3 perlunicode - Unicode support in Perl
7 WARNING: The implementation of Unicode support in Perl is incomplete.
8 Expect sudden and unannounced changes!
10 Beginning with version 5.6, Perl uses logically wide characters to
11 represent strings internally. This internal representation of strings
12 uses the UTF-8 encoding.
14 In future, Perl-level operations will expect to work with characters
15 rather than bytes, in general.
17 However, as strictly an interim compatibility measure, Perl v5.6 aims to
18 provide a safe migration path from byte semantics to character semantics
19 for programs. For operations where Perl can unambiguously decide that the
20 input data is characters, Perl now switches to character semantics.
21 For operations where this determination cannot be made without additional
22 information from the user, Perl decides in favor of compatibility, and
23 chooses to use byte semantics.
25 This behavior preserves compatibility with earlier versions of Perl,
26 which allowed byte semantics in Perl operations, but only as long as
27 none of the program's inputs are marked as being as source of Unicode
28 character data. Such data may come from filehandles, from calls to
29 external programs, from information provided by the system (such as %ENV),
30 or from literals and constants in the source text. Later, in
31 L</Character encodings for input and output>, we'll see how such
32 inputs may be marked as being Unicode character data sources.
34 One particular condition will enable character semantics on the entire
35 program, bypassing the compatibility mode: if the C<$^U> global flag is
36 set to C<1>, nearly all operations will use character semantics by
37 default. As an added convenience, if the C<utf8> pragma is used in the
38 C<main> package, C<$^U> is enabled automatically. [XXX: Should there
39 be a -C switch to enable $^U?]
41 Regardless of the above, the C<byte> pragma can always be used to force
42 byte semantics in a particular lexical scope. See L<byte>.
44 The C<utf8> pragma is primarily a compatibility device that enables
45 recognition of UTF-8 in literals encountered by the parser. It is also
46 used for enabling some of the more experimental Unicode support features.
47 Note that this pragma is only required until a future version of Perl
48 in which character semantics will become the default. This pragma may
49 then become a no-op. See L<utf8>.
51 Unless mentioned otherwise, Perl operators will use character semantics
52 when they are dealing with Unicode data, and byte semantics otherwise.
53 Thus, character semantics for these operations apply transparently; if
54 the input data came from a Unicode source (for example, by adding a
55 character encoding discipline to the filehandle whence it came, or a
56 literal UTF-8 string constant in the program), character semantics
57 apply; otherwise, byte semantics are in effect. To force byte semantics
58 on Unicode data, the C<byte> pragma should be used.
60 Under character semantics, many operations that formerly operated on
61 bytes change to operating on characters. For ASCII data this makes
62 no difference, because UTF-8 stores ASCII in single bytes, but for
63 any character greater than C<chr(127)>, the character is stored in
64 a sequence of two or more bytes, all of which have the high bit set.
65 But by and large, the user need not worry about this, because Perl
66 hides it from the user. A character in Perl is logically just a number
67 ranging from 0 to 2**32 or so. Larger characters encode to longer
68 sequences of bytes internally, but again, this is just an internal
69 detail which is hidden at the Perl level.
71 =head2 Effects of character semantics
73 Character semantics have the following effects:
79 Strings and patterns may contain characters that have an ordinal value
80 larger than 255. In Perl v5.6, this is only enabled if the lexical
81 scope has a C<use utf8> declaration (due to compatibility needs) but
82 future versions may enable this by default.
84 Presuming you use a Unicode editor to edit your program, such characters
85 will typically occur directly within the literal strings as UTF-8
86 characters, but you can also specify a particular character with an
87 extension of the C<\x> notation. UTF-8 characters are specified by
88 putting the hexadecimal code within curlies after the C<\x>. For instance,
89 a Unicode smiley face is C<\x{263A}>. A character in the Latin-1 range
90 (128..255) should be written C<\x{ab}> rather than C<\xab>, since the
91 former will turn into a two-byte UTF-8 code, while the latter will
92 continue to be interpreted as generating a 8-bit byte rather than a
93 character. In fact, if C<-w> is turned on, it will produce a warning
94 that you might be generating invalid UTF-8.
98 Identifiers within the Perl script may contain Unicode alphanumeric
99 characters, including ideographs. (You are currently on your own when
100 it comes to using the canonical forms of characters--Perl doesn't (yet)
101 attempt to canonicalize variable names for you.)
103 This also needs C<use utf8> currently. [XXX: Why?!? High-bit chars were
104 syntax errors when they occurred within identifiers in previous versions,
105 so this should probably be enabled by default.]
109 Regular expressions match characters instead of bytes. For instance,
110 "." matches a character instead of a byte. (However, the C<\C> pattern
111 is provided to force a match a single byte ("C<char>" in C, hence
114 Unicode support in regular expressions needs C<use utf8> currently.
115 [XXX: Because the SWASH routines need to be loaded. And the RE engine
116 appears to need an overhaul to dynamically match Unicode anyway--the
117 current RE compiler creates different nodes with and without C<use utf8>.]
121 Character classes in regular expressions match characters instead of
122 bytes, and match against the character properties specified in the
123 Unicode properties database. So C<\w> can be used to match an ideograph,
126 C<use utf8> is needed to enable this. See above.
130 Named Unicode properties and block ranges make be used as character
131 classes via the new C<\p{}> (matches property) and C<\P{}> (doesn't
132 match property) constructs. For instance, C<\p{Lu}> matches any
133 character with the Unicode uppercase property, while C<\p{M}> matches
134 any mark character. Single letter properties may omit the brackets, so
135 that can be written C<\pM> also. Many predefined character classes are
136 available, such as C<\p{IsMirrored}> and C<\p{InTibetan}>.
138 C<use utf8> is needed to enable this. See above.
142 The special pattern C<\X> match matches any extended Unicode sequence
143 (a "combining character sequence" in Standardese), where the first
144 character is a base character and subsequent characters are mark
145 characters that apply to the base character. It is equivalent to
148 C<use utf8> is needed to enable this. See above.
152 The C<tr///> operator translates characters instead of bytes. It can also
153 be forced to translate between 8-bit codes and UTF-8 regardless of the
154 surrounding utf8 state. For instance, if you know your input in Latin-1,
159 tr/\0-\xff//CU; # latin1 char to utf8
163 Similarly you could translate your output with
165 tr/\0-\x{ff}//UC; # utf8 to latin1 char
167 No, C<s///> doesn't take /U or /C (yet?).
169 C<use utf8> is needed to enable this. See above.
173 Case translation operators use the Unicode case translation tables
174 when provided character input. Note that C<uc()> translates to
175 uppercase, while C<ucfirst> translates to titlecase (for languages
176 that make the distinction). Naturally the corresponding backslash
177 sequences have the same semantics.
181 Most operators that deal with positions or lengths in the string will
182 automatically switch to using character positions, including C<chop()>,
183 C<substr()>, C<pos()>, C<index()>, C<rindex()>, C<sprintf()>,
184 C<write()>, and C<length()>. Operators that specifically don't switch
185 include C<vec()>, C<pack()>, and C<unpack()>. Operators that really
186 don't care include C<chomp()>, as well as any other operator that
187 treats a string as a bucket of bits, such as C<sort()>, and the
188 operators dealing with filenames.
192 The C<pack()>/C<unpack()> letters "C<c>" and "C<C>" do I<not> change,
193 since they're often used for byte-oriented formats. (Again, think
194 "C<char>" in the C language.) However, there is a new "C<U>" specifier
195 that will convert between UTF-8 characters and integers. (It works
196 outside of the utf8 pragma too.)
200 The C<chr()> and C<ord()> functions work on characters. This is like
201 C<pack("U")> and C<unpack("U")>, not like C<pack("C")> and
202 C<unpack("C")>. In fact, the latter are how you now emulate
203 byte-oriented C<chr()> and C<ord()> under utf8.
207 And finally, C<scalar reverse()> reverses by character rather than by byte.
211 =head2 Character encodings for input and output
213 [XXX: This feature is not yet implemented.]
217 As of yet, there is no method for automatically coercing input and
218 output to some encoding other than UTF-8. This is planned in the near
221 Whether an arbitrary piece of data will be treated as "characters" or
222 "bytes" by internal operations cannot be divined at the current time.
224 Use of locales with utf8 may lead to odd results. Currently there is
225 some attempt to apply 8-bit locale info to characters in the range
226 0..255, but this is demonstrably incorrect for locales that use
227 characters above that range (when mapped into Unicode). It will also
228 tend to run slower. Avoidance of locales is strongly encouraged.
232 L<byte>, L<utf8>, L<perlvar/"$^U">