56 |
# format in XML file |
# format in XML file |
57 |
my %type2tag = ( |
my %type2tag = ( |
58 |
'isis' => 'isis', |
'isis' => 'isis', |
59 |
'excel' => 'column' |
'excel' => 'column', |
60 |
|
'marc' => 'marc', |
61 |
|
'feed' => 'feed' |
62 |
); |
); |
63 |
|
|
64 |
sub data2xml { |
sub data2xml { |
178 |
# now try to parse variables from configuration file |
# now try to parse variables from configuration file |
179 |
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
180 |
|
|
181 |
|
my $delimiter = x($x->{delimiter}) || ' '; |
182 |
my $val = $cfg->val($database, x($x->{content})); |
my $val = $cfg->val($database, x($x->{content})); |
183 |
|
|
184 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$d,$i) = (1,1,0); # swish, display default |
187 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
188 |
|
|
189 |
if ($val) { |
if ($val) { |
190 |
$display_data .= $val if ($d); |
$display_data .= $delimiter.$val if ($d); |
191 |
$swish_data .= $val if ($s); |
$swish_data .= $val if ($s); |
192 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
193 |
} |
} |
284 |
} |
} |
285 |
} |
} |
286 |
|
|
287 |
|
my $fake_dir = 1; |
288 |
|
sub fakeprogress { |
289 |
|
my $current = shift @_; |
290 |
|
|
291 |
|
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
292 |
|
|
293 |
|
$last_p += $fake_dir; |
294 |
|
$fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0); |
295 |
|
if ($last_p % 10 == 0) { |
296 |
|
printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]); |
297 |
|
} |
298 |
|
} |
299 |
|
|
300 |
# now read database |
# now read database |
301 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
302 |
|
|
386 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
387 |
} |
} |
388 |
} |
} |
389 |
|
} elsif ($type_base eq "marc") { |
390 |
|
|
391 |
|
use MARC; |
392 |
|
|
393 |
|
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
394 |
|
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
395 |
|
|
396 |
|
# optional argument is format |
397 |
|
my $format = x($config->{format}) || 'usmarc'; |
398 |
|
|
399 |
|
print STDERR "Reading MARC file '$marc_file'\n"; |
400 |
|
|
401 |
|
my $marc = new MARC; |
402 |
|
my $nr = $marc->openmarc({ |
403 |
|
file=>$marc_file, format=>$format |
404 |
|
}) || die "Can't open MARC file '$marc_file'"; |
405 |
|
|
406 |
|
my $i=0; # record nr. |
407 |
|
|
408 |
|
my $rec; |
409 |
|
|
410 |
|
while ($marc->nextmarc(1)) { |
411 |
|
|
412 |
|
# XXX |
413 |
|
fakeprogress($i++); |
414 |
|
|
415 |
|
my $swishpath = $database."#".$i; |
416 |
|
|
417 |
|
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
418 |
|
$xml = $cp2utf->convert($xml); |
419 |
|
use bytes; # as opposed to chars |
420 |
|
print "Path-Name: $swishpath\n"; |
421 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
422 |
|
print "Document-Type: XML\n\n$xml\n"; |
423 |
|
} |
424 |
|
} |
425 |
|
} elsif ($type_base eq "feed") { |
426 |
|
|
427 |
|
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
428 |
|
my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!"; |
429 |
|
|
430 |
|
print STDERR "Reading feed from program '$prog'\n"; |
431 |
|
|
432 |
|
open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; |
433 |
|
|
434 |
|
my $i=1; # record nr. |
435 |
|
|
436 |
|
my $data; |
437 |
|
my $line=1; |
438 |
|
|
439 |
|
while (<FEED>) { |
440 |
|
chomp; |
441 |
|
|
442 |
|
if (/^$/) { |
443 |
|
my $swishpath = $database."#".$i++; |
444 |
|
|
445 |
|
if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { |
446 |
|
$xml = $cp2utf->convert($xml); |
447 |
|
use bytes; # as opposed to chars |
448 |
|
print "Path-Name: $swishpath\n"; |
449 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
450 |
|
print "Document-Type: XML\n\n$xml\n"; |
451 |
|
} |
452 |
|
$line = 1; |
453 |
|
$data = {}; |
454 |
|
next; |
455 |
|
} |
456 |
|
|
457 |
|
$line = $1 if (s/^(\d+):\s*//); |
458 |
|
$data->{$line++} = $_; |
459 |
|
|
460 |
|
fakeprogress($i); |
461 |
|
|
462 |
|
} |
463 |
} |
} |
464 |
} |
} |
465 |
|
|