From 8df3fe40b8e434ffd47a298bef845645ab480f86 Mon Sep 17 00:00:00 2001
From: dbs <dbs@6d9bc8c9-1ec2-4278-b937-99fde70a366f>
Date: Sun, 19 Apr 2009 21:20:57 +0000
Subject: [PATCH] Fix a few common yaz-marcdump conversion to XML errors

git-svn-id: svn://svn.open-ils.org/ILS-Contrib/conifer/trunk@358 6d9bc8c9-1ec2-4278-b937-99fde70a366f
---
 tools/migration-scripts/fix_bad_marcxml.pl | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

diff --git a/tools/migration-scripts/fix_bad_marcxml.pl b/tools/migration-scripts/fix_bad_marcxml.pl
index 9b202c6c44..44a0f9a0e3 100644
--- a/tools/migration-scripts/fix_bad_marcxml.pl
+++ b/tools/migration-scripts/fix_bad_marcxml.pl
@@ -3,19 +3,20 @@ use strict;
 use warnings;
 
 foreach my $file (@ARGV) {
-	clean_empty_datafields($file);
+	process_file($file);
 }
 
-sub clean_empty_datafields {
+sub process_file {
 	my $file = shift;
 
 	# Empty datafields anger MARC::File::XML
 	open(FH, '<', $file) or die $!;
 	open(CLEAN, '>', "$file.new");
 
-	my ($trim, $lastline) = (0, '');
+	my ($trim, $lastline, $lineno) = (0, '', 1);
 	while (<FH>) {
 		if ($_ =~ m#</datafield># and $lastline =~ m#<datafield#) {
+			print STDERR "Empty datafield at line $lineno of file $file\n";
 			$trim = 1;
 		} elsif ($trim) {
 			$trim = 0;
@@ -23,8 +24,25 @@ sub clean_empty_datafields {
 			print CLEAN $lastline;
 			$trim = 0;
 		}
+
+		# Given questionable input, yaz-marcdump creates invalid XML like this:
+		#   <datafield tag="500" ind1=" " ind2=" ">
+		#    <subfield code="a">In subtitle "sports" appears as "</subfield>
+		#    <subfield code="p">ort</subfield>
+		#    <subfield code=""">.</subfield>
+		#  </datafield>
+		#
+		# This will at least enable MARC::File::XML to process it:
+		if ($_ =~ m#<subfield code=""">#o) {
+			print STDERR "Bad subfield code \" at line $lineno of file $file\n";
+			$_ =~ s{<subfield code=""">}{<subfield code="a">}o;
+		} elsif ($_ =~ m#<subfield code="<">#o) {
+			print STDERR "Bad subfield code < at line $lineno of file $file\n";
+			$_ =~ s{<subfield code="<">}{<subfield code="a">}o;
+		}
 		
 		$lastline = $_;
+		$lineno++;
 	}
 	print CLEAN $lastline;
 
-- 
2.11.0