7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Lookup; |
use WebPAC::Lookup 0.03; |
11 |
use WebPAC::Input 0.03; |
use WebPAC::Input 0.07; |
12 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.03; |
13 |
use WebPAC::Normalize; |
use WebPAC::Normalize 0.11; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate; |
use WebPAC::Validate; |
16 |
|
use WebPAC::Output::MARC; |
17 |
use YAML qw/LoadFile/; |
use YAML qw/LoadFile/; |
18 |
use Getopt::Long; |
use Getopt::Long; |
19 |
use File::Path; |
use File::Path; |
20 |
use Time::HiRes qw/time/; |
use Time::HiRes qw/time/; |
21 |
use File::Slurp; |
use File::Slurp; |
|
use MARC::Record 2.0; # need 2.0 for utf-8 encoding see marcpm.sf.net |
|
|
use MARC::Lint; |
|
22 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
23 |
|
|
24 |
=head1 NAME |
=head1 NAME |
144 |
my @links; |
my @links; |
145 |
my $indexer; |
my $indexer; |
146 |
|
|
|
my $lint = new MARC::Lint if ($marc_lint); |
|
|
|
|
147 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
148 |
|
|
149 |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
221 |
|
|
222 |
die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs)); |
die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs)); |
223 |
|
|
224 |
my $lookup = new WebPAC::Lookup( |
my $lookup; |
225 |
lookup_file => $input->{lookup}, |
if ($input->{lookup}) { |
226 |
) if ($input->{lookup}); |
$lookup = new WebPAC::Lookup( |
227 |
|
lookup_file => $input->{lookup}, |
228 |
|
); |
229 |
|
delete( $input->{lookup} ); |
230 |
|
} |
231 |
|
|
232 |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
233 |
|
|
237 |
|
|
238 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
239 |
module => $input_module, |
module => $input_module, |
240 |
code_page => $config->{webpac}->{webpac_encoding}, |
encoding => $config->{webpac}->{webpac_encoding}, |
241 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
242 |
offset => $offset, |
offset => $offset, |
243 |
lookup => $lookup, |
lookup_coderef => sub { |
244 |
|
my $rec = shift || return; |
245 |
|
$lookup->add( $rec ); |
246 |
|
}, |
247 |
recode => $input->{recode}, |
recode => $input->{recode}, |
248 |
stats => $stats, |
stats => $stats, |
249 |
); |
); |
275 |
|
|
276 |
$log->info("Using $normalize_path for normalization..."); |
$log->info("Using $normalize_path for normalization..."); |
277 |
|
|
278 |
my $marc_fh; |
my $marc = new WebPAC::Output::MARC( |
279 |
if (my $path = $normalize->{output}) { |
path => $normalize->{output}, |
280 |
open($marc_fh, '>', $path) || |
lint => $marc_lint, |
281 |
$log->logdie("can't open MARC output $path: $!"); |
dump => $marc_dump, |
282 |
|
) if ($normalize->{output}); |
|
$log->info("Creating MARC export file $path", $marc_lint ? ' (with lint)' : '', "\n"); |
|
|
} |
|
283 |
|
|
284 |
# reset position in database |
# reset position in database |
285 |
$input_db->seek(1); |
$input_db->seek(1); |
321 |
type => $config->{$use_indexer}->{type}, |
type => $config->{$use_indexer}->{type}, |
322 |
) if ($indexer && $ds); |
) if ($indexer && $ds); |
323 |
|
|
324 |
if ($marc_fh) { |
if ($marc) { |
325 |
my $marc = new MARC::Record; |
my $i = 0; |
|
$marc->encoding( 'utf-8' ); |
|
|
my @marc_fields = WebPAC::Normalize::_get_marc_fields(); |
|
|
if (! @marc_fields) { |
|
|
$log->warn("MARC record $mfn is empty, skipping"); |
|
|
} else { |
|
|
$marc->add_fields( @marc_fields ); |
|
|
|
|
|
if ($marc_lint) { |
|
|
$lint->check_record( $marc ); |
|
|
my $err = join( "\n", $lint->warnings ); |
|
|
$log->error("MARC lint detected warning on MFN $mfn\n", |
|
|
"Original imput row: ",dump($row), "\n", |
|
|
"Normalized MARC row: ",dump(@marc_fields), "\n", |
|
|
"MARC lint warnings: ",$err,"\n" |
|
|
) if ($err); |
|
|
} |
|
|
|
|
|
if ($marc_dump) { |
|
|
$log->info("MARC record on MFN $mfn\n", |
|
|
"Original imput row: ",dump($row), "\n", |
|
|
"Normalized MARC row: ",dump(@marc_fields), "\n", |
|
|
); |
|
|
} |
|
326 |
|
|
327 |
print $marc_fh $marc->as_usmarc; |
while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) { |
328 |
|
$marc->add( |
329 |
|
id => $mfn . ( $i ? "/$i" : '' ), |
330 |
|
fields => $fields, |
331 |
|
leader => WebPAC::Normalize::marc_leader(), |
332 |
|
row => $row, |
333 |
|
); |
334 |
|
$i++; |
335 |
} |
} |
336 |
|
|
337 |
|
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
338 |
} |
} |
339 |
|
|
340 |
$total_rows++; |
$total_rows++; |
343 |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats); |
344 |
|
|
345 |
# close MARC file |
# close MARC file |
346 |
close($marc_fh) if ($marc_fh); |
$marc->finish if ($marc); |
347 |
|
|
348 |
} |
} |
349 |
|
|