use warnings;
use DBI;
-use Data::Dumper;
-use MARC::Record;
-use MARC::Batch;
-use MARC::File;
-use MARC::File::XML;
+#binmode STDIN, ':bytes';
use Env qw(
HOME PGHOST PGPORT PGUSER PGDATABASE MIGSCHEMA
MIGBASEWORKDIR MIGBASEGITDIR MIGGITDIR MIGWORKDIR
use Cwd 'abs_path';
use FindBin;
use UNIVERSAL;
-use Unicode::Normalize;
my $mig_bin = "$FindBin::Bin/";
use lib "$FindBin::Bin/";
use Mig;
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 $next_arg_is_stage = 0;
+my $stage_table = 'biblio_record_entry';
my $source = 'default';
my $file_is_xml = 0;
my $dbh = Mig::db_connect();
my $batch;
binmode STDIN, ':utf8';
+#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') {
- $append_is_false = 0;
- next;
- }
- if ($arg eq '--xml') {
- $file_is_xml = 1;
- next;
- }
}
-if ($append_is_false) {
- $dbh->do(qq{
- DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_stage;
- CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_stage (
- l_bib_id TEXT,
- x_source TEXT,
- x_warnings TEXT,
- x_migrate BOOLEAN DEFAULT TRUE
- ) INHERITS ($MIGSCHEMA.biblio_record_entry);
- });
-}
+my $bre_test = check_for_table($dbh,'biblio_record_entry');
+if ($bre_test == 0) { create_child_bre($dbh); }
+
+my $xmig_test = check_for_column($dbh,'biblio_record_entry','x_migrate');
+if ($xmig_test == 0) { add_column($dbh,'biblio_record_entry','x_migrate','BOOLEAN DEFAULT TRUE');
+
+my $xsource_test = check_for_column($dbh,'biblio_record_entry','x_source');
+if ($xsource_test == 0) { add_column($dbh,'biblio_record_entry','x_source','TEXT');
+
+my $last_xact;
+if ($source) { $last_xact = "'$MIGSCHEMA $source'" } else { $last_xact = "'$MIGSCHEMA'"; }
+
+#flatten out MARC XML FILE
+open my $xml, "<:encoding(utf8)", $infile or abort('could not open MARC XML file');
+$i = 0;
+my $record;
+while(my $line = <$xml>) {
+ if ($line =~ /^<\/?collection/) { next; }
+ chomp $line;
+ $record = $record . $line;
+ if ($line =~ /^<\/record/) {
+ stage_record($dbh,$record,$last_xact);
+ $record = '';
+ }
+
+close $xml;
-if ($file_is_xml) {
- $batch = MARC::Batch->new('XML',$infile);
-} else {
- $batch = MARC::Batch->new('USMARC',$infile);
-}
-$batch->strict_off();
-
-while ( my $record = $batch->next() ) {
- my $xml;
- if ($file_is_xml) { $xml = $record; }
- else { $xml = $record->as_xml_record(); }
- $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 $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);
-/);
+#load the MARC XML FILE TO STAGING
+report_progress("Records staged", $i) if 0 != $i % 100;
print "Finis.\n";
-sub clean_marc {
- my $xml = shift;
- $xml =~ s/\n//sog;
- $xml =~ s/^<\?xml.+\?\s*>//go;
- $xml =~ s/>\s+</></go;
- $xml =~ s/\p{Cc}//go;
- $xml = NFC($xml);
- $xml =~ s/&(?!\S+;)/&/gso;
- $xml =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
- $xml =~ s/[\x00-\x1f]//go;
- return $xml;
+# beyond here be functions
+
+sub create_child_bre {
+ my $dbh = shift;
+ $dbh->do("DO \$\$
+ DECLARE
+ t BOOLEAN;
+ BEGIN
+ SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry') INTO t;
+ IF t = FALSE THEN
+ PERFORM migration_tools.build_specific_base_staging_table ('$MIGSCHEMA','biblio.record_entry');
+ END IF;
+ END \$\$;");
+
+ return ();
}
-
sub abort {
my $msg = shift;
print STDERR "$0: $msg", "\n";
print STDERR "$msg\n";
}
}
+
+sub stage_record {
+ my $dbh = shift;
+ my $record = shift;
+ my $last_xact = shift;
+ $record = '$_$' . $record . '$_$';
+ my $sql = "INSERT INTO $MIGSCHEMA.biblio_record_entry (last_xact_id,marc) VALUES ($last_xact,$record);";
+ my $sth = $dbh->prepare($sql);
+ $sth->execute();
+ return;
+}
+
+sub check_for_table {
+ my $dbh = shift;
+ my $table = shift;
+ my $sql = "SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = '$table';";
+ my $sth = $dbh->prepare($sql);
+ $sth->execute();
+ my @sqlresult = $sth->fetchrow_array;
+ my $r = pop @sqlresult;
+ if ($r) { return $r; } else { return 0; }
+}
+
+sub check_for_column {
+ my $dbh = shift;
+ my $table = shift;
+ my $column = shift;
+ my $sql = "SELECT 1 FROM information_schema.columns WHERE table_schema = '$MIGSCHEMA' AND table_name = '$table' AND column_name = $column;";
+ my $sth = $dbh->prepare($sql);
+ $sth->execute();
+ my @sqlresult = $sth->fetchrow_array;
+ my $r = pop @sqlresult;
+ if ($r) { return $r; } else { return 0; }
+}
+
+sub add_column {
+ my $dbh = shift;
+ my $table = shift;
+ my $column = shift;
+ my $column_type = shift;
+ my $sql = "ALTER TABLE $MIGSCHEMA.$table ADD COLUMN $COLUMN $COLUMN_TYPE;";
+ my $r = check_for_column($dbh,$table,$column);
+ if ($r == 0) { abort('failed to create column'; } else { return $r; }
+}
+
projects / migration-tools.git / blobdiff