countries: only set short name if different

[sheet.git] / tools / mkcountries-geonames

diff --git a/tools/mkcountries-geonames b/tools/mkcountries-geonames
index b10bb83705bf8c2d3c3ce77314a44db65902be60..32883ec9511694ad739e7180c2b3a27843a1df01 100755 (executable)
--- a/tools/mkcountries-geonames
+++ b/tools/mkcountries-geonames
@@ -52,17 +52,18 @@ my %cc;  # map of country code to info array
 
 while (<>) {
        /^#/ and next;  # skip comments
-       my ($iso, $name, $cont) = (split /\t/)[0, 4, 8];
-       $cont =~ s/\AAN\z/aa/;  # different antarctica abbreviation
-       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]) {
+       my $abbr = do {
+               local $_ = $_->[2] // $_->[0];
                s/,.*//;
                s/(?<=.)\(.*\)\s*//;
                s/ republic\b//gi;
@@ -77,7 +78,9 @@ for (values %cc) {
                s/New /n./g;
                s/(\S)(\S+)-/$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)
-       }
+               $_;
+       };
+       $_->[2] = $abbr if $abbr ne $_->[0];  # short name if different
 }
 
 say "# automatically generated by $0";