62 |
If optional parametar C<limit_mfn> is set, it will read just 500 records |
If optional parametar C<limit_mfn> is set, it will read just 500 records |
63 |
from database in example above. |
from database in example above. |
64 |
|
|
65 |
Returns number of last record read into memory (size of database, really). |
Returns size of database, regardless of C<start_mfn> and C<limit_mfn> |
66 |
|
parametars, see also C<$isis->size>. |
67 |
|
|
68 |
=cut |
=cut |
69 |
|
|
76 |
$log->logcroak("need filename") if (! $arg->{'filename'}); |
$log->logcroak("need filename") if (! $arg->{'filename'}); |
77 |
my $code_page = $arg->{'code_page'} || '852'; |
my $code_page = $arg->{'code_page'} || '852'; |
78 |
|
|
|
$log->logdie("can't find database ",$arg->{'filename'}) unless (glob($arg->{'filename'}.'.*')); |
|
|
|
|
79 |
# store data in object |
# store data in object |
|
$self->{'isis_filename'} = $arg->{'filename'}; |
|
80 |
$self->{'isis_code_page'} = $code_page; |
$self->{'isis_code_page'} = $code_page; |
81 |
|
foreach my $v (qw/isis_filename start_mfn limit_mfn/) { |
82 |
#$self->{'isis_code_page'} = $code_page; |
$self->{$v} = $arg->{$v} if ($arg->{$v}); |
83 |
|
} |
84 |
|
|
85 |
# create Text::Iconv object |
# create Text::Iconv object |
86 |
my $cp = Text::Iconv->new($code_page,$self->{'code_page'}); |
my $cp = Text::Iconv->new($code_page,$self->{'code_page'}); |
88 |
$log->info("reading ISIS database '",$arg->{'filename'},"'"); |
$log->info("reading ISIS database '",$arg->{'filename'},"'"); |
89 |
$log->debug("isis code page: $code_page"); |
$log->debug("isis code page: $code_page"); |
90 |
|
|
91 |
my ($isis_db,$maxmfn); |
my ($isis_db,$db_size); |
92 |
|
|
93 |
if ($have_openisis) { |
if ($have_openisis) { |
94 |
$log->debug("using OpenIsis perl bindings"); |
$log->debug("using OpenIsis perl bindings"); |
95 |
$isis_db = OpenIsis::open($arg->{'filename'}); |
$isis_db = OpenIsis::open($arg->{'filename'}); |
96 |
$maxmfn = OpenIsis::maxRowid( $isis_db ) || 1; |
$db_size = OpenIsis::maxRowid( $isis_db ) || 1; |
97 |
} elsif ($have_biblio_isis) { |
} elsif ($have_biblio_isis) { |
98 |
$log->debug("using Biblio::Isis"); |
$log->debug("using Biblio::Isis"); |
99 |
use Biblio::Isis; |
use Biblio::Isis; |
105 |
$l = $cp->convert($l); |
$l = $cp->convert($l); |
106 |
return $l; |
return $l; |
107 |
}, |
}, |
108 |
); |
) or $log->logdie("can't find database ",$arg->{'filename'}); |
109 |
$maxmfn = $isis_db->count; |
|
110 |
|
$db_size = $isis_db->count; |
111 |
|
|
112 |
unless ($maxmfn) { |
unless ($db_size) { |
113 |
$log->logwarn("no records in database ", $arg->{'filename'}, ", skipping..."); |
$log->logwarn("no records in database ", $arg->{'filename'}, ", skipping..."); |
114 |
return; |
return; |
115 |
} |
} |
120 |
|
|
121 |
|
|
122 |
my $startmfn = 1; |
my $startmfn = 1; |
123 |
|
my $maxmfn = $db_size; |
124 |
|
|
125 |
if (my $s = $self->{'start_mfn'}) { |
if (my $s = $self->{'start_mfn'}) { |
126 |
$log->info("skipping to MFN $s"); |
$log->info("skipping to MFN $s"); |
129 |
$self->{'start_mfn'} = $startmfn; |
$self->{'start_mfn'} = $startmfn; |
130 |
} |
} |
131 |
|
|
132 |
$maxmfn = $startmfn + $self->{limit_mfn} if ($self->{limit_mfn}); |
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 |
|
|
138 |
|
# store size for later |
139 |
|
$self->{'size'} = ($maxmfn - $startmfn) ? ($maxmfn - $startmfn + 1) : 0; |
140 |
|
|
141 |
$log->info("processing ",($maxmfn-$startmfn)." records using ",( $have_openisis ? 'OpenIsis' : 'Biblio::Isis')); |
$log->info("processing ",($maxmfn-$startmfn)." records using ",( $have_openisis ? 'OpenIsis' : 'Biblio::Isis')); |
142 |
|
|
180 |
$log->logdie("hum? implementation missing?"); |
$log->logdie("hum? implementation missing?"); |
181 |
} |
} |
182 |
|
|
183 |
$log->confess("record $mfn empty?") unless ($rec); |
if (! $rec) { |
184 |
|
$log->warn("record $mfn empty? skipping..."); |
185 |
|
next; |
186 |
|
} |
187 |
|
|
188 |
# store |
# store |
189 |
if ($self->{'low_mem'}) { |
if ($self->{'low_mem'}) { |
193 |
} |
} |
194 |
|
|
195 |
# create lookup |
# create lookup |
196 |
$self->{'lookup'}->add( $rec ) if ($self->{'lookup'} && can($self->{'lookup'}->add)); |
$self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); |
197 |
|
|
198 |
$self->progress_bar($mfn,$maxmfn); |
$self->progress_bar($mfn,$maxmfn); |
199 |
|
|
205 |
$log->debug("max mfn: $maxmfn"); |
$log->debug("max mfn: $maxmfn"); |
206 |
|
|
207 |
# store max mfn and return it. |
# store max mfn and return it. |
208 |
return $self->{'max_mfn'} = $maxmfn; |
$self->{'max_mfn'} = $maxmfn; |
209 |
|
|
210 |
|
return $db_size; |
211 |
} |
} |
212 |
|
|
213 |
=head2 fetch_rec |
=head2 fetch |
214 |
|
|
215 |
Fetch next record from database. It will also displays progress bar. |
Fetch next record from database. It will also displays progress bar. |
216 |
|
|
217 |
my $rec = $webpac->fetch_rec; |
my $rec = $isis->fetch; |
218 |
|
|
219 |
You should rearly have the need to call this function directly. Instead use |
Record from this function should probably go to C<data_structure> for |
220 |
C<fetch_data_structure> which returns normalised data. |
normalisation. |
221 |
|
|
222 |
=cut |
=cut |
223 |
|
|
224 |
sub fetch_rec { |
sub fetch { |
225 |
my $self = shift; |
my $self = shift; |
226 |
|
|
227 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
244 |
|
|
245 |
$self->progress_bar($mfn,$self->{'max_mfn'}); |
$self->progress_bar($mfn,$self->{'max_mfn'}); |
246 |
|
|
247 |
|
my $rec; |
248 |
|
|
249 |
if ($self->{'low_mem'}) { |
if ($self->{'low_mem'}) { |
250 |
return $self->{'db'}->get($mfn); |
$rec = $self->{'db'}->get($mfn); |
251 |
} else { |
} else { |
252 |
return $self->{'data'}->{$mfn}; |
$rec = $self->{'data'}->{$mfn}; |
253 |
} |
} |
254 |
|
|
255 |
|
$rec ||= 0E0; |
256 |
} |
} |
257 |
|
|
258 |
=head2 fetch_data_structure |
=head2 pos |
259 |
|
|
260 |
|
Returns current record number (MFN). |
261 |
|
|
262 |
Fetch data structure of next record from database. |
print $isis->pos; |
263 |
|
|
264 |
my @ds = $webpac->fetch_data_structure; |
First record in database has position 1. |
265 |
|
|
266 |
=cut |
=cut |
267 |
|
|
268 |
sub fetch_data_structure { |
sub pos { |
269 |
my $self = shift; |
my $self = shift; |
270 |
|
return $self->{'current_mfn'}; |
271 |
|
} |
272 |
|
|
273 |
|
|
274 |
|
=head2 size |
275 |
|
|
276 |
return $self->data_structure( |
Returns number of records in database |
277 |
$self->fetch_rec(@_) |
|
278 |
); |
print $isis->size; |
279 |
|
|
280 |
|
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 |
|
=cut |
287 |
|
|
288 |
|
sub size { |
289 |
|
my $self = shift; |
290 |
|
return $self->{'size'}; |
291 |
} |
} |
292 |
|
|
293 |
=head2 mfn |
=head2 seek |
294 |
|
|
295 |
Returns current record number (MFN). |
Seek to specified MFN in file. |
296 |
|
|
297 |
|
$isis->seek(42); |
298 |
|
|
299 |
print $webpac->mfn; |
First record in database has position 1. |
300 |
|
|
301 |
=cut |
=cut |
302 |
|
|
303 |
sub mfn { |
sub seek { |
304 |
my $self = shift; |
my $self = shift; |
305 |
return $self->{'current_mfn'}; |
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 |
} |
} |
319 |
|
|
320 |
=head1 AUTHOR |
=head1 AUTHOR |