From faed9bcdeb4d89b38df644dcf549a1f9e981634c Mon Sep 17 00:00:00 2001
From: Dan Scott <dscott@laurentian.ca>
Date: Mon, 18 Apr 2016 14:37:36 -0400
Subject: [PATCH] Fold separate NOSME and NOSMW URIs to just OSM

Signed-off-by: Dan Scott <dscott@laurentian.ca>
---
 .../src/sql/Pg/update_marc_records_in_database.sql | 62 ++++++++++++++++++++++
 1 file changed, 62 insertions(+)

diff --git a/Open-ILS/src/sql/Pg/update_marc_records_in_database.sql b/Open-ILS/src/sql/Pg/update_marc_records_in_database.sql
index 3e116cfff4..f02372abfa 100644
--- a/Open-ILS/src/sql/Pg/update_marc_records_in_database.sql
+++ b/Open-ILS/src/sql/Pg/update_marc_records_in_database.sql
@@ -278,3 +278,65 @@ EXCEPTION
 END;
 $func$ LANGUAGE PLPGSQL;
 
+CREATE OR REPLACE FUNCTION conifer.squash_to_osm(record BIGINT) RETURNS TEXT AS $func$
+use strict;
+use MARC::Record;
+use MARC::File::XML (BinaryEncoding => 'UTF-8');
+use MARC::Charset;
+use Encode;
+use Unicode::Normalize;
+use LWP::UserAgent;
+use JSON::XS;
+use Digest::MD5;
+
+MARC::Charset->assume_unicode(1);
+
+my $md5 = Digest::MD5->new;
+my $json = new JSON::XS;
+$json->latin1(1);
+
+my $q = spi_prepare('SELECT marc FROM biblio.record_entry WHERE id = $1', 'BIGINT');
+my $marc = spi_exec_prepared($q, $_[0])->{rows}->[0]->{marc};
+my $record = MARC::Record->new_from_xml($marc);
+
+my @eights = $record->field('856');
+my $saw_nosme = 0;
+foreach my $ocho (@eights) {
+    my @ous = $ocho->subfield('9');
+    foreach my $ou (sort @ous) {
+        if ($ou eq 'NOSME') {
+            $ocho->update('9' => 'OSM');
+            $saw_nosme = 1;
+        }
+    }
+}
+foreach my $ocho (@eights) {
+    my @ous = $ocho->subfield('9');
+    foreach my $ou (sort @ous) {
+        if ($ou eq 'NOSMW' && $saw_nosme) {
+            $record->delete_field($ocho);
+        }
+    }
+}
+
+my $xml = $record->as_xml_record();
+$xml =~ s/\n//sgo;
+$xml =~ s/^<\?xml.+\?\s*>//go;
+$xml =~ s/>\s+</></go;
+$xml =~ s/\p{Cc}//go;
+
+# Embed a version of OpenILS::Application::AppUtils->entityize()
+# to avoid having to set PERL5LIB for PostgreSQL as well
+
+$xml = NFC($xml);
+
+# Convert raw ampersands to entities
+$xml =~ s/&(?!\S+;)/&amp;/gso;
+
+# Convert Unicode characters to entities
+$xml =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
+
+$xml =~ s/[\x00-\x1f]//go;
+
+return $xml;
+$func$ LANGUAGE PLPERLU;
-- 
2.11.0