From 7374d928fe95fa7ea93e2e8650da3a98bfce9738 Mon Sep 17 00:00:00 2001
From: dbs <dbs@6d9bc8c9-1ec2-4278-b937-99fde70a366f>
Date: Thu, 30 Apr 2009 15:23:38 +0000
Subject: [PATCH] Depending on when you process the XML, the annoying
 characters may be escaped

git-svn-id: svn://svn.open-ils.org/ILS-Contrib/conifer/trunk@424 6d9bc8c9-1ec2-4278-b937-99fde70a366f
---
 tools/migration-scripts/fix_bad_marcxml.pl | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/tools/migration-scripts/fix_bad_marcxml.pl b/tools/migration-scripts/fix_bad_marcxml.pl
index 20f174cd1c..2e45dbd3c5 100644
--- a/tools/migration-scripts/fix_bad_marcxml.pl
+++ b/tools/migration-scripts/fix_bad_marcxml.pl
@@ -33,15 +33,18 @@ sub process_file {
 		#  </datafield>
 		#
 		# This will at least enable MARC::File::XML to process it:
-		if ($_ =~ m#<subfield code=""">#o) {
+		if ($_ =~ m#<subfield code=""">#o or $_ =~ m#<subfield code="&quot;">#o ) {
 			print STDERR "Bad subfield code \" at line $lineno of file $file\n";
 			$_ =~ s{<subfield code=""">}{<subfield code="Z">}o;
-		} elsif ($_ =~ m#<subfield code="<">#o) {
+			$_ =~ s{<subfield code="&quot;">}{<subfield code="Z">}o;
+		} elsif ($_ =~ m#<subfield code="<">#o or $_ =~ m#<subfield code="&lt;">#o) {
 			print STDERR "Bad subfield code < at line $lineno of file $file\n";
 			$_ =~ s{<subfield code="<">}{<subfield code="Z">}o;
-		} elsif ($_ =~ m#<subfield code="&">#o) {
+			$_ =~ s{<subfield code="&lt;">}{<subfield code="Z">}o;
+		} elsif ($_ =~ m#<subfield code="&">#o or $_ =~ m#<subfield code="&amp;">#o) {
 			print STDERR "Bad subfield code & at line $lineno of file $file\n";
 			$_ =~ s{<subfield code="&">}{<subfield code="Z">}o;
+			$_ =~ s{<subfield code="&amp;">}{<subfield code="Z">}o;
 		} elsif ($_ =~ m#<datafield tag="(\d)" ind1="."#o) {
 			my $badtag = $1;
 			print STDERR "Bad datafield tag $badtag at line $lineno of file $file\n";
-- 
2.11.0