1 |
dpavlin |
6 |
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 |
dpavlin |
9 |
use Text::Iconv; |
9 |
dpavlin |
6 |
|
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 => $lookup_obj, |
55 |
|
|
); |
56 |
|
|
|
57 |
|
|
By default, ISIS code page is assumed to be C<852>. |
58 |
|
|
|
59 |
|
|
If optional parametar C<start_mfn> is set, this will be first MFN to read |
60 |
|
|
from database (so you can skip beginning of your database if you need to). |
61 |
|
|
|
62 |
|
|
If optional parametar C<limit_mfn> is set, it will read just 500 records |
63 |
|
|
from database in example above. |
64 |
|
|
|
65 |
dpavlin |
21 |
Returns size of database, regardless of C<start_mfn> and C<limit_mfn> |
66 |
|
|
parametars, see also C<$isis->size>. |
67 |
dpavlin |
6 |
|
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 |
|
|
# store data in object |
80 |
|
|
$self->{'isis_code_page'} = $code_page; |
81 |
dpavlin |
21 |
foreach my $v (qw/isis_filename start_mfn limit_mfn/) { |
82 |
|
|
$self->{$v} = $arg->{$v} if ($arg->{$v}); |
83 |
|
|
} |
84 |
dpavlin |
6 |
|
85 |
|
|
# create Text::Iconv object |
86 |
|
|
my $cp = Text::Iconv->new($code_page,$self->{'code_page'}); |
87 |
|
|
|
88 |
|
|
$log->info("reading ISIS database '",$arg->{'filename'},"'"); |
89 |
|
|
$log->debug("isis code page: $code_page"); |
90 |
|
|
|
91 |
dpavlin |
21 |
my ($isis_db,$db_size); |
92 |
dpavlin |
6 |
|
93 |
|
|
if ($have_openisis) { |
94 |
|
|
$log->debug("using OpenIsis perl bindings"); |
95 |
|
|
$isis_db = OpenIsis::open($arg->{'filename'}); |
96 |
dpavlin |
21 |
$db_size = OpenIsis::maxRowid( $isis_db ) || 1; |
97 |
dpavlin |
6 |
} elsif ($have_biblio_isis) { |
98 |
|
|
$log->debug("using Biblio::Isis"); |
99 |
|
|
use Biblio::Isis; |
100 |
|
|
$isis_db = new Biblio::Isis( |
101 |
|
|
isisdb => $arg->{'filename'}, |
102 |
|
|
include_deleted => 1, |
103 |
|
|
hash_filter => sub { |
104 |
|
|
my $l = shift || return; |
105 |
|
|
$l = $cp->convert($l); |
106 |
|
|
return $l; |
107 |
|
|
}, |
108 |
dpavlin |
265 |
) or $log->logdie("can't find database ",$arg->{'filename'}); |
109 |
|
|
|
110 |
dpavlin |
21 |
$db_size = $isis_db->count; |
111 |
dpavlin |
6 |
|
112 |
dpavlin |
21 |
unless ($db_size) { |
113 |
dpavlin |
6 |
$log->logwarn("no records in database ", $arg->{'filename'}, ", skipping..."); |
114 |
|
|
return; |
115 |
|
|
} |
116 |
|
|
|
117 |
|
|
} else { |
118 |
|
|
$log->logdie("Can't find supported ISIS library for perl. I suggent that you install Bilbio::Isis from CPAN."); |
119 |
|
|
} |
120 |
|
|
|
121 |
|
|
|
122 |
|
|
my $startmfn = 1; |
123 |
dpavlin |
21 |
my $maxmfn = $db_size; |
124 |
dpavlin |
6 |
|
125 |
|
|
if (my $s = $self->{'start_mfn'}) { |
126 |
|
|
$log->info("skipping to MFN $s"); |
127 |
|
|
$startmfn = $s; |
128 |
|
|
} else { |
129 |
|
|
$self->{'start_mfn'} = $startmfn; |
130 |
|
|
} |
131 |
|
|
|
132 |
dpavlin |
21 |
if ($self->{limit_mfn}) { |
133 |
|
|
$log->info("limiting to ",$self->{limit_mfn}," records"); |
134 |
|
|
$maxmfn = $startmfn + $self->{limit_mfn} - 1; |
135 |
|
|
$maxmfn = $db_size if ($maxmfn > $db_size); |
136 |
|
|
} |
137 |
dpavlin |
6 |
|
138 |
dpavlin |
21 |
# store size for later |
139 |
|
|
$self->{'size'} = ($maxmfn - $startmfn) ? ($maxmfn - $startmfn + 1) : 0; |
140 |
|
|
|
141 |
dpavlin |
6 |
$log->info("processing ",($maxmfn-$startmfn)." records using ",( $have_openisis ? 'OpenIsis' : 'Biblio::Isis')); |
142 |
|
|
|
143 |
|
|
|
144 |
|
|
# read database |
145 |
|
|
for (my $mfn = $startmfn; $mfn <= $maxmfn; $mfn++) { |
146 |
|
|
|
147 |
|
|
$log->debug("mfn: $mfn\n"); |
148 |
|
|
|
149 |
|
|
my $rec; |
150 |
|
|
|
151 |
|
|
if ($have_openisis) { |
152 |
|
|
|
153 |
|
|
# read record using OpenIsis |
154 |
|
|
my $row = OpenIsis::read( $isis_db, $mfn ); |
155 |
|
|
foreach my $k (keys %{$row}) { |
156 |
|
|
if ($k ne "mfn") { |
157 |
|
|
foreach my $l (@{$row->{$k}}) { |
158 |
|
|
$l = $cp->convert($l); |
159 |
|
|
# has subfields? |
160 |
|
|
my $val; |
161 |
|
|
if ($l =~ m/\^/) { |
162 |
|
|
foreach my $t (split(/\^/,$l)) { |
163 |
|
|
next if (! $t); |
164 |
|
|
$val->{substr($t,0,1)} = substr($t,1); |
165 |
|
|
} |
166 |
|
|
} else { |
167 |
|
|
$val = $l; |
168 |
|
|
} |
169 |
|
|
|
170 |
|
|
push @{$rec->{$k}}, $val; |
171 |
|
|
} |
172 |
|
|
} else { |
173 |
|
|
push @{$rec->{'000'}}, $mfn; |
174 |
|
|
} |
175 |
|
|
} |
176 |
|
|
|
177 |
|
|
} elsif ($have_biblio_isis) { |
178 |
|
|
$rec = $isis_db->to_hash($mfn); |
179 |
|
|
} else { |
180 |
|
|
$log->logdie("hum? implementation missing?"); |
181 |
|
|
} |
182 |
|
|
|
183 |
dpavlin |
113 |
if (! $rec) { |
184 |
|
|
$log->warn("record $mfn empty? skipping..."); |
185 |
|
|
next; |
186 |
|
|
} |
187 |
dpavlin |
6 |
|
188 |
|
|
# store |
189 |
|
|
if ($self->{'low_mem'}) { |
190 |
|
|
$self->{'db'}->put($mfn, $rec); |
191 |
|
|
} else { |
192 |
|
|
$self->{'data'}->{$mfn} = $rec; |
193 |
|
|
} |
194 |
|
|
|
195 |
|
|
# create lookup |
196 |
dpavlin |
251 |
$self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); |
197 |
dpavlin |
6 |
|
198 |
|
|
$self->progress_bar($mfn,$maxmfn); |
199 |
|
|
|
200 |
|
|
} |
201 |
|
|
|
202 |
|
|
$self->{'current_mfn'} = -1; |
203 |
|
|
$self->{'last_pcnt'} = 0; |
204 |
|
|
|
205 |
|
|
$log->debug("max mfn: $maxmfn"); |
206 |
|
|
|
207 |
|
|
# store max mfn and return it. |
208 |
dpavlin |
21 |
$self->{'max_mfn'} = $maxmfn; |
209 |
|
|
|
210 |
|
|
return $db_size; |
211 |
dpavlin |
6 |
} |
212 |
|
|
|
213 |
dpavlin |
10 |
=head2 fetch |
214 |
dpavlin |
6 |
|
215 |
|
|
Fetch next record from database. It will also displays progress bar. |
216 |
|
|
|
217 |
dpavlin |
10 |
my $rec = $isis->fetch; |
218 |
dpavlin |
6 |
|
219 |
dpavlin |
10 |
Record from this function should probably go to C<data_structure> for |
220 |
|
|
normalisation. |
221 |
dpavlin |
8 |
|
222 |
dpavlin |
6 |
=cut |
223 |
|
|
|
224 |
dpavlin |
10 |
sub fetch { |
225 |
dpavlin |
6 |
my $self = shift; |
226 |
|
|
|
227 |
|
|
my $log = $self->_get_logger(); |
228 |
|
|
|
229 |
|
|
$log->logconfess("it seems that you didn't load database!") unless ($self->{'current_mfn'}); |
230 |
|
|
|
231 |
|
|
if ($self->{'current_mfn'} == -1) { |
232 |
|
|
$self->{'current_mfn'} = $self->{'start_mfn'}; |
233 |
|
|
} else { |
234 |
|
|
$self->{'current_mfn'}++; |
235 |
|
|
} |
236 |
|
|
|
237 |
|
|
my $mfn = $self->{'current_mfn'}; |
238 |
|
|
|
239 |
|
|
if ($mfn > $self->{'max_mfn'}) { |
240 |
|
|
$self->{'current_mfn'} = $self->{'max_mfn'}; |
241 |
|
|
$log->debug("at EOF"); |
242 |
|
|
return; |
243 |
|
|
} |
244 |
|
|
|
245 |
|
|
$self->progress_bar($mfn,$self->{'max_mfn'}); |
246 |
|
|
|
247 |
dpavlin |
113 |
my $rec; |
248 |
|
|
|
249 |
dpavlin |
6 |
if ($self->{'low_mem'}) { |
250 |
dpavlin |
113 |
$rec = $self->{'db'}->get($mfn); |
251 |
dpavlin |
6 |
} else { |
252 |
dpavlin |
113 |
$rec = $self->{'data'}->{$mfn}; |
253 |
dpavlin |
6 |
} |
254 |
dpavlin |
113 |
|
255 |
|
|
$rec ||= 0E0; |
256 |
dpavlin |
6 |
} |
257 |
|
|
|
258 |
dpavlin |
10 |
=head2 pos |
259 |
dpavlin |
8 |
|
260 |
dpavlin |
10 |
Returns current record number (MFN). |
261 |
dpavlin |
8 |
|
262 |
dpavlin |
10 |
print $isis->pos; |
263 |
dpavlin |
8 |
|
264 |
dpavlin |
10 |
First record in database has position 1. |
265 |
|
|
|
266 |
dpavlin |
8 |
=cut |
267 |
|
|
|
268 |
dpavlin |
10 |
sub pos { |
269 |
dpavlin |
8 |
my $self = shift; |
270 |
dpavlin |
10 |
return $self->{'current_mfn'}; |
271 |
|
|
} |
272 |
dpavlin |
8 |
|
273 |
dpavlin |
10 |
|
274 |
|
|
=head2 size |
275 |
|
|
|
276 |
|
|
Returns number of records in database |
277 |
|
|
|
278 |
|
|
print $isis->size; |
279 |
|
|
|
280 |
dpavlin |
21 |
Result from this function can be used to loop through all records |
281 |
|
|
|
282 |
|
|
foreach my $mfn ( 1 ... $isis->size ) { ... } |
283 |
|
|
|
284 |
|
|
because it takes into account C<start_mfn> and C<limit_mfn>. |
285 |
|
|
|
286 |
dpavlin |
10 |
=cut |
287 |
|
|
|
288 |
|
|
sub size { |
289 |
|
|
my $self = shift; |
290 |
dpavlin |
21 |
return $self->{'size'}; |
291 |
dpavlin |
8 |
} |
292 |
|
|
|
293 |
dpavlin |
10 |
=head2 seek |
294 |
dpavlin |
6 |
|
295 |
dpavlin |
10 |
Seek to specified MFN in file. |
296 |
dpavlin |
6 |
|
297 |
dpavlin |
10 |
$isis->seek(42); |
298 |
dpavlin |
6 |
|
299 |
dpavlin |
10 |
First record in database has position 1. |
300 |
|
|
|
301 |
dpavlin |
6 |
=cut |
302 |
|
|
|
303 |
dpavlin |
10 |
sub seek { |
304 |
dpavlin |
6 |
my $self = shift; |
305 |
dpavlin |
10 |
my $pos = shift || return; |
306 |
|
|
|
307 |
|
|
my $log = $self->_get_logger(); |
308 |
|
|
|
309 |
|
|
if ($pos < 1) { |
310 |
|
|
$log->warn("seek before first record"); |
311 |
|
|
$pos = 1; |
312 |
|
|
} elsif ($pos > $self->{'max_mfn'}) { |
313 |
|
|
$log->warn("seek beyond last record"); |
314 |
|
|
$pos = $self->{'max_mfn'}; |
315 |
|
|
} |
316 |
|
|
|
317 |
|
|
return $self->{'current_mfn'} = (($pos - 1) || -1); |
318 |
dpavlin |
6 |
} |
319 |
|
|
|
320 |
|
|
=head1 AUTHOR |
321 |
|
|
|
322 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
323 |
|
|
|
324 |
|
|
=head1 COPYRIGHT & LICENSE |
325 |
|
|
|
326 |
|
|
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
327 |
|
|
|
328 |
|
|
This program is free software; you can redistribute it and/or modify it |
329 |
|
|
under the same terms as Perl itself. |
330 |
|
|
|
331 |
|
|
=cut |
332 |
|
|
|
333 |
|
|
1; # End of WebPAC::Input::ISIS |