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; |
|
4 |
|
|
5 |
my $config=XMLin(); |
webpac.pl - basic script for all WebPac needs |
6 |
|
|
7 |
print Dumper($config); |
=cut |
8 |
|
|
9 |
my %opts; |
use strict; |
10 |
|
use OpenIsis; |
11 |
getopts('d:m:q', \%opts); |
use Text::Iconv; |
12 |
|
use Data::Dumper; |
13 |
|
|
14 |
my $db_dir = $opts{d}; |
my $cp = Text::Iconv->new("852", "iso-8859-2"); |
15 |
|
|
16 |
die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
$|=1; |
17 |
|
|
18 |
#-------------------------------------------------------------------- |
my $db = OpenIsis::open( shift @ARGV || '/data/hidra/THS-500/THS' ); |
19 |
|
my $maxmfn = OpenIsis::maxRowid( $db ) || 1; |
20 |
|
|
21 |
my $last_tell=0; |
print "rows: $maxmfn\n\n"; |
22 |
|
|
23 |
my @isis_dirs = ( '.' ); # use dirname as database name |
my $data; |
24 |
|
|
25 |
|
my @lookup = ( |
26 |
|
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
27 |
|
{ 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
28 |
|
{ 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'ma:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
29 |
|
{ 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'da:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
30 |
|
); |
31 |
|
|
32 |
|
sub fill_in { |
33 |
|
my $tmp = shift || return; |
34 |
|
my $data = shift || return; |
35 |
|
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 $$data->{$f}->[$i]->{$sf}; |
44 |
|
} else { |
45 |
|
return ''; |
46 |
|
} |
47 |
|
} |
48 |
|
|
49 |
if ($opts{m}) { |
sub get_nosf { |
50 |
@isis_dirs = split(/,/,$opts{m}); |
my ($found,$data,$f,$i) = @_; |
51 |
} |
if ($$data->{$f}->[$i]) { |
52 |
|
$$found++; |
53 |
|
return $$data->{$f}->[$i]; |
54 |
|
} else { |
55 |
|
return ''; |
56 |
|
} |
57 |
|
} |
58 |
|
|
59 |
my @isis_dbs; |
$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; |
61 |
|
|
62 |
foreach (@isis_dirs) { |
if ($found) { |
63 |
if (-e "$common::isis_data/$db_dir/$_/LIBRI") { |
return $tmp; |
64 |
push @isis_dbs,"$common::isis_data/$db_dir/$_/LIBRI/LIBRI"; |
} else { |
65 |
} |
return; |
|
if (-e "$common::isis_data/$db_dir/$_/PERI") { |
|
|
push @isis_dbs,"$common::isis_data/$db_dir/$_/PERI/PERI"; |
|
|
} |
|
|
if (-e "$common::isis_data/$db_dir/$_/AMS") { |
|
|
push @isis_dbs,"$common::isis_data/$db_dir/$_/AMS/AMS"; |
|
|
} |
|
|
if (-e "$common::isis_data/$db_dir/$_/ARTI") { |
|
|
# push @isis_dbs,"$common::isis_data/$db_dir/$_/ARTI/ARTI"; |
|
66 |
} |
} |
67 |
} |
} |
68 |
|
|
69 |
foreach my $isis_db (@isis_dbs) { |
print "reading database...\n"; |
70 |
|
|
71 |
my $db = OpenIsis::open( "$isis_db" ) || warn "can't open '$isis_db'"; |
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
72 |
|
|
73 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
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 |
my $last_pcnt = 0; |
push @{$data->{$mfn}->{$k}}, $val if ($val && $val ne ""); |
92 |
|
} |
93 |
|
} |
94 |
|
|
95 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
} |
|
my $row = OpenIsis::read( $db, $row_id ); |
|
96 |
|
|
97 |
# output current process indicator |
foreach my $i (@lookup) { |
98 |
my $pcnt = int($row->{mfn} * 100 / $max_rowid); |
if ($i->{'eval'}) { |
99 |
if ($pcnt != $last_pcnt) { |
my $eval = fill_in($i->{'eval'},$data->{$mfn}); |
100 |
printf STDERR ("%5d / %5d -- %-2d %%\n",$row->{mfn},$max_rowid,$pcnt) if (! $opts{q}); |
my $key = fill_in($i->{'key'}, $data->{$mfn}); |
101 |
$last_pcnt = $pcnt; |
my $val = fill_in($i->{'val'}, $data->{$mfn}); |
102 |
|
if ($key && $val && eval $eval) { |
103 |
|
push @{$data->{lookup}->{$key}}, $val; |
104 |
|
} |
105 |
|
} else { |
106 |
|
my $key = fill_in($i->{'key'}, $data->{$mfn}); |
107 |
|
my $val = fill_in($i->{'val'}, $data->{$mfn}); |
108 |
|
if ($key && $val) { |
109 |
|
push @{$data->{lookup}->{$key}}, $val; |
110 |
|
} |
111 |
} |
} |
112 |
} |
} |
113 |
|
|
114 |
} |
} |
115 |
|
print "\n"; |
116 |
|
|
117 |
|
for (my $mfn = 1; $mfn <= $maxmfn; $mfn++) { |
118 |
|
my $d = $data->{$mfn}; |
119 |
|
print $d->{'250'}->[0]->{'a'},"\n"; |
120 |
|
my $t = fill_in('v553^1;;v553^a', $d); |
121 |
|
print " Uži pojam: $t\n" if ($t); |
122 |
|
$t = fill_in('[a:v251];;[d:[a:v251]]', $d); |
123 |
|
print " Uži pojam: $t\n" if ($t); |
124 |
|
} |
125 |
|
|
126 |
|
print Dumper($data); |