use MARC::Batch;
use MARC::File;
use MARC::File::XML;
-binmode STDIN, ':bytes';
+use MARC::Charset 'marc8_to_utf8';
+#binmode STDIN, ':bytes';
use Env qw(
HOME PGHOST PGPORT PGUSER PGDATABASE MIGSCHEMA
MIGBASEWORKDIR MIGBASEGITDIR MIGGITDIR MIGWORKDIR
pod2usage(-verbose => 1) if ! $ARGV[1];
my $next_arg_is_file = 0;
-my $append_is_false = 1;
+my $append = 0;
my $next_arg_is_source = 0;
my $source = 'default';
my $file_is_xml = 0;
my $batch;
binmode STDIN, ':utf8';
+my $ignore = MARC::Charset->ignore_errors();
+MARC::Charset->ignore_errors(1);
+#causes issues sometimes
+#may be useful other times ... still in flux
+#my $setting = MARC::Charset->assume_unicode();
+#MARC::Charset->assume_unicode(1);
+MARC::Charset->ignore_errors(1);
+
foreach my $arg (@ARGV) {
if ($arg eq '--stage_file') {
$next_arg_is_file = 1;
$next_arg_is_source = 1;
next;
}
- if ($next_arg_is_file) {
+ if ($next_arg_is_source) {
$source = $arg;
$next_arg_is_source = 0;
next;
}
if ($arg eq '--append') {
- $append_is_false = 0;
+ $append = 1;
next;
}
if ($arg eq '--xml') {
}
}
-create_child_table($dbh);
+create_child_table($dbh); #and test to see if it exists
# normal stage table creation
-if ($append_is_false) { create_stage_table($dbh); }
-
-#sanity check and create stage table if it doesn't exist
-my $query = "SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry_stage')";
-my $qsth = $dbh->prepare($query);
-$qsth->execute();
-my $f;
-while (my @row = $qsth->fetchrow_array) { $f = $row[0]; }
-if ($f eq 'f') { create_stage_table($dbh); }
-
-if ($append_is_false == 0) { create_stage_table($dbh); }
-
+if ($append == 0) {
+ drop_stage_table($dbh);
+ create_stage_table($dbh);
+ }
if ($file_is_xml) {
$batch = MARC::Batch->new('XML',$infile);
} else {
$batch->strict_off();
while ( my $record = $batch->next() ) {
- my $xml;
- if ($file_is_xml) { $xml = $record; }
- else { $xml = $record->as_xml_record(); }
+ my $xml = $record->as_xml_record();
+ $xml = marc8_to_utf8($xml);
$i++;
$xml = clean_marc($xml);
$xml = '$_$' . $xml . '$_$';
my @warnings = $batch->warnings();
my $warning_string;
if (@warnings) { $warning_string = "'" . join(':',@warnings) . "'"; } else { $warning_string = "'none'"; }
- my $sql = "INSERT INTO $MIGSCHEMA.biblio_record_entry_stage (marc,x_source,x_warnings) VALUES ($xml,$source,$warning_string);";
+ my $sql = "INSERT INTO $MIGSCHEMA.biblio_record_entry_legacy (marc,x_source,x_warnings) VALUES ($xml,'$source',$warning_string);";
my $sth = $dbh->prepare($sql);
$sth->execute();
report_progress("Records staged", $i) if 0 != $i % 100;
}
$dbh->do(qq/
- CREATE INDEX ${MIGSCHEMA}_biblio_record_entry_stage_idx ON
- $MIGSCHEMA.biblio_record_entry_stage (id);
+ CREATE INDEX ${MIGSCHEMA}_biblio_record_entry_legacy_idx ON
+ $MIGSCHEMA.biblio_record_entry_legacy (id);
/);
print "Finis.\n";
+
+sub drop_stage_table {
+ my $dbh = shift;
+ $dbh->do("DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_legacy;");
+ return();
+}
+
sub create_stage_table {
my $dbh = shift;
- $dbh->do("DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_stage;");
- $dbh->do("CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_stage (
+ $dbh->do("CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_legacy (
l_bib_id TEXT,
x_source TEXT,
x_warnings TEXT,
sub clean_marc {
my $xml = shift;
+ $xml = marc8_to_utf8($xml);
$xml =~ s/\n//sog;
$xml =~ s/^<\?xml.+\?\s*>//go;
$xml =~ s/>\s+</></go;