license and environment

[migration-tools.git] / marc-cleanup

diff --git a/marc-cleanup b/marc-cleanup
index 41aff58..0f7d638 100755 (executable)
--- a/marc-cleanup
+++ b/marc-cleanup
@@ -23,7 +23,7 @@ my %trash = ();   # hash for tags to be dumped
 initialize($conf);
 
 # read in trash tags file if it exists
-populate_trash() if (-e '.trashtags');
+populate_trash() if ($conf->{trash});
 
 my @record = ();  # current record storage
 my %recmeta = (); # metadata about current record
@@ -31,12 +31,12 @@ my @context= ();  # last 5 lines of file
 
 my $input = shift || 'incoming.marc.xml';
 
-open MARC, '<', $input;
-open my $NUMARC, '>', $conf->{output};
+open MARC, '<:utf8', $input;
+open my $NUMARC, '>:utf8', $conf->{output};
 print $NUMARC '<?xml version="1.0" encoding="UTF-8"?>',"\n";
 print $NUMARC '<collection xmlns="http://www.loc.gov/MARC21/slim">',"\n";
 
-open my $EXMARC, '>', $conf->{exception};
+open my $EXMARC, '>:utf8', $conf->{exception};
 print $EXMARC '<?xml version="1.0" encoding="UTF-8"?>',"\n";
 print $EXMARC '<collection xmlns="http://www.loc.gov/MARC21/slim">',"\n";
 open MARC2, '<', $input;
@@ -109,6 +109,10 @@ while (my $line = getline()) {
         $recmeta{ind2} = $3;
     }
 
+    # automatable subfield maladies
+    $line =~ s/code=" ">c/code="c">/;
+    $line =~ s/code=" ">\$/code="c"$>/;
+
     # and stow line back in record
     $record[-1] = $line;
 
@@ -126,10 +130,10 @@ while (my $line = getline()) {
     }
 
     # subfields can't be non-alphanumeric
-    if ($line =~ /<subfield code="(.+?)"/) {
+    if ($line =~ /<subfield code="(.*?)"/) {
         my $match = $1;
-        if ($match =~ /\P{IsAlnum}/) {
-            edit("Junk in subfield code", $line);
+        if ($match =~ /\P{IsAlnum}/ or $match eq '') {
+            edit("Junk in subfield code/Null subfield code", $line);
             next;
         }
     }
@@ -203,8 +207,8 @@ sub write_record {
     print $FH '<!-- ', $recmeta{explanation}, " -->\n"
       if(defined $recmeta{explanation});
 
-    # LOOP OVER %trash TO EXCISE UNWANTED TAGS1
-    if (keys %trash) {
+    # excise unwanted tags
+    if (keys %trash or $conf->{autoscrub}) {
         my @trimmed = ();
         my $istrash = 0;
         for my $line (@record) {
@@ -232,10 +236,11 @@ sub write_record {
     # add 903(?) with new record id
     my $renumber = '';
     if ($conf->{'renumber-from'}) {
-        $renumber = join('', '<datafield tag="', $conf->{'renumber-tag'}, '">',
+        $renumber = join('', '<datafield tag="', $conf->{'renumber-tag'},
+                         '" ind1=" " ind2=" ">',
                          '<subfield code="', $conf->{'renumber-subfield'}, '">',
                          $conf->{'renumber-from'}, '</subfield></datafield>');
-        $renumber .= "\n" unless $conf->{nocollapse};
+        $renumber .= "\n" if $conf->{nocollapse};
         push @record, $renumber;
         $conf->{'renumber-from'}++;
     }
@@ -385,8 +390,9 @@ sub quit { exit }
 # specifying a tag twice is an error, to help prevent typos
 
 sub populate_trash {
-    print $OUT ">>> TRASHTAGS FILE FOUND. LOADING TAGS TO BE STRIPPED FROM OUTPUT...\n";
-    open TRASH, '<', '.trashtags';
+    print $OUT ">>> TRASHTAGS FILE FOUND. LOADING TAGS TO BE STRIPPED FROM OUTPUT\n";
+    open TRASH, '<', $conf->{trash}
+      or die "Can't open trash tags file!\n";
     while (<TRASH>) {
         my $lastwasrange = 0;
         my %lastrange = ( high => 0, low => 0);
@@ -483,6 +489,7 @@ sub initialize {
                          'original-tag|ot=i',
                          'renumber-tag|rt=i',
                          'renumber-subfield|rt=i',
+                         'trash|t=s',
                          'help|h',
                        );
     show_help() unless $rc;