| 1 |
package WebPAC::Input::ISIS; |
| 2 |
|
| 3 |
use warnings; |
| 4 |
use strict; |
| 5 |
|
| 6 |
use WebPAC::Common; |
| 7 |
use base qw/WebPAC::Input WebPAC::Common/; |
| 8 |
use Text::Iconv; |
| 9 |
|
| 10 |
=head1 NAME |
| 11 |
|
| 12 |
WebPAC::Input::ISIS - support for CDS/ISIS source files |
| 13 |
|
| 14 |
=head1 VERSION |
| 15 |
|
| 16 |
Version 0.01 |
| 17 |
|
| 18 |
=cut |
| 19 |
|
| 20 |
our $VERSION = '0.01'; |
| 21 |
|
| 22 |
|
| 23 |
# auto-configure |
| 24 |
|
| 25 |
my ($have_biblio_isis, $have_openisis) = (0,0); |
| 26 |
|
| 27 |
eval "use Biblio::Isis 0.13;"; |
| 28 |
unless ($@) { |
| 29 |
$have_biblio_isis = 1 |
| 30 |
} else { |
| 31 |
eval "use OpenIsis;"; |
| 32 |
$have_openisis = 1 unless ($@); |
| 33 |
} |
| 34 |
|
| 35 |
=head1 SYNOPSIS |
| 36 |
|
| 37 |
Open CDS/ISIS, WinISIS or IsisMarc database using Biblio::Isis or OpenIsis |
| 38 |
module and read all records to memory. |
| 39 |
|
| 40 |
my $isis = new WebPAC::Input::ISIS(); |
| 41 |
$isis->open( filename => '/path/to/ISIS/ISIS' ); |
| 42 |
|
| 43 |
=head1 FUNCTIONS |
| 44 |
|
| 45 |
=head2 open |
| 46 |
|
| 47 |
This function will read whole database in memory and produce lookups. |
| 48 |
|
| 49 |
$isis->open( |
| 50 |
filename => '/data/ISIS/ISIS', |
| 51 |
code_page => '852', |
| 52 |
limit_mfn => 500, |
| 53 |
start_mfn => 6000, |
| 54 |
lookup => $lookup_obj, |
| 55 |
); |
| 56 |
|
| 57 |
By default, ISIS code page is assumed to be C<852>. |
| 58 |
|
| 59 |
If optional parametar C<start_mfn> is set, this will be first MFN to read |
| 60 |
from database (so you can skip beginning of your database if you need to). |
| 61 |
|
| 62 |
If optional parametar C<limit_mfn> is set, it will read just 500 records |
| 63 |
from database in example above. |
| 64 |
|
| 65 |
Returns size of database, regardless of C<start_mfn> and C<limit_mfn> |
| 66 |
parametars, see also C<$isis->size>. |
| 67 |
|
| 68 |
=cut |
| 69 |
|
| 70 |
sub open { |
| 71 |
my $self = shift; |
| 72 |
my $arg = {@_}; |
| 73 |
|
| 74 |
my $log = $self->_get_logger(); |
| 75 |
|
| 76 |
$log->logcroak("need filename") if (! $arg->{'filename'}); |
| 77 |
my $code_page = $arg->{'code_page'} || '852'; |
| 78 |
|
| 79 |
$log->logdie("can't find database ",$arg->{'filename'}) unless (glob($arg->{'filename'}.'.*')); |
| 80 |
|
| 81 |
# store data in object |
| 82 |
$self->{'isis_code_page'} = $code_page; |
| 83 |
foreach my $v (qw/isis_filename start_mfn limit_mfn/) { |
| 84 |
$self->{$v} = $arg->{$v} if ($arg->{$v}); |
| 85 |
} |
| 86 |
|
| 87 |
# create Text::Iconv object |
| 88 |
my $cp = Text::Iconv->new($code_page,$self->{'code_page'}); |
| 89 |
|
| 90 |
$log->info("reading ISIS database '",$arg->{'filename'},"'"); |
| 91 |
$log->debug("isis code page: $code_page"); |
| 92 |
|
| 93 |
my ($isis_db,$db_size); |
| 94 |
|
| 95 |
if ($have_openisis) { |
| 96 |
$log->debug("using OpenIsis perl bindings"); |
| 97 |
$isis_db = OpenIsis::open($arg->{'filename'}); |
| 98 |
$db_size = OpenIsis::maxRowid( $isis_db ) || 1; |
| 99 |
} elsif ($have_biblio_isis) { |
| 100 |
$log->debug("using Biblio::Isis"); |
| 101 |
use Biblio::Isis; |
| 102 |
$isis_db = new Biblio::Isis( |
| 103 |
isisdb => $arg->{'filename'}, |
| 104 |
include_deleted => 1, |
| 105 |
hash_filter => sub { |
| 106 |
my $l = shift || return; |
| 107 |
$l = $cp->convert($l); |
| 108 |
return $l; |
| 109 |
}, |
| 110 |
); |
| 111 |
$db_size = $isis_db->count; |
| 112 |
|
| 113 |
unless ($db_size) { |
| 114 |
$log->logwarn("no records in database ", $arg->{'filename'}, ", skipping..."); |
| 115 |
return; |
| 116 |
} |
| 117 |
|
| 118 |
} else { |
| 119 |
$log->logdie("Can't find supported ISIS library for perl. I suggent that you install Bilbio::Isis from CPAN."); |
| 120 |
} |
| 121 |
|
| 122 |
|
| 123 |
my $startmfn = 1; |
| 124 |
my $maxmfn = $db_size; |
| 125 |
|
| 126 |
if (my $s = $self->{'start_mfn'}) { |
| 127 |
$log->info("skipping to MFN $s"); |
| 128 |
$startmfn = $s; |
| 129 |
} else { |
| 130 |
$self->{'start_mfn'} = $startmfn; |
| 131 |
} |
| 132 |
|
| 133 |
if ($self->{limit_mfn}) { |
| 134 |
$log->info("limiting to ",$self->{limit_mfn}," records"); |
| 135 |
$maxmfn = $startmfn + $self->{limit_mfn} - 1; |
| 136 |
$maxmfn = $db_size if ($maxmfn > $db_size); |
| 137 |
} |
| 138 |
|
| 139 |
# store size for later |
| 140 |
$self->{'size'} = ($maxmfn - $startmfn) ? ($maxmfn - $startmfn + 1) : 0; |
| 141 |
|
| 142 |
$log->info("processing ",($maxmfn-$startmfn)." records using ",( $have_openisis ? 'OpenIsis' : 'Biblio::Isis')); |
| 143 |
|
| 144 |
|
| 145 |
# read database |
| 146 |
for (my $mfn = $startmfn; $mfn <= $maxmfn; $mfn++) { |
| 147 |
|
| 148 |
$log->debug("mfn: $mfn\n"); |
| 149 |
|
| 150 |
my $rec; |
| 151 |
|
| 152 |
if ($have_openisis) { |
| 153 |
|
| 154 |
# read record using OpenIsis |
| 155 |
my $row = OpenIsis::read( $isis_db, $mfn ); |
| 156 |
foreach my $k (keys %{$row}) { |
| 157 |
if ($k ne "mfn") { |
| 158 |
foreach my $l (@{$row->{$k}}) { |
| 159 |
$l = $cp->convert($l); |
| 160 |
# has subfields? |
| 161 |
my $val; |
| 162 |
if ($l =~ m/\^/) { |
| 163 |
foreach my $t (split(/\^/,$l)) { |
| 164 |
next if (! $t); |
| 165 |
$val->{substr($t,0,1)} = substr($t,1); |
| 166 |
} |
| 167 |
} else { |
| 168 |
$val = $l; |
| 169 |
} |
| 170 |
|
| 171 |
push @{$rec->{$k}}, $val; |
| 172 |
} |
| 173 |
} else { |
| 174 |
push @{$rec->{'000'}}, $mfn; |
| 175 |
} |
| 176 |
} |
| 177 |
|
| 178 |
} elsif ($have_biblio_isis) { |
| 179 |
$rec = $isis_db->to_hash($mfn); |
| 180 |
} else { |
| 181 |
$log->logdie("hum? implementation missing?"); |
| 182 |
} |
| 183 |
|
| 184 |
$log->confess("record $mfn empty?") unless ($rec); |
| 185 |
|
| 186 |
# store |
| 187 |
if ($self->{'low_mem'}) { |
| 188 |
$self->{'db'}->put($mfn, $rec); |
| 189 |
} else { |
| 190 |
$self->{'data'}->{$mfn} = $rec; |
| 191 |
} |
| 192 |
|
| 193 |
# create lookup |
| 194 |
$self->{'lookup'}->add( $rec ) if ($self->{'lookup'} && can($self->{'lookup'}->add)); |
| 195 |
|
| 196 |
$self->progress_bar($mfn,$maxmfn); |
| 197 |
|
| 198 |
} |
| 199 |
|
| 200 |
$self->{'current_mfn'} = -1; |
| 201 |
$self->{'last_pcnt'} = 0; |
| 202 |
|
| 203 |
$log->debug("max mfn: $maxmfn"); |
| 204 |
|
| 205 |
# store max mfn and return it. |
| 206 |
$self->{'max_mfn'} = $maxmfn; |
| 207 |
|
| 208 |
return $db_size; |
| 209 |
} |
| 210 |
|
| 211 |
=head2 fetch |
| 212 |
|
| 213 |
Fetch next record from database. It will also displays progress bar. |
| 214 |
|
| 215 |
my $rec = $isis->fetch; |
| 216 |
|
| 217 |
Record from this function should probably go to C<data_structure> for |
| 218 |
normalisation. |
| 219 |
|
| 220 |
=cut |
| 221 |
|
| 222 |
sub fetch { |
| 223 |
my $self = shift; |
| 224 |
|
| 225 |
my $log = $self->_get_logger(); |
| 226 |
|
| 227 |
$log->logconfess("it seems that you didn't load database!") unless ($self->{'current_mfn'}); |
| 228 |
|
| 229 |
if ($self->{'current_mfn'} == -1) { |
| 230 |
$self->{'current_mfn'} = $self->{'start_mfn'}; |
| 231 |
} else { |
| 232 |
$self->{'current_mfn'}++; |
| 233 |
} |
| 234 |
|
| 235 |
my $mfn = $self->{'current_mfn'}; |
| 236 |
|
| 237 |
if ($mfn > $self->{'max_mfn'}) { |
| 238 |
$self->{'current_mfn'} = $self->{'max_mfn'}; |
| 239 |
$log->debug("at EOF"); |
| 240 |
return; |
| 241 |
} |
| 242 |
|
| 243 |
$self->progress_bar($mfn,$self->{'max_mfn'}); |
| 244 |
|
| 245 |
if ($self->{'low_mem'}) { |
| 246 |
return $self->{'db'}->get($mfn); |
| 247 |
} else { |
| 248 |
return $self->{'data'}->{$mfn}; |
| 249 |
} |
| 250 |
} |
| 251 |
|
| 252 |
=head2 pos |
| 253 |
|
| 254 |
Returns current record number (MFN). |
| 255 |
|
| 256 |
print $isis->pos; |
| 257 |
|
| 258 |
First record in database has position 1. |
| 259 |
|
| 260 |
=cut |
| 261 |
|
| 262 |
sub pos { |
| 263 |
my $self = shift; |
| 264 |
return $self->{'current_mfn'}; |
| 265 |
} |
| 266 |
|
| 267 |
|
| 268 |
=head2 size |
| 269 |
|
| 270 |
Returns number of records in database |
| 271 |
|
| 272 |
print $isis->size; |
| 273 |
|
| 274 |
Result from this function can be used to loop through all records |
| 275 |
|
| 276 |
foreach my $mfn ( 1 ... $isis->size ) { ... } |
| 277 |
|
| 278 |
because it takes into account C<start_mfn> and C<limit_mfn>. |
| 279 |
|
| 280 |
=cut |
| 281 |
|
| 282 |
sub size { |
| 283 |
my $self = shift; |
| 284 |
return $self->{'size'}; |
| 285 |
} |
| 286 |
|
| 287 |
=head2 seek |
| 288 |
|
| 289 |
Seek to specified MFN in file. |
| 290 |
|
| 291 |
$isis->seek(42); |
| 292 |
|
| 293 |
First record in database has position 1. |
| 294 |
|
| 295 |
=cut |
| 296 |
|
| 297 |
sub seek { |
| 298 |
my $self = shift; |
| 299 |
my $pos = shift || return; |
| 300 |
|
| 301 |
my $log = $self->_get_logger(); |
| 302 |
|
| 303 |
if ($pos < 1) { |
| 304 |
$log->warn("seek before first record"); |
| 305 |
$pos = 1; |
| 306 |
} elsif ($pos > $self->{'max_mfn'}) { |
| 307 |
$log->warn("seek beyond last record"); |
| 308 |
$pos = $self->{'max_mfn'}; |
| 309 |
} |
| 310 |
|
| 311 |
return $self->{'current_mfn'} = (($pos - 1) || -1); |
| 312 |
} |
| 313 |
|
| 314 |
=head1 AUTHOR |
| 315 |
|
| 316 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
| 317 |
|
| 318 |
=head1 COPYRIGHT & LICENSE |
| 319 |
|
| 320 |
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
| 321 |
|
| 322 |
This program is free software; you can redistribute it and/or modify it |
| 323 |
under the same terms as Perl itself. |
| 324 |
|
| 325 |
=cut |
| 326 |
|
| 327 |
1; # End of WebPAC::Input::ISIS |
Parent Directory
| 
Revision Log