2 |
|
|
3 |
=head1 NAME |
=head1 NAME |
4 |
|
|
5 |
all2all.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; |
use Carp; |
12 |
|
|
13 |
use lib './lib'; |
use lib './lib'; |
14 |
use WebPac; |
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
|
17 |
my $webpac = new WebPac( |
my $webpac = new WebPAC( |
18 |
code_page => 'ISO-8859-2', |
code_page => 'ISO-8859-2', |
19 |
|
limit_mfn => 100, |
20 |
|
# debug => 1, |
21 |
|
) || die; |
22 |
|
|
23 |
|
my $log = $webpac->_get_logger(); |
24 |
|
|
25 |
|
my $index = new WebPAC::jsFind( |
26 |
|
index_path => './out/index', |
27 |
) || die; |
) || die; |
28 |
|
|
29 |
$|=1; |
$|=1; |
30 |
|
|
|
print "reading database\n"; |
|
31 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
32 |
filename => shift @ARGV || '/data/hidra/THS-500/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
33 |
lookup => [ |
lookup => [ |
34 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
35 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
36 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
37 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
38 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
39 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
40 |
], |
], |
41 |
); |
); |
42 |
|
|
43 |
print "rows: $maxmfn\n\n"; |
$log->info("rows: $maxmfn"); |
44 |
|
|
45 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
|
my $rec = $webpac->{'data'}->{$mfn} || die "no record with mfn $mfn"; |
|
46 |
|
|
47 |
print "-- ",$webpac->fill_in($rec,'v250^a (v901^a)'),"\n"; |
while (my $rec = $webpac->fetch_rec) { |
48 |
|
|
49 |
my @t = $webpac->fill_in($rec,'v553^1;;v553^a'); |
my @ds = $webpac->data_structure($rec); |
|
print " Uži pojam: ",join("\t\n",@t),"\n" if (@t); |
|
50 |
|
|
51 |
@t = $webpac->fill_in($rec,'[a:v251::];;[d:[a:v251::]]'); |
if ($log->is_debug) { |
52 |
@t = map { $webpac->lookup($_) => $_ } @t; |
$log->debug("rec = ",Dumper($rec)); |
53 |
print " Područje: ",join("\t\n",@t),"\n" if (@t); |
$log->debug("ds = ",Dumper(\@ds)); |
54 |
|
} |
55 |
|
|
56 |
|
next if (! @ds); |
57 |
|
|
58 |
|
my $filename = $webpac->{'current_filename'}; |
59 |
|
|
60 |
|
if ($filename) { |
61 |
|
open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); |
62 |
|
print OUT $webpac->output( |
63 |
|
template => 'html.tt', |
64 |
|
data => \@ds, |
65 |
|
headline => $webpac->{'headline'}, |
66 |
|
); |
67 |
|
close(OUT); |
68 |
|
} else { |
69 |
|
print $webpac->output( |
70 |
|
template => 'text.tt', |
71 |
|
data => \@ds, |
72 |
|
headline => $webpac->{'headline'}, |
73 |
|
); |
74 |
|
} |
75 |
|
|
76 |
|
my $headline = $webpac->{'headline'}; |
77 |
|
|
78 |
|
my $f = $filename; |
79 |
|
$f =~ s!out/!!; |
80 |
|
|
81 |
|
# save into index |
82 |
|
foreach my $ds (@ds) { |
83 |
|
next if (! $ds->{'swish'}); |
84 |
|
|
85 |
|
$index->insert( |
86 |
|
index_name => $ds->{'tag'}, |
87 |
|
path => $f, |
88 |
|
headline => $headline, |
89 |
|
words => join(" ",@{$ds->{'swish'}}) |
90 |
|
); |
91 |
|
} |
92 |
|
|
93 |
@t = $webpac->fill_in($rec,'[a:v561^4:v251:];;[d:[a:v561^4:v251:]]'); |
# print Dumper(\@ds); |
|
print " Mikrotezaurus: ",join("\t\n",@t),"\n" if (@t); |
|
94 |
|
|
|
@t = $webpac->fill_in($rec,'[a:v561^4:v562^4:v900];;[d:[a:v561^4:v562^4:v900]]'); |
|
|
print " Deskriptor: ",join("\t\n",@t),"\n" if (@t); |
|
95 |
} |
} |
96 |
|
|
97 |
print "## lookup ",Dumper($webpac->{'lookup'}); |
if ($log->is_debug) { |
98 |
print "## data ",Dumper($webpac->{'data'}); |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
99 |
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
100 |
|
} |
101 |
|
|
102 |
|
$index->close; |
103 |
|
|