line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Data::Decode; |
2
|
7
|
|
|
7
|
|
248032
|
use Moose; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
use Moose::Util::TypeConstraints; |
4
|
|
|
|
|
|
|
use namespace::clean -except => qw(meta); |
5
|
|
|
|
|
|
|
use Data::Decode::Exception; |
6
|
|
|
|
|
|
|
use Data::Decode::Types; |
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
our $VERSION = '0.00007_01'; |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
has decoder => ( |
11
|
|
|
|
|
|
|
is => 'ro', |
12
|
|
|
|
|
|
|
isa => 'Data::Decode::Decoder', |
13
|
|
|
|
|
|
|
required => 1, |
14
|
|
|
|
|
|
|
coerce => 1, |
15
|
|
|
|
|
|
|
); |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
sub import { |
18
|
|
|
|
|
|
|
my ( $self, @modules ) = @_; |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
foreach my $class (@modules) { |
21
|
|
|
|
|
|
|
if ($class !~ s/^\+//) { |
22
|
|
|
|
|
|
|
$class = "Data::Decode::$class"; |
23
|
|
|
|
|
|
|
Class::MOP::load_class($class); |
24
|
|
|
|
|
|
|
} |
25
|
|
|
|
|
|
|
} |
26
|
|
|
|
|
|
|
} |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
sub decode { |
29
|
|
|
|
|
|
|
my ($self, $data, $hints) = @_; |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
return () unless defined $data; |
32
|
|
|
|
|
|
|
$hints ||= {}; |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
my $ret = eval { |
35
|
|
|
|
|
|
|
$self->decoder->decode($self, $data, $hints); |
36
|
|
|
|
|
|
|
}; |
37
|
|
|
|
|
|
|
my $e; |
38
|
|
|
|
|
|
|
if ($e = Data::Decode::Exception::Deferred->caught() ) { |
39
|
|
|
|
|
|
|
# Just deferred. return () |
40
|
|
|
|
|
|
|
return (); |
41
|
|
|
|
|
|
|
} elsif ( $e = Exception::Class->caught() ) { |
42
|
|
|
|
|
|
|
# Oh, this we re-throw |
43
|
|
|
|
|
|
|
eval { $e->isa('Data::Decode::Exception') } ? |
44
|
|
|
|
|
|
|
$e->rethrow : die $e; |
45
|
|
|
|
|
|
|
} |
46
|
|
|
|
|
|
|
return $ret; |
47
|
|
|
|
|
|
|
} |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
1; |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
__END__ |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
=head1 NAME |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
Data::Decode - Pluggable Data Decoder |
56
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
=head1 SYNOPSIS |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
# simple usage (you probably won't use this form much) |
60
|
|
|
|
|
|
|
use Data::Decode qw( Encode::Guess ); |
61
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
my $decoder = Data::Decode->new( |
63
|
|
|
|
|
|
|
decoder => Data::Decode::Encode::Gues->new() |
64
|
|
|
|
|
|
|
); |
65
|
|
|
|
|
|
|
$decoder->decode($data); |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
# cascading several decoders |
68
|
|
|
|
|
|
|
use Data::Decode |
69
|
|
|
|
|
|
|
qw( HTTP::Response Encode::Guess ); |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
my $decoder = Data::Decode->new( |
72
|
|
|
|
|
|
|
decoder => [ |
73
|
|
|
|
|
|
|
Data::Decode::Encode::HTTP::Response->new(), |
74
|
|
|
|
|
|
|
Data::Decode::Encode::Guess->new(), |
75
|
|
|
|
|
|
|
] |
76
|
|
|
|
|
|
|
); |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
my $res = LWP::UserAgent->new->get("http://whatever.example.com"); |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
my $decoded = $decoder->decode($res->content, { response => $res }); |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
=head1 DESCRIPTION |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
Data::Decode implements a pluggable "decoder". The main aim is to provide |
85
|
|
|
|
|
|
|
a uniform interface to decode a given data while allowing the actual |
86
|
|
|
|
|
|
|
algorithm being used to be changed depending on your needs.. |
87
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
For now this is aimed at decoding miscellaneous text to perl's internal |
89
|
|
|
|
|
|
|
unicode encoding, but should be able to handle anything if you give it a |
90
|
|
|
|
|
|
|
proper plugin |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
=head1 DECODING TO UNICODE |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
Japanese, which is the language that I mainly deal with, has an annoying |
95
|
|
|
|
|
|
|
property: It can come in at least 4 different flavors (utf-8, shift-jis, |
96
|
|
|
|
|
|
|
euc-jp and iso-2022-jp). Even worse, vendors may have more vendor-specific |
97
|
|
|
|
|
|
|
symbols, such as the pictograms in mobile phones. |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
Ways to decode these strings into unicode varies between each environment |
100
|
|
|
|
|
|
|
and application. |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
Many modules require that the strings be normalized to unicode, but they |
103
|
|
|
|
|
|
|
all handle this normalization process differently, which is, well, not exactly |
104
|
|
|
|
|
|
|
an optimal solution. |
105
|
|
|
|
|
|
|
|
106
|
|
|
|
|
|
|
Data::Decode provides a uniform interface to this problem, and a few common |
107
|
|
|
|
|
|
|
ways decoding is handled. The actual decoding strategies are separated out |
108
|
|
|
|
|
|
|
from the surface interface, so other users who find a particular strategy to |
109
|
|
|
|
|
|
|
decode strings can then upload their way to CPAN, and everyone can benefit |
110
|
|
|
|
|
|
|
from it. |
111
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
=head1 CASCADING |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
Data::Decode comes with a simple chaining functionality. You can take as many |
115
|
|
|
|
|
|
|
decoders as you want, and you can stack them on top of each other. To enable |
116
|
|
|
|
|
|
|
this feature, just provide an array as the decoder, instead of a single object. |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
=head1 METHODS |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
=head2 new |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
Instantiates a new Data::Decode object. |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
=over 4 |
125
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
=item decoder |
127
|
|
|
|
|
|
|
|
128
|
|
|
|
|
|
|
Required. Takes in the object that encapsulates the actual decoding logic. |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
(WARNING: Subject to change - we may require an object that implements a role |
131
|
|
|
|
|
|
|
instead of just a function in the future. Beware!) The object must have a |
132
|
|
|
|
|
|
|
method named "decode", which takes in a reference to the Data::Decode object |
133
|
|
|
|
|
|
|
and a string to be decoded. An optional third parameter may be provided to |
134
|
|
|
|
|
|
|
specify any hints that could be used to figure out what to do. |
135
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
# a decode() method |
137
|
|
|
|
|
|
|
sub decode { |
138
|
|
|
|
|
|
|
my ($self, $decoder, $string, $hints) = @_; |
139
|
|
|
|
|
|
|
# $decoder = Data::Decode object |
140
|
|
|
|
|
|
|
# $string = a scalar to be decoded |
141
|
|
|
|
|
|
|
# $hints = a hashref of hints |
142
|
|
|
|
|
|
|
} |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
You may also specify the class names of the decoders -- in that case, an |
145
|
|
|
|
|
|
|
argument-less new() will be called upon the class name to instantiate the |
146
|
|
|
|
|
|
|
decoder. |
147
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
If you provide a list of decoders, Data::Decode::Chain will automatically be |
149
|
|
|
|
|
|
|
set for you. |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
my $decoder = Data::Decode->new( |
152
|
|
|
|
|
|
|
decoder => [ # This will turn into a Data::Decode::Chain object |
153
|
|
|
|
|
|
|
Decoder1->new(), |
154
|
|
|
|
|
|
|
Decoder2->new(), |
155
|
|
|
|
|
|
|
Decoder3->new(), |
156
|
|
|
|
|
|
|
... |
157
|
|
|
|
|
|
|
] |
158
|
|
|
|
|
|
|
); |
159
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
=back |
161
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
=head2 decode |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
Decodes a string. Takes in a string, and a hashref of hints to be used |
165
|
|
|
|
|
|
|
for decoding. The meaning or the usage of the hints may differ between |
166
|
|
|
|
|
|
|
the actual underlying decoders. |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
=head2 decoder |
169
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
Get the underlying decoder object. |
171
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
=head1 AUTHOR |
173
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
Daisuke Maki E<lt>daisuke@endeworks.jpE<gt> |
175
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
=head1 LICENSE |
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or modify it |
179
|
|
|
|
|
|
|
under the same terms as Perl itself. |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
See http://www.perl.com/perl/misc/Artistic.html |
182
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
=cut |