18 |
|
|
19 |
my $config; |
my $config; |
20 |
|
|
21 |
use index_DBI; # there is no other, right now ;-) |
#use index_DBI; # default DBI module for index |
22 |
|
use index_DBI_cache; # faster DBI module using memory cache |
23 |
my $index; |
my $index; |
24 |
|
|
25 |
my %opts; |
my %opts; |
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 { |
68 |
my $type = shift @_; |
my $type = shift @_; |
69 |
my $row = shift @_; |
my $row = shift @_; |
70 |
my $add_xml = shift @_; |
my $add_xml = shift @_; |
71 |
|
# needed to read values from configuration file |
72 |
|
my $cfg = shift @_; |
73 |
|
my $database = shift @_; |
74 |
|
|
75 |
my $xml; |
my $xml; |
76 |
|
|
175 |
} |
} |
176 |
} |
} |
177 |
|
|
178 |
|
# now try to parse variables from configuration file |
179 |
|
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
180 |
|
|
181 |
|
my $delimiter = x($x->{delimiter}) || ' '; |
182 |
|
my $val = $cfg->val($database, x($x->{content})); |
183 |
|
|
184 |
|
my ($s,$d,$i) = (1,1,0); # swish, display default |
185 |
|
$s = 0 if (lc($x->{type}) eq "display"); |
186 |
|
$d = 0 if (lc($x->{type}) eq "swish"); |
187 |
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
188 |
|
|
189 |
|
if ($val) { |
190 |
|
$display_data .= $delimiter.$val if ($d); |
191 |
|
$swish_data .= $val if ($s); |
192 |
|
$index->insert($field, $val, $path) if ($i); |
193 |
|
} |
194 |
|
|
195 |
|
} |
196 |
|
|
197 |
|
|
198 |
if ($display_data) { |
if ($display_data) { |
199 |
|
|
231 |
|
|
232 |
# dump formatted output in <html> |
# dump formatted output in <html> |
233 |
if ($html) { |
if ($html) { |
234 |
$xml .= xmlify("html",$html); |
#$xml .= xmlify("html",$html); |
235 |
|
$xml .= "<html><![CDATA[ $html ]]></html>"; |
236 |
} |
} |
237 |
|
|
238 |
if ($xml) { |
if ($xml) { |
266 |
|
|
267 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
268 |
|
|
269 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type} ], forcecontent => 1); |
# extract just type basic |
270 |
|
my $type_base = $type; |
271 |
|
$type_base =~ s/_.+$//g; |
272 |
|
|
273 |
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
274 |
|
|
275 |
# output current progress indicator |
# output current progress indicator |
276 |
my $last_p = 0; |
my $last_p = 0; |
285 |
} |
} |
286 |
} |
} |
287 |
|
|
288 |
|
my $fake_dir = 1; |
289 |
|
sub fakeprogress { |
290 |
|
my $current = shift @_; |
291 |
|
|
292 |
|
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
293 |
|
|
294 |
|
$last_p += $fake_dir; |
295 |
|
$fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0); |
296 |
|
if ($last_p % 10 == 0) { |
297 |
|
printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]); |
298 |
|
} |
299 |
|
} |
300 |
|
|
301 |
# now read database |
# now read database |
302 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
303 |
|
|
304 |
if ($type eq "isis") { |
if ($type_base eq "isis") { |
305 |
|
|
306 |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
307 |
|
|
308 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
322 |
|
|
323 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
324 |
|
|
325 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
326 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
327 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
328 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
333 |
} |
} |
334 |
print STDERR "\n"; |
print STDERR "\n"; |
335 |
|
|
336 |
} elsif ($type eq "excel") { |
} elsif ($type_base eq "excel") { |
337 |
use Spreadsheet::ParseExcel; |
use Spreadsheet::ParseExcel; |
338 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
339 |
|
|
379 |
|
|
380 |
next if (! $row); |
next if (! $row); |
381 |
|
|
382 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
383 |
|
$xml = $cp2utf->convert($xml); |
384 |
|
use bytes; # as opposed to chars |
385 |
|
print "Path-Name: $swishpath\n"; |
386 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
387 |
|
print "Document-Type: XML\n\n$xml\n"; |
388 |
|
} |
389 |
|
} |
390 |
|
} elsif ($type_base eq "marc") { |
391 |
|
|
392 |
|
use MARC; |
393 |
|
|
394 |
|
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
395 |
|
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
396 |
|
|
397 |
|
# optional argument is format |
398 |
|
my $format = x($config->{format}) || 'usmarc'; |
399 |
|
|
400 |
|
print STDERR "Reading MARC file '$marc_file'\n"; |
401 |
|
|
402 |
|
my $marc = new MARC; |
403 |
|
my $nr = $marc->openmarc({ |
404 |
|
file=>$marc_file, format=>$format |
405 |
|
}) || die "Can't open MARC file '$marc_file'"; |
406 |
|
|
407 |
|
my $i=0; # record nr. |
408 |
|
|
409 |
|
my $rec; |
410 |
|
|
411 |
|
while ($marc->nextmarc(1)) { |
412 |
|
|
413 |
|
# XXX |
414 |
|
fakeprogress($i++); |
415 |
|
|
416 |
|
my $swishpath = $database."#".$i; |
417 |
|
|
418 |
|
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
419 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
420 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
421 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
423 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
424 |
} |
} |
425 |
} |
} |
426 |
|
} elsif ($type_base eq "feed") { |
427 |
|
|
428 |
|
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
429 |
|
my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!"; |
430 |
|
|
431 |
|
print STDERR "Reading feed from program '$prog'\n"; |
432 |
|
|
433 |
|
open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; |
434 |
|
|
435 |
|
my $i=1; # record nr. |
436 |
|
|
437 |
|
my $data; |
438 |
|
my $line=1; |
439 |
|
|
440 |
|
while (<FEED>) { |
441 |
|
chomp; |
442 |
|
|
443 |
|
if (/^$/) { |
444 |
|
my $swishpath = $database."#".$i++; |
445 |
|
|
446 |
|
if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { |
447 |
|
$xml = $cp2utf->convert($xml); |
448 |
|
use bytes; # as opposed to chars |
449 |
|
print "Path-Name: $swishpath\n"; |
450 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
451 |
|
print "Document-Type: XML\n\n$xml\n"; |
452 |
|
} |
453 |
|
$line = 1; |
454 |
|
$data = {}; |
455 |
|
next; |
456 |
|
} |
457 |
|
|
458 |
|
$line = $1 if (s/^(\d+):\s*//); |
459 |
|
$data->{$line++} = $_; |
460 |
|
|
461 |
|
fakeprogress($i); |
462 |
|
|
463 |
|
} |
464 |
} |
} |
465 |
} |
} |
466 |
|
|
473 |
|
|
474 |
=head1 NAME |
=head1 NAME |
475 |
|
|
476 |
isis2xml.pl - read isis file and dump XML |
all2xml.pl - read various file formats and dump XML for SWISH-E |
477 |
|
|
478 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
479 |
|
|
480 |
This command will read ISIS data file using OpenIsis perl module and |
This command will read ISIS data file using OpenIsis perl module, MARC |
481 |
create XML file for usage with I<SWISH-E> |
records using MARC module and optionally Micro$oft Excel files to |
482 |
indexer. Dispite it's name, this script B<isn't general xml generator> |
create one XML file for usage with I<SWISH-E> indexer. Dispite it's name, |
483 |
from isis files (isis allready has something like that). Output of this |
this script B<isn't general xml generator> from isis files (isis allready |
484 |
script is tailor-made for SWISH-E. |
has something like that). Output of this script is tailor-made for SWISH-E. |
485 |
|
|
486 |
|
=head1 BUGS |
487 |
|
|
488 |
|
Documentation is really lacking. However, in true Open Source spirit, source |
489 |
|
is best documentation. I even made considerable effort to comment parts |
490 |
|
which are not intuitively clear, so... |
491 |
|
|
492 |
=head1 AUTHOR |
=head1 AUTHOR |
493 |
|
|