16 |
|
|
17 |
=head1 VERSION |
=head1 VERSION |
18 |
|
|
19 |
Version 0.17 |
Version 0.18 |
20 |
|
|
21 |
=cut |
=cut |
22 |
|
|
23 |
our $VERSION = '0.17'; |
our $VERSION = '0.18'; |
24 |
|
|
25 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
26 |
|
|
64 |
encoding => 'ISO-8859-2', |
encoding => 'ISO-8859-2', |
65 |
recode => 'char pairs', |
recode => 'char pairs', |
66 |
no_progress_bar => 1, |
no_progress_bar => 1, |
67 |
|
input_config => { |
68 |
|
mapping => [ 'foo', 'bar', 'baz' ], |
69 |
|
}, |
70 |
); |
); |
71 |
|
|
72 |
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 |
236 |
|
|
237 |
my $ll_db = $class->new( |
my $ll_db = $class->new( |
238 |
path => $arg->{path}, |
path => $arg->{path}, |
239 |
|
input_config => $arg->{input_config}, |
240 |
# filter => sub { |
# filter => sub { |
241 |
# my ($l,$f_nr) = @_; |
# my ($l,$f_nr) = @_; |
242 |
# return unless defined($l); |
# return unless defined($l); |
290 |
# return unless defined($l); |
# return unless defined($l); |
291 |
# return $l unless ($rec_regex && $f_nr); |
# return $l unless ($rec_regex && $f_nr); |
292 |
|
|
293 |
|
return unless ( defined($l) && defined($f_nr) ); |
294 |
|
|
295 |
warn "-=> $f_nr ## |$l|\n" if ($debug); |
warn "-=> $f_nr ## |$l|\n" if ($debug); |
296 |
$log->debug("-=> $f_nr ## $l"); |
$log->debug("-=> $f_nr ## $l"); |
297 |
|
|
523 |
|
|
524 |
my $out = join("\n", |
my $out = join("\n", |
525 |
map { |
map { |
526 |
my $f = $_ || die "no field"; |
my $f = $_; |
527 |
|
die "no field in ", dump( $s->{fld} ) unless defined( $f ); |
528 |
my $v = $s->{fld}->{$f} || die "no s->{fld}->{$f}"; |
my $v = $s->{fld}->{$f} || die "no s->{fld}->{$f}"; |
529 |
$max_fld = $v if ($v > $max_fld); |
$max_fld = $v if ($v > $max_fld); |
530 |
|
|
544 |
} |
} |
545 |
|
|
546 |
$o; |
$o; |
547 |
} sort { $a cmp $b } keys %{ $s->{fld} } |
} sort { $a <=> $b } keys %{ $s->{fld} } |
548 |
); |
); |
549 |
|
|
550 |
$log->debug( sub { dump($s) } ); |
$log->debug( sub { dump($s) } ); |
563 |
|
|
564 |
return unless $self->{ll_db}; |
return unless $self->{ll_db}; |
565 |
|
|
566 |
if ($self->{ll_db}->can('dump_rec')) { |
if ($self->{ll_db}->can('dump_ascii')) { |
567 |
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
568 |
} else { |
} else { |
569 |
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
570 |
} |
} |
571 |
} |
} |
572 |
|
|
573 |
=head2 modify_record_regexps |
=head2 _get_regex |
574 |
|
|
575 |
Generate hash with regexpes to be applied using L<filter>. |
Helper function called which create regexps to be execute on code. |
576 |
|
|
577 |
my $regexpes = $input->modify_record_regexps( |
_get_regex( 900, 'regex:[0-9]+' ,'numbers' ); |
578 |
900 => { '^a' => { ' : ' => '^b' } }, |
_get_regex( 900, '^b', ' : ^b' ); |
579 |
901 => { '*' => { '^b' => ' ; ' } }, |
|
580 |
); |
It supports perl regexps with C<regex:> prefix to from value and has |
581 |
|
additional logic to skip empty subfields. |
582 |
|
|
583 |
=cut |
=cut |
584 |
|
|
595 |
$from = '\Q' . $from . '\E'; |
$from = '\Q' . $from . '\E'; |
596 |
} |
} |
597 |
if ($sf =~ /^\^/) { |
if ($sf =~ /^\^/) { |
598 |
|
my $need_subfield_data = '*'; # no |
599 |
|
# if from is also subfield, require some data in between |
600 |
|
# to correctly skip empty subfields |
601 |
|
$need_subfield_data = '+' if ($from =~ m/^\\Q\^/); |
602 |
return |
return |
603 |
's/\Q'. $sf .'\E([^\^]*?)'. $from .'([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
's/\Q'. $sf .'\E([^\^]' . $need_subfield_data . '?)'. $from .'([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
604 |
} else { |
} else { |
605 |
return |
return |
606 |
's/'. $from .'/'. $to .'/g'; |
's/'. $from .'/'. $to .'/g'; |
607 |
} |
} |
608 |
} |
} |
609 |
|
|
610 |
|
|
611 |
|
=head2 modify_record_regexps |
612 |
|
|
613 |
|
Generate hash with regexpes to be applied using L<filter>. |
614 |
|
|
615 |
|
my $regexpes = $input->modify_record_regexps( |
616 |
|
900 => { '^a' => { ' : ' => '^b' } }, |
617 |
|
901 => { '*' => { '^b' => ' ; ' } }, |
618 |
|
); |
619 |
|
|
620 |
|
=cut |
621 |
|
|
622 |
sub modify_record_regexps { |
sub modify_record_regexps { |
623 |
my $self = shift; |
my $self = shift; |
624 |
my $modify_record = {@_}; |
my $modify_record = {@_}; |