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 C<use encoding> needed to upgrade non-Latin-1 byte strings
47 By default, there is a fundamental asymmetry in Perl's unicode model:
48 implicit upgrading from byte strings to Unicode strings assumes that
49 they were encoded in I<ISO 8859-1 (Latin-1)>, but Unicode strings are
50 downgraded with UTF-8 encoding. This happens because the first 256
51 codepoints in Unicode happens to agree with Latin-1.
53 If you wish to interpret byte strings as UTF-8 instead, use the
58 See L</"Byte and Character Semantics"> for more details.
62 =head2 Byte and Character Semantics
64 Beginning with version 5.6, Perl uses logically-wide characters to
65 represent strings internally.
67 In future, Perl-level operations will be expected to work with
68 characters rather than bytes.
70 However, as an interim compatibility measure, Perl aims to
71 provide a safe migration path from byte semantics to character
72 semantics for programs. For operations where Perl can unambiguously
73 decide that the input data are characters, Perl switches to
74 character semantics. For operations where this determination cannot
75 be made without additional information from the user, Perl decides in
76 favor of compatibility and chooses to use byte semantics.
78 This behavior preserves compatibility with earlier versions of Perl,
79 which allowed byte semantics in Perl operations only if
80 none of the program's inputs were marked as being as source of Unicode
81 character data. Such data may come from filehandles, from calls to
82 external programs, from information provided by the system (such as %ENV),
83 or from literals and constants in the source text.
85 The C<bytes> pragma will always, regardless of platform, force byte
86 semantics in a particular lexical scope. See L<bytes>.
88 The C<utf8> pragma is primarily a compatibility device that enables
89 recognition of UTF-(8|EBCDIC) in literals encountered by the parser.
90 Note that this pragma is only required while Perl defaults to byte
91 semantics; when character semantics become the default, this pragma
92 may become a no-op. See L<utf8>.
94 Unless explicitly stated, Perl operators use character semantics
95 for Unicode data and byte semantics for non-Unicode data.
96 The decision to use character semantics is made transparently. If
97 input data comes from a Unicode source--for example, if a character
98 encoding layer is added to a filehandle or a literal Unicode
99 string constant appears in a program--character semantics apply.
100 Otherwise, byte semantics are in effect. The C<bytes> pragma should
101 be used to force byte semantics on Unicode data.
103 If strings operating under byte semantics and strings with Unicode
104 character data are concatenated, the new string will be created by
105 decoding the byte strings as I<ISO 8859-1 (Latin-1)>, even if the
106 old Unicode string used EBCDIC. This translation is done without
107 regard to the system's native 8-bit encoding. To change this for
108 systems with non-Latin-1 and non-EBCDIC native encodings, use the
109 C<encoding> pragma. See L<encoding>.
111 Under character semantics, many operations that formerly operated on
112 bytes now operate on characters. A character in Perl is
113 logically just a number ranging from 0 to 2**31 or so. Larger
114 characters may encode into longer sequences of bytes internally, but
115 this internal detail is mostly hidden for Perl code.
116 See L<perluniintro> for more.
118 =head2 Effects of Character Semantics
120 Character semantics have the following effects:
126 Strings--including hash keys--and regular expression patterns may
127 contain characters that have an ordinal value larger than 255.
129 If you use a Unicode editor to edit your program, Unicode characters
130 may occur directly within the literal strings in one of the various
131 Unicode encodings (UTF-8, UTF-EBCDIC, UCS-2, etc.), but will be recognized
132 as such and converted to Perl's internal representation only if the
133 appropriate L<encoding> is specified.
135 Unicode characters can also be added to a string by using the
136 C<\x{...}> notation. The Unicode code for the desired character, in
137 hexadecimal, should be placed in the braces. For instance, a smiley
138 face is C<\x{263A}>. This encoding scheme only works for characters
139 with a code of 0x100 or above.
143 use charnames ':full';
145 you can use the C<\N{...}> notation and put the official Unicode
146 character name within the braces, such as C<\N{WHITE SMILING FACE}>.
151 If an appropriate L<encoding> is specified, identifiers within the
152 Perl script may contain Unicode alphanumeric characters, including
153 ideographs. Perl does not currently attempt to canonicalize variable
158 Regular expressions match characters instead of bytes. "." matches
159 a character instead of a byte. The C<\C> pattern is provided to force
160 a match a single byte--a C<char> in C, hence C<\C>.
164 Character classes in regular expressions match characters instead of
165 bytes and match against the character properties specified in the
166 Unicode properties database. C<\w> can be used to match a Japanese
167 ideograph, for instance.
169 (However, and as a limitation of the current implementation, using
170 C<\w> or C<\W> I<inside> a C<[...]> character class will still match
171 with byte semantics.)
175 Named Unicode properties, scripts, and block ranges may be used like
176 character classes via the C<\p{}> "matches property" construct and
177 the C<\P{}> negation, "doesn't match property".
179 For instance, C<\p{Lu}> matches any character with the Unicode "Lu"
180 (Letter, uppercase) property, while C<\p{M}> matches any character
181 with an "M" (mark--accents and such) property. Brackets are not
182 required for single letter properties, so C<\p{M}> is equivalent to
183 C<\pM>. Many predefined properties are available, such as
184 C<\p{Mirrored}> and C<\p{Tibetan}>.
186 The official Unicode script and block names have spaces and dashes as
187 separators, but for convenience you can use dashes, spaces, or
188 underbars, and case is unimportant. It is recommended, however, that
189 for consistency you use the following naming: the official Unicode
190 script, property, or block name (see below for the additional rules
191 that apply to block names) with whitespace and dashes removed, and the
192 words "uppercase-first-lowercase-rest". C<Latin-1 Supplement> thus
193 becomes C<Latin1Supplement>.
195 You can also use negation in both C<\p{}> and C<\P{}> by introducing a caret
196 (^) between the first brace and the property name: C<\p{^Tamil}> is
197 equal to C<\P{Tamil}>.
199 B<NOTE: the properties, scripts, and blocks listed here are as of
200 Unicode 3.2.0, March 2002, or Perl 5.8.0, July 2002. Unicode 4.0.0
201 came out in April 2003, and Perl 5.8.1 in September 2003.>
203 Here are the basic Unicode General Category properties, followed by their
204 long form. You can use either; C<\p{Lu}> and C<\p{UppercaseLetter}>,
205 for instance, are identical.
227 Pc ConnectorPunctuation
231 Pi InitialPunctuation
232 (may behave like Ps or Pe depending on usage)
234 (may behave like Ps or Pe depending on usage)
246 Zp ParagraphSeparator
251 Cs Surrogate (not usable)
255 Single-letter properties match all characters in any of the
256 two-letter sub-properties starting with the same letter.
257 C<L&> is a special case, which is an alias for C<Ll>, C<Lu>, and C<Lt>.
259 Because Perl hides the need for the user to understand the internal
260 representation of Unicode characters, there is no need to implement
261 the somewhat messy concept of surrogates. C<Cs> is therefore not
264 Because scripts differ in their directionality--Hebrew is
265 written right to left, for example--Unicode supplies these properties:
270 BidiLRE Left-to-Right Embedding
271 BidiLRO Left-to-Right Override
273 BidiAL Right-to-Left Arabic
274 BidiRLE Right-to-Left Embedding
275 BidiRLO Right-to-Left Override
276 BidiPDF Pop Directional Format
277 BidiEN European Number
278 BidiES European Number Separator
279 BidiET European Number Terminator
281 BidiCS Common Number Separator
282 BidiNSM Non-Spacing Mark
283 BidiBN Boundary Neutral
284 BidiB Paragraph Separator
285 BidiS Segment Separator
287 BidiON Other Neutrals
289 For example, C<\p{BidiR}> matches characters that are normally
290 written right to left.
296 The script names which can be used by C<\p{...}> and C<\P{...}>,
297 such as in C<\p{Latin}> or C<\p{Cyrillic}>, are as follows:
344 Extended property classes can supplement the basic
345 properties, defined by the F<PropList> Unicode database:
360 LogicalOrderException
361 NoncharacterCodePoint
363 OtherDefaultIgnorableCodePoint
375 and there are further derived properties:
377 Alphabetic Lu + Ll + Lt + Lm + Lo + OtherAlphabetic
378 Lowercase Ll + OtherLowercase
379 Uppercase Lu + OtherUppercase
382 ID_Start Lu + Ll + Lt + Lm + Lo + Nl
383 ID_Continue ID_Start + Mn + Mc + Nd + Pc
386 Assigned Any non-Cn character (i.e. synonym for \P{Cn})
387 Unassigned Synonym for \p{Cn}
388 Common Any character (or unassigned code point)
389 not explicitly assigned to a script
391 For backward compatibility (with Perl 5.6), all properties mentioned
392 so far may have C<Is> prepended to their name, so C<\P{IsLu}>, for
393 example, is equal to C<\P{Lu}>.
397 In addition to B<scripts>, Unicode also defines B<blocks> of
398 characters. The difference between scripts and blocks is that the
399 concept of scripts is closer to natural languages, while the concept
400 of blocks is more of an artificial grouping based on groups of 256
401 Unicode characters. For example, the C<Latin> script contains letters
402 from many blocks but does not contain all the characters from those
403 blocks. It does not, for example, contain digits, because digits are
404 shared across many scripts. Digits and similar groups, like
405 punctuation, are in a category called C<Common>.
407 For more about scripts, see the UTR #24:
409 http://www.unicode.org/unicode/reports/tr24/
411 For more about blocks, see:
413 http://www.unicode.org/Public/UNIDATA/Blocks.txt
415 Block names are given with the C<In> prefix. For example, the
416 Katakana block is referenced via C<\p{InKatakana}>. The C<In>
417 prefix may be omitted if there is no naming conflict with a script
418 or any other property, but it is recommended that C<In> always be used
419 for block tests to avoid confusion.
421 These block names are supported:
423 InAlphabeticPresentationForms
425 InArabicPresentationFormsA
426 InArabicPresentationFormsB
437 InByzantineMusicalSymbols
439 InCJKCompatibilityForms
440 InCJKCompatibilityIdeographs
441 InCJKCompatibilityIdeographsSupplement
442 InCJKRadicalsSupplement
443 InCJKSymbolsAndPunctuation
444 InCJKUnifiedIdeographs
445 InCJKUnifiedIdeographsExtensionA
446 InCJKUnifiedIdeographsExtensionB
448 InCombiningDiacriticalMarks
449 InCombiningDiacriticalMarksforSymbols
454 InCyrillicSupplementary
458 InEnclosedAlphanumerics
459 InEnclosedCJKLettersAndMonths
469 InHalfwidthAndFullwidthForms
470 InHangulCompatibilityJamo
475 InHighPrivateUseSurrogates
479 InIdeographicDescriptionCharacters
484 InKatakanaPhoneticExtensions
489 InLatinExtendedAdditional
494 InMathematicalAlphanumericSymbols
495 InMathematicalOperators
496 InMiscellaneousMathematicalSymbolsA
497 InMiscellaneousMathematicalSymbolsB
498 InMiscellaneousSymbols
499 InMiscellaneousTechnical
506 InOpticalCharacterRecognition
512 InSpacingModifierLetters
514 InSuperscriptsAndSubscripts
515 InSupplementalArrowsA
516 InSupplementalArrowsB
517 InSupplementalMathematicalOperators
518 InSupplementaryPrivateUseAreaA
519 InSupplementaryPrivateUseAreaB
529 InUnifiedCanadianAboriginalSyllabics
538 The special pattern C<\X> matches any extended Unicode
539 sequence--"a combining character sequence" in Standardese--where the
540 first character is a base character and subsequent characters are mark
541 characters that apply to the base character. C<\X> is equivalent to
546 The C<tr///> operator translates characters instead of bytes. Note
547 that the C<tr///CU> functionality has been removed. For similar
548 functionality see pack('U0', ...) and pack('C0', ...).
552 Case translation operators use the Unicode case translation tables
553 when character input is provided. Note that C<uc()>, or C<\U> in
554 interpolated strings, translates to uppercase, while C<ucfirst>,
555 or C<\u> in interpolated strings, translates to titlecase in languages
556 that make the distinction.
560 Most operators that deal with positions or lengths in a string will
561 automatically switch to using character positions, including
562 C<chop()>, C<chomp()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
563 C<sprintf()>, C<write()>, and C<length()>. Operators that
564 specifically do not switch include C<vec()>, C<pack()>, and
565 C<unpack()>. Operators that really don't care include
566 operators that treats strings as a bucket of bits such as C<sort()>,
567 and operators dealing with filenames.
571 The C<pack()>/C<unpack()> letters C<c> and C<C> do I<not> change,
572 since they are often used for byte-oriented formats. Again, think
573 C<char> in the C language.
575 There is a new C<U> specifier that converts between Unicode characters
580 The C<chr()> and C<ord()> functions work on characters, similar to
581 C<pack("U")> and C<unpack("U")>, I<not> C<pack("C")> and
582 C<unpack("C")>. C<pack("C")> and C<unpack("C")> are methods for
583 emulating byte-oriented C<chr()> and C<ord()> on Unicode strings.
584 While these methods reveal the internal encoding of Unicode strings,
585 that is not something one normally needs to care about at all.
589 The bit string operators, C<& | ^ ~>, can operate on character data.
590 However, for backward compatibility, such as when using bit string
591 operations when characters are all less than 256 in ordinal value, one
592 should not use C<~> (the bit complement) with characters of both
593 values less than 256 and values greater than 256. Most importantly,
594 DeMorgan's laws (C<~($x|$y) eq ~$x&~$y> and C<~($x&$y) eq ~$x|~$y>)
595 will not hold. The reason for this mathematical I<faux pas> is that
596 the complement cannot return B<both> the 8-bit (byte-wide) bit
597 complement B<and> the full character-wide bit complement.
601 lc(), uc(), lcfirst(), and ucfirst() work for the following cases:
607 the case mapping is from a single Unicode character to another
608 single Unicode character, or
612 the case mapping is from a single Unicode character to more
613 than one Unicode character.
617 Things to do with locales (Lithuanian, Turkish, Azeri) do B<not> work
618 since Perl does not understand the concept of Unicode locales.
620 See the Unicode Technical Report #21, Case Mappings, for more details.
628 And finally, C<scalar reverse()> reverses by character rather than by byte.
632 =head2 User-Defined Character Properties
634 You can define your own character properties by defining subroutines
635 whose names begin with "In" or "Is". The subroutines can be defined in
636 any package. The user-defined properties can be used in the regular
637 expression C<\p> and C<\P> constructs; if you are using a user-defined
638 property from a package other than the one you are in, you must specify
639 its package in the C<\p> or C<\P> construct.
641 # assuming property IsForeign defined in Lang::
642 package main; # property package name required
643 if ($txt =~ /\p{Lang::IsForeign}+/) { ... }
645 package Lang; # property package name not required
646 if ($txt =~ /\p{IsForeign}+/) { ... }
649 Note that the effect is compile-time and immutable once defined.
651 The subroutines must return a specially-formatted string, with one
652 or more newline-separated lines. Each line must be one of the following:
658 Two hexadecimal numbers separated by horizontal whitespace (space or
659 tabular characters) denoting a range of Unicode code points to include.
663 Something to include, prefixed by "+": a built-in character
664 property (prefixed by "utf8::") or a user-defined character property,
665 to represent all the characters in that property; two hexadecimal code
666 points for a range; or a single hexadecimal code point.
670 Something to exclude, prefixed by "-": an existing character
671 property (prefixed by "utf8::") or a user-defined character property,
672 to represent all the characters in that property; two hexadecimal code
673 points for a range; or a single hexadecimal code point.
677 Something to negate, prefixed "!": an existing character
678 property (prefixed by "utf8::") or a user-defined character property,
679 to represent all the characters in that property; two hexadecimal code
680 points for a range; or a single hexadecimal code point.
684 Something to intersect with, prefixed by "&": an existing character
685 property (prefixed by "utf8::") or a user-defined character property,
686 for all the characters except the characters in the property; two
687 hexadecimal code points for a range; or a single hexadecimal code point.
691 For example, to define a property that covers both the Japanese
692 syllabaries (hiragana and katakana), you can define
701 Imagine that the here-doc end marker is at the beginning of the line.
702 Now you can use C<\p{InKana}> and C<\P{InKana}>.
704 You could also have used the existing block property names:
713 Suppose you wanted to match only the allocated characters,
714 not the raw block ranges: in other words, you want to remove
725 The negation is useful for defining (surprise!) negated classes.
735 Intersection is useful for getting the common characters matched by
736 two (or more) classes.
745 It's important to remember not to use "&" for the first set -- that
746 would be intersecting with nothing (resulting in an empty set).
748 You can also define your own mappings to be used in the lc(),
749 lcfirst(), uc(), and ucfirst() (or their string-inlined versions).
750 The principle is the same: define subroutines in the C<main> package
751 with names like C<ToLower> (for lc() and lcfirst()), C<ToTitle> (for
752 the first character in ucfirst()), and C<ToUpper> (for uc(), and the
753 rest of the characters in ucfirst()).
755 The string returned by the subroutines needs now to be three
756 hexadecimal numbers separated by tabulators: start of the source
757 range, end of the source range, and start of the destination range.
766 defines an uc() mapping that causes only the characters "a", "b", and
767 "c" to be mapped to "A", "B", "C", all other characters will remain
770 If there is no source range to speak of, that is, the mapping is from
771 a single character to another single character, leave the end of the
772 source range empty, but the two tabulator characters are still needed.
781 defines a lc() mapping that causes only "A" to be mapped to "a", all
782 other characters will remain unchanged.
784 (For serious hackers only) If you want to introspect the default
785 mappings, you can find the data in the directory
786 C<$Config{privlib}>/F<unicore/To/>. The mapping data is returned as
787 the here-document, and the C<utf8::ToSpecFoo> are special exception
788 mappings derived from <$Config{privlib}>/F<unicore/SpecialCasing.txt>.
789 The C<Digit> and C<Fold> mappings that one can see in the directory
790 are not directly user-accessible, one can use either the
791 C<Unicode::UCD> module, or just match case-insensitively (that's when
792 the C<Fold> mapping is used).
794 A final note on the user-defined property tests and mappings: they
795 will be used only if the scalar has been marked as having Unicode
796 characters. Old byte-style strings will not be affected.
798 =head2 Character Encodings for Input and Output
802 =head2 Unicode Regular Expression Support Level
804 The following list of Unicode support for regular expressions describes
805 all the features currently supported. The references to "Level N"
806 and the section numbers refer to the Unicode Technical Report 18,
807 "Unicode Regular Expression Guidelines", version 6 (Unicode 3.2.0,
814 Level 1 - Basic Unicode Support
816 2.1 Hex Notation - done [1]
817 Named Notation - done [2]
818 2.2 Categories - done [3][4]
819 2.3 Subtraction - MISSING [5][6]
820 2.4 Simple Word Boundaries - done [7]
821 2.5 Simple Loose Matches - done [8]
822 2.6 End of Line - MISSING [9][10]
826 [ 3] . \p{...} \P{...}
827 [ 4] now scripts (see UTR#24 Script Names) in addition to blocks
829 [ 6] can use regular expression look-ahead [a]
830 or user-defined character properties [b] to emulate subtraction
831 [ 7] include Letters in word characters
832 [ 8] note that Perl does Full case-folding in matching, not Simple:
833 for example U+1F88 is equivalent with U+1F00 U+03B9,
834 not with 1F80. This difference matters for certain Greek
835 capital letters with certain modifiers: the Full case-folding
836 decomposes the letter, while the Simple case-folding would map
837 it to a single character.
838 [ 9] see UTR #13 Unicode Newline Guidelines
839 [10] should do ^ and $ also on \x{85}, \x{2028} and \x{2029}
840 (should also affect <>, $., and script line numbers)
841 (the \x{85}, \x{2028} and \x{2029} do match \s)
843 [a] You can mimic class subtraction using lookahead.
844 For example, what UTR #18 might write as
846 [{Greek}-[{UNASSIGNED}]]
848 in Perl can be written as:
850 (?!\p{Unassigned})\p{InGreekAndCoptic}
851 (?=\p{Assigned})\p{InGreekAndCoptic}
853 But in this particular example, you probably really want
857 which will match assigned characters known to be part of the Greek script.
859 Also see the Unicode::Regex::Set module, it does implement the full
860 UTR #18 grouping, intersection, union, and removal (subtraction) syntax.
862 [b] See L</"User-Defined Character Properties">.
866 Level 2 - Extended Unicode Support
868 3.1 Surrogates - MISSING [11]
869 3.2 Canonical Equivalents - MISSING [12][13]
870 3.3 Locale-Independent Graphemes - MISSING [14]
871 3.4 Locale-Independent Words - MISSING [15]
872 3.5 Locale-Independent Loose Matches - MISSING [16]
874 [11] Surrogates are solely a UTF-16 concept and Perl's internal
875 representation is UTF-8. The Encode module does UTF-16, though.
876 [12] see UTR#15 Unicode Normalization
877 [13] have Unicode::Normalize but not integrated to regexes
878 [14] have \X but at this level . should equal that
879 [15] need three classes, not just \w and \W
880 [16] see UTR#21 Case Mappings
884 Level 3 - Locale-Sensitive Support
886 4.1 Locale-Dependent Categories - MISSING
887 4.2 Locale-Dependent Graphemes - MISSING [16][17]
888 4.3 Locale-Dependent Words - MISSING
889 4.4 Locale-Dependent Loose Matches - MISSING
890 4.5 Locale-Dependent Ranges - MISSING
892 [16] see UTR#10 Unicode Collation Algorithms
893 [17] have Unicode::Collate but not integrated to regexes
897 =head2 Unicode Encodings
899 Unicode characters are assigned to I<code points>, which are abstract
900 numbers. To use these numbers, various encodings are needed.
908 UTF-8 is a variable-length (1 to 6 bytes, current character allocations
909 require 4 bytes), byte-order independent encoding. For ASCII (and we
910 really do mean 7-bit ASCII, not another 8-bit encoding), UTF-8 is
913 The following table is from Unicode 3.2.
915 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
917 U+0000..U+007F 00..7F
918 U+0080..U+07FF C2..DF 80..BF
919 U+0800..U+0FFF E0 A0..BF 80..BF
920 U+1000..U+CFFF E1..EC 80..BF 80..BF
921 U+D000..U+D7FF ED 80..9F 80..BF
922 U+D800..U+DFFF ******* ill-formed *******
923 U+E000..U+FFFF EE..EF 80..BF 80..BF
924 U+10000..U+3FFFF F0 90..BF 80..BF 80..BF
925 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
926 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
928 Note the C<A0..BF> in C<U+0800..U+0FFF>, the C<80..9F> in
929 C<U+D000...U+D7FF>, the C<90..B>F in C<U+10000..U+3FFFF>, and the
930 C<80...8F> in C<U+100000..U+10FFFF>. The "gaps" are caused by legal
931 UTF-8 avoiding non-shortest encodings: it is technically possible to
932 UTF-8-encode a single code point in different ways, but that is
933 explicitly forbidden, and the shortest possible encoding should always
934 be used. So that's what Perl does.
936 Another way to look at it is via bits:
938 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
941 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
942 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
943 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
945 As you can see, the continuation bytes all begin with C<10>, and the
946 leading bits of the start byte tell how many bytes the are in the
953 Like UTF-8 but EBCDIC-safe, in the way that UTF-8 is ASCII-safe.
957 UTF-16, UTF-16BE, UTF-16LE, Surrogates, and BOMs (Byte Order Marks)
959 The followings items are mostly for reference and general Unicode
960 knowledge, Perl doesn't use these constructs internally.
962 UTF-16 is a 2 or 4 byte encoding. The Unicode code points
963 C<U+0000..U+FFFF> are stored in a single 16-bit unit, and the code
964 points C<U+10000..U+10FFFF> in two 16-bit units. The latter case is
965 using I<surrogates>, the first 16-bit unit being the I<high
966 surrogate>, and the second being the I<low surrogate>.
968 Surrogates are code points set aside to encode the C<U+10000..U+10FFFF>
969 range of Unicode code points in pairs of 16-bit units. The I<high
970 surrogates> are the range C<U+D800..U+DBFF>, and the I<low surrogates>
971 are the range C<U+DC00..U+DFFF>. The surrogate encoding is
973 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
974 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
978 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
980 If you try to generate surrogates (for example by using chr()), you
981 will get a warning if warnings are turned on, because those code
982 points are not valid for a Unicode character.
984 Because of the 16-bitness, UTF-16 is byte-order dependent. UTF-16
985 itself can be used for in-memory computations, but if storage or
986 transfer is required either UTF-16BE (big-endian) or UTF-16LE
987 (little-endian) encodings must be chosen.
989 This introduces another problem: what if you just know that your data
990 is UTF-16, but you don't know which endianness? Byte Order Marks, or
991 BOMs, are a solution to this. A special character has been reserved
992 in Unicode to function as a byte order marker: the character with the
993 code point C<U+FEFF> is the BOM.
995 The trick is that if you read a BOM, you will know the byte order,
996 since if it was written on a big-endian platform, you will read the
997 bytes C<0xFE 0xFF>, but if it was written on a little-endian platform,
998 you will read the bytes C<0xFF 0xFE>. (And if the originating platform
999 was writing in UTF-8, you will read the bytes C<0xEF 0xBB 0xBF>.)
1001 The way this trick works is that the character with the code point
1002 C<U+FFFE> is guaranteed not to be a valid Unicode character, so the
1003 sequence of bytes C<0xFF 0xFE> is unambiguously "BOM, represented in
1004 little-endian format" and cannot be C<U+FFFE>, represented in big-endian
1009 UTF-32, UTF-32BE, UTF-32LE
1011 The UTF-32 family is pretty much like the UTF-16 family, expect that
1012 the units are 32-bit, and therefore the surrogate scheme is not
1013 needed. The BOM signatures will be C<0x00 0x00 0xFE 0xFF> for BE and
1014 C<0xFF 0xFE 0x00 0x00> for LE.
1020 Encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
1021 encoding. Unlike UTF-16, UCS-2 is not extensible beyond C<U+FFFF>,
1022 because it does not use surrogates. UCS-4 is a 32-bit encoding,
1023 functionally identical to UTF-32.
1029 A seven-bit safe (non-eight-bit) encoding, which is useful if the
1030 transport or storage is not eight-bit safe. Defined by RFC 2152.
1034 =head2 Security Implications of Unicode
1042 Unfortunately, the specification of UTF-8 leaves some room for
1043 interpretation of how many bytes of encoded output one should generate
1044 from one input Unicode character. Strictly speaking, the shortest
1045 possible sequence of UTF-8 bytes should be generated,
1046 because otherwise there is potential for an input buffer overflow at
1047 the receiving end of a UTF-8 connection. Perl always generates the
1048 shortest length UTF-8, and with warnings on Perl will warn about
1049 non-shortest length UTF-8 along with other malformations, such as the
1050 surrogates, which are not real Unicode code points.
1054 Regular expressions behave slightly differently between byte data and
1055 character (Unicode) data. For example, the "word character" character
1056 class C<\w> will work differently depending on if data is eight-bit bytes
1059 In the first case, the set of C<\w> characters is either small--the
1060 default set of alphabetic characters, digits, and the "_"--or, if you
1061 are using a locale (see L<perllocale>), the C<\w> might contain a few
1062 more letters according to your language and country.
1064 In the second case, the C<\w> set of characters is much, much larger.
1065 Most importantly, even in the set of the first 256 characters, it will
1066 probably match different characters: unlike most locales, which are
1067 specific to a language and country pair, Unicode classifies all the
1068 characters that are letters I<somewhere> as C<\w>. For example, your
1069 locale might not think that LATIN SMALL LETTER ETH is a letter (unless
1070 you happen to speak Icelandic), but Unicode does.
1072 As discussed elsewhere, Perl has one foot (two hooves?) planted in
1073 each of two worlds: the old world of bytes and the new world of
1074 characters, upgrading from bytes to characters when necessary.
1075 If your legacy code does not explicitly use Unicode, no automatic
1076 switch-over to characters should happen. Characters shouldn't get
1077 downgraded to bytes, either. It is possible to accidentally mix bytes
1078 and characters, however (see L<perluniintro>), in which case C<\w> in
1079 regular expressions might start behaving differently. Review your
1080 code. Use warnings and the C<strict> pragma.
1084 =head2 Unicode in Perl on EBCDIC
1086 The way Unicode is handled on EBCDIC platforms is still
1087 experimental. On such platforms, references to UTF-8 encoding in this
1088 document and elsewhere should be read as meaning the UTF-EBCDIC
1089 specified in Unicode Technical Report 16, unless ASCII vs. EBCDIC issues
1090 are specifically discussed. There is no C<utfebcdic> pragma or
1091 ":utfebcdic" layer; rather, "utf8" and ":utf8" are reused to mean
1092 the platform's "natural" 8-bit encoding of Unicode. See L<perlebcdic>
1093 for more discussion of the issues.
1097 Usually locale settings and Unicode do not affect each other, but
1098 there are a couple of exceptions:
1104 You can enable automatic UTF-8-ification of your standard file
1105 handles, default C<open()> layer, and C<@ARGV> by using either
1106 the C<-C> command line switch or the C<PERL_UNICODE> environment
1107 variable, see L<perlrun> for the documentation of the C<-C> switch.
1111 Perl tries really hard to work both with Unicode and the old
1112 byte-oriented world. Most often this is nice, but sometimes Perl's
1113 straddling of the proverbial fence causes problems.
1117 =head2 When Unicode Does Not Happen
1119 While Perl does have extensive ways to input and output in Unicode,
1120 and few other 'entry points' like the @ARGV which can be interpreted
1121 as Unicode (UTF-8), there still are many places where Unicode (in some
1122 encoding or another) could be given as arguments or received as
1123 results, or both, but it is not.
1125 The following are such interfaces. For all of these interfaces Perl
1126 currently (as of 5.8.3) simply assumes byte strings both as arguments
1127 and results, or UTF-8 strings if the C<encoding> pragma has been used.
1129 One reason why Perl does not attempt to resolve the role of Unicode in
1130 this cases is that the answers are highly dependent on the operating
1131 system and the file system(s). For example, whether filenames can be
1132 in Unicode, and in exactly what kind of encoding, is not exactly a
1133 portable concept. Similarly for the qx and system: how well will the
1134 'command line interface' (and which of them?) handle Unicode?
1140 chmod, chmod, chown, chroot, exec, link, lstat, mkdir,
1141 rename, rmdir, stat, symlink, truncate, unlink, utime, -X
1153 open, opendir, sysopen
1157 qx (aka the backtick operator), system
1165 =head2 Forcing Unicode in Perl (Or Unforcing Unicode in Perl)
1167 Sometimes (see L</"When Unicode Does Not Happen">) there are
1168 situations where you simply need to force Perl to believe that a byte
1169 string is UTF-8, or vice versa. The low-level calls
1170 utf8::upgrade($bytestring) and utf8::downgrade($utf8string) are
1173 Do not use them without careful thought, though: Perl may easily get
1174 very confused, angry, or even crash, if you suddenly change the 'nature'
1175 of scalar like that. Especially careful you have to be if you use the
1176 utf8::upgrade(): any random byte string is not valid UTF-8.
1178 =head2 Using Unicode in XS
1180 If you want to handle Perl Unicode in XS extensions, you may find the
1181 following C APIs useful. See also L<perlguts/"Unicode Support"> for an
1182 explanation about Unicode at the XS level, and L<perlapi> for the API
1189 C<DO_UTF8(sv)> returns true if the C<UTF8> flag is on and the bytes
1190 pragma is not in effect. C<SvUTF8(sv)> returns true is the C<UTF8>
1191 flag is on; the bytes pragma is ignored. The C<UTF8> flag being on
1192 does B<not> mean that there are any characters of code points greater
1193 than 255 (or 127) in the scalar or that there are even any characters
1194 in the scalar. What the C<UTF8> flag means is that the sequence of
1195 octets in the representation of the scalar is the sequence of UTF-8
1196 encoded code points of the characters of a string. The C<UTF8> flag
1197 being off means that each octet in this representation encodes a
1198 single character with code point 0..255 within the string. Perl's
1199 Unicode model is not to use UTF-8 until it is absolutely necessary.
1203 C<uvuni_to_utf8(buf, chr)> writes a Unicode character code point into
1204 a buffer encoding the code point as UTF-8, and returns a pointer
1205 pointing after the UTF-8 bytes.
1209 C<utf8_to_uvuni(buf, lenp)> reads UTF-8 encoded bytes from a buffer and
1210 returns the Unicode character code point and, optionally, the length of
1211 the UTF-8 byte sequence.
1215 C<utf8_length(start, end)> returns the length of the UTF-8 encoded buffer
1216 in characters. C<sv_len_utf8(sv)> returns the length of the UTF-8 encoded
1221 C<sv_utf8_upgrade(sv)> converts the string of the scalar to its UTF-8
1222 encoded form. C<sv_utf8_downgrade(sv)> does the opposite, if
1223 possible. C<sv_utf8_encode(sv)> is like sv_utf8_upgrade except that
1224 it does not set the C<UTF8> flag. C<sv_utf8_decode()> does the
1225 opposite of C<sv_utf8_encode()>. Note that none of these are to be
1226 used as general-purpose encoding or decoding interfaces: C<use Encode>
1227 for that. C<sv_utf8_upgrade()> is affected by the encoding pragma
1228 but C<sv_utf8_downgrade()> is not (since the encoding pragma is
1229 designed to be a one-way street).
1233 C<is_utf8_char(s)> returns true if the pointer points to a valid UTF-8
1238 C<is_utf8_string(buf, len)> returns true if C<len> bytes of the buffer
1243 C<UTF8SKIP(buf)> will return the number of bytes in the UTF-8 encoded
1244 character in the buffer. C<UNISKIP(chr)> will return the number of bytes
1245 required to UTF-8-encode the Unicode character code point. C<UTF8SKIP()>
1246 is useful for example for iterating over the characters of a UTF-8
1247 encoded buffer; C<UNISKIP()> is useful, for example, in computing
1248 the size required for a UTF-8 encoded buffer.
1252 C<utf8_distance(a, b)> will tell the distance in characters between the
1253 two pointers pointing to the same UTF-8 encoded buffer.
1257 C<utf8_hop(s, off)> will return a pointer to an UTF-8 encoded buffer
1258 that is C<off> (positive or negative) Unicode characters displaced
1259 from the UTF-8 buffer C<s>. Be careful not to overstep the buffer:
1260 C<utf8_hop()> will merrily run off the end or the beginning of the
1261 buffer if told to do so.
1265 C<pv_uni_display(dsv, spv, len, pvlim, flags)> and
1266 C<sv_uni_display(dsv, ssv, pvlim, flags)> are useful for debugging the
1267 output of Unicode strings and scalars. By default they are useful
1268 only for debugging--they display B<all> characters as hexadecimal code
1269 points--but with the flags C<UNI_DISPLAY_ISPRINT>,
1270 C<UNI_DISPLAY_BACKSLASH>, and C<UNI_DISPLAY_QQ> you can make the
1271 output more readable.
1275 C<ibcmp_utf8(s1, pe1, u1, l1, u1, s2, pe2, l2, u2)> can be used to
1276 compare two strings case-insensitively in Unicode. For case-sensitive
1277 comparisons you can just use C<memEQ()> and C<memNE()> as usual.
1281 For more information, see L<perlapi>, and F<utf8.c> and F<utf8.h>
1282 in the Perl source code distribution.
1286 =head2 Interaction with Locales
1288 Use of locales with Unicode data may lead to odd results. Currently,
1289 Perl attempts to attach 8-bit locale info to characters in the range
1290 0..255, but this technique is demonstrably incorrect for locales that
1291 use characters above that range when mapped into Unicode. Perl's
1292 Unicode support will also tend to run slower. Use of locales with
1293 Unicode is discouraged.
1295 =head2 Interaction with Extensions
1297 When Perl exchanges data with an extension, the extension should be
1298 able to understand the UTF-8 flag and act accordingly. If the
1299 extension doesn't know about the flag, it's likely that the extension
1300 will return incorrectly-flagged data.
1302 So if you're working with Unicode data, consult the documentation of
1303 every module you're using if there are any issues with Unicode data
1304 exchange. If the documentation does not talk about Unicode at all,
1305 suspect the worst and probably look at the source to learn how the
1306 module is implemented. Modules written completely in Perl shouldn't
1307 cause problems. Modules that directly or indirectly access code written
1308 in other programming languages are at risk.
1310 For affected functions, the simple strategy to avoid data corruption is
1311 to always make the encoding of the exchanged data explicit. Choose an
1312 encoding that you know the extension can handle. Convert arguments passed
1313 to the extensions to that encoding and convert results back from that
1314 encoding. Write wrapper functions that do the conversions for you, so
1315 you can later change the functions when the extension catches up.
1317 To provide an example, let's say the popular Foo::Bar::escape_html
1318 function doesn't deal with Unicode data yet. The wrapper function
1319 would convert the argument to raw UTF-8 and convert the result back to
1320 Perl's internal representation like so:
1322 sub my_escape_html ($) {
1324 return unless defined $what;
1325 Encode::decode_utf8(Foo::Bar::escape_html(Encode::encode_utf8($what)));
1328 Sometimes, when the extension does not convert data but just stores
1329 and retrieves them, you will be in a position to use the otherwise
1330 dangerous Encode::_utf8_on() function. Let's say the popular
1331 C<Foo::Bar> extension, written in C, provides a C<param> method that
1332 lets you store and retrieve data according to these prototypes:
1334 $self->param($name, $value); # set a scalar
1335 $value = $self->param($name); # retrieve a scalar
1337 If it does not yet provide support for any encoding, one could write a
1338 derived class with such a C<param> method:
1341 my($self,$name,$value) = @_;
1342 utf8::upgrade($name); # make sure it is UTF-8 encoded
1344 utf8::upgrade($value); # make sure it is UTF-8 encoded
1345 return $self->SUPER::param($name,$value);
1347 my $ret = $self->SUPER::param($name);
1348 Encode::_utf8_on($ret); # we know, it is UTF-8 encoded
1353 Some extensions provide filters on data entry/exit points, such as
1354 DB_File::filter_store_key and family. Look out for such filters in
1355 the documentation of your extensions, they can make the transition to
1356 Unicode data much easier.
1360 Some functions are slower when working on UTF-8 encoded strings than
1361 on byte encoded strings. All functions that need to hop over
1362 characters such as length(), substr() or index(), or matching regular
1363 expressions can work B<much> faster when the underlying data are
1366 In Perl 5.8.0 the slowness was often quite spectacular; in Perl 5.8.1
1367 a caching scheme was introduced which will hopefully make the slowness
1368 somewhat less spectacular, at least for some operations. In general,
1369 operations with UTF-8 encoded strings are still slower. As an example,
1370 the Unicode properties (character classes) like C<\p{Nd}> are known to
1371 be quite a bit slower (5-20 times) than their simpler counterparts
1372 like C<\d> (then again, there 268 Unicode characters matching C<Nd>
1373 compared with the 10 ASCII characters matching C<d>).
1375 =head2 Porting code from perl-5.6.X
1377 Perl 5.8 has a different Unicode model from 5.6. In 5.6 the programmer
1378 was required to use the C<utf8> pragma to declare that a given scope
1379 expected to deal with Unicode data and had to make sure that only
1380 Unicode data were reaching that scope. If you have code that is
1381 working with 5.6, you will need some of the following adjustments to
1382 your code. The examples are written such that the code will continue
1383 to work under 5.6, so you should be safe to try them out.
1389 A filehandle that should read or write UTF-8
1392 binmode $fh, ":utf8";
1397 A scalar that is going to be passed to some extension
1399 Be it Compress::Zlib, Apache::Request or any extension that has no
1400 mention of Unicode in the manpage, you need to make sure that the
1401 UTF-8 flag is stripped off. Note that at the time of this writing
1402 (October 2002) the mentioned modules are not UTF-8-aware. Please
1403 check the documentation to verify if this is still true.
1407 $val = Encode::encode_utf8($val); # make octets
1412 A scalar we got back from an extension
1414 If you believe the scalar comes back as UTF-8, you will most likely
1415 want the UTF-8 flag restored:
1419 $val = Encode::decode_utf8($val);
1424 Same thing, if you are really sure it is UTF-8
1428 Encode::_utf8_on($val);
1433 A wrapper for fetchrow_array and fetchrow_hashref
1435 When the database contains only UTF-8, a wrapper function or method is
1436 a convenient way to replace all your fetchrow_array and
1437 fetchrow_hashref calls. A wrapper function will also make it easier to
1438 adapt to future enhancements in your database driver. Note that at the
1439 time of this writing (October 2002), the DBI has no standardized way
1440 to deal with UTF-8 data. Please check the documentation to verify if
1444 my($self, $sth, $what) = @_; # $what is one of fetchrow_{array,hashref}
1450 my @arr = $sth->$what;
1452 defined && /[^\000-\177]/ && Encode::_utf8_on($_);
1456 my $ret = $sth->$what;
1458 for my $k (keys %$ret) {
1459 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret->{$k};
1463 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret;
1473 A large scalar that you know can only contain ASCII
1475 Scalars that contain only ASCII and are marked as UTF-8 are sometimes
1476 a drag to your program. If you recognize such a situation, just remove
1479 utf8::downgrade($val) if $] > 5.007;
1485 L<perluniintro>, L<encoding>, L<Encode>, L<open>, L<utf8>, L<bytes>,
1486 L<perlretut>, L<perlvar/"${^UNICODE}">