X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=marc_cleanup;h=58a68c163e0786ee2ef20ce154ce2de3f3b1e398;hp=4d7354e6f3c3373d514c5e684cdfcce2c1d3264f;hb=ed7dae56be7fb1dece7fdb5caaea8903c13b32cd;hpb=51e8d1e73795adf0fa387794c95495073c5a703e

diff --git a/marc_cleanup b/marc_cleanup
index 4d7354e..58a68c1 100755
--- a/marc_cleanup
+++ b/marc_cleanup
@@ -93,7 +93,7 @@ while ( buildrecord() ) {
     until ($ptr == $#record) {
         # get datafield/tag data if we have it
         $rc = stow_record_data() if ($c->{'renumber-from'} and $c->{'original-tag'});
-        return $rc if $rc;
+        next if $rc;
 
         # naked ampersands
         if ($record[$ptr] =~ /&/ && $record[$ptr] !~ /&\w+?;/)
@@ -111,11 +111,11 @@ 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 ($1)");
                 next;
             }
         }
-        # subfields can't be non-alphanumeric
+        # subfields can't be larger than 1 char (technically you could make the MARC format accomodate that:)
         if ($record[$ptr] =~ /<subfield code="(\w{2,})"/) {
             edit("Subfield code larger than 1 char");
             next;
@@ -177,6 +177,11 @@ sub do_automated_cleanups {
                 message("Short leader padded");
             }
         }
+        if ($c->{'force-utf8'}) {
+            if ($record[$ptr] =~ m|<leader>(.........).(.+)</leader>|) {
+                $record[$ptr] = "<leader>$1a$2</leader>\n";
+            }
+        }
         if ($record[$ptr] =~ m|<controlfield tag="008">(.+?)</control|) {
             #pad short 008
             my $content = $1;
@@ -193,9 +198,36 @@ sub do_automated_cleanups {
             message("Dollar sign corrected");
         }
 
+        # excessive trailing whitespace in subfield contents
+        if ($record[$ptr] =~ m|\s{10,}</subfield>|) {
+            $record[$ptr] =~ s|\s{10,}</subfield>|</subfield>|;
+            message("Trailing whitespace trimmed from subfield contents");
+        }
+
         # 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 ($c->{'fix-subfield'}) {
+            if ($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;
 }
@@ -609,6 +641,8 @@ sub initialize {
                          'renumber-subfield|rs=s',
                          'original-tag|ot=i',
                          'original-subfield|os=s',
+                         'fix-subfield|fs=s',
+                         'force-utf8',
                          'script',
                          'no-strip9',
                          'trashfile|t=s',
@@ -672,13 +706,17 @@ Options
                            and renumbering is in effect, an old-to-new mapping
                            file (old2new.map) will be generated.
 
-  --autoscrub  -a  Automatically remove non-numeric tags in data
-  --nocollapse -n  Don't compress records to one line on output
-  --no-strip9      Don't autoremove 901/903 tags in data
-  --trashfile  -t  File containing trash tag data (see --trashhelp)
-
-  --fullauto       No manual edits. All problematic records dumped to
-                   exception file.
+  --force-utf8             Rewrite each record so that they describe themselves as
+                           UTF-8 encoded
+  --autoscrub         -a   Automatically remove non-numeric tags in data
+  --fix-subfield      -fs  Subfield code to use in place of non-alphanumeric
+                           or empty subfield codes
+  --nocollapse        -n   Don't compress records to one line on output
+  --no-strip9              Don't autoremove 901/903 tags in data
+  --trashfile         -t   File containing trash tag data (see --trashhelp)
+
+  --fullauto               No manual edits. All problematic records dumped to
+                           exception file.
 
 HELP
 exit;