tools/mkcountries-geonames

   1 #!/usr/bin/env perl
   2 use 5.012;
   3 use warnings;
   4
   5 my %cc;  # map of country code to info array
   6
   7 while (<>) {
   8         /^#/ and next;  # skip comments
   9         my ($iso, $name, $cont) = (split /\t/)[0, 4, 8];
  10         my $class = "c-\L$cont";
  11         $cc{ lc $iso } = [ $name, $class ];
  12 }
  13
  14 for (values %cc) {
  15         for ($_->[2] //= $_->[0]) {
  16                 s/,.*//;
  17                 s/(?<=.)\(.*\)\s*//;
  18                 s/ republic\b//gi;
  19                 s/ islands?\b//gi;
  20                 s/\bthe //g;
  21                 s/ and / & /g and s/(?<=.)[a-z ]+//g;
  22                 s/\bsaint /st /gi;
  23                 s/South(?:ern)? /S-/g;
  24                 s/North(?:ern)? /N-/g;
  25                 s/New /n./g;
  26                 s/(\S)(\S+)-/$1-/g;  # strip most chars preceding dash
  27                 s/(\S{4}[b-df-hj-np-tv-xz])((?<!Austr)(?!land)\w{2,})/$1./g;  # abbreviate (at consonant)
  28         }
  29 }
  30
  31 say "# automatically generated by $0";
  32 use Data::Dump 'dd';
  33 $Data::Dump::INDENT = '';
  34 dd \%cc;
  35
  36 __END__
  37
  38 =head1 NAME
  39
  40 mkcountries-geonames - Create Perl include of country info from GeoNames data
  41
  42 =head1 SYNOPSIS
  43
  44         curl http://download.geonames.org/export/dump/countryInfo.txt |
  45         tools/mkcountryinfo > countries.inc.pl
  46