198 |
my $rec = $self->{'data'}->{$mfn}; |
my $rec = $self->{'data'}->{$mfn}; |
199 |
$self->create_lookup($rec, @{$arg->{'lookup'}}); |
$self->create_lookup($rec, @{$arg->{'lookup'}}); |
200 |
|
|
201 |
|
$self->progress_bar($mfn,$maxmfn); |
202 |
|
|
203 |
} |
} |
204 |
|
|
205 |
$self->{'current_mfn'} = 1; |
$self->{'current_mfn'} = 1; |
206 |
|
$self->{'last_pcnt'} = 0; |
207 |
|
|
208 |
# store max mfn and return it. |
# store max mfn and return it. |
209 |
return $self->{'max_mfn'} = $maxmfn; |
return $self->{'max_mfn'} = $maxmfn; |
231 |
return; |
return; |
232 |
} |
} |
233 |
|
|
234 |
|
$self->progress_bar($mfn,$self->{'max_mfn'}); |
235 |
|
|
236 |
return $self->{'data'}->{$mfn}; |
return $self->{'data'}->{$mfn}; |
237 |
} |
} |
238 |
|
|
239 |
|
=head2 progress_bar |
240 |
|
|
241 |
|
Draw progress bar on STDERR. |
242 |
|
|
243 |
|
$webpac->progress_bar($current, $max); |
244 |
|
|
245 |
|
=cut |
246 |
|
|
247 |
|
sub progress_bar { |
248 |
|
my $self = shift; |
249 |
|
|
250 |
|
my ($curr,$max) = @_; |
251 |
|
|
252 |
|
my $log = $self->_get_logger(); |
253 |
|
|
254 |
|
$log->logconfess("no current value!") if (! $curr); |
255 |
|
$log->logconfess("no maximum value!") if (! $max); |
256 |
|
|
257 |
|
if ($curr > $max) { |
258 |
|
$max = $curr; |
259 |
|
$log->debug("overflow to $curr"); |
260 |
|
} |
261 |
|
|
262 |
|
$self->{'last_pcnt'} ||= 1; |
263 |
|
|
264 |
|
$self->{'last_pcnt'} = $curr if ($curr < $self->{'last_pcnt'}); |
265 |
|
|
266 |
|
my $p = int($curr * 100 / $max); |
267 |
|
if ($p != $self->{'last_pcnt'}) { |
268 |
|
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$curr,$max,"=" x ($p/2).">", $p ); |
269 |
|
$self->{'last_pcnt'} = $p; |
270 |
|
} |
271 |
|
} |
272 |
|
|
273 |
=head2 open_import_xml |
=head2 open_import_xml |
274 |
|
|
275 |
Read file from C<import_xml/> directory and parse it. |
Read file from C<import_xml/> directory and parse it. |
293 |
|
|
294 |
$self->{'tag'} = $type2tag{$type_base}; |
$self->{'tag'} = $type2tag{$type_base}; |
295 |
|
|
296 |
$log->debug("using type '",$self->{'type'},"' tag <",$self->{'tag'},">") if ($self->{'debug'}); |
$log->info("using type '",$self->{'type'},"' tag <",$self->{'tag'},">"); |
297 |
|
|
298 |
my $f = "./import_xml/".$self->{'type'}.".xml"; |
my $f = "./import_xml/".$self->{'type'}.".xml"; |
299 |
$log->logconfess("import_xml file '$f' doesn't exist!") if (! -e "$f"); |
$log->logconfess("import_xml file '$f' doesn't exist!") if (! -e "$f"); |
300 |
|
|
301 |
$log->debug("reading '$f'") if ($self->{'debug'}); |
$log->info("reading '$f'"); |
302 |
|
|
303 |
|
$self->{'import_xml_file'} = $f; |
304 |
|
|
305 |
$self->{'import_xml'} = XMLin($f, |
$self->{'import_xml'} = XMLin($f, |
306 |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
307 |
); |
); |
308 |
|
|
309 |
|
$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }); |
310 |
|
|
311 |
} |
} |
312 |
|
|
313 |
=head2 create_lookup |
=head2 create_lookup |
374 |
|
|
375 |
if ($$rec->{$f}) { |
if ($$rec->{$f}) { |
376 |
return '' if (! $$rec->{$f}->[$i]); |
return '' if (! $$rec->{$f}->[$i]); |
377 |
|
no strict 'refs'; |
378 |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
379 |
$$found++ if (defined($$found)); |
$$found++ if (defined($$found)); |
380 |
return $$rec->{$f}->[$i]->{$sf}; |
return $$rec->{$f}->[$i]->{$sf}; |
415 |
This function B<does not> perform parsing of format to inteligenty skip |
This function B<does not> perform parsing of format to inteligenty skip |
416 |
delimiters before fields which aren't used. |
delimiters before fields which aren't used. |
417 |
|
|
418 |
|
This method will automatically decode UTF-8 string to local code page |
419 |
|
if needed. |
420 |
|
|
421 |
=cut |
=cut |
422 |
|
|
423 |
sub fill_in { |
sub fill_in { |
433 |
# FIXME remove for speedup? |
# FIXME remove for speedup? |
434 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
435 |
|
|
436 |
|
if (utf8::is_utf8($format)) { |
437 |
|
$format = $self->_x($format); |
438 |
|
} |
439 |
|
|
440 |
my $found = 0; |
my $found = 0; |
441 |
|
|
442 |
my $eval_code; |
my $eval_code; |
534 |
|
|
535 |
$i = 0 if (! $i); |
$i = 0 if (! $i); |
536 |
|
|
537 |
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'}); |
538 |
|
|
539 |
my @out; |
my @out; |
540 |
|
|
656 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
657 |
|
|
658 |
This method will also set C<$webpac->{'currnet_filename'}> if there is |
This method will also set C<$webpac->{'currnet_filename'}> if there is |
659 |
<filename> tag in C<import_xml>. |
<filename> tag in C<import_xml> and C<$webpac->{'headline'}> if there is |
660 |
|
<headline> tag. |
661 |
|
|
662 |
=cut |
=cut |
663 |
|
|
670 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
671 |
|
|
672 |
undef $self->{'currnet_filename'}; |
undef $self->{'currnet_filename'}; |
673 |
|
undef $self->{'headline'}; |
674 |
|
|
675 |
my @sorted_tags; |
my @sorted_tags; |
676 |
if ($self->{tags_by_order}) { |
if ($self->{tags_by_order}) { |
703 |
} |
} |
704 |
next if (! @v); |
next if (! @v); |
705 |
|
|
706 |
|
# use format? |
707 |
|
if ($tag->{'format_name'}) { |
708 |
|
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
709 |
|
} |
710 |
|
|
711 |
|
if ($field eq 'filename') { |
712 |
|
$self->{'current_filename'} = join('',@v); |
713 |
|
$log->debug("filename: ",$self->{'current_filename'}); |
714 |
|
} elsif ($field eq 'headline') { |
715 |
|
$self->{'headline'} .= join('',@v); |
716 |
|
$log->debug("headline: ",$self->{'headline'}); |
717 |
|
next; # don't return headline in data_structure! |
718 |
|
} |
719 |
|
|
720 |
# does tag have type? |
# does tag have type? |
721 |
if ($tag->{'type'}) { |
if ($tag->{'type'}) { |
722 |
push @{$row->{$tag->{'type'}}}, @v; |
push @{$row->{$tag->{'type'}}}, @v; |
725 |
push @{$row->{'swish'}}, @v; |
push @{$row->{'swish'}}, @v; |
726 |
} |
} |
727 |
|
|
|
if ($field eq 'filename') { |
|
|
$self->{'current_filename'} = join('',@v); |
|
|
$log->debug("filename: ",$self->{'current_filename'}); |
|
|
} |
|
728 |
|
|
729 |
} |
} |
730 |
|
|
731 |
if ($row) { |
if ($row) { |
732 |
$row->{'tag'} = $field; |
$row->{'tag'} = $field; |
733 |
|
|
734 |
|
# TODO: name_sigular, name_plural |
735 |
|
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
736 |
|
$row->{'name'} = $name ? $self->_x($name) : $field; |
737 |
|
|
738 |
push @ds, $row; |
push @ds, $row; |
739 |
|
|
740 |
$log->debug("row $field: ",sub { Dumper($row) }); |
$log->debug("row $field: ",sub { Dumper($row) }); |
775 |
return $out; |
return $out; |
776 |
} |
} |
777 |
|
|
778 |
|
=head2 output_file |
779 |
|
|
780 |
|
Create output from in-memory data structure using Template Toolkit template |
781 |
|
to a file. |
782 |
|
|
783 |
|
$webpac->output_file( |
784 |
|
file => 'out.txt', |
785 |
|
template => 'text.tt', |
786 |
|
data => @ds |
787 |
|
); |
788 |
|
|
789 |
|
=cut |
790 |
|
|
791 |
|
sub output_file { |
792 |
|
my $self = shift; |
793 |
|
|
794 |
|
my $args = {@_}; |
795 |
|
|
796 |
|
my $log = $self->_get_logger(); |
797 |
|
|
798 |
|
$log->logconfess("need file name") if (! $args->{'file'}); |
799 |
|
|
800 |
|
$log->debug("creating file ",$args->{'file'}); |
801 |
|
|
802 |
|
open(my $fh, ">", $args->{'file'}) || $log->logdie("can't open output file '$self->{'file'}': $!"); |
803 |
|
print $fh $self->output( |
804 |
|
template => $args->{'template'}, |
805 |
|
data => $args->{'data'}, |
806 |
|
) || $log->logdie("print: $!"); |
807 |
|
close($fh) || $log->logdie("close: $!"); |
808 |
|
} |
809 |
|
|
810 |
|
=head2 apply_format |
811 |
|
|
812 |
|
Apply format specified in tag with C<format_name="name"> and |
813 |
|
C<format_delimiter=";;">. |
814 |
|
|
815 |
|
my $text = $webpac->apply_format($format_name,$format_delimiter,$data); |
816 |
|
|
817 |
|
Formats can contain C<lookup{...}> if you need them. |
818 |
|
|
819 |
|
=cut |
820 |
|
|
821 |
|
sub apply_format { |
822 |
|
my $self = shift; |
823 |
|
|
824 |
|
my ($name,$delimiter,$data) = @_; |
825 |
|
|
826 |
|
my $log = $self->_get_logger(); |
827 |
|
|
828 |
|
if (! $self->{'import_xml'}->{'format'}->{$name}) { |
829 |
|
$log->warn("<format name=\"$name\"> is not defined in ",$self->{'import_xml_file'}); |
830 |
|
return $data; |
831 |
|
} |
832 |
|
|
833 |
|
$log->warn("no delimiter for format $name") if (! $delimiter); |
834 |
|
|
835 |
|
my $format = $self->_x($self->{'import_xml'}->{'format'}->{$name}->{'content'}) || $log->logdie("can't find format '$name'"); |
836 |
|
|
837 |
|
my @data = split(/\Q$delimiter\E/, $data); |
838 |
|
|
839 |
|
my $out = sprintf($format, @data); |
840 |
|
$log->debug("using format $name [$format] on $data to produce: $out"); |
841 |
|
|
842 |
|
if ($out =~ m/$LOOKUP_REGEX/o) { |
843 |
|
return $self->lookup($out); |
844 |
|
} else { |
845 |
|
return $out; |
846 |
|
} |
847 |
|
|
848 |
|
} |
849 |
|
|
850 |
|
|
851 |
# |
# |
852 |
# |
# |
853 |
# |
# |
900 |
return $va <=> $vb; |
return $va <=> $vb; |
901 |
} |
} |
902 |
|
|
903 |
|
=head2 _get_logger |
904 |
|
|
905 |
|
Get C<Log::Log4perl> object with a twist: domains are defined for each |
906 |
|
method |
907 |
|
|
908 |
|
my $log = $webpac->_get_logger(); |
909 |
|
|
910 |
|
=cut |
911 |
|
|
912 |
sub _get_logger { |
sub _get_logger { |
913 |
my $self = shift; |
my $self = shift; |
914 |
|
|
916 |
return get_logger($name); |
return get_logger($name); |
917 |
} |
} |
918 |
|
|
919 |
|
=head2 _x |
920 |
|
|
921 |
|
Convert string from UTF-8 to code page defined in C<import_xml>. |
922 |
|
|
923 |
|
my $text = $webpac->_x('utf8 text'); |
924 |
|
|
925 |
|
=cut |
926 |
|
|
927 |
|
sub _x { |
928 |
|
my $self = shift; |
929 |
|
my $utf8 = shift || return; |
930 |
|
|
931 |
|
return $self->{'utf2cp'}->convert($utf8) || |
932 |
|
$self->_get_logger()->logwarn("can't convert '$utf8'"); |
933 |
|
} |
934 |
|
|
935 |
# |
# |
936 |
# |
# |
937 |
# |
# |