1#
2# $Id: Encoder.pm,v 2.0 2004/05/16 20:55:17 dankogai Exp $
3#
4package Encode::Encoder;
5use strict;
6use warnings;
7our $VERSION = do { my @r = (q$Revision: 2.0 $ =~ /\d+/g); sprintf "%d."."%02d"  x $#r, @r };
8
9require Exporter;
10our @ISA = qw(Exporter);
11our @EXPORT_OK = qw ( encoder );
12
13our $AUTOLOAD;
14sub DEBUG () { 0 }
15use Encode qw(encode decode find_encoding from_to);
16use Carp;
17
18sub new{
19    my ($class, $data, $encname) = @_;
20    unless($encname){
21	$encname = Encode::is_utf8($data) ? 'utf8' : '';
22    }else{
23	my $obj = find_encoding($encname)
24	    or croak __PACKAGE__, ": unknown encoding: $encname";
25	$encname = $obj->name;
26    }
27    my $self = {
28		data     => $data,
29		encoding => $encname,
30	       };
31    bless $self => $class;
32}
33
34sub encoder{ __PACKAGE__->new(@_) }
35
36sub data{
37    my ($self, $data) = @_;
38    if (defined $data){
39	$self->{data} = $data;
40	return $data;
41    }else{
42	return $self->{data};
43    }
44}
45
46sub encoding{
47    my ($self, $encname) = @_;
48    if ($encname){
49	my $obj = find_encoding($encname)
50	    or confess __PACKAGE__, ": unknown encoding: $encname";
51	$self->{encoding} = $obj->name;
52	return $self;
53    }else{
54	return $self->{encoding}
55    }
56}
57
58sub bytes {
59    my ($self, $encname) = @_;
60    $encname ||= $self->{encoding};
61    my $obj = find_encoding($encname)
62	    or confess __PACKAGE__, ": unknown encoding: $encname";
63    $self->{data} = $obj->decode($self->{data}, 1);
64    $self->{encoding} = '' ;
65    return $self;
66}
67
68sub DESTROY{ # defined so it won't autoload.
69    DEBUG and warn shift;
70}
71
72sub AUTOLOAD {
73    my $self = shift;
74    my $type = ref($self)
75	or confess "$self is not an object";
76    my $myname = $AUTOLOAD;
77    $myname =~ s/.*://;   # strip fully-qualified portion
78    my $obj = find_encoding($myname)
79	    or confess __PACKAGE__, ": unknown encoding: $myname";
80    DEBUG and warn $self->{encoding}, " => ", $obj->name;
81    if ($self->{encoding}){
82	from_to($self->{data}, $self->{encoding}, $obj->name, 1);
83    }else{
84	$self->{data} = $obj->encode($self->{data}, 1);
85    }
86    $self->{encoding} = $obj->name;
87    return $self;
88}
89
90use overload
91    q("") => sub { $_[0]->{data} },
92    q(0+) => sub { use bytes (); bytes::length($_[0]->{data}) },
93    fallback => 1,
94    ;
95
961;
97__END__
98
99=head1 NAME
100
101Encode::Encoder -- Object Oriented Encoder
102
103=head1 SYNOPSIS
104
105  use Encode::Encoder;
106  # Encode::encode("ISO-8859-1", $data);
107  Encode::Encoder->new($data)->iso_8859_1; # OOP way
108  # shortcut
109  use Encode::Encoder qw(encoder);
110  encoder($data)->iso_8859_1;
111  # you can stack them!
112  encoder($data)->iso_8859_1->base64;  # provided base64() is defined
113  # you can use it as a decoder as well
114  encoder($base64)->bytes('base64')->latin1;
115  # stringified
116  print encoder($data)->utf8->latin1;  # prints the string in latin1
117  # numified
118  encoder("\x{abcd}\x{ef}g")->utf8 == 6; # true. bytes::length($data)
119
120=head1 ABSTRACT
121
122B<Encode::Encoder> allows you to use Encode in an object-oriented
123style.  This is not only more intuitive than a functional approach,
124but also handier when you want to stack encodings.  Suppose you want
125your UTF-8 string converted to Latin1 then Base64: you can simply say
126
127  my $base64 = encoder($utf8)->latin1->base64;
128
129instead of
130
131  my $latin1 = encode("latin1", $utf8);
132  my $base64 = encode_base64($utf8);
133
134or the lazier and more convoluted
135
136  my $base64 = encode_base64(encode("latin1", $utf8));
137
138=head1 Description
139
140Here is how to use this module.
141
142=over 4
143
144=item *
145
146There are at least two instance variables stored in a hash reference,
147{data} and {encoding}.
148
149=item *
150
151When there is no method, it takes the method name as the name of the
152encoding and encodes the instance I<data> with I<encoding>.  If successful,
153the instance I<encoding> is set accordingly.
154
155=item *
156
157You can retrieve the result via -E<gt>data but usually you don't have to
158because the stringify operator ("") is overridden to do exactly that.
159
160=back
161
162=head2 Predefined Methods
163
164This module predefines the methods below:
165
166=over 4
167
168=item $e = Encode::Encoder-E<gt>new([$data, $encoding]);
169
170returns an encoder object.  Its data is initialized with $data if
171present, and its encoding is set to $encoding if present.
172
173When $encoding is omitted, it defaults to utf8 if $data is already in
174utf8 or "" (empty string) otherwise.
175
176=item encoder()
177
178is an alias of Encode::Encoder-E<gt>new().  This one is exported on demand.
179
180=item $e-E<gt>data([$data])
181
182When $data is present, sets the instance data to $data and returns the
183object itself.  Otherwise, the current instance data is returned.
184
185=item $e-E<gt>encoding([$encoding])
186
187When $encoding is present, sets the instance encoding to $encoding and
188returns the object itself.  Otherwise, the current instance encoding is
189returned.
190
191=item $e-E<gt>bytes([$encoding])
192
193decodes instance data from $encoding, or the instance encoding if
194omitted.  If the conversion is successful, the instance encoding
195will be set to "".
196
197The name I<bytes> was deliberately picked to avoid namespace tainting
198-- this module may be used as a base class so method names that appear
199in Encode::Encoding are avoided.
200
201=back
202
203=head2 Example: base64 transcoder
204
205This module is designed to work with L<Encode::Encoding>.
206To make the Base64 transcoder example above really work, you could
207write a module like this:
208
209  package Encode::Base64;
210  use base 'Encode::Encoding';
211  __PACKAGE__->Define('base64');
212  use MIME::Base64;
213  sub encode{
214      my ($obj, $data) = @_;
215      return encode_base64($data);
216  }
217  sub decode{
218      my ($obj, $data) = @_;
219      return decode_base64($data);
220  }
221  1;
222  __END__
223
224And your caller module would be something like this:
225
226  use Encode::Encoder;
227  use Encode::Base64;
228
229  # now you can really do the following
230
231  encoder($data)->iso_8859_1->base64;
232  encoder($base64)->bytes('base64')->latin1;
233
234=head2 Operator Overloading
235
236This module overloads two operators, stringify ("") and numify (0+).
237
238Stringify dumps the data inside the object.
239
240Numify returns the number of bytes in the instance data.
241
242They come in handy when you want to print or find the size of data.
243
244=head1 SEE ALSO
245
246L<Encode>,
247L<Encode::Encoding>
248
249=cut
250