From c83ef1811e7455381f3da5094692af7e84c48369 Mon Sep 17 00:00:00 2001
From: Ian Young <ian@iay.org.uk>
Date: Tue, 12 Apr 2011 09:58:12 +0000
Subject: [PATCH] Prototype script to verify entity IDs against known
 registries.  Final version needs to use the public suffix list.

---
 attic/sift_entityids.pl | 64 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 64 insertions(+)
 create mode 100755 attic/sift_entityids.pl

diff --git a/attic/sift_entityids.pl b/attic/sift_entityids.pl
new file mode 100755
index 00000000..71c5127f
--- /dev/null
+++ b/attic/sift_entityids.pl
@@ -0,0 +1,64 @@
+#!/usr/bin/perl
+
+@registries = (
+	# gTLDs
+	'.com',
+	'.edu',
+	'.net',
+	'.org',
+	'.info',
+	
+	# ccTLDs which allow top-level registration
+	'.es',
+	'.eu',
+	'.nl',
+	'.tv',
+	
+	# ccTLD: cn
+	'.edu.cn',
+	
+	# ccTLD: jp
+	'.ac.jp',
+	
+	# ccTLD: my
+	'.edu.my',
+	
+	# CC TLD: uk
+	'.ac.uk',
+	'.bl.uk',
+	'.co.uk',
+	'.gov.uk',
+	'.org.uk',
+	'.parliament.uk',
+);
+
+LINE: while (<>) {
+	chop;
+
+	#
+	# Extract a domain from the entityID
+	#
+	if (/^https?:\/\/([^\:\/]+)/) {
+		$domain = $1;
+	} elsif (/^urn:mace:ac.uk:sdss.ac.uk:provider:(service|identity):([^:]+)/) {
+		$domain = $2;
+	} elsif (/^urn:mace:eduserv.org.uk:athens:federation:(uk|beta)$/) {
+		$domain = 'eduserv.org.uk';
+	} elsif (/^urn:mace:eduserv.org.uk:athens:provider:(.*)/) {
+		$domain = $1;
+	} else {
+		print "*** can't extract domain from $_\n";
+		next;
+	}
+	
+	#
+	# Now figure out the registrar involved with this domain.
+	#
+	foreach $registry (@registries) {
+		if (substr($domain, -length($registry)) eq $registry) {
+			# print "$domain matched $registry\n";
+			next LINE;
+		}
+	}
+	print "*** no registry match for $domain\n";
+}