view release on metacpan or  search on metacpan

lib/AI/MicroStructure/RemoteList.pm  view on Meta::CPAN

    no strict 'refs';

    my $src = ${"$class\::Remote"}{source};
    if ( ref $src eq 'ARRAY' ) {
        return @$src;
    }
    elsif ( ref $src eq 'HASH' ) {
        return grep $_,
              defined $_[1] && $_[1] ne ':all'
            ? ref $_[1] ? @$src{ @{ $_[1] } }
                        : $src->{ $_[1] }
            : values %$src;
    }
    return $src;
}

sub has_remotelist { return defined $_[0]->source(); }

# main method: return the list from the remote source
sub remote_list {
    my $class = ref $_[0] || $_[0];
    return unless $class->has_remotelist();

    # check that we can access the network
    eval {
        require LWP::UserAgent;
        die "version 5.802 required ($LWP::VERSION installed)\n"
            if $LWP::VERSION < 5.802;
    };
    if ($@) {
        carp "LWP::UserAgent not available: $@";
        return;
    }

    # figure out the default category (for an instance)
    my $category = ref $_[0] ? $_[1] || $_[0]->{category} : $_[1];

    # fetch the content
    my @items;
    my @srcs = $class->sources($category);
    my $ua   = LWP::UserAgent->new( env_proxy => 1 );
    foreach my $src (@srcs) {
        my $request = HTTP::Request->new(
            ref $src
            ? ( POST => $src->[0],
                [ content_type => 'application/x-www-form-urlencoded' ],
                $src->[1]
                )
            : ( GET => $src )
        );

        my $res = $ua->request( $request );
        if ( ! $res->is_success() ) {
            carp "Failed to get content at $src (" . $res->status_line();
            return;
        }

        # extract, cleanup and return the data
        # if decoding the content fails, we just deal with the raw content
        push @items =>
            $class->extract( $res->decoded_content() || $res->content(),
               $category || () );

    }

    # return unique items
    my %seen;
    return grep { !$seen{$_}++ } @items;
}

#
# transformation subroutines
#
sub tr_nonword {
    my $str = shift;
    $str =~ tr/a-zA-Z0-9_/_/c;
    $str;
}

sub tr_accent {
    my $str = shift;
    $str =~ tr{ÀÁÂÃÄÅÇÈÉÊËÌÍÎÏÑÒÓÔÕÖØÙÚÛÜÝàáâãäåçèéêëìíîïñòóôõöøùúûüýÿ}
              {AAAAAACEEEEIIIINOOOOOOUUUUYaaaaaaceeeeiiiinoooooouuuuyy};
    return $str;
}

my %utf2asc = (
    "æ"        => 'ae',
    "Æ"        => 'AE',
    "\xc5\xa0" => 'S',
    "\x{0160}" => 'S',
    # for pokemons
    "\x{0101}"     => 'a',
    "\x{012b}"     => 'i',
    "\x{014d}"     => 'o',
    "\x{016b}"     => 'u',
    "\xe2\x99\x80" => 'female',
    "\xe2\x99\x82" => 'male',
    "\x{2640}"     => 'female',
    "\x{2642}"     => 'male',
);
my $utf_re = qr/(@{[join( '|', sort keys %utf2asc )]})/;

sub tr_utf8_basic {
    my $str = shift;
    $str =~ s/$utf_re/$utf2asc{$1}/go;
    return $str;
}

1;

__END__

=head1 NAME

AI::MicroStructure::RemoteList - Retrieval of a remote source for a structure

=head1 SYNOPSIS

    package AI::MicroStructure::contributors;
    use strict;