Take some unknown locations (city and state) and call out to google to get some of the pertinent info

google_maps_locality.pl

#!/usr/bin/perl
use strict;
use Data::Dumper;
use JSON;

sub best_candidate {
	my ($json_obj) = @_;

	for( @{$json_obj->{results}} ){
		next if $_->{types}->[0] eq 'route';

		return $_ if( grep /locality|political/, @{$_->{types}} );
	}
	return undef;
}

sub address_components {
	my ($json_obj) = @_;

	my ( $city, $county, $state, $country );

	my $candidate = best_candidate( $json_obj );

	COMPONENT: for( @{$candidate->{address_components}} ){
		# Get the State?
		if( $_->{types}->[0] eq 'administrative_area_level_1' ){
			$state = $_->{short_name};
			next COMPONENT;
		}

		if( $_->{types}->[0] eq 'administrative_area_level_2' && $_->{types}->[1] eq 'political' ){
			$county = $_->{short_name};
			next COMPONENT;
		}

		if( $_->{types}->[0] eq 'locality' && $_->{types}->[1] eq 'political' ){
			$city = $_->{short_name};
			next COMPONENT;
		}

		if( $_->{types}->[0] eq 'country' && $_->{types}->[1] eq 'political' ){
			$country = $_->{short_name};
			next COMPONENT;
		}
	}

	return( $city, $county, $state, $country );
}

sub get_lat_long {
	my ($json_obj) = @_;
	return ( $json_obj->{results}->[0]->{geometry}->{location}->{lat}, $json_obj->{results}->[0]->{geometry}->{location}->{lng} );
}

sub get_json_obj {
	my $_ = shift;
	#print "Searching for [$_]\n";
	my @json_raw = `curl "http://maps.googleapis.com/maps/api/geocode/json?address=${_}&sensor=false&components=locality" 2>/dev/null`;

	chomp for @json_raw;

	# Get the components
	my $s = join( "\n", @json_raw );
	my $o = decode_json( $s );


}

my @entries = <>;
chomp for @entries;

LINE: for( @entries ){
	chomp;

	my $orig = $_;

	s/.*?,.*?,(.*)/$1/;

	my $o = get_json_obj( $_ );

	if( $o->{status} eq 'OVER_QUERY_LIMIT' ){
		warn( "Over Query Limit - sleeping for 5 seconds..\n" );
		sleep 5;
		redo LINE;
	}

	unless( $o->{status} eq 'OK' ){
		warn( "Failed to parse [$_], skipping...\n" );
		print "E:$_\n";
		next LINE;
	}

	my ($lat, $lng) = get_lat_long( $o );

	my ($city, $county, $state, $country) = address_components( $o );

	unless( $city ){
		$city = $county || $state;
	}

	unless( $country eq 'US' ){
		print Dumper( $o ),"\n";
		print "E:$orig\n";
		next LINE;
	}

	print "O:$city,$state,$country,$lat,$lng\n";

	unless( $lat ){
		print "E:$orig\n";
		next LINE;
	}
}