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 |
|
|
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 |
|
|
265 |
|
|
266 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
267 |
|
|
268 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type} ], forcecontent => 1); |
# extract just type basic |
269 |
|
my $type_base = $type; |
270 |
|
$type_base =~ s/_.+$//g; |
271 |
|
|
272 |
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
273 |
|
|
274 |
# output current progress indicator |
# output current progress indicator |
275 |
my $last_p = 0; |
my $last_p = 0; |
276 |
sub progress { |
sub progress { |
277 |
#return if (! $opts{q}); # FIXME |
#return if (! $opts{q}); # FIXME |
278 |
my $current = shift; |
my $current = shift; |
279 |
my $total = shift; |
my $total = shift || 1; |
280 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
281 |
if ($p != $last_p) { |
if ($p != $last_p) { |
282 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
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: $type...\n"; |
302 |
|
|
303 |
if ($type eq "isis") { |
if ($type_base eq "isis") { |
304 |
|
|
305 |
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!"; |
306 |
|
|
307 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
321 |
|
|
322 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
323 |
|
|
324 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
325 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
326 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
327 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
332 |
} |
} |
333 |
print STDERR "\n"; |
print STDERR "\n"; |
334 |
|
|
335 |
} elsif ($type eq "excel") { |
} elsif ($type_base eq "excel") { |
336 |
use Spreadsheet::ParseExcel; |
use Spreadsheet::ParseExcel; |
337 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
338 |
|
|
378 |
|
|
379 |
next if (! $row); |
next if (! $row); |
380 |
|
|
381 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
382 |
|
$xml = $cp2utf->convert($xml); |
383 |
|
use bytes; # as opposed to chars |
384 |
|
print "Path-Name: $swishpath\n"; |
385 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
386 |
|
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); |
$xml = $cp2utf->convert($xml); |
419 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
420 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
422 |
print "Document-Type: XML\n\n$xml\n"; |
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=0; # 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 |
|
$data->{$line++} = $_; |
458 |
|
|
459 |
|
fakeprogress($i++); |
460 |
|
|
461 |
|
} |
462 |
} |
} |
463 |
} |
} |
464 |
|
|