From: Jason Etheridge <jason@esilibrary.com>
Date: Wed, 24 Sep 2008 14:22:47 +0000 (+0000)
Subject: quick and dirty
X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=commitdiff_plain;h=5b673d2e1aeff09ecfb4a3bf2f8652d72e1cb967

quick and dirty
---

diff --git a/filter1per.pl b/filter1per.pl
new file mode 100755
index 0000000..ff2b20f
--- /dev/null
+++ b/filter1per.pl
@@ -0,0 +1,56 @@
+#!/usr/bin/perl
+
+use Time::HiRes qw/time/;
+use MARC::Record;
+use MARC::File::XML ( BinaryEncoding => 'utf-8' );
+
+my $tag = shift;
+my $subfield = shift;
+my $idfile = shift;
+my $marcfile = shift;
+my $found = shift;
+my $notfound = shift;
+if (! ($tag && $subfield && $idfile && $marcfile && $found ) ) {
+    print "filter1per.pl <tag> <subfield> <idfile> <marcfile> <output.found> [<output.notfound>]\n";
+    exit 0;
+}
+
+my %id;
+
+open F, "<$idfile";
+while (<F>) {
+	chomp;
+	$id{$_} = 1;
+}
+
+close F;
+
+my $M;
+open $M, '<:utf8', $marcfile;
+open $I, '>:utf8', $found;
+if ($notfound) { open $S, '>:utf8', $notfound; }
+
+my $starttime = time;
+my $count = 0;
+my $icount = 0;
+my $scount = 0;
+while (<$M>) {
+
+	/tag="$tag" ind1=" " ind2=" ">.*?<subfield code="$subfield">(\d+)</;
+	if ( $id{$1} ) {
+		print $I $_;
+		$icount++;
+	} else {
+        if ($notfound) {
+    		print $S $_;
+        }
+   		$scount++;
+	}
+	$count++;
+
+	unless ($count && $count % 100) {
+		print STDERR "\r$count\t(notfoundd: $scount, found: $icount)\t". $count / (time - $starttime);
+	}
+
+}
+print STDERR "\n";