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; |
|
use Text::Unaccent; |
|
|
require Unicode::Map8; |
|
12 |
|
|
13 |
my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
use lib './lib'; |
14 |
|
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
|
17 |
my %opts; |
my $webpac = new WebPAC( |
18 |
|
code_page => 'ISO-8859-2', |
19 |
|
limit_mfn => 100, |
20 |
|
# debug => 1, |
21 |
|
) || die; |
22 |
|
|
23 |
getopts('d:m:q', \%opts); |
my $log = $webpac->_get_logger(); |
24 |
|
|
25 |
my $db_dir = $opts{d} || "ps"; # FIX |
my $index = new WebPAC::jsFind( |
26 |
|
index_path => './out/index', |
27 |
|
) || die; |
28 |
|
|
29 |
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
$|=1; |
30 |
|
|
31 |
#print Dumper($config->{indexer}); |
my $maxmfn = $webpac->open_isis( |
32 |
#print "-" x 70,"\n"; |
filename => shift @ARGV || '/data/hidra/THS/THS', |
33 |
|
lookup => [ |
34 |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
35 |
|
# { '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' }, |
37 |
|
# { '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 |
# how to convert isis code page to UTF8? |
$log->info("rows: $maxmfn"); |
|
my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die; |
|
44 |
|
|
45 |
sub isis2xml { |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
46 |
|
|
47 |
my $row = shift @_; |
while (my $rec = $webpac->fetch_rec) { |
48 |
|
|
49 |
my $xml; |
my @ds = $webpac->data_structure($rec); |
50 |
|
|
51 |
sub isis_sf { |
if ($log->is_debug) { |
52 |
my $row = shift @_; |
$log->debug("rec = ",Dumper($rec)); |
53 |
my $isis_id = shift @_; |
$log->debug("ds = ",Dumper(\@ds)); |
|
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 |
|
|
return $row->{$isis_id}->[0]; |
|
|
} elsif ($sf->{$subfield}) { |
|
|
return $sf->{$subfield}; |
|
|
} |
|
|
} |
|
54 |
} |
} |
55 |
|
|
56 |
foreach my $field (keys %{$config->{indexer}}) { |
next if (! @ds); |
57 |
|
|
58 |
my $display_data = ""; |
my $filename = $webpac->{'current_filename'}; |
|
my $index_data = ""; |
|
59 |
|
|
60 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
if ($filename) { |
61 |
|
open(OUT,"> $filename") || $log->logdie("can't open output '$filename': $!"); |
62 |
my $display_tmp = ""; |
print OUT $webpac->output( |
63 |
my $index_tmp = ""; |
template => 'html.tt', |
64 |
|
data => \@ds, |
65 |
my $format = $x->{content}; |
); |
66 |
my $i = 1; # index only |
close(OUT); |
|
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 ); |
|
67 |
} else { |
} else { |
68 |
return; |
print $webpac->output( |
69 |
|
template => 'text.tt', |
70 |
|
data => \@ds, |
71 |
|
); |
72 |
} |
} |
|
} |
|
73 |
|
|
74 |
########################################################################## |
my $headline; |
75 |
|
foreach my $ds (@ds) { |
76 |
my $last_tell=0; |
if ($ds->{'tag'} eq 'headline') { |
77 |
|
$headline = join(" ",@{$ds->{'display'}}); |
78 |
my @isis_dirs = ( '.' ); # use dirname as database name |
last; |
79 |
|
} |
|
if ($opts{m}) { |
|
|
@isis_dirs = split(/,/,$opts{m}); |
|
|
} |
|
|
|
|
|
my @isis_dbs; |
|
|
|
|
|
foreach (@isis_dirs) { |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") { |
|
|
push @isis_dbs,$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"; |
|
80 |
} |
} |
|
} |
|
81 |
|
|
82 |
print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs); |
# save into index |
83 |
|
foreach my $ds (@ds) { |
84 |
my $db; |
next if (! $ds->{'swish'}); |
85 |
|
|
86 |
foreach my $isis_db (@isis_dbs) { |
$index->insert( |
87 |
|
index_name => $ds->{'tag'}, |
88 |
|
path => $filename, |
89 |
my $db = OpenIsis::open( $isis_db ); |
headline => $headline, |
90 |
if (0) { |
words => join(" ",@{$ds->{'swish'}}) |
91 |
# # FIX |
); |
|
# if (! $db ) { |
|
|
print STDERR "WARNING: can't open '$isis_db'\n"; |
|
|
next ; |
|
92 |
} |
} |
93 |
|
|
94 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
# print Dumper(\@ds); |
|
|
|
|
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
|
|
|
|
|
my $last_p = 0; |
|
95 |
|
|
|
# { 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"; |
|
96 |
} |
} |
97 |
|
|
98 |
|
if ($log->is_debug) { |
99 |
|
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
100 |
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
101 |
|
} |
102 |
|
|
103 |
1; |
$index->close; |
|
__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 |
|
104 |
|
|
|
=cut |
|