From: Jason Etheridge <jason@esilibrary.com>
Date: Thu, 29 Jun 2017 18:50:07 +0000 (-0400)
Subject: fix -fs, and more output when fixing stuff
X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=commitdiff_plain;h=00daef49e8646e7ecd55f44525237241aa9e17a0;hp=3d029e1c2c46a181fe7da27df11974c4e3d7cd49

fix -fs, and more output when fixing stuff

Signed-off-by: Jason Etheridge <jason@esilibrary.com>
---

diff --git a/marc_cleanup b/marc_cleanup
index e79180a..58a68c1 100755
--- a/marc_cleanup
+++ b/marc_cleanup
@@ -111,7 +111,7 @@ 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");
+                edit("Junk in subfield code/Null subfield code ($1)");
                 next;
             }
         }
@@ -205,13 +205,28 @@ sub do_automated_cleanups {
         }
 
         # automatable subfield maladies
-        $record[$ptr] =~ s/code=" ">c/code="c">/;
-        $record[$ptr] =~ s/code=" ">\$/code="c">\$/;
+        if ($record[$ptr] =~ /code=" ">c/) {
+            message('Fixing probable subfield c, scenario 1');
+            $record[$ptr] =~ s/code=" ">c/code="c">/;
+        }
+        if ($record[$ptr] =~ /code=" ">\$/) {
+            message('Fixing probable subfield c, scenario 2');
+            $record[$ptr] =~ s/code=" ">\$/code="c">\$/;
+        }
 
         if ($c->{'fix-subfield'}) {
-            $record[$ptr] =~ s/code="&amp;">/code="$c->{'fix-subfield'}">/;
-            $record[$ptr] =~ s/code="\P{IsAlnum}">/code="$c->{'fix-subfield'}">/;
-            $record[$ptr] =~ s/code="">/code="$c->{'fix-subfield'}">/;
+            if ($record[$ptr] =~ /code="&amp;">/) {
+                message('Fixing &amp; for subfield code');
+                $record[$ptr] =~ s/code="&amp;">/code="$c->{'fix-subfield'}">/;
+            }
+            if ($record[$ptr] =~ /code="(.*?\P{IsAlnum}.*?)">/) {
+                message("Fixing non-alphanumeric subfield code: $1 -> " . $c->{'fix-subfield'});
+                $record[$ptr] =~ s/code=".*?\P{IsAlnum}.*?">/code="$c->{'fix-subfield'}">/;
+            }
+            if ($record[$ptr] =~ /code="">/) {
+                message('Fixing null subfield code');
+                $record[$ptr] =~ s/code="">/code="$c->{'fix-subfield'}">/;
+            }
         }
     }
     return 0;