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; |
34 |
|
|
35 |
my $path; # this is name of database |
my $path; # this is name of database |
36 |
|
|
37 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
Text::Iconv->raise_error(0); # Conversion errors don't raise exceptions |
38 |
|
|
39 |
# this is encoding of all files on disk, including import_xml/*.xml file and |
# this is encoding of all files on disk, including import_xml/*.xml file and |
40 |
# filter/*.pm files! It will be used to store strings in perl internally! |
# filter/*.pm files! It will be used to store strings in perl internally! |
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 $xml; |
76 |
|
|
124 |
# init vars so that we go into while... |
# init vars so that we go into while... |
125 |
($swish,$display) = (1,1); |
($swish,$display) = (1,1); |
126 |
|
|
127 |
if ($swish || $display) { |
# while because of repeatable fields |
128 |
|
while ($swish || $display) { |
129 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
130 |
|
if ($repeat_off > 1000) { |
131 |
|
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
132 |
|
last; |
133 |
|
} |
134 |
|
|
135 |
# filter="name" ; filter this field through |
# filter="name" ; filter this field through |
136 |
# filter/[name].pm |
# filter/[name].pm |
137 |
my $filter = $x->{filter}; |
my $filter = $x->{filter}; |
181 |
} |
} |
182 |
} |
} |
183 |
|
|
184 |
|
# now try to parse variables from configuration file |
185 |
|
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
186 |
|
|
187 |
|
my $delimiter = x($x->{delimiter}) || ' '; |
188 |
|
my $val = $cfg->val($database, x($x->{content})); |
189 |
|
|
190 |
|
my ($s,$d,$i) = (1,1,0); # swish, display default |
191 |
|
$s = 0 if (lc($x->{type}) eq "display"); |
192 |
|
$d = 0 if (lc($x->{type}) eq "swish"); |
193 |
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
194 |
|
|
195 |
|
if ($val) { |
196 |
|
$display_data .= $delimiter.$val if ($d); |
197 |
|
$swish_data .= $val if ($s); |
198 |
|
$index->insert($field, $val, $path) if ($i); |
199 |
|
} |
200 |
|
|
201 |
|
} |
202 |
|
|
203 |
|
|
204 |
if ($display_data) { |
if ($display_data) { |
205 |
|
|
237 |
|
|
238 |
# dump formatted output in <html> |
# dump formatted output in <html> |
239 |
if ($html) { |
if ($html) { |
240 |
$xml .= xmlify("html",$html); |
#$xml .= xmlify("html",$html); |
241 |
|
$xml .= "<html><![CDATA[ $html ]]></html>"; |
242 |
} |
} |
243 |
|
|
244 |
if ($xml) { |
if ($xml) { |
272 |
|
|
273 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
274 |
|
|
275 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type} ], forcecontent => 1); |
# extract just type basic |
276 |
|
my $type_base = $type; |
277 |
|
$type_base =~ s/_.+$//g; |
278 |
|
|
279 |
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
280 |
|
|
281 |
# output current progress indicator |
# output current progress indicator |
282 |
my $last_p = 0; |
my $last_p = 0; |
283 |
sub progress { |
sub progress { |
284 |
#return if (! $opts{q}); # FIXME |
#return if (! $opts{q}); # FIXME |
285 |
my $current = shift; |
my $current = shift; |
286 |
my $total = shift; |
my $total = shift || 1; |
287 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
288 |
if ($p != $last_p) { |
if ($p != $last_p) { |
289 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
291 |
} |
} |
292 |
} |
} |
293 |
|
|
294 |
|
my $fake_dir = 1; |
295 |
|
sub fakeprogress { |
296 |
|
my $current = shift @_; |
297 |
|
|
298 |
|
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
299 |
|
|
300 |
|
$last_p += $fake_dir; |
301 |
|
$fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0); |
302 |
|
if ($last_p % 10 == 0) { |
303 |
|
printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]); |
304 |
|
} |
305 |
|
} |
306 |
|
|
307 |
# now read database |
# now read database |
308 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
309 |
|
|
310 |
if ($type eq "isis") { |
if ($type_base eq "isis") { |
311 |
|
|
312 |
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!"; |
313 |
|
|
314 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
328 |
|
|
329 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
330 |
|
|
331 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
332 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
333 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
334 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
339 |
} |
} |
340 |
print STDERR "\n"; |
print STDERR "\n"; |
341 |
|
|
342 |
} elsif ($type eq "excel") { |
} elsif ($type_base eq "excel") { |
343 |
use Spreadsheet::ParseExcel; |
use Spreadsheet::ParseExcel; |
344 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
345 |
|
|
385 |
|
|
386 |
next if (! $row); |
next if (! $row); |
387 |
|
|
388 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
389 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
390 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
391 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
393 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
394 |
} |
} |
395 |
} |
} |
396 |
|
} elsif ($type_base eq "marc") { |
397 |
|
|
398 |
|
use MARC; |
399 |
|
|
400 |
|
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
401 |
|
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
402 |
|
|
403 |
|
# optional argument is format |
404 |
|
my $format = x($config->{format}) || 'usmarc'; |
405 |
|
|
406 |
|
print STDERR "Reading MARC file '$marc_file'\n"; |
407 |
|
|
408 |
|
my $marc = new MARC; |
409 |
|
my $nr = $marc->openmarc({ |
410 |
|
file=>$marc_file, format=>$format |
411 |
|
}) || die "Can't open MARC file '$marc_file'"; |
412 |
|
|
413 |
|
my $i=0; # record nr. |
414 |
|
|
415 |
|
my $rec; |
416 |
|
|
417 |
|
while ($marc->nextmarc(1)) { |
418 |
|
|
419 |
|
# XXX |
420 |
|
fakeprogress($i++); |
421 |
|
|
422 |
|
my $swishpath = $database."#".$i; |
423 |
|
|
424 |
|
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
425 |
|
$xml = $cp2utf->convert($xml); |
426 |
|
use bytes; # as opposed to chars |
427 |
|
print "Path-Name: $swishpath\n"; |
428 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
429 |
|
print "Document-Type: XML\n\n$xml\n"; |
430 |
|
} |
431 |
|
} |
432 |
|
} elsif ($type_base eq "feed") { |
433 |
|
|
434 |
|
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
435 |
|
my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!"; |
436 |
|
|
437 |
|
print STDERR "Reading feed from program '$prog'\n"; |
438 |
|
|
439 |
|
open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; |
440 |
|
|
441 |
|
my $i=1; # record nr. |
442 |
|
|
443 |
|
my $data; |
444 |
|
my $line=1; |
445 |
|
|
446 |
|
while (<FEED>) { |
447 |
|
chomp; |
448 |
|
|
449 |
|
if (/^$/) { |
450 |
|
my $swishpath = $database."#".$i++; |
451 |
|
|
452 |
|
if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { |
453 |
|
$xml = $cp2utf->convert($xml); |
454 |
|
use bytes; # as opposed to chars |
455 |
|
print "Path-Name: $swishpath\n"; |
456 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
457 |
|
print "Document-Type: XML\n\n$xml\n"; |
458 |
|
} |
459 |
|
$line = 1; |
460 |
|
$data = {}; |
461 |
|
next; |
462 |
|
} |
463 |
|
|
464 |
|
$line = $1 if (s/^(\d+):\s*//); |
465 |
|
$data->{$line++} = $_; |
466 |
|
|
467 |
|
fakeprogress($i); |
468 |
|
|
469 |
|
} |
470 |
} |
} |
471 |
} |
} |
472 |
|
|
479 |
|
|
480 |
=head1 NAME |
=head1 NAME |
481 |
|
|
482 |
isis2xml.pl - read isis file and dump XML |
all2xml.pl - read various file formats and dump XML for SWISH-E |
483 |
|
|
484 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
485 |
|
|
486 |
This command will read ISIS data file using OpenIsis perl module and |
This command will read ISIS data file using OpenIsis perl module, MARC |
487 |
create XML file for usage with I<SWISH-E> |
records using MARC module and optionally Micro$oft Excel files to |
488 |
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, |
489 |
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 |
490 |
script is tailor-made for SWISH-E. |
has something like that). Output of this script is tailor-made for SWISH-E. |
491 |
|
|
492 |
|
=head1 BUGS |
493 |
|
|
494 |
|
Documentation is really lacking. However, in true Open Source spirit, source |
495 |
|
is best documentation. I even made considerable effort to comment parts |
496 |
|
which are not intuitively clear, so... |
497 |
|
|
498 |
=head1 AUTHOR |
=head1 AUTHOR |
499 |
|
|