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; |
10 |
use OpenIsis; |
use OpenIsis; |
11 |
use Getopt::Std; |
use Text::Iconv; |
12 |
use Data::Dumper; |
use Data::Dumper; |
13 |
use XML::Simple; |
use Carp; |
|
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
|
|
require Unicode::Map8; |
|
|
|
|
|
my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
|
|
|
|
|
my %opts; |
|
|
|
|
|
getopts('d:m:q', \%opts); |
|
|
|
|
|
my $db_dir = $opts{d} || "ps"; # FIX |
|
|
|
|
|
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
|
|
|
|
|
#print Dumper($config->{indexer}); |
|
|
#print "-" x 70,"\n"; |
|
|
|
|
|
# how to convert isis code page to UTF8? |
|
|
my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die; |
|
|
|
|
|
sub isis2xml { |
|
|
|
|
|
my $row = shift @_; |
|
|
|
|
|
my $xml; |
|
|
|
|
|
sub isis_sf { |
|
|
my $row = shift @_; |
|
|
my $isis_id = shift @_; |
|
|
my $subfield = shift @_; |
|
|
if ($row->{$isis_id}->[0]) { |
|
|
my $sf = OpenIsis::subfields($row->{$isis_id}->[0]); |
|
|
if (! defined $subfield || length($subfield) == 0) { |
|
|
# subfield list undef, empty or no defined subfields for this record |
|
|
my $all_sf = $row->{$isis_id}->[0]; |
|
|
$all_sf =~ s/\^./ /g; nuke definirions |
|
|
return $all_sf; |
|
|
} elsif ($sf->{$subfield}) { |
|
|
return $sf->{$subfield}; |
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
|
foreach my $field (keys %{$config->{indexer}}) { |
|
|
|
|
|
my $display_data = ""; |
|
|
my $index_data = ""; |
|
|
|
|
|
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
|
|
|
|
|
my $display_tmp = ""; |
|
|
my $index_tmp = ""; |
|
|
|
|
|
my $format = $x->{content}; |
|
|
my $i = 1; # index only |
|
|
my $d = 1; # display only |
|
|
$i = 0 if (lc($x->{type}) eq "display"); |
|
|
$d = 0 if (lc($x->{type}) eq "index"); |
|
|
#print "## i: $i d: $d ## $format ##"; |
|
|
# parse format |
|
|
my $prefix = ""; |
|
|
if ($format =~ s/^([^\d]+)//) { |
|
|
$prefix = $1; |
|
|
} |
|
|
while ($format) { |
|
|
if ($format =~ s/^(\d\d\d)(\w?)//) { |
|
|
my $isis_tmp = isis_sf($row,$1,$2); |
|
|
if ($isis_tmp) { |
|
|
# $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d); |
|
|
$display_tmp .= $prefix . $isis_tmp if ($d); |
|
|
$index_tmp .= $isis_tmp." " if ($i); |
|
|
#print " $isis_tmp <--\n"; |
|
|
} |
|
|
$prefix = ""; |
|
|
} elsif ($format =~ s/^([^\d]+)//) { |
|
|
$prefix = $1; |
|
|
} else { |
|
|
print STDERR "WARNING: unparsed format '$format'\n"; |
|
|
last; |
|
|
}; |
|
|
} |
|
|
# add suffix |
|
|
$display_tmp .= $prefix if ($display_tmp); |
|
|
|
|
|
# $display_data .= $display_tmp if ($display_tmp ne ""); |
|
|
# $index_data .= $index_tmp if ($index_tmp ne ""); |
|
|
$display_data .= $display_tmp; |
|
|
$index_data .= $index_tmp; |
|
|
|
|
|
} |
|
|
#print "--display:$display_data\n--index:$index_data\n"; |
|
|
$xml->{$field}->{display} .= $isis_map->tou($display_data)->utf8 if ($display_data); |
|
|
$xml->{$field}->{index} .= unac_string($config->{isis_codepage},$index_data) if ($index_data); |
|
|
|
|
|
} |
|
|
if ($xml) { |
|
|
return XMLout($xml, rootname => 'xml', noattr => 1 ); |
|
|
} else { |
|
|
return; |
|
|
} |
|
|
} |
|
14 |
|
|
15 |
########################################################################## |
use lib './lib'; |
16 |
|
use WebPac; |
17 |
|
|
18 |
my $last_tell=0; |
my $webpac = new WebPac( |
19 |
|
code_page => 'ISO-8859-2', |
20 |
|
) || die; |
21 |
|
|
22 |
my @isis_dirs = ( '.' ); # use dirname as database name |
$|=1; |
23 |
|
|
24 |
if ($opts{m}) { |
print "reading database\n"; |
25 |
@isis_dirs = split(/,/,$opts{m}); |
my $maxmfn = $webpac->open_isis( |
26 |
} |
filename => shift @ARGV || '/data/hidra/THS-500/THS', |
27 |
|
lookup => [ |
28 |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
29 |
|
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
30 |
|
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
31 |
|
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
32 |
|
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
33 |
|
], |
34 |
|
); |
35 |
|
|
36 |
my @isis_dbs; |
print "rows: $maxmfn\n\n"; |
37 |
|
|
38 |
foreach (@isis_dirs) { |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
39 |
if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") { |
my $rec = $webpac->{'data'}->{$mfn} || die "no record with mfn $mfn"; |
|
push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI"; |
|
|
} |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/PERI") { |
|
|
push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI"; |
|
|
} |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/AMS") { |
|
|
push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS"; |
|
|
} |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/ARTI") { |
|
|
# push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI"; |
|
|
} |
|
|
} |
|
40 |
|
|
41 |
print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs); |
print "-- ",$webpac->fill_in($rec,'v250^a (v901^a)'),"\n"; |
42 |
|
|
43 |
my $db; |
my @t = $webpac->fill_in($rec,'v553^1;;v553^a'); |
44 |
|
print " Uži pojam: ",join("\t\n",@t),"\n" if (@t); |
45 |
|
|
46 |
foreach my $isis_db (@isis_dbs) { |
@t = $webpac->fill_in($rec,'[a:v251::];;[d:[a:v251::]]'); |
47 |
|
@t = map { $webpac->lookup($_) => $_ } @t; |
48 |
|
print " Područje: ",join("\t\n",@t),"\n" if (@t); |
49 |
|
|
50 |
|
@t = $webpac->fill_in($rec,'[a:v561^4:v251:];;[d:[a:v561^4:v251:]]'); |
51 |
|
print " Mikrotezaurus: ",join("\t\n",@t),"\n" if (@t); |
52 |
|
|
53 |
my $db = OpenIsis::open( $isis_db ); |
@t = $webpac->fill_in($rec,'[a:v561^4:v562^4:v900];;[d:[a:v561^4:v562^4:v900]]'); |
54 |
if (0) { |
print " Deskriptor: ",join("\t\n",@t),"\n" if (@t); |
|
# # FIX |
|
|
# if (! $db ) { |
|
|
print STDERR "WARNING: can't open '$isis_db'\n"; |
|
|
next ; |
|
|
} |
|
|
|
|
|
my $max_rowid = OpenIsis::maxRowid( $db ); |
|
|
|
|
|
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
|
|
|
|
|
my $last_p = 0; |
|
|
|
|
|
# { my $row_id = 1; |
|
|
# FIX |
|
|
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
|
|
my $row = OpenIsis::read( $db, $row_id ); |
|
|
if ($row && $row->{mfn}) { |
|
|
|
|
|
# output current process indicator |
|
|
my $p = int($row->{mfn} * 100 / $max_rowid); |
|
|
if ($p != $last_p) { |
|
|
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q}); |
|
|
$last_p = $p; |
|
|
} |
|
|
|
|
|
if (my $xml = isis2xml($row)) { |
|
|
print "Path-Name: $isis_db#".$row->{mfn}."\n"; |
|
|
print "Content-Length: ".(length($xml)+1)."\n"; |
|
|
print "Document-Type: XML\n\n$xml\n"; |
|
|
} |
|
|
} |
|
|
} |
|
|
print STDERR "\n"; |
|
55 |
} |
} |
56 |
|
|
57 |
|
print "## lookup ",Dumper($webpac->{'lookup'}); |
58 |
1; |
print "## data ",Dumper($webpac->{'data'}); |
|
__END__ |
|
|
########################################################################## |
|
|
|
|
|
=head1 NAME |
|
|
|
|
|
isis2xml.pl - read isis file and dump XML |
|
|
|
|
|
=head1 DESCRIPTION |
|
|
|
|
|
This command will read ISIS data file using OpenIsis perl module and |
|
|
create XML file for usage with I<SWISH-E> |
|
|
indexer. Dispite it's name, this script B<isn't general xml generator> |
|
|
from isis files (isis allready has something like that). Output of this |
|
|
script is tailor-made for SWISH-E. |
|
|
|
|
|
=head1 AUTHOR |
|
|
|
|
|
Dobrica Pavlinusic <dpavlin@rot13.org> |
|
|
|
|
|
=head1 COPYRIGHT |
|
|
|
|
|
GNU Public License (GPL) v2 or later |
|
|
|
|
|
=head1 SEE ALSO |
|
|
|
|
|
SWISH-E web site at http://www.swish-e.org |
|
|
|
|
|
=cut |
|