21:0 ¬# # $Id: Encoder.pm,v 2.3 2013/09/14 07:51:59 dankogai Exp $ # package Encode::Encoder; use strict; use warnings; our $VERSION = do { my @r = ( q$Revision: 2.3 $ =~ /\d+/g ); sprintf "%d." . "%02d" x $#r, @r }; require Exporter; our @ISA = qw(Exporter); our @EXPORT_OK = qw ( encoder ); our $AUTOLOAD; use constant DEBUG => !!$ENV{PERL_ENCODE_DEBUG}; use Encode qw(encode decode find_encoding from_to); use Carp; sub new { my ( $class, $data, $encname ) = @_; unless ($encname) { $encname = Encode::is_utf8($data) ? 'utf8' : ''; } else { my $obj = find_encoding($encname) or croak __PACKAGE__, ": unknown encoding: $encname"; $encname = $obj->name; } my $self = { data => $data, encoding => $encname, }; bless $self => $class; } sub encoder { __PACKAGE__->new(@_) } sub data { my ( $self, $data ) = @_; if ( defined $data ) { $self->{data} = $data; return $data; } else { return $self->{data}; } } sub encoding { my ( $self, $encname ) = @_; if ($encname) { my $obj = find_encoding($encname) or confess __PACKAGE__, ": unknown encoding: $encname"; $self->{encoding} = $obj->name; return $self; } else { return $self->{encoding}; } } sub bytes { my ( $self, $encname ) = @_; $encname ||= $self->{encoding}; my $obj = find_encoding($encname) or confess __PACKAGE__, ": unknown encoding: $encname"; $self->{data} = $obj->decode( $self->{data}, 1 ); $self->{encoding} = ''; return $self; } sub DESTROY { # defined so it won't autoload. DEBUG and warn shift; } sub AUTOLOAD { my $self = shift; my $type = ref($self) or confess "$self is not an object"; my $myname = $AUTOLOAD; $myname =~ s/.*://; # strip fully-qualified portion my $obj = find_encoding($myname) or confess __PACKAGE__, ": unknown encoding: $myname"; DEBUG and warn $self->{encoding}, " => ", $obj->name; if ( $self->{encoding} ) { from_to( $self->{data}, $self->{encoding}, $obj->name, 1 ); } else { $self->{data} = $obj->encode( $self->{data}, 1 ); } $self->{encoding} = $obj->name; return $self; } use overload q("") => sub { $_[0]->{data} }, q(0+) => sub { use bytes(); bytes::length( $_[0]->{data} ) }, fallback => 1, ; 1; __END__ =head1 NAME Encode::Encoder -- Object Oriented Encoder =head1 SYNOPSIS use Encode::Encoder; # Encode::encode("ISO-8859-1", $data); Encode::Encoder->new($data)->iso_8859_1; # OOP way # shortcut use Encode::Encoder qw(encoder); encoder($data)->iso_8859_1; # you can stack them! encoder($data)->iso_8859_1->base64; # provided base64() is defined # you can use it as a decoder as well encoder($base64)->bytes('base64')->latin1; # stringified print encoder($data)->utf8->latin1; # prints the string in latin1 # numified encoder("\x{abcd}\x{ef}g")->utf8 == 6; # true. bytes::length($data) =head1 ABSTRACT B allows you to use Encode in an object-oriented style. This is not only more intuitive than a functional approach, but also handier when you want to stack encodings. Suppose you want your UTF-8 string converted to Latin1 then Base64: you can simply say my $base64 = encoder($utf8)->latin1->base64; instead of my $latin1 = encode("latin1", $utf8); my $base64 = encode_base64($utf8); or the lazier and more convoluted my $base64 = encode_base64(encode("latin1", $utf8)); =head1 Description Here is how to use this module. =over 4 =item * There are at least two instance variables stored in a hash reference, {data} and {encoding}. =item * When there is no method, it takes the method name as the name of the encoding and encodes the instance I with I. If successful, the instance I is set accordingly. =item * You can retrieve the result via -Edata but usually you don't have to because the stringify operator ("") is overridden to do exactly t