command line should be able to modify presets (fingerprinter --incoming -o someothero...

[migration-tools.git] / fingerprinter

diff --git a/fingerprinter b/fingerprinter
index 4118fd3..3b7e0ef 100755 (executable)
--- a/fingerprinter
+++ b/fingerprinter
@@ -5,8 +5,6 @@ use open ':utf8';
 
 use Getopt::Long;
 use MARC::Batch;
-use MARC::File::XML ( BinaryEncoding => 'utf-8' );
-use MARC::Field;
 use Unicode::Normalize;
 
 my $conf  = {}; # configuration hashref
@@ -75,8 +73,8 @@ sub populate_marc {
     # date1, date2
     my $my_008 = $record->field('008');
     $my_008 = $my_008->as_string() if ($my_008);
-    unless (length $my_008 == 40)
-      { print XF ">> Bad 008 field length in rec. $id\n"; return \%marc }
+    unless (defined $my_008 and length $my_008 == 40)
+      { print XF ">> Bad 008 length in rec ",$marc{id},"\n"; return \%marc }
     $marc{date1} = substr($my_008,7,4) if ($my_008);
     $marc{date2} = substr($my_008,11,4) if ($my_008); # UNUSED
 
@@ -113,7 +111,7 @@ sub populate_marc {
     $marc{desc} = $record->field('300')->subfield('a') if $record->field('300');
     $marc{pages} = $1 if (defined $marc{desc} and $marc{desc} =~ /(\d+)/);
     $marc{title} = $record->field('245')->subfield('a')
-      if defined $record->field('245');
+      if $record->field('245');
     $marc{edition} = $record->field('250')->subfield('a')
       if $record->field('250');
     if ($record->field('260')) {
@@ -261,9 +259,11 @@ sub dump_exception {
     }
     print XF "Record ", $marc->{id}, " did not make the cut: ";
     print XF "Missing item_form. " unless ($marc->{item_form});
-    print XF "Missing date1. " unless (defined $marc->{date1});
-    print XF "Invalid date1: ", $marc->{date1}
-      unless (defined $marc->{date1} and $marc->{date1} =~ /\d{4}/);
+    unless (defined $marc->{date1})
+      { print XF "Missing date1. " }
+    else
+      { print XF "Invalid date1: ", $marc->{date1}, " "
+          unless ($marc->{date1} =~ /\d{4}/); }
     print XF "Missing record_type. " unless ($marc->{record_type});
     print XF "Missing bib_lvl. " unless ($marc->{bib_lvl});
     print XF "Missing title. " unless ($marc->{title});
@@ -285,7 +285,22 @@ sub initialize {
     # set mode on existing filehandles
     binmode(STDIN, ':utf8');
 
+    # set defaults if told to do so
+    if ($c->{incoming}) {
+        $c->{tag} = 903;
+        $c->{subfield} = 'a';
+        $c->{output} = 'incoming.fp';
+        $c->{exception} = 'incoming.ex';
+    } elsif ($c->{incumbent}) {
+        $c->{tag} = 901;
+        $c->{subfield} = 'c';
+        $c->{output} = 'incumbent.fp';
+        $c->{exception} = 'incumbent.ex';
+    }
+
     my $rc = GetOptions( $c,
+                         'incoming',
+                         'incumbent',
                          'exception|x=s',
                          'output|o=s',
                          'runtype|r=s',
@@ -295,18 +310,16 @@ sub initialize {
                          'help|h',
                        );
     show_help() unless $rc;
+    show_help() if ($c->{help});
 
     my @keys = keys %{$c};
     show_help() unless (@ARGV and @keys);
-    for my $key ('runtype', 'tag', 'subfield', 'output', 'exception') {
-        push @missing, $key unless $c->{$key}
-    }
+    for my $key ('runtype', 'tag', 'subfield', 'output', 'exception')
+      { push @missing, $key unless $c->{$key} }
     if (@missing) {
         print "Required option: ", join(', ', @missing), " missing!\n";
         show_help();
     }
-
-    show_help() if ($c->{help});
 }
 
 
@@ -316,7 +329,7 @@ sub initialize {
 
 sub progress_ticker {
     return if $conf->{quiet};
-    printf("> %d (%d/s)\r", $count, ($count / (time - $start)))
+    printf("> %d (%d/s)\r", $count, ($count / (time - $start + 1)))
       if ($count % 100 == 0);
 }
 
@@ -328,7 +341,7 @@ Display usage message when things go wrong
 
 sub show_help {
 print <<HELP;
-Usage is: fingerprinter [REQUIRED ARGS] [OPTIONS] <filelist>
+Usage is: $0 [REQUIRED ARGS] [OPTIONS] <filelist>
 Req'd Arguments
   --runtype=(primary|full) -r  Do 'primary' or 'full' fingerprinting
   --tag=N                  -t  Which tag to use
@@ -336,7 +349,13 @@ Req'd Arguments
   --output=<file>          -o  Output filename
   --exceptions=<file>      -x  Exception report filename
 Options
-  --quiet  -q  Don't write status messages to STDOUT
+  --incoming     Set -r to 'full'; -t, -s, -o, -x to incoming defaults
+  --incumbent    Set -r to 'full'; -t, -s, -o, -x to incumbent defaults
+
+  Example: '$0 --incoming' is equivalent to
+           '$0 -r full -t 903 -s a -o incoming.fp -x incoming.ex'
+
+  --quiet    -q  Don't write status messages to STDOUT
 HELP
 exit 1;
 }