1 |
dpavlin |
1 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
dpavlin |
348 |
=head1 NAME |
4 |
|
|
|
5 |
dpavlin |
354 |
all2all.pl - basic script for all WebPAC needs |
6 |
dpavlin |
348 |
|
7 |
|
|
=cut |
8 |
|
|
|
9 |
dpavlin |
1 |
use strict; |
10 |
|
|
use Data::Dumper; |
11 |
dpavlin |
352 |
use Carp; |
12 |
dpavlin |
1 |
|
13 |
dpavlin |
352 |
use lib './lib'; |
14 |
dpavlin |
354 |
use WebPAC; |
15 |
dpavlin |
390 |
use WebPAC::jsFind; |
16 |
dpavlin |
348 |
|
17 |
dpavlin |
354 |
my $webpac = new WebPAC( |
18 |
dpavlin |
352 |
code_page => 'ISO-8859-2', |
19 |
dpavlin |
407 |
# limit_mfn => 100, |
20 |
dpavlin |
372 |
# debug => 1, |
21 |
dpavlin |
352 |
) || die; |
22 |
|
|
|
23 |
dpavlin |
373 |
my $log = $webpac->_get_logger(); |
24 |
|
|
|
25 |
dpavlin |
390 |
my $index = new WebPAC::jsFind( |
26 |
|
|
index_path => './out/index', |
27 |
dpavlin |
407 |
keys => 80, |
28 |
dpavlin |
390 |
) || die; |
29 |
|
|
|
30 |
dpavlin |
10 |
$|=1; |
31 |
dpavlin |
9 |
|
32 |
dpavlin |
352 |
my $maxmfn = $webpac->open_isis( |
33 |
dpavlin |
357 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
34 |
dpavlin |
352 |
lookup => [ |
35 |
dpavlin |
348 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
36 |
dpavlin |
352 |
# { 'eval' => '"v901^a" eq "Podruèje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
37 |
|
|
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
38 |
|
|
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
39 |
|
|
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
40 |
dpavlin |
374 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
41 |
dpavlin |
352 |
], |
42 |
dpavlin |
54 |
); |
43 |
dpavlin |
3 |
|
44 |
dpavlin |
372 |
$log->info("rows: $maxmfn"); |
45 |
dpavlin |
170 |
|
46 |
dpavlin |
363 |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
47 |
|
|
|
48 |
dpavlin |
362 |
while (my $rec = $webpac->fetch_rec) { |
49 |
dpavlin |
358 |
|
50 |
dpavlin |
370 |
my @ds = $webpac->data_structure($rec); |
51 |
dpavlin |
366 |
|
52 |
dpavlin |
372 |
if ($log->is_debug) { |
53 |
|
|
$log->debug("rec = ",Dumper($rec)); |
54 |
|
|
$log->debug("ds = ",Dumper(\@ds)); |
55 |
|
|
} |
56 |
|
|
|
57 |
dpavlin |
374 |
next if (! @ds); |
58 |
dpavlin |
366 |
|
59 |
dpavlin |
374 |
my $filename = $webpac->{'current_filename'}; |
60 |
|
|
|
61 |
|
|
if ($filename) { |
62 |
|
|
open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); |
63 |
|
|
print OUT $webpac->output( |
64 |
|
|
template => 'html.tt', |
65 |
|
|
data => \@ds, |
66 |
dpavlin |
398 |
headline => $webpac->{'headline'}, |
67 |
dpavlin |
374 |
); |
68 |
|
|
close(OUT); |
69 |
|
|
} else { |
70 |
|
|
print $webpac->output( |
71 |
|
|
template => 'text.tt', |
72 |
|
|
data => \@ds, |
73 |
dpavlin |
398 |
headline => $webpac->{'headline'}, |
74 |
dpavlin |
374 |
); |
75 |
|
|
} |
76 |
|
|
|
77 |
dpavlin |
398 |
my $headline = $webpac->{'headline'}; |
78 |
dpavlin |
390 |
|
79 |
dpavlin |
392 |
my $f = $filename; |
80 |
|
|
$f =~ s!out/!!; |
81 |
|
|
|
82 |
dpavlin |
390 |
# save into index |
83 |
|
|
foreach my $ds (@ds) { |
84 |
|
|
next if (! $ds->{'swish'}); |
85 |
|
|
|
86 |
|
|
$index->insert( |
87 |
|
|
index_name => $ds->{'tag'}, |
88 |
dpavlin |
392 |
path => $f, |
89 |
dpavlin |
390 |
headline => $headline, |
90 |
|
|
words => join(" ",@{$ds->{'swish'}}) |
91 |
|
|
); |
92 |
|
|
} |
93 |
|
|
|
94 |
|
|
# print Dumper(\@ds); |
95 |
|
|
|
96 |
dpavlin |
3 |
} |
97 |
|
|
|
98 |
dpavlin |
372 |
if ($log->is_debug) { |
99 |
|
|
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
100 |
|
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
101 |
|
|
} |
102 |
dpavlin |
390 |
|
103 |
|
|
$index->close; |
104 |
|
|
|