From 4e4e2c556fb0a8299cd4b751f3538a5632ae41c9 Mon Sep 17 00:00:00 2001
From: Shawn Boyette <sboyette@esilibrary.com>
Date: Mon, 10 Nov 2008 15:11:16 +0000
Subject: [PATCH] misc

---
 filter_record.ids          |  121 ++++++++++++++++++++++++++++++++++++++++++++
 marc-cleanup               |   13 +++--
 miker-filter_incoming.pl   |    6 ++-
 miker-filter_incumbents.pl |   10 ++--
 4 files changed, 139 insertions(+), 11 deletions(-)
 create mode 100644 filter_record.ids

diff --git a/filter_record.ids b/filter_record.ids
new file mode 100644
index 0000000..99c6171
--- /dev/null
+++ b/filter_record.ids
@@ -0,0 +1,121 @@
+#!/usr/bin/perl
+use warnings;
+use strict;
+
+use Getopt::Long;
+use Time::HiRes qw/time/;
+use MARC::Record;
+use MARC::File::XML ( BinaryEncoding => 'utf-8' );
+
+# THIS FILE EXTRACTS NONMATCHING RECORDS
+
+# configuration hashref
+my $conf  = ();
+#initialize($conf);
+
+my $idfile = shift;
+my $marcfile = shift;
+my $import = shift;
+my $shelve = shift;
+
+my %id;
+
+open F, "<$idfile";
+while (<F>) {
+	chomp;
+	$id{$_} = 1;
+}
+
+close F;
+
+my $M; my $I; my $S;
+open $M, '<:utf8', $marcfile;
+open $I, '>:utf8', $import;
+open $S, '>:utf8', $shelve;
+
+my $starttime = time;
+my $count = 0;
+my $icount = 0;
+my $scount = 0;
+while (<$M>) {
+    /tag="903" ind1=" " ind2=" ">.*?<subfield code="a">(\d+)</;
+    if ( $id{$1} ) {
+        print $S $_;
+        $scount++;
+    } else {
+        print $I $_;
+        $icount++;
+    }
+    $count++;
+
+    unless ($count && $count % 100) {
+        print STDERR "\r$count\t(shelved: $scount, import: $icount)\t". $count / (time - $starttime);
+    }
+}
+
+=head2 initialize
+
+Performs boring script initialization. Handles argument parsing,
+mostly.
+
+=cut
+
+sub initialize {
+    my ($c) = @_;
+    my @missing = ();
+
+    # set mode on existing filehandles
+    binmode(STDIN, ':utf8');
+
+    my $rc = GetOptions( $c,
+                         'incoming',
+                         'incumbent',
+                         'incoming-tag|incot=i',
+                         'incoming-subfield|incos=s',
+                         'incumbent-tag|incut=i',
+                         'incumbent-subfield|incus=s',
+                         'output|o=s',
+                         'help|h',
+                       );
+    show_help() unless $rc;
+    show_help() if ($c->{help});
+
+    $c->{'incoming-tag'}         = 903;
+    $c->{'incoming-subfield'}    = 'a';
+    $c->{'incoming-matchfile'}   = '';
+    $c->{'incoming-nomatchfile'} = '';
+    $c->{'incumbent-tag'}         = 901;
+    $c->{'incumbent-subfield'}    = 'a';
+    $c->{'incumbent-matchfile'}   = '';
+    $c->{'incumbent-nomatchfile'} = '';
+    my @keys = keys %{$c};
+    show_help() unless (@ARGV and @keys);
+    for my $key ('renumber-from', 'tag', 'subfield', 'output')
+      { push @missing, $key unless $c->{$key} }
+    if (@missing) {
+        print "Required option: ", join(', ', @missing), " missing!\n";
+        show_help();
+    }
+
+}
+
+
+=head2 show_help
+
+Display usage message when things go wrong
+
+=cut
+
+sub show_help {
+print <<HELP;
+Usage is: $0 [REQUIRED ARGS]
+Req'd Arguments
+  --renumber-from=N        -rf First id# of new sequence
+  --tag=N                  -t  Which tag to use
+  --subfield=X             -s  Which subfield to use
+  --output=<file>          -o  Output filename
+
+Any number of input files may be specified; one output file will result.
+HELP
+exit 1;
+}
diff --git a/marc-cleanup b/marc-cleanup
index b3e3b73..58b59a8 100755
--- a/marc-cleanup
+++ b/marc-cleanup
@@ -33,8 +33,9 @@ open my $EXMARC, '>:utf8', $conf->{exception};
 # edit(), below
 my %commands = ( c => \&print_context,
                  C => \&print_linecontext,
-                 k => \&kill_line,
                  o => \&show_original,
+                 f => \&flip_lines,
+                 k => \&kill_line,
                  m => \&merge_lines,
                  n => \&next_line,
                  p => \&prev_line,
@@ -189,6 +190,9 @@ sub edit {
     message($msg, 1);
     print_context();
 
+    # stow original problem line
+    $conf->{origline} = $record[$recptr];
+
     while (1) {
         my $line = $term->readline('marc-cleanup>');
         my @chunks = split /\s+/, $line;
@@ -343,6 +347,7 @@ sub merge_lines {
     @record = (@a, @b);
     # move record pointer to previous line
     prev_line();
+    print_linecontext();
     return 0;
 }
 
@@ -373,14 +378,14 @@ sub prev_line {
     return 0;
 }
 
-sub commit_edit { return 1 }
-
 sub show_original {
     my ($line_in) = @_;
-    print $OUT "\n$line_in\n";
+    print $OUT "\n", $conf->{origline}, "\n";
     return 0;
 }
 
+sub commit_edit { return 1 }
+
 sub help {
 print $OUT <<HELP;
 
diff --git a/miker-filter_incoming.pl b/miker-filter_incoming.pl
index 3adb8aa..99c6171 100644
--- a/miker-filter_incoming.pl
+++ b/miker-filter_incoming.pl
@@ -7,6 +7,8 @@ use Time::HiRes qw/time/;
 use MARC::Record;
 use MARC::File::XML ( BinaryEncoding => 'utf-8' );
 
+# THIS FILE EXTRACTS NONMATCHING RECORDS
+
 # configuration hashref
 my $conf  = ();
 #initialize($conf);
@@ -39,10 +41,10 @@ while (<$M>) {
     /tag="903" ind1=" " ind2=" ">.*?<subfield code="a">(\d+)</;
     if ( $id{$1} ) {
         print $S $_;
-        $icount++;
+        $scount++;
     } else {
         print $I $_;
-        $scount++;
+        $icount++;
     }
     $count++;
 
diff --git a/miker-filter_incumbents.pl b/miker-filter_incumbents.pl
index 3ae0ab7..aa304f3 100644
--- a/miker-filter_incumbents.pl
+++ b/miker-filter_incumbents.pl
@@ -3,13 +3,13 @@ use warnings;
 use strict;
 
 use Getopt::Long;
-#use Time::HiRes qw/time/;
-#use MARC::Record;
-#use MARC::File::XML ( BinaryEncoding => 'utf-8' );
+use Time::HiRes qw/time/;
+use MARC::Record;
+use MARC::File::XML ( BinaryEncoding => 'utf-8' );
 
 # configuration hashref
 my $conf  = ();
-initialize($conf);
+#initialize($conf);
 
 my $idfile = shift;
 my $marcfile = shift;
@@ -26,7 +26,7 @@ while (<F>) {
 
 close F;
 
-my $M;
+my $M; my $I; my $S;
 open $M, '<:utf8', $marcfile;
 open $I, '>:utf8', $import;
 open $S, '>:utf8', $shelve;
-- 
1.7.2.5