From b9e4ad4dde4619596902dbdb17d355ebc0a850dc Mon Sep 17 00:00:00 2001
From: Shawn Boyette <sboyette@esilibrary.com>
Date: Fri, 19 Sep 2008 20:05:57 +0000
Subject: [PATCH] naked amp detect

---
 yaz-cleanup |   27 ++++++++++++++++++++-------
 1 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/yaz-cleanup b/yaz-cleanup
index b3d5eda..1f9d93d 100755
--- a/yaz-cleanup
+++ b/yaz-cleanup
@@ -10,14 +10,16 @@ $| = 1;
 open MARC, '<', 'incoming.marc.xml';
 open NUMARC, '>', 'incoming.clean.marc.xml';
 
-until ($count == ($skip - 1)) {
-    my $t = <MARC>;
-    print NUMARC $t;
-    $count++;
-    printf("\rSpinning on to record %s (%2.2f%%)", $skip, ($count / $skip *100))
-      unless ($count % 1000);
+if ($skip) {
+    until ($count == ($skip - 1)) {
+        my $t = <MARC>;
+        print NUMARC $t;
+        $count++;
+        printf("\rSpinning on to record %s (%2.2f%%)", $skip, ($count / $skip *100))
+          unless ($count % 1000);
+    }
+    print "\nScrubbing resumes...\n" if $skip;
 }
-print "\nScrubbing resumes...\n" if $skip;
 
 my $line1 = <MARC>;
 
@@ -26,18 +28,29 @@ while (my $line2 = <MARC>) {
     # catch empty datafield elements
     if ($line1 =~ m/<datafield tag="..." ind1="." ind2=".">/) {
         if ($line2 =~ m|</datafield>|) {
+            print "Empty datafield scrubbed at line $count\n";
             $line1 = <MARC>;
             $count++;
             next;
         }
     }
 
+    # clean misplaced dollarsigns
+    if ($line1 =~ m|<subfield code="\$">c?\d+\.\d{2}|) {
+        $line1 =~ s|"\$">c?(\d+\.\d{2})|"c">\$$1|;
+        print "Dollar sign in subfield code corrected at line $count\n";
+    }
+
     # clean up tags with spaces in them
     $line1 =~ s/tag="  /tag="00/g;
     $line1 =~ s/tag=" /tag="0/g;
     $line1 =~ s/tag="-/tag="0/g;
     $line1 =~ s/tag="(\d\d) /tag="0$1/g;
 
+    # naked ampersands
+    die "Looks like naked ampersand at line $count: $line1"
+      if ($line1 =~ /&/ && $line1 !~ /&\w{1,7};/);
+
     # subfields can't be non-alphanumeric
     die "Junk in subfield at line $count: $line1"
       if $line1 =~ /<subfield code="[^[:alnum:]]"/;
-- 
1.7.2.5