210 |
|
|
211 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
212 |
|
|
213 |
|
sub chk_eval($) { |
214 |
|
my $data = shift; |
215 |
|
return if (! defined($data)); |
216 |
|
if ($data && $data =~ s/\s*eval{([^}]+)}\s*//) { |
217 |
|
if (eval "$1") { |
218 |
|
return $data; |
219 |
|
} else { |
220 |
|
return undef; |
221 |
|
} |
222 |
|
} else { |
223 |
|
return $data; |
224 |
|
} |
225 |
|
} |
226 |
|
|
227 |
sub mkformat($$) { |
sub mkformat($$) { |
228 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
229 |
my $data = shift || return; |
my $data = shift || return; |
230 |
my $format_name = x($x->{format_name}) || return $data; |
my $format_name = x($x->{format_name}) || return chk_eval($data); |
231 |
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
232 |
my $format_delimiter = x($x->{format_delimiter}); |
my $format_delimiter = x($x->{format_delimiter}); |
233 |
my @data; |
my @data; |
240 |
if ($fmt) { |
if ($fmt) { |
241 |
my $nr = scalar $fmt =~ s/%s/%s/g; |
my $nr = scalar $fmt =~ s/%s/%s/g; |
242 |
if (($#data+1) == $nr) { |
if (($#data+1) == $nr) { |
243 |
return sprintf($fmt,@data); |
return chk_eval(sprintf($fmt,@data)); |
244 |
} else { |
} else { |
245 |
print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
#print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
246 |
return $data; |
return chk_eval($data); |
247 |
} |
} |
248 |
} else { |
} else { |
249 |
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
271 |
$display = $new_display; |
$display = $new_display; |
272 |
$cache->{lhash}->{$display} = $new_display; |
$cache->{lhash}->{$display} = $new_display; |
273 |
} else { |
} else { |
274 |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
# print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
275 |
$display = ""; |
$display = ""; |
276 |
$cache->{lhash}->{$display} = $null; |
$cache->{lhash}->{$display} = $null; |
277 |
} |
} |
292 |
} |
} |
293 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
294 |
if ($swish) { |
if ($swish) { |
295 |
|
my $tmp = $swish; |
296 |
if ($filter && ($s || $se)) { |
if ($filter && ($s || $se)) { |
297 |
no strict 'refs'; |
no strict 'refs'; |
298 |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
$tmp = join(" ",&$filter($tmp)) if ($s || $se); |
|
$swish_data .= $tmp if ($s); |
|
|
if ($se) { |
|
|
if ($swish_exact_data) { |
|
|
$swish_exact_data .= "xxexx xxbxx ".$tmp; |
|
|
} else { |
|
|
$swish_exact_data .= $tmp; |
|
|
} |
|
|
} |
|
|
|
|
|
} else { |
|
|
$swish_data .= $swish if ($s); |
|
|
if ($se) { |
|
|
if ($swish_exact_data) { |
|
|
$swish_exact_data .= "xxexx xxbxx ".$swish; |
|
|
} else { |
|
|
$swish_exact_data .= $swish; |
|
|
} |
|
|
} |
|
299 |
} |
} |
300 |
|
|
301 |
|
$tmp = chk_eval($tmp); |
302 |
|
$swish_data .= $tmp if ($s && $tmp); |
303 |
|
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
304 |
} |
} |
305 |
|
|
306 |
# type="display" ; field for display |
# type="display" ; field for display |
307 |
if ($d && $display) { |
if ($d && $display) { |
308 |
|
my $ldel = $delimiter; |
309 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
310 |
$display_data .= $line_delimiter; |
$ldel = $line_delimiter; |
311 |
} |
} |
312 |
if ($filter) { |
if ($filter) { |
313 |
no strict 'refs'; |
no strict 'refs'; |
314 |
if ($display_data) { |
my @arr; |
315 |
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
foreach my $tmp (&$filter($display)) { |
316 |
} else { |
my $tmp2 = mkformat($x,$tmp); |
317 |
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
push @arr,$tmp2 if ($tmp2); |
318 |
} |
} |
319 |
|
$display_data .= $ldel if ($display_data && @arr); |
320 |
|
$display_data .= join($delimiter,@arr); |
321 |
} else { |
} else { |
322 |
if ($display_data) { |
$display_data .= $ldel if ($display_data); |
323 |
$display_data .= $delimiter.mkformat($x,$display); |
my $tmp = mkformat($x,$display); |
324 |
} else { |
$display_data .= $tmp if ($tmp); |
|
$display_data = mkformat($x,$display); |
|
|
} |
|
325 |
} |
} |
326 |
} |
} |
327 |
|
|
333 |
no strict 'refs'; |
no strict 'refs'; |
334 |
$idisplay = &$filter($idisplay); |
$idisplay = &$filter($idisplay); |
335 |
} |
} |
336 |
push @index_data, $idisplay if (! $iterate_by_page); |
$idisplay = chk_eval($idisplay); |
337 |
|
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
338 |
} |
} |
339 |
|
|
340 |
# store fields in lookup |
# store fields in lookup |
343 |
if ($lookup_key) { |
if ($lookup_key) { |
344 |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
345 |
} else { |
} else { |
346 |
$lookup_key = $display; |
if ($filter) { |
347 |
|
no strict 'refs'; |
348 |
|
$lookup_key = &$filter($display); |
349 |
|
} else { |
350 |
|
$lookup_key = $display; |
351 |
|
} |
352 |
} |
} |
353 |
} elsif (lc($x->{type}) eq "lookup_val") { |
} elsif (lc($x->{type}) eq "lookup_val") { |
354 |
if ($lookup_key) { |
if ($lookup_key) { |
355 |
$lhash{$lookup_key} = $display; |
if ($filter) { |
356 |
|
no strict 'refs'; |
357 |
|
$lhash{$lookup_key} = &$filter($display); |
358 |
|
} else { |
359 |
|
$lhash{$lookup_key} = $display; |
360 |
|
} |
361 |
} else { |
} else { |
362 |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
363 |
} |
} |
476 |
for (my $page=0; $page <= $nr_pages; $page++) { |
for (my $page=0; $page <= $nr_pages; $page++) { |
477 |
my $display_data; |
my $display_data; |
478 |
if ($cache->{format}->{$field}) { |
if ($cache->{format}->{$field}) { |
479 |
$display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
480 |
|
$display_data=$tmp if ($tmp); |
481 |
} else { |
} else { |
482 |
$display_data = $cache->{display_data}->{$field}->[$page]; |
$display_data = $cache->{display_data}->{$field}->[$page]; |
483 |
} |
} |
508 |
|
|
509 |
# add delimiters before and after word. |
# add delimiters before and after word. |
510 |
# That is required to produce exact match |
# That is required to produce exact match |
511 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
512 |
} |
} |
513 |
|
|
514 |
my $idel = $cache->{index_delimiter}->{$field}; |
my $idel = $cache->{index_delimiter}->{$field}; |
550 |
|
|
551 |
# add delimiters before and after word. |
# add delimiters before and after word. |
552 |
# That is required to produce exact match |
# That is required to produce exact match |
553 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
554 |
} |
} |
555 |
} |
} |
556 |
} |
} |
603 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
604 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
605 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
606 |
|
# delete memory cache for lookup file |
607 |
|
delete $cache->{lhash}; |
608 |
} |
} |
609 |
|
|
610 |
# open existing lookup file |
# open existing lookup file |
741 |
print STDERR "\n"; |
print STDERR "\n"; |
742 |
|
|
743 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
744 |
use Spreadsheet::ParseExcel; |
require Spreadsheet::ParseExcel; |
745 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
require Spreadsheet::ParseExcel::Utility; |
746 |
|
import Spreadsheet::ParseExcel::Utility qw(int2col); |
747 |
|
|
748 |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
749 |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
797 |
} |
} |
798 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
799 |
|
|
800 |
use MARC; |
require MARC; |
801 |
|
|
802 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
803 |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
804 |
|
|
805 |
# optional argument is format |
# optional argument is format |
806 |
my $format = x($config->{format}) || 'usmarc'; |
my $format = x($config->{marc_format}) || 'usmarc'; |
807 |
|
|
808 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
809 |
|
|
810 |
my $marc = new MARC; |
my $marc = new MARC; |
811 |
my $nr = $marc->openmarc({ |
my $nr = $marc->openmarc({ |
812 |
file=>$marc_file, format=>$format |
file=>$marc_file, format=>$format |
813 |
}) || die "Can't open MARC file '$marc_file'"; |
}) || die "Can't open MARC file '$marc_file' with format '$format'"; |
814 |
|
|
815 |
|
# read MARC file in memory |
816 |
|
$marc->nextmarc(-1); |
817 |
|
|
818 |
my $i=0; # record nr. |
my $max_rec = $marc->marc_count(); |
819 |
|
|
820 |
my $rec; |
for(my $i=1; $i<=$max_rec; $i++) { |
821 |
|
|
822 |
while ($marc->nextmarc(1)) { |
progress($i,$max_rec); |
823 |
|
|
824 |
# XXX |
# store value for marc_sf.pm |
825 |
fakeprogress($i++); |
$main::cache->{marc_record} = $i; |
826 |
|
|
827 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
828 |
|
|
834 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
835 |
} |
} |
836 |
} |
} |
837 |
|
|
838 |
|
print STDERR "\n"; |
839 |
|
|
840 |
} elsif ($type_base eq "feed") { |
} elsif ($type_base eq "feed") { |
841 |
|
|
842 |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |