From: Dan Scott <dscott@laurentian.ca>
Date: Sat, 27 Dec 2014 21:51:32 +0000 (-0500)
Subject: More example in-database MARC fixing
X-Git-Url: https://old-git.evergreen-ils.org/?a=commitdiff_plain;h=a1638d30c3acaa13d5f5c2f2b26786d2ad79ac0f;p=contrib%2FConifer.git

More example in-database MARC fixing

Generated the fix file with

\o ecco_uris_to_fix.sql
SELECT 'UPDATE biblio.record_entry SET marc = conifer.fix_ecco_uris(' || id ||
 ') WHERE id = ' || id || ';' FROM biblio.record_entry WHERE marc ~
 'galegroup.com/ecco' AND marc ~ 'wind05901' ORDER BY id ASC;
\o

And then running ecco_uris_to_fix.sql

Signed-off-by: Dan Scott <dscott@laurentian.ca>
---

diff --git a/Open-ILS/src/sql/Pg/fix_ecco_uris.sql b/Open-ILS/src/sql/Pg/fix_ecco_uris.sql
new file mode 100644
index 0000000000..19fd52693f
--- /dev/null
+++ b/Open-ILS/src/sql/Pg/fix_ecco_uris.sql
@@ -0,0 +1,112 @@
+CREATE OR REPLACE FUNCTION conifer.fix_ecco_uris(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;
+
+MARC::Charset->assume_unicode(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');
+foreach my $ocho (@eights) {
+    my @ous = $ocho->subfield('9');
+    foreach my $ou (@ous) {
+        if ($ou eq 'WINDSYS' or $ou eq 'OWA') {
+            $record->delete_field($ocho);
+        }
+        if ($ou eq 'OSUL') {
+            $ocho->update('9' => 'LUSYS');
+        }
+    }
+
+    # Fix URIs that raise SSL cert issues
+    my $uri = $ocho->subfield('u');
+    my $old_uri = $uri;
+    if ($uri =~ m#^https://librweb#) {
+        $uri =~ s{^https://librweb}{http://librweb};
+    }
+
+    # Fix URIs with a space after the URL proxy parameter
+    $uri =~ s{url= http}{url=http};
+
+    # Point at LU ECCO URIs
+    $uri =~ s{wind05901}{subd78095};
+
+    if ($old_uri ne $uri) {
+        $ocho->update('u' => $uri);
+    }
+
+    # Provide an indication of the platform
+    if ($uri =~ m#scholarsportal#) {
+        $ocho->update('y' => 'Available online / disponible en ligne (ScholarsPortal)');
+    } elsif ($uri =~ m#myilibrary#) {
+        $ocho->update('y' => 'Available online / disponible en ligne (MyiLibrary)');
+    } elsif ($uri =~ m#sagepub#) {
+        $ocho->update('y' => 'Available online / disponible en ligne (Sage)');
+    } elsif ($uri =~ m#galegroup.com/ecco#) {
+        $ocho->update('y' => 'Available online / disponible en ligne (Gale ECCO)');
+    }
+}
+
+# Do not need access notes anymore
+my @access = $record->field('506');
+foreach my $note (@access) {
+    my @ous = $note->subfield('9');
+    foreach my $ou (@ous) {
+        if ($ou eq 'OWA') {
+            $record->delete_field($note);
+        }
+    }
+}
+
+# Update provenance of the records
+my @provs = $record->field('040');
+foreach my $prov (@provs) {
+    $prov->delete_subfield(code => 'd', match => qr/CaOWA/);
+    my @subfields = $prov->subfield('d');
+    my $found = 0;
+    foreach my $subfield (@subfields) {
+        if ($subfield eq 'CaOSUL') {
+            $found = 1;
+        }
+    }
+    if (!$found) {
+        $prov->add_subfields('d' => 'CaOSUL');
+    }
+}
+
+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;
+
+CREATE OR REPLACE FUNCTION conifer.marc_version() RETURNS TEXT AS $func$
+use MARC::Record;
+return $MARC::Record::VERSION;
+$func$ LANGUAGE PLPERLU;
+