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 |
|
|
17 |
|
my $webpac = new WebPAC( |
18 |
|
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; |
28 |
|
|
29 |
$|=1; |
$|=1; |
30 |
|
|
31 |
my $db = OpenIsis::open( shift @ARGV || '/data/hidra/THS-500/THS' ); |
my $maxmfn = $webpac->open_isis( |
32 |
my $maxmfn = OpenIsis::maxRowid( $db ) || 1; |
filename => shift @ARGV || '/data/hidra/THS/THS', |
33 |
|
lookup => [ |
|
print "rows: $maxmfn\n\n"; |
|
|
|
|
|
my $data; |
|
|
|
|
|
my @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' => 'ma: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' => 'da: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' }, |
39 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
40 |
|
], |
41 |
); |
); |
42 |
|
|
43 |
sub fill_in { |
$log->info("rows: $maxmfn"); |
|
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 ($$data->{$f}->[$i]->{$sf}) { |
|
|
$$found++; |
|
|
return $$data->{$f}->[$i]->{$sf}; |
|
|
} else { |
|
|
return ''; |
|
|
} |
|
|
} |
|
44 |
|
|
45 |
sub get_nosf { |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
|
my ($found,$data,$f,$i) = @_; |
|
|
if ($$data->{$f}->[$i]) { |
|
|
$$found++; |
|
|
return $$data->{$f}->[$i]; |
|
|
} else { |
|
|
return ''; |
|
|
} |
|
|
} |
|
46 |
|
|
47 |
$tmp =~ s/v(\d+)\^(\w)/get_sf(\$found,\$data,$1,$2,$i)/ges; |
while (my $rec = $webpac->fetch_rec) { |
|
$tmp =~ s/v(\d+)/get_nosf(\$found,\$data,$1,$i)/ges; |
|
48 |
|
|
49 |
if ($found) { |
my @ds = $webpac->data_structure($rec); |
50 |
return $tmp; |
|
51 |
} else { |
if ($log->is_debug) { |
52 |
return; |
$log->debug("rec = ",Dumper($rec)); |
53 |
|
$log->debug("ds = ",Dumper(\@ds)); |
54 |
} |
} |
|
} |
|
55 |
|
|
56 |
print "reading database...\n"; |
next if (! @ds); |
57 |
|
|
58 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
my $filename = $webpac->{'current_filename'}; |
59 |
|
|
60 |
my $row = OpenIsis::read( $db, $mfn ); |
if ($filename) { |
61 |
print Dumper($row); |
open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); |
62 |
foreach my $k (keys %{$row}) { |
print OUT $webpac->output( |
63 |
if ($k ne "mfn") { |
template => 'html.tt', |
64 |
foreach my $l (@{$row->{$k}}) { |
data => \@ds, |
65 |
next if (! $l); |
); |
66 |
$l = $cp->convert($l); |
close(OUT); |
67 |
# has subfields? |
} else { |
68 |
my $val; |
print $webpac->output( |
69 |
if ($l =~ m/\^/) { |
template => 'text.tt', |
70 |
foreach my $t (split(/\^/,$l)) { |
data => \@ds, |
71 |
next if (! $t); |
); |
72 |
$val->{substr($t,0,1)} = substr($t,1); |
} |
|
} |
|
|
} else { |
|
|
$val = $l; |
|
|
} |
|
73 |
|
|
74 |
push @{$data->{$mfn}->{$k}}, $val if ($val && $val ne ""); |
my $headline; |
75 |
} |
foreach my $ds (@ds) { |
76 |
|
if ($ds->{'tag'} eq 'headline') { |
77 |
|
$headline = join(" ",@{$ds->{'display'}}); |
78 |
|
last; |
79 |
} |
} |
|
|
|
80 |
} |
} |
81 |
|
|
82 |
foreach my $i (@lookup) { |
my $f = $filename; |
83 |
if ($i->{'eval'}) { |
$f =~ s!out/!!; |
84 |
my $eval = fill_in($i->{'eval'},$data->{$mfn}); |
|
85 |
my $key = fill_in($i->{'key'}, $data->{$mfn}); |
# save into index |
86 |
my $val = fill_in($i->{'val'}, $data->{$mfn}); |
foreach my $ds (@ds) { |
87 |
if ($key && $val && eval $eval) { |
next if (! $ds->{'swish'}); |
88 |
push @{$data->{lookup}->{$key}}, $val; |
|
89 |
} |
$index->insert( |
90 |
} else { |
index_name => $ds->{'tag'}, |
91 |
my $key = fill_in($i->{'key'}, $data->{$mfn}); |
path => $f, |
92 |
my $val = fill_in($i->{'val'}, $data->{$mfn}); |
headline => $headline, |
93 |
if ($key && $val) { |
words => join(" ",@{$ds->{'swish'}}) |
94 |
push @{$data->{lookup}->{$key}}, $val; |
); |
|
} |
|
|
} |
|
95 |
} |
} |
96 |
|
|
97 |
|
# print Dumper(\@ds); |
98 |
|
|
99 |
} |
} |
|
print "\n"; |
|
100 |
|
|
101 |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
if ($log->is_debug) { |
102 |
my $d = $data->{$mfn}; |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
103 |
print $d->{'250'}->[0]->{'a'},"\n"; |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
|
my $t = fill_in('v553^1;;v553^a', $d); |
|
|
print " Uži pojam: $t\n" if ($t); |
|
|
$t = fill_in('[a:v251];;[d:[a:v251]]', $d); |
|
|
print " Uži pojam: $t\n" if ($t); |
|
104 |
} |
} |
105 |
|
|
106 |
print Dumper($data); |
$index->close; |
107 |
|
|