4 |
|
|
5 |
use Cwd qw/abs_path/; |
use Cwd qw/abs_path/; |
6 |
use File::Temp qw/tempdir/; |
use File::Temp qw/tempdir/; |
|
use Data::Dumper; |
|
7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
19 |
use Time::HiRes qw/time/; |
use Time::HiRes qw/time/; |
20 |
use File::Slurp; |
use File::Slurp; |
21 |
use MARC::Record 2.0; # need 2.0 for utf-8 encoding see marcpm.sf.net |
use MARC::Record 2.0; # need 2.0 for utf-8 encoding see marcpm.sf.net |
22 |
|
use MARC::Lint; |
23 |
|
use Data::Dump qw/dump/; |
24 |
|
|
25 |
=head1 NAME |
=head1 NAME |
26 |
|
|
72 |
|
|
73 |
Optional path to output file |
Optional path to output file |
74 |
|
|
75 |
|
=item --marc-lint |
76 |
|
|
77 |
|
By default turned on if C<--marc-normalize> is used. You can disable lint |
78 |
|
messages with C<--no-marc-lint>. |
79 |
|
|
80 |
|
=item --marc-dump |
81 |
|
|
82 |
|
Force dump or input and marc record for debugging. |
83 |
|
|
84 |
=back |
=back |
85 |
|
|
86 |
=cut |
=cut |
95 |
my $stats = 0; |
my $stats = 0; |
96 |
my $validate_path; |
my $validate_path; |
97 |
my ($marc_normalize, $marc_output); |
my ($marc_normalize, $marc_output); |
98 |
|
my $marc_lint = 1; |
99 |
|
my $marc_dump = 0; |
100 |
|
|
101 |
GetOptions( |
GetOptions( |
102 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
105 |
"one=s" => \$only_filter, |
"one=s" => \$only_filter, |
106 |
"only=s" => \$only_filter, |
"only=s" => \$only_filter, |
107 |
"config" => \$config, |
"config" => \$config, |
108 |
"debug" => \$debug, |
"debug+" => \$debug, |
109 |
"stats" => \$stats, |
"stats" => \$stats, |
110 |
"validate=s" => \$validate_path, |
"validate=s" => \$validate_path, |
111 |
"marc-normalize=s" => \$marc_normalize, |
"marc-normalize=s" => \$marc_normalize, |
112 |
"marc-output=s" => \$marc_output, |
"marc-output=s" => \$marc_output, |
113 |
|
"marc-lint!" => \$marc_lint, |
114 |
|
"marc-dump!" => \$marc_dump, |
115 |
); |
); |
116 |
|
|
117 |
$config = LoadFile($config); |
$config = LoadFile($config); |
118 |
|
|
119 |
print "config = ",Dumper($config) if ($debug); |
print "config = ",dump($config) if ($debug); |
120 |
|
|
121 |
die "no databases in config file!\n" unless ($config->{databases}); |
die "no databases in config file!\n" unless ($config->{databases}); |
122 |
|
|
145 |
my @links; |
my @links; |
146 |
my $indexer; |
my $indexer; |
147 |
|
|
148 |
|
my $lint = new MARC::Lint if ($marc_lint); |
149 |
|
|
150 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
151 |
|
|
152 |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
276 |
open($marc_fh, '>', $path) || |
open($marc_fh, '>', $path) || |
277 |
$log->logdie("can't open MARC output $path: $!"); |
$log->logdie("can't open MARC output $path: $!"); |
278 |
|
|
279 |
$log->info("Creating MARC export file $path\n"); |
$log->info("Creating MARC export file $path", $marc_lint ? ' (with lint)' : '', "\n"); |
280 |
} |
} |
281 |
|
|
282 |
# reset position in database |
# reset position in database |
300 |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors); |
301 |
} |
} |
302 |
|
|
|
|
|
303 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
304 |
row => $row, |
row => $row, |
305 |
rules => $rules, |
rules => $rules, |
322 |
if ($marc_fh) { |
if ($marc_fh) { |
323 |
my $marc = new MARC::Record; |
my $marc = new MARC::Record; |
324 |
$marc->encoding( 'utf-8' ); |
$marc->encoding( 'utf-8' ); |
325 |
$marc->add_fields( WebPAC::Normalize::_get_marc_fields() ); |
my @marc_fields = WebPAC::Normalize::_get_marc_fields(); |
326 |
print $marc_fh $marc->as_usmarc; |
if (! @marc_fields) { |
327 |
|
$log->warn("MARC record $mfn is empty, skipping"); |
328 |
|
} else { |
329 |
|
$marc->add_fields( @marc_fields ); |
330 |
|
|
331 |
|
# tweak leader |
332 |
|
if (my $new_l = WebPAC::Normalize::marc_leader()) { |
333 |
|
|
334 |
|
my $leader = $marc->leader; |
335 |
|
|
336 |
|
foreach my $o ( keys %$new_l ) { |
337 |
|
my $insert = $new_l->{$o}; |
338 |
|
$leader = substr($leader, 0, $o) . |
339 |
|
$insert . substr($leader, $o+length($insert)); |
340 |
|
} |
341 |
|
$marc->leader( $leader ); |
342 |
|
} |
343 |
|
|
344 |
|
if ($marc_lint) { |
345 |
|
$lint->check_record( $marc ); |
346 |
|
my $err = join( "\n", $lint->warnings ); |
347 |
|
$log->error("MARC lint detected warning on MFN $mfn\n", |
348 |
|
"<<<<< Original imput row:\n",dump($row), "\n", |
349 |
|
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump(@marc_fields), "\n", |
350 |
|
"!!!!! MARC lint warnings:\n",$err,"\n" |
351 |
|
) if ($err); |
352 |
|
} |
353 |
|
|
354 |
|
if ($marc_dump) { |
355 |
|
$log->info("MARC record on MFN $mfn\n", |
356 |
|
"<<<<< Original imput row:\n",dump($row), "\n", |
357 |
|
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump(@marc_fields), "\n", |
358 |
|
); |
359 |
|
} |
360 |
|
|
361 |
|
print $marc_fh $marc->as_usmarc; |
362 |
|
} |
363 |
} |
} |
364 |
|
|
365 |
$total_rows++; |
$total_rows++; |