188 |
|
|
189 |
push @{$self->{'data'}->{$mfn}->{$k}}, $val; |
push @{$self->{'data'}->{$mfn}->{$k}}, $val; |
190 |
} |
} |
191 |
|
} else { |
192 |
|
push @{$self->{'data'}->{$mfn}->{'000'}}, $mfn; |
193 |
} |
} |
194 |
|
|
195 |
} |
} |
254 |
|
|
255 |
$self->{'tag'} = $type2tag{$type_base}; |
$self->{'tag'} = $type2tag{$type_base}; |
256 |
|
|
257 |
$log->debug("using type '",$self->{'type'},"' tag <",$self->{'tag'},">") if ($self->{'debug'}); |
$log->info("using type '",$self->{'type'},"' tag <",$self->{'tag'},">"); |
258 |
|
|
259 |
my $f = "./import_xml/".$self->{'type'}.".xml"; |
my $f = "./import_xml/".$self->{'type'}.".xml"; |
260 |
$log->logconfess("import_xml file '$f' doesn't exist!") if (! -e "$f"); |
$log->logconfess("import_xml file '$f' doesn't exist!") if (! -e "$f"); |
261 |
|
|
262 |
$log->debug("reading '$f'") if ($self->{'debug'}); |
$log->info("reading '$f'"); |
263 |
|
|
264 |
|
$self->{'import_xml_file'} = $f; |
265 |
|
|
266 |
$self->{'import_xml'} = XMLin($f, |
$self->{'import_xml'} = XMLin($f, |
267 |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
268 |
); |
); |
269 |
|
|
270 |
|
$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }); |
271 |
|
|
272 |
} |
} |
273 |
|
|
274 |
=head2 create_lookup |
=head2 create_lookup |
375 |
This function B<does not> perform parsing of format to inteligenty skip |
This function B<does not> perform parsing of format to inteligenty skip |
376 |
delimiters before fields which aren't used. |
delimiters before fields which aren't used. |
377 |
|
|
378 |
|
This method will automatically decode UTF-8 string to local code page |
379 |
|
if needed. |
380 |
|
|
381 |
=cut |
=cut |
382 |
|
|
383 |
sub fill_in { |
sub fill_in { |
393 |
# FIXME remove for speedup? |
# FIXME remove for speedup? |
394 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
395 |
|
|
396 |
|
if (utf8::is_utf8($format)) { |
397 |
|
$format = $self->_x($format); |
398 |
|
} |
399 |
|
|
400 |
my $found = 0; |
my $found = 0; |
401 |
|
|
402 |
my $eval_code; |
my $eval_code; |
494 |
|
|
495 |
$i = 0 if (! $i); |
$i = 0 if (! $i); |
496 |
|
|
497 |
my $format = $self->{'utf2cp'}->convert($format_utf8) || $log->logconfess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'}); |
my $format = $self->_x($format_utf8) || $log->logconfess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'}); |
498 |
|
|
499 |
my @out; |
my @out; |
500 |
|
|
615 |
|
|
616 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
617 |
|
|
618 |
|
This method will also set C<$webpac->{'currnet_filename'}> if there is |
619 |
|
<filename> tag in C<import_xml>. |
620 |
|
|
621 |
=cut |
=cut |
622 |
|
|
623 |
sub data_structure { |
sub data_structure { |
628 |
my $rec = shift; |
my $rec = shift; |
629 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
630 |
|
|
631 |
|
undef $self->{'currnet_filename'}; |
632 |
|
|
633 |
my @sorted_tags; |
my @sorted_tags; |
634 |
if ($self->{tags_by_order}) { |
if ($self->{tags_by_order}) { |
635 |
@sorted_tags = @{$self->{tags_by_order}}; |
@sorted_tags = @{$self->{tags_by_order}}; |
661 |
} |
} |
662 |
next if (! @v); |
next if (! @v); |
663 |
|
|
664 |
|
# use format? |
665 |
|
if ($tag->{'format_name'}) { |
666 |
|
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
667 |
|
} |
668 |
|
|
669 |
# does tag have type? |
# does tag have type? |
670 |
if ($tag->{'type'}) { |
if ($tag->{'type'}) { |
671 |
push @{$row->{$tag->{'type'}}}, @v; |
push @{$row->{$tag->{'type'}}}, @v; |
674 |
push @{$row->{'swish'}}, @v; |
push @{$row->{'swish'}}, @v; |
675 |
} |
} |
676 |
|
|
677 |
|
if ($field eq 'filename') { |
678 |
|
$self->{'current_filename'} = join('',@v); |
679 |
|
$log->debug("filename: ",$self->{'current_filename'}); |
680 |
|
} |
681 |
|
|
682 |
} |
} |
683 |
|
|
684 |
if ($row) { |
if ($row) { |
685 |
$row->{'tag'} = $field; |
$row->{'tag'} = $field; |
686 |
|
|
687 |
|
# TODO: name_sigular, name_plural |
688 |
|
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
689 |
|
$row->{'name'} = $name ? $self->_x($name) : $field; |
690 |
|
|
691 |
push @ds, $row; |
push @ds, $row; |
692 |
|
|
693 |
$log->debug("row $field: ",sub { Dumper($row) }); |
$log->debug("row $field: ",sub { Dumper($row) }); |
694 |
} |
} |
695 |
|
|
728 |
return $out; |
return $out; |
729 |
} |
} |
730 |
|
|
731 |
|
=head2 apply_format |
732 |
|
|
733 |
|
Apply format specified in tag with C<format_name="name"> and |
734 |
|
C<format_delimiter=";;">. |
735 |
|
|
736 |
|
my $text = $webpac->apply_format($format_name,$format_delimiter,$data); |
737 |
|
|
738 |
|
Formats can contain C<lookup{...}> if you need them. |
739 |
|
|
740 |
|
=cut |
741 |
|
|
742 |
|
sub apply_format { |
743 |
|
my $self = shift; |
744 |
|
|
745 |
|
my ($name,$delimiter,$data) = @_; |
746 |
|
|
747 |
|
my $log = $self->_get_logger(); |
748 |
|
|
749 |
|
if (! $self->{'import_xml'}->{'format'}->{$name}) { |
750 |
|
$log->warn("<format name=\"$name\"> is not defined in ",$self->{'import_xml_file'}); |
751 |
|
return $data; |
752 |
|
} |
753 |
|
|
754 |
|
$log->warn("no delimiter for format $name") if (! $delimiter); |
755 |
|
|
756 |
|
my $format = $self->_x($self->{'import_xml'}->{'format'}->{$name}->{'content'}) || $log->logdie("can't find format '$name'"); |
757 |
|
|
758 |
|
my @data = split(/\Q$delimiter\E/, $data); |
759 |
|
|
760 |
|
my $out = sprintf($format, @data); |
761 |
|
$log->debug("using format $name [$format] on $data to produce: $out"); |
762 |
|
|
763 |
|
if ($out =~ m/$LOOKUP_REGEX/o) { |
764 |
|
return $self->lookup($out); |
765 |
|
} else { |
766 |
|
return $out; |
767 |
|
} |
768 |
|
|
769 |
|
} |
770 |
|
|
771 |
|
|
772 |
# |
# |
773 |
# |
# |
774 |
# |
# |
821 |
return $va <=> $vb; |
return $va <=> $vb; |
822 |
} |
} |
823 |
|
|
824 |
|
=head2 _get_logger |
825 |
|
|
826 |
|
Get C<Log::Log4perl> object with a twist: domains are defined for each |
827 |
|
method |
828 |
|
|
829 |
|
my $log = $webpac->_get_logger(); |
830 |
|
|
831 |
|
=cut |
832 |
|
|
833 |
sub _get_logger { |
sub _get_logger { |
834 |
my $self = shift; |
my $self = shift; |
835 |
|
|
836 |
my @c = caller(1); |
my $name = (caller(1))[3] || caller; |
837 |
return get_logger($c[3]); |
return get_logger($name); |
838 |
|
} |
839 |
|
|
840 |
|
=head2 _x |
841 |
|
|
842 |
|
Convert string from UTF-8 to code page defined in C<import_xml>. |
843 |
|
|
844 |
|
my $text = $webpac->_x('utf8 text'); |
845 |
|
|
846 |
|
=cut |
847 |
|
|
848 |
|
sub _x { |
849 |
|
my $self = shift; |
850 |
|
my $utf8 = shift || return; |
851 |
|
|
852 |
|
return $self->{'utf2cp'}->convert($utf8) || |
853 |
|
$self->_get_logger()->logwarn("can't convert '$utf8'"); |
854 |
} |
} |
855 |
|
|
856 |
# |
# |