3 perlunicode - Unicode support in Perl
7 =head2 Important Caveats
9 Unicode support is an extensive requirement. While Perl does not
10 implement the Unicode standard or the accompanying technical reports
11 from cover to cover, Perl does support many Unicode features.
13 People who want to learn to use Unicode in Perl, should probably read
14 L<the Perl Unicode tutorial, perlunitut|perlunitut>, before reading
15 this reference document.
19 =item Input and Output Layers
21 Perl knows when a filehandle uses Perl's internal Unicode encodings
22 (UTF-8, or UTF-EBCDIC if in EBCDIC) if the filehandle is opened with
23 the ":utf8" layer. Other encodings can be converted to Perl's
24 encoding on input or from Perl's encoding on output by use of the
25 ":encoding(...)" layer. See L<open>.
27 To indicate that Perl source itself is in UTF-8, use C<use utf8;>.
29 =item Regular Expressions
31 The regular expression compiler produces polymorphic opcodes. That is,
32 the pattern adapts to the data and automatically switches to the Unicode
33 character scheme when presented with data that is internally encoded in
34 UTF-8 -- or instead uses a traditional byte scheme when presented with
37 =item C<use utf8> still needed to enable UTF-8/UTF-EBCDIC in scripts
39 As a compatibility measure, the C<use utf8> pragma must be explicitly
40 included to enable recognition of UTF-8 in the Perl scripts themselves
41 (in string or regular expression literals, or in identifier names) on
42 ASCII-based machines or to recognize UTF-EBCDIC on EBCDIC-based
43 machines. B<These are the only times when an explicit C<use utf8>
44 is needed.> See L<utf8>.
46 =item BOM-marked scripts and UTF-16 scripts autodetected
48 If a Perl script begins marked with the Unicode BOM (UTF-16LE, UTF16-BE,
49 or UTF-8), or if the script looks like non-BOM-marked UTF-16 of either
50 endianness, Perl will correctly read in the script as Unicode.
51 (BOMless UTF-8 cannot be effectively recognized or differentiated from
52 ISO 8859-1 or other eight-bit encodings.)
54 =item C<use encoding> needed to upgrade non-Latin-1 byte strings
56 By default, there is a fundamental asymmetry in Perl's Unicode model:
57 implicit upgrading from byte strings to Unicode strings assumes that
58 they were encoded in I<ISO 8859-1 (Latin-1)>, but Unicode strings are
59 downgraded with UTF-8 encoding. This happens because the first 256
60 codepoints in Unicode happens to agree with Latin-1.
62 See L</"Byte and Character Semantics"> for more details.
66 =head2 Byte and Character Semantics
68 Beginning with version 5.6, Perl uses logically-wide characters to
69 represent strings internally.
71 In future, Perl-level operations will be expected to work with
72 characters rather than bytes.
74 However, as an interim compatibility measure, Perl aims to
75 provide a safe migration path from byte semantics to character
76 semantics for programs. For operations where Perl can unambiguously
77 decide that the input data are characters, Perl switches to
78 character semantics. For operations where this determination cannot
79 be made without additional information from the user, Perl decides in
80 favor of compatibility and chooses to use byte semantics.
82 Under byte semantics, when C<use locale> is in effect, Perl uses the
83 semantics associated with the current locale. Absent a C<use locale>, Perl
84 currently uses US-ASCII (or Basic Latin in Unicode terminology) byte semantics,
85 meaning that characters whose ordinal numbers are in the range 128 - 255 are
86 undefined except for their ordinal numbers. This means that none have case
87 (upper and lower), nor are any a member of character classes, like C<[:alpha:]>
89 (But all do belong to the C<\W> class or the Perl regular expression extension
92 This behavior preserves compatibility with earlier versions of Perl,
93 which allowed byte semantics in Perl operations only if
94 none of the program's inputs were marked as being as source of Unicode
95 character data. Such data may come from filehandles, from calls to
96 external programs, from information provided by the system (such as %ENV),
97 or from literals and constants in the source text.
99 The C<bytes> pragma will always, regardless of platform, force byte
100 semantics in a particular lexical scope. See L<bytes>.
102 The C<utf8> pragma is primarily a compatibility device that enables
103 recognition of UTF-(8|EBCDIC) in literals encountered by the parser.
104 Note that this pragma is only required while Perl defaults to byte
105 semantics; when character semantics become the default, this pragma
106 may become a no-op. See L<utf8>.
108 Unless explicitly stated, Perl operators use character semantics
109 for Unicode data and byte semantics for non-Unicode data.
110 The decision to use character semantics is made transparently. If
111 input data comes from a Unicode source--for example, if a character
112 encoding layer is added to a filehandle or a literal Unicode
113 string constant appears in a program--character semantics apply.
114 Otherwise, byte semantics are in effect. The C<bytes> pragma should
115 be used to force byte semantics on Unicode data.
117 If strings operating under byte semantics and strings with Unicode
118 character data are concatenated, the new string will have
119 character semantics. This can cause surprises: See L</BUGS>, below
121 Under character semantics, many operations that formerly operated on
122 bytes now operate on characters. A character in Perl is
123 logically just a number ranging from 0 to 2**31 or so. Larger
124 characters may encode into longer sequences of bytes internally, but
125 this internal detail is mostly hidden for Perl code.
126 See L<perluniintro> for more.
128 =head2 Effects of Character Semantics
130 Character semantics have the following effects:
136 Strings--including hash keys--and regular expression patterns may
137 contain characters that have an ordinal value larger than 255.
139 If you use a Unicode editor to edit your program, Unicode characters may
140 occur directly within the literal strings in UTF-8 encoding, or UTF-16.
141 (The former requires a BOM or C<use utf8>, the latter requires a BOM.)
143 Unicode characters can also be added to a string by using the C<\x{...}>
144 notation. The Unicode code for the desired character, in hexadecimal,
145 should be placed in the braces. For instance, a smiley face is
146 C<\x{263A}>. This encoding scheme works for all characters, but
147 for characters under 0x100, note that Perl may use an 8 bit encoding
148 internally, for optimization and/or backward compatibility.
152 use charnames ':full';
154 you can use the C<\N{...}> notation and put the official Unicode
155 character name within the braces, such as C<\N{WHITE SMILING FACE}>.
159 If an appropriate L<encoding> is specified, identifiers within the
160 Perl script may contain Unicode alphanumeric characters, including
161 ideographs. Perl does not currently attempt to canonicalize variable
166 Regular expressions match characters instead of bytes. "." matches
167 a character instead of a byte.
171 Character classes in regular expressions match characters instead of
172 bytes and match against the character properties specified in the
173 Unicode properties database. C<\w> can be used to match a Japanese
174 ideograph, for instance.
178 Named Unicode properties, scripts, and block ranges may be used like
179 character classes via the C<\p{}> "matches property" construct and
180 the C<\P{}> negation, "doesn't match property".
182 See L</"Unicode Character Properties"> for more details.
184 You can define your own character properties and use them
185 in the regular expression with the C<\p{}> or C<\P{}> construct.
187 See L</"User-Defined Character Properties"> for more details.
191 The special pattern C<\X> matches any extended Unicode
192 sequence--"a combining character sequence" in Standardese--where the
193 first character is a base character and subsequent characters are mark
194 characters that apply to the base character. C<\X> is equivalent to
199 The C<tr///> operator translates characters instead of bytes. Note
200 that the C<tr///CU> functionality has been removed. For similar
201 functionality see pack('U0', ...) and pack('C0', ...).
205 Case translation operators use the Unicode case translation tables
206 when character input is provided. Note that C<uc()>, or C<\U> in
207 interpolated strings, translates to uppercase, while C<ucfirst>,
208 or C<\u> in interpolated strings, translates to titlecase in languages
209 that make the distinction.
213 Most operators that deal with positions or lengths in a string will
214 automatically switch to using character positions, including
215 C<chop()>, C<chomp()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
216 C<sprintf()>, C<write()>, and C<length()>. An operator that
217 specifically does not switch is C<vec()>. Operators that really don't
218 care include operators that treat strings as a bucket of bits such as
219 C<sort()>, and operators dealing with filenames.
223 The C<pack()>/C<unpack()> letter C<C> does I<not> change, since it is often
224 used for byte-oriented formats. Again, think C<char> in the C language.
226 There is a new C<U> specifier that converts between Unicode characters
227 and code points. There is also a C<W> specifier that is the equivalent of
228 C<chr>/C<ord> and properly handles character values even if they are above 255.
232 The C<chr()> and C<ord()> functions work on characters, similar to
233 C<pack("W")> and C<unpack("W")>, I<not> C<pack("C")> and
234 C<unpack("C")>. C<pack("C")> and C<unpack("C")> are methods for
235 emulating byte-oriented C<chr()> and C<ord()> on Unicode strings.
236 While these methods reveal the internal encoding of Unicode strings,
237 that is not something one normally needs to care about at all.
241 The bit string operators, C<& | ^ ~>, can operate on character data.
242 However, for backward compatibility, such as when using bit string
243 operations when characters are all less than 256 in ordinal value, one
244 should not use C<~> (the bit complement) with characters of both
245 values less than 256 and values greater than 256. Most importantly,
246 DeMorgan's laws (C<~($x|$y) eq ~$x&~$y> and C<~($x&$y) eq ~$x|~$y>)
247 will not hold. The reason for this mathematical I<faux pas> is that
248 the complement cannot return B<both> the 8-bit (byte-wide) bit
249 complement B<and> the full character-wide bit complement.
253 lc(), uc(), lcfirst(), and ucfirst() work for the following cases:
259 the case mapping is from a single Unicode character to another
260 single Unicode character, or
264 the case mapping is from a single Unicode character to more
265 than one Unicode character.
269 Things to do with locales (Lithuanian, Turkish, Azeri) do B<not> work
270 since Perl does not understand the concept of Unicode locales.
272 See the Unicode Technical Report #21, Case Mappings, for more details.
274 But you can also define your own mappings to be used in the lc(),
275 lcfirst(), uc(), and ucfirst() (or their string-inlined versions).
277 See L</"User-Defined Case Mappings"> for more details.
285 And finally, C<scalar reverse()> reverses by character rather than by byte.
289 =head2 Unicode Character Properties
291 Named Unicode properties, scripts, and block ranges may be used like
292 character classes via the C<\p{}> "matches property" construct and
293 the C<\P{}> negation, "doesn't match property".
295 For instance, C<\p{Lu}> matches any character with the Unicode "Lu"
296 (Letter, uppercase) property, while C<\p{M}> matches any character
297 with an "M" (mark--accents and such) property. Brackets are not
298 required for single letter properties, so C<\p{M}> is equivalent to
299 C<\pM>. Many predefined properties are available, such as
300 C<\p{Mirrored}> and C<\p{Tibetan}>.
302 The official Unicode script and block names have spaces and dashes as
303 separators, but for convenience you can use dashes, spaces, or
304 underbars, and case is unimportant. It is recommended, however, that
305 for consistency you use the following naming: the official Unicode
306 script, property, or block name (see below for the additional rules
307 that apply to block names) with whitespace and dashes removed, and the
308 words "uppercase-first-lowercase-rest". C<Latin-1 Supplement> thus
309 becomes C<Latin1Supplement>.
311 You can also use negation in both C<\p{}> and C<\P{}> by introducing a caret
312 (^) between the first brace and the property name: C<\p{^Tamil}> is
313 equal to C<\P{Tamil}>.
315 B<NOTE: the properties, scripts, and blocks listed here are as of
316 Unicode 5.0.0 in July 2006.>
320 =item General Category
322 Here are the basic Unicode General Category properties, followed by their
323 long form. You can use either; C<\p{Lu}> and C<\p{UppercaseLetter}>,
324 for instance, are identical.
347 Pc ConnectorPunctuation
351 Pi InitialPunctuation
352 (may behave like Ps or Pe depending on usage)
354 (may behave like Ps or Pe depending on usage)
366 Zp ParagraphSeparator
371 Cs Surrogate (not usable)
375 Single-letter properties match all characters in any of the
376 two-letter sub-properties starting with the same letter.
377 C<LC> and C<L&> are special cases, which are aliases for the set of
378 C<Ll>, C<Lu>, and C<Lt>.
380 Because Perl hides the need for the user to understand the internal
381 representation of Unicode characters, there is no need to implement
382 the somewhat messy concept of surrogates. C<Cs> is therefore not
385 =item Bidirectional Character Types
387 Because scripts differ in their directionality--Hebrew is
388 written right to left, for example--Unicode supplies these properties in
394 LRE Left-to-Right Embedding
395 LRO Left-to-Right Override
397 AL Right-to-Left Arabic
398 RLE Right-to-Left Embedding
399 RLO Right-to-Left Override
400 PDF Pop Directional Format
402 ES European Number Separator
403 ET European Number Terminator
405 CS Common Number Separator
408 B Paragraph Separator
413 For example, C<\p{BidiClass:R}> matches characters that are normally
414 written right to left.
418 The script names which can be used by C<\p{...}> and C<\P{...}>,
419 such as in C<\p{Latin}> or C<\p{Cyrillic}>, are as follows:
487 =item Extended property classes
489 Extended property classes can supplement the basic
490 properties, defined by the F<PropList> Unicode database:
504 LogicalOrderException
505 NoncharacterCodePoint
507 OtherDefaultIgnorableCodePoint
525 and there are further derived properties:
527 Alphabetic = Lu + Ll + Lt + Lm + Lo + Nl + OtherAlphabetic
528 Lowercase = Ll + OtherLowercase
529 Uppercase = Lu + OtherUppercase
530 Math = Sm + OtherMath
532 IDStart = Lu + Ll + Lt + Lm + Lo + Nl + OtherIDStart
533 IDContinue = IDStart + Mn + Mc + Nd + Pc + OtherIDContinue
535 DefaultIgnorableCodePoint
536 = OtherDefaultIgnorableCodePoint
537 + Cf + Cc + Cs + Noncharacters + VariationSelector
538 - WhiteSpace - FFF9..FFFB (Annotation Characters)
540 Any = Any code points (i.e. U+0000 to U+10FFFF)
541 Assigned = Any non-Cn code points (i.e. synonym for \P{Cn})
542 Unassigned = Synonym for \p{Cn}
543 ASCII = ASCII (i.e. U+0000 to U+007F)
545 Common = Any character (or unassigned code point)
546 not explicitly assigned to a script
548 =item Use of "Is" Prefix
550 For backward compatibility (with Perl 5.6), all properties mentioned
551 so far may have C<Is> prepended to their name, so C<\P{IsLu}>, for
552 example, is equal to C<\P{Lu}>.
556 In addition to B<scripts>, Unicode also defines B<blocks> of
557 characters. The difference between scripts and blocks is that the
558 concept of scripts is closer to natural languages, while the concept
559 of blocks is more of an artificial grouping based on groups of 256
560 Unicode characters. For example, the C<Latin> script contains letters
561 from many blocks but does not contain all the characters from those
562 blocks. It does not, for example, contain digits, because digits are
563 shared across many scripts. Digits and similar groups, like
564 punctuation, are in a category called C<Common>.
566 For more about scripts, see the UAX#24 "Script Names":
568 http://www.unicode.org/reports/tr24/
570 For more about blocks, see:
572 http://www.unicode.org/Public/UNIDATA/Blocks.txt
574 Block names are given with the C<In> prefix. For example, the
575 Katakana block is referenced via C<\p{InKatakana}>. The C<In>
576 prefix may be omitted if there is no naming conflict with a script
577 or any other property, but it is recommended that C<In> always be used
578 for block tests to avoid confusion.
580 These block names are supported:
583 InAlphabeticPresentationForms
584 InAncientGreekMusicalNotation
585 InAncientGreekNumbers
587 InArabicPresentationFormsA
588 InArabicPresentationFormsB
602 InByzantineMusicalSymbols
604 InCJKCompatibilityForms
605 InCJKCompatibilityIdeographs
606 InCJKCompatibilityIdeographsSupplement
607 InCJKRadicalsSupplement
609 InCJKSymbolsAndPunctuation
610 InCJKUnifiedIdeographs
611 InCJKUnifiedIdeographsExtensionA
612 InCJKUnifiedIdeographsExtensionB
614 InCombiningDiacriticalMarks
615 InCombiningDiacriticalMarksSupplement
616 InCombiningDiacriticalMarksforSymbols
620 InCountingRodNumerals
622 InCuneiformNumbersAndPunctuation
630 InEnclosedAlphanumerics
631 InEnclosedCJKLettersAndMonths
645 InHalfwidthAndFullwidthForms
646 InHangulCompatibilityJamo
651 InHighPrivateUseSurrogates
655 InIdeographicDescriptionCharacters
660 InKatakanaPhoneticExtensions
667 InLatinExtendedAdditional
677 InMathematicalAlphanumericSymbols
678 InMathematicalOperators
679 InMiscellaneousMathematicalSymbolsA
680 InMiscellaneousMathematicalSymbolsB
681 InMiscellaneousSymbols
682 InMiscellaneousSymbolsAndArrows
683 InMiscellaneousTechnical
684 InModifierToneLetters
694 InOpticalCharacterRecognition
700 InPhoneticExtensionsSupplement
706 InSpacingModifierLetters
708 InSuperscriptsAndSubscripts
709 InSupplementalArrowsA
710 InSupplementalArrowsB
711 InSupplementalMathematicalOperators
712 InSupplementalPunctuation
713 InSupplementaryPrivateUseAreaA
714 InSupplementaryPrivateUseAreaB
729 InUnifiedCanadianAboriginalSyllabics
731 InVariationSelectorsSupplement
735 InYijingHexagramSymbols
739 =head2 User-Defined Character Properties
741 You can define your own character properties by defining subroutines
742 whose names begin with "In" or "Is". The subroutines can be defined in
743 any package. The user-defined properties can be used in the regular
744 expression C<\p> and C<\P> constructs; if you are using a user-defined
745 property from a package other than the one you are in, you must specify
746 its package in the C<\p> or C<\P> construct.
748 # assuming property IsForeign defined in Lang::
749 package main; # property package name required
750 if ($txt =~ /\p{Lang::IsForeign}+/) { ... }
752 package Lang; # property package name not required
753 if ($txt =~ /\p{IsForeign}+/) { ... }
756 Note that the effect is compile-time and immutable once defined.
758 The subroutines must return a specially-formatted string, with one
759 or more newline-separated lines. Each line must be one of the following:
765 A single hexadecimal number denoting a Unicode code point to include.
769 Two hexadecimal numbers separated by horizontal whitespace (space or
770 tabular characters) denoting a range of Unicode code points to include.
774 Something to include, prefixed by "+": a built-in character
775 property (prefixed by "utf8::") or a user-defined character property,
776 to represent all the characters in that property; two hexadecimal code
777 points for a range; or a single hexadecimal code point.
781 Something to exclude, prefixed by "-": an existing character
782 property (prefixed by "utf8::") or a user-defined character property,
783 to represent all the characters in that property; two hexadecimal code
784 points for a range; or a single hexadecimal code point.
788 Something to negate, prefixed "!": an existing character
789 property (prefixed by "utf8::") or a user-defined character property,
790 to represent all the characters in that property; two hexadecimal code
791 points for a range; or a single hexadecimal code point.
795 Something to intersect with, prefixed by "&": an existing character
796 property (prefixed by "utf8::") or a user-defined character property,
797 for all the characters except the characters in the property; two
798 hexadecimal code points for a range; or a single hexadecimal code point.
802 For example, to define a property that covers both the Japanese
803 syllabaries (hiragana and katakana), you can define
812 Imagine that the here-doc end marker is at the beginning of the line.
813 Now you can use C<\p{InKana}> and C<\P{InKana}>.
815 You could also have used the existing block property names:
824 Suppose you wanted to match only the allocated characters,
825 not the raw block ranges: in other words, you want to remove
836 The negation is useful for defining (surprise!) negated classes.
846 Intersection is useful for getting the common characters matched by
847 two (or more) classes.
856 It's important to remember not to use "&" for the first set -- that
857 would be intersecting with nothing (resulting in an empty set).
859 =head2 User-Defined Case Mappings
861 You can also define your own mappings to be used in the lc(),
862 lcfirst(), uc(), and ucfirst() (or their string-inlined versions).
863 The principle is similar to that of user-defined character
864 properties: to define subroutines in the C<main> package
865 with names like C<ToLower> (for lc() and lcfirst()), C<ToTitle> (for
866 the first character in ucfirst()), and C<ToUpper> (for uc(), and the
867 rest of the characters in ucfirst()).
869 The string returned by the subroutines needs now to be three
870 hexadecimal numbers separated by tabulators: start of the source
871 range, end of the source range, and start of the destination range.
880 defines an uc() mapping that causes only the characters "a", "b", and
881 "c" to be mapped to "A", "B", "C", all other characters will remain
884 If there is no source range to speak of, that is, the mapping is from
885 a single character to another single character, leave the end of the
886 source range empty, but the two tabulator characters are still needed.
895 defines a lc() mapping that causes only "A" to be mapped to "a", all
896 other characters will remain unchanged.
898 (For serious hackers only) If you want to introspect the default
899 mappings, you can find the data in the directory
900 C<$Config{privlib}>/F<unicore/To/>. The mapping data is returned as
901 the here-document, and the C<utf8::ToSpecFoo> are special exception
902 mappings derived from <$Config{privlib}>/F<unicore/SpecialCasing.txt>.
903 The C<Digit> and C<Fold> mappings that one can see in the directory
904 are not directly user-accessible, one can use either the
905 C<Unicode::UCD> module, or just match case-insensitively (that's when
906 the C<Fold> mapping is used).
908 A final note on the user-defined case mappings: they will be used
909 only if the scalar has been marked as having Unicode characters.
910 Old byte-style strings will not be affected.
912 =head2 Character Encodings for Input and Output
916 =head2 Unicode Regular Expression Support Level
918 The following list of Unicode support for regular expressions describes
919 all the features currently supported. The references to "Level N"
920 and the section numbers refer to the Unicode Technical Standard #18,
921 "Unicode Regular Expressions", version 11, in May 2005.
927 Level 1 - Basic Unicode Support
929 RL1.1 Hex Notation - done [1]
930 RL1.2 Properties - done [2][3]
931 RL1.2a Compatibility Properties - done [4]
932 RL1.3 Subtraction and Intersection - MISSING [5]
933 RL1.4 Simple Word Boundaries - done [6]
934 RL1.5 Simple Loose Matches - done [7]
935 RL1.6 Line Boundaries - MISSING [8]
936 RL1.7 Supplementary Code Points - done [9]
940 [3] supports not only minimal list (general category, scripts,
941 Alphabetic, Lowercase, Uppercase, WhiteSpace,
942 NoncharacterCodePoint, DefaultIgnorableCodePoint, Any,
943 ASCII, Assigned), but also bidirectional types, blocks, etc.
944 (see L</"Unicode Character Properties">)
945 [4] \d \D \s \S \w \W \X [:prop:] [:^prop:]
946 [5] can use regular expression look-ahead [a] or
947 user-defined character properties [b] to emulate set operations
949 [7] note that Perl does Full case-folding in matching, not Simple:
950 for example U+1F88 is equivalent to U+1F00 U+03B9,
951 not with 1F80. This difference matters mainly for certain Greek
952 capital letters with certain modifiers: the Full case-folding
953 decomposes the letter, while the Simple case-folding would map
954 it to a single character.
955 [8] should do ^ and $ also on U+000B (\v in C), FF (\f), CR (\r),
956 CRLF (\r\n), NEL (U+0085), LS (U+2028), and PS (U+2029);
957 should also affect <>, $., and script line numbers;
958 should not split lines within CRLF [c] (i.e. there is no empty
959 line between \r and \n)
960 [9] UTF-8/UTF-EBDDIC used in perl allows not only U+10000 to U+10FFFF
961 but also beyond U+10FFFF [d]
963 [a] You can mimic class subtraction using lookahead.
964 For example, what UTS#18 might write as
966 [{Greek}-[{UNASSIGNED}]]
968 in Perl can be written as:
970 (?!\p{Unassigned})\p{InGreekAndCoptic}
971 (?=\p{Assigned})\p{InGreekAndCoptic}
973 But in this particular example, you probably really want
977 which will match assigned characters known to be part of the Greek script.
979 Also see the Unicode::Regex::Set module, it does implement the full
980 UTS#18 grouping, intersection, union, and removal (subtraction) syntax.
982 [b] '+' for union, '-' for removal (set-difference), '&' for intersection
983 (see L</"User-Defined Character Properties">)
985 [c] Try the C<:crlf> layer (see L<PerlIO>).
987 [d] Avoid C<use warning 'utf8';> (or say C<no warning 'utf8';>) to allow
988 U+FFFF (C<\x{FFFF}>).
992 Level 2 - Extended Unicode Support
994 RL2.1 Canonical Equivalents - MISSING [10][11]
995 RL2.2 Default Grapheme Clusters - MISSING [12][13]
996 RL2.3 Default Word Boundaries - MISSING [14]
997 RL2.4 Default Loose Matches - MISSING [15]
998 RL2.5 Name Properties - MISSING [16]
999 RL2.6 Wildcard Properties - MISSING
1001 [10] see UAX#15 "Unicode Normalization Forms"
1002 [11] have Unicode::Normalize but not integrated to regexes
1003 [12] have \X but at this level . should equal that
1004 [13] UAX#29 "Text Boundaries" considers CRLF and Hangul syllable
1005 clusters as a single grapheme cluster.
1006 [14] see UAX#29, Word Boundaries
1007 [15] see UAX#21 "Case Mappings"
1008 [16] have \N{...} but neither compute names of CJK Ideographs
1009 and Hangul Syllables nor use a loose match [e]
1011 [e] C<\N{...}> allows namespaces (see L<charnames>).
1015 Level 3 - Tailored Support
1017 RL3.1 Tailored Punctuation - MISSING
1018 RL3.2 Tailored Grapheme Clusters - MISSING [17][18]
1019 RL3.3 Tailored Word Boundaries - MISSING
1020 RL3.4 Tailored Loose Matches - MISSING
1021 RL3.5 Tailored Ranges - MISSING
1022 RL3.6 Context Matching - MISSING [19]
1023 RL3.7 Incremental Matches - MISSING
1024 ( RL3.8 Unicode Set Sharing )
1025 RL3.9 Possible Match Sets - MISSING
1026 RL3.10 Folded Matching - MISSING [20]
1027 RL3.11 Submatchers - MISSING
1029 [17] see UAX#10 "Unicode Collation Algorithms"
1030 [18] have Unicode::Collate but not integrated to regexes
1031 [19] have (?<=x) and (?=x), but look-aheads or look-behinds should see
1032 outside of the target substring
1033 [20] need insensitive matching for linguistic features other than case;
1034 for example, hiragana to katakana, wide and narrow, simplified Han
1035 to traditional Han (see UTR#30 "Character Foldings")
1039 =head2 Unicode Encodings
1041 Unicode characters are assigned to I<code points>, which are abstract
1042 numbers. To use these numbers, various encodings are needed.
1050 UTF-8 is a variable-length (1 to 6 bytes, current character allocations
1051 require 4 bytes), byte-order independent encoding. For ASCII (and we
1052 really do mean 7-bit ASCII, not another 8-bit encoding), UTF-8 is
1055 The following table is from Unicode 3.2.
1057 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
1059 U+0000..U+007F 00..7F
1060 U+0080..U+07FF C2..DF 80..BF
1061 U+0800..U+0FFF E0 A0..BF 80..BF
1062 U+1000..U+CFFF E1..EC 80..BF 80..BF
1063 U+D000..U+D7FF ED 80..9F 80..BF
1064 U+D800..U+DFFF ******* ill-formed *******
1065 U+E000..U+FFFF EE..EF 80..BF 80..BF
1066 U+10000..U+3FFFF F0 90..BF 80..BF 80..BF
1067 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
1068 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
1070 Note the C<A0..BF> in C<U+0800..U+0FFF>, the C<80..9F> in
1071 C<U+D000...U+D7FF>, the C<90..B>F in C<U+10000..U+3FFFF>, and the
1072 C<80...8F> in C<U+100000..U+10FFFF>. The "gaps" are caused by legal
1073 UTF-8 avoiding non-shortest encodings: it is technically possible to
1074 UTF-8-encode a single code point in different ways, but that is
1075 explicitly forbidden, and the shortest possible encoding should always
1076 be used. So that's what Perl does.
1078 Another way to look at it is via bits:
1080 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
1083 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
1084 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
1085 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
1087 As you can see, the continuation bytes all begin with C<10>, and the
1088 leading bits of the start byte tell how many bytes the are in the
1095 Like UTF-8 but EBCDIC-safe, in the way that UTF-8 is ASCII-safe.
1099 UTF-16, UTF-16BE, UTF-16LE, Surrogates, and BOMs (Byte Order Marks)
1101 The followings items are mostly for reference and general Unicode
1102 knowledge, Perl doesn't use these constructs internally.
1104 UTF-16 is a 2 or 4 byte encoding. The Unicode code points
1105 C<U+0000..U+FFFF> are stored in a single 16-bit unit, and the code
1106 points C<U+10000..U+10FFFF> in two 16-bit units. The latter case is
1107 using I<surrogates>, the first 16-bit unit being the I<high
1108 surrogate>, and the second being the I<low surrogate>.
1110 Surrogates are code points set aside to encode the C<U+10000..U+10FFFF>
1111 range of Unicode code points in pairs of 16-bit units. The I<high
1112 surrogates> are the range C<U+D800..U+DBFF>, and the I<low surrogates>
1113 are the range C<U+DC00..U+DFFF>. The surrogate encoding is
1115 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
1116 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
1120 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
1122 If you try to generate surrogates (for example by using chr()), you
1123 will get a warning if warnings are turned on, because those code
1124 points are not valid for a Unicode character.
1126 Because of the 16-bitness, UTF-16 is byte-order dependent. UTF-16
1127 itself can be used for in-memory computations, but if storage or
1128 transfer is required either UTF-16BE (big-endian) or UTF-16LE
1129 (little-endian) encodings must be chosen.
1131 This introduces another problem: what if you just know that your data
1132 is UTF-16, but you don't know which endianness? Byte Order Marks, or
1133 BOMs, are a solution to this. A special character has been reserved
1134 in Unicode to function as a byte order marker: the character with the
1135 code point C<U+FEFF> is the BOM.
1137 The trick is that if you read a BOM, you will know the byte order,
1138 since if it was written on a big-endian platform, you will read the
1139 bytes C<0xFE 0xFF>, but if it was written on a little-endian platform,
1140 you will read the bytes C<0xFF 0xFE>. (And if the originating platform
1141 was writing in UTF-8, you will read the bytes C<0xEF 0xBB 0xBF>.)
1143 The way this trick works is that the character with the code point
1144 C<U+FFFE> is guaranteed not to be a valid Unicode character, so the
1145 sequence of bytes C<0xFF 0xFE> is unambiguously "BOM, represented in
1146 little-endian format" and cannot be C<U+FFFE>, represented in big-endian
1151 UTF-32, UTF-32BE, UTF-32LE
1153 The UTF-32 family is pretty much like the UTF-16 family, expect that
1154 the units are 32-bit, and therefore the surrogate scheme is not
1155 needed. The BOM signatures will be C<0x00 0x00 0xFE 0xFF> for BE and
1156 C<0xFF 0xFE 0x00 0x00> for LE.
1162 Encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
1163 encoding. Unlike UTF-16, UCS-2 is not extensible beyond C<U+FFFF>,
1164 because it does not use surrogates. UCS-4 is a 32-bit encoding,
1165 functionally identical to UTF-32.
1171 A seven-bit safe (non-eight-bit) encoding, which is useful if the
1172 transport or storage is not eight-bit safe. Defined by RFC 2152.
1176 =head2 Security Implications of Unicode
1184 Unfortunately, the specification of UTF-8 leaves some room for
1185 interpretation of how many bytes of encoded output one should generate
1186 from one input Unicode character. Strictly speaking, the shortest
1187 possible sequence of UTF-8 bytes should be generated,
1188 because otherwise there is potential for an input buffer overflow at
1189 the receiving end of a UTF-8 connection. Perl always generates the
1190 shortest length UTF-8, and with warnings on Perl will warn about
1191 non-shortest length UTF-8 along with other malformations, such as the
1192 surrogates, which are not real Unicode code points.
1196 Regular expressions behave slightly differently between byte data and
1197 character (Unicode) data. For example, the "word character" character
1198 class C<\w> will work differently depending on if data is eight-bit bytes
1201 In the first case, the set of C<\w> characters is either small--the
1202 default set of alphabetic characters, digits, and the "_"--or, if you
1203 are using a locale (see L<perllocale>), the C<\w> might contain a few
1204 more letters according to your language and country.
1206 In the second case, the C<\w> set of characters is much, much larger.
1207 Most importantly, even in the set of the first 256 characters, it will
1208 probably match different characters: unlike most locales, which are
1209 specific to a language and country pair, Unicode classifies all the
1210 characters that are letters I<somewhere> as C<\w>. For example, your
1211 locale might not think that LATIN SMALL LETTER ETH is a letter (unless
1212 you happen to speak Icelandic), but Unicode does.
1214 As discussed elsewhere, Perl has one foot (two hooves?) planted in
1215 each of two worlds: the old world of bytes and the new world of
1216 characters, upgrading from bytes to characters when necessary.
1217 If your legacy code does not explicitly use Unicode, no automatic
1218 switch-over to characters should happen. Characters shouldn't get
1219 downgraded to bytes, either. It is possible to accidentally mix bytes
1220 and characters, however (see L<perluniintro>), in which case C<\w> in
1221 regular expressions might start behaving differently. Review your
1222 code. Use warnings and the C<strict> pragma.
1226 =head2 Unicode in Perl on EBCDIC
1228 The way Unicode is handled on EBCDIC platforms is still
1229 experimental. On such platforms, references to UTF-8 encoding in this
1230 document and elsewhere should be read as meaning the UTF-EBCDIC
1231 specified in Unicode Technical Report 16, unless ASCII vs. EBCDIC issues
1232 are specifically discussed. There is no C<utfebcdic> pragma or
1233 ":utfebcdic" layer; rather, "utf8" and ":utf8" are reused to mean
1234 the platform's "natural" 8-bit encoding of Unicode. See L<perlebcdic>
1235 for more discussion of the issues.
1239 Usually locale settings and Unicode do not affect each other, but
1240 there are a couple of exceptions:
1246 You can enable automatic UTF-8-ification of your standard file
1247 handles, default C<open()> layer, and C<@ARGV> by using either
1248 the C<-C> command line switch or the C<PERL_UNICODE> environment
1249 variable, see L<perlrun> for the documentation of the C<-C> switch.
1253 Perl tries really hard to work both with Unicode and the old
1254 byte-oriented world. Most often this is nice, but sometimes Perl's
1255 straddling of the proverbial fence causes problems.
1259 =head2 When Unicode Does Not Happen
1261 While Perl does have extensive ways to input and output in Unicode,
1262 and few other 'entry points' like the @ARGV which can be interpreted
1263 as Unicode (UTF-8), there still are many places where Unicode (in some
1264 encoding or another) could be given as arguments or received as
1265 results, or both, but it is not.
1267 The following are such interfaces. For all of these interfaces Perl
1268 currently (as of 5.8.3) simply assumes byte strings both as arguments
1269 and results, or UTF-8 strings if the C<encoding> pragma has been used.
1271 One reason why Perl does not attempt to resolve the role of Unicode in
1272 this cases is that the answers are highly dependent on the operating
1273 system and the file system(s). For example, whether filenames can be
1274 in Unicode, and in exactly what kind of encoding, is not exactly a
1275 portable concept. Similarly for the qx and system: how well will the
1276 'command line interface' (and which of them?) handle Unicode?
1282 chdir, chmod, chown, chroot, exec, link, lstat, mkdir,
1283 rename, rmdir, stat, symlink, truncate, unlink, utime, -X
1295 open, opendir, sysopen
1299 qx (aka the backtick operator), system
1307 =head2 Forcing Unicode in Perl (Or Unforcing Unicode in Perl)
1309 Sometimes (see L</"When Unicode Does Not Happen">) there are
1310 situations where you simply need to force a byte
1311 string into UTF-8, or vice versa. The low-level calls
1312 utf8::upgrade($bytestring) and utf8::downgrade($utf8string[, FAIL_OK]) are
1315 Note that utf8::downgrade() can fail if the string contains characters
1316 that don't fit into a byte.
1318 =head2 Using Unicode in XS
1320 If you want to handle Perl Unicode in XS extensions, you may find the
1321 following C APIs useful. See also L<perlguts/"Unicode Support"> for an
1322 explanation about Unicode at the XS level, and L<perlapi> for the API
1329 C<DO_UTF8(sv)> returns true if the C<UTF8> flag is on and the bytes
1330 pragma is not in effect. C<SvUTF8(sv)> returns true if the C<UTF8>
1331 flag is on; the bytes pragma is ignored. The C<UTF8> flag being on
1332 does B<not> mean that there are any characters of code points greater
1333 than 255 (or 127) in the scalar or that there are even any characters
1334 in the scalar. What the C<UTF8> flag means is that the sequence of
1335 octets in the representation of the scalar is the sequence of UTF-8
1336 encoded code points of the characters of a string. The C<UTF8> flag
1337 being off means that each octet in this representation encodes a
1338 single character with code point 0..255 within the string. Perl's
1339 Unicode model is not to use UTF-8 until it is absolutely necessary.
1343 C<uvchr_to_utf8(buf, chr)> writes a Unicode character code point into
1344 a buffer encoding the code point as UTF-8, and returns a pointer
1345 pointing after the UTF-8 bytes. It works appropriately on EBCDIC machines.
1349 C<utf8_to_uvchr(buf, lenp)> reads UTF-8 encoded bytes from a buffer and
1350 returns the Unicode character code point and, optionally, the length of
1351 the UTF-8 byte sequence. It works appropriately on EBCDIC machines.
1355 C<utf8_length(start, end)> returns the length of the UTF-8 encoded buffer
1356 in characters. C<sv_len_utf8(sv)> returns the length of the UTF-8 encoded
1361 C<sv_utf8_upgrade(sv)> converts the string of the scalar to its UTF-8
1362 encoded form. C<sv_utf8_downgrade(sv)> does the opposite, if
1363 possible. C<sv_utf8_encode(sv)> is like sv_utf8_upgrade except that
1364 it does not set the C<UTF8> flag. C<sv_utf8_decode()> does the
1365 opposite of C<sv_utf8_encode()>. Note that none of these are to be
1366 used as general-purpose encoding or decoding interfaces: C<use Encode>
1367 for that. C<sv_utf8_upgrade()> is affected by the encoding pragma
1368 but C<sv_utf8_downgrade()> is not (since the encoding pragma is
1369 designed to be a one-way street).
1373 C<is_utf8_char(s)> returns true if the pointer points to a valid UTF-8
1378 C<is_utf8_string(buf, len)> returns true if C<len> bytes of the buffer
1383 C<UTF8SKIP(buf)> will return the number of bytes in the UTF-8 encoded
1384 character in the buffer. C<UNISKIP(chr)> will return the number of bytes
1385 required to UTF-8-encode the Unicode character code point. C<UTF8SKIP()>
1386 is useful for example for iterating over the characters of a UTF-8
1387 encoded buffer; C<UNISKIP()> is useful, for example, in computing
1388 the size required for a UTF-8 encoded buffer.
1392 C<utf8_distance(a, b)> will tell the distance in characters between the
1393 two pointers pointing to the same UTF-8 encoded buffer.
1397 C<utf8_hop(s, off)> will return a pointer to a UTF-8 encoded buffer
1398 that is C<off> (positive or negative) Unicode characters displaced
1399 from the UTF-8 buffer C<s>. Be careful not to overstep the buffer:
1400 C<utf8_hop()> will merrily run off the end or the beginning of the
1401 buffer if told to do so.
1405 C<pv_uni_display(dsv, spv, len, pvlim, flags)> and
1406 C<sv_uni_display(dsv, ssv, pvlim, flags)> are useful for debugging the
1407 output of Unicode strings and scalars. By default they are useful
1408 only for debugging--they display B<all> characters as hexadecimal code
1409 points--but with the flags C<UNI_DISPLAY_ISPRINT>,
1410 C<UNI_DISPLAY_BACKSLASH>, and C<UNI_DISPLAY_QQ> you can make the
1411 output more readable.
1415 C<ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2)> can be used to
1416 compare two strings case-insensitively in Unicode. For case-sensitive
1417 comparisons you can just use C<memEQ()> and C<memNE()> as usual.
1421 For more information, see L<perlapi>, and F<utf8.c> and F<utf8.h>
1422 in the Perl source code distribution.
1426 =head2 Interaction with Locales
1428 Use of locales with Unicode data may lead to odd results. Currently,
1429 Perl attempts to attach 8-bit locale info to characters in the range
1430 0..255, but this technique is demonstrably incorrect for locales that
1431 use characters above that range when mapped into Unicode. Perl's
1432 Unicode support will also tend to run slower. Use of locales with
1433 Unicode is discouraged.
1435 =head2 Problems with characters whose ordinal numbers are in the range 128 - 255 with no Locale specified
1437 Without a locale specified, unlike all other characters or code points,
1438 these characters have very different semantics in byte semantics versus
1439 character semantics.
1440 In character semantics they are interpreted as Unicode code points, which means
1441 they are viewed as Latin-1 (ISO-8859-1).
1442 In byte semantics, they are considered to be unassigned characters,
1443 meaning that the only semantics they have is their
1444 ordinal numbers, and that they are not members of various character classes.
1445 None are considered to match C<\w> for example, but all match C<\W>.
1446 Besides these class matches,
1447 the known operations that this affects are those that change the case,
1448 regular expression matching while ignoring case,
1450 This can lead to unexpected results in which a string's semantics suddenly
1451 change if a code point above 255 is appended to or removed from it,
1452 which changes the string's semantics from byte to character or vice versa.
1453 This behavior is scheduled to change in version 5.12, but in the meantime,
1454 a workaround is to always call utf8::upgrade($string), or to use the
1455 standard modules L<Encode> or L<charnames>.
1457 =head2 Interaction with Extensions
1459 When Perl exchanges data with an extension, the extension should be
1460 able to understand the UTF8 flag and act accordingly. If the
1461 extension doesn't know about the flag, it's likely that the extension
1462 will return incorrectly-flagged data.
1464 So if you're working with Unicode data, consult the documentation of
1465 every module you're using if there are any issues with Unicode data
1466 exchange. If the documentation does not talk about Unicode at all,
1467 suspect the worst and probably look at the source to learn how the
1468 module is implemented. Modules written completely in Perl shouldn't
1469 cause problems. Modules that directly or indirectly access code written
1470 in other programming languages are at risk.
1472 For affected functions, the simple strategy to avoid data corruption is
1473 to always make the encoding of the exchanged data explicit. Choose an
1474 encoding that you know the extension can handle. Convert arguments passed
1475 to the extensions to that encoding and convert results back from that
1476 encoding. Write wrapper functions that do the conversions for you, so
1477 you can later change the functions when the extension catches up.
1479 To provide an example, let's say the popular Foo::Bar::escape_html
1480 function doesn't deal with Unicode data yet. The wrapper function
1481 would convert the argument to raw UTF-8 and convert the result back to
1482 Perl's internal representation like so:
1484 sub my_escape_html ($) {
1486 return unless defined $what;
1487 Encode::decode_utf8(Foo::Bar::escape_html(Encode::encode_utf8($what)));
1490 Sometimes, when the extension does not convert data but just stores
1491 and retrieves them, you will be in a position to use the otherwise
1492 dangerous Encode::_utf8_on() function. Let's say the popular
1493 C<Foo::Bar> extension, written in C, provides a C<param> method that
1494 lets you store and retrieve data according to these prototypes:
1496 $self->param($name, $value); # set a scalar
1497 $value = $self->param($name); # retrieve a scalar
1499 If it does not yet provide support for any encoding, one could write a
1500 derived class with such a C<param> method:
1503 my($self,$name,$value) = @_;
1504 utf8::upgrade($name); # make sure it is UTF-8 encoded
1505 if (defined $value) {
1506 utf8::upgrade($value); # make sure it is UTF-8 encoded
1507 return $self->SUPER::param($name,$value);
1509 my $ret = $self->SUPER::param($name);
1510 Encode::_utf8_on($ret); # we know, it is UTF-8 encoded
1515 Some extensions provide filters on data entry/exit points, such as
1516 DB_File::filter_store_key and family. Look out for such filters in
1517 the documentation of your extensions, they can make the transition to
1518 Unicode data much easier.
1522 Some functions are slower when working on UTF-8 encoded strings than
1523 on byte encoded strings. All functions that need to hop over
1524 characters such as length(), substr() or index(), or matching regular
1525 expressions can work B<much> faster when the underlying data are
1528 In Perl 5.8.0 the slowness was often quite spectacular; in Perl 5.8.1
1529 a caching scheme was introduced which will hopefully make the slowness
1530 somewhat less spectacular, at least for some operations. In general,
1531 operations with UTF-8 encoded strings are still slower. As an example,
1532 the Unicode properties (character classes) like C<\p{Nd}> are known to
1533 be quite a bit slower (5-20 times) than their simpler counterparts
1534 like C<\d> (then again, there 268 Unicode characters matching C<Nd>
1535 compared with the 10 ASCII characters matching C<d>).
1537 =head2 Possible problems on EBCDIC platforms
1539 In earlier versions, when byte and character data were concatenated,
1540 the new string was sometimes created by
1541 decoding the byte strings as I<ISO 8859-1 (Latin-1)>, even if the
1542 old Unicode string used EBCDIC.
1544 If you find any of these, please report them as bugs.
1546 =head2 Porting code from perl-5.6.X
1548 Perl 5.8 has a different Unicode model from 5.6. In 5.6 the programmer
1549 was required to use the C<utf8> pragma to declare that a given scope
1550 expected to deal with Unicode data and had to make sure that only
1551 Unicode data were reaching that scope. If you have code that is
1552 working with 5.6, you will need some of the following adjustments to
1553 your code. The examples are written such that the code will continue
1554 to work under 5.6, so you should be safe to try them out.
1560 A filehandle that should read or write UTF-8
1563 binmode $fh, ":encoding(utf8)";
1568 A scalar that is going to be passed to some extension
1570 Be it Compress::Zlib, Apache::Request or any extension that has no
1571 mention of Unicode in the manpage, you need to make sure that the
1572 UTF8 flag is stripped off. Note that at the time of this writing
1573 (October 2002) the mentioned modules are not UTF-8-aware. Please
1574 check the documentation to verify if this is still true.
1578 $val = Encode::encode_utf8($val); # make octets
1583 A scalar we got back from an extension
1585 If you believe the scalar comes back as UTF-8, you will most likely
1586 want the UTF8 flag restored:
1590 $val = Encode::decode_utf8($val);
1595 Same thing, if you are really sure it is UTF-8
1599 Encode::_utf8_on($val);
1604 A wrapper for fetchrow_array and fetchrow_hashref
1606 When the database contains only UTF-8, a wrapper function or method is
1607 a convenient way to replace all your fetchrow_array and
1608 fetchrow_hashref calls. A wrapper function will also make it easier to
1609 adapt to future enhancements in your database driver. Note that at the
1610 time of this writing (October 2002), the DBI has no standardized way
1611 to deal with UTF-8 data. Please check the documentation to verify if
1615 my($self, $sth, $what) = @_; # $what is one of fetchrow_{array,hashref}
1621 my @arr = $sth->$what;
1623 defined && /[^\000-\177]/ && Encode::_utf8_on($_);
1627 my $ret = $sth->$what;
1629 for my $k (keys %$ret) {
1630 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret->{$k};
1634 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret;
1644 A large scalar that you know can only contain ASCII
1646 Scalars that contain only ASCII and are marked as UTF-8 are sometimes
1647 a drag to your program. If you recognize such a situation, just remove
1650 utf8::downgrade($val) if $] > 5.007;
1656 L<perlunitut>, L<perluniintro>, L<Encode>, L<open>, L<utf8>, L<bytes>,
1657 L<perlretut>, L<perlvar/"${^UNICODE}">