1package Encode::Unicode;
2
3use strict;
4use warnings;
5
6our $VERSION = do { my @r = ( q$Revision: 2.18 $ =~ /\d+/g ); sprintf "%d." . "%02d" x $#r, @r };
7
8use XSLoader;
9XSLoader::load( __PACKAGE__, $VERSION );
10
11#
12# Object Generator 8 transcoders all at once!
13#
14
15use Encode ();
16
17our %BOM_Unknown = map { $_ => 1 } qw(UTF-16 UTF-32);
18
19for my $name (
20    qw(UTF-16 UTF-16BE UTF-16LE
21    UTF-32 UTF-32BE UTF-32LE
22    UCS-2BE  UCS-2LE)
23  )
24{
25    my ( $size, $endian, $ucs2, $mask );
26    $name =~ /^(\w+)-(\d+)(\w*)$/o;
27    if ( $ucs2 = ( $1 eq 'UCS' ) ) {
28        $size = 2;
29    }
30    else {
31        $size = $2 / 8;
32    }
33    $endian = ( $3 eq 'BE' ) ? 'n' : ( $3 eq 'LE' ) ? 'v' : '';
34    $size == 4 and $endian = uc($endian);
35
36    my $obj = bless {
37        Name   => $name,
38        size   => $size,
39        endian => $endian,
40        ucs2   => $ucs2,
41    } => __PACKAGE__;
42    Encode::define_encoding($obj, $name);
43}
44
45use parent qw(Encode::Encoding);
46
47sub renew {
48    my $self = shift;
49    $BOM_Unknown{ $self->name } or return $self;
50    my $clone = bless {%$self} => ref($self);
51    $clone->{renewed}++;    # so the caller knows it is renewed.
52    return $clone;
53}
54
551;
56__END__
57
58=head1 NAME
59
60Encode::Unicode -- Various Unicode Transformation Formats
61
62=cut
63
64=head1 SYNOPSIS
65
66    use Encode qw/encode decode/;
67    $ucs2 = encode("UCS-2BE", $utf8);
68    $utf8 = decode("UCS-2BE", $ucs2);
69
70=head1 ABSTRACT
71
72This module implements all Character Encoding Schemes of Unicode that
73are officially documented by Unicode Consortium (except, of course,
74for UTF-8, which is a native format in perl).
75
76=over 4
77
78=item L<http://www.unicode.org/glossary/> says:
79
80I<Character Encoding Scheme> A character encoding form plus byte
81serialization. There are Seven character encoding schemes in Unicode:
82UTF-8, UTF-16, UTF-16BE, UTF-16LE, UTF-32 (UCS-4), UTF-32BE (UCS-4BE) and
83UTF-32LE (UCS-4LE), and UTF-7.
84
85Since UTF-7 is a 7-bit (re)encoded version of UTF-16BE, It is not part of
86Unicode's Character Encoding Scheme.  It is separately implemented in
87Encode::Unicode::UTF7.  For details see L<Encode::Unicode::UTF7>.
88
89=item Quick Reference
90
91                Decodes from ord(N)           Encodes chr(N) to...
92       octet/char BOM S.P d800-dfff  ord > 0xffff     \x{1abcd} ==
93  ---------------+-----------------+------------------------------
94  UCS-2BE       2   N   N  is bogus                  Not Available
95  UCS-2LE       2   N   N     bogus                  Not Available
96  UTF-16      2/4   Y   Y  is   S.P           S.P            BE/LE
97  UTF-16BE    2/4   N   Y       S.P           S.P    0xd82a,0xdfcd
98  UTF-16LE    2/4   N   Y       S.P           S.P    0x2ad8,0xcddf
99  UTF-32        4   Y   -  is bogus         As is            BE/LE
100  UTF-32BE      4   N   -     bogus         As is       0x0001abcd
101  UTF-32LE      4   N   -     bogus         As is       0xcdab0100
102  UTF-8       1-4   -   -     bogus   >= 4 octets   \xf0\x9a\af\8d
103  ---------------+-----------------+------------------------------
104
105=back
106
107=head1 Size, Endianness, and BOM
108
109You can categorize these CES by 3 criteria:  size of each character,
110endianness, and Byte Order Mark.
111
112=head2 by size
113
114UCS-2 is a fixed-length encoding with each character taking 16 bits.
115It B<does not> support I<surrogate pairs>.  When a surrogate pair
116is encountered during decode(), its place is filled with \x{FFFD}
117if I<CHECK> is 0, or the routine croaks if I<CHECK> is 1.  When a
118character whose ord value is larger than 0xFFFF is encountered,
119its place is filled with \x{FFFD} if I<CHECK> is 0, or the routine
120croaks if I<CHECK> is 1.
121
122UTF-16 is almost the same as UCS-2 but it supports I<surrogate pairs>.
123When it encounters a high surrogate (0xD800-0xDBFF), it fetches the
124following low surrogate (0xDC00-0xDFFF) and C<desurrogate>s them to
125form a character.  Bogus surrogates result in death.  When \x{10000}
126or above is encountered during encode(), it C<ensurrogate>s them and
127pushes the surrogate pair to the output stream.
128
129UTF-32 (UCS-4) is a fixed-length encoding with each character taking 32 bits.
130Since it is 32-bit, there is no need for I<surrogate pairs>.
131
132=head2 by endianness
133
134The first (and now failed) goal of Unicode was to map all character
135repertoires into a fixed-length integer so that programmers are happy.
136Since each character is either a I<short> or I<long> in C, you have to
137pay attention to the endianness of each platform when you pass data
138to one another.
139
140Anything marked as BE is Big Endian (or network byte order) and LE is
141Little Endian (aka VAX byte order).  For anything not marked either
142BE or LE, a character called Byte Order Mark (BOM) indicating the
143endianness is prepended to the string.
144
145CAVEAT: Though BOM in utf8 (\xEF\xBB\xBF) is valid, it is meaningless
146and as of this writing Encode suite just leave it as is (\x{FeFF}).
147
148=over 4
149
150=item BOM as integer when fetched in network byte order
151
152              16         32 bits/char
153  -------------------------
154  BE      0xFeFF 0x0000FeFF
155  LE      0xFFFe 0xFFFe0000
156  -------------------------
157
158=back
159
160This modules handles the BOM as follows.
161
162=over 4
163
164=item *
165
166When BE or LE is explicitly stated as the name of encoding, BOM is
167simply treated as a normal character (ZERO WIDTH NO-BREAK SPACE).
168
169=item *
170
171When BE or LE is omitted during decode(), it checks if BOM is at the
172beginning of the string; if one is found, the endianness is set to
173what the BOM says.
174
175=item *
176
177Default Byte Order
178
179When no BOM is found, Encode 2.76 and blow croaked.  Since Encode
1802.77, it falls back to BE accordingly to RFC2781 and the Unicode
181Standard version 8.0
182
183=item *
184
185When BE or LE is omitted during encode(), it returns a BE-encoded
186string with BOM prepended.  So when you want to encode a whole text
187file, make sure you encode() the whole text at once, not line by line
188or each line, not file, will have a BOM prepended.
189
190=item *
191
192C<UCS-2> is an exception.  Unlike others, this is an alias of UCS-2BE.
193UCS-2 is already registered by IANA and others that way.
194
195=back
196
197=head1 Surrogate Pairs
198
199To say the least, surrogate pairs were the biggest mistake of the
200Unicode Consortium.  But according to the late Douglas Adams in I<The
201Hitchhiker's Guide to the Galaxy> Trilogy, C<In the beginning the
202Universe was created. This has made a lot of people very angry and
203been widely regarded as a bad move>.  Their mistake was not of this
204magnitude so let's forgive them.
205
206(I don't dare make any comparison with Unicode Consortium and the
207Vogons here ;)  Or, comparing Encode to Babel Fish is completely
208appropriate -- if you can only stick this into your ear :)
209
210Surrogate pairs were born when the Unicode Consortium finally
211admitted that 16 bits were not big enough to hold all the world's
212character repertoires.  But they already made UCS-2 16-bit.  What
213do we do?
214
215Back then, the range 0xD800-0xDFFF was not allocated.  Let's split
216that range in half and use the first half to represent the C<upper
217half of a character> and the second half to represent the C<lower
218half of a character>.  That way, you can represent 1024 * 1024 =
2191048576 more characters.  Now we can store character ranges up to
220\x{10ffff} even with 16-bit encodings.  This pair of half-character is
221now called a I<surrogate pair> and UTF-16 is the name of the encoding
222that embraces them.
223
224Here is a formula to ensurrogate a Unicode character \x{10000} and
225above;
226
227  $hi = ($uni - 0x10000) / 0x400 + 0xD800;
228  $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
229
230And to desurrogate;
231
232 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
233
234Note this move has made \x{D800}-\x{DFFF} into a forbidden zone but
235perl does not prohibit the use of characters within this range.  To perl,
236every one of \x{0000_0000} up to \x{ffff_ffff} (*) is I<a character>.
237
238  (*) or \x{ffff_ffff_ffff_ffff} if your perl is compiled with 64-bit
239  integer support!
240
241=head1 Error Checking
242
243Unlike most encodings which accept various ways to handle errors,
244Unicode encodings simply croaks.
245
246  % perl -MEncode -e'$_ = "\xfe\xff\xd8\xd9\xda\xdb\0\n"' \
247         -e'Encode::from_to($_, "utf16","shift_jis", 0); print'
248  UTF-16:Malformed LO surrogate d8d9 at /path/to/Encode.pm line 184.
249  % perl -MEncode -e'$a = "BOM missing"' \
250         -e' Encode::from_to($a, "utf16", "shift_jis", 0); print'
251  UTF-16:Unrecognised BOM 424f at /path/to/Encode.pm line 184.
252
253Unlike other encodings where mappings are not one-to-one against
254Unicode, UTFs are supposed to map 100% against one another.  So Encode
255is more strict on UTFs.
256
257Consider that "division by zero" of Encode :)
258
259=head1 SEE ALSO
260
261L<Encode>, L<Encode::Unicode::UTF7>, L<http://www.unicode.org/glossary/>,
262L<http://www.unicode.org/unicode/faq/utf_bom.html>,
263
264RFC 2781 L<http://www.ietf.org/rfc/rfc2781.txt>,
265
266The whole Unicode standard L<http://www.unicode.org/unicode/uni2book/u2.html>
267
268Ch. 15, pp. 403 of C<Programming Perl (3rd Edition)>
269by Larry Wall, Tom Christiansen, Jon Orwant;
270O'Reilly & Associates; ISBN 0-596-00027-8
271
272=cut
273