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.
15 =item Input and Output Layers
17 Perl knows when a filehandle uses Perl's internal Unicode encodings
18 (UTF-8, or UTF-EBCDIC if in EBCDIC) if the filehandle is opened with
19 the ":utf8" layer. Other encodings can be converted to Perl's
20 encoding on input or from Perl's encoding on output by use of the
21 ":encoding(...)" layer. See L<open>.
23 To indicate that Perl source itself is using a particular encoding,
26 =item Regular Expressions
28 The regular expression compiler produces polymorphic opcodes. That is,
29 the pattern adapts to the data and automatically switches to the Unicode
30 character scheme when presented with Unicode data--or instead uses
31 a traditional byte scheme when presented with byte data.
33 =item C<use utf8> still needed to enable UTF-8/UTF-EBCDIC in scripts
35 As a compatibility measure, the C<use utf8> pragma must be explicitly
36 included to enable recognition of UTF-8 in the Perl scripts themselves
37 (in string or regular expression literals, or in identifier names) on
38 ASCII-based machines or to recognize UTF-EBCDIC on EBCDIC-based
39 machines. B<These are the only times when an explicit C<use utf8>
40 is needed.> See L<utf8>.
42 You can also use the C<encoding> pragma to change the default encoding
43 of the data in your script; see L<encoding>.
45 =item BOM-marked scripts and UTF-16 scripts autodetected
47 If a Perl script begins marked with the Unicode BOM (UTF-16LE, UTF16-BE,
48 or UTF-8), or if the script looks like non-BOM-marked UTF-16 of either
49 endianness, Perl will correctly read in the script as Unicode.
50 (BOMless UTF-8 cannot be effectively recognized or differentiated from
51 ISO 8859-1 or other eight-bit encodings.)
53 =item C<use encoding> needed to upgrade non-Latin-1 byte strings
55 By default, there is a fundamental asymmetry in Perl's unicode model:
56 implicit upgrading from byte strings to Unicode strings assumes that
57 they were encoded in I<ISO 8859-1 (Latin-1)>, but Unicode strings are
58 downgraded with UTF-8 encoding. This happens because the first 256
59 codepoints in Unicode happens to agree with Latin-1.
61 If you wish to interpret byte strings as UTF-8 instead, use the
66 See L</"Byte and Character Semantics"> for more details.
70 =head2 Byte and Character Semantics
72 Beginning with version 5.6, Perl uses logically-wide characters to
73 represent strings internally.
75 In future, Perl-level operations will be expected to work with
76 characters rather than bytes.
78 However, as an interim compatibility measure, Perl aims to
79 provide a safe migration path from byte semantics to character
80 semantics for programs. For operations where Perl can unambiguously
81 decide that the input data are characters, Perl switches to
82 character semantics. For operations where this determination cannot
83 be made without additional information from the user, Perl decides in
84 favor of compatibility and chooses to use byte semantics.
86 This behavior preserves compatibility with earlier versions of Perl,
87 which allowed byte semantics in Perl operations only if
88 none of the program's inputs were marked as being as source of Unicode
89 character data. Such data may come from filehandles, from calls to
90 external programs, from information provided by the system (such as %ENV),
91 or from literals and constants in the source text.
93 The C<bytes> pragma will always, regardless of platform, force byte
94 semantics in a particular lexical scope. See L<bytes>.
96 The C<utf8> pragma is primarily a compatibility device that enables
97 recognition of UTF-(8|EBCDIC) in literals encountered by the parser.
98 Note that this pragma is only required while Perl defaults to byte
99 semantics; when character semantics become the default, this pragma
100 may become a no-op. See L<utf8>.
102 Unless explicitly stated, Perl operators use character semantics
103 for Unicode data and byte semantics for non-Unicode data.
104 The decision to use character semantics is made transparently. If
105 input data comes from a Unicode source--for example, if a character
106 encoding layer is added to a filehandle or a literal Unicode
107 string constant appears in a program--character semantics apply.
108 Otherwise, byte semantics are in effect. The C<bytes> pragma should
109 be used to force byte semantics on Unicode data.
111 If strings operating under byte semantics and strings with Unicode
112 character data are concatenated, the new string will be created by
113 decoding the byte strings as I<ISO 8859-1 (Latin-1)>, even if the
114 old Unicode string used EBCDIC. This translation is done without
115 regard to the system's native 8-bit encoding. To change this for
116 systems with non-Latin-1 and non-EBCDIC native encodings, use the
117 C<encoding> pragma. See L<encoding>.
119 Under character semantics, many operations that formerly operated on
120 bytes now operate on characters. A character in Perl is
121 logically just a number ranging from 0 to 2**31 or so. Larger
122 characters may encode into longer sequences of bytes internally, but
123 this internal detail is mostly hidden for Perl code.
124 See L<perluniintro> for more.
126 =head2 Effects of Character Semantics
128 Character semantics have the following effects:
134 Strings--including hash keys--and regular expression patterns may
135 contain characters that have an ordinal value larger than 255.
137 If you use a Unicode editor to edit your program, Unicode characters
138 may occur directly within the literal strings in one of the various
139 Unicode encodings (UTF-8, UTF-EBCDIC, UCS-2, etc.), but will be recognized
140 as such and converted to Perl's internal representation only if the
141 appropriate L<encoding> is specified.
143 Unicode characters can also be added to a string by using the
144 C<\x{...}> notation. The Unicode code for the desired character, in
145 hexadecimal, should be placed in the braces. For instance, a smiley
146 face is C<\x{263A}>. This encoding scheme only works for characters
147 with a code of 0x100 or above.
151 use charnames ':full';
153 you can use the C<\N{...}> notation and put the official Unicode
154 character name within the braces, such as C<\N{WHITE SMILING FACE}>.
158 If an appropriate L<encoding> is specified, identifiers within the
159 Perl script may contain Unicode alphanumeric characters, including
160 ideographs. Perl does not currently attempt to canonicalize variable
165 Regular expressions match characters instead of bytes. "." matches
166 a character instead of a byte. The C<\C> pattern is provided to force
167 a match a single byte--a C<char> in C, hence C<\C>.
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.
176 (However, and as a limitation of the current implementation, using
177 C<\w> or C<\W> I<inside> a C<[...]> character class will still match
178 with byte semantics.)
182 Named Unicode properties, scripts, and block ranges may be used like
183 character classes via the C<\p{}> "matches property" construct and
184 the C<\P{}> negation, "doesn't match property".
186 See L</"Unicode Character Properties"> for more details.
188 You can define your own character properties and use them
189 in the regular expression with the C<\p{}> or C<\P{}> construct.
191 See L</"User-Defined Character Properties"> for more details.
195 The special pattern C<\X> matches any extended Unicode
196 sequence--"a combining character sequence" in Standardese--where the
197 first character is a base character and subsequent characters are mark
198 characters that apply to the base character. C<\X> is equivalent to
203 The C<tr///> operator translates characters instead of bytes. Note
204 that the C<tr///CU> functionality has been removed. For similar
205 functionality see pack('U0', ...) and pack('C0', ...).
209 Case translation operators use the Unicode case translation tables
210 when character input is provided. Note that C<uc()>, or C<\U> in
211 interpolated strings, translates to uppercase, while C<ucfirst>,
212 or C<\u> in interpolated strings, translates to titlecase in languages
213 that make the distinction.
217 Most operators that deal with positions or lengths in a string will
218 automatically switch to using character positions, including
219 C<chop()>, C<chomp()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
220 C<sprintf()>, C<write()>, and C<length()>. An operator that
221 specifically does not switch is C<vec()>. Operators that really don't
222 care include operators that treat strings as a bucket of bits such as
223 C<sort()>, and operators dealing with filenames.
227 The C<pack()>/C<unpack()> letter C<C> does I<not> change, since it is often
228 used for byte-oriented formats. Again, think C<char> in the C language.
230 There is a new C<U> specifier that converts between Unicode characters
231 and code points. There is also a C<W> specifier that is the equivalent of
232 C<chr>/C<ord> and properly handles character values even if they are above 255.
236 The C<chr()> and C<ord()> functions work on characters, similar to
237 C<pack("W")> and C<unpack("W")>, I<not> C<pack("C")> and
238 C<unpack("C")>. C<pack("C")> and C<unpack("C")> are methods for
239 emulating byte-oriented C<chr()> and C<ord()> on Unicode strings.
240 While these methods reveal the internal encoding of Unicode strings,
241 that is not something one normally needs to care about at all.
245 The bit string operators, C<& | ^ ~>, can operate on character data.
246 However, for backward compatibility, such as when using bit string
247 operations when characters are all less than 256 in ordinal value, one
248 should not use C<~> (the bit complement) with characters of both
249 values less than 256 and values greater than 256. Most importantly,
250 DeMorgan's laws (C<~($x|$y) eq ~$x&~$y> and C<~($x&$y) eq ~$x|~$y>)
251 will not hold. The reason for this mathematical I<faux pas> is that
252 the complement cannot return B<both> the 8-bit (byte-wide) bit
253 complement B<and> the full character-wide bit complement.
257 lc(), uc(), lcfirst(), and ucfirst() work for the following cases:
263 the case mapping is from a single Unicode character to another
264 single Unicode character, or
268 the case mapping is from a single Unicode character to more
269 than one Unicode character.
273 Things to do with locales (Lithuanian, Turkish, Azeri) do B<not> work
274 since Perl does not understand the concept of Unicode locales.
276 See the Unicode Technical Report #21, Case Mappings, for more details.
278 But you can also define your own mappings to be used in the lc(),
279 lcfirst(), uc(), and ucfirst() (or their string-inlined versions).
281 See L</"User-Defined Case Mappings"> for more details.
289 And finally, C<scalar reverse()> reverses by character rather than by byte.
293 =head2 Unicode Character Properties
295 Named Unicode properties, scripts, and block ranges may be used like
296 character classes via the C<\p{}> "matches property" construct and
297 the C<\P{}> negation, "doesn't match property".
299 For instance, C<\p{Lu}> matches any character with the Unicode "Lu"
300 (Letter, uppercase) property, while C<\p{M}> matches any character
301 with an "M" (mark--accents and such) property. Brackets are not
302 required for single letter properties, so C<\p{M}> is equivalent to
303 C<\pM>. Many predefined properties are available, such as
304 C<\p{Mirrored}> and C<\p{Tibetan}>.
306 The official Unicode script and block names have spaces and dashes as
307 separators, but for convenience you can use dashes, spaces, or
308 underbars, and case is unimportant. It is recommended, however, that
309 for consistency you use the following naming: the official Unicode
310 script, property, or block name (see below for the additional rules
311 that apply to block names) with whitespace and dashes removed, and the
312 words "uppercase-first-lowercase-rest". C<Latin-1 Supplement> thus
313 becomes C<Latin1Supplement>.
315 You can also use negation in both C<\p{}> and C<\P{}> by introducing a caret
316 (^) between the first brace and the property name: C<\p{^Tamil}> is
317 equal to C<\P{Tamil}>.
319 B<NOTE: the properties, scripts, and blocks listed here are as of
320 Unicode 3.2.0, March 2002, or Perl 5.8.0, July 2002. Unicode 4.0.0
321 came out in April 2003, and Perl 5.8.1 in September 2003.>
325 =item General Category
327 Here are the basic Unicode General Category properties, followed by their
328 long form. You can use either; C<\p{Lu}> and C<\p{UppercaseLetter}>,
329 for instance, are identical.
352 Pc ConnectorPunctuation
356 Pi InitialPunctuation
357 (may behave like Ps or Pe depending on usage)
359 (may behave like Ps or Pe depending on usage)
371 Zp ParagraphSeparator
376 Cs Surrogate (not usable)
380 Single-letter properties match all characters in any of the
381 two-letter sub-properties starting with the same letter.
382 C<LC> and C<L&> are special cases, which are aliases for the set of
383 C<Ll>, C<Lu>, and C<Lt>.
385 Because Perl hides the need for the user to understand the internal
386 representation of Unicode characters, there is no need to implement
387 the somewhat messy concept of surrogates. C<Cs> is therefore not
390 =item Bidirectional Character Types
392 Because scripts differ in their directionality--Hebrew is
393 written right to left, for example--Unicode supplies these properties in
399 LRE Left-to-Right Embedding
400 LRO Left-to-Right Override
402 AL Right-to-Left Arabic
403 RLE Right-to-Left Embedding
404 RLO Right-to-Left Override
405 PDF Pop Directional Format
407 ES European Number Separator
408 ET European Number Terminator
410 CS Common Number Separator
413 B Paragraph Separator
418 For example, C<\p{BidiClass:R}> matches characters that are normally
419 written right to left.
423 The script names which can be used by C<\p{...}> and C<\P{...}>,
424 such as in C<\p{Latin}> or C<\p{Cyrillic}>, are as follows:
471 =item Extended property classes
473 Extended property classes can supplement the basic
474 properties, defined by the F<PropList> Unicode database:
489 LogicalOrderException
490 NoncharacterCodePoint
492 OtherDefaultIgnorableCodePoint
504 and there are further derived properties:
506 Alphabetic Lu + Ll + Lt + Lm + Lo + OtherAlphabetic
507 Lowercase Ll + OtherLowercase
508 Uppercase Lu + OtherUppercase
511 ID_Start Lu + Ll + Lt + Lm + Lo + Nl
512 ID_Continue ID_Start + Mn + Mc + Nd + Pc
515 Assigned Any non-Cn character (i.e. synonym for \P{Cn})
516 Unassigned Synonym for \p{Cn}
517 Common Any character (or unassigned code point)
518 not explicitly assigned to a script
520 =item Use of "Is" Prefix
522 For backward compatibility (with Perl 5.6), all properties mentioned
523 so far may have C<Is> prepended to their name, so C<\P{IsLu}>, for
524 example, is equal to C<\P{Lu}>.
528 In addition to B<scripts>, Unicode also defines B<blocks> of
529 characters. The difference between scripts and blocks is that the
530 concept of scripts is closer to natural languages, while the concept
531 of blocks is more of an artificial grouping based on groups of 256
532 Unicode characters. For example, the C<Latin> script contains letters
533 from many blocks but does not contain all the characters from those
534 blocks. It does not, for example, contain digits, because digits are
535 shared across many scripts. Digits and similar groups, like
536 punctuation, are in a category called C<Common>.
538 For more about scripts, see the UTR #24:
540 http://www.unicode.org/unicode/reports/tr24/
542 For more about blocks, see:
544 http://www.unicode.org/Public/UNIDATA/Blocks.txt
546 Block names are given with the C<In> prefix. For example, the
547 Katakana block is referenced via C<\p{InKatakana}>. The C<In>
548 prefix may be omitted if there is no naming conflict with a script
549 or any other property, but it is recommended that C<In> always be used
550 for block tests to avoid confusion.
552 These block names are supported:
554 InAlphabeticPresentationForms
556 InArabicPresentationFormsA
557 InArabicPresentationFormsB
568 InByzantineMusicalSymbols
570 InCJKCompatibilityForms
571 InCJKCompatibilityIdeographs
572 InCJKCompatibilityIdeographsSupplement
573 InCJKRadicalsSupplement
574 InCJKSymbolsAndPunctuation
575 InCJKUnifiedIdeographs
576 InCJKUnifiedIdeographsExtensionA
577 InCJKUnifiedIdeographsExtensionB
579 InCombiningDiacriticalMarks
580 InCombiningDiacriticalMarksforSymbols
585 InCyrillicSupplementary
589 InEnclosedAlphanumerics
590 InEnclosedCJKLettersAndMonths
600 InHalfwidthAndFullwidthForms
601 InHangulCompatibilityJamo
606 InHighPrivateUseSurrogates
610 InIdeographicDescriptionCharacters
615 InKatakanaPhoneticExtensions
620 InLatinExtendedAdditional
625 InMathematicalAlphanumericSymbols
626 InMathematicalOperators
627 InMiscellaneousMathematicalSymbolsA
628 InMiscellaneousMathematicalSymbolsB
629 InMiscellaneousSymbols
630 InMiscellaneousTechnical
637 InOpticalCharacterRecognition
643 InSpacingModifierLetters
645 InSuperscriptsAndSubscripts
646 InSupplementalArrowsA
647 InSupplementalArrowsB
648 InSupplementalMathematicalOperators
649 InSupplementaryPrivateUseAreaA
650 InSupplementaryPrivateUseAreaB
660 InUnifiedCanadianAboriginalSyllabics
667 =head2 User-Defined Character Properties
669 You can define your own character properties by defining subroutines
670 whose names begin with "In" or "Is". The subroutines can be defined in
671 any package. The user-defined properties can be used in the regular
672 expression C<\p> and C<\P> constructs; if you are using a user-defined
673 property from a package other than the one you are in, you must specify
674 its package in the C<\p> or C<\P> construct.
676 # assuming property IsForeign defined in Lang::
677 package main; # property package name required
678 if ($txt =~ /\p{Lang::IsForeign}+/) { ... }
680 package Lang; # property package name not required
681 if ($txt =~ /\p{IsForeign}+/) { ... }
684 Note that the effect is compile-time and immutable once defined.
686 The subroutines must return a specially-formatted string, with one
687 or more newline-separated lines. Each line must be one of the following:
693 Two hexadecimal numbers separated by horizontal whitespace (space or
694 tabular characters) denoting a range of Unicode code points to include.
698 Something to include, prefixed by "+": a built-in character
699 property (prefixed by "utf8::") or a user-defined character property,
700 to represent all the characters in that property; two hexadecimal code
701 points for a range; or a single hexadecimal code point.
705 Something to exclude, prefixed by "-": an existing character
706 property (prefixed by "utf8::") or a user-defined character property,
707 to represent all the characters in that property; two hexadecimal code
708 points for a range; or a single hexadecimal code point.
712 Something to negate, prefixed "!": an existing character
713 property (prefixed by "utf8::") or a user-defined character property,
714 to represent all the characters in that property; two hexadecimal code
715 points for a range; or a single hexadecimal code point.
719 Something to intersect with, prefixed by "&": an existing character
720 property (prefixed by "utf8::") or a user-defined character property,
721 for all the characters except the characters in the property; two
722 hexadecimal code points for a range; or a single hexadecimal code point.
726 For example, to define a property that covers both the Japanese
727 syllabaries (hiragana and katakana), you can define
736 Imagine that the here-doc end marker is at the beginning of the line.
737 Now you can use C<\p{InKana}> and C<\P{InKana}>.
739 You could also have used the existing block property names:
748 Suppose you wanted to match only the allocated characters,
749 not the raw block ranges: in other words, you want to remove
760 The negation is useful for defining (surprise!) negated classes.
770 Intersection is useful for getting the common characters matched by
771 two (or more) classes.
780 It's important to remember not to use "&" for the first set -- that
781 would be intersecting with nothing (resulting in an empty set).
783 A final note on the user-defined property tests: they will be used
784 only if the scalar has been marked as having Unicode characters.
785 Old byte-style strings will not be affected.
787 =head2 User-Defined Case Mappings
789 You can also define your own mappings to be used in the lc(),
790 lcfirst(), uc(), and ucfirst() (or their string-inlined versions).
791 The principle is similar to that of user-defined character
792 properties: to define subroutines in the C<main> package
793 with names like C<ToLower> (for lc() and lcfirst()), C<ToTitle> (for
794 the first character in ucfirst()), and C<ToUpper> (for uc(), and the
795 rest of the characters in ucfirst()).
797 The string returned by the subroutines needs now to be three
798 hexadecimal numbers separated by tabulators: start of the source
799 range, end of the source range, and start of the destination range.
808 defines an uc() mapping that causes only the characters "a", "b", and
809 "c" to be mapped to "A", "B", "C", all other characters will remain
812 If there is no source range to speak of, that is, the mapping is from
813 a single character to another single character, leave the end of the
814 source range empty, but the two tabulator characters are still needed.
823 defines a lc() mapping that causes only "A" to be mapped to "a", all
824 other characters will remain unchanged.
826 (For serious hackers only) If you want to introspect the default
827 mappings, you can find the data in the directory
828 C<$Config{privlib}>/F<unicore/To/>. The mapping data is returned as
829 the here-document, and the C<utf8::ToSpecFoo> are special exception
830 mappings derived from <$Config{privlib}>/F<unicore/SpecialCasing.txt>.
831 The C<Digit> and C<Fold> mappings that one can see in the directory
832 are not directly user-accessible, one can use either the
833 C<Unicode::UCD> module, or just match case-insensitively (that's when
834 the C<Fold> mapping is used).
836 A final note on the user-defined case mappings: they will be used
837 only if the scalar has been marked as having Unicode characters.
838 Old byte-style strings will not be affected.
840 =head2 Character Encodings for Input and Output
844 =head2 Unicode Regular Expression Support Level
846 The following list of Unicode support for regular expressions describes
847 all the features currently supported. The references to "Level N"
848 and the section numbers refer to the Unicode Technical Report 18,
849 "Unicode Regular Expression Guidelines", version 6 (Unicode 3.2.0,
856 Level 1 - Basic Unicode Support
858 2.1 Hex Notation - done [1]
859 Named Notation - done [2]
860 2.2 Categories - done [3][4]
861 2.3 Subtraction - MISSING [5][6]
862 2.4 Simple Word Boundaries - done [7]
863 2.5 Simple Loose Matches - done [8]
864 2.6 End of Line - MISSING [9][10]
868 [ 3] . \p{...} \P{...}
869 [ 4] support for scripts (see UTR#24 Script Names), blocks,
870 binary properties, enumerated non-binary properties, and
871 numeric properties (as listed in UTR#18 Other Properties)
873 [ 6] can use regular expression look-ahead [a]
874 or user-defined character properties [b] to emulate subtraction
875 [ 7] include Letters in word characters
876 [ 8] note that Perl does Full case-folding in matching, not Simple:
877 for example U+1F88 is equivalent with U+1F00 U+03B9,
878 not with 1F80. This difference matters for certain Greek
879 capital letters with certain modifiers: the Full case-folding
880 decomposes the letter, while the Simple case-folding would map
881 it to a single character.
882 [ 9] see UTR #13 Unicode Newline Guidelines
883 [10] should do ^ and $ also on \x{85}, \x{2028} and \x{2029}
884 (should also affect <>, $., and script line numbers)
885 (the \x{85}, \x{2028} and \x{2029} do match \s)
887 [a] You can mimic class subtraction using lookahead.
888 For example, what UTR #18 might write as
890 [{Greek}-[{UNASSIGNED}]]
892 in Perl can be written as:
894 (?!\p{Unassigned})\p{InGreekAndCoptic}
895 (?=\p{Assigned})\p{InGreekAndCoptic}
897 But in this particular example, you probably really want
901 which will match assigned characters known to be part of the Greek script.
903 Also see the Unicode::Regex::Set module, it does implement the full
904 UTR #18 grouping, intersection, union, and removal (subtraction) syntax.
906 [b] See L</"User-Defined Character Properties">.
910 Level 2 - Extended Unicode Support
912 3.1 Surrogates - MISSING [11]
913 3.2 Canonical Equivalents - MISSING [12][13]
914 3.3 Locale-Independent Graphemes - MISSING [14]
915 3.4 Locale-Independent Words - MISSING [15]
916 3.5 Locale-Independent Loose Matches - MISSING [16]
918 [11] Surrogates are solely a UTF-16 concept and Perl's internal
919 representation is UTF-8. The Encode module does UTF-16, though.
920 [12] see UTR#15 Unicode Normalization
921 [13] have Unicode::Normalize but not integrated to regexes
922 [14] have \X but at this level . should equal that
923 [15] need three classes, not just \w and \W
924 [16] see UTR#21 Case Mappings
928 Level 3 - Locale-Sensitive Support
930 4.1 Locale-Dependent Categories - MISSING
931 4.2 Locale-Dependent Graphemes - MISSING [16][17]
932 4.3 Locale-Dependent Words - MISSING
933 4.4 Locale-Dependent Loose Matches - MISSING
934 4.5 Locale-Dependent Ranges - MISSING
936 [16] see UTR#10 Unicode Collation Algorithms
937 [17] have Unicode::Collate but not integrated to regexes
941 =head2 Unicode Encodings
943 Unicode characters are assigned to I<code points>, which are abstract
944 numbers. To use these numbers, various encodings are needed.
952 UTF-8 is a variable-length (1 to 6 bytes, current character allocations
953 require 4 bytes), byte-order independent encoding. For ASCII (and we
954 really do mean 7-bit ASCII, not another 8-bit encoding), UTF-8 is
957 The following table is from Unicode 3.2.
959 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
961 U+0000..U+007F 00..7F
962 U+0080..U+07FF C2..DF 80..BF
963 U+0800..U+0FFF E0 A0..BF 80..BF
964 U+1000..U+CFFF E1..EC 80..BF 80..BF
965 U+D000..U+D7FF ED 80..9F 80..BF
966 U+D800..U+DFFF ******* ill-formed *******
967 U+E000..U+FFFF EE..EF 80..BF 80..BF
968 U+10000..U+3FFFF F0 90..BF 80..BF 80..BF
969 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
970 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
972 Note the C<A0..BF> in C<U+0800..U+0FFF>, the C<80..9F> in
973 C<U+D000...U+D7FF>, the C<90..B>F in C<U+10000..U+3FFFF>, and the
974 C<80...8F> in C<U+100000..U+10FFFF>. The "gaps" are caused by legal
975 UTF-8 avoiding non-shortest encodings: it is technically possible to
976 UTF-8-encode a single code point in different ways, but that is
977 explicitly forbidden, and the shortest possible encoding should always
978 be used. So that's what Perl does.
980 Another way to look at it is via bits:
982 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
985 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
986 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
987 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
989 As you can see, the continuation bytes all begin with C<10>, and the
990 leading bits of the start byte tell how many bytes the are in the
997 Like UTF-8 but EBCDIC-safe, in the way that UTF-8 is ASCII-safe.
1001 UTF-16, UTF-16BE, UTF-16LE, Surrogates, and BOMs (Byte Order Marks)
1003 The followings items are mostly for reference and general Unicode
1004 knowledge, Perl doesn't use these constructs internally.
1006 UTF-16 is a 2 or 4 byte encoding. The Unicode code points
1007 C<U+0000..U+FFFF> are stored in a single 16-bit unit, and the code
1008 points C<U+10000..U+10FFFF> in two 16-bit units. The latter case is
1009 using I<surrogates>, the first 16-bit unit being the I<high
1010 surrogate>, and the second being the I<low surrogate>.
1012 Surrogates are code points set aside to encode the C<U+10000..U+10FFFF>
1013 range of Unicode code points in pairs of 16-bit units. The I<high
1014 surrogates> are the range C<U+D800..U+DBFF>, and the I<low surrogates>
1015 are the range C<U+DC00..U+DFFF>. The surrogate encoding is
1017 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
1018 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
1022 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
1024 If you try to generate surrogates (for example by using chr()), you
1025 will get a warning if warnings are turned on, because those code
1026 points are not valid for a Unicode character.
1028 Because of the 16-bitness, UTF-16 is byte-order dependent. UTF-16
1029 itself can be used for in-memory computations, but if storage or
1030 transfer is required either UTF-16BE (big-endian) or UTF-16LE
1031 (little-endian) encodings must be chosen.
1033 This introduces another problem: what if you just know that your data
1034 is UTF-16, but you don't know which endianness? Byte Order Marks, or
1035 BOMs, are a solution to this. A special character has been reserved
1036 in Unicode to function as a byte order marker: the character with the
1037 code point C<U+FEFF> is the BOM.
1039 The trick is that if you read a BOM, you will know the byte order,
1040 since if it was written on a big-endian platform, you will read the
1041 bytes C<0xFE 0xFF>, but if it was written on a little-endian platform,
1042 you will read the bytes C<0xFF 0xFE>. (And if the originating platform
1043 was writing in UTF-8, you will read the bytes C<0xEF 0xBB 0xBF>.)
1045 The way this trick works is that the character with the code point
1046 C<U+FFFE> is guaranteed not to be a valid Unicode character, so the
1047 sequence of bytes C<0xFF 0xFE> is unambiguously "BOM, represented in
1048 little-endian format" and cannot be C<U+FFFE>, represented in big-endian
1053 UTF-32, UTF-32BE, UTF-32LE
1055 The UTF-32 family is pretty much like the UTF-16 family, expect that
1056 the units are 32-bit, and therefore the surrogate scheme is not
1057 needed. The BOM signatures will be C<0x00 0x00 0xFE 0xFF> for BE and
1058 C<0xFF 0xFE 0x00 0x00> for LE.
1064 Encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
1065 encoding. Unlike UTF-16, UCS-2 is not extensible beyond C<U+FFFF>,
1066 because it does not use surrogates. UCS-4 is a 32-bit encoding,
1067 functionally identical to UTF-32.
1073 A seven-bit safe (non-eight-bit) encoding, which is useful if the
1074 transport or storage is not eight-bit safe. Defined by RFC 2152.
1078 =head2 Security Implications of Unicode
1086 Unfortunately, the specification of UTF-8 leaves some room for
1087 interpretation of how many bytes of encoded output one should generate
1088 from one input Unicode character. Strictly speaking, the shortest
1089 possible sequence of UTF-8 bytes should be generated,
1090 because otherwise there is potential for an input buffer overflow at
1091 the receiving end of a UTF-8 connection. Perl always generates the
1092 shortest length UTF-8, and with warnings on Perl will warn about
1093 non-shortest length UTF-8 along with other malformations, such as the
1094 surrogates, which are not real Unicode code points.
1098 Regular expressions behave slightly differently between byte data and
1099 character (Unicode) data. For example, the "word character" character
1100 class C<\w> will work differently depending on if data is eight-bit bytes
1103 In the first case, the set of C<\w> characters is either small--the
1104 default set of alphabetic characters, digits, and the "_"--or, if you
1105 are using a locale (see L<perllocale>), the C<\w> might contain a few
1106 more letters according to your language and country.
1108 In the second case, the C<\w> set of characters is much, much larger.
1109 Most importantly, even in the set of the first 256 characters, it will
1110 probably match different characters: unlike most locales, which are
1111 specific to a language and country pair, Unicode classifies all the
1112 characters that are letters I<somewhere> as C<\w>. For example, your
1113 locale might not think that LATIN SMALL LETTER ETH is a letter (unless
1114 you happen to speak Icelandic), but Unicode does.
1116 As discussed elsewhere, Perl has one foot (two hooves?) planted in
1117 each of two worlds: the old world of bytes and the new world of
1118 characters, upgrading from bytes to characters when necessary.
1119 If your legacy code does not explicitly use Unicode, no automatic
1120 switch-over to characters should happen. Characters shouldn't get
1121 downgraded to bytes, either. It is possible to accidentally mix bytes
1122 and characters, however (see L<perluniintro>), in which case C<\w> in
1123 regular expressions might start behaving differently. Review your
1124 code. Use warnings and the C<strict> pragma.
1128 =head2 Unicode in Perl on EBCDIC
1130 The way Unicode is handled on EBCDIC platforms is still
1131 experimental. On such platforms, references to UTF-8 encoding in this
1132 document and elsewhere should be read as meaning the UTF-EBCDIC
1133 specified in Unicode Technical Report 16, unless ASCII vs. EBCDIC issues
1134 are specifically discussed. There is no C<utfebcdic> pragma or
1135 ":utfebcdic" layer; rather, "utf8" and ":utf8" are reused to mean
1136 the platform's "natural" 8-bit encoding of Unicode. See L<perlebcdic>
1137 for more discussion of the issues.
1141 Usually locale settings and Unicode do not affect each other, but
1142 there are a couple of exceptions:
1148 You can enable automatic UTF-8-ification of your standard file
1149 handles, default C<open()> layer, and C<@ARGV> by using either
1150 the C<-C> command line switch or the C<PERL_UNICODE> environment
1151 variable, see L<perlrun> for the documentation of the C<-C> switch.
1155 Perl tries really hard to work both with Unicode and the old
1156 byte-oriented world. Most often this is nice, but sometimes Perl's
1157 straddling of the proverbial fence causes problems.
1161 =head2 When Unicode Does Not Happen
1163 While Perl does have extensive ways to input and output in Unicode,
1164 and few other 'entry points' like the @ARGV which can be interpreted
1165 as Unicode (UTF-8), there still are many places where Unicode (in some
1166 encoding or another) could be given as arguments or received as
1167 results, or both, but it is not.
1169 The following are such interfaces. For all of these interfaces Perl
1170 currently (as of 5.8.3) simply assumes byte strings both as arguments
1171 and results, or UTF-8 strings if the C<encoding> pragma has been used.
1173 One reason why Perl does not attempt to resolve the role of Unicode in
1174 this cases is that the answers are highly dependent on the operating
1175 system and the file system(s). For example, whether filenames can be
1176 in Unicode, and in exactly what kind of encoding, is not exactly a
1177 portable concept. Similarly for the qx and system: how well will the
1178 'command line interface' (and which of them?) handle Unicode?
1184 chdir, chmod, chown, chroot, exec, link, lstat, mkdir,
1185 rename, rmdir, stat, symlink, truncate, unlink, utime, -X
1197 open, opendir, sysopen
1201 qx (aka the backtick operator), system
1209 =head2 Forcing Unicode in Perl (Or Unforcing Unicode in Perl)
1211 Sometimes (see L</"When Unicode Does Not Happen">) there are
1212 situations where you simply need to force Perl to believe that a byte
1213 string is UTF-8, or vice versa. The low-level calls
1214 utf8::upgrade($bytestring) and utf8::downgrade($utf8string) are
1217 Do not use them without careful thought, though: Perl may easily get
1218 very confused, angry, or even crash, if you suddenly change the 'nature'
1219 of scalar like that. Especially careful you have to be if you use the
1220 utf8::upgrade(): any random byte string is not valid UTF-8.
1222 =head2 Using Unicode in XS
1224 If you want to handle Perl Unicode in XS extensions, you may find the
1225 following C APIs useful. See also L<perlguts/"Unicode Support"> for an
1226 explanation about Unicode at the XS level, and L<perlapi> for the API
1233 C<DO_UTF8(sv)> returns true if the C<UTF8> flag is on and the bytes
1234 pragma is not in effect. C<SvUTF8(sv)> returns true is the C<UTF8>
1235 flag is on; the bytes pragma is ignored. The C<UTF8> flag being on
1236 does B<not> mean that there are any characters of code points greater
1237 than 255 (or 127) in the scalar or that there are even any characters
1238 in the scalar. What the C<UTF8> flag means is that the sequence of
1239 octets in the representation of the scalar is the sequence of UTF-8
1240 encoded code points of the characters of a string. The C<UTF8> flag
1241 being off means that each octet in this representation encodes a
1242 single character with code point 0..255 within the string. Perl's
1243 Unicode model is not to use UTF-8 until it is absolutely necessary.
1247 C<uvuni_to_utf8(buf, chr)> writes a Unicode character code point into
1248 a buffer encoding the code point as UTF-8, and returns a pointer
1249 pointing after the UTF-8 bytes.
1253 C<utf8_to_uvuni(buf, lenp)> reads UTF-8 encoded bytes from a buffer and
1254 returns the Unicode character code point and, optionally, the length of
1255 the UTF-8 byte sequence.
1259 C<utf8_length(start, end)> returns the length of the UTF-8 encoded buffer
1260 in characters. C<sv_len_utf8(sv)> returns the length of the UTF-8 encoded
1265 C<sv_utf8_upgrade(sv)> converts the string of the scalar to its UTF-8
1266 encoded form. C<sv_utf8_downgrade(sv)> does the opposite, if
1267 possible. C<sv_utf8_encode(sv)> is like sv_utf8_upgrade except that
1268 it does not set the C<UTF8> flag. C<sv_utf8_decode()> does the
1269 opposite of C<sv_utf8_encode()>. Note that none of these are to be
1270 used as general-purpose encoding or decoding interfaces: C<use Encode>
1271 for that. C<sv_utf8_upgrade()> is affected by the encoding pragma
1272 but C<sv_utf8_downgrade()> is not (since the encoding pragma is
1273 designed to be a one-way street).
1277 C<is_utf8_char(s)> returns true if the pointer points to a valid UTF-8
1282 C<is_utf8_string(buf, len)> returns true if C<len> bytes of the buffer
1287 C<UTF8SKIP(buf)> will return the number of bytes in the UTF-8 encoded
1288 character in the buffer. C<UNISKIP(chr)> will return the number of bytes
1289 required to UTF-8-encode the Unicode character code point. C<UTF8SKIP()>
1290 is useful for example for iterating over the characters of a UTF-8
1291 encoded buffer; C<UNISKIP()> is useful, for example, in computing
1292 the size required for a UTF-8 encoded buffer.
1296 C<utf8_distance(a, b)> will tell the distance in characters between the
1297 two pointers pointing to the same UTF-8 encoded buffer.
1301 C<utf8_hop(s, off)> will return a pointer to an UTF-8 encoded buffer
1302 that is C<off> (positive or negative) Unicode characters displaced
1303 from the UTF-8 buffer C<s>. Be careful not to overstep the buffer:
1304 C<utf8_hop()> will merrily run off the end or the beginning of the
1305 buffer if told to do so.
1309 C<pv_uni_display(dsv, spv, len, pvlim, flags)> and
1310 C<sv_uni_display(dsv, ssv, pvlim, flags)> are useful for debugging the
1311 output of Unicode strings and scalars. By default they are useful
1312 only for debugging--they display B<all> characters as hexadecimal code
1313 points--but with the flags C<UNI_DISPLAY_ISPRINT>,
1314 C<UNI_DISPLAY_BACKSLASH>, and C<UNI_DISPLAY_QQ> you can make the
1315 output more readable.
1319 C<ibcmp_utf8(s1, pe1, u1, l1, u1, s2, pe2, l2, u2)> can be used to
1320 compare two strings case-insensitively in Unicode. For case-sensitive
1321 comparisons you can just use C<memEQ()> and C<memNE()> as usual.
1325 For more information, see L<perlapi>, and F<utf8.c> and F<utf8.h>
1326 in the Perl source code distribution.
1330 =head2 Interaction with Locales
1332 Use of locales with Unicode data may lead to odd results. Currently,
1333 Perl attempts to attach 8-bit locale info to characters in the range
1334 0..255, but this technique is demonstrably incorrect for locales that
1335 use characters above that range when mapped into Unicode. Perl's
1336 Unicode support will also tend to run slower. Use of locales with
1337 Unicode is discouraged.
1339 =head2 Interaction with Extensions
1341 When Perl exchanges data with an extension, the extension should be
1342 able to understand the UTF-8 flag and act accordingly. If the
1343 extension doesn't know about the flag, it's likely that the extension
1344 will return incorrectly-flagged data.
1346 So if you're working with Unicode data, consult the documentation of
1347 every module you're using if there are any issues with Unicode data
1348 exchange. If the documentation does not talk about Unicode at all,
1349 suspect the worst and probably look at the source to learn how the
1350 module is implemented. Modules written completely in Perl shouldn't
1351 cause problems. Modules that directly or indirectly access code written
1352 in other programming languages are at risk.
1354 For affected functions, the simple strategy to avoid data corruption is
1355 to always make the encoding of the exchanged data explicit. Choose an
1356 encoding that you know the extension can handle. Convert arguments passed
1357 to the extensions to that encoding and convert results back from that
1358 encoding. Write wrapper functions that do the conversions for you, so
1359 you can later change the functions when the extension catches up.
1361 To provide an example, let's say the popular Foo::Bar::escape_html
1362 function doesn't deal with Unicode data yet. The wrapper function
1363 would convert the argument to raw UTF-8 and convert the result back to
1364 Perl's internal representation like so:
1366 sub my_escape_html ($) {
1368 return unless defined $what;
1369 Encode::decode_utf8(Foo::Bar::escape_html(Encode::encode_utf8($what)));
1372 Sometimes, when the extension does not convert data but just stores
1373 and retrieves them, you will be in a position to use the otherwise
1374 dangerous Encode::_utf8_on() function. Let's say the popular
1375 C<Foo::Bar> extension, written in C, provides a C<param> method that
1376 lets you store and retrieve data according to these prototypes:
1378 $self->param($name, $value); # set a scalar
1379 $value = $self->param($name); # retrieve a scalar
1381 If it does not yet provide support for any encoding, one could write a
1382 derived class with such a C<param> method:
1385 my($self,$name,$value) = @_;
1386 utf8::upgrade($name); # make sure it is UTF-8 encoded
1388 utf8::upgrade($value); # make sure it is UTF-8 encoded
1389 return $self->SUPER::param($name,$value);
1391 my $ret = $self->SUPER::param($name);
1392 Encode::_utf8_on($ret); # we know, it is UTF-8 encoded
1397 Some extensions provide filters on data entry/exit points, such as
1398 DB_File::filter_store_key and family. Look out for such filters in
1399 the documentation of your extensions, they can make the transition to
1400 Unicode data much easier.
1404 Some functions are slower when working on UTF-8 encoded strings than
1405 on byte encoded strings. All functions that need to hop over
1406 characters such as length(), substr() or index(), or matching regular
1407 expressions can work B<much> faster when the underlying data are
1410 In Perl 5.8.0 the slowness was often quite spectacular; in Perl 5.8.1
1411 a caching scheme was introduced which will hopefully make the slowness
1412 somewhat less spectacular, at least for some operations. In general,
1413 operations with UTF-8 encoded strings are still slower. As an example,
1414 the Unicode properties (character classes) like C<\p{Nd}> are known to
1415 be quite a bit slower (5-20 times) than their simpler counterparts
1416 like C<\d> (then again, there 268 Unicode characters matching C<Nd>
1417 compared with the 10 ASCII characters matching C<d>).
1419 =head2 Porting code from perl-5.6.X
1421 Perl 5.8 has a different Unicode model from 5.6. In 5.6 the programmer
1422 was required to use the C<utf8> pragma to declare that a given scope
1423 expected to deal with Unicode data and had to make sure that only
1424 Unicode data were reaching that scope. If you have code that is
1425 working with 5.6, you will need some of the following adjustments to
1426 your code. The examples are written such that the code will continue
1427 to work under 5.6, so you should be safe to try them out.
1433 A filehandle that should read or write UTF-8
1436 binmode $fh, ":utf8";
1441 A scalar that is going to be passed to some extension
1443 Be it Compress::Zlib, Apache::Request or any extension that has no
1444 mention of Unicode in the manpage, you need to make sure that the
1445 UTF-8 flag is stripped off. Note that at the time of this writing
1446 (October 2002) the mentioned modules are not UTF-8-aware. Please
1447 check the documentation to verify if this is still true.
1451 $val = Encode::encode_utf8($val); # make octets
1456 A scalar we got back from an extension
1458 If you believe the scalar comes back as UTF-8, you will most likely
1459 want the UTF-8 flag restored:
1463 $val = Encode::decode_utf8($val);
1468 Same thing, if you are really sure it is UTF-8
1472 Encode::_utf8_on($val);
1477 A wrapper for fetchrow_array and fetchrow_hashref
1479 When the database contains only UTF-8, a wrapper function or method is
1480 a convenient way to replace all your fetchrow_array and
1481 fetchrow_hashref calls. A wrapper function will also make it easier to
1482 adapt to future enhancements in your database driver. Note that at the
1483 time of this writing (October 2002), the DBI has no standardized way
1484 to deal with UTF-8 data. Please check the documentation to verify if
1488 my($self, $sth, $what) = @_; # $what is one of fetchrow_{array,hashref}
1494 my @arr = $sth->$what;
1496 defined && /[^\000-\177]/ && Encode::_utf8_on($_);
1500 my $ret = $sth->$what;
1502 for my $k (keys %$ret) {
1503 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret->{$k};
1507 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret;
1517 A large scalar that you know can only contain ASCII
1519 Scalars that contain only ASCII and are marked as UTF-8 are sometimes
1520 a drag to your program. If you recognize such a situation, just remove
1523 utf8::downgrade($val) if $] > 5.007;
1529 L<perluniintro>, L<encoding>, L<Encode>, L<open>, L<utf8>, L<bytes>,
1530 L<perlretut>, L<perlvar/"${^UNICODE}">