X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/49069763e748ef46dcb2587647a6f5c26bf04214..d96b16650836046f79dc76641681445f97bd60ce:/tools/mkcountries-geonames

diff --git a/tools/mkcountries-geonames b/tools/mkcountries-geonames
index 0eaf880..d9b938a 100755
--- a/tools/mkcountries-geonames
+++ b/tools/mkcountries-geonames
@@ -11,6 +11,23 @@ while (<>) {
 	$cc{ lc $iso } = [ $name, $class ];
 }
 
+for (values %cc) {
+	for ($_->[2] //= $_->[0]) {
+		s/,.*//;
+		s/(?<=.)\(.*\)\s*//;
+		s/ republic\b//gi;
+		s/ islands?\b//gi;
+		s/\bthe //g;
+		s/ and / & /g and s/(?<=.)[a-z ]+//g;
+		s/\bsaint /st /gi;
+		s/South(?:ern)? /S-/g;
+		s/North(?:ern)? /N-/g;
+		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)
+	}
+}
+
 say "# automatically generated by $0";
 use Data::Dump 'dd';
 $Data::Dump::INDENT = '';