2 |
|
|
3 |
=head1 NAME |
=head1 NAME |
4 |
|
|
5 |
webpac.pl - basic script for all WebPac needs |
all2all.pl - basic script for all WebPAC needs |
6 |
|
|
7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
|
use OpenIsis; |
|
|
use Text::Iconv; |
|
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
|
use Carp; |
12 |
|
|
13 |
my $cp = Text::Iconv->new("852", "iso-8859-2"); |
use lib './lib'; |
14 |
|
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
use WebPAC::Index; |
17 |
|
|
18 |
|
my $webpac = new WebPAC( |
19 |
|
code_page => 'ISO-8859-2', |
20 |
|
limit_mfn => 500, |
21 |
|
# debug => 1, |
22 |
|
) || die; |
23 |
|
|
24 |
|
my $log = $webpac->_get_logger(); |
25 |
|
|
26 |
|
my $index = new WebPAC::jsFind( |
27 |
|
index_path => './out/index', |
28 |
|
keys => 10, |
29 |
|
) || die; |
30 |
|
|
31 |
|
my $thes; |
32 |
|
|
33 |
$|=1; |
$|=1; |
34 |
|
|
35 |
my $db = OpenIsis::open( shift @ARGV || '/data/hidra/THS-500/THS' ); |
my $maxmfn = $webpac->open_isis( |
36 |
my $maxmfn = OpenIsis::maxRowid( $db ) || 1; |
filename => shift @ARGV || '/data/hidra/THS/THS', |
37 |
|
lookup => [ |
38 |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
39 |
|
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
40 |
|
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
41 |
|
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
42 |
|
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
43 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
44 |
|
], |
45 |
|
); |
46 |
|
|
47 |
print "rows: $maxmfn\n\n"; |
$log->info("rows: $maxmfn"); |
48 |
|
|
49 |
my $data; |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
50 |
|
|
51 |
my @lookup = ( |
while (my $rec = $webpac->fetch_rec) { |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
|
|
{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
|
|
{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'ma:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
|
|
{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'da:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
|
|
); |
|
52 |
|
|
53 |
sub fill_in { |
my @ds = $webpac->data_structure($rec); |
|
my $tmp = shift || return; |
|
|
my $data = shift || return; |
|
|
my $i = shift || 0; |
|
|
|
|
|
my $found = 0; |
|
|
|
|
|
sub get_sf { |
|
|
my ($found,$data,$f,$sf,$i) = @_; |
|
|
if ($$data->{$f} && $$data->{$f}->[$i]->{$sf}) { |
|
|
$$found++; |
|
|
return $$data->{$f}->[$i]->{$sf}; |
|
|
} else { |
|
|
return ''; |
|
|
} |
|
|
} |
|
54 |
|
|
55 |
sub get_nosf { |
if (0 && $log->is_debug) { |
56 |
my ($found,$data,$f,$i) = @_; |
$log->debug("rec = ",Dumper($rec)); |
57 |
if ($$data->{$f} && $$data->{$f}->[$i]) { |
$log->debug("ds = ",Dumper(\@ds)); |
|
$$found++; |
|
|
return $$data->{$f}->[$i]; |
|
|
} else { |
|
|
return ''; |
|
|
} |
|
58 |
} |
} |
59 |
|
|
60 |
$tmp =~ s/v(\d+)\^(\w)/get_sf(\$found,\$data,$1,$2,$i)/ges; |
next if (! @ds); |
|
$tmp =~ s/v(\d+)/get_nosf(\$found,\$data,$1,$i)/ges; |
|
61 |
|
|
62 |
if ($found) { |
my $filename = $webpac->{'current_filename'}; |
63 |
return $tmp; |
|
64 |
|
if ($filename) { |
65 |
|
$webpac->output_file( |
66 |
|
file => $filename, |
67 |
|
template => 'html.tt', |
68 |
|
data => \@ds, |
69 |
|
headline => $webpac->{'headline'}, |
70 |
|
); |
71 |
} else { |
} else { |
72 |
return; |
print $webpac->output( |
73 |
|
template => 'text.tt', |
74 |
|
data => \@ds, |
75 |
|
headline => $webpac->{'headline'}, |
76 |
|
); |
77 |
} |
} |
|
} |
|
78 |
|
|
79 |
print "reading database...\n"; |
my $headline = $webpac->{'headline'}; |
80 |
|
|
81 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
my $f = $filename; |
82 |
|
$f =~ s!out/!!; |
83 |
|
|
84 |
my $row = OpenIsis::read( $db, $mfn ); |
# save into index |
85 |
foreach my $k (keys %{$row}) { |
foreach my $ds (@ds) { |
86 |
if ($k ne "mfn") { |
next if (! $ds->{'swish'}); |
87 |
foreach my $l (@{$row->{$k}}) { |
|
88 |
$l = $cp->convert($l); |
$index->insert( |
89 |
# has subfields? |
index_name => $ds->{'tag'}, |
90 |
my $val; |
path => $f, |
91 |
if ($l =~ m/\^/) { |
headline => $headline, |
92 |
foreach my $t (split(/\^/,$l)) { |
words => join(" ",@{$ds->{'swish'}}) |
93 |
next if (! $t); |
); |
94 |
$val->{substr($t,0,1)} = substr($t,1); |
} |
|
} |
|
|
} else { |
|
|
$val = $l; |
|
|
} |
|
95 |
|
|
96 |
push @{$data->{$mfn}->{$k}}, $val; |
# save into sorted index (thesaurus) |
97 |
} |
foreach my $ds (@ds) { |
98 |
|
next if (! $ds->{'index'}); |
99 |
|
|
100 |
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
101 |
|
|
102 |
|
foreach my $h (@{$ds->{'index'}}) { |
103 |
|
$thes->{$ds->{'tag'}}->insert( |
104 |
|
path => $f, |
105 |
|
headline => $h, |
106 |
|
); |
107 |
} |
} |
|
|
|
108 |
} |
} |
109 |
|
|
110 |
|
# print Dumper(\@ds); |
111 |
|
|
112 |
foreach my $i (@lookup) { |
} |
113 |
my $d = $data->{$mfn}; |
|
114 |
if ($i->{'eval'}) { |
foreach my $t (keys %{$thes}) { |
115 |
my $eval = fill_in($i->{'eval'},$d); |
|
116 |
my $key = fill_in($i->{'key'}, $d); |
my @e = $thes->{$t}->elements; |
117 |
my $val = fill_in($i->{'val'}, $d); |
if (! @e) { |
118 |
if ($key && $val && eval $eval) { |
$log->logwarn("no elements in sorted index $t?"); |
119 |
push @{$data->{lookup}->{$key}}, $val; |
next; |
|
} |
|
|
} else { |
|
|
my $key = fill_in($i->{'key'}, $d); |
|
|
my $val = fill_in($i->{'val'}, $d); |
|
|
if ($key && $val) { |
|
|
push @{$data->{lookup}->{$key}}, $val; |
|
|
} |
|
|
} |
|
120 |
} |
} |
121 |
|
|
122 |
|
my $file = "./out/bfilter/$t.txt"; |
123 |
|
$log->info("saving sorted index $t to '$file' [".scalar(@e)." elements]"); |
124 |
|
|
125 |
|
$webpac->output_file( |
126 |
|
file => $file, |
127 |
|
template => 'index.tt', |
128 |
|
data => \@e, |
129 |
|
index_name => $t, |
130 |
|
); |
131 |
} |
} |
|
print "\n"; |
|
132 |
|
|
133 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
if (0 && $log->is_debug) { |
134 |
my $d = $data->{$mfn}; |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
135 |
print $d->{'250'}->[0]->{'a'},"\n"; |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
136 |
my $t = fill_in('v553^1;;v553^a', $d); |
foreach my $t (keys %{$thes}) { |
137 |
print " Uži pojam: $t\n" if ($t); |
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
138 |
$t = fill_in('[a:v251];;[d:[a:v251]]', $d); |
} |
|
print " Uži pojam: $t\n" if ($t); |
|
139 |
} |
} |
140 |
|
|
141 |
print Dumper($data); |
$index->close; |
142 |
|
|