16 |
|
|
17 |
=head1 VERSION |
=head1 VERSION |
18 |
|
|
19 |
Version 0.11 |
Version 0.12 |
20 |
|
|
21 |
=cut |
=cut |
22 |
|
|
23 |
our $VERSION = '0.11'; |
our $VERSION = '0.12'; |
24 |
|
|
25 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
26 |
|
|
107 |
#eval $self->{module} .'->import'; |
#eval $self->{module} .'->import'; |
108 |
|
|
109 |
# check if required subclasses are implemented |
# check if required subclasses are implemented |
110 |
foreach my $subclass (qw/open_db fetch_rec init/) { |
foreach my $subclass (qw/open_db fetch_rec init dump_rec/) { |
111 |
my $n = $self->{module} . '::' . $subclass; |
my $n = $self->{module} . '::' . $subclass; |
112 |
if (! defined &{ $n }) { |
if (! defined &{ $n }) { |
113 |
my $missing = "missing $subclass in $self->{module}"; |
my $missing = "missing $subclass in $self->{module}"; |
172 |
900 => { '^a' => { ' : ' => '^b' } }, |
900 => { '^a' => { ' : ' => '^b' } }, |
173 |
901 => { '*' => { '^b' => ' ; ' } }, |
901 => { '*' => { '^b' => ' ; ' } }, |
174 |
}, |
}, |
175 |
|
modify_file => 'conf/modify/mapping.map', |
176 |
); |
); |
177 |
|
|
178 |
By default, C<code_page> is assumed to be C<cp852>. |
By default, C<code_page> is assumed to be C<cp852>. |
190 |
delimiters to subfields, as well as oprations on fields (if subfield is |
delimiters to subfields, as well as oprations on fields (if subfield is |
191 |
defined as C<*>. |
defined as C<*>. |
192 |
|
|
193 |
|
C<modify_file> is alternative for C<modify_records> above which preserves order and offers |
194 |
|
(hopefully) simplier sintax than YAML or perl (see L</modify_file_regex>). This option |
195 |
|
overrides C<modify_records> if both exists for same input. |
196 |
|
|
197 |
Returns size of database, regardless of C<offset> and C<limit> |
Returns size of database, regardless of C<offset> and C<limit> |
198 |
parametars, see also C<size>. |
parametars, see also C<size>. |
199 |
|
|
240 |
|
|
241 |
} |
} |
242 |
|
|
243 |
my $rec_regex = $self->modify_record_regexps(%{ $arg->{modify_records} }); |
my $rec_regex; |
244 |
$log->debug("rec_regex: ", Dumper($rec_regex)); |
if (my $p = $arg->{modify_file}) { |
245 |
|
$log->debug("using modify_file $p"); |
246 |
|
$rec_regex = $self->modify_file_regexps( $p ); |
247 |
|
} elsif (my $h = $arg->{modify_records}) { |
248 |
|
$log->debug("using modify_records ", Dumper( $h )); |
249 |
|
$rec_regex = $self->modify_record_regexps(%{ $h }); |
250 |
|
} |
251 |
|
$log->debug("rec_regex: ", Dumper($rec_regex)) if ($rec_regex); |
252 |
|
|
253 |
my ($db, $size) = $self->{open_db}->( $self, |
my ($db, $size) = $self->{open_db}->( $self, |
254 |
path => $arg->{path}, |
path => $arg->{path}, |
298 |
|
|
299 |
$log->debug("position: $pos\n"); |
$log->debug("position: $pos\n"); |
300 |
|
|
301 |
my $rec = $self->{fetch_rec}->($self, $db, $pos, sub { |
my $rec = $self->{fetch_rec}->($self, $pos, sub { |
302 |
my ($l,$f_nr) = @_; |
my ($l,$f_nr) = @_; |
303 |
# return unless defined($l); |
# return unless defined($l); |
304 |
# return $l unless ($rec_regex && $f_nr); |
# return $l unless ($rec_regex && $f_nr); |
348 |
if ($self->{stats}) { |
if ($self->{stats}) { |
349 |
|
|
350 |
# fetch clean record with regexpes applied for statistics |
# fetch clean record with regexpes applied for statistics |
351 |
my $rec = $self->{fetch_rec}->($self, $db, $pos); |
my $rec = $self->{fetch_rec}->($self, $pos); |
352 |
|
|
353 |
foreach my $fld (keys %{ $rec }) { |
foreach my $fld (keys %{ $rec }) { |
354 |
$self->{_stats}->{fld}->{ $fld }++; |
$self->{_stats}->{fld}->{ $fld }++; |
546 |
return $out; |
return $out; |
547 |
} |
} |
548 |
|
|
549 |
|
=head2 dump |
550 |
|
|
551 |
|
Display humanly readable dump of record |
552 |
|
|
553 |
|
=cut |
554 |
|
|
555 |
|
sub dump { |
556 |
|
my $self = shift; |
557 |
|
|
558 |
|
return $self->{dump_rec}->($self, $self->{pos}); |
559 |
|
|
560 |
|
} |
561 |
|
|
562 |
=head2 modify_record_regexps |
=head2 modify_record_regexps |
563 |
|
|
564 |
Generate hash with regexpes to be applied using L<filter>. |
Generate hash with regexpes to be applied using l<filter>. |
565 |
|
|
566 |
my $regexpes = $input->modify_record_regexps( |
my $regexpes = $input->modify_record_regexps( |
567 |
900 => { '^a' => { ' : ' => '^b' } }, |
900 => { '^a' => { ' : ' => '^b' } }, |
570 |
|
|
571 |
=cut |
=cut |
572 |
|
|
573 |
|
sub _get_regex { |
574 |
|
my ($sf,$from,$to) = @_; |
575 |
|
if ($sf =~ /^\^/) { |
576 |
|
return |
577 |
|
's/\Q'. $sf .'\E([^\^]*?)\Q'. $from .'\E([^\^]*?)/'. $sf .'$1'. $to .'$2/'; |
578 |
|
} else { |
579 |
|
return |
580 |
|
's/\Q'. $from .'\E/'. $to .'/g'; |
581 |
|
} |
582 |
|
} |
583 |
|
|
584 |
sub modify_record_regexps { |
sub modify_record_regexps { |
585 |
my $self = shift; |
my $self = shift; |
586 |
my $modify_record = {@_}; |
my $modify_record = {@_}; |
587 |
|
|
588 |
my $regexpes; |
my $regexpes; |
589 |
|
|
590 |
|
my $log = $self->_get_logger(); |
591 |
|
|
592 |
foreach my $f (keys %$modify_record) { |
foreach my $f (keys %$modify_record) { |
593 |
warn "--- f: $f\n"; |
$log->debug("field: $f"); |
594 |
|
|
595 |
foreach my $sf (keys %{ $modify_record->{$f} }) { |
foreach my $sf (keys %{ $modify_record->{$f} }) { |
596 |
warn "---- sf: $sf\n"; |
$log->debug("subfield: $sf"); |
597 |
|
|
598 |
foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { |
foreach my $from (keys %{ $modify_record->{$f}->{$sf} }) { |
599 |
my $to = $modify_record->{$f}->{$sf}->{$from}; |
my $to = $modify_record->{$f}->{$sf}->{$from}; |
600 |
#die "no field?" unless defined($to); |
#die "no field?" unless defined($to); |
601 |
warn "----- transform: |$from| -> |$to|\n"; |
$log->debug("transform: |$from| -> |$to|"); |
|
|
|
|
if ($sf =~ /^\^/) { |
|
|
my $regex = |
|
|
's/\Q'. $sf .'\E([^\^]+)\Q'. $from .'\E([^\^]+)/'. $sf .'$1'. $to .'$2/g'; |
|
|
push @{ $regexpes->{$f} }, $regex; |
|
|
warn ">>>>> $regex [sf]\n"; |
|
|
} else { |
|
|
my $regex = |
|
|
's/\Q'. $from .'\E/'. $to .'/g'; |
|
|
push @{ $regexpes->{$f} }, $regex; |
|
|
warn ">>>>> $regex [global]\n"; |
|
|
} |
|
602 |
|
|
603 |
|
my $regex = _get_regex($sf,$from,$to); |
604 |
|
push @{ $regexpes->{$f} }, $regex; |
605 |
|
$log->debug("regex: $regex"); |
606 |
} |
} |
607 |
} |
} |
608 |
} |
} |
609 |
|
|
610 |
|
return $regexpes; |
611 |
|
} |
612 |
|
|
613 |
|
=head2 modify_file_regexps |
614 |
|
|
615 |
|
Generate hash with regexpes to be applied using l<filter> from |
616 |
|
pseudo hash/yaml format for regex mappings. |
617 |
|
|
618 |
|
It should be obvious: |
619 |
|
|
620 |
|
200 |
621 |
|
'^a' |
622 |
|
' : ' => '^e' |
623 |
|
' = ' => '^d' |
624 |
|
|
625 |
|
In field I<200> find C<'^a'> and then C<' : '>, and replace it with C<'^e'>. |
626 |
|
In field I<200> find C<'^a'> and then C<' = '>, and replace it with C<'^d'>. |
627 |
|
|
628 |
|
my $regexpes = $input->modify_file_regexps( 'conf/modify/common.pl' ); |
629 |
|
|
630 |
|
On undef path it will just return. |
631 |
|
|
632 |
|
=cut |
633 |
|
|
634 |
|
sub modify_file_regexps { |
635 |
|
my $self = shift; |
636 |
|
|
637 |
|
my $modify_path = shift || return; |
638 |
|
|
639 |
|
my $log = $self->_get_logger(); |
640 |
|
|
641 |
|
my $regexpes; |
642 |
|
|
643 |
|
CORE::open(my $fh, $modify_path) || $log->die("can't open modify file $modify_path: $!"); |
644 |
|
|
645 |
|
my ($f,$sf); |
646 |
|
|
647 |
|
while(<$fh>) { |
648 |
|
chomp; |
649 |
|
next if (/^#/ || /^\s*$/); |
650 |
|
|
651 |
|
if (/^\s*(\d+)\s*$/) { |
652 |
|
$f = $1; |
653 |
|
$log->debug("field: $f"); |
654 |
|
next; |
655 |
|
} elsif (/^\s*'([^']*)'\s*$/) { |
656 |
|
$sf = $1; |
657 |
|
$log->die("can't define subfiled before field in: $_") unless ($f); |
658 |
|
$log->debug("subfield: $sf"); |
659 |
|
} elsif (/^\s*'([^']*)'\s*=>\s*'([^']*)'\s*$/) { |
660 |
|
my ($from,$to) = ($1, $2); |
661 |
|
|
662 |
|
$log->debug("transform: |$from| -> |$to|"); |
663 |
|
|
664 |
|
my $regex = _get_regex($sf,$from,$to); |
665 |
|
push @{ $regexpes->{$f} }, $regex; |
666 |
|
$log->debug("regex: $regex"); |
667 |
|
} |
668 |
|
} |
669 |
|
|
670 |
return $regexpes; |
return $regexpes; |
671 |
} |
} |
672 |
|
|