228 |
if (($#data+1) == $nr) { |
if (($#data+1) == $nr) { |
229 |
return sprintf($fmt,@data); |
return sprintf($fmt,@data); |
230 |
} else { |
} else { |
231 |
print STDERR "mkformat: [$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"; |
232 |
return $data; |
return $data; |
233 |
} |
} |
234 |
} else { |
} else { |
257 |
$display = $new_display; |
$display = $new_display; |
258 |
$cache->{lhash}->{$display} = $new_display; |
$cache->{lhash}->{$display} = $new_display; |
259 |
} else { |
} else { |
260 |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
# print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
261 |
$display = ""; |
$display = ""; |
262 |
$cache->{lhash}->{$display} = $null; |
$cache->{lhash}->{$display} = $null; |
263 |
} |
} |
278 |
} |
} |
279 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
280 |
if ($swish) { |
if ($swish) { |
281 |
|
my $tmp = $swish; |
282 |
if ($filter && ($s || $se)) { |
if ($filter && ($s || $se)) { |
283 |
no strict 'refs'; |
no strict 'refs'; |
284 |
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; |
|
|
} |
|
|
} |
|
285 |
} |
} |
286 |
|
|
287 |
|
$swish_data .= $tmp if ($s && $tmp); |
288 |
|
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
289 |
} |
} |
290 |
|
|
291 |
# type="display" ; field for display |
# type="display" ; field for display |
292 |
if ($d && $display) { |
if ($d && $display) { |
293 |
|
my $ldel = $delimiter; |
294 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
295 |
$display_data .= $line_delimiter; |
$ldel = $line_delimiter; |
296 |
} |
} |
297 |
if ($filter) { |
if ($filter) { |
298 |
no strict 'refs'; |
no strict 'refs'; |
299 |
if ($display_data) { |
my @arr; |
300 |
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
foreach my $tmp (&$filter($display)) { |
301 |
} else { |
my $tmp2 = mkformat($x,$tmp); |
302 |
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
push @arr,$tmp2 if ($tmp2); |
303 |
} |
} |
304 |
|
$display_data .= $ldel if ($display_data && @arr); |
305 |
|
$display_data .= join($delimiter,@arr); |
306 |
} else { |
} else { |
307 |
if ($display_data) { |
$display_data .= $ldel if ($display_data); |
308 |
$display_data .= $delimiter.mkformat($x,$display); |
my $tmp = mkformat($x,$display); |
309 |
} else { |
$display_data .= $tmp if ($tmp); |
|
$display_data = mkformat($x,$display); |
|
|
} |
|
310 |
} |
} |
311 |
} |
} |
312 |
|
|
318 |
no strict 'refs'; |
no strict 'refs'; |
319 |
$idisplay = &$filter($idisplay); |
$idisplay = &$filter($idisplay); |
320 |
} |
} |
321 |
push @index_data, $idisplay if (! $iterate_by_page); |
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
322 |
} |
} |
323 |
|
|
324 |
# store fields in lookup |
# store fields in lookup |
327 |
if ($lookup_key) { |
if ($lookup_key) { |
328 |
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!)"; |
329 |
} else { |
} else { |
330 |
$lookup_key = $display; |
if ($filter) { |
331 |
|
no strict 'refs'; |
332 |
|
$lookup_key = &$filter($display); |
333 |
|
} else { |
334 |
|
$lookup_key = $display; |
335 |
|
} |
336 |
} |
} |
337 |
} elsif (lc($x->{type}) eq "lookup_val") { |
} elsif (lc($x->{type}) eq "lookup_val") { |
338 |
if ($lookup_key) { |
if ($lookup_key) { |
339 |
$lhash{$lookup_key} = $display; |
if ($filter) { |
340 |
|
no strict 'refs'; |
341 |
|
$lhash{$lookup_key} = &$filter($display); |
342 |
|
} else { |
343 |
|
$lhash{$lookup_key} = $display; |
344 |
|
} |
345 |
} else { |
} else { |
346 |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
347 |
} |
} |
460 |
for (my $page=0; $page <= $nr_pages; $page++) { |
for (my $page=0; $page <= $nr_pages; $page++) { |
461 |
my $display_data; |
my $display_data; |
462 |
if ($cache->{format}->{$field}) { |
if ($cache->{format}->{$field}) { |
463 |
$display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
464 |
|
$display_data=$tmp if ($tmp); |
465 |
} else { |
} else { |
466 |
$display_data = $cache->{display_data}->{$field}->[$page]; |
$display_data = $cache->{display_data}->{$field}->[$page]; |
467 |
} |
} |
492 |
|
|
493 |
# add delimiters before and after word. |
# add delimiters before and after word. |
494 |
# That is required to produce exact match |
# That is required to produce exact match |
495 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
496 |
} |
} |
497 |
|
|
498 |
my $idel = $cache->{index_delimiter}->{$field}; |
my $idel = $cache->{index_delimiter}->{$field}; |
534 |
|
|
535 |
# add delimiters before and after word. |
# add delimiters before and after word. |
536 |
# That is required to produce exact match |
# That is required to produce exact match |
537 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
538 |
} |
} |
539 |
} |
} |
540 |
} |
} |
587 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
588 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
589 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
590 |
|
# delete memory cache for lookup file |
591 |
|
delete $cache->{lhash}; |
592 |
} |
} |
593 |
|
|
594 |
# open existing lookup file |
# open existing lookup file |
725 |
print STDERR "\n"; |
print STDERR "\n"; |
726 |
|
|
727 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
728 |
use Spreadsheet::ParseExcel; |
require Spreadsheet::ParseExcel; |
729 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
require Spreadsheet::ParseExcel::Utility; |
730 |
|
import Spreadsheet::ParseExcel::Utility qw(int2col); |
731 |
|
|
732 |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
733 |
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 "$database doesn't have 'excel_file' defined!"; |
781 |
} |
} |
782 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
783 |
|
|
784 |
use MARC; |
require MARC; |
785 |
|
|
786 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
787 |
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!"; |
788 |
|
|
789 |
# optional argument is format |
# optional argument is format |
790 |
my $format = x($config->{format}) || 'usmarc'; |
my $format = x($config->{marc_format}) || 'usmarc'; |
791 |
|
|
792 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
793 |
|
|
794 |
my $marc = new MARC; |
my $marc = new MARC; |
795 |
my $nr = $marc->openmarc({ |
my $nr = $marc->openmarc({ |
796 |
file=>$marc_file, format=>$format |
file=>$marc_file, format=>$format |
797 |
}) || die "Can't open MARC file '$marc_file'"; |
}) || die "Can't open MARC file '$marc_file' with format '$format'"; |
798 |
|
|
799 |
|
# read MARC file in memory |
800 |
|
$marc->nextmarc(-1); |
801 |
|
|
802 |
my $i=0; # record nr. |
my $max_rec = $marc->marc_count(); |
803 |
|
|
804 |
my $rec; |
for(my $i=1; $i<=$max_rec; $i++) { |
805 |
|
|
806 |
while ($marc->nextmarc(1)) { |
progress($i,$max_rec); |
807 |
|
|
808 |
# XXX |
# store value for marc_sf.pm |
809 |
fakeprogress($i++); |
$main::cache->{marc_record} = $i; |
810 |
|
|
811 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
812 |
|
|
818 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
819 |
} |
} |
820 |
} |
} |
821 |
|
|
822 |
|
print STDERR "\n"; |
823 |
|
|
824 |
} elsif ($type_base eq "feed") { |
} elsif ($type_base eq "feed") { |
825 |
|
|
826 |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |