File Coverage

blib/lib/Geo/Coder/Abbreviations.pm

Criterion	Covered	Total	%
statement	50	72	69.4
branch	20	44	45.4
condition	9	15	60.0
subroutine	8	8	100.0
pod	3	3	100.0
total	90	142	63.3

package Geo::Coder::Abbreviations;

use warnings;
use strict;
use JSON::MaybeXS;
use LWP::Simple::WithCache;

=head1 NAME

Geo::Coder::Abbreviations - Quick and Dirty Interface to https://github.com/mapbox/geocoder-abbreviations

=head1 VERSION

Version 0.09

=cut

our %abbreviations;
our $VERSION = '0.09';

# This is giving 404 errors at the moment
#	https://github.com/mapbox/mapbox-java/issues/1460
# our location = 'https://raw.githubusercontent.com/mapbox/geocoder-abbreviations/master/tokens/en.json';
use constant LOCATION => 'https://raw.githubusercontent.com/allison-strandberg/geocoder-abbreviations/master/tokens/en.json';

=head1 SYNOPSIS

Provides an interface to https://github.com/mapbox/geocoder-abbreviations.
One small function for now, I'll add others later.

=head1 SUBROUTINES/METHODS

=head2 new

Creates a Geo::Coder::Abbreviations object.
It takes no arguments.
If you have L installed it will load much faster,
otherwise it will download the database from the Internet
when the class is first instantiated.

=cut

sub new {
	my $proto = shift;
	my $class = ref($proto) || $proto;

	if(!defined($class)) {
		# Using Geo::Coder::Abbreviations->new(), not Geo::Coder::Abbreviations::new()
		# carp(__PACKAGE__, ' use ->new() not ::new() to instantiate');
		# return;

		# FIXME: this only works when no arguments are given
		$class = __PACKAGE__;
	} elsif(ref($class)) {
		# clone the given object
		# return bless { %{$class}, %args }, ref($class);
		return bless { %{$class} }, ref($class);
	}

	unless(scalar keys(%abbreviations)) {
		if(eval { require HTTP::Cache::Transparent; }) {
			require File::Spec;	# That should be installed

			File::Spec->import();
			HTTP::Cache::Transparent->import();

			my $cache_dir;
			if($cache_dir = ($ENV{'CACHE_DIR'} || $ENV{'CACHEDIR'})) {
				mkdir $cache_dir, 02700 if(!-d $cache_dir);
				$cache_dir = File::Spec->catfile($cache_dir, 'http-cache-transparent');
			} else {
				# $cache_dir = File::Spec->catfile(File::Spec->tmpdir(), 'cache', 'http-cache-transparent');
				$cache_dir = File::Spec->catfile(File::HomeDir->my_home(), '.cache', 'http-cache-transparent');
			}

			HTTP::Cache::Transparent::init({
				BasePath => $cache_dir,
				# Verbose => $opts{'v'} ? 1 : 0,
				NoUpdate => 60 * 60 * 24,
				MaxAge => 30 * 24
			}) || die "$0: $cache_dir $!";
		}

		# TODO:	Support other languages
		my $data = LWP::Simple::WithCache::get(LOCATION);

		if(!defined($data)) {
			# die 'error downloading from ', LOCATION;
			$data = join('', grep(!/^\s*(#|$)/, ));
		}
		%abbreviations = map {
			my %rc = ();
			if(defined($_->{'type'}) && ($_->{'type'} eq 'way')) {
				foreach my $token(@{$_->{'tokens'}}) {
					$rc{uc($token)} = uc($_->{'canonical'});
				}
			}
			%rc;
		} @{JSON::MaybeXS->new()->utf8()->decode($data)};

		# %abbreviations = map { (defined($_->{'type'}) && ($_->{'type'} eq 'way')) ? (uc($_->{'full'}) => uc($_->{'canonical'})) : () } @{JSON::MaybeXS->new()->utf8()->decode($data)};
	}

	return bless {
		table => \%abbreviations
	}, $class;
}

=head2 abbreviate

Abbreviate a place.

    use Geo::Coder::Abbreviations;

    my $abbr = Geo::Coder::Abbreviations->new();
    print $abbr->abbreviate('Road'), "\n";	# prints 'RD'
    print $abbr->abbreviate('RD'), "\n";	# prints 'RD'

=cut

sub abbreviate {
	my $self = shift;

	return $self->{'table'}->{uc(shift)};
}

=head2 normalize

Normalize and abbreviate street names - useful for comparisons

    print $abbr->normalize({ street => '10 Downing Street' }), "\n";	# prints '10 DOWNING ST'

Can be run as a class method

    print Geo::Coder::Abbreviations('1600 Pennsylvania Avenue NW'), "\n";	# prints '1600 Pennsylvia Ave NW'

=cut

sub normalize
{
	my $self = shift;
	my %params;

	# Try hard to support whatever API that the user wants to use
	if(!ref($self)) {
		if(scalar(@_)) {
			return(__PACKAGE__->new()->normalize(@_));
		} elsif(!defined($self)) {
			# Geo::Coder::Abbreviations->normalize()
			Carp::croak('Usage: ', __PACKAGE__, '::normalize(street => $street)');
		} elsif($self eq __PACKAGE__) {
			Carp::croak("Usage: $self", '::normalize(street => $street)');
		}
		return(__PACKAGE__->new()->normalize($self));
	} elsif(ref($self) eq 'HASH') {
		return(__PACKAGE__->new()->noralize($self));
	} elsif(ref($_[0]) eq 'HASH') {
		%params = %{$_[0]};
	# } elsif(ref($_[0]) && (ref($_[0] !~ /::/))) {
	} elsif(ref($_[0])) {
		Carp::croak('Usage: ', __PACKAGE__, '::normalize(street => $street)');
	} elsif(scalar(@_) && (scalar(@_) % 2 == 0)) {
		%params = @_;
	} else {
		$params{'street'} = shift;
	}

	my $street = $params{'street'};

	$street = uc($street);
	if($street =~ /(.+)\s+(.+)\s+(.+)/) {
		my $a;
		if((lc($2) ne 'cross') && ($a = $self->abbreviate($2))) {
			$street = "$1 $a $3";
		} elsif($a = $self->abbreviate($3)) {
			$street = "$1 $2 $a";
		}
	} elsif($street =~ /(.+)\s(.+)$/) {
		if(my $a = $self->abbreviate($2)) {
			$street = "$1 $a";
		}
	}
	$street =~ s/^0+//;     # Turn 04th St into 4th St
	return $street;
}

=head1 SEE ALSO

L
L
L

=head1 AUTHOR

Nigel Horne, C<<  >>

=head1 BUGS

You may need to ensure you don't translate "Cross Street" to "X ST".
See t/abbreviations.t.

=head1 SUPPORT

You can find documentation for this module with the perldoc command.

    perldoc Geo::Coder::Abbreviations

You can also look for information at:

=over 4

=item * RT: CPAN's request tracker

L

=item * Search CPAN

L

=back

=head1 ACKNOWLEDGMENTS

=head1 LICENSE AND COPYRIGHT

Copyright 2020-2024 Nigel Horne.

This program is released under the following licence: GPL2

=cut

1; # End of Geo::Coder::Abbreviations

# https://raw.githubusercontent.com/mapbox/geocoder-abbreviations/master/tokens/en.json is giving 404 errors at the moment
# so here is a copy until it's back

__DATA__