Commit | Line | Data |
---|---|---|
393fec97 GS |
1 | =head1 NAME |
2 | ||
3 | perlunicode - Unicode support in Perl | |
4 | ||
5 | =head1 DESCRIPTION | |
6 | ||
0a1f2d14 | 7 | =head2 Important Caveats |
21bad921 | 8 | |
c349b1b9 JH |
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. | |
21bad921 | 12 | |
13a2d996 | 13 | =over 4 |
21bad921 GS |
14 | |
15 | =item Input and Output Disciplines | |
16 | ||
75daf61c JH |
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. | |
0a1f2d14 | 19 | Other encodings can be converted to perl's encoding on input, or from |
c349b1b9 JH |
20 | perl's encoding on output by use of the ":encoding(...)" layer. |
21 | See L<open>. | |
22 | ||
d1be9408 | 23 | To mark the Perl source itself as being in a particular encoding, |
c349b1b9 | 24 | see L<encoding>. |
21bad921 GS |
25 | |
26 | =item Regular Expressions | |
27 | ||
c349b1b9 JH |
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. | |
21bad921 | 32 | |
ad0029c4 | 33 | =item C<use utf8> still needed to enable UTF-8/UTF-EBCDIC in scripts |
21bad921 | 34 | |
c349b1b9 JH |
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 | |
3e4dbfed | 37 | based machines, or to recognize UTF-EBCDIC on EBCDIC based machines. |
c349b1b9 JH |
38 | B<NOTE: this should be the only place where an explicit C<use utf8> |
39 | is needed>. | |
21bad921 | 40 | |
1768d7eb | 41 | You can also use the C<encoding> pragma to change the default encoding |
6ec9efec | 42 | of the data in your script; see L<encoding>. |
1768d7eb | 43 | |
21bad921 GS |
44 | =back |
45 | ||
46 | =head2 Byte and Character semantics | |
393fec97 GS |
47 | |
48 | Beginning with version 5.6, Perl uses logically wide characters to | |
3e4dbfed | 49 | represent strings internally. |
393fec97 | 50 | |
75daf61c JH |
51 | In future, Perl-level operations can be expected to work with |
52 | characters rather than bytes, in general. | |
393fec97 | 53 | |
75daf61c JH |
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. | |
8cbd9a7a GS |
61 | |
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), | |
21bad921 | 67 | or from literals and constants in the source text. |
8cbd9a7a | 68 | |
c349b1b9 | 69 | On Windows platforms, if the C<-C> command line switch is used, (or the |
75daf61c JH |
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 | |
c349b1b9 JH |
72 | currently only implemented on Windows since other platforms lack an |
73 | API standard on this area. | |
8cbd9a7a | 74 | |
75daf61c JH |
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>. | |
8cbd9a7a GS |
77 | |
78 | The C<utf8> pragma is primarily a compatibility device that enables | |
75daf61c | 79 | recognition of UTF-(8|EBCDIC) in literals encountered by the parser. |
7dedd01f JH |
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>. | |
8cbd9a7a GS |
83 | |
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 | |
3e4dbfed | 89 | literal Unicode string constant in the program), character semantics |
8cbd9a7a | 90 | apply; otherwise, byte semantics are in effect. To force byte semantics |
8058d7ab | 91 | on Unicode data, the C<bytes> pragma should be used. |
393fec97 | 92 | |
0a378802 JH |
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 | |
3e4dbfed JF |
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> | |
0a378802 | 99 | pragma, see L<encoding>. |
7dedd01f | 100 | |
feda178f JH |
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. | |
393fec97 | 107 | |
8cbd9a7a | 108 | =head2 Effects of character semantics |
393fec97 GS |
109 | |
110 | Character semantics have the following effects: | |
111 | ||
112 | =over 4 | |
113 | ||
114 | =item * | |
115 | ||
116 | Strings and patterns may contain characters that have an ordinal value | |
21bad921 | 117 | larger than 255. |
393fec97 | 118 | |
feda178f JH |
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. | |
3e4dbfed JF |
124 | |
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. | |
129 | ||
130 | Additionally, if you | |
131 | use charnames ':full'; | |
132 | you can use the C<\N{...}> notation, putting the official Unicode character | |
133 | name within the curlies. For example, C<\N{WHITE SMILING FACE}>. | |
134 | This works for all characters that have names. | |
393fec97 GS |
135 | |
136 | =item * | |
137 | ||
3e4dbfed JF |
138 | If an appropriate L<encoding> is specified, |
139 | identifiers within the Perl script may contain Unicode alphanumeric | |
393fec97 | 140 | characters, including ideographs. (You are currently on your own when |
75daf61c JH |
141 | it comes to using the canonical forms of characters--Perl doesn't |
142 | (yet) attempt to canonicalize variable names for you.) | |
393fec97 | 143 | |
393fec97 GS |
144 | =item * |
145 | ||
146 | Regular expressions match characters instead of bytes. For instance, | |
147 | "." matches a character instead of a byte. (However, the C<\C> pattern | |
75daf61c | 148 | is provided to force a match a single byte ("C<char>" in C, hence C<\C>).) |
393fec97 | 149 | |
393fec97 GS |
150 | =item * |
151 | ||
152 | Character classes in regular expressions match characters instead of | |
153 | bytes, and match against the character properties specified in the | |
75daf61c JH |
154 | Unicode properties database. So C<\w> can be used to match an |
155 | ideograph, for instance. | |
393fec97 | 156 | |
393fec97 GS |
157 | =item * |
158 | ||
eb0cc9e3 JH |
159 | Named Unicode properties, scripts, and block ranges may be used like |
160 | character classes via the new C<\p{}> (matches property) and C<\P{}> | |
161 | (doesn't match property) constructs. For instance, C<\p{Lu}> matches any | |
feda178f JH |
162 | character with the Unicode "Lu" (Letter, uppercase) property, while |
163 | C<\p{M}> matches any character with a "M" (mark -- accents and such) | |
eb0cc9e3 JH |
164 | property. Single letter properties may omit the brackets, so that can be |
165 | written C<\pM> also. Many predefined properties are available, such | |
166 | as C<\p{Mirrored}> and C<\p{Tibetan}>. | |
4193bef7 | 167 | |
cfc01aea | 168 | The official Unicode script and block names have spaces and dashes as |
eb0cc9e3 JH |
169 | separators, but for convenience you can have dashes, spaces, and underbars |
170 | at every word division, and you need not care about correct casing. It is | |
171 | recommended, however, that for consistency you use the following naming: | |
172 | the official Unicode script, block, or property name (see below for the | |
173 | additional rules that apply to block names), with whitespace and dashes | |
174 | removed, and the words "uppercase-first-lowercase-rest". That is, "Latin-1 | |
175 | Supplement" becomes "Latin1Supplement". | |
4193bef7 | 176 | |
a1cc1cb1 | 177 | You can also negate both C<\p{}> and C<\P{}> by introducing a caret |
eb0cc9e3 JH |
178 | (^) between the first curly and the property name: C<\p{^Tamil}> is |
179 | equal to C<\P{Tamil}>. | |
4193bef7 | 180 | |
eb0cc9e3 JH |
181 | Here are the basic Unicode General Category properties, followed by their |
182 | long form (you can use either, e.g. C<\p{Lu}> and C<\p{LowercaseLetter}> | |
183 | are identical). | |
393fec97 | 184 | |
d73e5302 JH |
185 | Short Long |
186 | ||
187 | L Letter | |
eb0cc9e3 JH |
188 | Lu UppercaseLetter |
189 | Ll LowercaseLetter | |
190 | Lt TitlecaseLetter | |
191 | Lm ModifierLetter | |
192 | Lo OtherLetter | |
d73e5302 JH |
193 | |
194 | M Mark | |
eb0cc9e3 JH |
195 | Mn NonspacingMark |
196 | Mc SpacingMark | |
197 | Me EnclosingMark | |
d73e5302 JH |
198 | |
199 | N Number | |
eb0cc9e3 JH |
200 | Nd DecimalNumber |
201 | Nl LetterNumber | |
202 | No OtherNumber | |
d73e5302 JH |
203 | |
204 | P Punctuation | |
eb0cc9e3 JH |
205 | Pc ConnectorPunctuation |
206 | Pd DashPunctuation | |
207 | Ps OpenPunctuation | |
208 | Pe ClosePunctuation | |
209 | Pi InitialPunctuation | |
d73e5302 | 210 | (may behave like Ps or Pe depending on usage) |
eb0cc9e3 | 211 | Pf FinalPunctuation |
d73e5302 | 212 | (may behave like Ps or Pe depending on usage) |
eb0cc9e3 | 213 | Po OtherPunctuation |
d73e5302 JH |
214 | |
215 | S Symbol | |
eb0cc9e3 JH |
216 | Sm MathSymbol |
217 | Sc CurrencySymbol | |
218 | Sk ModifierSymbol | |
219 | So OtherSymbol | |
d73e5302 JH |
220 | |
221 | Z Separator | |
eb0cc9e3 JH |
222 | Zs SpaceSeparator |
223 | Zl LineSeparator | |
224 | Zp ParagraphSeparator | |
d73e5302 JH |
225 | |
226 | C Other | |
e150c829 JH |
227 | Cc Control |
228 | Cf Format | |
eb0cc9e3 JH |
229 | Cs Surrogate (not usable) |
230 | Co PrivateUse | |
e150c829 | 231 | Cn Unassigned |
1ac13f9a | 232 | |
3e4dbfed JF |
233 | The single-letter properties match all characters in any of the |
234 | two-letter sub-properties starting with the same letter. | |
1ac13f9a | 235 | There's also C<L&> which is an alias for C<Ll>, C<Lu>, and C<Lt>. |
32293815 | 236 | |
eb0cc9e3 JH |
237 | Because Perl hides the need for the user to understand the internal |
238 | representation of Unicode characters, it has no need to support the | |
239 | somewhat messy concept of surrogates. Therefore, the C<Cs> property is not | |
240 | supported. | |
d73e5302 | 241 | |
eb0cc9e3 JH |
242 | Because scripts differ in their directionality (for example Hebrew is |
243 | written right to left), Unicode supplies these properties: | |
32293815 | 244 | |
eb0cc9e3 | 245 | Property Meaning |
92e830a9 | 246 | |
d73e5302 JH |
247 | BidiL Left-to-Right |
248 | BidiLRE Left-to-Right Embedding | |
249 | BidiLRO Left-to-Right Override | |
250 | BidiR Right-to-Left | |
251 | BidiAL Right-to-Left Arabic | |
252 | BidiRLE Right-to-Left Embedding | |
253 | BidiRLO Right-to-Left Override | |
254 | BidiPDF Pop Directional Format | |
255 | BidiEN European Number | |
256 | BidiES European Number Separator | |
257 | BidiET European Number Terminator | |
258 | BidiAN Arabic Number | |
259 | BidiCS Common Number Separator | |
260 | BidiNSM Non-Spacing Mark | |
261 | BidiBN Boundary Neutral | |
262 | BidiB Paragraph Separator | |
263 | BidiS Segment Separator | |
264 | BidiWS Whitespace | |
265 | BidiON Other Neutrals | |
32293815 | 266 | |
eb0cc9e3 JH |
267 | For example, C<\p{BidiR}> matches all characters that are normally |
268 | written right to left. | |
269 | ||
210b36aa AMS |
270 | =back |
271 | ||
2796c109 JH |
272 | =head2 Scripts |
273 | ||
eb0cc9e3 JH |
274 | The scripts available via C<\p{...}> and C<\P{...}>, for example |
275 | C<\p{Latin}> or \p{Cyrillic>, are as follows: | |
2796c109 | 276 | |
1ac13f9a | 277 | Arabic |
e9ad1727 | 278 | Armenian |
1ac13f9a | 279 | Bengali |
e9ad1727 | 280 | Bopomofo |
eb0cc9e3 | 281 | CanadianAboriginal |
e9ad1727 JH |
282 | Cherokee |
283 | Cyrillic | |
284 | Deseret | |
285 | Devanagari | |
286 | Ethiopic | |
287 | Georgian | |
288 | Gothic | |
289 | Greek | |
1ac13f9a | 290 | Gujarati |
e9ad1727 JH |
291 | Gurmukhi |
292 | Han | |
293 | Hangul | |
294 | Hebrew | |
295 | Hiragana | |
296 | Inherited | |
1ac13f9a | 297 | Kannada |
e9ad1727 JH |
298 | Katakana |
299 | Khmer | |
1ac13f9a | 300 | Lao |
e9ad1727 JH |
301 | Latin |
302 | Malayalam | |
303 | Mongolian | |
1ac13f9a | 304 | Myanmar |
1ac13f9a | 305 | Ogham |
eb0cc9e3 | 306 | OldItalic |
e9ad1727 | 307 | Oriya |
1ac13f9a | 308 | Runic |
e9ad1727 JH |
309 | Sinhala |
310 | Syriac | |
311 | Tamil | |
312 | Telugu | |
313 | Thaana | |
314 | Thai | |
315 | Tibetan | |
1ac13f9a | 316 | Yi |
1ac13f9a JH |
317 | |
318 | There are also extended property classes that supplement the basic | |
319 | properties, defined by the F<PropList> Unicode database: | |
320 | ||
e9ad1727 | 321 | ASCII_Hex_Digit |
eb0cc9e3 | 322 | BidiControl |
1ac13f9a | 323 | Dash |
1ac13f9a JH |
324 | Diacritic |
325 | Extender | |
eb0cc9e3 | 326 | HexDigit |
e9ad1727 JH |
327 | Hyphen |
328 | Ideographic | |
eb0cc9e3 JH |
329 | JoinControl |
330 | NoncharacterCodePoint | |
331 | OtherAlphabetic | |
332 | OtherLowercase | |
333 | OtherMath | |
334 | OtherUppercase | |
335 | QuotationMark | |
336 | WhiteSpace | |
1ac13f9a JH |
337 | |
338 | and further derived properties: | |
339 | ||
eb0cc9e3 JH |
340 | Alphabetic Lu + Ll + Lt + Lm + Lo + OtherAlphabetic |
341 | Lowercase Ll + OtherLowercase | |
342 | Uppercase Lu + OtherUppercase | |
343 | Math Sm + OtherMath | |
1ac13f9a JH |
344 | |
345 | ID_Start Lu + Ll + Lt + Lm + Lo + Nl | |
346 | ID_Continue ID_Start + Mn + Mc + Nd + Pc | |
347 | ||
348 | Any Any character | |
eb0cc9e3 JH |
349 | Assigned Any non-Cn character (i.e. synonym for C<\P{Cn}>) |
350 | Unassigned Synonym for C<\p{Cn}> | |
1ac13f9a | 351 | Common Any character (or unassigned code point) |
e150c829 | 352 | not explicitly assigned to a script |
2796c109 | 353 | |
eb0cc9e3 JH |
354 | For backward compatability, all properties mentioned so far may have C<Is> |
355 | prepended to their name (e.g. C<\P{IsLu}> is equal to C<\P{Lu}>). | |
356 | ||
2796c109 JH |
357 | =head2 Blocks |
358 | ||
eb0cc9e3 JH |
359 | In addition to B<scripts>, Unicode also defines B<blocks> of characters. |
360 | The difference between scripts and blocks is that the scripts concept is | |
361 | closer to natural languages, while the blocks concept is more an artificial | |
362 | grouping based on groups of mostly 256 Unicode characters. For example, the | |
363 | C<Latin> script contains letters from many blocks. On the other hand, the | |
364 | C<Latin> script does not contain all the characters from those blocks. It | |
365 | does not, for example, contain digits because digits are shared across many | |
366 | scripts. Digits and other similar groups, like punctuation, are in a | |
367 | category called C<Common>. | |
2796c109 | 368 | |
cfc01aea JF |
369 | For more about scripts, see the UTR #24: |
370 | ||
371 | http://www.unicode.org/unicode/reports/tr24/ | |
372 | ||
373 | For more about blocks, see: | |
374 | ||
375 | http://www.unicode.org/Public/UNIDATA/Blocks.txt | |
2796c109 | 376 | |
eb0cc9e3 | 377 | Blocks names are given with the C<In> prefix. For example, the |
92e830a9 | 378 | Katakana block is referenced via C<\p{InKatakana}>. The C<In> |
eb0cc9e3 JH |
379 | prefix may be omitted if there is no nameing conflict with a script |
380 | or any other property, but it is recommended that C<In> always be used | |
381 | to avoid confusion. | |
382 | ||
383 | These block names are supported: | |
384 | ||
385 | InAlphabeticPresentationForms | |
386 | InArabicBlock | |
387 | InArabicPresentationFormsA | |
388 | InArabicPresentationFormsB | |
389 | InArmenianBlock | |
390 | InArrows | |
391 | InBasicLatin | |
392 | InBengaliBlock | |
393 | InBlockElements | |
394 | InBopomofoBlock | |
395 | InBopomofoExtended | |
396 | InBoxDrawing | |
397 | InBraillePatterns | |
398 | InByzantineMusicalSymbols | |
399 | InCJKCompatibility | |
400 | InCJKCompatibilityForms | |
401 | InCJKCompatibilityIdeographs | |
402 | InCJKCompatibilityIdeographsSupplement | |
403 | InCJKRadicalsSupplement | |
404 | InCJKSymbolsAndPunctuation | |
405 | InCJKUnifiedIdeographs | |
406 | InCJKUnifiedIdeographsExtensionA | |
407 | InCJKUnifiedIdeographsExtensionB | |
408 | InCherokeeBlock | |
409 | InCombiningDiacriticalMarks | |
410 | InCombiningHalfMarks | |
411 | InCombiningMarksForSymbols | |
412 | InControlPictures | |
413 | InCurrencySymbols | |
414 | InCyrillicBlock | |
415 | InDeseretBlock | |
416 | InDevanagariBlock | |
417 | InDingbats | |
418 | InEnclosedAlphanumerics | |
419 | InEnclosedCJKLettersAndMonths | |
420 | InEthiopicBlock | |
421 | InGeneralPunctuation | |
422 | InGeometricShapes | |
423 | InGeorgianBlock | |
424 | InGothicBlock | |
425 | InGreekBlock | |
426 | InGreekExtended | |
427 | InGujaratiBlock | |
428 | InGurmukhiBlock | |
429 | InHalfwidthAndFullwidthForms | |
430 | InHangulCompatibilityJamo | |
431 | InHangulJamo | |
432 | InHangulSyllables | |
433 | InHebrewBlock | |
434 | InHighPrivateUseSurrogates | |
435 | InHighSurrogates | |
436 | InHiraganaBlock | |
437 | InIPAExtensions | |
438 | InIdeographicDescriptionCharacters | |
439 | InKanbun | |
440 | InKangxiRadicals | |
441 | InKannadaBlock | |
442 | InKatakanaBlock | |
443 | InKhmerBlock | |
444 | InLaoBlock | |
445 | InLatin1Supplement | |
446 | InLatinExtendedAdditional | |
447 | InLatinExtended-A | |
448 | InLatinExtended-B | |
449 | InLetterlikeSymbols | |
450 | InLowSurrogates | |
451 | InMalayalamBlock | |
452 | InMathematicalAlphanumericSymbols | |
453 | InMathematicalOperators | |
454 | InMiscellaneousSymbols | |
455 | InMiscellaneousTechnical | |
456 | InMongolianBlock | |
457 | InMusicalSymbols | |
458 | InMyanmarBlock | |
459 | InNumberForms | |
460 | InOghamBlock | |
461 | InOldItalicBlock | |
462 | InOpticalCharacterRecognition | |
463 | InOriyaBlock | |
464 | InPrivateUse | |
465 | InRunicBlock | |
466 | InSinhalaBlock | |
467 | InSmallFormVariants | |
468 | InSpacingModifierLetters | |
469 | InSpecials | |
470 | InSuperscriptsAndSubscripts | |
471 | InSyriacBlock | |
472 | InTags | |
473 | InTamilBlock | |
474 | InTeluguBlock | |
475 | InThaanaBlock | |
476 | InThaiBlock | |
477 | InTibetanBlock | |
478 | InUnifiedCanadianAboriginalSyllabics | |
479 | InYiRadicals | |
480 | InYiSyllables | |
32293815 | 481 | |
210b36aa AMS |
482 | =over 4 |
483 | ||
393fec97 GS |
484 | =item * |
485 | ||
c29a771d | 486 | The special pattern C<\X> matches any extended Unicode sequence |
393fec97 GS |
487 | (a "combining character sequence" in Standardese), where the first |
488 | character is a base character and subsequent characters are mark | |
489 | characters that apply to the base character. It is equivalent to | |
490 | C<(?:\PM\pM*)>. | |
491 | ||
393fec97 GS |
492 | =item * |
493 | ||
383e7cdd JH |
494 | The C<tr///> operator translates characters instead of bytes. Note |
495 | that the C<tr///CU> functionality has been removed, as the interface | |
496 | was a mistake. For similar functionality see pack('U0', ...) and | |
497 | pack('C0', ...). | |
393fec97 | 498 | |
393fec97 GS |
499 | =item * |
500 | ||
501 | Case translation operators use the Unicode case translation tables | |
44bc797b JH |
502 | when provided character input. Note that C<uc()> (also known as C<\U> |
503 | in doublequoted strings) translates to uppercase, while C<ucfirst> | |
504 | (also known as C<\u> in doublequoted strings) translates to titlecase | |
505 | (for languages that make the distinction). Naturally the | |
506 | corresponding backslash sequences have the same semantics. | |
393fec97 GS |
507 | |
508 | =item * | |
509 | ||
510 | Most operators that deal with positions or lengths in the string will | |
75daf61c JH |
511 | automatically switch to using character positions, including |
512 | C<chop()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>, | |
513 | C<sprintf()>, C<write()>, and C<length()>. Operators that | |
514 | specifically don't switch include C<vec()>, C<pack()>, and | |
515 | C<unpack()>. Operators that really don't care include C<chomp()>, as | |
516 | well as any other operator that treats a string as a bucket of bits, | |
517 | such as C<sort()>, and the operators dealing with filenames. | |
393fec97 GS |
518 | |
519 | =item * | |
520 | ||
521 | The C<pack()>/C<unpack()> letters "C<c>" and "C<C>" do I<not> change, | |
522 | since they're often used for byte-oriented formats. (Again, think | |
523 | "C<char>" in the C language.) However, there is a new "C<U>" specifier | |
3e4dbfed | 524 | that will convert between Unicode characters and integers. |
393fec97 GS |
525 | |
526 | =item * | |
527 | ||
528 | The C<chr()> and C<ord()> functions work on characters. This is like | |
529 | C<pack("U")> and C<unpack("U")>, not like C<pack("C")> and | |
530 | C<unpack("C")>. In fact, the latter are how you now emulate | |
35bcd338 | 531 | byte-oriented C<chr()> and C<ord()> for Unicode strings. |
3e4dbfed JF |
532 | (Note that this reveals the internal encoding of Unicode strings, |
533 | which is not something one normally needs to care about at all.) | |
393fec97 GS |
534 | |
535 | =item * | |
536 | ||
a1ca4561 YST |
537 | The bit string operators C<& | ^ ~> can operate on character data. |
538 | However, for backward compatibility reasons (bit string operations | |
75daf61c JH |
539 | when the characters all are less than 256 in ordinal value) one should |
540 | not mix C<~> (the bit complement) and characters both less than 256 and | |
a1ca4561 YST |
541 | equal or greater than 256. Most importantly, the DeMorgan's laws |
542 | (C<~($x|$y) eq ~$x&~$y>, C<~($x&$y) eq ~$x|~$y>) won't hold. | |
543 | Another way to look at this is that the complement cannot return | |
75daf61c | 544 | B<both> the 8-bit (byte) wide bit complement B<and> the full character |
a1ca4561 YST |
545 | wide bit complement. |
546 | ||
547 | =item * | |
548 | ||
983ffd37 JH |
549 | lc(), uc(), lcfirst(), and ucfirst() work for the following cases: |
550 | ||
551 | =over 8 | |
552 | ||
553 | =item * | |
554 | ||
555 | the case mapping is from a single Unicode character to another | |
556 | single Unicode character | |
557 | ||
558 | =item * | |
559 | ||
560 | the case mapping is from a single Unicode character to more | |
561 | than one Unicode character | |
562 | ||
563 | =back | |
564 | ||
210b36aa | 565 | What doesn't yet work are the following cases: |
983ffd37 JH |
566 | |
567 | =over 8 | |
568 | ||
569 | =item * | |
570 | ||
571 | the "final sigma" (Greek) | |
572 | ||
573 | =item * | |
574 | ||
575 | anything to with locales (Lithuanian, Turkish, Azeri) | |
576 | ||
577 | =back | |
578 | ||
579 | See the Unicode Technical Report #21, Case Mappings, for more details. | |
ac1256e8 JH |
580 | |
581 | =item * | |
582 | ||
393fec97 GS |
583 | And finally, C<scalar reverse()> reverses by character rather than by byte. |
584 | ||
585 | =back | |
586 | ||
8cbd9a7a GS |
587 | =head2 Character encodings for input and output |
588 | ||
7221edc9 | 589 | See L<Encode>. |
8cbd9a7a | 590 | |
c29a771d | 591 | =head2 Unicode Regular Expression Support Level |
776f8809 JH |
592 | |
593 | The following list of Unicode regular expression support describes | |
594 | feature by feature the Unicode support implemented in Perl as of Perl | |
595 | 5.8.0. The "Level N" and the section numbers refer to the Unicode | |
596 | Technical Report 18, "Unicode Regular Expression Guidelines". | |
597 | ||
598 | =over 4 | |
599 | ||
600 | =item * | |
601 | ||
602 | Level 1 - Basic Unicode Support | |
603 | ||
604 | 2.1 Hex Notation - done [1] | |
3bfdc84c | 605 | Named Notation - done [2] |
776f8809 JH |
606 | 2.2 Categories - done [3][4] |
607 | 2.3 Subtraction - MISSING [5][6] | |
608 | 2.4 Simple Word Boundaries - done [7] | |
78d3e1bf | 609 | 2.5 Simple Loose Matches - done [8] |
776f8809 JH |
610 | 2.6 End of Line - MISSING [9][10] |
611 | ||
612 | [ 1] \x{...} | |
613 | [ 2] \N{...} | |
eb0cc9e3 | 614 | [ 3] . \p{...} \P{...} |
29bdacb8 | 615 | [ 4] now scripts (see UTR#24 Script Names) in addition to blocks |
776f8809 | 616 | [ 5] have negation |
29bdacb8 | 617 | [ 6] can use look-ahead to emulate subtraction (*) |
776f8809 | 618 | [ 7] include Letters in word characters |
e0f9d4a8 JH |
619 | [ 8] note that perl does Full casefolding in matching, not Simple: |
620 | for example U+1F88 is equivalent with U+1F000 U+03B9, | |
621 | not with 1F80. This difference matters for certain Greek | |
622 | capital letters with certain modifiers: the Full casefolding | |
623 | decomposes the letter, while the Simple casefolding would map | |
624 | it to a single character. | |
776f8809 | 625 | [ 9] see UTR#13 Unicode Newline Guidelines |
ec83e909 JH |
626 | [10] should do ^ and $ also on \x{85}, \x{2028} and \x{2029}) |
627 | (should also affect <>, $., and script line numbers) | |
3bfdc84c | 628 | (the \x{85}, \x{2028} and \x{2029} do match \s) |
7207e29d | 629 | |
dbe420b4 JH |
630 | (*) You can mimic class subtraction using lookahead. |
631 | For example, what TR18 might write as | |
29bdacb8 | 632 | |
dbe420b4 JH |
633 | [{Greek}-[{UNASSIGNED}]] |
634 | ||
635 | in Perl can be written as: | |
636 | ||
eb0cc9e3 JH |
637 | (?!\p{Unassigned})\p{InGreek} |
638 | (?=\p{Assigned})\p{InGreek} | |
dbe420b4 JH |
639 | |
640 | But in this particular example, you probably really want | |
641 | ||
642 | \p{Greek} | |
643 | ||
644 | which will match assigned characters known to be part of the Greek script. | |
29bdacb8 | 645 | |
776f8809 JH |
646 | =item * |
647 | ||
648 | Level 2 - Extended Unicode Support | |
649 | ||
650 | 3.1 Surrogates - MISSING | |
651 | 3.2 Canonical Equivalents - MISSING [11][12] | |
652 | 3.3 Locale-Independent Graphemes - MISSING [13] | |
653 | 3.4 Locale-Independent Words - MISSING [14] | |
654 | 3.5 Locale-Independent Loose Matches - MISSING [15] | |
655 | ||
656 | [11] see UTR#15 Unicode Normalization | |
657 | [12] have Unicode::Normalize but not integrated to regexes | |
658 | [13] have \X but at this level . should equal that | |
659 | [14] need three classes, not just \w and \W | |
660 | [15] see UTR#21 Case Mappings | |
661 | ||
662 | =item * | |
663 | ||
664 | Level 3 - Locale-Sensitive Support | |
665 | ||
666 | 4.1 Locale-Dependent Categories - MISSING | |
667 | 4.2 Locale-Dependent Graphemes - MISSING [16][17] | |
668 | 4.3 Locale-Dependent Words - MISSING | |
669 | 4.4 Locale-Dependent Loose Matches - MISSING | |
670 | 4.5 Locale-Dependent Ranges - MISSING | |
671 | ||
672 | [16] see UTR#10 Unicode Collation Algorithms | |
673 | [17] have Unicode::Collate but not integrated to regexes | |
674 | ||
675 | =back | |
676 | ||
c349b1b9 JH |
677 | =head2 Unicode Encodings |
678 | ||
679 | Unicode characters are assigned to I<code points> which are abstract | |
86bbd6d1 | 680 | numbers. To use these numbers various encodings are needed. |
c349b1b9 JH |
681 | |
682 | =over 4 | |
683 | ||
c29a771d | 684 | =item * |
5cb3728c RB |
685 | |
686 | UTF-8 | |
c349b1b9 | 687 | |
3e4dbfed JF |
688 | UTF-8 is a variable-length (1 to 6 bytes, current character allocations |
689 | require 4 bytes), byteorder independent encoding. For ASCII, UTF-8 is | |
690 | transparent (and we really do mean 7-bit ASCII, not another 8-bit encoding). | |
c349b1b9 | 691 | |
8c007b5a | 692 | The following table is from Unicode 3.2. |
05632f9a JH |
693 | |
694 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte | |
695 | ||
8c007b5a JH |
696 | U+0000..U+007F 00..7F |
697 | U+0080..U+07FF C2..DF 80..BF | |
05632f9a | 698 | U+0800..U+0FFF E0 A0..BF 80..BFÂ Â |
8c007b5a JH |
699 | U+1000..U+CFFF E1..EC 80..BF 80..BFÂ Â |
700 | U+D000..U+D7FF ED 80..9F 80..BFÂ Â | |
701 | U+D800..U+DFFF ******* ill-formed ******* | |
702 | U+E000..U+FFFF EE..EF 80..BF 80..BFÂ Â | |
05632f9a JH |
703 | U+10000..U+3FFFF F0 90..BF 80..BF 80..BF |
704 | U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF | |
705 | U+100000..U+10FFFF F4 80..8F 80..BF 80..BF | |
706 | ||
8c007b5a JH |
707 | Note the A0..BF in U+0800..U+0FFF, the 80..9F in U+D000...U+D7FF, |
708 | the 90..BF in U+10000..U+3FFFF, and the 80...8F in U+100000..U+10FFFF. | |
05632f9a JH |
709 | Or, another way to look at it, as bits: |
710 | ||
711 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte | |
712 | ||
713 | 0aaaaaaa 0aaaaaaa | |
714 | 00000bbbbbaaaaaa 110bbbbb 10aaaaaa | |
715 | ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa | |
716 | 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa | |
717 | ||
718 | As you can see, the continuation bytes all begin with C<10>, and the | |
8c007b5a | 719 | leading bits of the start byte tell how many bytes the are in the |
05632f9a JH |
720 | encoded character. |
721 | ||
c29a771d | 722 | =item * |
5cb3728c RB |
723 | |
724 | UTF-EBCDIC | |
dbe420b4 | 725 | |
fe854a6f | 726 | Like UTF-8, but EBCDIC-safe, as UTF-8 is ASCII-safe. |
dbe420b4 | 727 | |
c29a771d | 728 | =item * |
5cb3728c RB |
729 | |
730 | UTF-16, UTF-16BE, UTF16-LE, Surrogates, and BOMs (Byte Order Marks) | |
c349b1b9 | 731 | |
dbe420b4 JH |
732 | (The followings items are mostly for reference, Perl doesn't |
733 | use them internally.) | |
734 | ||
c349b1b9 JH |
735 | UTF-16 is a 2 or 4 byte encoding. The Unicode code points |
736 | 0x0000..0xFFFF are stored in two 16-bit units, and the code points | |
dbe420b4 | 737 | 0x010000..0x10FFFF in two 16-bit units. The latter case is |
c349b1b9 JH |
738 | using I<surrogates>, the first 16-bit unit being the I<high |
739 | surrogate>, and the second being the I<low surrogate>. | |
740 | ||
741 | Surrogates are code points set aside to encode the 0x01000..0x10FFFF | |
742 | range of Unicode code points in pairs of 16-bit units. The I<high | |
743 | surrogates> are the range 0xD800..0xDBFF, and the I<low surrogates> | |
744 | are the range 0xDC00..0xDFFFF. The surrogate encoding is | |
745 | ||
746 | $hi = ($uni - 0x10000) / 0x400 + 0xD800; | |
747 | $lo = ($uni - 0x10000) % 0x400 + 0xDC00; | |
748 | ||
749 | and the decoding is | |
750 | ||
751 | $uni = 0x10000 + ($hi - 0xD8000) * 0x400 + ($lo - 0xDC00); | |
752 | ||
feda178f JH |
753 | If you try to generate surrogates (for example by using chr()), you |
754 | will get a warning if warnings are turned on (C<-w> or C<use | |
755 | warnings;>) because those code points are not valid for a Unicode | |
756 | character. | |
9466bab6 | 757 | |
86bbd6d1 | 758 | Because of the 16-bitness, UTF-16 is byteorder dependent. UTF-16 |
c349b1b9 | 759 | itself can be used for in-memory computations, but if storage or |
86bbd6d1 | 760 | transfer is required, either UTF-16BE (Big Endian) or UTF-16LE |
c349b1b9 JH |
761 | (Little Endian) must be chosen. |
762 | ||
763 | This introduces another problem: what if you just know that your data | |
764 | is UTF-16, but you don't know which endianness? Byte Order Marks | |
765 | (BOMs) are a solution to this. A special character has been reserved | |
86bbd6d1 PN |
766 | in Unicode to function as a byte order marker: the character with the |
767 | code point 0xFEFF is the BOM. | |
042da322 | 768 | |
c349b1b9 JH |
769 | The trick is that if you read a BOM, you will know the byte order, |
770 | since if it was written on a big endian platform, you will read the | |
86bbd6d1 PN |
771 | bytes 0xFE 0xFF, but if it was written on a little endian platform, |
772 | you will read the bytes 0xFF 0xFE. (And if the originating platform | |
773 | was writing in UTF-8, you will read the bytes 0xEF 0xBB 0xBF.) | |
042da322 | 774 | |
86bbd6d1 PN |
775 | The way this trick works is that the character with the code point |
776 | 0xFFFE is guaranteed not to be a valid Unicode character, so the | |
777 | sequence of bytes 0xFF 0xFE is unambiguously "BOM, represented in | |
042da322 JH |
778 | little-endian format" and cannot be "0xFFFE, represented in big-endian |
779 | format". | |
c349b1b9 | 780 | |
c29a771d | 781 | =item * |
5cb3728c RB |
782 | |
783 | UTF-32, UTF-32BE, UTF32-LE | |
c349b1b9 JH |
784 | |
785 | The UTF-32 family is pretty much like the UTF-16 family, expect that | |
042da322 JH |
786 | the units are 32-bit, and therefore the surrogate scheme is not |
787 | needed. The BOM signatures will be 0x00 0x00 0xFE 0xFF for BE and | |
788 | 0xFF 0xFE 0x00 0x00 for LE. | |
c349b1b9 | 789 | |
c29a771d | 790 | =item * |
5cb3728c RB |
791 | |
792 | UCS-2, UCS-4 | |
c349b1b9 | 793 | |
86bbd6d1 PN |
794 | Encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit |
795 | encoding, UCS-4 is a 32-bit encoding. Unlike UTF-16, UCS-2 | |
796 | is not extensible beyond 0xFFFF, because it does not use surrogates. | |
c349b1b9 | 797 | |
c29a771d | 798 | =item * |
5cb3728c RB |
799 | |
800 | UTF-7 | |
c349b1b9 JH |
801 | |
802 | A seven-bit safe (non-eight-bit) encoding, useful if the | |
803 | transport/storage is not eight-bit safe. Defined by RFC 2152. | |
804 | ||
95a1a48b JH |
805 | =back |
806 | ||
bf0fa0b2 JH |
807 | =head2 Security Implications of Malformed UTF-8 |
808 | ||
809 | Unfortunately, the specification of UTF-8 leaves some room for | |
810 | interpretation of how many bytes of encoded output one should generate | |
811 | from one input Unicode character. Strictly speaking, one is supposed | |
812 | to always generate the shortest possible sequence of UTF-8 bytes, | |
feda178f JH |
813 | because otherwise there is potential for input buffer overflow at |
814 | the receiving end of a UTF-8 connection. Perl always generates the | |
815 | shortest length UTF-8, and with warnings on (C<-w> or C<use | |
816 | warnings;>) Perl will warn about non-shortest length UTF-8 (and other | |
817 | malformations, too, such as the surrogates, which are not real | |
818 | Unicode code points.) | |
bf0fa0b2 | 819 | |
c349b1b9 JH |
820 | =head2 Unicode in Perl on EBCDIC |
821 | ||
822 | The way Unicode is handled on EBCDIC platforms is still rather | |
86bbd6d1 | 823 | experimental. On such a platform, references to UTF-8 encoding in this |
c349b1b9 JH |
824 | document and elsewhere should be read as meaning UTF-EBCDIC as |
825 | specified in Unicode Technical Report 16 unless ASCII vs EBCDIC issues | |
826 | are specifically discussed. There is no C<utfebcdic> pragma or | |
86bbd6d1 PN |
827 | ":utfebcdic" layer, rather, "utf8" and ":utf8" are re-used to mean |
828 | the platform's "natural" 8-bit encoding of Unicode. See L<perlebcdic> | |
829 | for more discussion of the issues. | |
c349b1b9 | 830 | |
95a1a48b JH |
831 | =head2 Using Unicode in XS |
832 | ||
833 | If you want to handle Perl Unicode in XS extensions, you may find | |
90f968e0 | 834 | the following C APIs useful (see perlapi for details): |
95a1a48b JH |
835 | |
836 | =over 4 | |
837 | ||
838 | =item * | |
839 | ||
f1e62f77 AT |
840 | DO_UTF8(sv) returns true if the UTF8 flag is on and the bytes pragma |
841 | is not in effect. SvUTF8(sv) returns true is the UTF8 flag is on, the | |
842 | bytes pragma is ignored. The UTF8 flag being on does B<not> mean that | |
b31c5e31 AT |
843 | there are any characters of code points greater than 255 (or 127) in |
844 | the scalar, or that there even are any characters in the scalar. | |
845 | What the UTF8 flag means is that the sequence of octets in the | |
846 | representation of the scalar is the sequence of UTF-8 encoded | |
847 | code points of the characters of a string. The UTF8 flag being | |
848 | off means that each octet in this representation encodes a single | |
849 | character with codepoint 0..255 within the string. Perl's Unicode | |
850 | model is not to use UTF-8 until it's really necessary. | |
95a1a48b JH |
851 | |
852 | =item * | |
853 | ||
854 | uvuni_to_utf8(buf, chr) writes a Unicode character code point into a | |
cfc01aea | 855 | buffer encoding the code point as UTF-8, and returns a pointer |
95a1a48b JH |
856 | pointing after the UTF-8 bytes. |
857 | ||
858 | =item * | |
859 | ||
860 | utf8_to_uvuni(buf, lenp) reads UTF-8 encoded bytes from a buffer and | |
861 | returns the Unicode character code point (and optionally the length of | |
862 | the UTF-8 byte sequence). | |
863 | ||
864 | =item * | |
865 | ||
90f968e0 JH |
866 | utf8_length(start, end) returns the length of the UTF-8 encoded buffer |
867 | in characters. sv_len_utf8(sv) returns the length of the UTF-8 encoded | |
95a1a48b JH |
868 | scalar. |
869 | ||
870 | =item * | |
871 | ||
872 | sv_utf8_upgrade(sv) converts the string of the scalar to its UTF-8 | |
873 | encoded form. sv_utf8_downgrade(sv) does the opposite (if possible). | |
874 | sv_utf8_encode(sv) is like sv_utf8_upgrade but the UTF8 flag does not | |
875 | get turned on. sv_utf8_decode() does the opposite of sv_utf8_encode(). | |
13a6c0e0 JH |
876 | Note that none of these are to be used as general purpose encoding/decoding |
877 | interfaces: use Encode for that. sv_utf8_upgrade() is affected by the | |
878 | encoding pragma, but sv_utf8_downgrade() is not (since the encoding | |
879 | pragma is designed to be a one-way street). | |
95a1a48b JH |
880 | |
881 | =item * | |
882 | ||
90f968e0 JH |
883 | is_utf8_char(s) returns true if the pointer points to a valid UTF-8 |
884 | character. | |
95a1a48b JH |
885 | |
886 | =item * | |
887 | ||
888 | is_utf8_string(buf, len) returns true if the len bytes of the buffer | |
889 | are valid UTF-8. | |
890 | ||
891 | =item * | |
892 | ||
893 | UTF8SKIP(buf) will return the number of bytes in the UTF-8 encoded | |
894 | character in the buffer. UNISKIP(chr) will return the number of bytes | |
90f968e0 JH |
895 | required to UTF-8-encode the Unicode character code point. UTF8SKIP() |
896 | is useful for example for iterating over the characters of a UTF-8 | |
897 | encoded buffer; UNISKIP() is useful for example in computing | |
898 | the size required for a UTF-8 encoded buffer. | |
95a1a48b JH |
899 | |
900 | =item * | |
901 | ||
902 | utf8_distance(a, b) will tell the distance in characters between the | |
903 | two pointers pointing to the same UTF-8 encoded buffer. | |
904 | ||
905 | =item * | |
906 | ||
907 | utf8_hop(s, off) will return a pointer to an UTF-8 encoded buffer that | |
908 | is C<off> (positive or negative) Unicode characters displaced from the | |
90f968e0 JH |
909 | UTF-8 buffer C<s>. Be careful not to overstep the buffer: utf8_hop() |
910 | will merrily run off the end or the beginning if told to do so. | |
95a1a48b | 911 | |
d2cc3551 JH |
912 | =item * |
913 | ||
914 | pv_uni_display(dsv, spv, len, pvlim, flags) and sv_uni_display(dsv, | |
915 | ssv, pvlim, flags) are useful for debug output of Unicode strings and | |
90f968e0 JH |
916 | scalars. By default they are useful only for debug: they display |
917 | B<all> characters as hexadecimal code points, but with the flags | |
918 | UNI_DISPLAY_ISPRINT and UNI_DISPLAY_BACKSLASH you can make the output | |
919 | more readable. | |
d2cc3551 JH |
920 | |
921 | =item * | |
922 | ||
90f968e0 JH |
923 | ibcmp_utf8(s1, pe1, u1, l1, u1, s2, pe2, l2, u2) can be used to |
924 | compare two strings case-insensitively in Unicode. | |
925 | (For case-sensitive comparisons you can just use memEQ() and memNE() | |
926 | as usual.) | |
d2cc3551 | 927 | |
c349b1b9 JH |
928 | =back |
929 | ||
95a1a48b JH |
930 | For more information, see L<perlapi>, and F<utf8.c> and F<utf8.h> |
931 | in the Perl source code distribution. | |
932 | ||
c29a771d JH |
933 | =head1 BUGS |
934 | ||
935 | Use of locales with Unicode data may lead to odd results. Currently | |
936 | there is some attempt to apply 8-bit locale info to characters in the | |
937 | range 0..255, but this is demonstrably incorrect for locales that use | |
938 | characters above that range when mapped into Unicode. It will also | |
939 | tend to run slower. Avoidance of locales is strongly encouraged. | |
940 | ||
941 | Some functions are slower when working on UTF-8 encoded strings than | |
942 | on byte encoded strings. All functions that need to hop over | |
943 | characters such as length(), substr() or index() can work B<much> | |
944 | faster when the underlying data are byte-encoded. Witness the | |
945 | following benchmark: | |
946 | ||
947 | % perl -e ' | |
948 | use Benchmark; | |
949 | use strict; | |
950 | our $l = 10000; | |
951 | our $u = our $b = "x" x $l; | |
952 | substr($u,0,1) = "\x{100}"; | |
953 | timethese(-2,{ | |
954 | LENGTH_B => q{ length($b) }, | |
955 | LENGTH_U => q{ length($u) }, | |
956 | SUBSTR_B => q{ substr($b, $l/4, $l/2) }, | |
957 | SUBSTR_U => q{ substr($u, $l/4, $l/2) }, | |
958 | }); | |
959 | ' | |
960 | Benchmark: running LENGTH_B, LENGTH_U, SUBSTR_B, SUBSTR_U for at least 2 CPU seconds... | |
961 | LENGTH_B: 2 wallclock secs ( 2.36 usr + 0.00 sys = 2.36 CPU) @ 5649983.05/s (n=13333960) | |
962 | LENGTH_U: 2 wallclock secs ( 2.11 usr + 0.00 sys = 2.11 CPU) @ 12155.45/s (n=25648) | |
963 | SUBSTR_B: 3 wallclock secs ( 2.16 usr + 0.00 sys = 2.16 CPU) @ 374480.09/s (n=808877) | |
964 | SUBSTR_U: 2 wallclock secs ( 2.11 usr + 0.00 sys = 2.11 CPU) @ 6791.00/s (n=14329) | |
965 | ||
966 | The numbers show an incredible slowness on long UTF-8 strings and you | |
967 | should carefully avoid to use these functions within tight loops. For | |
968 | example if you want to iterate over characters, it is infinitely | |
969 | better to split into an array than to use substr, as the following | |
970 | benchmark shows: | |
971 | ||
972 | % perl -e ' | |
973 | use Benchmark; | |
974 | use strict; | |
975 | our $l = 10000; | |
976 | our $u = our $b = "x" x $l; | |
977 | substr($u,0,1) = "\x{100}"; | |
978 | timethese(-5,{ | |
979 | SPLIT_B => q{ for my $c (split //, $b){} }, | |
980 | SPLIT_U => q{ for my $c (split //, $u){} }, | |
981 | SUBSTR_B => q{ for my $i (0..length($b)-1){my $c = substr($b,$i,1);} }, | |
982 | SUBSTR_U => q{ for my $i (0..length($u)-1){my $c = substr($u,$i,1);} }, | |
983 | }); | |
984 | ' | |
985 | Benchmark: running SPLIT_B, SPLIT_U, SUBSTR_B, SUBSTR_U for at least 5 CPU seconds... | |
986 | SPLIT_B: 6 wallclock secs ( 5.29 usr + 0.00 sys = 5.29 CPU) @ 56.14/s (n=297) | |
987 | SPLIT_U: 5 wallclock secs ( 5.17 usr + 0.01 sys = 5.18 CPU) @ 55.21/s (n=286) | |
988 | SUBSTR_B: 5 wallclock secs ( 5.34 usr + 0.00 sys = 5.34 CPU) @ 123.22/s (n=658) | |
989 | SUBSTR_U: 7 wallclock secs ( 6.20 usr + 0.00 sys = 6.20 CPU) @ 0.81/s (n=5) | |
990 | ||
991 | You see, the algorithm based on substr() was faster with byte encoded | |
992 | data but it is pathologically slow with UTF-8 data. | |
993 | ||
393fec97 GS |
994 | =head1 SEE ALSO |
995 | ||
72ff2908 JH |
996 | L<perluniintro>, L<encoding>, L<Encode>, L<open>, L<utf8>, L<bytes>, |
997 | L<perlretut>, L<perlvar/"${^WIDE_SYSTEM_CALLS}"> | |
393fec97 GS |
998 | |
999 | =cut |