X-Git-Url: http://git.shiar.net/sheet.git/blobdiff_plain/c133e6ad3027e095811931f3e0d21b818865e0e8..3f38a7192491a73a1527179cdd68218f023aed61:/tools/mkcountries-geonames?ds=inline

diff --git a/tools/mkcountries-geonames b/tools/mkcountries-geonames
index b7fb27e..13f6e65 100755
--- a/tools/mkcountries-geonames
+++ b/tools/mkcountries-geonames
@@ -1,7 +1,9 @@
 #!/usr/bin/env perl
-use 5.012;
+use 5.014;
 use warnings;
 
+our $VERSION = '1.04';
+
 my %cc;  # map of country code to info array
 %cc = (
 	# transitional reservations
@@ -16,15 +18,16 @@ my %cc;  # map of country code to info array
 	# exceptional reservations
 	ac => ["exceptionally reserved: Ascension Island", "c-oc Xr", "Ascension Island", 'sh'],
 	cp => ["exceptionally reserved: Clipperton Island", "c-na Xr", "Clipperton Island"],
+	cq => ["exceptionally reserved by gb: Island of Sark", "c-eu Xr", "Island of Sark", 'gg'],
 	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", "Eurozone"],
-	fx => ["exceptionally reserved: Metropolitan France", "c-eu Xr", "Metropolitan France", 'fr'],
+	fx => ["exceptionally reserved by fr: 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"],
 	ta => ["exceptionally reserved: Tristan da Cunha", "c-oc Xr", "Tristan da Cunha", 'sh'],
-	uk => ["exceptionally reserved: United Kingdom", "c-eu Xr", "(United Kingdom)", 'gb'],
+	uk => ["exceptionally reserved by gb: United Kingdom", "c-eu Xr", "(United Kingdom)", 'gb'],
 	un => ["exceptionally reserved: United Nations", "Xi Xr", "United Nations"],
 
 	# indeterminate reservations
@@ -61,10 +64,20 @@ 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
+	xa => ["user-assigned by Russia: Abkhazia", "c-az Co Xi", "Abkhazia", 'ge'], # also IC alias by CH
+	xo => ["user-assigned by Russia: South Ossetia", "c-az Co Xi", "South Ossetia", 'ge'],
+	xi => ["user-assigned for UK/EU trade: Northern Ireland", "c-eu Co Xi", "Northern Ireland", 'gb'],
+	xn => ["user-assigned by WIPO: Nordic Patent Institute", "c-eu Co Xi", "Nordic", 'dk is no se'],
+	xu => ["user-assigned by WIPO: international org UPOV", "Co Xi", "UPOV"],
+	xv => ["user-assigned by WIPO: Visegrad Patent Institute", "c-eu Co Xi", "Visegrad", 'cz hu pl sk'],
+	xx => ["user-assigned by WIPO: unknown state or other entity", "Co Xi", "other"],
+	xz => ["UN/LOCODE semantics: international waters", "Co Xi", "international"],
+	qm => ["user-assigned by ISRC: United States alternative", "c-na Co Xi", "USA"],
 	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"],
+	oo => ["escape to indicate additional code", "Co Xi", "escape"],
 );
 
 while (<>) {
@@ -85,15 +98,17 @@ for (values %cc) {
 		s/(?<=.)\(.*\)\s*//;
 		s/ republic\b//gi;
 		s/ islands?\b//gi;
-		s/\bthe //g;
+		s/\bthe //gi;
 		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)
 		$_;
 	};