From: Galen Charlton <gmc@esilibrary.com>
Date: Tue, 13 Apr 2010 19:18:52 +0000 (+0000)
Subject: recover gracefully after an XML parsing error
X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=commitdiff_plain;h=26d762670d7610ed9ccb1c36a8dbab3169e4e2a3

recover gracefully after an XML parsing error
---

diff --git a/fingerprinter b/fingerprinter
index 93df981..dcf2e29 100755
--- a/fingerprinter
+++ b/fingerprinter
@@ -27,7 +27,15 @@ for my $file (@ARGV) {
     $batch->strict_off();
     $batch->warnings_off();
 
-    while ( $record = $batch->next ) {
+    my $record;
+    while ( 1 ) {
+        eval { $record = $batch->next; };
+        if ($@) {
+            import MARC::File::XML;
+            print "bad record\n";
+            next;
+        }
+        last unless $record;
         $count++; progress_ticker();
         my $marc = undef;
         unless ( defined $record )
diff --git a/marc_cleanup b/marc_cleanup
index 89f8ab0..53801f3 100755
--- a/marc_cleanup
+++ b/marc_cleanup
@@ -92,12 +92,12 @@ while ( buildrecord() ) {
         }
 
         # subfields can't be non-alphanumeric
-        if ($record[$ptr] =~ /<subfield code="(.*?)"/) {
-            if ($1 =~ /\P{IsAlnum}/ or $1 eq '') {
-                edit("Junk in subfield code/Null subfield code");
-                next;
-            }
-        }
+        #if ($record[$ptr] =~ /<subfield code="(.*?)"/) {
+            #if ($1 =~ /\P{IsAlnum}/ or $1 eq '') {
+                #edit("Junk in subfield code/Null subfield code");
+                #next;
+            #}
+        #}
         # subfields can't be non-alphanumeric
         if ($record[$ptr] =~ /<subfield code="(\w{2,})"/) {
             edit("Subfield code larger than 1 char");