 |
dpavlin |
52 |
create => 1, |
17 |
dpavlin |
50 |
); |
18 |
dpavlin |
29 |
|
19 |
|
|
|
20 |
|
|
my $header = 1; |
21 |
|
|
my ($title, $year) = ('',undef); |
22 |
|
|
my $in = 0; |
23 |
|
|
my $trivia = ''; |
24 |
|
|
my @qv; |
25 |
|
|
|
26 |
|
|
my ($all_years,$all_titles,$all_quotes); |
27 |
|
|
|
28 |
|
|
sub qv { |
29 |
|
|
my $t = shift || return; |
30 |
|
|
|
31 |
|
|
sub qv_print { |
32 |
|
|
my $v = shift || return ''; |
33 |
|
|
# $v =~ s/(.*)\s*,\s*(.+)/$2 $1/g; |
34 |
|
|
push @qv, $v; |
35 |
|
|
$all_quotes->{$v}++; |
36 |
|
|
return ''; |
37 |
|
|
} |
38 |
dpavlin |
30 |
$t =~ s#([_'"])([^_'"]+?)\1 \(qv\)#qv_print($2)#ge; |
39 |
dpavlin |
29 |
} |
40 |
|
|
|
41 |
|
|
my $i = 0; |
42 |
|
|
|
43 |
|
|
while(<$t>) { |
44 |
|
|
if ($header && /^=====+/) { |
45 |
|
|
$header = 0; |
46 |
|
|
next; |
47 |
|
|
} |
48 |
|
|
next if $header; |
49 |
|
|
|
50 |
|
|
if (/^#\s+(.*)\s*$/) { |
51 |
|
|
$title = $1; |
52 |
|
|
if ($title =~ m#^("*)(.*)\1\s*\((\d+)\)(:?\s*\(\w+\))*$#) { |
53 |
dpavlin |
30 |
($title, $year) = ($2,$3); |
54 |
|
|
$all_titles->{$title}++; |
55 |
dpavlin |
29 |
$all_years->{$3}++; |
56 |
|
|
} else { |
57 |
|
|
$year = undef; |
58 |
|
|
} |
59 |
|
|
|
60 |
|
|
print "# $title ", ( $year ? "[$year]" : "" ), "\n"; |
61 |
|
|
next; |
62 |
|
|
|
63 |
|
|
} elsif (/^-\s(.*)\s*$/) { |
64 |
|
|
$in = 1; |
65 |
|
|
$trivia = "$1\n"; |
66 |
|
|
qv($1); |
67 |
|
|
} elsif (/^\s\s(.*)\s*$/) { |
68 |
|
|
$trivia .= "$1\n"; |
69 |
|
|
qv($1); |
70 |
|
|
} elsif (/^$/ && $in) { |
71 |
|
|
$i++; |
72 |
|
|
print "$i "; |
73 |
|
|
|
74 |
|
|
# create a document object |
75 |
dpavlin |
50 |
my $doc = Search::Estraier::Document->new; |
76 |
dpavlin |
29 |
|
77 |
|
|
# add attributes to the document object |
78 |
|
|
$doc->add_attr('@uri', "file://localhost/trivia/$i"); |
79 |
dpavlin |
30 |
|
80 |
dpavlin |
29 |
$doc->add_attr('@title', $title); |
81 |
dpavlin |
30 |
$doc->add_hidden_text( |
82 |
|
|
($title x $title_rank) |
83 |
|
|
); |
84 |
|
|
|
85 |
dpavlin |
29 |
$doc->add_attr('@size', length($trivia)); |
86 |
|
|
|
87 |
|
|
$doc->add_attr('year', $year) if ($year); |
88 |
|
|
foreach my $q (@qv) { |
89 |
|
|
$doc->add_attr('quote', $q); |
90 |
|
|
$doc->add_hidden_text($q); |
91 |
|
|
} |
92 |
|
|
|
93 |
|
|
# add the body text to the document object |
94 |
|
|
$doc->add_text($trivia); |
95 |
|
|
|
96 |
|
|
|
97 |
|
|
# register the document object to the database |
98 |
dpavlin |
41 |
$node->put_doc($doc); |
99 |
dpavlin |
29 |
|
100 |
|
|
$trivia = ''; |
101 |
|
|
@qv = (); |
102 |
|
|
$in = 0; |
103 |
|
|
} else { |
104 |
|
|
print "#$_\n"; |
105 |
|
|
} |
106 |
|
|
|
107 |
|
|
# last if ($i > 1000); # XXX remove this! |
108 |
|
|
} |
109 |
|
|
|
110 |
|
|
sub dump_data($$) { |
111 |
|
|
my ($name,$hash) = @_; |
112 |
|
|
|
113 |
|
|
open(my $fh, "> $name") || die "can't open $name: $!"; |
114 |
|
|
|
115 |
|
|
foreach my $k (sort keys %{$hash}) { |
116 |
|
|
print $fh "$k\t",$hash->{$k},"\n"; |
117 |
|
|
} |
118 |
|
|
|
119 |
|
|
close($fh); |
120 |
|
|
} |
121 |
|
|
|
122 |
dpavlin |
59 |
#dump_data('titles.data', $all_titles); |
123 |
|
|
#dump_data('quotes.data', $all_quotes); |
124 |
|
|
#dump_data('years.data', $all_years); |
125 |
dpavlin |
29 |
|