From 2660b87046558453cc856ab885070b1477d7e541 Mon Sep 17 00:00:00 2001
From: Don McMorris <dmcmorris@esilibrary.com>
Date: Tue, 5 Aug 2008 21:13:43 +0000
Subject: [PATCH] miker's script

---
 miker-filter_incumbents.pl |   48 ++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 48 insertions(+), 0 deletions(-)
 create mode 100644 miker-filter_incumbents.pl

diff --git a/miker-filter_incumbents.pl b/miker-filter_incumbents.pl
new file mode 100644
index 0000000..c68b9c7
--- /dev/null
+++ b/miker-filter_incumbents.pl
@@ -0,0 +1,48 @@
+#!/usr/bin/perl
+
+use Time::HiRes qw/time/;
+use MARC::Record;
+use MARC::File::XML ( BinaryEncoding => 'utf-8' );
+
+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;
+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="901" ind1=" " ind2=" "><subfield code="a">(\d+)</;
+	if ( $id{$1} ) {
+		print $I $_;
+		$icount++;
+	} else {
+		print $S $_;
+		$scount++;
+	}
+	$count++;
+
+	unless ($count && $count % 100) {
+		print STDERR "\r$count\t(shelved: $scount, import: $icount)\t". $count / (time - $starttime);
+	}
+
+}
+		
-- 
1.7.2.5