1 |
#!/usr/bin/perl -w |
#!/usr/bin/perl -w |
2 |
|
|
3 |
|
=head1 NAME |
4 |
|
|
5 |
|
all2all.pl - basic script for all WebPAC needs |
6 |
|
|
7 |
|
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
|
use OpenIsis; |
|
|
use Getopt::Std; |
|
10 |
use Data::Dumper; |
use Data::Dumper; |
11 |
use XML::Simple; |
use Carp; |
12 |
|
|
13 |
my $config=XMLin(); |
use lib './lib'; |
14 |
|
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
use WebPAC::Index; |
17 |
|
|
18 |
print Dumper($config); |
my $webpac = new WebPAC( |
19 |
|
code_page => 'ISO-8859-2', |
20 |
|
limit_mfn => 100, |
21 |
|
# debug => 1, |
22 |
|
) || die; |
23 |
|
|
24 |
my %opts; |
my $log = $webpac->_get_logger(); |
25 |
|
|
26 |
getopts('d:m:q', \%opts); |
my $index = new WebPAC::jsFind( |
27 |
|
index_path => './out/index', |
28 |
|
keys => 80, |
29 |
|
) || die; |
30 |
|
|
31 |
my $db_dir = $opts{d}; |
my $thes; |
32 |
|
|
33 |
die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
$|=1; |
34 |
|
|
35 |
#-------------------------------------------------------------------- |
my $maxmfn = $webpac->open_isis( |
36 |
|
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 |
my $last_tell=0; |
$log->info("rows: $maxmfn"); |
48 |
|
|
49 |
my @isis_dirs = ( '.' ); # use dirname as database name |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
50 |
|
|
51 |
if ($opts{m}) { |
while (my $rec = $webpac->fetch_rec) { |
|
@isis_dirs = split(/,/,$opts{m}); |
|
|
} |
|
52 |
|
|
53 |
my @isis_dbs; |
my @ds = $webpac->data_structure($rec); |
54 |
|
|
55 |
foreach (@isis_dirs) { |
if ($log->is_debug) { |
56 |
if (-e "$common::isis_data/$db_dir/$_/LIBRI") { |
$log->debug("rec = ",Dumper($rec)); |
57 |
push @isis_dbs,"$common::isis_data/$db_dir/$_/LIBRI/LIBRI"; |
$log->debug("ds = ",Dumper(\@ds)); |
58 |
} |
} |
59 |
if (-e "$common::isis_data/$db_dir/$_/PERI") { |
|
60 |
push @isis_dbs,"$common::isis_data/$db_dir/$_/PERI/PERI"; |
next if (! @ds); |
61 |
} |
|
62 |
if (-e "$common::isis_data/$db_dir/$_/AMS") { |
my $filename = $webpac->{'current_filename'}; |
63 |
push @isis_dbs,"$common::isis_data/$db_dir/$_/AMS/AMS"; |
|
64 |
} |
if ($filename) { |
65 |
if (-e "$common::isis_data/$db_dir/$_/ARTI") { |
open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); |
66 |
# push @isis_dbs,"$common::isis_data/$db_dir/$_/ARTI/ARTI"; |
print OUT $webpac->output( |
67 |
|
template => 'html.tt', |
68 |
|
data => \@ds, |
69 |
|
headline => $webpac->{'headline'}, |
70 |
|
); |
71 |
|
close(OUT); |
72 |
|
} else { |
73 |
|
print $webpac->output( |
74 |
|
template => 'text.tt', |
75 |
|
data => \@ds, |
76 |
|
headline => $webpac->{'headline'}, |
77 |
|
); |
78 |
} |
} |
|
} |
|
79 |
|
|
80 |
foreach my $isis_db (@isis_dbs) { |
my $headline = $webpac->{'headline'}; |
81 |
|
|
82 |
my $db = OpenIsis::open( "$isis_db" ) || warn "can't open '$isis_db'"; |
my $f = $filename; |
83 |
|
$f =~ s!out/!!; |
84 |
|
|
85 |
|
# save into index |
86 |
|
foreach my $ds (@ds) { |
87 |
|
next if (! $ds->{'swish'}); |
88 |
|
|
89 |
|
$index->insert( |
90 |
|
index_name => $ds->{'tag'}, |
91 |
|
path => $f, |
92 |
|
headline => $headline, |
93 |
|
words => join(" ",@{$ds->{'swish'}}) |
94 |
|
); |
95 |
|
} |
96 |
|
|
97 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
# save into sorted index (thesaurus) |
98 |
|
foreach my $ds (@ds) { |
99 |
|
next if (! $ds->{'index'}); |
100 |
|
|
101 |
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
102 |
|
|
103 |
|
$thes->{$ds->{'tag'}}->insert( |
104 |
|
path => $f, |
105 |
|
headline => $headline, |
106 |
|
); |
107 |
|
} |
108 |
|
|
109 |
my $last_pcnt = 0; |
# print Dumper(\@ds); |
110 |
|
|
111 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
} |
|
my $row = OpenIsis::read( $db, $row_id ); |
|
112 |
|
|
113 |
# output current process indicator |
if ($log->is_debug) { |
114 |
my $pcnt = int($row->{mfn} * 100 / $max_rowid); |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
115 |
if ($pcnt != $last_pcnt) { |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
116 |
printf STDERR ("%5d / %5d -- %-2d %%\n",$row->{mfn},$max_rowid,$pcnt) if (! $opts{q}); |
foreach my $t (keys %{$thes}) { |
117 |
$last_pcnt = $pcnt; |
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
|
} |
|
118 |
} |
} |
119 |
} |
} |
120 |
|
|
121 |
|
$index->close; |
122 |
|
|