3 |
use warnings; |
use warnings; |
4 |
use strict; |
use strict; |
5 |
|
|
6 |
|
use blib; |
7 |
|
|
8 |
use WebPAC::Common; |
use WebPAC::Common; |
9 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
10 |
use Text::Iconv; |
use Text::Iconv; |
16 |
|
|
17 |
=head1 VERSION |
=head1 VERSION |
18 |
|
|
19 |
Version 0.04 |
Version 0.05 |
20 |
|
|
21 |
=cut |
=cut |
22 |
|
|
23 |
our $VERSION = '0.04'; |
our $VERSION = '0.05'; |
24 |
|
|
25 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
26 |
|
|
67 |
code_page => 'ISO-8859-2', |
code_page => 'ISO-8859-2', |
68 |
low_mem => 1, |
low_mem => 1, |
69 |
recode => 'char pairs', |
recode => 'char pairs', |
70 |
|
no_progress_bar => 1, |
71 |
); |
); |
72 |
|
|
73 |
C<module> is low-level file format module. See L<WebPAC::Input::Isis> and |
C<module> is low-level file format module. See L<WebPAC::Input::Isis> and |
79 |
|
|
80 |
Default is not to use C<low_mem> options (see L<MEMORY USAGE> below). |
Default is not to use C<low_mem> options (see L<MEMORY USAGE> below). |
81 |
|
|
82 |
|
C<recode> is optional string constisting of character or words pairs that |
83 |
|
should be replaced in input stream. |
84 |
|
|
85 |
|
C<no_progress_bar> disables progress bar output on C<STDOUT> |
86 |
|
|
87 |
This function will also call low-level C<init> if it exists with same |
This function will also call low-level C<init> if it exists with same |
88 |
parametars. |
parametars. |
89 |
|
|
162 |
limit => 500, |
limit => 500, |
163 |
offset => 6000, |
offset => 6000, |
164 |
lookup => $lookup_obj, |
lookup => $lookup_obj, |
165 |
|
stats => 1, |
166 |
); |
); |
167 |
|
|
168 |
By default, C<code_page> is assumed to be C<852>. |
By default, C<code_page> is assumed to be C<852>. |
171 |
|
|
172 |
C<limit> is optional parametar to read just C<limit> records from database |
C<limit> is optional parametar to read just C<limit> records from database |
173 |
|
|
174 |
|
C<stats> create optional report about usage of fields and subfields |
175 |
|
|
176 |
Returns size of database, regardless of C<offset> and C<limit> |
Returns size of database, regardless of C<offset> and C<limit> |
177 |
parametars, see also C<size>. |
parametars, see also C<size>. |
178 |
|
|
227 |
my ($db, $size) = $self->{open_db}->( $self, |
my ($db, $size) = $self->{open_db}->( $self, |
228 |
path => $arg->{path}, |
path => $arg->{path}, |
229 |
filter => $filter_ref, |
filter => $filter_ref, |
230 |
|
%{ $arg }, |
231 |
); |
); |
232 |
|
|
233 |
unless ($db) { |
unless (defined($db)) { |
234 |
$log->logwarn("can't open database $arg->{path}, skipping..."); |
$log->logwarn("can't open database $arg->{path}, skipping..."); |
235 |
return; |
return; |
236 |
} |
} |
244 |
my $to_rec = $size; |
my $to_rec = $size; |
245 |
|
|
246 |
if (my $s = $self->{offset}) { |
if (my $s = $self->{offset}) { |
247 |
$log->info("skipping to MFN $s"); |
$log->debug("skipping to MFN $s"); |
248 |
$from_rec = $s; |
$from_rec = $s; |
249 |
} else { |
} else { |
250 |
$self->{offset} = $from_rec; |
$self->{offset} = $from_rec; |
259 |
# store size for later |
# store size for later |
260 |
$self->{size} = ($to_rec - $from_rec) ? ($to_rec - $from_rec + 1) : 0; |
$self->{size} = ($to_rec - $from_rec) ? ($to_rec - $from_rec + 1) : 0; |
261 |
|
|
262 |
$log->info("processing $self->{size}/$size records [$from_rec-$to_rec] convert $code_page -> $self->{code_page}"); |
$log->info("processing $self->{size}/$size records [$from_rec-$to_rec] convert $code_page -> $self->{code_page}", $self->{stats} ? ' [stats]' : ''); |
263 |
|
|
264 |
# read database |
# read database |
265 |
for (my $pos = $from_rec; $pos <= $to_rec; $pos++) { |
for (my $pos = $from_rec; $pos <= $to_rec; $pos++) { |
285 |
# create lookup |
# create lookup |
286 |
$self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); |
$self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); |
287 |
|
|
288 |
$self->progress_bar($pos,$to_rec); |
# update counters for statistics |
289 |
|
if ($self->{stats}) { |
290 |
|
map { |
291 |
|
my $fld = $_; |
292 |
|
$self->{_stats}->{fld}->{ $fld }++; |
293 |
|
if (ref($rec->{ $fld }) eq 'ARRAY') { |
294 |
|
map { |
295 |
|
if (ref($_) eq 'HASH') { |
296 |
|
map { |
297 |
|
$self->{_stats}->{sf}->{ $fld }->{ $_ }++; |
298 |
|
} keys %{ $_ }; |
299 |
|
} else { |
300 |
|
$self->{_stats}->{repeatable}->{ $fld }++; |
301 |
|
} |
302 |
|
} @{ $rec->{$fld} }; |
303 |
|
} |
304 |
|
} keys %{ $rec }; |
305 |
|
} |
306 |
|
|
307 |
|
$self->progress_bar($pos,$to_rec) unless ($self->{no_progress_bar}); |
308 |
|
|
309 |
} |
} |
310 |
|
|
350 |
return; |
return; |
351 |
} |
} |
352 |
|
|
353 |
$self->progress_bar($mfn,$self->{max_pos}); |
$self->progress_bar($mfn,$self->{max_pos}) unless ($self->{no_progress_bar}); |
354 |
|
|
355 |
my $rec; |
my $rec; |
356 |
|
|
425 |
return $self->{pos} = (($pos - 1) || -1); |
return $self->{pos} = (($pos - 1) || -1); |
426 |
} |
} |
427 |
|
|
428 |
|
=head2 stats |
429 |
|
|
430 |
|
Dump statistics about field and subfield usage |
431 |
|
|
432 |
|
print $input->stats; |
433 |
|
|
434 |
|
=cut |
435 |
|
|
436 |
|
sub stats { |
437 |
|
my $self = shift; |
438 |
|
|
439 |
|
my $log = $self->_get_logger(); |
440 |
|
|
441 |
|
my $s = $self->{_stats}; |
442 |
|
if (! $s) { |
443 |
|
$log->warn("called stats, but there is no statistics collected"); |
444 |
|
return; |
445 |
|
} |
446 |
|
|
447 |
|
my $max_fld = 0; |
448 |
|
|
449 |
|
my $out = join("\n", |
450 |
|
map { |
451 |
|
my $f = $_ || die "no field"; |
452 |
|
my $v = $s->{fld}->{$f} || die "no s->{fld}->{$f}"; |
453 |
|
$max_fld = $v if ($v > $max_fld); |
454 |
|
|
455 |
|
my $o = sprintf("%4s %d ~", $f, $v); |
456 |
|
|
457 |
|
if (defined($s->{sf}->{$f})) { |
458 |
|
map { |
459 |
|
$o .= sprintf(" %s:%d", $_, $s->{sf}->{$f}->{$_}); |
460 |
|
} sort keys %{ $s->{sf}->{$f} }; |
461 |
|
} |
462 |
|
|
463 |
|
if (my $v_r = $s->{repeatable}->{$f}) { |
464 |
|
$o .= " ($v_r)" if ($v_r != $v); |
465 |
|
} |
466 |
|
|
467 |
|
$o; |
468 |
|
} sort { $a cmp $b } keys %{ $s->{fld} } |
469 |
|
); |
470 |
|
|
471 |
|
$log->debug( sub { Dumper($s) } ); |
472 |
|
|
473 |
|
return $out; |
474 |
|
} |
475 |
|
|
476 |
=head1 MEMORY USAGE |
=head1 MEMORY USAGE |
477 |
|
|