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 |
); |
); |
62 |
|
|
63 |
sub data2xml { |
sub data2xml { |
67 |
my $type = shift @_; |
my $type = shift @_; |
68 |
my $row = shift @_; |
my $row = shift @_; |
69 |
my $add_xml = shift @_; |
my $add_xml = shift @_; |
70 |
|
# needed to read values from configuration file |
71 |
|
my $cfg = shift @_; |
72 |
|
my $database = shift @_; |
73 |
|
|
74 |
my $xml; |
my $xml; |
75 |
|
|
174 |
} |
} |
175 |
} |
} |
176 |
|
|
177 |
|
# now try to parse variables from configuration file |
178 |
|
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
179 |
|
|
180 |
|
my $delimiter = x($x->{delimiter}) || ' '; |
181 |
|
my $val = $cfg->val($database, x($x->{content})); |
182 |
|
|
183 |
|
my ($s,$d,$i) = (1,1,0); # swish, display default |
184 |
|
$s = 0 if (lc($x->{type}) eq "display"); |
185 |
|
$d = 0 if (lc($x->{type}) eq "swish"); |
186 |
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
187 |
|
|
188 |
|
if ($val) { |
189 |
|
$display_data .= $delimiter.$val if ($d); |
190 |
|
$swish_data .= $val if ($s); |
191 |
|
$index->insert($field, $val, $path) if ($i); |
192 |
|
} |
193 |
|
|
194 |
|
} |
195 |
|
|
196 |
|
|
197 |
if ($display_data) { |
if ($display_data) { |
198 |
|
|
264 |
|
|
265 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
266 |
|
|
267 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type} ], forcecontent => 1); |
# extract just type basic |
268 |
|
my $type_base = $type; |
269 |
|
$type_base =~ s/_.+$//g; |
270 |
|
|
271 |
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
272 |
|
|
273 |
# output current progress indicator |
# output current progress indicator |
274 |
my $last_p = 0; |
my $last_p = 0; |
275 |
sub progress { |
sub progress { |
276 |
#return if (! $opts{q}); # FIXME |
#return if (! $opts{q}); # FIXME |
277 |
my $current = shift; |
my $current = shift; |
278 |
my $total = shift; |
my $total = shift || 1; |
279 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
280 |
if ($p != $last_p) { |
if ($p != $last_p) { |
281 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
286 |
# now read database |
# now read database |
287 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
288 |
|
|
289 |
if ($type eq "isis") { |
if ($type_base eq "isis") { |
290 |
|
|
291 |
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!"; |
292 |
|
|
293 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
307 |
|
|
308 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
309 |
|
|
310 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
311 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
312 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
313 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
318 |
} |
} |
319 |
print STDERR "\n"; |
print STDERR "\n"; |
320 |
|
|
321 |
} elsif ($type eq "excel") { |
} elsif ($type_base eq "excel") { |
322 |
use Spreadsheet::ParseExcel; |
use Spreadsheet::ParseExcel; |
323 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
324 |
|
|
364 |
|
|
365 |
next if (! $row); |
next if (! $row); |
366 |
|
|
367 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
368 |
|
$xml = $cp2utf->convert($xml); |
369 |
|
use bytes; # as opposed to chars |
370 |
|
print "Path-Name: $swishpath\n"; |
371 |
|
print "Content-Length: ".(length($xml)+1)."\n"; |
372 |
|
print "Document-Type: XML\n\n$xml\n"; |
373 |
|
} |
374 |
|
} |
375 |
|
} elsif ($type_base eq "marc") { |
376 |
|
## XXX |
377 |
|
use MARC; |
378 |
|
|
379 |
|
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
380 |
|
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
381 |
|
|
382 |
|
# optional argument is format |
383 |
|
my $format = x($config->{format}) || 'usmarc'; |
384 |
|
|
385 |
|
my %id_stored; # to aviod duplicates |
386 |
|
|
387 |
|
print STDERR "Reading MARC file '$marc_file'\n"; |
388 |
|
|
389 |
|
my $marc = new MARC; |
390 |
|
my $nr = $marc->openmarc({ |
391 |
|
file=>$marc_file, format=>$format |
392 |
|
}) || die "Can't open MARC file '$marc_file'"; |
393 |
|
|
394 |
|
my $i=0; # record nr. |
395 |
|
my $inc=1; |
396 |
|
my $max_i=1000; |
397 |
|
|
398 |
|
my $rec; |
399 |
|
|
400 |
|
while ($marc->nextmarc(1)) { |
401 |
|
|
402 |
|
# XXX |
403 |
|
progress($i, $max_i); |
404 |
|
$i += $inc; |
405 |
|
$inc = -$inc if ($i > $max_i || $i < 0); |
406 |
|
|
407 |
|
my $swishpath = $database."#".$i; |
408 |
|
|
409 |
|
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
410 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
411 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
412 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |