From c16c4d4c903563ce83f4c6df5d6c51a4c6742b2c Mon Sep 17 00:00:00 2001
From: Rogan Hamby <rhamby@equinoxinitiative.org>
Date: Tue, 14 Jul 2020 12:56:29 -0400
Subject: [PATCH] enhancements to mig prepbibs

---
 emig.d/bin/mig-prepbibs |   20 ++++++++++++++++----
 kmig.d/bin/mig-prepbibs |   19 +++++++++++++++----
 2 files changed, 31 insertions(+), 8 deletions(-)

diff --git a/emig.d/bin/mig-prepbibs b/emig.d/bin/mig-prepbibs
index 69c2674..93d1c66 100755
--- a/emig.d/bin/mig-prepbibs
+++ b/emig.d/bin/mig-prepbibs
@@ -45,19 +45,31 @@ pod2usage(-verbose => 2) if defined $ARGV[0] && $ARGV[0] eq '--help';
 pod2usage(-verbose => 1) if ! $ARGV[1];
 
 my $infile;
+my $i = 0;
 
 my $ret = GetOptions(
     'file:s'           		 => \$infile
 );
 
 my $outfile = $infile . '.prepped_xml';
+my $errorfile = $infile . '.errors_xml';
 
 open my $outfh, '>:utf8', $outfile or die "Can't open output file $!\n";
-
-my $batch = MARC::Batch->new( 'USMARC', $infile );
+open my $errorfh, '>:utf8', $errorfile or die "Can't open error file $!\n";
+
+my $filetype = `file $infile`;
+my $batch;
+if ($filetype =~ m/MARC21/) {
+	$batch = MARC::Batch->new( 'USMARC', $infile );
+} else {
+	$batch = MARC::Batch->new( 'XML', $infile );
+}
 $batch->strict_off();
 
-while ( my $record = $batch->next() ) {
+my $record;
+while ( eval {$record = $batch->next()} or do { if ($@) { print $errorfh $record->as_xml(); next; } } ) {
+	$i++;
+	print "Processed: $i\n" if ($i % 100) == 0;
 	my @nnn = $record->field('852');
 	foreach my $n (@nnn) {
 		$n->set_tag('851');
@@ -68,7 +80,7 @@ while ( my $record = $batch->next() ) {
 
 close ($infile);
 close ($outfile);
-
+close ($errorfile);
 
 ########### functions
 
diff --git a/kmig.d/bin/mig-prepbibs b/kmig.d/bin/mig-prepbibs
index b3a89f1..7c86c38 100755
--- a/kmig.d/bin/mig-prepbibs
+++ b/kmig.d/bin/mig-prepbibs
@@ -45,19 +45,30 @@ pod2usage(-verbose => 2) if defined $ARGV[0] && $ARGV[0] eq '--help';
 pod2usage(-verbose => 1) if ! $ARGV[1];
 
 my $infile;
+my $i = 0;
 
 my $ret = GetOptions(
     'file:s'           		 => \$infile
 );
 
 my $outfile = $infile . '.prepped_xml';
+my $errorfile = $infile . '.errors_xml';
 
 open my $outfh, '>:utf8', $outfile or die "Can't open output file $!\n";
-
-my $batch = MARC::Batch->new( 'USMARC', $infile );
+open my $errorfh, '>:utf8', $errorfile or die "Can't open error file $!\n";
+
+my $filetype = `file $infile`;
+my $batch;
+if ($filetype =~ m/MARC21/) {
+	$batch = MARC::Batch->new( 'USMARC', $infile );
+} else {
+	$batch = MARC::Batch->new( 'XML', $infile );
+}
 $batch->strict_off();
 
-while ( my $record = $batch->next() ) {
+while ( eval {$record = $batch->next()} or do { if ($@) { print $errorfh $record->as_xml(); next; } } ) {
+ 	$i++;
+ 	print "Processed: $i\n" if ($i % 100) == 0;
 	my @nnn = $record->field('999');
 	foreach my $n (@nnn) {
 		$n->set_tag('998');
@@ -68,7 +79,7 @@ while ( my $record = $batch->next() ) {
 
 close ($infile);
 close ($outfile);
-
+close ($errorfile);
 
 ########### functions
 
-- 
1.7.2.5