countries: reference tld if different from iso code

[sheet.git] / tools / mkcountries-geonames

diff --git a/tools/mkcountries-geonames b/tools/mkcountries-geonames
index d9b938a03ce070c4a10df882ba21107f948eddf6..91daae753722d030e40af9340c1d6221f59377c0 100755 (executable)
--- a/tools/mkcountries-geonames
+++ b/tools/mkcountries-geonames
@@ -3,14 +3,64 @@ use 5.012;
 use warnings;
 
 my %cc;  # map of country code to info array
+%cc = (
+       # exceptional reservations
+       ac => ["Ascension Island", "c-oc Xr", undef, 'sh'],
+       cp => ["Clipperton Island", "c-na Xr"],
+       dg => ["Diego Garcia", "c-as Xr", undef, 'io'],
+       ea => ["Ceuta and Melilla", "c-af Xr"],
+       eu => ["European Union", "c-eu Xr"],
+       fx => ["Metropolitan France", "c-eu Xr", undef, 'fr'],
+       ic => ["Canary Islands", "c-af Xr"],
+       su => ["former USSR", "c-eu Xr"],
+       ta => ["Tristan da Cunha", "c-oc Xr", undef, 'sh'],
+       uk => ["(United Kingdom)", "c-eu Xr", undef, 'gb'],
+
+       # indeterminate reservations
+       dy => ["(Benin)", "c-af Xr", undef, 'bj'],
+       ew => ["(Estonia)", "c-eu Xr", undef, 'ee'],
+       fl => ["(Liechtenstein)", "c-eu Xr", undef, 'li'],
+       ja => ["(Jamaica)", "c-na Xr", undef, 'jm'],
+       lf => ["Libya Fezzan", "c-af Xr", "Fezzan", 'ly'],
+       pi => ["(Philippines)", "c-as Xr", undef, 'ph'],
+       ra => ["(Argentina)", "c-sa Xr", undef, 'ar'],
+       rb => ["Bolivia/Botswana", "c-xx Xr"],
+       rc => ["(RoC)", "c-as Xr", undef, 'tw'],
+       rh => ["(Haiti)", "c-na Xr",undef, 'ht'],
+       ri => ["(Indonesia)", "c-as Xr", undef, 'id'],
+       rl => ["(Lebanon)", "c-as Xr", undef, 'lb'],
+       rm => ["(Madagascar)", "c-af Xr", undef, 'mg'],
+       rn => ["(Niger)", "c-af Xr",undef, 'ne'],
+       rp => ["(Philippines)", "c-as Xr", undef, 'ph'],
+       wg => ["(Grenada)", "c-na Xr", undef, 'gd'],
+       wl => ["(Saint Lucia)", "c-na Xr", "(Saint Luc.)", 'lc'],
+       wv => ["(Saint Vincent)", "c-na Xr", "(Saint Vin.)", 'vc'],
+       yv => ["(Venezuela)", "c-sa Xr", undef, 've'],
+
+       # WIPO, agreed not to use
+       ap => ["African Regional Industrial Property Organization", "Xi", "ARIPO"], # c-af
+       bx => ["Benelux Office for Intellectual Property", "Xi", "BOIP"], # c-eu
+       ef => ["European Community Patent Convention", "Xi", "CPC"], # c-eu
+       em => ["European Trademark Office", "Xi", "OHIM"], # c-eu
+       ep => ["European Patent Organization", "Xi", "EPOrg"], # c-eu
+       ev => ["Eurasian Patent Organization", "Xi", "EAPO"], # c-as
+       gc => ["Gulf Patent Office", "Xi", "GCCPO"], # c-as
+       ib => ["International Bureau of WIPO", "Xi", "IB WIPO"],
+       oa => ["African Intellectual Property Organization", "Xi", "OAPI"], # c-af
+       wo => ["World Intellectual Property Organization", "Xi", "WIPO"],
+);
 
 while (<>) {
        /^#/ and next;  # skip comments
-       my ($iso, $name, $cont) = (split /\t/)[0, 4, 8];
-       my $class = "c-\L$cont";
-       $cc{ lc $iso } = [ $name, $class ];
+       my ($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;
 }
 
+$cc{io}->[2] = "Chagos Islands";
+$cc{um}->[2] = "U.S. isl.";
+
 for (values %cc) {
        for ($_->[2] //= $_->[0]) {
                s/,.*//;
@@ -19,7 +69,9 @@ for (values %cc) {
                s/ islands?\b//gi;
                s/\bthe //g;
                s/ and / & /g and s/(?<=.)[a-z ]+//g;
-               s/\bsaint /st /gi;
+               s/ of / /g;
+               s/\bsa?int /st /gi;
+               s/United /Un. /gi;
                s/South(?:ern)? /S-/g;
                s/North(?:ern)? /N-/g;
                s/New /n./g;