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 |
|
|
282 |
$log->debug("position: $pos\n"); |
$log->debug("position: $pos\n"); |
283 |
|
|
284 |
my $rec = $ll_db->fetch_rec($pos, sub { |
my $rec = $ll_db->fetch_rec($pos, sub { |
285 |
my ($l,$f_nr) = @_; |
my ($l,$f_nr,$debug) = @_; |
286 |
# return unless defined($l); |
# return unless defined($l); |
287 |
# return $l unless ($rec_regex && $f_nr); |
# return $l unless ($rec_regex && $f_nr); |
288 |
|
|
289 |
|
return unless ( defined($l) && defined($f_nr) ); |
290 |
|
|
291 |
|
warn "-=> $f_nr ## |$l|\n" if ($debug); |
292 |
$log->debug("-=> $f_nr ## $l"); |
$log->debug("-=> $f_nr ## $l"); |
293 |
|
|
294 |
# codepage conversion and recode_regex |
# codepage conversion and recode_regex |
301 |
my $c = 0; |
my $c = 0; |
302 |
foreach my $r (@{ $rec_regex->{$f_nr} }) { |
foreach my $r (@{ $rec_regex->{$f_nr} }) { |
303 |
my $old_l = $l; |
my $old_l = $l; |
304 |
eval '$l =~ ' . $r; |
$log->logconfess("expected regex in ", dump( $r )) unless defined($r->{regex}); |
305 |
|
eval '$l =~ ' . $r->{regex}; |
306 |
if ($old_l ne $l) { |
if ($old_l ne $l) { |
307 |
$log->debug("REGEX on $f_nr eval \$l =~ $r\n## old l: [$old_l]\n## new l: [$l]"); |
my $d = "|$old_l| -> |$l| "; # . $r->{regex}; |
308 |
|
$d .= ' +' . $r->{line} . ' ' . $r->{file} if defined($r->{line}); |
309 |
|
$d .= ' ' . $r->{debug} if defined($r->{debug}); |
310 |
|
$log->debug("MODIFY $d"); |
311 |
|
warn "*** $d\n" if ($debug); |
312 |
|
|
313 |
} |
} |
314 |
$log->error("error applying regex: $r") if ($@); |
$log->error("error applying regex: $r") if ($@); |
315 |
} |
} |
316 |
} |
} |
317 |
|
|
318 |
$log->debug("<=- $f_nr ## $l"); |
$log->debug("<=- $f_nr ## |$l|"); |
319 |
|
warn "<=- $f_nr ## $l\n" if ($debug); |
320 |
return $l; |
return $l; |
321 |
}); |
}); |
322 |
|
|
519 |
|
|
520 |
my $out = join("\n", |
my $out = join("\n", |
521 |
map { |
map { |
522 |
my $f = $_ || die "no field"; |
my $f = $_; |
523 |
|
die "no field in ", dump( $s->{fld} ) unless defined( $f ); |
524 |
my $v = $s->{fld}->{$f} || die "no s->{fld}->{$f}"; |
my $v = $s->{fld}->{$f} || die "no s->{fld}->{$f}"; |
525 |
$max_fld = $v if ($v > $max_fld); |
$max_fld = $v if ($v > $max_fld); |
526 |
|
|
540 |
} |
} |
541 |
|
|
542 |
$o; |
$o; |
543 |
} sort { $a cmp $b } keys %{ $s->{fld} } |
} sort { $a <=> $b } keys %{ $s->{fld} } |
544 |
); |
); |
545 |
|
|
546 |
$log->debug( sub { dump($s) } ); |
$log->debug( sub { dump($s) } ); |
559 |
|
|
560 |
return unless $self->{ll_db}; |
return unless $self->{ll_db}; |
561 |
|
|
562 |
if ($self->{ll_db}->can('dump_rec')) { |
if ($self->{ll_db}->can('dump_ascii')) { |
563 |
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
564 |
} else { |
} else { |
565 |
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
566 |
} |
} |
567 |
} |
} |
568 |
|
|
569 |
=head2 modify_record_regexps |
=head2 _get_regex |
570 |
|
|
571 |
Generate hash with regexpes to be applied using l<filter>. |
Helper function called which create regexps to be execute on code. |
572 |
|
|
573 |
my $regexpes = $input->modify_record_regexps( |
_get_regex( 900, 'regex:[0-9]+' ,'numbers' ); |
574 |
900 => { '^a' => { ' : ' => '^b' } }, |
_get_regex( 900, '^b', ' : ^b' ); |
575 |
901 => { '*' => { '^b' => ' ; ' } }, |
|
576 |
); |
It supports perl regexps with C<regex:> prefix to from value and has |
577 |
|
additional logic to skip empty subfields. |
578 |
|
|
579 |
=cut |
=cut |
580 |
|
|
591 |
$from = '\Q' . $from . '\E'; |
$from = '\Q' . $from . '\E'; |
592 |
} |
} |
593 |
if ($sf =~ /^\^/) { |
if ($sf =~ /^\^/) { |
594 |
|
my $need_subfield_data = '*'; # no |
595 |
|
# if from is also subfield, require some data in between |
596 |
|
# to correctly skip empty subfields |
597 |
|
$need_subfield_data = '+' if ($from =~ m/^\\Q\^/); |
598 |
return |
return |
599 |
's/\Q'. $sf .'\E([^\^]*?)'. $from .'([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
's/\Q'. $sf .'\E([^\^]' . $need_subfield_data . '?)'. $from .'([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
600 |
} else { |
} else { |
601 |
return |
return |
602 |
's/'. $from .'/'. $to .'/g'; |
's/'. $from .'/'. $to .'/g'; |
603 |
} |
} |
604 |
} |
} |
605 |
|
|
606 |
|
|
607 |
|
=head2 modify_record_regexps |
608 |
|
|
609 |
|
Generate hash with regexpes to be applied using L<filter>. |
610 |
|
|
611 |
|
my $regexpes = $input->modify_record_regexps( |
612 |
|
900 => { '^a' => { ' : ' => '^b' } }, |
613 |
|
901 => { '*' => { '^b' => ' ; ' } }, |
614 |
|
); |
615 |
|
|
616 |
|
=cut |
617 |
|
|
618 |
sub modify_record_regexps { |
sub modify_record_regexps { |
619 |
my $self = shift; |
my $self = shift; |
620 |
my $modify_record = {@_}; |
my $modify_record = {@_}; |
632 |
foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { |
foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { |
633 |
my $to = $modify_record->{$f}->{$sf}->{$from}; |
my $to = $modify_record->{$f}->{$sf}->{$from}; |
634 |
#die "no field?" unless defined($to); |
#die "no field?" unless defined($to); |
635 |
$log->debug("transform: |$from| -> |$to|"); |
my $d = "|$from| -> |$to|"; |
636 |
|
$log->debug("transform: $d"); |
637 |
|
|
638 |
my $regex = _get_regex($sf,$from,$to); |
my $regex = _get_regex($sf,$from,$to); |
639 |
push @{ $regexpes->{$f} }, $regex; |
push @{ $regexpes->{$f} }, { regex => $regex, debug => $d }; |
640 |
$log->debug("regex: $regex"); |
$log->debug("regex: $regex"); |
641 |
} |
} |
642 |
} |
} |
647 |
|
|
648 |
=head2 modify_file_regexps |
=head2 modify_file_regexps |
649 |
|
|
650 |
Generate hash with regexpes to be applied using l<filter> from |
Generate hash with regexpes to be applied using L<filter> from |
651 |
pseudo hash/yaml format for regex mappings. |
pseudo hash/yaml format for regex mappings. |
652 |
|
|
653 |
It should be obvious: |
It should be obvious: |
697 |
$log->debug("transform: |$from| -> |$to|"); |
$log->debug("transform: |$from| -> |$to|"); |
698 |
|
|
699 |
my $regex = _get_regex($sf,$from,$to); |
my $regex = _get_regex($sf,$from,$to); |
700 |
push @{ $regexpes->{$f} }, $regex; |
push @{ $regexpes->{$f} }, { |
701 |
|
regex => $regex, |
702 |
|
file => $modify_path, |
703 |
|
line => $., |
704 |
|
}; |
705 |
$log->debug("regex: $regex"); |
$log->debug("regex: $regex"); |
706 |
} |
} |
707 |
} |
} |