From: Shawn Boyette <sboyette@esilibrary.com>
Date: Wed, 24 Sep 2008 16:38:52 +0000 (+0000)
Subject: check for non-numerics in datafield tag
X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=commitdiff_plain;h=3f97f1f937ea404d1fde89dd4abcbfdc30b07674

check for non-numerics in datafield tag
---

diff --git a/marc-cleanup b/marc-cleanup
index 8071de5..401ae67 100755
--- a/marc-cleanup
+++ b/marc-cleanup
@@ -76,7 +76,7 @@ while (my $line = getline()) {
     $line =~ s/tag="(\d\d) /tag="0$1/g;
 
     # stow tag data if we're looking at it
-    if ($line =~ m/<datafield tag="(\d{3})" ind1="(.)" ind2="(.)">/) {
+    if ($line =~ m/<datafield tag="(.{3})" ind1="(.)" ind2="(.)">/) {
         $reccontext{tag}  = $1;
         $reccontext{ind1} = $2;
         $reccontext{ind2} = $3;
@@ -86,6 +86,15 @@ while (my $line = getline()) {
     if ($line =~ /&/ && $line !~ /&\w+?;/)
       { edit("Naked ampersand", $line); next }
 
+    # tags must be numeric
+    if ($line =~ /<datafield tag="(.+?)"/) {
+        my $match = $1;
+        if ($match =~ /\D/) {
+            edit("Non-numerics in tag", $line);
+            next;
+        }
+    }
+
     # subfields can't be non-alphanumeric
     if ($line =~ /<subfield code="(.+?)"/) {
         my $match = $1;