5 use lib '/openils/lib/perl5/';
8 use OpenSRF::Application;
9 use OpenSRF::EX qw/:try/;
10 use OpenSRF::AppSession;
11 use OpenSRF::MultiSession;
12 use OpenSRF::Utils::SettingsClient;
13 use OpenILS::Application::AppUtils;
14 use OpenILS::Utils::Fieldmapper;
15 use Digest::MD5 qw/md5_hex/;
18 use Unicode::Normalize;
22 use Time::HiRes qw/time/;
25 use MARC::File::XML ( BinaryEncoding => 'utf-8' );
28 #MARC::Charset->ignore_errors(1);
30 my ($id_field, $recid, $user, $config, $marctype, $keyfile, $dontuse_file, $enc, $force_enc, @files, @trash_fields) =
31 ('', 1, 1, '/openils/conf/bootstrap.conf', 'USMARC');
34 'marctype=s' => \$marctype,
35 'startid=i' => \$recid,
36 'idfield=s' => \$id_field,
38 'encoding=s' => \$enc,
39 'hard_encoding' => \$force_enc,
40 'keyfile=s' => \$keyfile,
41 'config=s' => \$config,
43 'trash=s' => \@trash_fields,
44 'dontuse=s' => \$dontuse_file
48 MARC::Charset->ignore_errors(1);
49 MARC::Charset->assume_encoding($enc);
52 if (uc($marctype) eq 'XML') {
58 @files = @ARGV if (!@files);
75 OpenSRF::System->bootstrap_client( config_file => $config );
76 Fieldmapper->import(IDL => OpenSRF::Utils::SettingsClient->new->config_value("IDL"));
80 open F, $keyfile or die "Couldn't open key file $keyfile";
82 if ( /^(\d+)\|(\S+)/o ) {
91 open F, $dontuse_file or die "Couldn't open used-id file $dontuse_file";
101 select STDERR; $| = 1;
102 select STDOUT; $| = 1;
104 my $batch = new MARC::Batch ( $marctype, @files );
105 $batch->strict_off();
106 $batch->warnings_off();
109 my $starttime = time;
112 while ( try { $rec = $batch->next } otherwise { $rec = -1 } ) {
113 next if ($rec == -1);
117 while ($used_ids{$recid}) {
120 $used_ids{$recid} = 1;
123 my $field = $rec->field($id_field);
125 if ($field->is_control_field) {
128 $id = $field->subfield('a');
140 if (my $tcn = $keymap{$id}) {
141 $rec->delete_field( $_ ) for ($rec->field($id_field));
142 $rec->append_fields( MARC::Field->new( $id_field, '', '', 'a', $tcn ) );
150 ($rec, $tcn) = preprocess($rec);
151 $rec->delete_field( $_ ) for ($rec->field($id_field));
152 $rec->append_fields( $tcn );
158 my $tcn_value = $rec->subfield('901' => 'a') || "SYS$id";
159 my $tcn_source = $rec->subfield('901' => 'b') || 'System';
161 (my $xml = $rec->as_xml_record()) =~ s/\n//sog;
162 $xml =~ s/^<\?xml.+\?\s*>//go;
163 $xml =~ s/>\s+</></go;
164 $xml =~ s/\p{Cc}//go;
165 $xml = entityize($xml);
167 my $bib = new Fieldmapper::biblio::record_entry;
172 $bib->creator($user);
173 $bib->create_date('now');
175 $bib->edit_date('now');
176 $bib->tcn_source($tcn_source);
177 $bib->tcn_value($tcn_value);
178 $bib->last_xact_id('IMPORT-'.$starttime);
180 print JSON->perl2JSON($bib)."\n";
181 $dontuse_id{$tcn_value} = 1;
185 if (!($count % 50)) {
186 print STDERR "\r$count\t". $count / (time - $starttime);
193 my ($id, $source, $value) = ('','','');
196 my $f = $rec->field('001');
197 $id = $f->data if ($f);
200 if (!$id || $dontuse_id{$source.$id}) {
201 my $f = $rec->field('000');
202 $id = $f->data if ($f);
203 $source = 'g' if ($f); # only PG seems to use this
206 if (!$id || $dontuse_id{$source.$id}) {
207 my $f = $rec->field('020');
208 $id = $f->subfield('a') if ($f);
209 $source = 'i' if ($f);
212 if (!$id || $dontuse_id{$source.$id}) {
213 my $f = $rec->field('022');
214 $id = $f->subfield('a') if ($f);
215 $source = 'i' if ($f);
218 if (!$id || $dontuse_id{$source.$id}) {
219 my $f = $rec->field('010');
220 $id = $f->subfield('a') if ($f);
221 $source = 'l' if ($f);
225 # my $f = $rec->field($id_field);
226 # $id = $f->subfield('a') if ($f);
229 $rec->delete_field($_) for ($rec->field($id_field, @trash_fields));
234 $id =~ s/^(\S+).*$/$1/o;
236 $id = $source.$id if ($source);
238 ($source, $value) = $id =~ /^(.)(.+)$/o;
239 if ($id =~ /^o(\d+)$/o) {
245 if ($id && $dontuse_id{$id}) {
246 warn "\n!!! ID $id is already in use\n";
255 my $tcn = MARC::Field->new(
258 b => do { $source_map{$source} || 'System' },
268 if ($form and $form eq 'D') {
269 $stuff = NFD($stuff);
271 $stuff = NFC($stuff);
274 $stuff =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;