34 |
|
|
35 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
36 |
config_file => 'name.conf', |
config_file => 'name.conf', |
37 |
[code_page => 'ISO-8859-2',] |
code_page => 'ISO-8859-2', |
38 |
[low_mem => 1,] |
low_mem => 1, |
39 |
|
filter => { |
40 |
|
'lower' => sub { lc($_[0]) }, |
41 |
|
}, |
42 |
); |
); |
43 |
|
|
44 |
Default C<code_page> is C<ISO-8859-2>. |
Default C<code_page> is C<ISO-8859-2>. |
45 |
|
|
46 |
Default is not to use C<low_mem> options (see L<MEMORY USAGE> below). |
Default is not to use C<low_mem> options (see L<MEMORY USAGE> below). |
47 |
|
|
48 |
|
There is optinal parametar C<filter> which specify different filters which |
49 |
|
can be applied using C<filter{name}> notation. |
50 |
|
|
51 |
This method will also read configuration files |
This method will also read configuration files |
52 |
C<global.conf> (used by indexer and Web font-end) |
C<global.conf> (used by indexer and Web font-end) |
53 |
and configuration file specified by C<config_file> |
and configuration file specified by C<config_file> |
129 |
$log->debug("removed '$db_file' from last run"); |
$log->debug("removed '$db_file' from last run"); |
130 |
} |
} |
131 |
|
|
132 |
use DBM::Deep; |
require DBM::Deep; |
133 |
|
|
134 |
my $db = new DBM::Deep $db_file; |
my $db = new DBM::Deep $db_file; |
135 |
|
|
144 |
$self->{'db'} = $db; |
$self->{'db'} = $db; |
145 |
} |
} |
146 |
|
|
147 |
|
$log->debug("filters defined: ",Dumper($self->{'filter'})); |
148 |
|
|
149 |
return $self; |
return $self; |
150 |
} |
} |
151 |
|
|
193 |
$log->logcroak("need filename") if (! $arg->{'filename'}); |
$log->logcroak("need filename") if (! $arg->{'filename'}); |
194 |
my $code_page = $arg->{'code_page'} || '852'; |
my $code_page = $arg->{'code_page'} || '852'; |
195 |
|
|
196 |
|
$log->logdie("can't find database ",$arg->{'filename'}) unless (glob($arg->{'filename'}.'.*')); |
197 |
|
|
198 |
# store data in object |
# store data in object |
199 |
$self->{'isis_filename'} = $arg->{'filename'}; |
$self->{'isis_filename'} = $arg->{'filename'}; |
200 |
$self->{'isis_code_page'} = $code_page; |
$self->{'isis_code_page'} = $code_page; |
217 |
if (my $s = $self->{'start_mfn'}) { |
if (my $s = $self->{'start_mfn'}) { |
218 |
$log->info("skipping to MFN $s"); |
$log->info("skipping to MFN $s"); |
219 |
$startmfn = $s; |
$startmfn = $s; |
220 |
|
} else { |
221 |
|
$self->{'start_mfn'} = $startmfn; |
222 |
} |
} |
223 |
|
|
224 |
$maxmfn = $startmfn + $self->{limit_mfn} if ($self->{limit_mfn}); |
$maxmfn = $startmfn + $self->{limit_mfn} if ($self->{limit_mfn}); |
274 |
|
|
275 |
} |
} |
276 |
|
|
277 |
$self->{'current_mfn'} = $startmfn; |
$self->{'current_mfn'} = -1; |
278 |
$self->{'last_pcnt'} = 0; |
$self->{'last_pcnt'} = 0; |
279 |
|
|
280 |
$log->debug("max mfn: $maxmfn"); |
$log->debug("max mfn: $maxmfn"); |
297 |
|
|
298 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
299 |
|
|
300 |
my $mfn = $self->{'current_mfn'}++ || $log->logconfess("it seems that you didn't load database!"); |
$log->logconfess("it seems that you didn't load database!") unless ($self->{'current_mfn'}); |
301 |
|
|
302 |
|
if ($self->{'current_mfn'} == -1) { |
303 |
|
$self->{'current_mfn'} = $self->{'start_mfn'}; |
304 |
|
} else { |
305 |
|
$self->{'current_mfn'}++; |
306 |
|
} |
307 |
|
|
308 |
|
my $mfn = $self->{'current_mfn'}; |
309 |
|
|
310 |
if ($mfn > $self->{'max_mfn'}) { |
if ($mfn > $self->{'max_mfn'}) { |
311 |
$self->{'current_mfn'} = $self->{'max_mfn'}; |
$self->{'current_mfn'} = $self->{'max_mfn'}; |
322 |
} |
} |
323 |
} |
} |
324 |
|
|
325 |
|
=head2 mfn |
326 |
|
|
327 |
|
Returns current record number (MFN). |
328 |
|
|
329 |
|
print $webpac->mfn; |
330 |
|
|
331 |
|
=cut |
332 |
|
|
333 |
|
sub mfn { |
334 |
|
my $self = shift; |
335 |
|
return $self->{'current_mfn'}; |
336 |
|
} |
337 |
|
|
338 |
=head2 progress_bar |
=head2 progress_bar |
339 |
|
|
340 |
Draw progress bar on STDERR. |
Draw progress bar on STDERR. |
360 |
|
|
361 |
$self->{'last_pcnt'} ||= 1; |
$self->{'last_pcnt'} ||= 1; |
362 |
|
|
363 |
my $p = int($curr * 100 / $max); |
my $p = int($curr * 100 / $max) || 1; |
364 |
|
|
365 |
# reset on re-run |
# reset on re-run |
366 |
if ($p < $self->{'last_pcnt'}) { |
if ($p < $self->{'last_pcnt'}) { |
367 |
$self->{'last_pcnt'} = $p; |
$self->{'last_pcnt'} = $p; |
368 |
$self->{'last_t'} = time(); |
$self->{'last_t'} = time(); |
369 |
$self->{'last_curr'} = 1; |
$self->{'last_curr'} = undef; |
370 |
} |
} |
371 |
|
|
372 |
|
$self->{'last_t'} ||= time(); |
373 |
|
|
374 |
if ($p != $self->{'last_pcnt'}) { |
if ($p != $self->{'last_pcnt'}) { |
375 |
|
|
376 |
my $last_curr = $self->{'last_curr'} || $curr; |
my $last_curr = $self->{'last_curr'} || $curr; |
583 |
# remove eval{...} from beginning |
# remove eval{...} from beginning |
584 |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
585 |
|
|
586 |
|
my $filter_name; |
587 |
|
# remove filter{...} from beginning |
588 |
|
$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s); |
589 |
|
|
590 |
# do actual replacement of placeholders |
# do actual replacement of placeholders |
591 |
$format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges; |
$format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges; |
592 |
|
|
596 |
my $eval = $self->fill_in($rec,$eval_code,$i); |
my $eval = $self->fill_in($rec,$eval_code,$i); |
597 |
return if (! $self->_eval($eval)); |
return if (! $self->_eval($eval)); |
598 |
} |
} |
599 |
|
if ($filter_name && $self->{'filter'}->{$filter_name}) { |
600 |
|
$log->debug("filter '$filter_name' for $format"); |
601 |
|
$format = $self->{'filter'}->{$filter_name}->($format); |
602 |
|
return unless(defined($format)); |
603 |
|
$log->debug("filter result: $format"); |
604 |
|
} |
605 |
# do we have lookups? |
# do we have lookups? |
606 |
if ($format =~ /$LOOKUP_REGEX/o) { |
if ($format =~ /$LOOKUP_REGEX/o) { |
607 |
$log->debug("format '$format' has lookup"); |
$log->debug("format '$format' has lookup"); |
694 |
# remove eval{...} from beginning |
# remove eval{...} from beginning |
695 |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
696 |
|
|
697 |
|
my $filter_name; |
698 |
|
# remove filter{...} from beginning |
699 |
|
$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s); |
700 |
|
|
701 |
my $prefix; |
my $prefix; |
702 |
my $all_found=0; |
my $all_found=0; |
703 |
|
|
732 |
|
|
733 |
if ($eval_code) { |
if ($eval_code) { |
734 |
my $eval = $self->fill_in($rec,$eval_code,$i); |
my $eval = $self->fill_in($rec,$eval_code,$i); |
735 |
$log->debug("about to eval{",$eval,"} format: $out"); |
$log->debug("about to eval{$eval} format: $out"); |
736 |
return if (! $self->_eval($eval)); |
return if (! $self->_eval($eval)); |
737 |
} |
} |
738 |
|
|
739 |
|
if ($filter_name && $self->{'filter'}->{$filter_name}) { |
740 |
|
$log->debug("about to filter{$filter_name} format: $out"); |
741 |
|
$out = $self->{'filter'}->{$filter_name}->($out); |
742 |
|
return unless(defined($out)); |
743 |
|
$log->debug("filter result: $out"); |
744 |
|
} |
745 |
|
|
746 |
return $out; |
return $out; |
747 |
} |
} |
808 |
return @arr; |
return @arr; |
809 |
} |
} |
810 |
|
|
811 |
|
=head2 sort_arr |
812 |
|
|
813 |
|
Sort array ignoring case and html in data |
814 |
|
|
815 |
|
my @sorted = $webpac->sort_arr(@unsorted); |
816 |
|
|
817 |
|
=cut |
818 |
|
|
819 |
|
sub sort_arr { |
820 |
|
my $self = shift; |
821 |
|
|
822 |
|
my $log = $self->_get_logger(); |
823 |
|
|
824 |
|
# FIXME add Schwartzian Transformation? |
825 |
|
|
826 |
|
my @sorted = sort { |
827 |
|
$a =~ s#<[^>]+/*>##; |
828 |
|
$b =~ s#<[^>]+/*>##; |
829 |
|
lc($b) cmp lc($a) |
830 |
|
} @_; |
831 |
|
$log->debug("sorted values: ",sub { join(", ",@sorted) }); |
832 |
|
|
833 |
|
return @sorted; |
834 |
|
} |
835 |
|
|
836 |
|
|
837 |
=head2 data_structure |
=head2 data_structure |
838 |
|
|
889 |
} |
} |
890 |
next if (! @v); |
next if (! @v); |
891 |
|
|
892 |
|
if ($tag->{'sort'}) { |
893 |
|
@v = $self->sort_arr(@v); |
894 |
|
$log->warn("sort within tag is usually not what you want!"); |
895 |
|
} |
896 |
|
|
897 |
# use format? |
# use format? |
898 |
if ($tag->{'format_name'}) { |
if ($tag->{'format_name'}) { |
899 |
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
908 |
next; # don't return headline in data_structure! |
next; # don't return headline in data_structure! |
909 |
} |
} |
910 |
|
|
911 |
# does tag have type? |
# delimiter will join repeatable fields |
912 |
if ($tag->{'type'}) { |
if ($tag->{'delimiter'}) { |
913 |
push @{$row->{$tag->{'type'}}}, @v; |
@v = ( join($tag->{'delimiter'}, @v) ); |
914 |
} else { |
} |
915 |
push @{$row->{'display'}}, @v; |
|
916 |
push @{$row->{'swish'}}, @v; |
# default types |
917 |
|
my @types = qw(display swish); |
918 |
|
# override by type attribute |
919 |
|
@types = ( $tag->{'type'} ) if ($tag->{'type'}); |
920 |
|
|
921 |
|
foreach my $type (@types) { |
922 |
|
# append to previous line? |
923 |
|
$log->debug("type: $type ",sub { join(" ",@v) }, $row->{'append'} || 'no append'); |
924 |
|
if ($tag->{'append'}) { |
925 |
|
|
926 |
|
# I will delimit appended part with |
927 |
|
# delimiter (or ,) |
928 |
|
my $d = $tag->{'delimiter'}; |
929 |
|
# default delimiter |
930 |
|
$d ||= ", "; |
931 |
|
|
932 |
|
my $last = pop @{$row->{$type}}; |
933 |
|
$d = "" if (! $last); |
934 |
|
$last .= $d . join($d, @v); |
935 |
|
push @{$row->{$type}}, $last; |
936 |
|
|
937 |
|
} else { |
938 |
|
push @{$row->{$type}}, @v; |
939 |
|
} |
940 |
} |
} |
941 |
|
|
942 |
|
|
949 |
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
950 |
$row->{'name'} = $name ? $self->_x($name) : $field; |
$row->{'name'} = $name ? $self->_x($name) : $field; |
951 |
|
|
952 |
|
# post-sort all values in field |
953 |
|
if ($self->{'import_xml'}->{'indexer'}->{$field}->{'sort'}) { |
954 |
|
$log->warn("sort at field tag not implemented"); |
955 |
|
} |
956 |
|
|
957 |
push @ds, $row; |
push @ds, $row; |
958 |
|
|
959 |
$log->debug("row $field: ",sub { Dumper($row) }); |
$log->debug("row $field: ",sub { Dumper($row) }); |
1099 |
|
|
1100 |
$log->debug("eval: ",$code," [",$ret,"]"); |
$log->debug("eval: ",$code," [",$ret,"]"); |
1101 |
|
|
1102 |
return $ret || 0; |
return $ret || undef; |
1103 |
} |
} |
1104 |
|
|
1105 |
=head2 _sort_by_order |
=head2 _sort_by_order |