X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/27fe25db96c70578d86ee1130b1d3788dd8e4c53..856d772d7f54045dbd78ef73f1d22c1bfe78e165:/tools/mkcountries-geonames

diff --git a/tools/mkcountries-geonames b/tools/mkcountries-geonames
index 2f12451..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
@@ -19,7 +19,7 @@ my %cc;  # map of country code to info array
 	dg => ["exceptionally reserved: Diego Garcia", "c-as Xr", "Diego Garcia", 'io'],
 	ea => ["exceptionally reserved: Ceuta and Melilla", "c-af Xr", "Ceuta and Melilla"],
 	eu => ["exceptionally reserved: European Union", "c-eu Xr", "European Union"],
-	ez => ["exceptionally reserved: European OTC derivatives", "c-eu Xr", "European OTC"],
+	ez => ["exceptionally reserved: European OTC derivatives", "c-eu Xr", "Eurozone"],
 	fx => ["exceptionally reserved: Metropolitan France", "c-eu Xr", "Metropolitan France", 'fr'],
 	ic => ["exceptionally reserved: Canary Islands", "c-af Xr", "Canary Islands"],
 	su => ["exceptionally reserved: former USSR", "c-eu Xr", "USSR"],
@@ -35,7 +35,7 @@ my %cc;  # map of country code to info array
 	lf => ["indeterminately reserved: Libya Fezzan", "c-af Xr", "Fezzan", 'ly'],
 	pi => ["indeterminately reserved: Philippines", "c-as Xr", "(Philippines)", 'ph'],
 	ra => ["indeterminately reserved: Argentina", "c-sa Xr", "(Argentina)", 'ar'],
-	rb => ["indeterminately reserved: Botswana, Bolivia", "c-sa Xr", "(Bots...)"],
+	rb => ["indeterminately reserved: Botswana, Bolivia", "c-sa Xr", "(Bots...)", 'bw bo'],
 	rc => ["indeterminately reserved: Republic of China", "c-as Xr", "(RoC)", 'tw'],
 	rh => ["indeterminately reserved: Haiti", "c-na Xr", "(Haiti)", 'ht'],
 	ri => ["indeterminately reserved: Indonesia", "c-as Xr", "(Indonesia)", 'id'],
@@ -60,6 +60,12 @@ my %cc;  # map of country code to info array
 	ib => ["not used: International Bureau of WIPO", "Xi", "IB"],
 	oa => ["not used: African Intellectual Property Organization", "Xi", "OAPI"], # c-af
 	wo => ["not used: World Intellectual Property Organization", "Xi", "WIPO"],
+
+	# 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"],
 );
 
 while (<>) {
@@ -67,7 +73,7 @@ while (<>) {
 	my ($iso, $name, $cont, $tld) = (split /\t/)[0, 4, 8, 9];
 	my @info = ($name, "c-\L$cont");
 	$info[3] = $tld if $tld =~ s/\A\.// and $tld ne lc $iso;
-	$cc{ lc $iso } = \@info;
+	$cc{ lc $iso } //= \@info;
 }
 
 $cc{io}->[2] = "Chagos Islands";
@@ -84,11 +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)
 		$_;
 	};