3 perlunicode - Unicode support in Perl
7 =head2 Important Caveats
9 WARNING: While the implementation of Unicode support in Perl is now
10 fairly complete it is still evolving to some extent.
12 In particular the way Unicode is handled on EBCDIC platforms is still
13 rather experimental. On such a platform references to UTF-8 encoding
14 in this document and elsewhere should be read as meaning UTF-EBCDIC as
15 specified in Unicode Technical Report 16 unless ASCII vs EBCDIC issues
16 are specifically discussed. There is no C<utfebcdic> pragma or
17 ":utfebcdic" layer, rather "utf8" and ":utf8" are re-used to mean
18 platform's "natural" 8-bit encoding of Unicode. See L<perlebcdic> for
19 more discussion of the issues.
21 The following areas are still under development.
25 =item Input and Output Disciplines
27 A filehandle can be marked as containing perl's internal Unicode
28 encoding (UTF-8 or UTF-EBCDIC) by opening it with the ":utf8" layer.
29 Other encodings can be converted to perl's encoding on input, or from
30 perl's encoding on output by use of the ":encoding()" layer. There is
31 not yet a clean way to mark the Perl source itself as being in an
34 =item Regular Expressions
36 The regular expression compiler does now attempt to produce
37 polymorphic opcodes. That is the pattern should now adapt to the data
38 and automatically switch to the Unicode character scheme when
39 presented with Unicode data, or a traditional byte scheme when
40 presented with byte data. The implementation is still new and
41 (particularly on EBCDIC platforms) may need further work.
43 =item C<use utf8> still needed to enable UTF-8/UTF-EBCDIC in scripts
45 The C<utf8> pragma implements the tables used for Unicode support.
46 These tables are automatically loaded on demand, so the C<utf8> pragma
47 need not normally be used.
49 However, as a compatibility measure, this pragma must be explicitly
50 used to enable recognition of UTF-8 in the Perl scripts themselves on
51 ASCII based machines or recognize UTF-EBCDIC on EBCDIC based machines.
52 B<NOTE: this should be the only place where an explicit C<use utf8> is
57 =head2 Byte and Character semantics
59 Beginning with version 5.6, Perl uses logically wide characters to
60 represent strings internally. This internal representation of strings
61 uses either the UTF-8 or the UTF-EBCDIC encoding.
63 In future, Perl-level operations can be expected to work with
64 characters rather than bytes, in general.
66 However, as strictly an interim compatibility measure, Perl aims to
67 provide a safe migration path from byte semantics to character
68 semantics for programs. For operations where Perl can unambiguously
69 decide that the input data is characters, Perl now switches to
70 character semantics. For operations where this determination cannot
71 be made without additional information from the user, Perl decides in
72 favor of compatibility, and chooses to use byte semantics.
74 This behavior preserves compatibility with earlier versions of Perl,
75 which allowed byte semantics in Perl operations, but only as long as
76 none of the program's inputs are marked as being as source of Unicode
77 character data. Such data may come from filehandles, from calls to
78 external programs, from information provided by the system (such as %ENV),
79 or from literals and constants in the source text.
81 If the C<-C> command line switch is used, (or the
82 ${^WIDE_SYSTEM_CALLS} global flag is set to C<1>), all system calls
83 will use the corresponding wide character APIs. Note that this is
84 currently only implemented on Windows since other platforms API
85 standard on this area.
87 Regardless of the above, the C<bytes> pragma can always be used to
88 force byte semantics in a particular lexical scope. See L<bytes>.
90 The C<utf8> pragma is primarily a compatibility device that enables
91 recognition of UTF-(8|EBCDIC) in literals encountered by the parser.
92 Note that this pragma is only required until a future version of Perl
93 in which character semantics will become the default. This pragma may
94 then become a no-op. See L<utf8>.
96 Unless mentioned otherwise, Perl operators will use character semantics
97 when they are dealing with Unicode data, and byte semantics otherwise.
98 Thus, character semantics for these operations apply transparently; if
99 the input data came from a Unicode source (for example, by adding a
100 character encoding discipline to the filehandle whence it came, or a
101 literal UTF-8 string constant in the program), character semantics
102 apply; otherwise, byte semantics are in effect. To force byte semantics
103 on Unicode data, the C<bytes> pragma should be used.
105 Notice that if you have a string with byte semantics and you then
106 add character data into it, the bytes will be upgraded I<as if they
107 were ISO 8859-1 (Latin-1)> (or if in EBCDIC, after a translation
110 Under character semantics, many operations that formerly operated on
111 bytes change to operating on characters. For ASCII data this makes no
112 difference, because UTF-8 stores ASCII in single bytes, but for any
113 character greater than C<chr(127)>, the character B<may> be stored in
114 a sequence of two or more bytes, all of which have the high bit set.
116 For C1 controls or Latin 1 characters on an EBCDIC platform the
117 character may be stored in a UTF-EBCDIC multi byte sequence. But by
118 and large, the user need not worry about this, because Perl hides it
119 from the user. A character in Perl is logically just a number ranging
120 from 0 to 2**32 or so. Larger characters encode to longer sequences
121 of bytes internally, but again, this is just an internal detail which
122 is hidden at the Perl level.
124 =head2 Effects of character semantics
126 Character semantics have the following effects:
132 Strings and patterns may contain characters that have an ordinal value
135 Presuming you use a Unicode editor to edit your program, such
136 characters will typically occur directly within the literal strings as
137 UTF-8 (or UTF-EBCDIC on EBCDIC platforms) characters, but you can also
138 specify a particular character with an extension of the C<\x>
139 notation. UTF-X characters are specified by putting the hexadecimal
140 code within curlies after the C<\x>. For instance, a Unicode smiley
145 Identifiers within the Perl script may contain Unicode alphanumeric
146 characters, including ideographs. (You are currently on your own when
147 it comes to using the canonical forms of characters--Perl doesn't
148 (yet) attempt to canonicalize variable names for you.)
152 Regular expressions match characters instead of bytes. For instance,
153 "." matches a character instead of a byte. (However, the C<\C> pattern
154 is provided to force a match a single byte ("C<char>" in C, hence C<\C>).)
158 Character classes in regular expressions match characters instead of
159 bytes, and match against the character properties specified in the
160 Unicode properties database. So C<\w> can be used to match an
161 ideograph, for instance.
165 Named Unicode properties and block ranges make be used as character
166 classes via the new C<\p{}> (matches property) and C<\P{}> (doesn't
167 match property) constructs. For instance, C<\p{Lu}> matches any
168 character with the Unicode uppercase property, while C<\p{M}> matches
169 any mark character. Single letter properties may omit the brackets,
170 so that can be written C<\pM> also. Many predefined character classes
171 are available, such as C<\p{IsMirrored}> and C<\p{InTibetan}>. The
172 recommended names of the C<In> classes are the official Unicode script
173 and block names but with all non-alphanumeric characters removed, for
174 example the block name C<"Latin-1 Supplement"> becomes
175 C<\p{InLatin1Supplement}>.
177 Here is the list as of Unicode 3.1.0 (the two-letter classes) and
178 as defined by Perl (the one-letter classes) (in Unicode materials
179 what Perl calls C<L> is often called C<L&>):
189 Mc Mark, Spacing Combining
192 Nd Number, Decimal Digit
196 Pc Punctuation, Connector
199 Pe Punctuation, Close
200 Pi Punctuation, Initial quote
201 (may behave like Ps or Pe depending on usage)
202 Pf Punctuation, Final quote
203 (may behave like Ps or Pe depending on usage)
204 Po Punctuation, Other
213 Zp Separator, Paragraph
218 Co Other, Private Use
219 Cn Other, Not Assigned (Unicode defines no Cn characters)
221 Additionally, because scripts differ in their directionality
222 (for example Hebrew is written right to left), all characters
223 have their directionality defined:
226 BidiLRE Left-to-Right Embedding
227 BidiLRO Left-to-Right Override
229 BidiAL Right-to-Left Arabic
230 BidiRLE Right-to-Left Embedding
231 BidiRLO Right-to-Left Override
232 BidiPDF Pop Directional Format
233 BidiEN European Number
234 BidiES European Number Separator
235 BidiET European Number Terminator
237 BidiCS Common Number Separator
238 BidiNSM Non-Spacing Mark
239 BidiBN Boundary Neutral
240 BidiB Paragraph Separator
241 BidiS Segment Separator
243 BidiON Other Neutrals
247 The scripts available for C<\p{In...}> and C<\P{In...}>, for example
248 \p{InCyrillic>, are as follows, for example C<\p{InLatin}> or C<\P{InHan}>:
293 In addition to B<scripts>, Unicode also defines B<blocks> of
294 characters. The difference between scripts and blocks is that the
295 former concept is closer to natural languages, while the latter
296 concept is more an artificial grouping based on groups of 256 Unicode
297 characters. For example, the C<Latin> script contains letters from
298 many blocks, but it does not contain all the characters from those
299 blocks, it does not for example contain digits.
301 For more about scripts see the UTR #24:
302 http://www.unicode.org/unicode/reports/tr24/
303 For more about blocks see
304 http://www.unicode.org/Public/UNIDATA/Blocks.txt
306 Because there are overlaps in naming (there are, for example, both
307 a script called C<Katakana> and a block called C<Katakana>, the block
308 version has C<Block> appended to its name, C<\p{InKatakanaBlock}>.
310 Notice that this definition was introduced in Perl 5.8.0: in Perl
311 5.6.0 only the blocks were used; in Perl 5.8.0 scripts became the
312 preferential character class definition; this meant that the
313 definitions of some character classes changed (the ones in the
314 below list that have the C<Block> appended).
321 SpacingModifierLetters
322 CombiningDiacriticalMarks
348 UnifiedCanadianAboriginalSyllabics
353 LatinExtendedAdditional
356 SuperscriptsandSubscripts
358 CombiningMarksforSymbols
362 MathematicalOperators
363 MiscellaneousTechnical
365 OpticalCharacterRecognition
366 EnclosedAlphanumerics
373 CJKRadicalsSupplement
375 IdeographicDescriptionCharacters
376 CJKSymbolsandPunctuation
380 HangulCompatibilityJamo
383 EnclosedCJKLettersandMonths
385 CJKUnifiedIdeographsExtensionA
391 HighPrivateUseSurrogates
394 CJKCompatibilityIdeographs
395 AlphabeticPresentationForms
396 ArabicPresentationFormsA
398 CJKCompatibilityForms
400 ArabicPresentationFormsB
402 HalfwidthandFullwidthForms
406 ByzantineMusicalSymbols
408 MathematicalAlphanumericSymbols
409 CJKUnifiedIdeographsExtensionB
410 CJKCompatibilityIdeographsSupplement
415 The special pattern C<\X> match matches any extended Unicode sequence
416 (a "combining character sequence" in Standardese), where the first
417 character is a base character and subsequent characters are mark
418 characters that apply to the base character. It is equivalent to
423 The C<tr///> operator translates characters instead of bytes. Note
424 that the C<tr///CU> functionality has been removed, as the interface
425 was a mistake. For similar functionality see pack('U0', ...) and
430 Case translation operators use the Unicode case translation tables
431 when provided character input. Note that C<uc()> translates to
432 uppercase, while C<ucfirst> translates to titlecase (for languages
433 that make the distinction). Naturally the corresponding backslash
434 sequences have the same semantics.
438 Most operators that deal with positions or lengths in the string will
439 automatically switch to using character positions, including
440 C<chop()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
441 C<sprintf()>, C<write()>, and C<length()>. Operators that
442 specifically don't switch include C<vec()>, C<pack()>, and
443 C<unpack()>. Operators that really don't care include C<chomp()>, as
444 well as any other operator that treats a string as a bucket of bits,
445 such as C<sort()>, and the operators dealing with filenames.
449 The C<pack()>/C<unpack()> letters "C<c>" and "C<C>" do I<not> change,
450 since they're often used for byte-oriented formats. (Again, think
451 "C<char>" in the C language.) However, there is a new "C<U>" specifier
452 that will convert between UTF-8 characters and integers. (It works
453 outside of the utf8 pragma too.)
457 The C<chr()> and C<ord()> functions work on characters. This is like
458 C<pack("U")> and C<unpack("U")>, not like C<pack("C")> and
459 C<unpack("C")>. In fact, the latter are how you now emulate
460 byte-oriented C<chr()> and C<ord()> for Unicode strings.
461 (Note that this reveals the internal UTF-8 encoding of strings and
462 you are not supposed to do that unless you know what you are doing.)
466 The bit string operators C<& | ^ ~> can operate on character data.
467 However, for backward compatibility reasons (bit string operations
468 when the characters all are less than 256 in ordinal value) one should
469 not mix C<~> (the bit complement) and characters both less than 256 and
470 equal or greater than 256. Most importantly, the DeMorgan's laws
471 (C<~($x|$y) eq ~$x&~$y>, C<~($x&$y) eq ~$x|~$y>) won't hold.
472 Another way to look at this is that the complement cannot return
473 B<both> the 8-bit (byte) wide bit complement B<and> the full character
478 lc(), uc(), lcfirst(), and ucfirst() work only for some of the
479 simplest cases, where the mapping goes from a single Unicode character
480 to another single Unicode character, and where the mapping does not
481 depend on surrounding characters, or on locales. More complex cases,
482 where for example one character maps into several, are not yet
483 implemented. See the Unicode Technical Report #21, Case Mappings,
484 for more details. The Unicode::UCD module (part of Perl since 5.8.0)
485 casespec() and casefold() interfaces supply information about the more
490 And finally, C<scalar reverse()> reverses by character rather than by byte.
494 =head2 Character encodings for input and output
500 As of yet, there is no method for automatically coercing input and
501 output to some encoding other than UTF-8 or UTF-EBCDIC. This is planned
502 in the near future, however.
504 Whether an arbitrary piece of data will be treated as "characters" or
505 "bytes" by internal operations cannot be divined at the current time.
507 Use of locales with utf8 may lead to odd results. Currently there is
508 some attempt to apply 8-bit locale info to characters in the range
509 0..255, but this is demonstrably incorrect for locales that use
510 characters above that range (when mapped into Unicode). It will also
511 tend to run slower. Avoidance of locales is strongly encouraged.
513 =head1 UNICODE REGULAR EXPRESSION SUPPORT LEVEL
515 The following list of Unicode regular expression support describes
516 feature by feature the Unicode support implemented in Perl as of Perl
517 5.8.0. The "Level N" and the section numbers refer to the Unicode
518 Technical Report 18, "Unicode Regular Expression Guidelines".
524 Level 1 - Basic Unicode Support
526 2.1 Hex Notation - done [1]
527 Named Notation - done [2]
528 2.2 Categories - done [3][4]
529 2.3 Subtraction - MISSING [5][6]
530 2.4 Simple Word Boundaries - done [7]
531 2.5 Simple Loose Matches - MISSING [8]
532 2.6 End of Line - MISSING [9][10]
536 [ 3] . \p{Is...} \P{Is...}
537 [ 4] now scripts (see UTR#24 Script Names) in addition to blocks
539 [ 6] can use look-ahead to emulate subtracion
540 [ 7] include Letters in word characters
541 [ 8] see UTR#21 Case Mappings
542 [ 9] see UTR#13 Unicode Newline Guidelines
543 [10] should do ^ and $ also on \x{2028} and \x{2029}
547 Level 2 - Extended Unicode Support
549 3.1 Surrogates - MISSING
550 3.2 Canonical Equivalents - MISSING [11][12]
551 3.3 Locale-Independent Graphemes - MISSING [13]
552 3.4 Locale-Independent Words - MISSING [14]
553 3.5 Locale-Independent Loose Matches - MISSING [15]
555 [11] see UTR#15 Unicode Normalization
556 [12] have Unicode::Normalize but not integrated to regexes
557 [13] have \X but at this level . should equal that
558 [14] need three classes, not just \w and \W
559 [15] see UTR#21 Case Mappings
563 Level 3 - Locale-Sensitive Support
565 4.1 Locale-Dependent Categories - MISSING
566 4.2 Locale-Dependent Graphemes - MISSING [16][17]
567 4.3 Locale-Dependent Words - MISSING
568 4.4 Locale-Dependent Loose Matches - MISSING
569 4.5 Locale-Dependent Ranges - MISSING
571 [16] see UTR#10 Unicode Collation Algorithms
572 [17] have Unicode::Collate but not integrated to regexes
578 L<bytes>, L<utf8>, L<perlretut>, L<perlvar/"${^WIDE_SYSTEM_CALLS}">