82 |
|
|
83 |
# sort subrouting using order="" attribute |
# sort subrouting using order="" attribute |
84 |
sub by_order { |
sub by_order { |
85 |
return 0 if (! $config->{indexer}->{$a}->{order}); |
my $va = $config->{indexer}->{$a}->{order} || |
86 |
return 0 if (! $config->{indexer}->{$b}->{order}); |
$config->{indexer}->{$a}; |
87 |
|
my $vb = $config->{indexer}->{$b}->{order} || |
88 |
|
$config->{indexer}->{$b}; |
89 |
|
|
90 |
return $config->{indexer}->{$a}->{order} <=> |
return $va <=> $vb; |
|
$config->{indexer}->{$b}->{order} ; |
|
91 |
} |
} |
92 |
|
|
93 |
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
94 |
|
|
95 |
$field=x($field); |
$field=x($field); |
|
|
|
96 |
$field_usage{$field}++; |
$field_usage{$field}++; |
97 |
|
|
98 |
my $swish_data = ""; |
my $swish_data = ""; |
124 |
# init vars so that we go into while... |
# init vars so that we go into while... |
125 |
($swish,$display) = (1,1); |
($swish,$display) = (1,1); |
126 |
|
|
127 |
if ($swish || $display) { |
# placeholder for all repeatable entries for index |
128 |
|
my @index_data; |
129 |
|
my $index_filter; |
130 |
|
|
131 |
|
# while because of repeatable fields |
132 |
|
while ($swish || $display) { |
133 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
134 |
|
if ($repeat_off > 1000) { |
135 |
|
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
136 |
|
last; |
137 |
|
} |
138 |
|
|
139 |
# filter="name" ; filter this field through |
# filter="name" ; filter this field through |
140 |
# filter/[name].pm |
# filter/[name].pm |
141 |
my $filter = $x->{filter}; |
my $filter = $x->{filter}; |
160 |
} |
} |
161 |
if ($filter) { |
if ($filter) { |
162 |
no strict 'refs'; |
no strict 'refs'; |
163 |
$display_data .= join($delimiter,&$filter($display)); |
$display_data .= $delimiter.&$filter($display); |
164 |
} else { |
} else { |
165 |
if ($display_data) { |
$display_data .= $delimiter.$display; |
|
$display_data .= $delimiter.$display; |
|
|
} else { |
|
|
$display_data .= $display; |
|
|
} |
|
166 |
} |
} |
167 |
} |
} |
168 |
|
|
169 |
# type="index" ; insert into index |
# type="index" ; insert into index |
170 |
if ($i && $display) { |
if ($i && $display) { |
171 |
my $index_data = $display; |
push @index_data, $display; |
172 |
if ($filter) { |
$index_filter = $filter if ($filter); |
173 |
no strict 'refs'; |
} |
174 |
foreach my $d (&$filter($index_data)) { |
} |
175 |
$index->insert($field, $d, $path); |
|
176 |
} |
# fill data in index |
177 |
} else { |
if (@index_data) { |
178 |
$index->insert($field, $index_data, $path); |
if ($index_filter) { |
179 |
|
no strict 'refs'; |
180 |
|
foreach my $d (@index_data) { |
181 |
|
$index->insert($field, &$index_filter($d), $path); |
182 |
|
} |
183 |
|
} else { |
184 |
|
foreach my $d (@index_data) { |
185 |
|
$index->insert($field, $d, $path); |
186 |
} |
} |
187 |
} |
} |
188 |
} |
} |
244 |
|
|
245 |
# dump formatted output in <html> |
# dump formatted output in <html> |
246 |
if ($html) { |
if ($html) { |
247 |
$xml .= xmlify("html",$html); |
#$xml .= xmlify("html",$html); |
248 |
|
$xml .= "<html><![CDATA[ $html ]]></html>"; |
249 |
} |
} |
250 |
|
|
251 |
if ($xml) { |
if ($xml) { |
272 |
$cfg_global->val('global', 'dbi_passwd') || '', |
$cfg_global->val('global', 'dbi_passwd') || '', |
273 |
); |
); |
274 |
|
|
275 |
|
my $show_progress = $cfg_global->val('global', 'show_progress'); |
276 |
|
|
277 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
278 |
|
|
279 |
my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; |
my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; |
290 |
# output current progress indicator |
# output current progress indicator |
291 |
my $last_p = 0; |
my $last_p = 0; |
292 |
sub progress { |
sub progress { |
293 |
#return if (! $opts{q}); # FIXME |
return if (! $show_progress); |
294 |
my $current = shift; |
my $current = shift; |
295 |
my $total = shift || 1; |
my $total = shift || 1; |
296 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
302 |
|
|
303 |
my $fake_dir = 1; |
my $fake_dir = 1; |
304 |
sub fakeprogress { |
sub fakeprogress { |
305 |
|
return if (! $show_progress); |
306 |
my $current = shift @_; |
my $current = shift @_; |
307 |
|
|
308 |
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
324 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
325 |
my $db = OpenIsis::open( $isis_db ); |
my $db = OpenIsis::open( $isis_db ); |
326 |
|
|
327 |
|
# check if .txt database for OpenIsis is zero length, |
328 |
|
# if so, erase it and re-open database |
329 |
|
sub check_txt_db { |
330 |
|
my $isis_db = shift || die "need isis database name"; |
331 |
|
my $reopen = 0; |
332 |
|
|
333 |
|
if (-e $isis_db.".TXT") { |
334 |
|
print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n"; |
335 |
|
unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!"; |
336 |
|
$reopen++; |
337 |
|
} |
338 |
|
if (-e $isis_db.".PTR") { |
339 |
|
print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n"; |
340 |
|
unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!"; |
341 |
|
$reopen++; |
342 |
|
} |
343 |
|
return OpenIsis::open( $isis_db ) if ($reopen); |
344 |
|
} |
345 |
|
|
346 |
|
# EOF error |
347 |
|
if ($db == -1) { |
348 |
|
$db = check_txt_db($isis_db); |
349 |
|
if ($db == -1) { |
350 |
|
print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n"; |
351 |
|
next; |
352 |
|
} |
353 |
|
} |
354 |
|
|
355 |
|
# OpenIsis::ERR_BADF |
356 |
|
if ($db == -4) { |
357 |
|
print STDERR "FATAL: OpenIsis can't find file $isis_db\n"; |
358 |
|
next; |
359 |
|
# OpenIsis::ERR_IO |
360 |
|
} elsif ($db == -5) { |
361 |
|
print STDERR "FATAL: OpenIsis can't access file $isis_db\n"; |
362 |
|
next; |
363 |
|
} elsif ($db < 0) { |
364 |
|
print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n"; |
365 |
|
next; |
366 |
|
} |
367 |
|
|
368 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
my $max_rowid = OpenIsis::maxRowid( $db ); |
369 |
|
|
370 |
|
# if 0 records, try to rease isis .txt database |
371 |
|
if ($max_rowid == 0) { |
372 |
|
# force removal of database |
373 |
|
$db = check_txt_db($isis_db); |
374 |
|
$max_rowid = OpenIsis::maxRowid( $db ); |
375 |
|
} |
376 |
|
|
377 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
378 |
|
|
379 |
my $path = $database; |
my $path = $database; |
395 |
} |
} |
396 |
} |
} |
397 |
} |
} |
398 |
|
# for this to work with current version of OpenIsis (0.9.0) |
399 |
|
# you might need my patch from |
400 |
|
# http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff |
401 |
|
OpenIsis::close($db); |
402 |
print STDERR "\n"; |
print STDERR "\n"; |
403 |
|
|
404 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
500 |
|
|
501 |
open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; |
open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; |
502 |
|
|
503 |
my $i=0; # record nr. |
my $i=1; # record nr. |
504 |
|
|
505 |
my $data; |
my $data; |
506 |
my $line=1; |
my $line=1; |
509 |
chomp; |
chomp; |
510 |
|
|
511 |
if (/^$/) { |
if (/^$/) { |
512 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i++; |
513 |
|
|
514 |
if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { |
if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { |
515 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
523 |
next; |
next; |
524 |
} |
} |
525 |
|
|
526 |
|
$line = $1 if (s/^(\d+):\s*//); |
527 |
$data->{$line++} = $_; |
$data->{$line++} = $_; |
528 |
|
|
529 |
fakeprogress($i++); |
fakeprogress($i); |
530 |
|
|
531 |
} |
} |
532 |
} |
} |
541 |
|
|
542 |
=head1 NAME |
=head1 NAME |
543 |
|
|
544 |
isis2xml.pl - read isis file and dump XML |
all2xml.pl - read various file formats and dump XML for SWISH-E |
545 |
|
|
546 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
547 |
|
|
548 |
This command will read ISIS data file using OpenIsis perl module and |
This command will read ISIS data file using OpenIsis perl module, MARC |
549 |
create XML file for usage with I<SWISH-E> |
records using MARC module and optionally Micro$oft Excel files to |
550 |
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, |
551 |
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 |
552 |
script is tailor-made for SWISH-E. |
has something like that). Output of this script is tailor-made for SWISH-E. |
553 |
|
|
554 |
|
=head1 BUGS |
555 |
|
|
556 |
|
Documentation is really lacking. However, in true Open Source spirit, source |
557 |
|
is best documentation. I even made considerable effort to comment parts |
558 |
|
which are not intuitively clear, so... |
559 |
|
|
560 |
=head1 AUTHOR |
=head1 AUTHOR |
561 |
|
|