3 perlunicode - Unicode support in Perl
7 If you haven't already, before reading this document, you should become
8 familiar with both L<perlunitut> and L<perluniintro>.
10 Unicode aims to B<UNI>-fy the en-B<CODE>-ings of all the world's
11 character sets into a single Standard. For quite a few of the various
12 coding standards that existed when Unicode was first created, converting
13 from each to Unicode essentially meant adding a constant to each code
14 point in the original standard, and converting back meant just
15 subtracting that same constant. For ASCII and ISO-8859-1, the constant
16 is 0. For ISO-8859-5, (Cyrillic) the constant is 864; for Hebrew
17 (ISO-8859-8), it's 1488; Thai (ISO-8859-11), 3424; and so forth. This
18 made it easy to do the conversions, and facilitated the adoption of
21 And it worked; nowadays, those legacy standards are rarely used. Most
22 everyone uses Unicode.
24 Unicode is a comprehensive standard. It specifies many things outside
25 the scope of Perl, such as how to display sequences of characters. For
26 a full discussion of all aspects of Unicode, see
27 L<http://www.unicode.org>.
29 =head2 Important Caveats
31 Even though some of this section may not be understandable to you on
32 first reading, we think it's important enough to highlight some of the
33 gotchas before delving further, so here goes:
35 Unicode support is an extensive requirement. While Perl does not
36 implement the Unicode standard or the accompanying technical reports
37 from cover to cover, Perl does support many Unicode features.
39 Also, the use of Unicode may present security issues that aren't obvious.
40 Read L<Unicode Security Considerations|http://www.unicode.org/reports/tr36>.
44 =item Safest if you C<use feature 'unicode_strings'>
46 In order to preserve backward compatibility, Perl does not turn
47 on full internal Unicode support unless the pragma
48 L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
49 is specified. (This is automatically
50 selected if you S<C<use 5.012>> or higher.) Failure to do this can
51 trigger unexpected surprises. See L</The "Unicode Bug"> below.
53 This pragma doesn't affect I/O. Nor does it change the internal
54 representation of strings, only their interpretation. There are still
55 several places where Unicode isn't fully supported, such as in
58 =item Input and Output Layers
60 Use the C<:encoding(...)> layer to read from and write to
61 filehandles using the specified encoding. (See L<open>.)
63 =item You should convert your non-ASCII, non-UTF-8 Perl scripts to be
68 =item C<use utf8> still needed to enable L<UTF-8|/Unicode Encodings> in scripts
70 If your Perl script is itself encoded in L<UTF-8|/Unicode Encodings>,
71 the S<C<use utf8>> pragma must be explicitly included to enable
72 recognition of that (in string or regular expression literals, or in
73 identifier names). B<This is the only time when an explicit S<C<use
74 utf8>> is needed.> (See L<utf8>).
76 =item C<BOM>-marked scripts and L<UTF-16|/Unicode Encodings> scripts autodetected
78 However, if a Perl script begins with the Unicode C<BOM> (UTF-16LE,
79 UTF16-BE, or UTF-8), or if the script looks like non-C<BOM>-marked
80 UTF-16 of either endianness, Perl will correctly read in the script as
81 the appropriate Unicode encoding. (C<BOM>-less UTF-8 cannot be
82 effectively recognized or differentiated from ISO 8859-1 or other
87 =head2 Byte and Character Semantics
89 Before Unicode, most encodings used 8 bits (a single byte) to encode
90 each character. Thus a character was a byte, and a byte was a
91 character, and there could be only 256 or fewer possible characters.
92 "Byte Semantics" in the title of this section refers to
93 this behavior. There was no need to distinguish between "Byte" and
96 Then along comes Unicode which has room for over a million characters
97 (and Perl allows for even more). This means that a character may
98 require more than a single byte to represent it, and so the two terms
99 are no longer equivalent. What matter are the characters as whole
100 entities, and not usually the bytes that comprise them. That's what the
101 term "Character Semantics" in the title of this section refers to.
103 Perl had to change internally to decouple "bytes" from "characters".
104 It is important that you too change your ideas, if you haven't already,
105 so that "byte" and "character" no longer mean the same thing in your
108 The basic building block of Perl strings has always been a "character".
109 The changes basically come down to that the implementation no longer
110 thinks that a character is always just a single byte.
112 There are various things to note:
118 String handling functions, for the most part, continue to operate in
119 terms of characters. C<length()>, for example, returns the number of
120 characters in a string, just as before. But that number no longer is
121 necessarily the same as the number of bytes in the string (there may be
122 more bytes than characters). The other such functions include
123 C<chop()>, C<chomp()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
124 C<sort()>, C<sprintf()>, and C<write()>.
132 the bit-oriented C<vec>
138 the byte-oriented C<pack>/C<unpack> C<"C"> format
140 However, the C<W> specifier does operate on whole characters, as does the
145 some operators that interact with the platform's operating system
147 Operators dealing with filenames are examples.
151 when the functions are called from within the scope of the
152 S<C<L<use bytes|bytes>>> pragma
154 Likely, you should use this only for debugging anyway.
160 Strings--including hash keys--and regular expression patterns may
161 contain characters that have ordinal values larger than 255.
163 If you use a Unicode editor to edit your program, Unicode characters may
164 occur directly within the literal strings in UTF-8 encoding, or UTF-16.
165 (The former requires a C<BOM> or C<use utf8>, the latter requires a C<BOM>.)
167 L<perluniintro/Creating Unicode> gives other ways to place non-ASCII
168 characters in your strings.
172 The C<chr()> and C<ord()> functions work on whole characters.
176 Regular expressions match whole characters. For example, C<"."> matches
177 a whole character instead of only a single byte.
181 The C<tr///> operator translates whole characters. (Note that the
182 C<tr///CU> functionality has been removed. For similar functionality to
183 that, see C<pack('U0', ...)> and C<pack('C0', ...)>).
187 C<scalar reverse()> reverses by character rather than by byte.
191 The bit string operators, C<& | ^ ~> and (starting in v5.22)
192 C<&. |. ^. ~.> can operate on characters that don't fit into a byte.
193 However, the current behavior is likely to change. You should not use
194 these operators on strings that are encoded in UTF-8. If you're not
195 sure about the encoding of a string, downgrade it before using any of
196 these operators; you can use
197 L<C<utf8::utf8_downgrade()>|utf8/Utility functions>.
201 The bottom line is that Perl has always practiced "Character Semantics",
202 but with the advent of Unicode, that is now different than "Byte
205 =head2 ASCII Rules versus Unicode Rules
207 Before Unicode, when a character was a byte was a character,
208 Perl knew only about the 128 characters defined by ASCII, code points 0
209 through 127 (except for under S<C<use locale>>). That left the code
210 points 128 to 255 as unassigned, and available for whatever use a
211 program might want. The only semantics they have is their ordinal
212 numbers, and that they are members of none of the non-negative character
213 classes. None are considered to match C<\w> for example, but all match
216 Unicode, of course, assigns each of those code points a particular
217 meaning (along with ones above 255). To preserve backward
218 compatibility, Perl only uses the Unicode meanings when there is some
219 indication that Unicode is what is intended; otherwise the non-ASCII
220 code points remain treated as if they are unassigned.
222 Here are the ways that Perl knows that a string should be treated as
229 Within the scope of S<C<use utf8>>
231 If the whole program is Unicode (signified by using 8-bit B<U>nicode
232 B<T>ransformation B<F>ormat), then all strings within it must be
238 L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
240 This pragma was created so you can explicitly tell Perl that operations
241 executed within its scope are to use Unicode rules. More operations are
242 affected with newer perls. See L</The "Unicode Bug">.
246 Within the scope of S<C<use 5.012>> or higher
248 This implicitly turns on S<C<use feature 'unicode_strings'>>.
253 L<S<C<use locale 'not_characters'>>|perllocale/Unicode and UTF-8>,
254 or L<S<C<use locale>>|perllocale> and the current
255 locale is a UTF-8 locale.
257 The former is defined to imply Unicode handling; and the latter
258 indicates a Unicode locale, hence a Unicode interpretation of all
263 When the string contains a Unicode-only code point
265 Perl has never accepted code points above 255 without them being
266 Unicode, so their use implies Unicode for the whole string.
270 When the string contains a Unicode named code point C<\N{...}>
272 The C<\N{...}> construct explicitly refers to a Unicode code point,
273 even if it is one that is also in ASCII. Therefore the string
274 containing it must be Unicode.
278 When the string has come from an external source marked as
281 The L<C<-C>|perlrun/-C [numberE<sol>list]> command line option can
282 specify that certain inputs to the program are Unicode, and the values
283 of this can be read by your Perl code, see L<perlvar/"${^UNICODE}">.
285 =item * When the string has been upgraded to UTF-8
287 The function L<C<utf8::utf8_upgrade()>|utf8/Utility functions>
288 can be explicitly used to permanently (unless a subsequent
289 C<utf8::utf8_downgrade()> is called) cause a string to be treated as
292 =item * There are additional methods for regular expression patterns
294 A pattern that is compiled with the C<< /u >> or C<< /a >> modifiers is
295 treated as Unicode (though there are some restrictions with C<< /a >>).
296 Under the C<< /d >> and C<< /l >> modifiers, there are several other
297 indications for Unicode; see L<perlre/Character set modifiers>.
301 Note that all of the above are overridden within the scope of
302 C<L<use bytes|bytes>>; but you should be using this pragma only for
305 Note also that some interactions with the platform's operating system
306 never use Unicode rules.
308 When Unicode rules are in effect:
314 Case translation operators use the Unicode case translation tables.
316 Note that C<uc()>, or C<\U> in interpolated strings, translates to
317 uppercase, while C<ucfirst>, or C<\u> in interpolated strings,
318 translates to titlecase in languages that make the distinction (which is
319 equivalent to uppercase in languages without the distinction).
321 There is a CPAN module, C<L<Unicode::Casing>>, which allows you to
322 define your own mappings to be used in C<lc()>, C<lcfirst()>, C<uc()>,
323 C<ucfirst()>, and C<fc> (or their double-quoted string inlined versions
324 such as C<\U>). (Prior to Perl 5.16, this functionality was partially
325 provided in the Perl core, but suffered from a number of insurmountable
326 drawbacks, so the CPAN module was written instead.)
330 Character classes in regular expressions match based on the character
331 properties specified in the Unicode properties database.
333 C<\w> can be used to match a Japanese ideograph, for instance; and
334 C<[[:digit:]]> a Bengali number.
338 Named Unicode properties, scripts, and block ranges may be used (like
339 bracketed character classes) by using the C<\p{}> "matches property"
340 construct and the C<\P{}> negation, "doesn't match property".
342 See L</"Unicode Character Properties"> for more details.
344 You can define your own character properties and use them
345 in the regular expression with the C<\p{}> or C<\P{}> construct.
346 See L</"User-Defined Character Properties"> for more details.
350 =head2 Extended Grapheme Clusters (Logical characters)
352 Consider a character, say C<H>. It could appear with various marks around it,
353 such as an acute accent, or a circumflex, or various hooks, circles, arrows,
354 I<etc.>, above, below, to one side or the other, I<etc>. There are many
355 possibilities among the world's languages. The number of combinations is
356 astronomical, and if there were a character for each combination, it would
357 soon exhaust Unicode's more than a million possible characters. So Unicode
358 took a different approach: there is a character for the base C<H>, and a
359 character for each of the possible marks, and these can be variously combined
360 to get a final logical character. So a logical character--what appears to be a
361 single character--can be a sequence of more than one individual characters.
362 The Unicode standard calls these "extended grapheme clusters" (which
363 is an improved version of the no-longer much used "grapheme cluster");
364 Perl furnishes the C<\X> regular expression construct to match such
365 sequences in their entirety.
367 But Unicode's intent is to unify the existing character set standards and
368 practices, and several pre-existing standards have single characters that
369 mean the same thing as some of these combinations, like ISO-8859-1,
370 which has quite a few of them. For example, C<"LATIN CAPITAL LETTER E
371 WITH ACUTE"> was already in this standard when Unicode came along.
372 Unicode therefore added it to its repertoire as that single character.
373 But this character is considered by Unicode to be equivalent to the
374 sequence consisting of the character C<"LATIN CAPITAL LETTER E">
375 followed by the character C<"COMBINING ACUTE ACCENT">.
377 C<"LATIN CAPITAL LETTER E WITH ACUTE"> is called a "pre-composed"
378 character, and its equivalence with the "E" and the "COMBINING ACCENT"
379 sequence is called canonical equivalence. All pre-composed characters
380 are said to have a decomposition (into the equivalent sequence), and the
381 decomposition type is also called canonical. A string may be comprised
382 as much as possible of precomposed characters, or it may be comprised of
383 entirely decomposed characters. Unicode calls these respectively,
384 "Normalization Form Composed" (NFC) and "Normalization Form Decomposed".
385 The C<L<Unicode::Normalize>> module contains functions that convert
386 between the two. A string may also have both composed characters and
387 decomposed characters; this module can be used to make it all one or the
390 You may be presented with strings in any of these equivalent forms.
391 There is currently nothing in Perl 5 that ignores the differences. So
392 you'll have to specially hanlde it. The usual advice is to convert your
393 inputs to C<NFD> before processing further.
395 For more detailed information, see L<http://unicode.org/reports/tr15/>.
397 =head2 Unicode Character Properties
399 (The only time that Perl considers a sequence of individual code
400 points as a single logical character is in the C<\X> construct, already
401 mentioned above. Therefore "character" in this discussion means a single
404 Very nearly all Unicode character properties are accessible through
405 regular expressions by using the C<\p{}> "matches property" construct
406 and the C<\P{}> "doesn't match property" for its negation.
408 For instance, C<\p{Uppercase}> matches any single character with the Unicode
409 C<"Uppercase"> property, while C<\p{L}> matches any character with a
410 C<General_Category> of C<"L"> (letter) property (see
411 L</General_Category> below). Brackets are not
412 required for single letter property names, so C<\p{L}> is equivalent to C<\pL>.
414 More formally, C<\p{Uppercase}> matches any single character whose Unicode
415 C<Uppercase> property value is C<True>, and C<\P{Uppercase}> matches any character
416 whose C<Uppercase> property value is C<False>, and they could have been written as
417 C<\p{Uppercase=True}> and C<\p{Uppercase=False}>, respectively.
419 This formality is needed when properties are not binary; that is, if they can
420 take on more values than just C<True> and C<False>. For example, the
421 C<Bidi_Class> property (see L</"Bidirectional Character Types"> below),
422 can take on several different
423 values, such as C<Left>, C<Right>, C<Whitespace>, and others. To match these, one needs
424 to specify both the property name (C<Bidi_Class>), AND the value being
426 (C<Left>, C<Right>, I<etc.>). This is done, as in the examples above, by having the
427 two components separated by an equal sign (or interchangeably, a colon), like
428 C<\p{Bidi_Class: Left}>.
430 All Unicode-defined character properties may be written in these compound forms
431 of C<\p{I<property>=I<value>}> or C<\p{I<property>:I<value>}>, but Perl provides some
432 additional properties that are written only in the single form, as well as
433 single-form short-cuts for all binary properties and certain others described
434 below, in which you may omit the property name and the equals or colon
437 Most Unicode character properties have at least two synonyms (or aliases if you
438 prefer): a short one that is easier to type and a longer one that is more
439 descriptive and hence easier to understand. Thus the C<"L"> and
440 C<"Letter"> properties above are equivalent and can be used
441 interchangeably. Likewise, C<"Upper"> is a synonym for C<"Uppercase">,
442 and we could have written C<\p{Uppercase}> equivalently as C<\p{Upper}>.
443 Also, there are typically various synonyms for the values the property
444 can be. For binary properties, C<"True"> has 3 synonyms: C<"T">,
445 C<"Yes">, and C<"Y">; and C<"False"> has correspondingly C<"F">,
446 C<"No">, and C<"N">. But be careful. A short form of a value for one
447 property may not mean the same thing as the same short form for another.
448 Thus, for the C<L</General_Category>> property, C<"L"> means
449 C<"Letter">, but for the L<C<Bidi_Class>|/Bidirectional Character Types>
450 property, C<"L"> means C<"Left">. A complete list of properties and
451 synonyms is in L<perluniprops>.
453 Upper/lower case differences in property names and values are irrelevant;
454 thus C<\p{Upper}> means the same thing as C<\p{upper}> or even C<\p{UpPeR}>.
455 Similarly, you can add or subtract underscores anywhere in the middle of a
456 word, so that these are also equivalent to C<\p{U_p_p_e_r}>. And white space
457 is irrelevant adjacent to non-word characters, such as the braces and the equals
458 or colon separators, so C<\p{ Upper }> and C<\p{ Upper_case : Y }> are
459 equivalent to these as well. In fact, white space and even
460 hyphens can usually be added or deleted anywhere. So even C<\p{ Up-per case = Yes}> is
461 equivalent. All this is called "loose-matching" by Unicode. The few places
462 where stricter matching is used is in the middle of numbers, and in the Perl
463 extension properties that begin or end with an underscore. Stricter matching
464 cares about white space (except adjacent to non-word characters),
465 hyphens, and non-interior underscores.
467 You can also use negation in both C<\p{}> and C<\P{}> by introducing a caret
468 (C<^>) between the first brace and the property name: C<\p{^Tamil}> is
469 equal to C<\P{Tamil}>.
471 Almost all properties are immune to case-insensitive matching. That is,
472 adding a C</i> regular expression modifier does not change what they
473 match. There are two sets that are affected.
477 and C<Titlecase_Letter>,
478 all of which match C<Cased_Letter> under C</i> matching.
479 And the second set is
483 all of which match C<Cased> under C</i> matching.
484 This set also includes its subsets C<PosixUpper> and C<PosixLower> both
485 of which under C</i> match C<PosixAlpha>.
486 (The difference between these sets is that some things, such as Roman
487 numerals, come in both upper and lower case so they are C<Cased>, but
488 aren't considered letters, so they aren't C<Cased_Letter>'s.)
490 See L</Beyond Unicode code points> for special considerations when
491 matching Unicode properties against non-Unicode code points.
493 =head3 B<General_Category>
495 Every Unicode character is assigned a general category, which is the "most
496 usual categorization of a character" (from
497 L<http://www.unicode.org/reports/tr44>).
499 The compound way of writing these is like C<\p{General_Category=Number}>
500 (short: C<\p{gc:n}>). But Perl furnishes shortcuts in which everything up
501 through the equal or colon separator is omitted. So you can instead just write
504 Here are the short and long forms of the values the C<General Category> property
510 LC, L& Cased_Letter (that is: [\p{Ll}\p{Lu}\p{Lt}])
523 Nd Decimal_Number (also Digit)
527 P Punctuation (also Punct)
528 Pc Connector_Punctuation
532 Pi Initial_Punctuation
533 (may behave like Ps or Pe depending on usage)
535 (may behave like Ps or Pe depending on usage)
547 Zp Paragraph_Separator
550 Cc Control (also Cntrl)
556 Single-letter properties match all characters in any of the
557 two-letter sub-properties starting with the same letter.
558 C<LC> and C<L&> are special: both are aliases for the set consisting of everything matched by C<Ll>, C<Lu>, and C<Lt>.
560 =head3 B<Bidirectional Character Types>
562 Because scripts differ in their directionality (Hebrew and Arabic are
563 written right to left, for example) Unicode supplies a C<Bidi_Class> property.
564 Some of the values this property can have are:
569 LRE Left-to-Right Embedding
570 LRO Left-to-Right Override
573 RLE Right-to-Left Embedding
574 RLO Right-to-Left Override
575 PDF Pop Directional Format
577 ES European Separator
578 ET European Terminator
583 B Paragraph Separator
588 This property is always written in the compound form.
589 For example, C<\p{Bidi_Class:R}> matches characters that are normally
590 written right to left. Unlike the
591 C<L</General_Category>> property, this
592 property can have more values added in a future Unicode release. Those
593 listed above comprised the complete set for many Unicode releases, but
594 others were added in Unicode 6.3; you can always find what the
595 current ones are in L<perluniprops>. And
596 L<http://www.unicode.org/reports/tr9/> describes how to use them.
600 The world's languages are written in many different scripts. This sentence
601 (unless you're reading it in translation) is written in Latin, while Russian is
602 written in Cyrillic, and Greek is written in, well, Greek; Japanese mainly in
603 Hiragana or Katakana. There are many more.
605 The Unicode C<Script> and C<Script_Extensions> properties give what script a
606 given character is in. Either property can be specified with the
608 C<\p{Script=Hebrew}> (short: C<\p{sc=hebr}>), or
609 C<\p{Script_Extensions=Javanese}> (short: C<\p{scx=java}>).
610 In addition, Perl furnishes shortcuts for all
611 C<Script> property names. You can omit everything up through the equals
612 (or colon), and simply write C<\p{Latin}> or C<\P{Cyrillic}>.
613 (This is not true for C<Script_Extensions>, which is required to be
614 written in the compound form.)
616 The difference between these two properties involves characters that are
617 used in multiple scripts. For example the digits '0' through '9' are
618 used in many parts of the world. These are placed in a script named
619 C<Common>. Other characters are used in just a few scripts. For
620 example, the C<"KATAKANA-HIRAGANA DOUBLE HYPHEN"> is used in both Japanese
621 scripts, Katakana and Hiragana, but nowhere else. The C<Script>
622 property places all characters that are used in multiple scripts in the
623 C<Common> script, while the C<Script_Extensions> property places those
624 that are used in only a few scripts into each of those scripts; while
625 still using C<Common> for those used in many scripts. Thus both these
628 "0" =~ /\p{sc=Common}/ # Matches
629 "0" =~ /\p{scx=Common}/ # Matches
631 and only the first of these match:
633 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Common} # Matches
634 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Common} # No match
636 And only the last two of these match:
638 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Hiragana} # No match
639 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Katakana} # No match
640 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Hiragana} # Matches
641 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Katakana} # Matches
643 C<Script_Extensions> is thus an improved C<Script>, in which there are
644 fewer characters in the C<Common> script, and correspondingly more in
645 other scripts. It is new in Unicode version 6.0, and its data are likely
646 to change significantly in later releases, as things get sorted out.
647 New code should probably be using C<Script_Extensions> and not plain
650 (Actually, besides C<Common>, the C<Inherited> script, contains
651 characters that are used in multiple scripts. These are modifier
652 characters which inherit the script value
653 of the controlling character. Some of these are used in many scripts,
654 and so go into C<Inherited> in both C<Script> and C<Script_Extensions>.
655 Others are used in just a few scripts, so are in C<Inherited> in
656 C<Script>, but not in C<Script_Extensions>.)
658 It is worth stressing that there are several different sets of digits in
659 Unicode that are equivalent to 0-9 and are matchable by C<\d> in a
660 regular expression. If they are used in a single language only, they
661 are in that language's C<Script> and C<Script_Extension>. If they are
662 used in more than one script, they will be in C<sc=Common>, but only
663 if they are used in many scripts should they be in C<scx=Common>.
665 A complete list of scripts and their shortcuts is in L<perluniprops>.
667 =head3 B<Use of the C<"Is"> Prefix>
669 For backward compatibility (with Perl 5.6), all properties writable
670 without using the compound form mentioned
671 so far may have C<Is> or C<Is_> prepended to their name, so C<\P{Is_Lu}>, for
672 example, is equal to C<\P{Lu}>, and C<\p{IsScript:Arabic}> is equal to
677 In addition to B<scripts>, Unicode also defines B<blocks> of
678 characters. The difference between scripts and blocks is that the
679 concept of scripts is closer to natural languages, while the concept
680 of blocks is more of an artificial grouping based on groups of Unicode
681 characters with consecutive ordinal values. For example, the C<"Basic Latin">
682 block is all the characters whose ordinals are between 0 and 127, inclusive; in
683 other words, the ASCII characters. The C<"Latin"> script contains some letters
684 from this as well as several other blocks, like C<"Latin-1 Supplement">,
685 C<"Latin Extended-A">, I<etc.>, but it does not contain all the characters from
686 those blocks. It does not, for example, contain the digits 0-9, because
687 those digits are shared across many scripts, and hence are in the
690 For more about scripts versus blocks, see UAX#24 "Unicode Script Property":
691 L<http://www.unicode.org/reports/tr24>
693 The C<Script> or C<Script_Extensions> properties are likely to be the
694 ones you want to use when processing
695 natural language; the C<Block> property may occasionally be useful in working
696 with the nuts and bolts of Unicode.
698 Block names are matched in the compound form, like C<\p{Block: Arrows}> or
699 C<\p{Blk=Hebrew}>. Unlike most other properties, only a few block names have a
700 Unicode-defined short name.
702 Perl also defines single form synonyms for the block property in cases
703 where these do not conflict with something else. But don't use any of
704 these, because they are unstable. Since these are Perl extensions, they
705 are subordinate to official Unicode property names; Unicode doesn't know
706 nor care about Perl's extensions. It may happen that a name that
707 currently means the Perl extension will later be changed without warning
708 to mean a different Unicode property in a future version of the perl
709 interpreter that uses a later Unicode release, and your code would no
710 longer work. The extensions are mentioned here for completeness: Take
711 the block name and prefix it with one of: C<In> (for example
712 C<\p{Blk=Arrows}> can currently be written as C<\p{In_Arrows}>); or
713 sometimes C<Is> (like C<\p{Is_Arrows}>); or sometimes no prefix at all
714 (C<\p{Arrows}>). As of this writing (Unicode 8.0) there are no
715 conflicts with using the C<In_> prefix, but there are plenty with the
716 other two forms. For example, C<\p{Is_Hebrew}> and C<\p{Hebrew}> mean
717 C<\p{Script=Hebrew}> which is NOT the same thing as C<\p{Blk=Hebrew}>. Our
718 advice used to be to use the C<In_> prefix as a single form way of
719 specifying a block. But Unicode 8.0 added properties whose names begin
720 with C<In>, and it's now clear that it's only luck that's so far
721 prevented a conflict. Using C<In> is only marginally less typing than
722 C<Blk:>, and the latter's meaning is clearer anyway, and guaranteed to
723 never conflict. So don't take chances. Use C<\p{Blk=foo}> for new
724 code. And be sure that block is what you really really want to do. In
725 most cases scripts are what you want instead.
727 A complete list of blocks is in L<perluniprops>.
729 =head3 B<Other Properties>
731 There are many more properties than the very basic ones described here.
732 A complete list is in L<perluniprops>.
734 Unicode defines all its properties in the compound form, so all single-form
735 properties are Perl extensions. Most of these are just synonyms for the
736 Unicode ones, but some are genuine extensions, including several that are in
737 the compound form. And quite a few of these are actually recommended by Unicode
738 (in L<http://www.unicode.org/reports/tr18>).
740 This section gives some details on all extensions that aren't just
741 synonyms for compound-form Unicode properties
742 (for those properties, you'll have to refer to the
743 L<Unicode Standard|http://www.unicode.org/reports/tr44>.
749 This matches every possible code point. It is equivalent to C<qr/./s>.
750 Unlike all the other non-user-defined C<\p{}> property matches, no
751 warning is ever generated if this is property is matched against a
752 non-Unicode code point (see L</Beyond Unicode code points> below).
754 =item B<C<\p{Alnum}>>
756 This matches any C<\p{Alphabetic}> or C<\p{Decimal_Number}> character.
760 This matches any of the 1_114_112 Unicode code points. It is a synonym
763 =item B<C<\p{ASCII}>>
765 This matches any of the 128 characters in the US-ASCII character set,
766 which is a subset of Unicode.
768 =item B<C<\p{Assigned}>>
770 This matches any assigned code point; that is, any code point whose L<general
771 category|/General_Category> is not C<Unassigned> (or equivalently, not C<Cn>).
773 =item B<C<\p{Blank}>>
775 This is the same as C<\h> and C<\p{HorizSpace}>: A character that changes the
776 spacing horizontally.
778 =item B<C<\p{Decomposition_Type: Non_Canonical}>> (Short: C<\p{Dt=NonCanon}>)
780 Matches a character that has a non-canonical decomposition.
782 The L</Extended Grapheme Clusters (Logical characters)> section above
783 talked about canonical decompositions. However, many more characters
784 have a different type of decomposition, a "compatible" or
785 "non-canonical" decomposition. The sequences that form these
786 decompositions are not considered canonically equivalent to the
787 pre-composed character. An example is the C<"SUPERSCRIPT ONE">. It is
788 somewhat like a regular digit 1, but not exactly; its decomposition into
789 the digit 1 is called a "compatible" decomposition, specifically a
790 "super" decomposition. There are several such compatibility
791 decompositions (see L<http://www.unicode.org/reports/tr44>), including
792 one called "compat", which means some miscellaneous type of
793 decomposition that doesn't fit into the other decomposition categories
794 that Unicode has chosen.
796 Note that most Unicode characters don't have a decomposition, so their
797 decomposition type is C<"None">.
799 For your convenience, Perl has added the C<Non_Canonical> decomposition
800 type to mean any of the several compatibility decompositions.
802 =item B<C<\p{Graph}>>
804 Matches any character that is graphic. Theoretically, this means a character
805 that on a printer would cause ink to be used.
807 =item B<C<\p{HorizSpace}>>
809 This is the same as C<\h> and C<\p{Blank}>: a character that changes the
810 spacing horizontally.
814 This is a synonym for C<\p{Present_In=*}>
816 =item B<C<\p{PerlSpace}>>
818 This is the same as C<\s>, restricted to ASCII, namely C<S<[ \f\n\r\t]>>
819 and starting in Perl v5.18, a vertical tab.
821 Mnemonic: Perl's (original) space
823 =item B<C<\p{PerlWord}>>
825 This is the same as C<\w>, restricted to ASCII, namely C<[A-Za-z0-9_]>
827 Mnemonic: Perl's (original) word.
829 =item B<C<\p{Posix...}>>
831 There are several of these, which are equivalents, using the C<\p{}>
832 notation, for Posix classes and are described in
833 L<perlrecharclass/POSIX Character Classes>.
835 =item B<C<\p{Present_In: *}>> (Short: C<\p{In=*}>)
837 This property is used when you need to know in what Unicode version(s) a
840 The "*" above stands for some two digit Unicode version number, such as
841 C<1.1> or C<4.0>; or the "*" can also be C<Unassigned>. This property will
842 match the code points whose final disposition has been settled as of the
843 Unicode release given by the version number; C<\p{Present_In: Unassigned}>
844 will match those code points whose meaning has yet to be assigned.
846 For example, C<U+0041> C<"LATIN CAPITAL LETTER A"> was present in the very first
847 Unicode release available, which is C<1.1>, so this property is true for all
848 valid "*" versions. On the other hand, C<U+1EFF> was not assigned until version
849 5.1 when it became C<"LATIN SMALL LETTER Y WITH LOOP">, so the only "*" that
850 would match it are 5.1, 5.2, and later.
852 Unicode furnishes the C<Age> property from which this is derived. The problem
853 with Age is that a strict interpretation of it (which Perl takes) has it
854 matching the precise release a code point's meaning is introduced in. Thus
855 C<U+0041> would match only 1.1; and C<U+1EFF> only 5.1. This is not usually what
858 Some non-Perl implementations of the Age property may change its meaning to be
859 the same as the Perl C<Present_In> property; just be aware of that.
861 Another confusion with both these properties is that the definition is not
862 that the code point has been I<assigned>, but that the meaning of the code point
863 has been I<determined>. This is because 66 code points will always be
864 unassigned, and so the C<Age> for them is the Unicode version in which the decision
865 to make them so was made. For example, C<U+FDD0> is to be permanently
866 unassigned to a character, and the decision to do that was made in version 3.1,
867 so C<\p{Age=3.1}> matches this character, as also does C<\p{Present_In: 3.1}> and up.
869 =item B<C<\p{Print}>>
871 This matches any character that is graphical or blank, except controls.
873 =item B<C<\p{SpacePerl}>>
875 This is the same as C<\s>, including beyond ASCII.
877 Mnemonic: Space, as modified by Perl. (It doesn't include the vertical tab
878 until v5.18, which both the Posix standard and Unicode consider white space.)
880 =item B<C<\p{Title}>> and B<C<\p{Titlecase}>>
882 Under case-sensitive matching, these both match the same code points as
883 C<\p{General Category=Titlecase_Letter}> (C<\p{gc=lt}>). The difference
884 is that under C</i> caseless matching, these match the same as
885 C<\p{Cased}>, whereas C<\p{gc=lt}> matches C<\p{Cased_Letter>).
887 =item B<C<\p{Unicode}>>
889 This matches any of the 1_114_112 Unicode code points.
892 =item B<C<\p{VertSpace}>>
894 This is the same as C<\v>: A character that changes the spacing vertically.
898 This is the same as C<\w>, including over 100_000 characters beyond ASCII.
900 =item B<C<\p{XPosix...}>>
902 There are several of these, which are the standard Posix classes
903 extended to the full Unicode range. They are described in
904 L<perlrecharclass/POSIX Character Classes>.
909 =head2 User-Defined Character Properties
911 You can define your own binary character properties by defining subroutines
912 whose names begin with C<"In"> or C<"Is">. (The experimental feature
913 L<perlre/(?[ ])> provides an alternative which allows more complex
914 definitions.) The subroutines can be defined in any
915 package. The user-defined properties can be used in the regular expression
916 C<\p{}> and C<\P{}> constructs; if you are using a user-defined property from a
917 package other than the one you are in, you must specify its package in the
918 C<\p{}> or C<\P{}> construct.
920 # assuming property Is_Foreign defined in Lang::
921 package main; # property package name required
922 if ($txt =~ /\p{Lang::IsForeign}+/) { ... }
924 package Lang; # property package name not required
925 if ($txt =~ /\p{IsForeign}+/) { ... }
928 Note that the effect is compile-time and immutable once defined.
929 However, the subroutines are passed a single parameter, which is 0 if
930 case-sensitive matching is in effect and non-zero if caseless matching
931 is in effect. The subroutine may return different values depending on
932 the value of the flag, and one set of values will immutably be in effect
933 for all case-sensitive matches, and the other set for all case-insensitive
936 Note that if the regular expression is tainted, then Perl will die rather
937 than calling the subroutine when the name of the subroutine is
938 determined by the tainted data.
940 The subroutines must return a specially-formatted string, with one
941 or more newline-separated lines. Each line must be one of the following:
947 A single hexadecimal number denoting a code point to include.
951 Two hexadecimal numbers separated by horizontal whitespace (space or
952 tabular characters) denoting a range of code points to include.
956 Something to include, prefixed by C<"+">: a built-in character
957 property (prefixed by C<"utf8::">) or a fully qualified (including package
958 name) user-defined character property,
959 to represent all the characters in that property; two hexadecimal code
960 points for a range; or a single hexadecimal code point.
964 Something to exclude, prefixed by C<"-">: an existing character
965 property (prefixed by C<"utf8::">) or a fully qualified (including package
966 name) user-defined character property,
967 to represent all the characters in that property; two hexadecimal code
968 points for a range; or a single hexadecimal code point.
972 Something to negate, prefixed C<"!">: an existing character
973 property (prefixed by C<"utf8::">) or a fully qualified (including package
974 name) user-defined character property,
975 to represent all the characters in that property; two hexadecimal code
976 points for a range; or a single hexadecimal code point.
980 Something to intersect with, prefixed by C<"&">: an existing character
981 property (prefixed by C<"utf8::">) or a fully qualified (including package
982 name) user-defined character property,
983 for all the characters except the characters in the property; two
984 hexadecimal code points for a range; or a single hexadecimal code point.
988 For example, to define a property that covers both the Japanese
989 syllabaries (hiragana and katakana), you can define
998 Imagine that the here-doc end marker is at the beginning of the line.
999 Now you can use C<\p{InKana}> and C<\P{InKana}>.
1001 You could also have used the existing block property names:
1010 Suppose you wanted to match only the allocated characters,
1011 not the raw block ranges: in other words, you want to remove
1012 the unassigned characters:
1022 The negation is useful for defining (surprise!) negated classes.
1032 This will match all non-Unicode code points, since every one of them is
1033 not in Kana. You can use intersection to exclude these, if desired, as
1034 this modified example shows:
1045 C<&utf8::Any> must be the last line in the definition.
1047 Intersection is used generally for getting the common characters matched
1048 by two (or more) classes. It's important to remember not to use C<"&"> for
1049 the first set; that would be intersecting with nothing, resulting in an
1052 Unlike non-user-defined C<\p{}> property matches, no warning is ever
1053 generated if these properties are matched against a non-Unicode code
1054 point (see L</Beyond Unicode code points> below).
1056 =head2 User-Defined Case Mappings (for serious hackers only)
1058 B<This feature has been removed as of Perl 5.16.>
1059 The CPAN module C<L<Unicode::Casing>> provides better functionality without
1060 the drawbacks that this feature had. If you are using a Perl earlier
1061 than 5.16, this feature was most fully documented in the 5.14 version of
1063 L<http://perldoc.perl.org/5.14.0/perlunicode.html#User-Defined-Case-Mappings-%28for-serious-hackers-only%29>
1065 =head2 Character Encodings for Input and Output
1069 =head2 Unicode Regular Expression Support Level
1071 The following list of Unicode supported features for regular expressions describes
1072 all features currently directly supported by core Perl. The references to "Level N"
1073 and the section numbers refer to the Unicode Technical Standard #18,
1074 "Unicode Regular Expressions", version 13, from August 2008.
1080 Level 1 - Basic Unicode Support
1082 RL1.1 Hex Notation - done [1]
1083 RL1.2 Properties - done [2][3]
1084 RL1.2a Compatibility Properties - done [4]
1085 RL1.3 Subtraction and Intersection - experimental [5]
1086 RL1.4 Simple Word Boundaries - done [6]
1087 RL1.5 Simple Loose Matches - done [7]
1088 RL1.6 Line Boundaries - MISSING [8][9]
1089 RL1.7 Supplementary Code Points - done [10]
1093 =item [1] C<\N{U+...}> and C<\x{...}>
1095 =item [2] C<\p{...}> C<\P{...}>
1097 =item [3] supports not only minimal list, but all Unicode character
1098 properties (see Unicode Character Properties above)
1100 =item [4] C<\d> C<\D> C<\s> C<\S> C<\w> C<\W> C<\X> C<[:I<prop>:]>
1103 =item [5] The experimental feature starting in v5.18 C<"(?[...])"> accomplishes
1106 See L<perlre/(?[ ])>. If you don't want to use an experimental
1107 feature, you can use one of the following:
1113 Regular expression look-ahead
1115 You can mimic class subtraction using lookahead.
1116 For example, what UTS#18 might write as
1118 [{Block=Greek}-[{UNASSIGNED}]]
1120 in Perl can be written as:
1122 (?!\p{Unassigned})\p{Block=Greek}
1123 (?=\p{Assigned})\p{Block=Greek}
1125 But in this particular example, you probably really want
1129 which will match assigned characters known to be part of the Greek script.
1133 CPAN module C<L<Unicode::Regex::Set>>
1135 It does implement the full UTS#18 grouping, intersection, union, and
1136 removal (subtraction) syntax.
1140 L</"User-Defined Character Properties">
1142 C<"+"> for union, C<"-"> for removal (set-difference), C<"&"> for intersection
1146 =item [6] C<\b> C<\B>
1149 Note that Perl does Full case-folding in matching, not Simple:
1151 For example C<U+1F88> is equivalent to C<U+1F00 U+03B9>, instead of just
1152 C<U+1F80>. This difference matters mainly for certain Greek capital
1153 letters with certain modifiers: the Full case-folding decomposes the
1154 letter, while the Simple case-folding would map it to a single
1158 Perl treats C<\n> as the start- and end-line delimiter. Unicode
1159 specifies more characters that should be so-interpreted.
1170 C<^> and C<$> in regular expression patterns are supposed to match all
1172 These characters also don't, but should, affect C<< <> >> C<$.>, and
1173 script line numbers.
1175 Also, lines should not be split within C<CRLF> (i.e. there is no
1176 empty line between C<\r> and C<\n>). For C<CRLF>, try the C<:crlf>
1177 layer (see L<PerlIO>).
1179 =item [9] But C<L<Unicode::LineBreak>> is available.
1181 This module supplies line breaking conformant with
1182 L<UAX#14 "Unicode Line Breaking Algorithm"|http://www.unicode.org/reports/tr14>.
1185 UTF-8/UTF-EBDDIC used in Perl allows not only C<U+10000> to
1186 C<U+10FFFF> but also beyond C<U+10FFFF>
1192 Level 2 - Extended Unicode Support
1194 RL2.1 Canonical Equivalents - MISSING [10][11]
1195 RL2.2 Default Grapheme Clusters - MISSING [12]
1196 RL2.3 Default Word Boundaries - DONE [14]
1197 RL2.4 Default Loose Matches - MISSING [15]
1198 RL2.5 Name Properties - DONE
1199 RL2.6 Wildcard Properties - MISSING
1201 [10] see UAX#15 "Unicode Normalization Forms"
1202 [11] have Unicode::Normalize but not integrated to regexes
1203 [12] have \X and \b{gcb} but we don't have a "Grapheme Cluster
1205 [14] see UAX#29, Word Boundaries
1206 [15] This is covered in Chapter 3.13 (in Unicode 6.0)
1210 Level 3 - Tailored Support
1212 RL3.1 Tailored Punctuation - MISSING
1213 RL3.2 Tailored Grapheme Clusters - MISSING [17][18]
1214 RL3.3 Tailored Word Boundaries - MISSING
1215 RL3.4 Tailored Loose Matches - MISSING
1216 RL3.5 Tailored Ranges - MISSING
1217 RL3.6 Context Matching - MISSING [19]
1218 RL3.7 Incremental Matches - MISSING
1219 ( RL3.8 Unicode Set Sharing )
1220 RL3.9 Possible Match Sets - MISSING
1221 RL3.10 Folded Matching - MISSING [20]
1222 RL3.11 Submatchers - MISSING
1224 [17] see UAX#10 "Unicode Collation Algorithms"
1225 [18] have Unicode::Collate but not integrated to regexes
1226 [19] have (?<=x) and (?=x), but look-aheads or look-behinds
1227 should see outside of the target substring
1228 [20] need insensitive matching for linguistic features other
1229 than case; for example, hiragana to katakana, wide and
1230 narrow, simplified Han to traditional Han (see UTR#30
1231 "Character Foldings")
1235 =head2 Unicode Encodings
1237 Unicode characters are assigned to I<code points>, which are abstract
1238 numbers. To use these numbers, various encodings are needed.
1246 UTF-8 is a variable-length (1 to 4 bytes), byte-order independent
1247 encoding. In most of Perl's documentation, including elsewhere in this
1248 document, the term "UTF-8" means also "UTF-EBCDIC". But in this section,
1249 "UTF-8" refers only to the encoding used on ASCII platforms. It is a
1250 superset of 7-bit US-ASCII, so anything encoded in ASCII has the
1251 identical representation when encoded in UTF-8.
1253 The following table is from Unicode 3.2.
1255 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
1257 U+0000..U+007F 00..7F
1258 U+0080..U+07FF * C2..DF 80..BF
1259 U+0800..U+0FFF E0 * A0..BF 80..BF
1260 U+1000..U+CFFF E1..EC 80..BF 80..BF
1261 U+D000..U+D7FF ED 80..9F 80..BF
1262 U+D800..U+DFFF +++++ utf16 surrogates, not legal utf8 +++++
1263 U+E000..U+FFFF EE..EF 80..BF 80..BF
1264 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
1265 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
1266 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
1268 Note the gaps marked by "*" before several of the byte entries above. These are
1269 caused by legal UTF-8 avoiding non-shortest encodings: it is technically
1270 possible to UTF-8-encode a single code point in different ways, but that is
1271 explicitly forbidden, and the shortest possible encoding should always be used
1272 (and that is what Perl does).
1274 Another way to look at it is via bits:
1276 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
1279 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
1280 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
1281 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
1283 As you can see, the continuation bytes all begin with C<"10">, and the
1284 leading bits of the start byte tell how many bytes there are in the
1287 The original UTF-8 specification allowed up to 6 bytes, to allow
1288 encoding of numbers up to C<0x7FFF_FFFF>. Perl continues to allow those,
1289 and has extended that up to 13 bytes to encode code points up to what
1290 can fit in a 64-bit word. However, Perl will warn if you output any of
1291 these as being non-portable; and under strict UTF-8 input protocols,
1298 Like UTF-8, but EBCDIC-safe, in the way that UTF-8 is ASCII-safe.
1299 This means that all the basic characters (which includes all
1300 those that have ASCII equivalents (like C<"A">, C<"0">, C<"%">, I<etc.>)
1301 are the same in both EBCDIC and UTF-EBCDIC.)
1303 UTF-EBCDIC is used on EBCDIC platforms. The largest Unicode code points
1304 take 5 bytes to represent (instead of 4 in UTF-8), and Perl extends it
1305 to a maximum of 7 bytes to encode pode points up to what can fit in a
1306 32-bit word (instead of 13 bytes and a 64-bit word in UTF-8).
1310 UTF-16, UTF-16BE, UTF-16LE, Surrogates, and C<BOM>'s (Byte Order Marks)
1312 The followings items are mostly for reference and general Unicode
1313 knowledge, Perl doesn't use these constructs internally.
1315 Like UTF-8, UTF-16 is a variable-width encoding, but where
1316 UTF-8 uses 8-bit code units, UTF-16 uses 16-bit code units.
1317 All code points occupy either 2 or 4 bytes in UTF-16: code points
1318 C<U+0000..U+FFFF> are stored in a single 16-bit unit, and code
1319 points C<U+10000..U+10FFFF> in two 16-bit units. The latter case is
1320 using I<surrogates>, the first 16-bit unit being the I<high
1321 surrogate>, and the second being the I<low surrogate>.
1323 Surrogates are code points set aside to encode the C<U+10000..U+10FFFF>
1324 range of Unicode code points in pairs of 16-bit units. The I<high
1325 surrogates> are the range C<U+D800..U+DBFF> and the I<low surrogates>
1326 are the range C<U+DC00..U+DFFF>. The surrogate encoding is
1328 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
1329 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
1333 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
1335 Because of the 16-bitness, UTF-16 is byte-order dependent. UTF-16
1336 itself can be used for in-memory computations, but if storage or
1337 transfer is required either UTF-16BE (big-endian) or UTF-16LE
1338 (little-endian) encodings must be chosen.
1340 This introduces another problem: what if you just know that your data
1341 is UTF-16, but you don't know which endianness? Byte Order Marks, or
1342 C<BOM>'s, are a solution to this. A special character has been reserved
1343 in Unicode to function as a byte order marker: the character with the
1344 code point C<U+FEFF> is the C<BOM>.
1346 The trick is that if you read a C<BOM>, you will know the byte order,
1347 since if it was written on a big-endian platform, you will read the
1348 bytes C<0xFE 0xFF>, but if it was written on a little-endian platform,
1349 you will read the bytes C<0xFF 0xFE>. (And if the originating platform
1350 was writing in ASCII platform UTF-8, you will read the bytes
1353 The way this trick works is that the character with the code point
1354 C<U+FFFE> is not supposed to be in input streams, so the
1355 sequence of bytes C<0xFF 0xFE> is unambiguously "C<BOM>, represented in
1356 little-endian format" and cannot be C<U+FFFE>, represented in big-endian
1359 Surrogates have no meaning in Unicode outside their use in pairs to
1360 represent other code points. However, Perl allows them to be
1361 represented individually internally, for example by saying
1362 C<chr(0xD801)>, so that all code points, not just those valid for open
1364 representable. Unicode does define semantics for them, such as their
1365 C<L</General_Category>> is C<"Cs">. But because their use is somewhat dangerous,
1366 Perl will warn (using the warning category C<"surrogate">, which is a
1367 sub-category of C<"utf8">) if an attempt is made
1368 to do things like take the lower case of one, or match
1369 case-insensitively, or to output them. (But don't try this on Perls
1374 UTF-32, UTF-32BE, UTF-32LE
1376 The UTF-32 family is pretty much like the UTF-16 family, except that
1377 the units are 32-bit, and therefore the surrogate scheme is not
1378 needed. UTF-32 is a fixed-width encoding. The C<BOM> signatures are
1379 C<0x00 0x00 0xFE 0xFF> for BE and C<0xFF 0xFE 0x00 0x00> for LE.
1385 Legacy, fixed-width encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
1386 encoding. Unlike UTF-16, UCS-2 is not extensible beyond C<U+FFFF>,
1387 because it does not use surrogates. UCS-4 is a 32-bit encoding,
1388 functionally identical to UTF-32 (the difference being that
1389 UCS-4 forbids neither surrogates nor code points larger than C<0x10_FFFF>).
1395 A seven-bit safe (non-eight-bit) encoding, which is useful if the
1396 transport or storage is not eight-bit safe. Defined by RFC 2152.
1400 =head2 Noncharacter code points
1402 66 code points are set aside in Unicode as "noncharacter code points".
1403 These all have the C<Unassigned> (C<Cn>) C<L</General_Category>>, and
1404 no character will ever be assigned to any of them. They are the 32 code
1405 points between C<U+FDD0> and C<U+FDEF> inclusive, and the 34 code
1416 Until Unicode 7.0, the noncharacters were "B<forbidden> for use in open
1417 interchange of Unicode text data", so that code that processed those
1418 streams could use these code points as sentinels that could be mixed in
1419 with character data, and would always be distinguishable from that data.
1420 (Emphasis above and in the next paragraph are added in this document.)
1422 Unicode 7.0 changed the wording so that they are "B<not recommended> for
1423 use in open interchange of Unicode text data". The 7.0 Standard goes on
1428 "If a noncharacter is received in open interchange, an application is
1429 not required to interpret it in any way. It is good practice, however,
1430 to recognize it as a noncharacter and to take appropriate action, such
1431 as replacing it with C<U+FFFD> replacement character, to indicate the
1432 problem in the text. It is not recommended to simply delete
1433 noncharacter code points from such text, because of the potential
1434 security issues caused by deleting uninterpreted characters. (See
1435 conformance clause C7 in Section 3.2, Conformance Requirements, and
1436 L<Unicode Technical Report #36, "Unicode Security
1437 Considerations"|http://www.unicode.org/reports/tr36/#Substituting_for_Ill_Formed_Subsequences>)."
1441 This change was made because it was found that various commercial tools
1442 like editors, or for things like source code control, had been written
1443 so that they would not handle program files that used these code points,
1444 effectively precluding their use almost entirely! And that was never
1445 the intent. They've always been meant to be usable within an
1446 application, or cooperating set of applications, at will.
1448 If you're writing code, such as an editor, that is supposed to be able
1449 to handle any Unicode text data, then you shouldn't be using these code
1450 points yourself, and instead allow them in the input. If you need
1451 sentinels, they should instead be something that isn't legal Unicode.
1452 For UTF-8 data, you can use the bytes 0xC1 and 0xC2 as sentinels, as
1453 they never appear in well-formed UTF-8. (There are equivalents for
1454 UTF-EBCDIC). You can also store your Unicode code points in integer
1455 variables and use negative values as sentinels.
1457 If you're not writing such a tool, then whether you accept noncharacters
1458 as input is up to you (though the Standard recommends that you not). If
1459 you do strict input stream checking with Perl, these code points
1460 continue to be forbidden. This is to maintain backward compatibility
1461 (otherwise potential security holes could open up, as an unsuspecting
1462 application that was written assuming the noncharacters would be
1463 filtered out before getting to it, could now, without warning, start
1464 getting them). To do strict checking, you can use the layer
1465 C<:encoding('UTF-8')>.
1467 Perl continues to warn (using the warning category C<"nonchar">, which
1468 is a sub-category of C<"utf8">) if an attempt is made to output
1471 =head2 Beyond Unicode code points
1473 The maximum Unicode code point is C<U+10FFFF>, and Unicode only defines
1474 operations on code points up through that. But Perl works on code
1475 points up to the maximum permissible unsigned number available on the
1476 platform. However, Perl will not accept these from input streams unless
1477 lax rules are being used, and will warn (using the warning category
1478 C<"non_unicode">, which is a sub-category of C<"utf8">) if any are output.
1480 Since Unicode rules are not defined on these code points, if a
1481 Unicode-defined operation is done on them, Perl uses what we believe are
1482 sensible rules, while generally warning, using the C<"non_unicode">
1483 category. For example, C<uc("\x{11_0000}")> will generate such a
1484 warning, returning the input parameter as its result, since Perl defines
1485 the uppercase of every non-Unicode code point to be the code point
1486 itself. (All the case changing operations, not just uppercasing, work
1489 The situation with matching Unicode properties in regular expressions,
1490 the C<\p{}> and C<\P{}> constructs, against these code points is not as
1491 clear cut, and how these are handled has changed as we've gained
1494 One possibility is to treat any match against these code points as
1495 undefined. But since Perl doesn't have the concept of a match being
1496 undefined, it converts this to failing or C<FALSE>. This is almost, but
1497 not quite, what Perl did from v5.14 (when use of these code points
1498 became generally reliable) through v5.18. The difference is that Perl
1499 treated all C<\p{}> matches as failing, but all C<\P{}> matches as
1502 One problem with this is that it leads to unexpected, and confusting
1503 results in some cases:
1505 chr(0x110000) =~ \p{ASCII_Hex_Digit=True} # Failed on <= v5.18
1506 chr(0x110000) =~ \p{ASCII_Hex_Digit=False} # Failed! on <= v5.18
1508 That is, it treated both matches as undefined, and converted that to
1509 false (raising a warning on each). The first case is the expected
1510 result, but the second is likely counterintuitive: "How could both be
1511 false when they are complements?" Another problem was that the
1512 implementation optimized many Unicode property matches down to already
1513 existing simpler, faster operations, which don't raise the warning. We
1514 chose to not forgo those optimizations, which help the vast majority of
1515 matches, just to generate a warning for the unlikely event that an
1516 above-Unicode code point is being matched against.
1518 As a result of these problems, starting in v5.20, what Perl does is
1519 to treat non-Unicode code points as just typical unassigned Unicode
1520 characters, and matches accordingly. (Note: Unicode has atypical
1521 unassigned code points. For example, it has noncharacter code points,
1522 and ones that, when they do get assigned, are destined to be written
1523 Right-to-left, as Arabic and Hebrew are. Perl assumes that no
1524 non-Unicode code point has any atypical properties.)
1526 Perl, in most cases, will raise a warning when matching an above-Unicode
1527 code point against a Unicode property when the result is C<TRUE> for
1528 C<\p{}>, and C<FALSE> for C<\P{}>. For example:
1530 chr(0x110000) =~ \p{ASCII_Hex_Digit=True} # Fails, no warning
1531 chr(0x110000) =~ \p{ASCII_Hex_Digit=False} # Succeeds, with warning
1533 In both these examples, the character being matched is non-Unicode, so
1534 Unicode doesn't define how it should match. It clearly isn't an ASCII
1535 hex digit, so the first example clearly should fail, and so it does,
1536 with no warning. But it is arguable that the second example should have
1537 an undefined, hence C<FALSE>, result. So a warning is raised for it.
1539 Thus the warning is raised for many fewer cases than in earlier Perls,
1540 and only when what the result is could be arguable. It turns out that
1541 none of the optimizations made by Perl (or are ever likely to be made)
1542 cause the warning to be skipped, so it solves both problems of Perl's
1543 earlier approach. The most commonly used property that is affected by
1544 this change is C<\p{Unassigned}> which is a short form for
1545 C<\p{General_Category=Unassigned}>. Starting in v5.20, all non-Unicode
1546 code points are considered C<Unassigned>. In earlier releases the
1547 matches failed because the result was considered undefined.
1549 The only place where the warning is not raised when it might ought to
1550 have been is if optimizations cause the whole pattern match to not even
1551 be attempted. For example, Perl may figure out that for a string to
1552 match a certain regular expression pattern, the string has to contain
1553 the substring C<"foobar">. Before attempting the match, Perl may look
1554 for that substring, and if not found, immediately fail the match without
1555 actually trying it; so no warning gets generated even if the string
1556 contains an above-Unicode code point.
1558 This behavior is more "Do what I mean" than in earlier Perls for most
1559 applications. But it catches fewer issues for code that needs to be
1560 strictly Unicode compliant. Therefore there is an additional mode of
1561 operation available to accommodate such code. This mode is enabled if a
1562 regular expression pattern is compiled within the lexical scope where
1563 the C<"non_unicode"> warning class has been made fatal, say by:
1565 use warnings FATAL => "non_unicode"
1567 (see L<warnings>). In this mode of operation, Perl will raise the
1568 warning for all matches against a non-Unicode code point (not just the
1569 arguable ones), and it skips the optimizations that might cause the
1570 warning to not be output. (It currently still won't warn if the match
1571 isn't even attempted, like in the C<"foobar"> example above.)
1573 In summary, Perl now normally treats non-Unicode code points as typical
1574 Unicode unassigned code points for regular expression matches, raising a
1575 warning only when it is arguable what the result should be. However, if
1576 this warning has been made fatal, it isn't skipped.
1578 There is one exception to all this. C<\p{All}> looks like a Unicode
1579 property, but it is a Perl extension that is defined to be true for all
1580 possible code points, Unicode or not, so no warning is ever generated
1581 when matching this against a non-Unicode code point. (Prior to v5.20,
1582 it was an exact synonym for C<\p{Any}>, matching code points C<0>
1583 through C<0x10FFFF>.)
1585 =head2 Security Implications of Unicode
1588 L<Unicode Security Considerations|http://www.unicode.org/reports/tr36>.
1590 Also, note the following:
1598 Unfortunately, the original specification of UTF-8 leaves some room for
1599 interpretation of how many bytes of encoded output one should generate
1600 from one input Unicode character. Strictly speaking, the shortest
1601 possible sequence of UTF-8 bytes should be generated,
1602 because otherwise there is potential for an input buffer overflow at
1603 the receiving end of a UTF-8 connection. Perl always generates the
1604 shortest length UTF-8, and with warnings on, Perl will warn about
1605 non-shortest length UTF-8 along with other malformations, such as the
1606 surrogates, which are not Unicode code points valid for interchange.
1610 Regular expression pattern matching may surprise you if you're not
1611 accustomed to Unicode. Starting in Perl 5.14, several pattern
1612 modifiers are available to control this, called the character set
1613 modifiers. Details are given in L<perlre/Character set modifiers>.
1617 As discussed elsewhere, Perl has one foot (two hooves?) planted in
1618 each of two worlds: the old world of ASCII and single-byte locales, and
1619 the new world of Unicode, upgrading when necessary.
1620 If your legacy code does not explicitly use Unicode, no automatic
1621 switch-over to Unicode should happen.
1623 =head2 Unicode in Perl on EBCDIC
1625 Unicode is supported on EBCDIC platforms. See L<perlebcdic>.
1627 Unless ASCII vs. EBCDIC issues are specifically being discussed,
1628 references to UTF-8 encoding in this document and elsewhere should be
1629 read as meaning UTF-EBCDIC on EBCDIC platforms.
1630 See L<perlebcdic/Unicode and UTF>.
1632 Because UTF-EBCDIC is so similar to UTF-8, the differences are mostly
1633 hidden from you; S<C<use utf8>> (and NOT something like
1634 S<C<use utfebcdic>>) declares the the script is in the platform's
1635 "native" 8-bit encoding of Unicode. (Similarly for the C<":utf8">
1640 See L<perllocale/Unicode and UTF-8>
1642 =head2 When Unicode Does Not Happen
1644 There are still many places where Unicode (in some encoding or
1645 another) could be given as arguments or received as results, or both in
1646 Perl, but it is not, in spite of Perl having extensive ways to input and
1647 output in Unicode, and a few other "entry points" like the C<@ARGV>
1648 array (which can sometimes be interpreted as UTF-8).
1650 The following are such interfaces. Also, see L</The "Unicode Bug">.
1651 For all of these interfaces Perl
1652 currently (as of v5.16.0) simply assumes byte strings both as arguments
1653 and results, or UTF-8 strings if the (deprecated) C<encoding> pragma has been used.
1655 One reason that Perl does not attempt to resolve the role of Unicode in
1656 these situations is that the answers are highly dependent on the operating
1657 system and the file system(s). For example, whether filenames can be
1658 in Unicode and in exactly what kind of encoding, is not exactly a
1659 portable concept. Similarly for C<qx> and C<system>: how well will the
1660 "command-line interface" (and which of them?) handle Unicode?
1666 C<chdir>, C<chmod>, C<chown>, C<chroot>, C<exec>, C<link>, C<lstat>, C<mkdir>,
1667 C<rename>, C<rmdir>, C<stat>, C<symlink>, C<truncate>, C<unlink>, C<utime>, C<-X>
1675 C<glob> (aka the C<E<lt>*E<gt>>)
1679 C<open>, C<opendir>, C<sysopen>
1683 C<qx> (aka the backtick operator), C<system>
1687 C<readdir>, C<readlink>
1691 =head2 The "Unicode Bug"
1693 The term, "Unicode bug" has been applied to an inconsistency with the
1694 code points in the C<Latin-1 Supplement> block, that is, between
1695 128 and 255. Without a locale specified, unlike all other characters or
1696 code points, these characters can have very different semantics
1697 depending on the rules in effect. (Characters whose code points are
1698 above 255 force Unicode rules; whereas the rules for ASCII characters
1699 are the same under both ASCII and Unicode rules.)
1701 Under Unicode rules, these upper-Latin1 characters are interpreted as
1702 Unicode code points, which means they have the same semantics as Latin-1
1703 (ISO-8859-1) and C1 controls.
1705 As explained in L</ASCII Rules versus Unicode Rules>, under ASCII rules,
1706 they are considered to be unassigned characters.
1708 This can lead to unexpected results. For example, a string's
1709 semantics can suddenly change if a code point above 255 is appended to
1710 it, which changes the rules from ASCII to Unicode. As an
1711 example, consider the following program and its output:
1714 no feature 'unicode_strings';
1717 for ($s1, $s2, $s1.$s2) {
1725 If there's no C<\w> in C<s1> nor in C<s2>, why does their concatenation
1728 This anomaly stems from Perl's attempt to not disturb older programs that
1729 didn't use Unicode, along with Perl's desire to add Unicode support
1730 seamlessly. But the result turned out to not be seamless. (By the way,
1731 you can choose to be warned when things like this happen. See
1732 C<L<encoding::warnings>>.)
1734 L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
1735 was added, starting in Perl v5.12, to address this problem. It affects
1742 Changing the case of a scalar, that is, using C<uc()>, C<ucfirst()>, C<lc()>,
1743 and C<lcfirst()>, or C<\L>, C<\U>, C<\u> and C<\l> in double-quotish
1744 contexts, such as regular expression substitutions.
1746 Under C<unicode_strings> starting in Perl 5.12.0, Unicode rules are
1747 generally used. See L<perlfunc/lc> for details on how this works
1748 in combination with various other pragmas.
1752 Using caseless (C</i>) regular expression matching.
1754 Starting in Perl 5.14.0, regular expressions compiled within
1755 the scope of C<unicode_strings> use Unicode rules
1756 even when executed or compiled into larger
1757 regular expressions outside the scope.
1761 Matching any of several properties in regular expressions.
1763 These properties are C<\b> (without braces), C<\B> (without braces),
1764 C<\s>, C<\S>, C<\w>, C<\W>, and all the Posix character classes
1765 I<except> C<[[:ascii:]]>.
1767 Starting in Perl 5.14.0, regular expressions compiled within
1768 the scope of C<unicode_strings> use Unicode rules
1769 even when executed or compiled into larger
1770 regular expressions outside the scope.
1774 In C<quotemeta> or its inline equivalent C<\Q>.
1776 Starting in Perl 5.16.0, consistent quoting rules are used within the
1777 scope of C<unicode_strings>, as described in L<perlfunc/quotemeta>.
1778 Prior to that, or outside its scope, no code points above 127 are quoted
1779 in UTF-8 encoded strings, but in byte encoded strings, code points
1780 between 128-255 are always quoted.
1784 You can see from the above that the effect of C<unicode_strings>
1785 increased over several Perl releases. (And Perl's support for Unicode
1786 continues to improve; it's best to use the latest available release in
1787 order to get the most complete and accurate results possible.) Note that
1788 C<unicode_strings> is automatically chosen if you S<C<use 5.012>> or
1791 For Perls earlier than those described above, or when a string is passed
1792 to a function outside the scope of C<unicode_strings>, see the next section.
1794 =head2 Forcing Unicode in Perl (Or Unforcing Unicode in Perl)
1796 Sometimes (see L</"When Unicode Does Not Happen"> or L</The "Unicode Bug">)
1797 there are situations where you simply need to force a byte
1798 string into UTF-8, or vice versa. The standard module L<Encode> can be
1799 used for this, or the low-level calls
1800 L<C<utf8::upgrade($bytestring)>|utf8/Utility functions> and
1801 L<C<utf8::downgrade($utf8string[, FAIL_OK])>|utf8/Utility functions>.
1803 Note that C<utf8::downgrade()> can fail if the string contains characters
1804 that don't fit into a byte.
1806 Calling either function on a string that already is in the desired state is a
1809 L</ASCII Rules versus Unicode Rules> gives all the ways that a string is
1810 made to use Unicode rules.
1812 =head2 Using Unicode in XS
1814 See L<perlguts/"Unicode Support"> for an introduction to Unicode at
1815 the XS level, and L<perlapi/Unicode Support> for the API details.
1817 =head2 Hacking Perl to work on earlier Unicode versions (for very serious hackers only)
1819 Perl by default comes with the latest supported Unicode version built-in, but
1820 the goal is to allow you to change to use any earlier one. In Perls
1821 v5.20 and v5.22, however, the earliest usable version is Unicode 5.1.
1822 Perl v5.18 is able to handle all earlier versions.
1824 Download the files in the desired version of Unicode from the Unicode web
1825 site L<http://www.unicode.org>). These should replace the existing files in
1826 F<lib/unicore> in the Perl source tree. Follow the instructions in
1827 F<README.perl> in that directory to change some of their names, and then build
1828 perl (see L<INSTALL>).
1830 =head2 Porting code from perl-5.6.X
1832 Perls starting in 5.8 have a different Unicode model from 5.6. In 5.6 the
1833 programmer was required to use the C<utf8> pragma to declare that a
1834 given scope expected to deal with Unicode data and had to make sure that
1835 only Unicode data were reaching that scope. If you have code that is
1836 working with 5.6, you will need some of the following adjustments to
1837 your code. The examples are written such that the code will continue to
1838 work under 5.6, so you should be safe to try them out.
1844 A filehandle that should read or write UTF-8
1847 binmode $fh, ":encoding(utf8)";
1852 A scalar that is going to be passed to some extension
1854 Be it C<Compress::Zlib>, C<Apache::Request> or any extension that has no
1855 mention of Unicode in the manpage, you need to make sure that the
1856 UTF8 flag is stripped off. Note that at the time of this writing
1857 (January 2012) the mentioned modules are not UTF-8-aware. Please
1858 check the documentation to verify if this is still true.
1862 $val = Encode::encode_utf8($val); # make octets
1867 A scalar we got back from an extension
1869 If you believe the scalar comes back as UTF-8, you will most likely
1870 want the UTF8 flag restored:
1874 $val = Encode::decode_utf8($val);
1879 Same thing, if you are really sure it is UTF-8
1883 Encode::_utf8_on($val);
1888 A wrapper for L<DBI> C<fetchrow_array> and C<fetchrow_hashref>
1890 When the database contains only UTF-8, a wrapper function or method is
1891 a convenient way to replace all your C<fetchrow_array> and
1892 C<fetchrow_hashref> calls. A wrapper function will also make it easier to
1893 adapt to future enhancements in your database driver. Note that at the
1894 time of this writing (January 2012), the DBI has no standardized way
1895 to deal with UTF-8 data. Please check the L<DBI documentation|DBI> to verify if
1899 # $what is one of fetchrow_{array,hashref}
1900 my($self, $sth, $what) = @_;
1906 my @arr = $sth->$what;
1908 defined && /[^\000-\177]/ && Encode::_utf8_on($_);
1912 my $ret = $sth->$what;
1914 for my $k (keys %$ret) {
1917 && Encode::_utf8_on($_) for $ret->{$k};
1921 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret;
1931 A large scalar that you know can only contain ASCII
1933 Scalars that contain only ASCII and are marked as UTF-8 are sometimes
1934 a drag to your program. If you recognize such a situation, just remove
1937 utf8::downgrade($val) if $] > 5.008;
1943 See also L</The "Unicode Bug"> above.
1945 =head2 Interaction with Extensions
1947 When Perl exchanges data with an extension, the extension should be
1948 able to understand the UTF8 flag and act accordingly. If the
1949 extension doesn't recognize that flag, it's likely that the extension
1950 will return incorrectly-flagged data.
1952 So if you're working with Unicode data, consult the documentation of
1953 every module you're using if there are any issues with Unicode data
1954 exchange. If the documentation does not talk about Unicode at all,
1955 suspect the worst and probably look at the source to learn how the
1956 module is implemented. Modules written completely in Perl shouldn't
1957 cause problems. Modules that directly or indirectly access code written
1958 in other programming languages are at risk.
1960 For affected functions, the simple strategy to avoid data corruption is
1961 to always make the encoding of the exchanged data explicit. Choose an
1962 encoding that you know the extension can handle. Convert arguments passed
1963 to the extensions to that encoding and convert results back from that
1964 encoding. Write wrapper functions that do the conversions for you, so
1965 you can later change the functions when the extension catches up.
1967 To provide an example, let's say the popular C<Foo::Bar::escape_html>
1968 function doesn't deal with Unicode data yet. The wrapper function
1969 would convert the argument to raw UTF-8 and convert the result back to
1970 Perl's internal representation like so:
1972 sub my_escape_html ($) {
1974 return unless defined $what;
1975 Encode::decode_utf8(Foo::Bar::escape_html(
1976 Encode::encode_utf8($what)));
1979 Sometimes, when the extension does not convert data but just stores
1980 and retrieves it, you will be able to use the otherwise
1981 dangerous L<C<Encode::_utf8_on()>|Encode/_utf8_on> function. Let's say
1982 the popular C<Foo::Bar> extension, written in C, provides a C<param>
1983 method that lets you store and retrieve data according to these prototypes:
1985 $self->param($name, $value); # set a scalar
1986 $value = $self->param($name); # retrieve a scalar
1988 If it does not yet provide support for any encoding, one could write a
1989 derived class with such a C<param> method:
1992 my($self,$name,$value) = @_;
1993 utf8::upgrade($name); # make sure it is UTF-8 encoded
1994 if (defined $value) {
1995 utf8::upgrade($value); # make sure it is UTF-8 encoded
1996 return $self->SUPER::param($name,$value);
1998 my $ret = $self->SUPER::param($name);
1999 Encode::_utf8_on($ret); # we know, it is UTF-8 encoded
2004 Some extensions provide filters on data entry/exit points, such as
2005 C<DB_File::filter_store_key> and family. Look out for such filters in
2006 the documentation of your extensions; they can make the transition to
2007 Unicode data much easier.
2011 Some functions are slower when working on UTF-8 encoded strings than
2012 on byte encoded strings. All functions that need to hop over
2013 characters such as C<length()>, C<substr()> or C<index()>, or matching
2014 regular expressions can work B<much> faster when the underlying data are
2017 In Perl 5.8.0 the slowness was often quite spectacular; in Perl 5.8.1
2018 a caching scheme was introduced which improved the situation. In general,
2019 operations with UTF-8 encoded strings are still slower. As an example,
2020 the Unicode properties (character classes) like C<\p{Nd}> are known to
2021 be quite a bit slower (5-20 times) than their simpler counterparts
2022 like C<[0-9]> (then again, there are hundreds of Unicode characters matching
2023 C<Nd> compared with the 10 ASCII characters matching C<[0-9]>).
2027 L<perlunitut>, L<perluniintro>, L<perluniprops>, L<Encode>, L<open>, L<utf8>, L<bytes>,
2028 L<perlretut>, L<perlvar/"${^UNICODE}">,
2029 L<http://www.unicode.org/reports/tr44>).