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 Disciplines
17 A filehandle can be marked as containing perl's internal Unicode
18 encoding (UTF-8 or UTF-EBCDIC) by opening it with the ":utf8" layer.
19 Other encodings can be converted to perl's encoding on input, or from
20 perl's encoding on output by use of the ":encoding(...)" layer.
23 To mark the Perl source itself as being in 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 switch to the Unicode
30 character scheme when presented with Unicode data, or a traditional
31 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, this pragma must be explicitly used to
36 enable recognition of UTF-8 in the Perl scripts themselves on ASCII
37 based machines, or to recognize UTF-EBCDIC on EBCDIC based machines.
38 B<NOTE: this should be the only place where an explicit C<use utf8>
41 You can also use the C<encoding> pragma to change the default encoding
42 of the data in your script; see L<encoding>.
46 =head2 Byte and Character semantics
48 Beginning with version 5.6, Perl uses logically wide characters to
49 represent strings internally.
51 In future, Perl-level operations can be expected to work with
52 characters rather than bytes, in general.
54 However, as strictly an interim compatibility measure, Perl aims to
55 provide a safe migration path from byte semantics to character
56 semantics for programs. For operations where Perl can unambiguously
57 decide that the input data is characters, Perl now switches to
58 character semantics. For operations where this determination cannot
59 be made without additional information from the user, Perl decides in
60 favor of compatibility, and chooses to use byte semantics.
62 This behavior preserves compatibility with earlier versions of Perl,
63 which allowed byte semantics in Perl operations, but only as long as
64 none of the program's inputs are marked as being as source of Unicode
65 character data. Such data may come from filehandles, from calls to
66 external programs, from information provided by the system (such as %ENV),
67 or from literals and constants in the source text.
69 On Windows platforms, if the C<-C> command line switch is used, (or the
70 ${^WIDE_SYSTEM_CALLS} global flag is set to C<1>), all system calls
71 will use the corresponding wide character APIs. Note that this is
72 currently only implemented on Windows since other platforms lack an
73 API standard on this area.
75 Regardless of the above, the C<bytes> pragma can always be used to
76 force byte semantics in a particular lexical scope. See L<bytes>.
78 The C<utf8> pragma is primarily a compatibility device that enables
79 recognition of UTF-(8|EBCDIC) in literals encountered by the parser.
80 Note that this pragma is only required until a future version of Perl
81 in which character semantics will become the default. This pragma may
82 then become a no-op. See L<utf8>.
84 Unless mentioned otherwise, Perl operators will use character semantics
85 when they are dealing with Unicode data, and byte semantics otherwise.
86 Thus, character semantics for these operations apply transparently; if
87 the input data came from a Unicode source (for example, by adding a
88 character encoding discipline to the filehandle whence it came, or a
89 literal Unicode string constant in the program), character semantics
90 apply; otherwise, byte semantics are in effect. To force byte semantics
91 on Unicode data, the C<bytes> pragma should be used.
93 Notice that if you concatenate strings with byte semantics and strings
94 with Unicode character data, the bytes will by default be upgraded
95 I<as if they were ISO 8859-1 (Latin-1)> (or if in EBCDIC, after a
96 translation to ISO 8859-1). This is done without regard to the
97 system's native 8-bit encoding, so to change this for systems with
98 non-Latin-1 (or non-EBCDIC) native encodings, use the C<encoding>
99 pragma, see L<encoding>.
101 Under character semantics, many operations that formerly operated on
102 bytes change to operating on characters. A character in Perl is
103 logically just a number ranging from 0 to 2**31 or so. Larger
104 characters may encode to longer sequences of bytes internally, but
105 this is just an internal detail which is hidden at the Perl level.
106 See L<perluniintro> for more on this.
108 =head2 Effects of character semantics
110 Character semantics have the following effects:
116 Strings (including hash keys) and regular expression patterns may
117 contain characters that have an ordinal value larger than 255.
119 If you use a Unicode editor to edit your program, Unicode characters
120 may occur directly within the literal strings in one of the various
121 Unicode encodings (UTF-8, UTF-EBCDIC, UCS-2, etc.), but are recognized
122 as such (and converted to Perl's internal representation) only if the
123 appropriate L<encoding> is specified.
125 You can also get Unicode characters into a string by using the C<\x{...}>
126 notation, putting the Unicode code for the desired character, in
127 hexadecimal, into the curlies. For instance, a smiley face is C<\x{263A}>.
128 This works only for characters with a code 0x100 and above.
132 use charnames ':full';
134 you can use the C<\N{...}> notation, putting the official Unicode character
135 name within the curlies. For example, C<\N{WHITE SMILING FACE}>.
136 This works for all characters that have names.
140 If Unicode is used in hash keys, there is a subtle effect on the hashes.
141 The hash becomes "Unicode-sticky" so that keys retrieved from the hash
142 (either by %hash, each %hash, or keys %hash) will be in Unicode, not
143 in bytes, even when the keys were bytes went they "went in". This
144 "stickiness" persists unless the hash is completely emptied, either by
145 using delete() or clearing the with undef() or assigning an empty list
146 to the hash. Most of the time this difference is negligible, but
147 there are few places where it matters: for example the regular
148 expression character classes like C<\w> behave differently for
149 bytes and characters.
153 If an appropriate L<encoding> is specified, identifiers within the
154 Perl script may contain Unicode alphanumeric characters, including
155 ideographs. (You are currently on your own when it comes to using the
156 canonical forms of characters--Perl doesn't (yet) attempt to
157 canonicalize variable names for you.)
161 Regular expressions match characters instead of bytes. For instance,
162 "." matches a character instead of a byte. (However, the C<\C> pattern
163 is provided to force a match a single byte ("C<char>" in C, hence C<\C>).)
167 Character classes in regular expressions match characters instead of
168 bytes, and match against the character properties specified in the
169 Unicode properties database. So C<\w> can be used to match an
170 ideograph, for instance.
174 Named Unicode properties, scripts, and block ranges may be used like
175 character classes via the new C<\p{}> (matches property) and C<\P{}>
176 (doesn't match property) constructs. For instance, C<\p{Lu}> matches any
177 character with the Unicode "Lu" (Letter, uppercase) property, while
178 C<\p{M}> matches any character with a "M" (mark -- accents and such)
179 property. Single letter properties may omit the brackets, so that can be
180 written C<\pM> also. Many predefined properties are available, such
181 as C<\p{Mirrored}> and C<\p{Tibetan}>.
183 The official Unicode script and block names have spaces and dashes as
184 separators, but for convenience you can have dashes, spaces, and underbars
185 at every word division, and you need not care about correct casing. It is
186 recommended, however, that for consistency you use the following naming:
187 the official Unicode script, block, or property name (see below for the
188 additional rules that apply to block names), with whitespace and dashes
189 removed, and the words "uppercase-first-lowercase-rest". That is, "Latin-1
190 Supplement" becomes "Latin1Supplement".
192 You can also negate both C<\p{}> and C<\P{}> by introducing a caret
193 (^) between the first curly and the property name: C<\p{^Tamil}> is
194 equal to C<\P{Tamil}>.
196 Here are the basic Unicode General Category properties, followed by their
197 long form (you can use either, e.g. C<\p{Lu}> and C<\p{LowercaseLetter}>
220 Pc ConnectorPunctuation
224 Pi InitialPunctuation
225 (may behave like Ps or Pe depending on usage)
227 (may behave like Ps or Pe depending on usage)
239 Zp ParagraphSeparator
244 Cs Surrogate (not usable)
248 The single-letter properties match all characters in any of the
249 two-letter sub-properties starting with the same letter.
250 There's also C<L&> which is an alias for C<Ll>, C<Lu>, and C<Lt>.
252 Because Perl hides the need for the user to understand the internal
253 representation of Unicode characters, it has no need to support the
254 somewhat messy concept of surrogates. Therefore, the C<Cs> property is not
257 Because scripts differ in their directionality (for example Hebrew is
258 written right to left), Unicode supplies these properties:
263 BidiLRE Left-to-Right Embedding
264 BidiLRO Left-to-Right Override
266 BidiAL Right-to-Left Arabic
267 BidiRLE Right-to-Left Embedding
268 BidiRLO Right-to-Left Override
269 BidiPDF Pop Directional Format
270 BidiEN European Number
271 BidiES European Number Separator
272 BidiET European Number Terminator
274 BidiCS Common Number Separator
275 BidiNSM Non-Spacing Mark
276 BidiBN Boundary Neutral
277 BidiB Paragraph Separator
278 BidiS Segment Separator
280 BidiON Other Neutrals
282 For example, C<\p{BidiR}> matches all characters that are normally
283 written right to left.
289 The scripts available via C<\p{...}> and C<\P{...}>, for example
290 C<\p{Latin}> or \p{Cyrillic>, are as follows:
333 There are also extended property classes that supplement the basic
334 properties, defined by the F<PropList> Unicode database:
345 NoncharacterCodePoint
353 and further derived properties:
355 Alphabetic Lu + Ll + Lt + Lm + Lo + OtherAlphabetic
356 Lowercase Ll + OtherLowercase
357 Uppercase Lu + OtherUppercase
360 ID_Start Lu + Ll + Lt + Lm + Lo + Nl
361 ID_Continue ID_Start + Mn + Mc + Nd + Pc
364 Assigned Any non-Cn character (i.e. synonym for C<\P{Cn}>)
365 Unassigned Synonym for C<\p{Cn}>
366 Common Any character (or unassigned code point)
367 not explicitly assigned to a script
369 For backward compatability, all properties mentioned so far may have C<Is>
370 prepended to their name (e.g. C<\P{IsLu}> is equal to C<\P{Lu}>).
374 In addition to B<scripts>, Unicode also defines B<blocks> of characters.
375 The difference between scripts and blocks is that the scripts concept is
376 closer to natural languages, while the blocks concept is more an artificial
377 grouping based on groups of mostly 256 Unicode characters. For example, the
378 C<Latin> script contains letters from many blocks. On the other hand, the
379 C<Latin> script does not contain all the characters from those blocks. It
380 does not, for example, contain digits because digits are shared across many
381 scripts. Digits and other similar groups, like punctuation, are in a
382 category called C<Common>.
384 For more about scripts, see the UTR #24:
386 http://www.unicode.org/unicode/reports/tr24/
388 For more about blocks, see:
390 http://www.unicode.org/Public/UNIDATA/Blocks.txt
392 Blocks names are given with the C<In> prefix. For example, the
393 Katakana block is referenced via C<\p{InKatakana}>. The C<In>
394 prefix may be omitted if there is no nameing conflict with a script
395 or any other property, but it is recommended that C<In> always be used
398 These block names are supported:
400 InAlphabeticPresentationForms
402 InArabicPresentationFormsA
403 InArabicPresentationFormsB
413 InByzantineMusicalSymbols
415 InCJKCompatibilityForms
416 InCJKCompatibilityIdeographs
417 InCJKCompatibilityIdeographsSupplement
418 InCJKRadicalsSupplement
419 InCJKSymbolsAndPunctuation
420 InCJKUnifiedIdeographs
421 InCJKUnifiedIdeographsExtensionA
422 InCJKUnifiedIdeographsExtensionB
424 InCombiningDiacriticalMarks
426 InCombiningMarksForSymbols
433 InEnclosedAlphanumerics
434 InEnclosedCJKLettersAndMonths
444 InHalfwidthAndFullwidthForms
445 InHangulCompatibilityJamo
449 InHighPrivateUseSurrogates
453 InIdeographicDescriptionCharacters
461 InLatinExtendedAdditional
467 InMathematicalAlphanumericSymbols
468 InMathematicalOperators
469 InMiscellaneousSymbols
470 InMiscellaneousTechnical
477 InOpticalCharacterRecognition
483 InSpacingModifierLetters
485 InSuperscriptsAndSubscripts
493 InUnifiedCanadianAboriginalSyllabics
501 The special pattern C<\X> matches any extended Unicode sequence
502 (a "combining character sequence" in Standardese), where the first
503 character is a base character and subsequent characters are mark
504 characters that apply to the base character. It is equivalent to
509 The C<tr///> operator translates characters instead of bytes. Note
510 that the C<tr///CU> functionality has been removed, as the interface
511 was a mistake. For similar functionality see pack('U0', ...) and
516 Case translation operators use the Unicode case translation tables
517 when provided character input. Note that C<uc()> (also known as C<\U>
518 in doublequoted strings) translates to uppercase, while C<ucfirst>
519 (also known as C<\u> in doublequoted strings) translates to titlecase
520 (for languages that make the distinction). Naturally the
521 corresponding backslash sequences have the same semantics.
525 Most operators that deal with positions or lengths in the string will
526 automatically switch to using character positions, including
527 C<chop()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
528 C<sprintf()>, C<write()>, and C<length()>. Operators that
529 specifically don't switch include C<vec()>, C<pack()>, and
530 C<unpack()>. Operators that really don't care include C<chomp()>, as
531 well as any other operator that treats a string as a bucket of bits,
532 such as C<sort()>, and the operators dealing with filenames.
536 The C<pack()>/C<unpack()> letters "C<c>" and "C<C>" do I<not> change,
537 since they're often used for byte-oriented formats. (Again, think
538 "C<char>" in the C language.) However, there is a new "C<U>" specifier
539 that will convert between Unicode characters and integers.
543 The C<chr()> and C<ord()> functions work on characters. This is like
544 C<pack("U")> and C<unpack("U")>, not like C<pack("C")> and
545 C<unpack("C")>. In fact, the latter are how you now emulate
546 byte-oriented C<chr()> and C<ord()> for Unicode strings.
547 (Note that this reveals the internal encoding of Unicode strings,
548 which is not something one normally needs to care about at all.)
552 The bit string operators C<& | ^ ~> can operate on character data.
553 However, for backward compatibility reasons (bit string operations
554 when the characters all are less than 256 in ordinal value) one should
555 not mix C<~> (the bit complement) and characters both less than 256 and
556 equal or greater than 256. Most importantly, the DeMorgan's laws
557 (C<~($x|$y) eq ~$x&~$y>, C<~($x&$y) eq ~$x|~$y>) won't hold.
558 Another way to look at this is that the complement cannot return
559 B<both> the 8-bit (byte) wide bit complement B<and> the full character
564 lc(), uc(), lcfirst(), and ucfirst() work for the following cases:
570 the case mapping is from a single Unicode character to another
571 single Unicode character
575 the case mapping is from a single Unicode character to more
576 than one Unicode character
580 What doesn't yet work are the following cases:
586 the "final sigma" (Greek)
590 anything to with locales (Lithuanian, Turkish, Azeri)
594 See the Unicode Technical Report #21, Case Mappings, for more details.
598 And finally, C<scalar reverse()> reverses by character rather than by byte.
602 =head2 Character encodings for input and output
606 =head2 Unicode Regular Expression Support Level
608 The following list of Unicode regular expression support describes
609 feature by feature the Unicode support implemented in Perl as of Perl
610 5.8.0. The "Level N" and the section numbers refer to the Unicode
611 Technical Report 18, "Unicode Regular Expression Guidelines".
617 Level 1 - Basic Unicode Support
619 2.1 Hex Notation - done [1]
620 Named Notation - done [2]
621 2.2 Categories - done [3][4]
622 2.3 Subtraction - MISSING [5][6]
623 2.4 Simple Word Boundaries - done [7]
624 2.5 Simple Loose Matches - done [8]
625 2.6 End of Line - MISSING [9][10]
629 [ 3] . \p{...} \P{...}
630 [ 4] now scripts (see UTR#24 Script Names) in addition to blocks
632 [ 6] can use look-ahead to emulate subtraction (*)
633 [ 7] include Letters in word characters
634 [ 8] note that perl does Full casefolding in matching, not Simple:
635 for example U+1F88 is equivalent with U+1F000 U+03B9,
636 not with 1F80. This difference matters for certain Greek
637 capital letters with certain modifiers: the Full casefolding
638 decomposes the letter, while the Simple casefolding would map
639 it to a single character.
640 [ 9] see UTR#13 Unicode Newline Guidelines
641 [10] should do ^ and $ also on \x{85}, \x{2028} and \x{2029})
642 (should also affect <>, $., and script line numbers)
643 (the \x{85}, \x{2028} and \x{2029} do match \s)
645 (*) You can mimic class subtraction using lookahead.
646 For example, what TR18 might write as
648 [{Greek}-[{UNASSIGNED}]]
650 in Perl can be written as:
652 (?!\p{Unassigned})\p{InGreek}
653 (?=\p{Assigned})\p{InGreek}
655 But in this particular example, you probably really want
659 which will match assigned characters known to be part of the Greek script.
663 Level 2 - Extended Unicode Support
665 3.1 Surrogates - MISSING
666 3.2 Canonical Equivalents - MISSING [11][12]
667 3.3 Locale-Independent Graphemes - MISSING [13]
668 3.4 Locale-Independent Words - MISSING [14]
669 3.5 Locale-Independent Loose Matches - MISSING [15]
671 [11] see UTR#15 Unicode Normalization
672 [12] have Unicode::Normalize but not integrated to regexes
673 [13] have \X but at this level . should equal that
674 [14] need three classes, not just \w and \W
675 [15] see UTR#21 Case Mappings
679 Level 3 - Locale-Sensitive Support
681 4.1 Locale-Dependent Categories - MISSING
682 4.2 Locale-Dependent Graphemes - MISSING [16][17]
683 4.3 Locale-Dependent Words - MISSING
684 4.4 Locale-Dependent Loose Matches - MISSING
685 4.5 Locale-Dependent Ranges - MISSING
687 [16] see UTR#10 Unicode Collation Algorithms
688 [17] have Unicode::Collate but not integrated to regexes
692 =head2 Unicode Encodings
694 Unicode characters are assigned to I<code points> which are abstract
695 numbers. To use these numbers various encodings are needed.
703 UTF-8 is a variable-length (1 to 6 bytes, current character allocations
704 require 4 bytes), byteorder independent encoding. For ASCII, UTF-8 is
705 transparent (and we really do mean 7-bit ASCII, not another 8-bit encoding).
707 The following table is from Unicode 3.2.
709 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
711 U+0000..U+007F 00..7F
712 U+0080..U+07FF C2..DF 80..BF
713 U+0800..U+0FFF E0 A0..BF 80..BF
714 U+1000..U+CFFF E1..EC 80..BF 80..BF
715 U+D000..U+D7FF ED 80..9F 80..BF
716 U+D800..U+DFFF ******* ill-formed *******
717 U+E000..U+FFFF EE..EF 80..BF 80..BF
718 U+10000..U+3FFFF F0 90..BF 80..BF 80..BF
719 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
720 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
722 Note the A0..BF in U+0800..U+0FFF, the 80..9F in U+D000...U+D7FF,
723 the 90..BF in U+10000..U+3FFFF, and the 80...8F in U+100000..U+10FFFF.
724 Or, another way to look at it, as bits:
726 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
729 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
730 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
731 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
733 As you can see, the continuation bytes all begin with C<10>, and the
734 leading bits of the start byte tell how many bytes the are in the
741 Like UTF-8, but EBCDIC-safe, as UTF-8 is ASCII-safe.
745 UTF-16, UTF-16BE, UTF16-LE, Surrogates, and BOMs (Byte Order Marks)
747 (The followings items are mostly for reference, Perl doesn't
748 use them internally.)
750 UTF-16 is a 2 or 4 byte encoding. The Unicode code points
751 0x0000..0xFFFF are stored in two 16-bit units, and the code points
752 0x010000..0x10FFFF in two 16-bit units. The latter case is
753 using I<surrogates>, the first 16-bit unit being the I<high
754 surrogate>, and the second being the I<low surrogate>.
756 Surrogates are code points set aside to encode the 0x01000..0x10FFFF
757 range of Unicode code points in pairs of 16-bit units. The I<high
758 surrogates> are the range 0xD800..0xDBFF, and the I<low surrogates>
759 are the range 0xDC00..0xDFFFF. The surrogate encoding is
761 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
762 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
766 $uni = 0x10000 + ($hi - 0xD8000) * 0x400 + ($lo - 0xDC00);
768 If you try to generate surrogates (for example by using chr()), you
769 will get a warning if warnings are turned on (C<-w> or C<use
770 warnings;>) because those code points are not valid for a Unicode
773 Because of the 16-bitness, UTF-16 is byteorder dependent. UTF-16
774 itself can be used for in-memory computations, but if storage or
775 transfer is required, either UTF-16BE (Big Endian) or UTF-16LE
776 (Little Endian) must be chosen.
778 This introduces another problem: what if you just know that your data
779 is UTF-16, but you don't know which endianness? Byte Order Marks
780 (BOMs) are a solution to this. A special character has been reserved
781 in Unicode to function as a byte order marker: the character with the
782 code point 0xFEFF is the BOM.
784 The trick is that if you read a BOM, you will know the byte order,
785 since if it was written on a big endian platform, you will read the
786 bytes 0xFE 0xFF, but if it was written on a little endian platform,
787 you will read the bytes 0xFF 0xFE. (And if the originating platform
788 was writing in UTF-8, you will read the bytes 0xEF 0xBB 0xBF.)
790 The way this trick works is that the character with the code point
791 0xFFFE is guaranteed not to be a valid Unicode character, so the
792 sequence of bytes 0xFF 0xFE is unambiguously "BOM, represented in
793 little-endian format" and cannot be "0xFFFE, represented in big-endian
798 UTF-32, UTF-32BE, UTF32-LE
800 The UTF-32 family is pretty much like the UTF-16 family, expect that
801 the units are 32-bit, and therefore the surrogate scheme is not
802 needed. The BOM signatures will be 0x00 0x00 0xFE 0xFF for BE and
803 0xFF 0xFE 0x00 0x00 for LE.
809 Encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
810 encoding, UCS-4 is a 32-bit encoding. Unlike UTF-16, UCS-2
811 is not extensible beyond 0xFFFF, because it does not use surrogates.
817 A seven-bit safe (non-eight-bit) encoding, useful if the
818 transport/storage is not eight-bit safe. Defined by RFC 2152.
822 =head2 Security Implications of Malformed UTF-8
824 Unfortunately, the specification of UTF-8 leaves some room for
825 interpretation of how many bytes of encoded output one should generate
826 from one input Unicode character. Strictly speaking, one is supposed
827 to always generate the shortest possible sequence of UTF-8 bytes,
828 because otherwise there is potential for input buffer overflow at
829 the receiving end of a UTF-8 connection. Perl always generates the
830 shortest length UTF-8, and with warnings on (C<-w> or C<use
831 warnings;>) Perl will warn about non-shortest length UTF-8 (and other
832 malformations, too, such as the surrogates, which are not real
833 Unicode code points.)
835 =head2 Unicode in Perl on EBCDIC
837 The way Unicode is handled on EBCDIC platforms is still rather
838 experimental. On such a platform, references to UTF-8 encoding in this
839 document and elsewhere should be read as meaning UTF-EBCDIC as
840 specified in Unicode Technical Report 16 unless ASCII vs EBCDIC issues
841 are specifically discussed. There is no C<utfebcdic> pragma or
842 ":utfebcdic" layer, rather, "utf8" and ":utf8" are re-used to mean
843 the platform's "natural" 8-bit encoding of Unicode. See L<perlebcdic>
844 for more discussion of the issues.
848 Usually locale settings and Unicode do not affect each other, but
849 there are a couple of exceptions:
855 If your locale environment variables (LANGUAGE, LC_ALL, LC_CTYPE, LANG)
856 contain the strings 'UTF-8' or 'UTF8' (case-insensitive matching),
857 the default encoding of your STDIN, STDOUT, and STDERR, and of
858 B<any subsequent file open>, is UTF-8.
862 Perl tries really hard to work both with Unicode and the old byte
863 oriented world: most often this is nice, but sometimes this causes
868 =head2 Using Unicode in XS
870 If you want to handle Perl Unicode in XS extensions, you may find
871 the following C APIs useful (see perlapi for details):
877 DO_UTF8(sv) returns true if the UTF8 flag is on and the bytes pragma
878 is not in effect. SvUTF8(sv) returns true is the UTF8 flag is on, the
879 bytes pragma is ignored. The UTF8 flag being on does B<not> mean that
880 there are any characters of code points greater than 255 (or 127) in
881 the scalar, or that there even are any characters in the scalar.
882 What the UTF8 flag means is that the sequence of octets in the
883 representation of the scalar is the sequence of UTF-8 encoded
884 code points of the characters of a string. The UTF8 flag being
885 off means that each octet in this representation encodes a single
886 character with codepoint 0..255 within the string. Perl's Unicode
887 model is not to use UTF-8 until it's really necessary.
891 uvuni_to_utf8(buf, chr) writes a Unicode character code point into a
892 buffer encoding the code point as UTF-8, and returns a pointer
893 pointing after the UTF-8 bytes.
897 utf8_to_uvuni(buf, lenp) reads UTF-8 encoded bytes from a buffer and
898 returns the Unicode character code point (and optionally the length of
899 the UTF-8 byte sequence).
903 utf8_length(start, end) returns the length of the UTF-8 encoded buffer
904 in characters. sv_len_utf8(sv) returns the length of the UTF-8 encoded
909 sv_utf8_upgrade(sv) converts the string of the scalar to its UTF-8
910 encoded form. sv_utf8_downgrade(sv) does the opposite (if possible).
911 sv_utf8_encode(sv) is like sv_utf8_upgrade but the UTF8 flag does not
912 get turned on. sv_utf8_decode() does the opposite of sv_utf8_encode().
913 Note that none of these are to be used as general purpose encoding/decoding
914 interfaces: use Encode for that. sv_utf8_upgrade() is affected by the
915 encoding pragma, but sv_utf8_downgrade() is not (since the encoding
916 pragma is designed to be a one-way street).
920 is_utf8_char(s) returns true if the pointer points to a valid UTF-8
925 is_utf8_string(buf, len) returns true if the len bytes of the buffer
930 UTF8SKIP(buf) will return the number of bytes in the UTF-8 encoded
931 character in the buffer. UNISKIP(chr) will return the number of bytes
932 required to UTF-8-encode the Unicode character code point. UTF8SKIP()
933 is useful for example for iterating over the characters of a UTF-8
934 encoded buffer; UNISKIP() is useful for example in computing
935 the size required for a UTF-8 encoded buffer.
939 utf8_distance(a, b) will tell the distance in characters between the
940 two pointers pointing to the same UTF-8 encoded buffer.
944 utf8_hop(s, off) will return a pointer to an UTF-8 encoded buffer that
945 is C<off> (positive or negative) Unicode characters displaced from the
946 UTF-8 buffer C<s>. Be careful not to overstep the buffer: utf8_hop()
947 will merrily run off the end or the beginning if told to do so.
951 pv_uni_display(dsv, spv, len, pvlim, flags) and sv_uni_display(dsv,
952 ssv, pvlim, flags) are useful for debug output of Unicode strings and
953 scalars. By default they are useful only for debug: they display
954 B<all> characters as hexadecimal code points, but with the flags
955 UNI_DISPLAY_ISPRINT and UNI_DISPLAY_BACKSLASH you can make the output
960 ibcmp_utf8(s1, pe1, u1, l1, u1, s2, pe2, l2, u2) can be used to
961 compare two strings case-insensitively in Unicode.
962 (For case-sensitive comparisons you can just use memEQ() and memNE()
967 For more information, see L<perlapi>, and F<utf8.c> and F<utf8.h>
968 in the Perl source code distribution.
972 Use of locales with Unicode data may lead to odd results. Currently
973 there is some attempt to apply 8-bit locale info to characters in the
974 range 0..255, but this is demonstrably incorrect for locales that use
975 characters above that range when mapped into Unicode. It will also
976 tend to run slower. Use of locales with Unicode is discouraged.
978 Some functions are slower when working on UTF-8 encoded strings than
979 on byte encoded strings. All functions that need to hop over
980 characters such as length(), substr() or index() can work B<much>
981 faster when the underlying data are byte-encoded. Witness the
988 our $u = our $b = "x" x $l;
989 substr($u,0,1) = "\x{100}";
991 LENGTH_B => q{ length($b) },
992 LENGTH_U => q{ length($u) },
993 SUBSTR_B => q{ substr($b, $l/4, $l/2) },
994 SUBSTR_U => q{ substr($u, $l/4, $l/2) },
997 Benchmark: running LENGTH_B, LENGTH_U, SUBSTR_B, SUBSTR_U for at least 2 CPU seconds...
998 LENGTH_B: 2 wallclock secs ( 2.36 usr + 0.00 sys = 2.36 CPU) @ 5649983.05/s (n=13333960)
999 LENGTH_U: 2 wallclock secs ( 2.11 usr + 0.00 sys = 2.11 CPU) @ 12155.45/s (n=25648)
1000 SUBSTR_B: 3 wallclock secs ( 2.16 usr + 0.00 sys = 2.16 CPU) @ 374480.09/s (n=808877)
1001 SUBSTR_U: 2 wallclock secs ( 2.11 usr + 0.00 sys = 2.11 CPU) @ 6791.00/s (n=14329)
1003 The numbers show an incredible slowness on long UTF-8 strings and you
1004 should carefully avoid to use these functions within tight loops. For
1005 example if you want to iterate over characters, it is infinitely
1006 better to split into an array than to use substr, as the following
1013 our $u = our $b = "x" x $l;
1014 substr($u,0,1) = "\x{100}";
1016 SPLIT_B => q{ for my $c (split //, $b){} },
1017 SPLIT_U => q{ for my $c (split //, $u){} },
1018 SUBSTR_B => q{ for my $i (0..length($b)-1){my $c = substr($b,$i,1);} },
1019 SUBSTR_U => q{ for my $i (0..length($u)-1){my $c = substr($u,$i,1);} },
1022 Benchmark: running SPLIT_B, SPLIT_U, SUBSTR_B, SUBSTR_U for at least 5 CPU seconds...
1023 SPLIT_B: 6 wallclock secs ( 5.29 usr + 0.00 sys = 5.29 CPU) @ 56.14/s (n=297)
1024 SPLIT_U: 5 wallclock secs ( 5.17 usr + 0.01 sys = 5.18 CPU) @ 55.21/s (n=286)
1025 SUBSTR_B: 5 wallclock secs ( 5.34 usr + 0.00 sys = 5.34 CPU) @ 123.22/s (n=658)
1026 SUBSTR_U: 7 wallclock secs ( 6.20 usr + 0.00 sys = 6.20 CPU) @ 0.81/s (n=5)
1028 You see, the algorithm based on substr() was faster with byte encoded
1029 data but it is pathologically slow with UTF-8 data.
1033 L<perluniintro>, L<encoding>, L<Encode>, L<open>, L<utf8>, L<bytes>,
1034 L<perlretut>, L<perlvar/"${^WIDE_SYSTEM_CALLS}">