1 |
#!/usr/bin/perl -w |
#!/usr/bin/perl -w |
2 |
|
|
3 |
use strict; |
=head1 NAME |
|
use OpenIsis; |
|
|
use Getopt::Std; |
|
|
use Data::Dumper; |
|
|
use XML::Simple; |
|
|
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; |
|
|
|
|
|
# usage: |
|
|
# -d directory name |
|
|
# -m multiple directories |
|
|
# -q quiet |
|
|
# -s run swish |
|
|
|
|
|
getopts('d:m:qs', \%opts); |
|
4 |
|
|
5 |
my $db_dir = $opts{d} || "ps"; # FIX |
webpac.pl - basic script for all WebPac needs |
6 |
|
|
7 |
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
=cut |
8 |
|
|
9 |
#print Dumper($config->{indexer}); |
use strict; |
10 |
#print "-" x 70,"\n"; |
use OpenIsis; |
11 |
|
use Text::Iconv; |
12 |
|
use Data::Dumper; |
13 |
|
|
14 |
# how to convert isis code page to UTF8? |
my $cp = Text::Iconv->new("852", "iso-8859-2"); |
|
my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die; |
|
15 |
|
|
16 |
sub isis2xml { |
$|=1; |
17 |
|
|
18 |
my $row = shift @_; |
my $db = OpenIsis::open( shift @ARGV || '/data/hidra/THS-500/THS' ); |
19 |
|
my $maxmfn = OpenIsis::maxRowid( $db ) || 1; |
20 |
|
|
21 |
my $xml; |
print "rows: $maxmfn\n\n"; |
|
$xml->{db_dir} = [ $db_dir ]; # FIX remove? |
|
22 |
|
|
23 |
sub isis_sf { |
my $data; |
24 |
my $row = shift @_; |
|
25 |
my $isis_id = shift @_; |
my @lookup = ( |
26 |
my $subfield = shift @_; |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
27 |
if ($row->{$isis_id}->[0]) { |
{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
28 |
my $sf = OpenIsis::subfields($row->{$isis_id}->[0]); |
{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'ma:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
29 |
if (! defined $subfield || length($subfield) == 0) { |
{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'da:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
30 |
# subfield list undef, empty or no defined subfields for this record |
); |
31 |
my $all_sf = $row->{$isis_id}->[0]; |
|
32 |
$all_sf =~ s/\^./ /g; nuke definirions |
sub fill_in { |
33 |
return $all_sf; |
my $tmp = shift || return; |
34 |
} elsif ($sf->{$subfield}) { |
my $data = shift || return; |
35 |
return $sf->{$subfield}; |
my $i = shift || 0; |
36 |
} |
|
37 |
|
my $found = 0; |
38 |
|
|
39 |
|
sub get_sf { |
40 |
|
my ($found,$data,$f,$sf,$i) = @_; |
41 |
|
if ($$data->{$f}->[$i]->{$sf}) { |
42 |
|
$$found++; |
43 |
|
return $$data->{$f}->[$i]->{$sf}; |
44 |
|
} else { |
45 |
|
return ''; |
46 |
} |
} |
47 |
} |
} |
48 |
|
|
49 |
foreach my $field (keys %{$config->{indexer}}) { |
sub get_nosf { |
50 |
|
my ($found,$data,$f,$i) = @_; |
51 |
my $display_data = ""; |
if ($$data->{$f}->[$i]) { |
52 |
my $index_data = ""; |
$$found++; |
53 |
|
return $$data->{$f}->[$i]; |
54 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
} else { |
55 |
|
return ''; |
|
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 .= $index_tmp; |
|
|
|
|
56 |
} |
} |
|
#print "--display:$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); |
|
|
$xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data); |
|
|
$xml->{$field."_index"} = [ unac_string($config->{isis_codepage},$index_data)." jabuka" ] if ($index_data); |
|
|
|
|
57 |
} |
} |
|
if ($xml) { |
|
|
return XMLout($xml, rootname => 'xml', keeproot => 0, noattr => 0 ); |
|
|
} else { |
|
|
return; |
|
|
} |
|
|
} |
|
|
|
|
|
########################################################################## |
|
58 |
|
|
59 |
my $last_tell=0; |
$tmp =~ s/v(\d+)\^(\w)/get_sf(\$found,\$data,$1,$2,$i)/ges; |
60 |
|
$tmp =~ s/v(\d+)/get_nosf(\$found,\$data,$1,$i)/ges; |
|
my @isis_dirs = ( '.' ); # use dirname as database name |
|
|
|
|
|
if ($opts{m}) { |
|
|
@isis_dirs = split(/,/,$opts{m}); |
|
|
} |
|
61 |
|
|
62 |
my @isis_dbs; |
if ($found) { |
63 |
|
return $tmp; |
64 |
foreach (@isis_dirs) { |
} else { |
65 |
if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") { |
return; |
|
push @isis_dbs,$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"; |
|
66 |
} |
} |
67 |
} |
} |
68 |
|
|
69 |
print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs); |
print "reading database...\n"; |
70 |
|
|
71 |
my $db; |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
72 |
|
|
73 |
foreach my $isis_db (@isis_dbs) { |
my $row = OpenIsis::read( $db, $mfn ); |
74 |
|
print Dumper($row); |
75 |
|
foreach my $k (keys %{$row}) { |
76 |
|
if ($k ne "mfn") { |
77 |
|
foreach my $l (@{$row->{$k}}) { |
78 |
|
next if (! $l); |
79 |
|
$l = $cp->convert($l); |
80 |
|
# has subfields? |
81 |
|
my $val; |
82 |
|
if ($l =~ m/\^/) { |
83 |
|
foreach my $t (split(/\^/,$l)) { |
84 |
|
next if (! $t); |
85 |
|
$val->{substr($t,0,1)} = substr($t,1); |
86 |
|
} |
87 |
|
} else { |
88 |
|
$val = $l; |
89 |
|
} |
90 |
|
|
91 |
|
push @{$data->{$mfn}->{$k}}, $val if ($val && $val ne ""); |
92 |
|
} |
93 |
|
} |
94 |
|
|
|
my $db = OpenIsis::open( $isis_db ); |
|
|
if (0) { |
|
|
# # FIX |
|
|
# if (! $db ) { |
|
|
print STDERR "WARNING: can't open '$isis_db'\n"; |
|
|
next ; |
|
95 |
} |
} |
96 |
|
|
97 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
foreach my $i (@lookup) { |
98 |
|
if ($i->{'eval'}) { |
99 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
my $eval = fill_in($i->{'eval'},$data->{$mfn}); |
100 |
|
my $key = fill_in($i->{'key'}, $data->{$mfn}); |
101 |
my $last_p = 0; |
my $val = fill_in($i->{'val'}, $data->{$mfn}); |
102 |
|
if ($key && $val && eval $eval) { |
103 |
# { my $row_id = 1; |
push @{$data->{lookup}->{$key}}, $val; |
|
# 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; |
|
104 |
} |
} |
105 |
|
} else { |
106 |
if (my $xml = isis2xml($row)) { |
my $key = fill_in($i->{'key'}, $data->{$mfn}); |
107 |
my $path = $isis_db; |
my $val = fill_in($i->{'val'}, $data->{$mfn}); |
108 |
$path =~ s#$config->{isis_data}/*##g; |
if ($key && $val) { |
109 |
my $out = "Path-Name: $path#".$row->{mfn}."\n"; |
push @{$data->{lookup}->{$key}}, $val; |
|
$out .= "Content-Length: ".(length($xml)+1)."\n"; |
|
|
$out .= "Document-Type: XML\n\n$xml\n"; |
|
|
print $out; |
|
110 |
} |
} |
111 |
} |
} |
112 |
} |
} |
|
print STDERR "\n"; |
|
|
} |
|
113 |
|
|
114 |
|
} |
115 |
|
print "\n"; |
116 |
|
|
117 |
1; |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
118 |
__END__ |
my $d = $data->{$mfn}; |
119 |
########################################################################## |
print $d->{'250'}->[0]->{'a'},"\n"; |
120 |
|
my $t = fill_in('v553^1;;v553^a', $d); |
121 |
=head1 NAME |
print " Uži pojam: $t\n" if ($t); |
122 |
|
$t = fill_in('[a:v251];;[d:[a:v251]]', $d); |
123 |
isis2xml.pl - read isis file and dump XML |
print " Uži pojam: $t\n" if ($t); |
124 |
|
} |
|
=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 |
|
125 |
|
|
126 |
=cut |
print Dumper($data); |