77 |
|
|
78 |
$self->{indexer_config_file} = new Config::IniFiles( -file => $self->{config_file} ) || croak "can't open '$self->{config_file}'"; |
$self->{indexer_config_file} = new Config::IniFiles( -file => $self->{config_file} ) || croak "can't open '$self->{config_file}'"; |
79 |
|
|
80 |
|
$self->{'utf2cp'} = Text::Iconv->new('UTF-8' ,$self->{'code_page'}); |
81 |
return $self; |
return $self; |
82 |
} |
} |
83 |
|
|
210 |
my $arg = {@_}; |
my $arg = {@_}; |
211 |
confess "need type to load file from import_xml/" if (! $arg->{'type'}); |
confess "need type to load file from import_xml/" if (! $arg->{'type'}); |
212 |
|
|
213 |
my $type = $arg->{'type'}; |
$self->{'type'} = $arg->{'type'}; |
214 |
|
|
215 |
my $type_base = $type; |
my $type_base = $arg->{'type'}; |
216 |
$type_base =~ s/_.*$//g; |
$type_base =~ s/_.*$//g; |
217 |
|
|
218 |
my $f = "./import_xml/$type.xml"; |
$self->{'tag'} = $type2tag{$type_base}; |
219 |
|
|
220 |
|
print STDERR "using type ",$self->{'type'}," tag ",$self->{'tag'},"\n" if ($self->{'debug'}); |
221 |
|
|
222 |
|
my $f = "./import_xml/".$self->{'type'}.".xml"; |
223 |
confess "import_xml file '$f' doesn't exist!" if (! -e "$f"); |
confess "import_xml file '$f' doesn't exist!" if (! -e "$f"); |
224 |
|
|
225 |
print STDERR "reading '$f'\n" if ($self->{'debug'}); |
print STDERR "reading '$f'\n" if ($self->{'debug'}); |
226 |
|
|
227 |
$self->{'import_xml'} = XMLin($f, |
$self->{'import_xml'} = XMLin($f, |
228 |
ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
229 |
ForceContent => 1 |
ForceContent => 1 |
230 |
); |
); |
231 |
|
|
292 |
return $$rec->{$f}->[$i]->{$sf}; |
return $$rec->{$f}->[$i]->{$sf}; |
293 |
} elsif ($$rec->{$f}->[$i]) { |
} elsif ($$rec->{$f}->[$i]) { |
294 |
$$found++ if (defined($$found)); |
$$found++ if (defined($$found)); |
295 |
return $$rec->{$f}->[$i]; |
# it still might have subfield, just |
296 |
|
# not specified, so we'll dump all |
297 |
|
if ($$rec->{$f}->[$i] =~ /HASH/o) { |
298 |
|
my $out; |
299 |
|
foreach my $k (keys %{$$rec->{$f}->[$i]}) { |
300 |
|
$out .= $$rec->{$f}->[$i]->{$k}." "; |
301 |
|
} |
302 |
|
return $out; |
303 |
|
} else { |
304 |
|
return $$rec->{$f}->[$i]; |
305 |
|
} |
306 |
} |
} |
307 |
} else { |
} else { |
308 |
return ''; |
return ''; |
422 |
sub parse { |
sub parse { |
423 |
my $self = shift; |
my $self = shift; |
424 |
|
|
425 |
my ($rec, $format, $i) = @_; |
my ($rec, $format_utf8, $i) = @_; |
426 |
|
|
427 |
|
return if (! $format_utf8); |
428 |
|
|
429 |
confess("need HASH as first argument!") if ($rec !~ /HASH/o); |
confess("need HASH as first argument!") if ($rec !~ /HASH/o); |
430 |
|
confess("need utf2cp Text::Iconv object!") if (! $self->{'utf2cp'}); |
431 |
|
|
432 |
$i = 0 if (! $i); |
$i = 0 if (! $i); |
433 |
|
|
434 |
|
my $format = $self->{'utf2cp'}->convert($format_utf8) || confess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'}); |
435 |
|
|
436 |
my @out; |
my @out; |
437 |
|
|
438 |
my $eval_code; |
my $eval_code; |
474 |
return $out; |
return $out; |
475 |
} |
} |
476 |
|
|
477 |
|
=head2 data_structure |
478 |
|
|
479 |
|
Create in-memory data structure which represents layout from C<import_xml>. |
480 |
|
It is used later to produce output. |
481 |
|
|
482 |
|
my $ds = $webpac->data_structure($rec); |
483 |
|
|
484 |
|
=cut |
485 |
|
|
486 |
|
# private method _sort_by_order |
487 |
|
# sort subrouting using order="" attribute |
488 |
|
sub _sort_by_order { |
489 |
|
my $self = shift; |
490 |
|
|
491 |
|
my $va = $self->{'import_xml'}->{'indexer'}->{$a}->{'order'} || |
492 |
|
$self->{'import_xml'}->{'indexer'}->{$a}; |
493 |
|
my $vb = $self->{'import_xml'}->{'indexer'}->{$b}->{'order'} || |
494 |
|
$self->{'import_xml'}->{'indexer'}->{$b}; |
495 |
|
|
496 |
|
return $va <=> $vb; |
497 |
|
} |
498 |
|
|
499 |
|
sub data_structure { |
500 |
|
my $self = shift; |
501 |
|
|
502 |
|
my $rec = shift; |
503 |
|
confess("need HASH as first argument!") if ($rec !~ /HASH/o); |
504 |
|
|
505 |
|
my @sorted_tags; |
506 |
|
if ($self->{tags_by_order}) { |
507 |
|
@sorted_tags = @{$self->{tags_by_order}}; |
508 |
|
} else { |
509 |
|
@sorted_tags = sort { $self->_sort_by_order } keys %{$self->{'import_xml'}->{'indexer'}}; |
510 |
|
$self->{tags_by_order} = \@sorted_tags; |
511 |
|
} |
512 |
|
|
513 |
|
my $ds; |
514 |
|
|
515 |
|
foreach my $field (@sorted_tags) { |
516 |
|
|
517 |
|
my $row; |
518 |
|
my $i = 0; |
519 |
|
|
520 |
|
#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}); |
521 |
|
|
522 |
|
foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) { |
523 |
|
|
524 |
|
my $v = $self->parse($rec,$tag->{'content'},$i); |
525 |
|
print "## $i:",$tag->{'content'}," = ",($v || 'null'),"\n"; |
526 |
|
|
527 |
|
next if (!$v || $v && $v eq ''); |
528 |
|
|
529 |
|
# does tag have type? |
530 |
|
if ($tag->{'type'}) { |
531 |
|
push @{$row->{$tag->{'type'}}}, $v; |
532 |
|
} else { |
533 |
|
push @{$row->{'display'}}, $v; |
534 |
|
push @{$row->{'swish'}}, $v; |
535 |
|
} |
536 |
|
} |
537 |
|
|
538 |
|
push @{$ds->{$field}}, $row if ($row); |
539 |
|
|
540 |
|
} |
541 |
|
|
542 |
|
print Dumper($ds); |
543 |
|
|
544 |
|
} |
545 |
|
|
546 |
1; |
1; |