7 |
|
|
8 |
use WebPAC::Common; |
use WebPAC::Common; |
9 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
10 |
use Data::Dumper; |
use Data::Dump qw/dump/; |
11 |
use Encode qw/from_to/; |
use Encode qw/from_to/; |
12 |
|
|
13 |
=head1 NAME |
=head1 NAME |
16 |
|
|
17 |
=head1 VERSION |
=head1 VERSION |
18 |
|
|
19 |
Version 0.14 |
Version 0.17 |
20 |
|
|
21 |
=cut |
=cut |
22 |
|
|
23 |
our $VERSION = '0.14'; |
our $VERSION = '0.17'; |
24 |
|
|
25 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
26 |
|
|
102 |
|
|
103 |
require $module_path; |
require $module_path; |
104 |
|
|
|
# check if required subclasses are implemented |
|
|
foreach my $subclass (qw/open_db fetch_rec init dump_rec/) { |
|
|
# FIXME |
|
|
} |
|
|
|
|
105 |
$self->{'encoding'} ||= 'ISO-8859-2'; |
$self->{'encoding'} ||= 'ISO-8859-2'; |
106 |
|
|
107 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
224 |
$log->debug("using modify_file $p"); |
$log->debug("using modify_file $p"); |
225 |
$rec_regex = $self->modify_file_regexps( $p ); |
$rec_regex = $self->modify_file_regexps( $p ); |
226 |
} elsif (my $h = $arg->{modify_records}) { |
} elsif (my $h = $arg->{modify_records}) { |
227 |
$log->debug("using modify_records ", Dumper( $h )); |
$log->debug("using modify_records ", sub { dump( $h ) }); |
228 |
$rec_regex = $self->modify_record_regexps(%{ $h }); |
$rec_regex = $self->modify_record_regexps(%{ $h }); |
229 |
} |
} |
230 |
$log->debug("rec_regex: ", Dumper($rec_regex)) if ($rec_regex); |
$log->debug("rec_regex: ", sub { dump($rec_regex) }) if ($rec_regex); |
231 |
|
|
232 |
my $class = $self->{module} || $log->logconfess("can't get low-level module name!"); |
my $class = $self->{module} || $log->logconfess("can't get low-level module name!"); |
233 |
|
|
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 |
|
warn "-=> $f_nr ## |$l|\n" if ($debug); |
290 |
$log->debug("-=> $f_nr ## $l"); |
$log->debug("-=> $f_nr ## $l"); |
291 |
|
|
292 |
# codepage conversion and recode_regex |
# codepage conversion and recode_regex |
299 |
my $c = 0; |
my $c = 0; |
300 |
foreach my $r (@{ $rec_regex->{$f_nr} }) { |
foreach my $r (@{ $rec_regex->{$f_nr} }) { |
301 |
my $old_l = $l; |
my $old_l = $l; |
302 |
eval '$l =~ ' . $r; |
$log->logconfess("expected regex in ", dump( $r )) unless defined($r->{regex}); |
303 |
|
eval '$l =~ ' . $r->{regex}; |
304 |
if ($old_l ne $l) { |
if ($old_l ne $l) { |
305 |
$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}; |
306 |
|
$d .= ' +' . $r->{line} . ' ' . $r->{file} if defined($r->{line}); |
307 |
|
$d .= ' ' . $r->{debug} if defined($r->{debug}); |
308 |
|
$log->debug("MODIFY $d"); |
309 |
|
warn "*** $d\n" if ($debug); |
310 |
|
|
311 |
} |
} |
312 |
$log->error("error applying regex: $r") if ($@); |
$log->error("error applying regex: $r") if ($@); |
313 |
} |
} |
314 |
} |
} |
315 |
|
|
316 |
$log->debug("<=- $f_nr ## $l"); |
$log->debug("<=- $f_nr ## |$l|"); |
317 |
|
warn "<=- $f_nr ## $l\n" if ($debug); |
318 |
return $l; |
return $l; |
319 |
}); |
}); |
320 |
|
|
321 |
$log->debug(sub { Dumper($rec) }); |
$log->debug(sub { dump($rec) }); |
322 |
|
|
323 |
if (! $rec) { |
if (! $rec) { |
324 |
$log->warn("record $pos empty? skipping..."); |
$log->warn("record $pos empty? skipping..."); |
477 |
|
|
478 |
sub seek { |
sub seek { |
479 |
my $self = shift; |
my $self = shift; |
480 |
my $pos = shift || return; |
my $pos = shift; |
481 |
|
|
482 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
483 |
|
|
484 |
|
$log->logconfess("called without pos") unless defined($pos); |
485 |
|
|
486 |
if ($pos < 1) { |
if ($pos < 1) { |
487 |
$log->warn("seek before first record"); |
$log->warn("seek before first record"); |
488 |
$pos = 1; |
$pos = 1; |
540 |
} sort { $a cmp $b } keys %{ $s->{fld} } |
} sort { $a cmp $b } keys %{ $s->{fld} } |
541 |
); |
); |
542 |
|
|
543 |
$log->debug( sub { Dumper($s) } ); |
$log->debug( sub { dump($s) } ); |
544 |
|
|
545 |
return $out; |
return $out; |
546 |
} |
} |
547 |
|
|
548 |
=head2 dump |
=head2 dump_ascii |
549 |
|
|
550 |
Display humanly readable dump of record |
Display humanly readable dump of record |
551 |
|
|
552 |
=cut |
=cut |
553 |
|
|
554 |
sub dump { |
sub dump_ascii { |
555 |
my $self = shift; |
my $self = shift; |
556 |
|
|
557 |
return $self->{ll_db}->dump_rec( $self->{pos} ); |
return unless $self->{ll_db}; |
558 |
|
|
559 |
|
if ($self->{ll_db}->can('dump_rec')) { |
560 |
|
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
561 |
|
} else { |
562 |
|
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
563 |
|
} |
564 |
} |
} |
565 |
|
|
566 |
=head2 modify_record_regexps |
=head2 modify_record_regexps |
567 |
|
|
568 |
Generate hash with regexpes to be applied using l<filter>. |
Generate hash with regexpes to be applied using L<filter>. |
569 |
|
|
570 |
my $regexpes = $input->modify_record_regexps( |
my $regexpes = $input->modify_record_regexps( |
571 |
900 => { '^a' => { ' : ' => '^b' } }, |
900 => { '^a' => { ' : ' => '^b' } }, |
576 |
|
|
577 |
sub _get_regex { |
sub _get_regex { |
578 |
my ($sf,$from,$to) = @_; |
my ($sf,$from,$to) = @_; |
579 |
|
|
580 |
|
# protect / |
581 |
|
$from =~ s!/!\\/!gs; |
582 |
|
$to =~ s!/!\\/!gs; |
583 |
|
|
584 |
|
if ($from =~ m/^regex:(.+)$/) { |
585 |
|
$from = $1; |
586 |
|
} else { |
587 |
|
$from = '\Q' . $from . '\E'; |
588 |
|
} |
589 |
if ($sf =~ /^\^/) { |
if ($sf =~ /^\^/) { |
590 |
return |
return |
591 |
's/\Q'. $sf .'\E([^\^]*?)\Q'. $from .'\E([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
's/\Q'. $sf .'\E([^\^]*?)'. $from .'([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
592 |
} else { |
} else { |
593 |
return |
return |
594 |
's/\Q'. $from .'\E/'. $to .'/g'; |
's/'. $from .'/'. $to .'/g'; |
595 |
} |
} |
596 |
} |
} |
597 |
|
|
612 |
foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { |
foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { |
613 |
my $to = $modify_record->{$f}->{$sf}->{$from}; |
my $to = $modify_record->{$f}->{$sf}->{$from}; |
614 |
#die "no field?" unless defined($to); |
#die "no field?" unless defined($to); |
615 |
$log->debug("transform: |$from| -> |$to|"); |
my $d = "|$from| -> |$to|"; |
616 |
|
$log->debug("transform: $d"); |
617 |
|
|
618 |
my $regex = _get_regex($sf,$from,$to); |
my $regex = _get_regex($sf,$from,$to); |
619 |
push @{ $regexpes->{$f} }, $regex; |
push @{ $regexpes->{$f} }, { regex => $regex, debug => $d }; |
620 |
$log->debug("regex: $regex"); |
$log->debug("regex: $regex"); |
621 |
} |
} |
622 |
} |
} |
627 |
|
|
628 |
=head2 modify_file_regexps |
=head2 modify_file_regexps |
629 |
|
|
630 |
Generate hash with regexpes to be applied using l<filter> from |
Generate hash with regexpes to be applied using L<filter> from |
631 |
pseudo hash/yaml format for regex mappings. |
pseudo hash/yaml format for regex mappings. |
632 |
|
|
633 |
It should be obvious: |
It should be obvious: |
677 |
$log->debug("transform: |$from| -> |$to|"); |
$log->debug("transform: |$from| -> |$to|"); |
678 |
|
|
679 |
my $regex = _get_regex($sf,$from,$to); |
my $regex = _get_regex($sf,$from,$to); |
680 |
push @{ $regexpes->{$f} }, $regex; |
push @{ $regexpes->{$f} }, { |
681 |
|
regex => $regex, |
682 |
|
file => $modify_path, |
683 |
|
line => $., |
684 |
|
}; |
685 |
$log->debug("regex: $regex"); |
$log->debug("regex: $regex"); |
686 |
} |
} |
687 |
} |
} |