X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/c133e6ad3027e095811931f3e0d21b818865e0e8..856d772d7f54045dbd78ef73f1d22c1bfe78e165:/tools/mkcountries-geonames

diff --git a/tools/mkcountries-geonames b/tools/mkcountries-geonames
index b7fb27e..56f85c4 100755
--- a/tools/mkcountries-geonames
+++ b/tools/mkcountries-geonames
@@ -1,5 +1,5 @@
 #!/usr/bin/env perl
-use 5.012;
+use 5.014;
 use warnings;
 
 my %cc;  # map of country code to info array
@@ -61,7 +61,8 @@ my %cc;  # map of country code to info array
 	oa => ["not used: African Intellectual Property Organization", "Xi", "OAPI"], # c-af
 	wo => ["not used: World Intellectual Property Organization", "Xi", "WIPO"],
 
-	# unicode identifiers
+	# common user-assigned
+	xz => ["UN/LOCODE semantics: international waters", "Co Xi", "international"],
 	qo => ["Unicode semantics: Outlying Oceania", "c-oc Co Xi", "Oceania"],
 	qu => ["Unicode semantics: European Union deprecated reserve", "c-eu Co Xi", "EU", 'eu'],
 	zz => ["Unicode semantics: unknown or invalid territory", "Co Xi","unknown"],
@@ -89,11 +90,13 @@ for (values %cc) {
 		s/ and / & /g and s/(?<=.)[a-z ]+//g;
 		s/ of / /g;
 		s/\bsa?int /st /gi;
-		s/United /Un. /gi;
+		s/Un\Kited /. /gi;
 		s/South(?:ern)? /S-/g;
 		s/North(?:ern)? /N-/g;
+		s/West(?:ern)? /W-/g;
+		s/East(?:ern)? /E-/g;
 		s/New /n./g;
-		s/(\S)(\S+)-/$1-/g;  # strip most chars preceding dash
+		s/(\w)(\w+)-/$1-/g;  # strip most chars preceding dash
 		s/(\S{4}[b-df-hj-np-tv-xz])((?<!Austr)(?!land)\w{2,})/$1./g;  # abbreviate (at consonant)
 		$_;
 	};