This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perlapi: Clarify entry for hv_store()
[perl5.git] / pod / perlunicode.pod
CommitLineData
393fec97
GS
1=head1 NAME
2
3perlunicode - Unicode support in Perl
4
5=head1 DESCRIPTION
6
a6a7eedc
KW
7If you haven't already, before reading this document, you should become
8familiar with both L<perlunitut> and L<perluniintro>.
9
10Unicode aims to B<UNI>-fy the en-B<CODE>-ings of all the world's
11character sets into a single Standard. For quite a few of the various
12coding standards that existed when Unicode was first created, converting
13from each to Unicode essentially meant adding a constant to each code
14point in the original standard, and converting back meant just
15subtracting that same constant. For ASCII and ISO-8859-1, the constant
16is 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
18made it easy to do the conversions, and facilitated the adoption of
19Unicode.
20
21And it worked; nowadays, those legacy standards are rarely used. Most
22everyone uses Unicode.
23
24Unicode is a comprehensive standard. It specifies many things outside
25the scope of Perl, such as how to display sequences of characters. For
26a full discussion of all aspects of Unicode, see
27L<http://www.unicode.org>.
28
0a1f2d14 29=head2 Important Caveats
21bad921 30
a6a7eedc
KW
31Even though some of this section may not be understandable to you on
32first reading, we think it's important enough to highlight some of the
33gotchas before delving further, so here goes:
34
376d9008 35Unicode support is an extensive requirement. While Perl does not
c349b1b9
JH
36implement the Unicode standard or the accompanying technical reports
37from cover to cover, Perl does support many Unicode features.
21bad921 38
f57d8456
KW
39Also, the use of Unicode may present security issues that aren't
40obvious, see L</Security Implications of Unicode>.
9d1c51c1 41
13a2d996 42=over 4
21bad921 43
a9130ea9 44=item Safest if you C<use feature 'unicode_strings'>
42581d5d
KW
45
46In order to preserve backward compatibility, Perl does not turn
47on full internal Unicode support unless the pragma
b65e6125
KW
48L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
49is specified. (This is automatically
50selected if you S<C<use 5.012>> or higher.) Failure to do this can
42581d5d
KW
51trigger unexpected surprises. See L</The "Unicode Bug"> below.
52
2269d15c
KW
53This pragma doesn't affect I/O. Nor does it change the internal
54representation of strings, only their interpretation. There are still
55several places where Unicode isn't fully supported, such as in
56filenames.
42581d5d 57
fae2c0fb 58=item Input and Output Layers
21bad921 59
a6a7eedc
KW
60Use the C<:encoding(...)> layer to read from and write to
61filehandles using the specified encoding. (See L<open>.)
c349b1b9 62
01c3fbbc 63=item You must convert your non-ASCII, non-UTF-8 Perl scripts to be
a6a7eedc 64UTF-8.
21bad921 65
01c3fbbc
TC
66The L<encoding> module has been deprecated since perl 5.18 and the
67perl internals it requires have been removed with perl 5.26.
21bad921 68
a6a7eedc 69=item C<use utf8> still needed to enable L<UTF-8|/Unicode Encodings> in scripts
21bad921 70
a6a7eedc
KW
71If your Perl script is itself encoded in L<UTF-8|/Unicode Encodings>,
72the S<C<use utf8>> pragma must be explicitly included to enable
73recognition of that (in string or regular expression literals, or in
74identifier names). B<This is the only time when an explicit S<C<use
75utf8>> is needed.> (See L<utf8>).
7aa207d6 76
27c74dfd
KW
77If a Perl script begins with the bytes that form the UTF-8 encoding of
78the Unicode BYTE ORDER MARK (C<BOM>, see L</Unicode Encodings>), those
79bytes are completely ignored.
80
81=item L<UTF-16|/Unicode Encodings> scripts autodetected
7aa207d6 82
fea12a3e 83If a Perl script begins with the Unicode C<BOM> (UTF-16LE,
27c74dfd 84UTF16-BE), or if the script looks like non-C<BOM>-marked
a6a7eedc 85UTF-16 of either endianness, Perl will correctly read in the script as
27c74dfd 86the appropriate Unicode encoding.
990e18f7 87
21bad921
GS
88=back
89
376d9008 90=head2 Byte and Character Semantics
393fec97 91
a6a7eedc
KW
92Before Unicode, most encodings used 8 bits (a single byte) to encode
93each character. Thus a character was a byte, and a byte was a
94character, and there could be only 256 or fewer possible characters.
95"Byte Semantics" in the title of this section refers to
96this behavior. There was no need to distinguish between "Byte" and
97"Character".
98
99Then along comes Unicode which has room for over a million characters
100(and Perl allows for even more). This means that a character may
101require more than a single byte to represent it, and so the two terms
102are no longer equivalent. What matter are the characters as whole
103entities, and not usually the bytes that comprise them. That's what the
104term "Character Semantics" in the title of this section refers to.
105
106Perl had to change internally to decouple "bytes" from "characters".
107It is important that you too change your ideas, if you haven't already,
108so that "byte" and "character" no longer mean the same thing in your
109mind.
110
111The basic building block of Perl strings has always been a "character".
112The changes basically come down to that the implementation no longer
113thinks that a character is always just a single byte.
114
115There are various things to note:
393fec97
GS
116
117=over 4
118
119=item *
120
a6a7eedc
KW
121String handling functions, for the most part, continue to operate in
122terms of characters. C<length()>, for example, returns the number of
123characters in a string, just as before. But that number no longer is
124necessarily the same as the number of bytes in the string (there may be
125more bytes than characters). The other such functions include
126C<chop()>, C<chomp()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
127C<sort()>, C<sprintf()>, and C<write()>.
128
129The exceptions are:
130
131=over 4
132
133=item *
134
135the bit-oriented C<vec>
136
137E<nbsp>
138
139=item *
140
141the byte-oriented C<pack>/C<unpack> C<"C"> format
142
143However, the C<W> specifier does operate on whole characters, as does the
144C<U> specifier.
145
146=item *
147
148some operators that interact with the platform's operating system
149
150Operators dealing with filenames are examples.
151
152=item *
153
154when the functions are called from within the scope of the
155S<C<L<use bytes|bytes>>> pragma
156
157Likely, you should use this only for debugging anyway.
158
159=back
160
161=item *
162
376d9008 163Strings--including hash keys--and regular expression patterns may
b65e6125 164contain characters that have ordinal values larger than 255.
393fec97 165
2575c402
JW
166If you use a Unicode editor to edit your program, Unicode characters may
167occur directly within the literal strings in UTF-8 encoding, or UTF-16.
27c74dfd 168(The former requires a C<use utf8>, the latter may require a C<BOM>.)
3e4dbfed 169
a6a7eedc
KW
170L<perluniintro/Creating Unicode> gives other ways to place non-ASCII
171characters in your strings.
6f335b04 172
a6a7eedc 173=item *
fbb93542 174
a6a7eedc 175The C<chr()> and C<ord()> functions work on whole characters.
376d9008 176
393fec97
GS
177=item *
178
a6a7eedc
KW
179Regular expressions match whole characters. For example, C<"."> matches
180a whole character instead of only a single byte.
393fec97 181
393fec97
GS
182=item *
183
a6a7eedc
KW
184The C<tr///> operator translates whole characters. (Note that the
185C<tr///CU> functionality has been removed. For similar functionality to
186that, see C<pack('U0', ...)> and C<pack('C0', ...)>).
393fec97 187
393fec97
GS
188=item *
189
a6a7eedc 190C<scalar reverse()> reverses by character rather than by byte.
393fec97 191
393fec97
GS
192=item *
193
a6a7eedc 194The bit string operators, C<& | ^ ~> and (starting in v5.22)
fac71630
KW
195C<&. |. ^. ~.> can operate on bit strings encoded in UTF-8, but this
196can give unexpected results if any of the strings contain code points
197above 0xFF. Starting in v5.28, it is a fatal error to have such an
198operand. Otherwise, the operation is performed on a non-UTF-8 copy of
199the operand. If you're not sure about the encoding of a string,
200downgrade it before using any of these operators; you can use
a6a7eedc 201L<C<utf8::utf8_downgrade()>|utf8/Utility functions>.
822502e5 202
a6a7eedc 203=back
822502e5 204
a6a7eedc
KW
205The bottom line is that Perl has always practiced "Character Semantics",
206but with the advent of Unicode, that is now different than "Byte
207Semantics".
208
209=head2 ASCII Rules versus Unicode Rules
210
211Before Unicode, when a character was a byte was a character,
212Perl knew only about the 128 characters defined by ASCII, code points 0
b57dd509
KW
213through 127 (except for under L<S<C<use locale>>|perllocale>). That
214left the code
a6a7eedc
KW
215points 128 to 255 as unassigned, and available for whatever use a
216program might want. The only semantics they have is their ordinal
217numbers, and that they are members of none of the non-negative character
218classes. None are considered to match C<\w> for example, but all match
219C<\W>.
822502e5 220
a6a7eedc
KW
221Unicode, of course, assigns each of those code points a particular
222meaning (along with ones above 255). To preserve backward
223compatibility, Perl only uses the Unicode meanings when there is some
224indication that Unicode is what is intended; otherwise the non-ASCII
225code points remain treated as if they are unassigned.
226
227Here are the ways that Perl knows that a string should be treated as
228Unicode:
229
230=over
822502e5
ST
231
232=item *
233
a6a7eedc
KW
234Within the scope of S<C<use utf8>>
235
236If the whole program is Unicode (signified by using 8-bit B<U>nicode
e423fa83 237B<T>ransformation B<F>ormat), then all literal strings within it must be
a6a7eedc 238Unicode.
822502e5
ST
239
240=item *
241
a6a7eedc
KW
242Within the scope of
243L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
244
245This pragma was created so you can explicitly tell Perl that operations
246executed within its scope are to use Unicode rules. More operations are
247affected with newer perls. See L</The "Unicode Bug">.
822502e5
ST
248
249=item *
250
a6a7eedc
KW
251Within the scope of S<C<use 5.012>> or higher
252
253This implicitly turns on S<C<use feature 'unicode_strings'>>.
822502e5
ST
254
255=item *
256
a6a7eedc
KW
257Within the scope of
258L<S<C<use locale 'not_characters'>>|perllocale/Unicode and UTF-8>,
259or L<S<C<use locale>>|perllocale> and the current
260locale is a UTF-8 locale.
822502e5 261
a6a7eedc
KW
262The former is defined to imply Unicode handling; and the latter
263indicates a Unicode locale, hence a Unicode interpretation of all
264strings within it.
822502e5
ST
265
266=item *
267
a6a7eedc
KW
268When the string contains a Unicode-only code point
269
270Perl has never accepted code points above 255 without them being
271Unicode, so their use implies Unicode for the whole string.
822502e5
ST
272
273=item *
274
a6a7eedc
KW
275When the string contains a Unicode named code point C<\N{...}>
276
277The C<\N{...}> construct explicitly refers to a Unicode code point,
278even if it is one that is also in ASCII. Therefore the string
279containing it must be Unicode.
822502e5
ST
280
281=item *
282
a6a7eedc
KW
283When the string has come from an external source marked as
284Unicode
285
286The L<C<-C>|perlrun/-C [numberE<sol>list]> command line option can
287specify that certain inputs to the program are Unicode, and the values
288of this can be read by your Perl code, see L<perlvar/"${^UNICODE}">.
289
290=item * When the string has been upgraded to UTF-8
291
292The function L<C<utf8::utf8_upgrade()>|utf8/Utility functions>
293can be explicitly used to permanently (unless a subsequent
294C<utf8::utf8_downgrade()> is called) cause a string to be treated as
295Unicode.
296
297=item * There are additional methods for regular expression patterns
298
f6cf4627
KW
299A pattern that is compiled with the C<< /u >> or C<< /a >> modifiers is
300treated as Unicode (though there are some restrictions with C<< /a >>).
301Under the C<< /d >> and C<< /l >> modifiers, there are several other
302indications for Unicode; see L<perlre/Character set modifiers>.
822502e5
ST
303
304=back
305
a6a7eedc
KW
306Note that all of the above are overridden within the scope of
307C<L<use bytes|bytes>>; but you should be using this pragma only for
308debugging.
309
310Note also that some interactions with the platform's operating system
311never use Unicode rules.
312
313When Unicode rules are in effect:
314
822502e5
ST
315=over 4
316
317=item *
318
a6a7eedc
KW
319Case translation operators use the Unicode case translation tables.
320
321Note that C<uc()>, or C<\U> in interpolated strings, translates to
322uppercase, while C<ucfirst>, or C<\u> in interpolated strings,
323translates to titlecase in languages that make the distinction (which is
324equivalent to uppercase in languages without the distinction).
325
326There is a CPAN module, C<L<Unicode::Casing>>, which allows you to
327define your own mappings to be used in C<lc()>, C<lcfirst()>, C<uc()>,
328C<ucfirst()>, and C<fc> (or their double-quoted string inlined versions
329such as C<\U>). (Prior to Perl 5.16, this functionality was partially
330provided in the Perl core, but suffered from a number of insurmountable
331drawbacks, so the CPAN module was written instead.)
332
333=item *
334
335Character classes in regular expressions match based on the character
336properties specified in the Unicode properties database.
337
338C<\w> can be used to match a Japanese ideograph, for instance; and
339C<[[:digit:]]> a Bengali number.
340
341=item *
342
343Named Unicode properties, scripts, and block ranges may be used (like
344bracketed character classes) by using the C<\p{}> "matches property"
345construct and the C<\P{}> negation, "doesn't match property".
346
347See L</"Unicode Character Properties"> for more details.
348
349You can define your own character properties and use them
350in the regular expression with the C<\p{}> or C<\P{}> construct.
351See L</"User-Defined Character Properties"> for more details.
822502e5
ST
352
353=back
354
a6a7eedc
KW
355=head2 Extended Grapheme Clusters (Logical characters)
356
357Consider a character, say C<H>. It could appear with various marks around it,
358such as an acute accent, or a circumflex, or various hooks, circles, arrows,
359I<etc.>, above, below, to one side or the other, I<etc>. There are many
360possibilities among the world's languages. The number of combinations is
361astronomical, and if there were a character for each combination, it would
362soon exhaust Unicode's more than a million possible characters. So Unicode
363took a different approach: there is a character for the base C<H>, and a
364character for each of the possible marks, and these can be variously combined
365to get a final logical character. So a logical character--what appears to be a
366single character--can be a sequence of more than one individual characters.
367The Unicode standard calls these "extended grapheme clusters" (which
368is an improved version of the no-longer much used "grapheme cluster");
369Perl furnishes the C<\X> regular expression construct to match such
370sequences in their entirety.
371
372But Unicode's intent is to unify the existing character set standards and
373practices, and several pre-existing standards have single characters that
374mean the same thing as some of these combinations, like ISO-8859-1,
375which has quite a few of them. For example, C<"LATIN CAPITAL LETTER E
376WITH ACUTE"> was already in this standard when Unicode came along.
377Unicode therefore added it to its repertoire as that single character.
378But this character is considered by Unicode to be equivalent to the
379sequence consisting of the character C<"LATIN CAPITAL LETTER E">
380followed by the character C<"COMBINING ACUTE ACCENT">.
381
382C<"LATIN CAPITAL LETTER E WITH ACUTE"> is called a "pre-composed"
383character, and its equivalence with the "E" and the "COMBINING ACCENT"
384sequence is called canonical equivalence. All pre-composed characters
385are said to have a decomposition (into the equivalent sequence), and the
386decomposition type is also called canonical. A string may be comprised
387as much as possible of precomposed characters, or it may be comprised of
388entirely decomposed characters. Unicode calls these respectively,
389"Normalization Form Composed" (NFC) and "Normalization Form Decomposed".
390The C<L<Unicode::Normalize>> module contains functions that convert
391between the two. A string may also have both composed characters and
392decomposed characters; this module can be used to make it all one or the
393other.
394
395You may be presented with strings in any of these equivalent forms.
396There is currently nothing in Perl 5 that ignores the differences. So
dabde021 397you'll have to specially handle it. The usual advice is to convert your
a6a7eedc
KW
398inputs to C<NFD> before processing further.
399
400For more detailed information, see L<http://unicode.org/reports/tr15/>.
401
822502e5
ST
402=head2 Unicode Character Properties
403
ee88f7b6 404(The only time that Perl considers a sequence of individual code
9d1c51c1
KW
405points as a single logical character is in the C<\X> construct, already
406mentioned above. Therefore "character" in this discussion means a single
ee88f7b6
KW
407Unicode code point.)
408
409Very nearly all Unicode character properties are accessible through
410regular expressions by using the C<\p{}> "matches property" construct
411and the C<\P{}> "doesn't match property" for its negation.
51f494cc 412
9d1c51c1 413For instance, C<\p{Uppercase}> matches any single character with the Unicode
a9130ea9
KW
414C<"Uppercase"> property, while C<\p{L}> matches any character with a
415C<General_Category> of C<"L"> (letter) property (see
416L</General_Category> below). Brackets are not
9d1c51c1 417required for single letter property names, so C<\p{L}> is equivalent to C<\pL>.
51f494cc 418
9d1c51c1 419More formally, C<\p{Uppercase}> matches any single character whose Unicode
a9130ea9
KW
420C<Uppercase> property value is C<True>, and C<\P{Uppercase}> matches any character
421whose C<Uppercase> property value is C<False>, and they could have been written as
9d1c51c1 422C<\p{Uppercase=True}> and C<\p{Uppercase=False}>, respectively.
51f494cc 423
b19eb496 424This formality is needed when properties are not binary; that is, if they can
a9130ea9
KW
425take on more values than just C<True> and C<False>. For example, the
426C<Bidi_Class> property (see L</"Bidirectional Character Types"> below),
427can take on several different
428values, such as C<Left>, C<Right>, C<Whitespace>, and others. To match these, one needs
429to specify both the property name (C<Bidi_Class>), AND the value being
5bff2035 430matched against
b65e6125 431(C<Left>, C<Right>, I<etc.>). This is done, as in the examples above, by having the
9f815e24 432two components separated by an equal sign (or interchangeably, a colon), like
51f494cc
KW
433C<\p{Bidi_Class: Left}>.
434
435All Unicode-defined character properties may be written in these compound forms
a9130ea9 436of C<\p{I<property>=I<value>}> or C<\p{I<property>:I<value>}>, but Perl provides some
51f494cc
KW
437additional properties that are written only in the single form, as well as
438single-form short-cuts for all binary properties and certain others described
439below, in which you may omit the property name and the equals or colon
440separator.
441
442Most Unicode character properties have at least two synonyms (or aliases if you
b19eb496 443prefer): a short one that is easier to type and a longer one that is more
a9130ea9
KW
444descriptive and hence easier to understand. Thus the C<"L"> and
445C<"Letter"> properties above are equivalent and can be used
446interchangeably. Likewise, C<"Upper"> is a synonym for C<"Uppercase">,
447and we could have written C<\p{Uppercase}> equivalently as C<\p{Upper}>.
448Also, there are typically various synonyms for the values the property
449can be. For binary properties, C<"True"> has 3 synonyms: C<"T">,
450C<"Yes">, and C<"Y">; and C<"False"> has correspondingly C<"F">,
451C<"No">, and C<"N">. But be careful. A short form of a value for one
452property may not mean the same thing as the same short form for another.
453Thus, for the C<L</General_Category>> property, C<"L"> means
454C<"Letter">, but for the L<C<Bidi_Class>|/Bidirectional Character Types>
455property, C<"L"> means C<"Left">. A complete list of properties and
456synonyms is in L<perluniprops>.
51f494cc 457
b19eb496 458Upper/lower case differences in property names and values are irrelevant;
51f494cc
KW
459thus C<\p{Upper}> means the same thing as C<\p{upper}> or even C<\p{UpPeR}>.
460Similarly, you can add or subtract underscores anywhere in the middle of a
461word, so that these are also equivalent to C<\p{U_p_p_e_r}>. And white space
462is irrelevant adjacent to non-word characters, such as the braces and the equals
b19eb496
TC
463or colon separators, so C<\p{ Upper }> and C<\p{ Upper_case : Y }> are
464equivalent to these as well. In fact, white space and even
465hyphens can usually be added or deleted anywhere. So even C<\p{ Up-per case = Yes}> is
51f494cc 466equivalent. All this is called "loose-matching" by Unicode. The few places
b19eb496 467where stricter matching is used is in the middle of numbers, and in the Perl
51f494cc 468extension properties that begin or end with an underscore. Stricter matching
b19eb496 469cares about white space (except adjacent to non-word characters),
51f494cc 470hyphens, and non-interior underscores.
4193bef7 471
376d9008 472You can also use negation in both C<\p{}> and C<\P{}> by introducing a caret
a9130ea9 473(C<^>) between the first brace and the property name: C<\p{^Tamil}> is
eb0cc9e3 474equal to C<\P{Tamil}>.
4193bef7 475
56ca34ca
KW
476Almost all properties are immune to case-insensitive matching. That is,
477adding a C</i> regular expression modifier does not change what they
478match. There are two sets that are affected.
479The first set is
480C<Uppercase_Letter>,
481C<Lowercase_Letter>,
482and C<Titlecase_Letter>,
483all of which match C<Cased_Letter> under C</i> matching.
484And the second set is
485C<Uppercase>,
486C<Lowercase>,
487and C<Titlecase>,
488all of which match C<Cased> under C</i> matching.
489This set also includes its subsets C<PosixUpper> and C<PosixLower> both
a9130ea9 490of which under C</i> match C<PosixAlpha>.
56ca34ca 491(The difference between these sets is that some things, such as Roman
b65e6125
KW
492numerals, come in both upper and lower case so they are C<Cased>, but
493aren't considered letters, so they aren't C<Cased_Letter>'s.)
56ca34ca 494
2d88a86a
KW
495See L</Beyond Unicode code points> for special considerations when
496matching Unicode properties against non-Unicode code points.
94b42e47 497
51f494cc 498=head3 B<General_Category>
14bb0a9a 499
51f494cc
KW
500Every Unicode character is assigned a general category, which is the "most
501usual categorization of a character" (from
502L<http://www.unicode.org/reports/tr44>).
822502e5 503
9f815e24 504The compound way of writing these is like C<\p{General_Category=Number}>
b65e6125 505(short: C<\p{gc:n}>). But Perl furnishes shortcuts in which everything up
51f494cc
KW
506through the equal or colon separator is omitted. So you can instead just write
507C<\pN>.
822502e5 508
a9130ea9
KW
509Here are the short and long forms of the values the C<General Category> property
510can have:
393fec97 511
d73e5302
JH
512 Short Long
513
514 L Letter
51f494cc
KW
515 LC, L& Cased_Letter (that is: [\p{Ll}\p{Lu}\p{Lt}])
516 Lu Uppercase_Letter
517 Ll Lowercase_Letter
518 Lt Titlecase_Letter
519 Lm Modifier_Letter
520 Lo Other_Letter
d73e5302
JH
521
522 M Mark
51f494cc
KW
523 Mn Nonspacing_Mark
524 Mc Spacing_Mark
525 Me Enclosing_Mark
d73e5302
JH
526
527 N Number
51f494cc
KW
528 Nd Decimal_Number (also Digit)
529 Nl Letter_Number
530 No Other_Number
531
532 P Punctuation (also Punct)
533 Pc Connector_Punctuation
534 Pd Dash_Punctuation
535 Ps Open_Punctuation
536 Pe Close_Punctuation
537 Pi Initial_Punctuation
d73e5302 538 (may behave like Ps or Pe depending on usage)
51f494cc 539 Pf Final_Punctuation
d73e5302 540 (may behave like Ps or Pe depending on usage)
51f494cc 541 Po Other_Punctuation
d73e5302
JH
542
543 S Symbol
51f494cc
KW
544 Sm Math_Symbol
545 Sc Currency_Symbol
546 Sk Modifier_Symbol
547 So Other_Symbol
d73e5302
JH
548
549 Z Separator
51f494cc
KW
550 Zs Space_Separator
551 Zl Line_Separator
552 Zp Paragraph_Separator
d73e5302
JH
553
554 C Other
d88362ca 555 Cc Control (also Cntrl)
e150c829 556 Cf Format
6d4f9cf2 557 Cs Surrogate
51f494cc 558 Co Private_Use
e150c829 559 Cn Unassigned
1ac13f9a 560
376d9008 561Single-letter properties match all characters in any of the
3e4dbfed 562two-letter sub-properties starting with the same letter.
b19eb496 563C<LC> and C<L&> are special: both are aliases for the set consisting of everything matched by C<Ll>, C<Lu>, and C<Lt>.
32293815 564
51f494cc 565=head3 B<Bidirectional Character Types>
822502e5 566
b19eb496 567Because scripts differ in their directionality (Hebrew and Arabic are
a9130ea9 568written right to left, for example) Unicode supplies a C<Bidi_Class> property.
1850f57f 569Some of the values this property can have are:
32293815 570
88af3b93 571 Value Meaning
92e830a9 572
12ac2576
JP
573 L Left-to-Right
574 LRE Left-to-Right Embedding
575 LRO Left-to-Right Override
576 R Right-to-Left
51f494cc 577 AL Arabic Letter
12ac2576
JP
578 RLE Right-to-Left Embedding
579 RLO Right-to-Left Override
580 PDF Pop Directional Format
581 EN European Number
51f494cc
KW
582 ES European Separator
583 ET European Terminator
12ac2576 584 AN Arabic Number
51f494cc 585 CS Common Separator
12ac2576
JP
586 NSM Non-Spacing Mark
587 BN Boundary Neutral
588 B Paragraph Separator
589 S Segment Separator
590 WS Whitespace
591 ON Other Neutrals
592
51f494cc
KW
593This property is always written in the compound form.
594For example, C<\p{Bidi_Class:R}> matches characters that are normally
1850f57f 595written right to left. Unlike the
a9130ea9 596C<L</General_Category>> property, this
1850f57f
KW
597property can have more values added in a future Unicode release. Those
598listed above comprised the complete set for many Unicode releases, but
599others were added in Unicode 6.3; you can always find what the
20ada7da 600current ones are in L<perluniprops>. And
1850f57f 601L<http://www.unicode.org/reports/tr9/> describes how to use them.
eb0cc9e3 602
51f494cc
KW
603=head3 B<Scripts>
604
b19eb496 605The world's languages are written in many different scripts. This sentence
e1b711da 606(unless you're reading it in translation) is written in Latin, while Russian is
c69ca1d4 607written in Cyrillic, and Greek is written in, well, Greek; Japanese mainly in
e1b711da 608Hiragana or Katakana. There are many more.
51f494cc 609
48791bf1
KW
610The Unicode C<Script> and C<Script_Extensions> properties give what
611script a given character is in. The C<Script_Extensions> property is an
612improved version of C<Script>, as demonstrated below. Either property
613can be specified with the compound form like
82aed44a
KW
614C<\p{Script=Hebrew}> (short: C<\p{sc=hebr}>), or
615C<\p{Script_Extensions=Javanese}> (short: C<\p{scx=java}>).
616In addition, Perl furnishes shortcuts for all
48791bf1
KW
617C<Script_Extensions> property names. You can omit everything up through
618the equals (or colon), and simply write C<\p{Latin}> or C<\P{Cyrillic}>.
619(This is not true for C<Script>, which is required to be
620written in the compound form. Prior to Perl v5.26, the single form
621returned the plain old C<Script> version, but was changed because
622C<Script_Extensions> gives better results.)
82aed44a
KW
623
624The difference between these two properties involves characters that are
625used in multiple scripts. For example the digits '0' through '9' are
626used in many parts of the world. These are placed in a script named
627C<Common>. Other characters are used in just a few scripts. For
a9130ea9 628example, the C<"KATAKANA-HIRAGANA DOUBLE HYPHEN"> is used in both Japanese
82aed44a
KW
629scripts, Katakana and Hiragana, but nowhere else. The C<Script>
630property places all characters that are used in multiple scripts in the
631C<Common> script, while the C<Script_Extensions> property places those
632that are used in only a few scripts into each of those scripts; while
633still using C<Common> for those used in many scripts. Thus both these
634match:
635
636 "0" =~ /\p{sc=Common}/ # Matches
637 "0" =~ /\p{scx=Common}/ # Matches
638
639and only the first of these match:
640
641 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Common} # Matches
642 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Common} # No match
643
644And only the last two of these match:
645
646 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Hiragana} # No match
647 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Katakana} # No match
648 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Hiragana} # Matches
649 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Katakana} # Matches
650
651C<Script_Extensions> is thus an improved C<Script>, in which there are
652fewer characters in the C<Common> script, and correspondingly more in
653other scripts. It is new in Unicode version 6.0, and its data are likely
654to change significantly in later releases, as things get sorted out.
b65e6125 655New code should probably be using C<Script_Extensions> and not plain
48791bf1
KW
656C<Script>. If you compile perl with a Unicode release that doesn't have
657C<Script_Extensions>, the single form Perl extensions will instead refer
658to the plain C<Script> property. If you compile with a version of
659Unicode that doesn't have the C<Script> property, these extensions will
660not be defined at all.
82aed44a
KW
661
662(Actually, besides C<Common>, the C<Inherited> script, contains
663characters that are used in multiple scripts. These are modifier
b65e6125 664characters which inherit the script value
82aed44a
KW
665of the controlling character. Some of these are used in many scripts,
666and so go into C<Inherited> in both C<Script> and C<Script_Extensions>.
667Others are used in just a few scripts, so are in C<Inherited> in
668C<Script>, but not in C<Script_Extensions>.)
669
670It is worth stressing that there are several different sets of digits in
671Unicode that are equivalent to 0-9 and are matchable by C<\d> in a
672regular expression. If they are used in a single language only, they
48791bf1 673are in that language's C<Script> and C<Script_Extensions>. If they are
82aed44a
KW
674used in more than one script, they will be in C<sc=Common>, but only
675if they are used in many scripts should they be in C<scx=Common>.
51f494cc 676
48791bf1
KW
677The explanation above has omitted some detail; refer to UAX#24 "Unicode
678Script Property": L<http://www.unicode.org/reports/tr24>.
679
51f494cc
KW
680A complete list of scripts and their shortcuts is in L<perluniprops>.
681
a9130ea9 682=head3 B<Use of the C<"Is"> Prefix>
822502e5 683
7b0ac457 684For backward compatibility (with ancient Perl 5.6), all properties writable
b65e6125 685without using the compound form mentioned
51f494cc
KW
686so far may have C<Is> or C<Is_> prepended to their name, so C<\P{Is_Lu}>, for
687example, is equal to C<\P{Lu}>, and C<\p{IsScript:Arabic}> is equal to
688C<\p{Arabic}>.
eb0cc9e3 689
51f494cc 690=head3 B<Blocks>
2796c109 691
1bfb14c4
JH
692In addition to B<scripts>, Unicode also defines B<blocks> of
693characters. The difference between scripts and blocks is that the
694concept of scripts is closer to natural languages, while the concept
51f494cc 695of blocks is more of an artificial grouping based on groups of Unicode
a9130ea9 696characters with consecutive ordinal values. For example, the C<"Basic Latin">
b65e6125 697block is all the characters whose ordinals are between 0 and 127, inclusive; in
a9130ea9
KW
698other words, the ASCII characters. The C<"Latin"> script contains some letters
699from this as well as several other blocks, like C<"Latin-1 Supplement">,
b65e6125 700C<"Latin Extended-A">, I<etc.>, but it does not contain all the characters from
7be67b37 701those blocks. It does not, for example, contain the digits 0-9, because
82aed44a
KW
702those digits are shared across many scripts, and hence are in the
703C<Common> script.
51f494cc
KW
704
705For more about scripts versus blocks, see UAX#24 "Unicode Script Property":
706L<http://www.unicode.org/reports/tr24>
707
48791bf1 708The C<Script_Extensions> or C<Script> properties are likely to be the
82aed44a 709ones you want to use when processing
a9130ea9 710natural language; the C<Block> property may occasionally be useful in working
b19eb496 711with the nuts and bolts of Unicode.
51f494cc
KW
712
713Block names are matched in the compound form, like C<\p{Block: Arrows}> or
b19eb496 714C<\p{Blk=Hebrew}>. Unlike most other properties, only a few block names have a
6b5cf123
KW
715Unicode-defined short name.
716
717Perl also defines single form synonyms for the block property in cases
718where these do not conflict with something else. But don't use any of
719these, because they are unstable. Since these are Perl extensions, they
720are subordinate to official Unicode property names; Unicode doesn't know
721nor care about Perl's extensions. It may happen that a name that
722currently means the Perl extension will later be changed without warning
723to mean a different Unicode property in a future version of the perl
724interpreter that uses a later Unicode release, and your code would no
725longer work. The extensions are mentioned here for completeness: Take
726the block name and prefix it with one of: C<In> (for example
727C<\p{Blk=Arrows}> can currently be written as C<\p{In_Arrows}>); or
728sometimes C<Is> (like C<\p{Is_Arrows}>); or sometimes no prefix at all
48791bf1 729(C<\p{Arrows}>). As of this writing (Unicode 9.0) there are no
6b5cf123
KW
730conflicts with using the C<In_> prefix, but there are plenty with the
731other two forms. For example, C<\p{Is_Hebrew}> and C<\p{Hebrew}> mean
48791bf1
KW
732C<\p{Script_Extensions=Hebrew}> which is NOT the same thing as
733C<\p{Blk=Hebrew}>. Our
6b5cf123
KW
734advice used to be to use the C<In_> prefix as a single form way of
735specifying a block. But Unicode 8.0 added properties whose names begin
736with C<In>, and it's now clear that it's only luck that's so far
737prevented a conflict. Using C<In> is only marginally less typing than
738C<Blk:>, and the latter's meaning is clearer anyway, and guaranteed to
739never conflict. So don't take chances. Use C<\p{Blk=foo}> for new
740code. And be sure that block is what you really really want to do. In
741most cases scripts are what you want instead.
742
743A complete list of blocks is in L<perluniprops>.
51f494cc 744
9f815e24
KW
745=head3 B<Other Properties>
746
747There are many more properties than the very basic ones described here.
748A complete list is in L<perluniprops>.
749
750Unicode defines all its properties in the compound form, so all single-form
b19eb496
TC
751properties are Perl extensions. Most of these are just synonyms for the
752Unicode ones, but some are genuine extensions, including several that are in
9f815e24
KW
753the compound form. And quite a few of these are actually recommended by Unicode
754(in L<http://www.unicode.org/reports/tr18>).
755
5bff2035
KW
756This section gives some details on all extensions that aren't just
757synonyms for compound-form Unicode properties
758(for those properties, you'll have to refer to the
9f815e24
KW
759L<Unicode Standard|http://www.unicode.org/reports/tr44>.
760
761=over
762
763=item B<C<\p{All}>>
764
2d88a86a
KW
765This matches every possible code point. It is equivalent to C<qr/./s>.
766Unlike all the other non-user-defined C<\p{}> property matches, no
767warning is ever generated if this is property is matched against a
768non-Unicode code point (see L</Beyond Unicode code points> below).
9f815e24
KW
769
770=item B<C<\p{Alnum}>>
771
772This matches any C<\p{Alphabetic}> or C<\p{Decimal_Number}> character.
773
774=item B<C<\p{Any}>>
775
2d88a86a
KW
776This matches any of the 1_114_112 Unicode code points. It is a synonym
777for C<\p{Unicode}>.
9f815e24 778
42581d5d
KW
779=item B<C<\p{ASCII}>>
780
781This matches any of the 128 characters in the US-ASCII character set,
782which is a subset of Unicode.
783
9f815e24
KW
784=item B<C<\p{Assigned}>>
785
a9130ea9
KW
786This matches any assigned code point; that is, any code point whose L<general
787category|/General_Category> is not C<Unassigned> (or equivalently, not C<Cn>).
9f815e24
KW
788
789=item B<C<\p{Blank}>>
790
791This is the same as C<\h> and C<\p{HorizSpace}>: A character that changes the
792spacing horizontally.
793
794=item B<C<\p{Decomposition_Type: Non_Canonical}>> (Short: C<\p{Dt=NonCanon}>)
795
796Matches a character that has a non-canonical decomposition.
797
a6a7eedc
KW
798The L</Extended Grapheme Clusters (Logical characters)> section above
799talked about canonical decompositions. However, many more characters
800have a different type of decomposition, a "compatible" or
801"non-canonical" decomposition. The sequences that form these
802decompositions are not considered canonically equivalent to the
803pre-composed character. An example is the C<"SUPERSCRIPT ONE">. It is
804somewhat like a regular digit 1, but not exactly; its decomposition into
805the digit 1 is called a "compatible" decomposition, specifically a
9f815e24 806"super" decomposition. There are several such compatibility
b65e6125
KW
807decompositions (see L<http://www.unicode.org/reports/tr44>), including
808one called "compat", which means some miscellaneous type of
809decomposition that doesn't fit into the other decomposition categories
810that Unicode has chosen.
9f815e24
KW
811
812Note that most Unicode characters don't have a decomposition, so their
a9130ea9 813decomposition type is C<"None">.
9f815e24 814
b19eb496
TC
815For your convenience, Perl has added the C<Non_Canonical> decomposition
816type to mean any of the several compatibility decompositions.
9f815e24
KW
817
818=item B<C<\p{Graph}>>
819
820Matches any character that is graphic. Theoretically, this means a character
821that on a printer would cause ink to be used.
822
823=item B<C<\p{HorizSpace}>>
824
b19eb496 825This is the same as C<\h> and C<\p{Blank}>: a character that changes the
9f815e24
KW
826spacing horizontally.
827
42581d5d 828=item B<C<\p{In=*}>>
9f815e24
KW
829
830This is a synonym for C<\p{Present_In=*}>
831
832=item B<C<\p{PerlSpace}>>
833
d28d8023 834This is the same as C<\s>, restricted to ASCII, namely C<S<[ \f\n\r\t]>>
779cf272 835and starting in Perl v5.18, a vertical tab.
9f815e24
KW
836
837Mnemonic: Perl's (original) space
838
839=item B<C<\p{PerlWord}>>
840
841This is the same as C<\w>, restricted to ASCII, namely C<[A-Za-z0-9_]>
842
843Mnemonic: Perl's (original) word.
844
42581d5d 845=item B<C<\p{Posix...}>>
9f815e24 846
b65e6125
KW
847There are several of these, which are equivalents, using the C<\p{}>
848notation, for Posix classes and are described in
42581d5d 849L<perlrecharclass/POSIX Character Classes>.
9f815e24
KW
850
851=item B<C<\p{Present_In: *}>> (Short: C<\p{In=*}>)
852
853This property is used when you need to know in what Unicode version(s) a
854character is.
855
856The "*" above stands for some two digit Unicode version number, such as
857C<1.1> or C<4.0>; or the "*" can also be C<Unassigned>. This property will
858match the code points whose final disposition has been settled as of the
859Unicode release given by the version number; C<\p{Present_In: Unassigned}>
860will match those code points whose meaning has yet to be assigned.
861
a9130ea9 862For example, C<U+0041> C<"LATIN CAPITAL LETTER A"> was present in the very first
9f815e24
KW
863Unicode release available, which is C<1.1>, so this property is true for all
864valid "*" versions. On the other hand, C<U+1EFF> was not assigned until version
a9130ea9 8655.1 when it became C<"LATIN SMALL LETTER Y WITH LOOP">, so the only "*" that
9f815e24
KW
866would match it are 5.1, 5.2, and later.
867
868Unicode furnishes the C<Age> property from which this is derived. The problem
869with Age is that a strict interpretation of it (which Perl takes) has it
870matching the precise release a code point's meaning is introduced in. Thus
871C<U+0041> would match only 1.1; and C<U+1EFF> only 5.1. This is not usually what
872you want.
873
874Some non-Perl implementations of the Age property may change its meaning to be
a9130ea9 875the same as the Perl C<Present_In> property; just be aware of that.
9f815e24
KW
876
877Another confusion with both these properties is that the definition is not
b19eb496
TC
878that the code point has been I<assigned>, but that the meaning of the code point
879has been I<determined>. This is because 66 code points will always be
a9130ea9 880unassigned, and so the C<Age> for them is the Unicode version in which the decision
b19eb496 881to make them so was made. For example, C<U+FDD0> is to be permanently
9f815e24 882unassigned to a character, and the decision to do that was made in version 3.1,
b19eb496 883so C<\p{Age=3.1}> matches this character, as also does C<\p{Present_In: 3.1}> and up.
9f815e24
KW
884
885=item B<C<\p{Print}>>
886
ae5b72c8 887This matches any character that is graphical or blank, except controls.
9f815e24
KW
888
889=item B<C<\p{SpacePerl}>>
890
891This is the same as C<\s>, including beyond ASCII.
892
4d4acfba 893Mnemonic: Space, as modified by Perl. (It doesn't include the vertical tab
779cf272 894until v5.18, which both the Posix standard and Unicode consider white space.)
9f815e24 895
4364919a
KW
896=item B<C<\p{Title}>> and B<C<\p{Titlecase}>>
897
898Under case-sensitive matching, these both match the same code points as
899C<\p{General Category=Titlecase_Letter}> (C<\p{gc=lt}>). The difference
900is that under C</i> caseless matching, these match the same as
901C<\p{Cased}>, whereas C<\p{gc=lt}> matches C<\p{Cased_Letter>).
902
2d88a86a
KW
903=item B<C<\p{Unicode}>>
904
905This matches any of the 1_114_112 Unicode code points.
906C<\p{Any}>.
907
9f815e24
KW
908=item B<C<\p{VertSpace}>>
909
910This is the same as C<\v>: A character that changes the spacing vertically.
911
912=item B<C<\p{Word}>>
913
b19eb496 914This is the same as C<\w>, including over 100_000 characters beyond ASCII.
9f815e24 915
42581d5d
KW
916=item B<C<\p{XPosix...}>>
917
b19eb496 918There are several of these, which are the standard Posix classes
42581d5d
KW
919extended to the full Unicode range. They are described in
920L<perlrecharclass/POSIX Character Classes>.
921
9f815e24
KW
922=back
923
a9130ea9 924
376d9008 925=head2 User-Defined Character Properties
491fd90a 926
51f494cc 927You can define your own binary character properties by defining subroutines
a9130ea9 928whose names begin with C<"In"> or C<"Is">. (The experimental feature
9d1a5160
KW
929L<perlre/(?[ ])> provides an alternative which allows more complex
930definitions.) The subroutines can be defined in any
51f494cc 931package. The user-defined properties can be used in the regular expression
a9130ea9 932C<\p{}> and C<\P{}> constructs; if you are using a user-defined property from a
51f494cc 933package other than the one you are in, you must specify its package in the
a9130ea9 934C<\p{}> or C<\P{}> construct.
bac0b425 935
51f494cc 936 # assuming property Is_Foreign defined in Lang::
bac0b425
JP
937 package main; # property package name required
938 if ($txt =~ /\p{Lang::IsForeign}+/) { ... }
939
940 package Lang; # property package name not required
941 if ($txt =~ /\p{IsForeign}+/) { ... }
942
943
944Note that the effect is compile-time and immutable once defined.
b19eb496
TC
945However, the subroutines are passed a single parameter, which is 0 if
946case-sensitive matching is in effect and non-zero if caseless matching
56ca34ca
KW
947is in effect. The subroutine may return different values depending on
948the value of the flag, and one set of values will immutably be in effect
b19eb496 949for all case-sensitive matches, and the other set for all case-insensitive
56ca34ca 950matches.
491fd90a 951
b19eb496 952Note that if the regular expression is tainted, then Perl will die rather
a9130ea9 953than calling the subroutine when the name of the subroutine is
0e9be77f
DM
954determined by the tainted data.
955
376d9008
JB
956The subroutines must return a specially-formatted string, with one
957or more newline-separated lines. Each line must be one of the following:
491fd90a
JH
958
959=over 4
960
961=item *
962
df9e1087 963A single hexadecimal number denoting a code point to include.
510254c9
A
964
965=item *
966
99a6b1f0 967Two hexadecimal numbers separated by horizontal whitespace (space or
73b95840
KW
968tabular characters) denoting a range of code points to include. The
969second number must not be smaller than the first.
491fd90a
JH
970
971=item *
972
a9130ea9
KW
973Something to include, prefixed by C<"+">: a built-in character
974property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 975name) user-defined character property,
bac0b425
JP
976to represent all the characters in that property; two hexadecimal code
977points for a range; or a single hexadecimal code point.
491fd90a
JH
978
979=item *
980
a9130ea9
KW
981Something to exclude, prefixed by C<"-">: an existing character
982property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 983name) user-defined character property,
bac0b425
JP
984to represent all the characters in that property; two hexadecimal code
985points for a range; or a single hexadecimal code point.
491fd90a
JH
986
987=item *
988
a9130ea9
KW
989Something to negate, prefixed C<"!">: an existing character
990property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 991name) user-defined character property,
bac0b425
JP
992to represent all the characters in that property; two hexadecimal code
993points for a range; or a single hexadecimal code point.
994
995=item *
996
a9130ea9
KW
997Something to intersect with, prefixed by C<"&">: an existing character
998property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 999name) user-defined character property,
bac0b425
JP
1000for all the characters except the characters in the property; two
1001hexadecimal code points for a range; or a single hexadecimal code point.
491fd90a
JH
1002
1003=back
1004
1005For example, to define a property that covers both the Japanese
1006syllabaries (hiragana and katakana), you can define
1007
1008 sub InKana {
d88362ca 1009 return <<END;
d5822f25
A
1010 3040\t309F
1011 30A0\t30FF
491fd90a
JH
1012 END
1013 }
1014
d5822f25
A
1015Imagine that the here-doc end marker is at the beginning of the line.
1016Now you can use C<\p{InKana}> and C<\P{InKana}>.
491fd90a
JH
1017
1018You could also have used the existing block property names:
1019
1020 sub InKana {
d88362ca 1021 return <<'END';
491fd90a
JH
1022 +utf8::InHiragana
1023 +utf8::InKatakana
1024 END
1025 }
1026
1027Suppose you wanted to match only the allocated characters,
d5822f25 1028not the raw block ranges: in other words, you want to remove
b65e6125 1029the unassigned characters:
491fd90a
JH
1030
1031 sub InKana {
d88362ca 1032 return <<'END';
491fd90a
JH
1033 +utf8::InHiragana
1034 +utf8::InKatakana
1035 -utf8::IsCn
1036 END
1037 }
1038
1039The negation is useful for defining (surprise!) negated classes.
1040
1041 sub InNotKana {
d88362ca 1042 return <<'END';
491fd90a
JH
1043 !utf8::InHiragana
1044 -utf8::InKatakana
1045 +utf8::IsCn
1046 END
1047 }
1048
461020ad
KW
1049This will match all non-Unicode code points, since every one of them is
1050not in Kana. You can use intersection to exclude these, if desired, as
1051this modified example shows:
bac0b425 1052
461020ad 1053 sub InNotKana {
bac0b425 1054 return <<'END';
461020ad
KW
1055 !utf8::InHiragana
1056 -utf8::InKatakana
1057 +utf8::IsCn
1058 &utf8::Any
bac0b425
JP
1059 END
1060 }
1061
461020ad
KW
1062C<&utf8::Any> must be the last line in the definition.
1063
1064Intersection is used generally for getting the common characters matched
a9130ea9 1065by two (or more) classes. It's important to remember not to use C<"&"> for
461020ad 1066the first set; that would be intersecting with nothing, resulting in an
5acbde07 1067empty set. (Similarly using C<"-"> for the first set does nothing).
461020ad 1068
2d88a86a
KW
1069Unlike non-user-defined C<\p{}> property matches, no warning is ever
1070generated if these properties are matched against a non-Unicode code
1071point (see L</Beyond Unicode code points> below).
bac0b425 1072
68585b5e 1073=head2 User-Defined Case Mappings (for serious hackers only)
822502e5 1074
5d1892be 1075B<This feature has been removed as of Perl 5.16.>
a9130ea9 1076The CPAN module C<L<Unicode::Casing>> provides better functionality without
5d1892be
KW
1077the drawbacks that this feature had. If you are using a Perl earlier
1078than 5.16, this feature was most fully documented in the 5.14 version of
1079this pod:
1080L<http://perldoc.perl.org/5.14.0/perlunicode.html#User-Defined-Case-Mappings-%28for-serious-hackers-only%29>
3a2263fe 1081
376d9008 1082=head2 Character Encodings for Input and Output
8cbd9a7a 1083
7221edc9 1084See L<Encode>.
8cbd9a7a 1085
c29a771d 1086=head2 Unicode Regular Expression Support Level
776f8809 1087
b19eb496 1088The following list of Unicode supported features for regular expressions describes
fea12a3e
KW
1089all features currently directly supported by core Perl. The references
1090to "Level I<N>" and the section numbers refer to
1091L<UTS#18 "Unicode Regular Expressions"|http://www.unicode.org/reports/tr18>,
1092version 13, November 2013.
1093
1094=head3 Level 1 - Basic Unicode Support
1095
1096 RL1.1 Hex Notation - Done [1]
1097 RL1.2 Properties - Done [2]
1098 RL1.2a Compatibility Properties - Done [3]
1099 RL1.3 Subtraction and Intersection - Experimental [4]
1100 RL1.4 Simple Word Boundaries - Done [5]
1101 RL1.5 Simple Loose Matches - Done [6]
1102 RL1.6 Line Boundaries - Partial [7]
1103 RL1.7 Supplementary Code Points - Done [8]
755789c0 1104
6f33e417
KW
1105=over 4
1106
a6a7eedc 1107=item [1] C<\N{U+...}> and C<\x{...}>
6f33e417 1108
fea12a3e
KW
1109=item [2]
1110C<\p{...}> C<\P{...}>. This requirement is for a minimal list of
1111properties. Perl supports these and all other Unicode character
1112properties, as R2.7 asks (see L</"Unicode Character Properties"> above).
6f33e417 1113
fea12a3e
KW
1114=item [3]
1115Perl has C<\d> C<\D> C<\s> C<\S> C<\w> C<\W> C<\X> C<[:I<prop>:]>
1116C<[:^I<prop>:]>, plus all the properties specified by
1117L<http://www.unicode.org/reports/tr18/#Compatibility_Properties>. These
1118are described above in L</Other Properties>
6f33e417 1119
fea12a3e 1120=item [4]
6f33e417 1121
fea12a3e 1122The experimental feature C<"(?[...])"> starting in v5.18 accomplishes
a6a7eedc 1123this.
6f33e417 1124
a6a7eedc
KW
1125See L<perlre/(?[ ])>. If you don't want to use an experimental
1126feature, you can use one of the following:
6f33e417
KW
1127
1128=over 4
1129
a6a7eedc 1130=item *
f67a5002 1131Regular expression lookahead
6f33e417
KW
1132
1133You can mimic class subtraction using lookahead.
8158862b 1134For example, what UTS#18 might write as
29bdacb8 1135
209c9685 1136 [{Block=Greek}-[{UNASSIGNED}]]
dbe420b4
JH
1137
1138in Perl can be written as:
1139
209c9685
KW
1140 (?!\p{Unassigned})\p{Block=Greek}
1141 (?=\p{Assigned})\p{Block=Greek}
dbe420b4
JH
1142
1143But in this particular example, you probably really want
1144
209c9685 1145 \p{Greek}
dbe420b4
JH
1146
1147which will match assigned characters known to be part of the Greek script.
29bdacb8 1148
a6a7eedc
KW
1149=item *
1150
1151CPAN module C<L<Unicode::Regex::Set>>
8158862b 1152
6f33e417
KW
1153It does implement the full UTS#18 grouping, intersection, union, and
1154removal (subtraction) syntax.
8158862b 1155
a6a7eedc
KW
1156=item *
1157
1158L</"User-Defined Character Properties">
6f33e417 1159
a9130ea9 1160C<"+"> for union, C<"-"> for removal (set-difference), C<"&"> for intersection
6f33e417
KW
1161
1162=back
1163
fea12a3e
KW
1164=item [5]
1165C<\b> C<\B> meet most, but not all, the details of this requirement, but
1166C<\b{wb}> and C<\B{wb}> do, as well as the stricter R2.3.
1167
1168=item [6]
6f33e417 1169
a6a7eedc 1170Note that Perl does Full case-folding in matching, not Simple:
6f33e417 1171
a6a7eedc
KW
1172For example C<U+1F88> is equivalent to C<U+1F00 U+03B9>, instead of just
1173C<U+1F80>. This difference matters mainly for certain Greek capital
a9130ea9
KW
1174letters with certain modifiers: the Full case-folding decomposes the
1175letter, while the Simple case-folding would map it to a single
1176character.
6f33e417 1177
fea12a3e
KW
1178=item [7]
1179
1180The reason this is considered to be only partially implemented is that
1181Perl has L<C<qrE<sol>\b{lb}E<sol>>|perlrebackslash/\b{lb}> and
1182C<L<Unicode::LineBreak>> that are conformant with
1183L<UAX#14 "Unicode Line Breaking Algorithm"|http://www.unicode.org/reports/tr14>.
1184The regular expression construct provides default behavior, while the
1185heavier-weight module provides customizable line breaking.
1186
1187But Perl treats C<\n> as the start- and end-line
1188delimiter, whereas Unicode specifies more characters that should be
1189so-interpreted.
6f33e417 1190
a6a7eedc 1191These are:
6f33e417 1192
a6a7eedc
KW
1193 VT U+000B (\v in C)
1194 FF U+000C (\f)
1195 CR U+000D (\r)
1196 NEL U+0085
1197 LS U+2028
1198 PS U+2029
6f33e417 1199
a6a7eedc
KW
1200C<^> and C<$> in regular expression patterns are supposed to match all
1201these, but don't.
1202These characters also don't, but should, affect C<< <> >> C<$.>, and
1203script line numbers.
6f33e417 1204
a6a7eedc
KW
1205Also, lines should not be split within C<CRLF> (i.e. there is no
1206empty line between C<\r> and C<\n>). For C<CRLF>, try the C<:crlf>
1207layer (see L<PerlIO>).
1208
fea12a3e 1209=item [8]
a9130ea9
KW
1210UTF-8/UTF-EBDDIC used in Perl allows not only C<U+10000> to
1211C<U+10FFFF> but also beyond C<U+10FFFF>
6f33e417
KW
1212
1213=back
5ca1ac52 1214
fea12a3e 1215=head3 Level 2 - Extended Unicode Support
776f8809 1216
fea12a3e
KW
1217 RL2.1 Canonical Equivalents - Retracted [9]
1218 by Unicode
1219 RL2.2 Extended Grapheme Clusters - Partial [10]
1220 RL2.3 Default Word Boundaries - Done [11]
1221 RL2.4 Default Case Conversion - Done
1222 RL2.5 Name Properties - Done
1223 RL2.6 Wildcard Properties - Missing
1224 RL2.7 Full Properties - Done
776f8809 1225
fea12a3e 1226=over 4
8158862b 1227
fea12a3e
KW
1228=item [9]
1229Unicode has rewritten this portion of UTS#18 to say that getting
1230canonical equivalence (see UAX#15
1231L<"Unicode Normalization Forms"|http://www.unicode.org/reports/tr15>)
1232is basically to be done at the programmer level. Use NFD to write
1233both your regular expressions and text to match them against (you
1234can use L<Unicode::Normalize>).
776f8809 1235
fea12a3e
KW
1236=item [10]
1237Perl has C<\X> and C<\b{gcb}> but we don't have a "Grapheme Cluster Mode".
1238
1239=item [11] see
1240L<UAX#29 "Unicode Text Segmentation"|http://www.unicode.org/reports/tr29>,
1241
1242=back
1243
1244=head3 Level 3 - Tailored Support
1245
1246 RL3.1 Tailored Punctuation - Missing
1247 RL3.2 Tailored Grapheme Clusters - Missing [12]
1248 RL3.3 Tailored Word Boundaries - Missing
1249 RL3.4 Tailored Loose Matches - Retracted by Unicode
1250 RL3.5 Tailored Ranges - Retracted by Unicode
1251 RL3.6 Context Matching - Missing [13]
1252 RL3.7 Incremental Matches - Missing
1253 RL3.8 Unicode Set Sharing - Unicode is proposing
1254 to retract this
1255 RL3.9 Possible Match Sets - Missing
1256 RL3.10 Folded Matching - Retracted by Unicode
1257 RL3.11 Submatchers - Missing
1258
1259=over 4
1260
1261=item [12]
1262Perl has L<Unicode::Collate>, but it isn't integrated with regular
1263expressions. See
1264L<UTS#10 "Unicode Collation Algorithms"|http://www.unicode.org/reports/tr10>.
776f8809 1265
fea12a3e
KW
1266=item [13]
1267Perl has C<(?<=x)> and C<(?=x)>, but lookaheads or lookbehinds should
1268see outside of the target substring
776f8809
JH
1269
1270=back
1271
c349b1b9
JH
1272=head2 Unicode Encodings
1273
376d9008
JB
1274Unicode characters are assigned to I<code points>, which are abstract
1275numbers. To use these numbers, various encodings are needed.
c349b1b9
JH
1276
1277=over 4
1278
c29a771d 1279=item *
5cb3728c
RB
1280
1281UTF-8
c349b1b9 1282
6d4f9cf2 1283UTF-8 is a variable-length (1 to 4 bytes), byte-order independent
a6a7eedc
KW
1284encoding. In most of Perl's documentation, including elsewhere in this
1285document, the term "UTF-8" means also "UTF-EBCDIC". But in this section,
1286"UTF-8" refers only to the encoding used on ASCII platforms. It is a
1287superset of 7-bit US-ASCII, so anything encoded in ASCII has the
1288identical representation when encoded in UTF-8.
c349b1b9 1289
8c007b5a 1290The following table is from Unicode 3.2.
05632f9a 1291
755789c0 1292 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
05632f9a 1293
d88362ca 1294 U+0000..U+007F 00..7F
e1b711da 1295 U+0080..U+07FF * C2..DF 80..BF
d88362ca 1296 U+0800..U+0FFF E0 * A0..BF 80..BF
ec90690f
ST
1297 U+1000..U+CFFF E1..EC 80..BF 80..BF
1298 U+D000..U+D7FF ED 80..9F 80..BF
755789c0 1299 U+D800..U+DFFF +++++ utf16 surrogates, not legal utf8 +++++
ec90690f 1300 U+E000..U+FFFF EE..EF 80..BF 80..BF
d88362ca
KW
1301 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
1302 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
1303 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
e1b711da 1304
b19eb496 1305Note the gaps marked by "*" before several of the byte entries above. These are
e1b711da
KW
1306caused by legal UTF-8 avoiding non-shortest encodings: it is technically
1307possible to UTF-8-encode a single code point in different ways, but that is
1308explicitly forbidden, and the shortest possible encoding should always be used
1309(and that is what Perl does).
37361303 1310
376d9008 1311Another way to look at it is via bits:
05632f9a 1312
755789c0 1313 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
05632f9a 1314
755789c0
KW
1315 0aaaaaaa 0aaaaaaa
1316 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
1317 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
1318 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
05632f9a 1319
a9130ea9 1320As you can see, the continuation bytes all begin with C<"10">, and the
e1b711da 1321leading bits of the start byte tell how many bytes there are in the
05632f9a
JH
1322encoded character.
1323
6d4f9cf2 1324The original UTF-8 specification allowed up to 6 bytes, to allow
a9130ea9 1325encoding of numbers up to C<0x7FFF_FFFF>. Perl continues to allow those,
6d4f9cf2
KW
1326and has extended that up to 13 bytes to encode code points up to what
1327can fit in a 64-bit word. However, Perl will warn if you output any of
b19eb496 1328these as being non-portable; and under strict UTF-8 input protocols,
760c7c2f
KW
1329they are forbidden. In addition, it is deprecated to use a code point
1330larger than what a signed integer variable on your system can hold. On
133132-bit ASCII systems, this means C<0x7FFF_FFFF> is the legal maximum
1332going forward (much higher on 64-bit systems).
6d4f9cf2 1333
c29a771d 1334=item *
5cb3728c
RB
1335
1336UTF-EBCDIC
dbe420b4 1337
b65e6125 1338Like UTF-8, but EBCDIC-safe, in the way that UTF-8 is ASCII-safe.
a6a7eedc
KW
1339This means that all the basic characters (which includes all
1340those that have ASCII equivalents (like C<"A">, C<"0">, C<"%">, I<etc.>)
1341are the same in both EBCDIC and UTF-EBCDIC.)
1342
c0236afe
KW
1343UTF-EBCDIC is used on EBCDIC platforms. It generally requires more
1344bytes to represent a given code point than UTF-8 does; the largest
1345Unicode code points take 5 bytes to represent (instead of 4 in UTF-8),
1346and, extended for 64-bit words, it uses 14 bytes instead of 13 bytes in
1347UTF-8.
dbe420b4 1348
c29a771d 1349=item *
5cb3728c 1350
b65e6125 1351UTF-16, UTF-16BE, UTF-16LE, Surrogates, and C<BOM>'s (Byte Order Marks)
c349b1b9 1352
1bfb14c4
JH
1353The followings items are mostly for reference and general Unicode
1354knowledge, Perl doesn't use these constructs internally.
dbe420b4 1355
b19eb496
TC
1356Like UTF-8, UTF-16 is a variable-width encoding, but where
1357UTF-8 uses 8-bit code units, UTF-16 uses 16-bit code units.
1358All code points occupy either 2 or 4 bytes in UTF-16: code points
1359C<U+0000..U+FFFF> are stored in a single 16-bit unit, and code
1bfb14c4 1360points C<U+10000..U+10FFFF> in two 16-bit units. The latter case is
c349b1b9
JH
1361using I<surrogates>, the first 16-bit unit being the I<high
1362surrogate>, and the second being the I<low surrogate>.
1363
376d9008 1364Surrogates are code points set aside to encode the C<U+10000..U+10FFFF>
c349b1b9 1365range of Unicode code points in pairs of 16-bit units. The I<high
9f815e24 1366surrogates> are the range C<U+D800..U+DBFF> and the I<low surrogates>
376d9008 1367are the range C<U+DC00..U+DFFF>. The surrogate encoding is
c349b1b9 1368
d88362ca
KW
1369 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
1370 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
c349b1b9
JH
1371
1372and the decoding is
1373
d88362ca 1374 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
c349b1b9 1375
376d9008 1376Because of the 16-bitness, UTF-16 is byte-order dependent. UTF-16
c349b1b9 1377itself can be used for in-memory computations, but if storage or
376d9008
JB
1378transfer is required either UTF-16BE (big-endian) or UTF-16LE
1379(little-endian) encodings must be chosen.
c349b1b9
JH
1380
1381This introduces another problem: what if you just know that your data
376d9008 1382is UTF-16, but you don't know which endianness? Byte Order Marks, or
b65e6125 1383C<BOM>'s, are a solution to this. A special character has been reserved
86bbd6d1 1384in Unicode to function as a byte order marker: the character with the
a9130ea9 1385code point C<U+FEFF> is the C<BOM>.
042da322 1386
a9130ea9 1387The trick is that if you read a C<BOM>, you will know the byte order,
376d9008
JB
1388since if it was written on a big-endian platform, you will read the
1389bytes C<0xFE 0xFF>, but if it was written on a little-endian platform,
1390you will read the bytes C<0xFF 0xFE>. (And if the originating platform
b65e6125
KW
1391was writing in ASCII platform UTF-8, you will read the bytes
1392C<0xEF 0xBB 0xBF>.)
042da322 1393
86bbd6d1 1394The way this trick works is that the character with the code point
6d4f9cf2 1395C<U+FFFE> is not supposed to be in input streams, so the
a9130ea9 1396sequence of bytes C<0xFF 0xFE> is unambiguously "C<BOM>, represented in
1bfb14c4 1397little-endian format" and cannot be C<U+FFFE>, represented in big-endian
6d4f9cf2
KW
1398format".
1399
1400Surrogates have no meaning in Unicode outside their use in pairs to
1401represent other code points. However, Perl allows them to be
1402represented individually internally, for example by saying
f651977e
TC
1403C<chr(0xD801)>, so that all code points, not just those valid for open
1404interchange, are
6d4f9cf2 1405representable. Unicode does define semantics for them, such as their
a9130ea9
KW
1406C<L</General_Category>> is C<"Cs">. But because their use is somewhat dangerous,
1407Perl will warn (using the warning category C<"surrogate">, which is a
1408sub-category of C<"utf8">) if an attempt is made
6d4f9cf2
KW
1409to do things like take the lower case of one, or match
1410case-insensitively, or to output them. (But don't try this on Perls
1411before 5.14.)
c349b1b9 1412
c29a771d 1413=item *
5cb3728c 1414
1e54db1a 1415UTF-32, UTF-32BE, UTF-32LE
c349b1b9 1416
b65e6125 1417The UTF-32 family is pretty much like the UTF-16 family, except that
042da322 1418the units are 32-bit, and therefore the surrogate scheme is not
a9130ea9 1419needed. UTF-32 is a fixed-width encoding. The C<BOM> signatures are
b19eb496 1420C<0x00 0x00 0xFE 0xFF> for BE and C<0xFF 0xFE 0x00 0x00> for LE.
c349b1b9 1421
c29a771d 1422=item *
5cb3728c
RB
1423
1424UCS-2, UCS-4
c349b1b9 1425
b19eb496 1426Legacy, fixed-width encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
376d9008 1427encoding. Unlike UTF-16, UCS-2 is not extensible beyond C<U+FFFF>,
339cfa0e 1428because it does not use surrogates. UCS-4 is a 32-bit encoding,
b19eb496 1429functionally identical to UTF-32 (the difference being that
a9130ea9 1430UCS-4 forbids neither surrogates nor code points larger than C<0x10_FFFF>).
c349b1b9 1431
c29a771d 1432=item *
5cb3728c
RB
1433
1434UTF-7
c349b1b9 1435
376d9008
JB
1436A seven-bit safe (non-eight-bit) encoding, which is useful if the
1437transport or storage is not eight-bit safe. Defined by RFC 2152.
c349b1b9 1438
95a1a48b
JH
1439=back
1440
57e88091 1441=head2 Noncharacter code points
6d4f9cf2 1442
57e88091 144366 code points are set aside in Unicode as "noncharacter code points".
a9130ea9 1444These all have the C<Unassigned> (C<Cn>) C<L</General_Category>>, and
57e88091
KW
1445no character will ever be assigned to any of them. They are the 32 code
1446points between C<U+FDD0> and C<U+FDEF> inclusive, and the 34 code
1447points:
1448
1449 U+FFFE U+FFFF
1450 U+1FFFE U+1FFFF
1451 U+2FFFE U+2FFFF
1452 ...
1453 U+EFFFE U+EFFFF
1454 U+FFFFE U+FFFFF
1455 U+10FFFE U+10FFFF
1456
1457Until Unicode 7.0, the noncharacters were "B<forbidden> for use in open
1458interchange of Unicode text data", so that code that processed those
1459streams could use these code points as sentinels that could be mixed in
1460with character data, and would always be distinguishable from that data.
1461(Emphasis above and in the next paragraph are added in this document.)
1462
1463Unicode 7.0 changed the wording so that they are "B<not recommended> for
1464use in open interchange of Unicode text data". The 7.0 Standard goes on
1465to say:
1466
1467=over 4
1468
1469"If a noncharacter is received in open interchange, an application is
1470not required to interpret it in any way. It is good practice, however,
1471to recognize it as a noncharacter and to take appropriate action, such
1472as replacing it with C<U+FFFD> replacement character, to indicate the
1473problem in the text. It is not recommended to simply delete
1474noncharacter code points from such text, because of the potential
1475security issues caused by deleting uninterpreted characters. (See
1476conformance clause C7 in Section 3.2, Conformance Requirements, and
1477L<Unicode Technical Report #36, "Unicode Security
1478Considerations"|http://www.unicode.org/reports/tr36/#Substituting_for_Ill_Formed_Subsequences>)."
1479
1480=back
1481
1482This change was made because it was found that various commercial tools
1483like editors, or for things like source code control, had been written
1484so that they would not handle program files that used these code points,
1485effectively precluding their use almost entirely! And that was never
1486the intent. They've always been meant to be usable within an
1487application, or cooperating set of applications, at will.
1488
1489If you're writing code, such as an editor, that is supposed to be able
1490to handle any Unicode text data, then you shouldn't be using these code
1491points yourself, and instead allow them in the input. If you need
1492sentinels, they should instead be something that isn't legal Unicode.
1493For UTF-8 data, you can use the bytes 0xC1 and 0xC2 as sentinels, as
1494they never appear in well-formed UTF-8. (There are equivalents for
1495UTF-EBCDIC). You can also store your Unicode code points in integer
1496variables and use negative values as sentinels.
1497
1498If you're not writing such a tool, then whether you accept noncharacters
1499as input is up to you (though the Standard recommends that you not). If
1500you do strict input stream checking with Perl, these code points
1501continue to be forbidden. This is to maintain backward compatibility
1502(otherwise potential security holes could open up, as an unsuspecting
1503application that was written assuming the noncharacters would be
1504filtered out before getting to it, could now, without warning, start
1505getting them). To do strict checking, you can use the layer
1506C<:encoding('UTF-8')>.
1507
1508Perl continues to warn (using the warning category C<"nonchar">, which
1509is a sub-category of C<"utf8">) if an attempt is made to output
1510noncharacters.
42581d5d
KW
1511
1512=head2 Beyond Unicode code points
1513
a9130ea9
KW
1514The maximum Unicode code point is C<U+10FFFF>, and Unicode only defines
1515operations on code points up through that. But Perl works on code
42581d5d
KW
1516points up to the maximum permissible unsigned number available on the
1517platform. However, Perl will not accept these from input streams unless
1518lax rules are being used, and will warn (using the warning category
2d88a86a
KW
1519C<"non_unicode">, which is a sub-category of C<"utf8">) if any are output.
1520
1521Since Unicode rules are not defined on these code points, if a
1522Unicode-defined operation is done on them, Perl uses what we believe are
1523sensible rules, while generally warning, using the C<"non_unicode">
1524category. For example, C<uc("\x{11_0000}")> will generate such a
1525warning, returning the input parameter as its result, since Perl defines
1526the uppercase of every non-Unicode code point to be the code point
b65e6125
KW
1527itself. (All the case changing operations, not just uppercasing, work
1528this way.)
2d88a86a
KW
1529
1530The situation with matching Unicode properties in regular expressions,
1531the C<\p{}> and C<\P{}> constructs, against these code points is not as
1532clear cut, and how these are handled has changed as we've gained
1533experience.
1534
1535One possibility is to treat any match against these code points as
1536undefined. But since Perl doesn't have the concept of a match being
1537undefined, it converts this to failing or C<FALSE>. This is almost, but
1538not quite, what Perl did from v5.14 (when use of these code points
1539became generally reliable) through v5.18. The difference is that Perl
1540treated all C<\p{}> matches as failing, but all C<\P{}> matches as
1541succeeding.
1542
f66ccb6c 1543One problem with this is that it leads to unexpected, and confusing
2d88a86a
KW
1544results in some cases:
1545
1546 chr(0x110000) =~ \p{ASCII_Hex_Digit=True} # Failed on <= v5.18
1547 chr(0x110000) =~ \p{ASCII_Hex_Digit=False} # Failed! on <= v5.18
1548
1549That is, it treated both matches as undefined, and converted that to
1550false (raising a warning on each). The first case is the expected
1551result, but the second is likely counterintuitive: "How could both be
1552false when they are complements?" Another problem was that the
1553implementation optimized many Unicode property matches down to already
1554existing simpler, faster operations, which don't raise the warning. We
1555chose to not forgo those optimizations, which help the vast majority of
1556matches, just to generate a warning for the unlikely event that an
1557above-Unicode code point is being matched against.
1558
1559As a result of these problems, starting in v5.20, what Perl does is
1560to treat non-Unicode code points as just typical unassigned Unicode
1561characters, and matches accordingly. (Note: Unicode has atypical
57e88091 1562unassigned code points. For example, it has noncharacter code points,
2d88a86a
KW
1563and ones that, when they do get assigned, are destined to be written
1564Right-to-left, as Arabic and Hebrew are. Perl assumes that no
1565non-Unicode code point has any atypical properties.)
1566
1567Perl, in most cases, will raise a warning when matching an above-Unicode
1568code point against a Unicode property when the result is C<TRUE> for
1569C<\p{}>, and C<FALSE> for C<\P{}>. For example:
1570
1571 chr(0x110000) =~ \p{ASCII_Hex_Digit=True} # Fails, no warning
1572 chr(0x110000) =~ \p{ASCII_Hex_Digit=False} # Succeeds, with warning
1573
1574In both these examples, the character being matched is non-Unicode, so
1575Unicode doesn't define how it should match. It clearly isn't an ASCII
1576hex digit, so the first example clearly should fail, and so it does,
1577with no warning. But it is arguable that the second example should have
1578an undefined, hence C<FALSE>, result. So a warning is raised for it.
1579
1580Thus the warning is raised for many fewer cases than in earlier Perls,
1581and only when what the result is could be arguable. It turns out that
1582none of the optimizations made by Perl (or are ever likely to be made)
1583cause the warning to be skipped, so it solves both problems of Perl's
1584earlier approach. The most commonly used property that is affected by
1585this change is C<\p{Unassigned}> which is a short form for
1586C<\p{General_Category=Unassigned}>. Starting in v5.20, all non-Unicode
1587code points are considered C<Unassigned>. In earlier releases the
1588matches failed because the result was considered undefined.
1589
1590The only place where the warning is not raised when it might ought to
1591have been is if optimizations cause the whole pattern match to not even
1592be attempted. For example, Perl may figure out that for a string to
1593match a certain regular expression pattern, the string has to contain
1594the substring C<"foobar">. Before attempting the match, Perl may look
1595for that substring, and if not found, immediately fail the match without
1596actually trying it; so no warning gets generated even if the string
1597contains an above-Unicode code point.
1598
1599This behavior is more "Do what I mean" than in earlier Perls for most
1600applications. But it catches fewer issues for code that needs to be
1601strictly Unicode compliant. Therefore there is an additional mode of
1602operation available to accommodate such code. This mode is enabled if a
1603regular expression pattern is compiled within the lexical scope where
1604the C<"non_unicode"> warning class has been made fatal, say by:
1605
1606 use warnings FATAL => "non_unicode"
1607
44ecbbd8 1608(see L<warnings>). In this mode of operation, Perl will raise the
2d88a86a
KW
1609warning for all matches against a non-Unicode code point (not just the
1610arguable ones), and it skips the optimizations that might cause the
1611warning to not be output. (It currently still won't warn if the match
1612isn't even attempted, like in the C<"foobar"> example above.)
1613
1614In summary, Perl now normally treats non-Unicode code points as typical
1615Unicode unassigned code points for regular expression matches, raising a
1616warning only when it is arguable what the result should be. However, if
1617this warning has been made fatal, it isn't skipped.
1618
1619There is one exception to all this. C<\p{All}> looks like a Unicode
1620property, but it is a Perl extension that is defined to be true for all
1621possible code points, Unicode or not, so no warning is ever generated
1622when matching this against a non-Unicode code point. (Prior to v5.20,
1623it was an exact synonym for C<\p{Any}>, matching code points C<0>
1624through C<0x10FFFF>.)
6d4f9cf2 1625
0d7c09bb
JH
1626=head2 Security Implications of Unicode
1627
b65e6125
KW
1628First, read
1629L<Unicode Security Considerations|http://www.unicode.org/reports/tr36>.
1630
e1b711da
KW
1631Also, note the following:
1632
0d7c09bb
JH
1633=over 4
1634
1635=item *
1636
1637Malformed UTF-8
bf0fa0b2 1638
f57d8456
KW
1639UTF-8 is very structured, so many combinations of bytes are invalid. In
1640the past, Perl tried to soldier on and make some sense of invalid
1641combinations, but this can lead to security holes, so now, if the Perl
1642core needs to process an invalid combination, it will either raise a
1643fatal error, or will replace those bytes by the sequence that forms the
1644Unicode REPLACEMENT CHARACTER, for which purpose Unicode created it.
1645
1646Every code point can be represented by more than one possible
1647syntactically valid UTF-8 sequence. Early on, both Unicode and Perl
1648considered any of these to be valid, but now, all sequences longer
1649than the shortest possible one are considered to be malformed.
1650
1651Unicode considers many code points to be illegal, or to be avoided.
1652Perl generally accepts them, once they have passed through any input
1653filters that may try to exclude them. These have been discussed above
1654(see "Surrogates" under UTF-16 in L</Unicode Encodings>,
1655L</Noncharacter code points>, and L</Beyond Unicode code points>).
bf0fa0b2 1656
0d7c09bb
JH
1657=item *
1658
68693f9e 1659Regular expression pattern matching may surprise you if you're not
b19eb496
TC
1660accustomed to Unicode. Starting in Perl 5.14, several pattern
1661modifiers are available to control this, called the character set
42581d5d
KW
1662modifiers. Details are given in L<perlre/Character set modifiers>.
1663
1664=back
0d7c09bb 1665
376d9008 1666As discussed elsewhere, Perl has one foot (two hooves?) planted in
a6a7eedc
KW
1667each of two worlds: the old world of ASCII and single-byte locales, and
1668the new world of Unicode, upgrading when necessary.
376d9008 1669If your legacy code does not explicitly use Unicode, no automatic
a6a7eedc 1670switch-over to Unicode should happen.
0d7c09bb 1671
c349b1b9
JH
1672=head2 Unicode in Perl on EBCDIC
1673
a6a7eedc
KW
1674Unicode is supported on EBCDIC platforms. See L<perlebcdic>.
1675
1676Unless ASCII vs. EBCDIC issues are specifically being discussed,
1677references to UTF-8 encoding in this document and elsewhere should be
1678read as meaning UTF-EBCDIC on EBCDIC platforms.
1679See L<perlebcdic/Unicode and UTF>.
1680
1681Because UTF-EBCDIC is so similar to UTF-8, the differences are mostly
1682hidden from you; S<C<use utf8>> (and NOT something like
dabde021 1683S<C<use utfebcdic>>) declares the script is in the platform's
a6a7eedc
KW
1684"native" 8-bit encoding of Unicode. (Similarly for the C<":utf8">
1685layer.)
c349b1b9 1686
b310b053
JH
1687=head2 Locales
1688
42581d5d 1689See L<perllocale/Unicode and UTF-8>
b310b053 1690
1aad1664
JH
1691=head2 When Unicode Does Not Happen
1692
b65e6125
KW
1693There are still many places where Unicode (in some encoding or
1694another) could be given as arguments or received as results, or both in
1695Perl, but it is not, in spite of Perl having extensive ways to input and
1696output in Unicode, and a few other "entry points" like the C<@ARGV>
1697array (which can sometimes be interpreted as UTF-8).
1aad1664 1698
e1b711da
KW
1699The following are such interfaces. Also, see L</The "Unicode Bug">.
1700For all of these interfaces Perl
b9cedb1b 1701currently (as of v5.16.0) simply assumes byte strings both as arguments
b65e6125 1702and results, or UTF-8 strings if the (deprecated) C<encoding> pragma has been used.
1aad1664 1703
b19eb496
TC
1704One reason that Perl does not attempt to resolve the role of Unicode in
1705these situations is that the answers are highly dependent on the operating
1aad1664 1706system and the file system(s). For example, whether filenames can be
b19eb496
TC
1707in Unicode and in exactly what kind of encoding, is not exactly a
1708portable concept. Similarly for C<qx> and C<system>: how well will the
1709"command-line interface" (and which of them?) handle Unicode?
1aad1664
JH
1710
1711=over 4
1712
557a2462
RB
1713=item *
1714
a9130ea9
KW
1715C<chdir>, C<chmod>, C<chown>, C<chroot>, C<exec>, C<link>, C<lstat>, C<mkdir>,
1716C<rename>, C<rmdir>, C<stat>, C<symlink>, C<truncate>, C<unlink>, C<utime>, C<-X>
557a2462
RB
1717
1718=item *
1719
a9130ea9 1720C<%ENV>
557a2462
RB
1721
1722=item *
1723
a9130ea9 1724C<glob> (aka the C<E<lt>*E<gt>>)
557a2462
RB
1725
1726=item *
1aad1664 1727
a9130ea9 1728C<open>, C<opendir>, C<sysopen>
1aad1664 1729
557a2462 1730=item *
1aad1664 1731
a9130ea9 1732C<qx> (aka the backtick operator), C<system>
1aad1664 1733
557a2462 1734=item *
1aad1664 1735
a9130ea9 1736C<readdir>, C<readlink>
1aad1664
JH
1737
1738=back
1739
e1b711da
KW
1740=head2 The "Unicode Bug"
1741
a6a7eedc
KW
1742The term, "Unicode bug" has been applied to an inconsistency with the
1743code points in the C<Latin-1 Supplement> block, that is, between
1744128 and 255. Without a locale specified, unlike all other characters or
1745code points, these characters can have very different semantics
1746depending on the rules in effect. (Characters whose code points are
1747above 255 force Unicode rules; whereas the rules for ASCII characters
1748are the same under both ASCII and Unicode rules.)
1749
1750Under Unicode rules, these upper-Latin1 characters are interpreted as
1751Unicode code points, which means they have the same semantics as Latin-1
1752(ISO-8859-1) and C1 controls.
1753
1754As explained in L</ASCII Rules versus Unicode Rules>, under ASCII rules,
1755they are considered to be unassigned characters.
1756
1757This can lead to unexpected results. For example, a string's
1758semantics can suddenly change if a code point above 255 is appended to
1759it, which changes the rules from ASCII to Unicode. As an
1760example, consider the following program and its output:
1761
1762 $ perl -le'
f434f357 1763 no feature "unicode_strings";
a6a7eedc
KW
1764 $s1 = "\xC2";
1765 $s2 = "\x{2660}";
1766 for ($s1, $s2, $s1.$s2) {
1767 print /\w/ || 0;
1768 }
1769 '
1770 0
1771 0
1772 1
1773
1774If there's no C<\w> in C<s1> nor in C<s2>, why does their concatenation
1775have one?
1776
1777This anomaly stems from Perl's attempt to not disturb older programs that
1778didn't use Unicode, along with Perl's desire to add Unicode support
1779seamlessly. But the result turned out to not be seamless. (By the way,
1780you can choose to be warned when things like this happen. See
1781C<L<encoding::warnings>>.)
1782
1783L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
1784was added, starting in Perl v5.12, to address this problem. It affects
1785these things:
e1b711da
KW
1786
1787=over 4
1788
1789=item *
1790
1791Changing the case of a scalar, that is, using C<uc()>, C<ucfirst()>, C<lc()>,
2e2b2571
KW
1792and C<lcfirst()>, or C<\L>, C<\U>, C<\u> and C<\l> in double-quotish
1793contexts, such as regular expression substitutions.
a6a7eedc
KW
1794
1795Under C<unicode_strings> starting in Perl 5.12.0, Unicode rules are
2e2b2571
KW
1796generally used. See L<perlfunc/lc> for details on how this works
1797in combination with various other pragmas.
e1b711da
KW
1798
1799=item *
1800
2e2b2571 1801Using caseless (C</i>) regular expression matching.
a6a7eedc 1802
2e2b2571 1803Starting in Perl 5.14.0, regular expressions compiled within
a6a7eedc 1804the scope of C<unicode_strings> use Unicode rules
2e2b2571
KW
1805even when executed or compiled into larger
1806regular expressions outside the scope.
e1b711da
KW
1807
1808=item *
1809
a6a7eedc
KW
1810Matching any of several properties in regular expressions.
1811
1812These properties are C<\b> (without braces), C<\B> (without braces),
1813C<\s>, C<\S>, C<\w>, C<\W>, and all the Posix character classes
630d17dc 1814I<except> C<[[:ascii:]]>.
a6a7eedc 1815
2e2b2571 1816Starting in Perl 5.14.0, regular expressions compiled within
a6a7eedc 1817the scope of C<unicode_strings> use Unicode rules
2e2b2571
KW
1818even when executed or compiled into larger
1819regular expressions outside the scope.
e1b711da
KW
1820
1821=item *
1822
a6a7eedc
KW
1823In C<quotemeta> or its inline equivalent C<\Q>.
1824
2e2b2571
KW
1825Starting in Perl 5.16.0, consistent quoting rules are used within the
1826scope of C<unicode_strings>, as described in L<perlfunc/quotemeta>.
a6a7eedc
KW
1827Prior to that, or outside its scope, no code points above 127 are quoted
1828in UTF-8 encoded strings, but in byte encoded strings, code points
1829between 128-255 are always quoted.
eb88ed9e 1830
d6c970c7
AC
1831=item *
1832
1833In the C<..> or L<range|perlop/Range Operators> operator.
1834
1835Starting in Perl 5.26.0, the range operator on strings treats their lengths
1836consistently within the scope of C<unicode_strings>. Prior to that, or
1837outside its scope, it could produce strings whose length in characters
1838exceeded that of the right-hand side, where the right-hand side took up more
1839bytes than the correct range endpoint.
1840
20ae58f7
AC
1841=item *
1842
1843In L<< C<split>'s special-case whitespace splitting|perlfunc/split >>.
1844
1845Starting in Perl 5.28.0, the C<split> function with a pattern specified as
1846a string containing a single space handles whitespace characters consistently
1847within the scope of of C<unicode_strings>. Prior to that, or outside its scope,
1848characters that are whitespace according to Unicode rules but not according to
1849ASCII rules were treated as field contents rather than field separators when
1850they appear in byte-encoded strings.
1851
e1b711da
KW
1852=back
1853
a6a7eedc
KW
1854You can see from the above that the effect of C<unicode_strings>
1855increased over several Perl releases. (And Perl's support for Unicode
1856continues to improve; it's best to use the latest available release in
1857order to get the most complete and accurate results possible.) Note that
1858C<unicode_strings> is automatically chosen if you S<C<use 5.012>> or
1859higher.
e1b711da 1860
2e2b2571 1861For Perls earlier than those described above, or when a string is passed
a6a7eedc 1862to a function outside the scope of C<unicode_strings>, see the next section.
e1b711da 1863
1aad1664
JH
1864=head2 Forcing Unicode in Perl (Or Unforcing Unicode in Perl)
1865
e1b711da
KW
1866Sometimes (see L</"When Unicode Does Not Happen"> or L</The "Unicode Bug">)
1867there are situations where you simply need to force a byte
a6a7eedc
KW
1868string into UTF-8, or vice versa. The standard module L<Encode> can be
1869used for this, or the low-level calls
a9130ea9 1870L<C<utf8::upgrade($bytestring)>|utf8/Utility functions> and
a6a7eedc 1871L<C<utf8::downgrade($utf8string[, FAIL_OK])>|utf8/Utility functions>.
1aad1664 1872
a9130ea9 1873Note that C<utf8::downgrade()> can fail if the string contains characters
2bbc8d55 1874that don't fit into a byte.
1aad1664 1875
e1b711da
KW
1876Calling either function on a string that already is in the desired state is a
1877no-op.
1878
a6a7eedc
KW
1879L</ASCII Rules versus Unicode Rules> gives all the ways that a string is
1880made to use Unicode rules.
95a1a48b 1881
37b3b608 1882=head2 Using Unicode in XS
c349b1b9 1883
37b3b608
KW
1884See L<perlguts/"Unicode Support"> for an introduction to Unicode at
1885the XS level, and L<perlapi/Unicode Support> for the API details.
95a1a48b 1886
e1b711da
KW
1887=head2 Hacking Perl to work on earlier Unicode versions (for very serious hackers only)
1888
a6a7eedc
KW
1889Perl by default comes with the latest supported Unicode version built-in, but
1890the goal is to allow you to change to use any earlier one. In Perls
1891v5.20 and v5.22, however, the earliest usable version is Unicode 5.1.
c55dd03d 1892Perl v5.18 and v5.24 are able to handle all earlier versions.
e1b711da 1893
42581d5d 1894Download the files in the desired version of Unicode from the Unicode web
e1b711da 1895site L<http://www.unicode.org>). These should replace the existing files in
b19eb496 1896F<lib/unicore> in the Perl source tree. Follow the instructions in
116693e8 1897F<README.perl> in that directory to change some of their names, and then build
26e391dd 1898perl (see L<INSTALL>).
116693e8 1899
c8d992ba
A
1900=head2 Porting code from perl-5.6.X
1901
a6a7eedc
KW
1902Perls starting in 5.8 have a different Unicode model from 5.6. In 5.6 the
1903programmer was required to use the C<utf8> pragma to declare that a
1904given scope expected to deal with Unicode data and had to make sure that
1905only Unicode data were reaching that scope. If you have code that is
c8d992ba 1906working with 5.6, you will need some of the following adjustments to
a6a7eedc
KW
1907your code. The examples are written such that the code will continue to
1908work under 5.6, so you should be safe to try them out.
c8d992ba 1909
755789c0 1910=over 3
c8d992ba
A
1911
1912=item *
1913
1914A filehandle that should read or write UTF-8
1915
b9cedb1b 1916 if ($] > 5.008) {
6d8e7450 1917 binmode $fh, ":encoding(UTF-8)";
c8d992ba
A
1918 }
1919
1920=item *
1921
1922A scalar that is going to be passed to some extension
1923
a9130ea9 1924Be it C<Compress::Zlib>, C<Apache::Request> or any extension that has no
c8d992ba 1925mention of Unicode in the manpage, you need to make sure that the
2575c402 1926UTF8 flag is stripped off. Note that at the time of this writing
b9cedb1b 1927(January 2012) the mentioned modules are not UTF-8-aware. Please
c8d992ba
A
1928check the documentation to verify if this is still true.
1929
b9cedb1b 1930 if ($] > 5.008) {
c8d992ba 1931 require Encode;
8e179dd8 1932 $val = Encode::encode("UTF-8", $val); # make octets
c8d992ba
A
1933 }
1934
1935=item *
1936
1937A scalar we got back from an extension
1938
1939If you believe the scalar comes back as UTF-8, you will most likely
2575c402 1940want the UTF8 flag restored:
c8d992ba 1941
b9cedb1b 1942 if ($] > 5.008) {
c8d992ba 1943 require Encode;
8e179dd8 1944 $val = Encode::decode("UTF-8", $val);
c8d992ba
A
1945 }
1946
1947=item *
1948
1949Same thing, if you are really sure it is UTF-8
1950
b9cedb1b 1951 if ($] > 5.008) {
c8d992ba
A
1952 require Encode;
1953 Encode::_utf8_on($val);
1954 }
1955
1956=item *
1957
a9130ea9 1958A wrapper for L<DBI> C<fetchrow_array> and C<fetchrow_hashref>
c8d992ba
A
1959
1960When the database contains only UTF-8, a wrapper function or method is
a9130ea9
KW
1961a convenient way to replace all your C<fetchrow_array> and
1962C<fetchrow_hashref> calls. A wrapper function will also make it easier to
c8d992ba 1963adapt to future enhancements in your database driver. Note that at the
b9cedb1b 1964time of this writing (January 2012), the DBI has no standardized way
a9130ea9 1965to deal with UTF-8 data. Please check the L<DBI documentation|DBI> to verify if
c8d992ba
A
1966that is still true.
1967
1968 sub fetchrow {
d88362ca
KW
1969 # $what is one of fetchrow_{array,hashref}
1970 my($self, $sth, $what) = @_;
b9cedb1b 1971 if ($] < 5.008) {
c8d992ba
A
1972 return $sth->$what;
1973 } else {
1974 require Encode;
1975 if (wantarray) {
1976 my @arr = $sth->$what;
1977 for (@arr) {
1978 defined && /[^\000-\177]/ && Encode::_utf8_on($_);
1979 }
1980 return @arr;
1981 } else {
1982 my $ret = $sth->$what;
1983 if (ref $ret) {
1984 for my $k (keys %$ret) {
d88362ca
KW
1985 defined
1986 && /[^\000-\177]/
1987 && Encode::_utf8_on($_) for $ret->{$k};
c8d992ba
A
1988 }
1989 return $ret;
1990 } else {
1991 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret;
1992 return $ret;
1993 }
1994 }
1995 }
1996 }
1997
1998
1999=item *
2000
2001A large scalar that you know can only contain ASCII
2002
2003Scalars that contain only ASCII and are marked as UTF-8 are sometimes
2004a drag to your program. If you recognize such a situation, just remove
2575c402 2005the UTF8 flag:
c8d992ba 2006
b9cedb1b 2007 utf8::downgrade($val) if $] > 5.008;
c8d992ba
A
2008
2009=back
2010
a6a7eedc
KW
2011=head1 BUGS
2012
2013See also L</The "Unicode Bug"> above.
2014
2015=head2 Interaction with Extensions
2016
2017When Perl exchanges data with an extension, the extension should be
2018able to understand the UTF8 flag and act accordingly. If the
2019extension doesn't recognize that flag, it's likely that the extension
2020will return incorrectly-flagged data.
2021
2022So if you're working with Unicode data, consult the documentation of
2023every module you're using if there are any issues with Unicode data
2024exchange. If the documentation does not talk about Unicode at all,
2025suspect the worst and probably look at the source to learn how the
2026module is implemented. Modules written completely in Perl shouldn't
2027cause problems. Modules that directly or indirectly access code written
2028in other programming languages are at risk.
2029
2030For affected functions, the simple strategy to avoid data corruption is
2031to always make the encoding of the exchanged data explicit. Choose an
2032encoding that you know the extension can handle. Convert arguments passed
2033to the extensions to that encoding and convert results back from that
2034encoding. Write wrapper functions that do the conversions for you, so
2035you can later change the functions when the extension catches up.
2036
2037To provide an example, let's say the popular C<Foo::Bar::escape_html>
2038function doesn't deal with Unicode data yet. The wrapper function
2039would convert the argument to raw UTF-8 and convert the result back to
2040Perl's internal representation like so:
2041
2042 sub my_escape_html ($) {
2043 my($what) = shift;
2044 return unless defined $what;
8e179dd8
P
2045 Encode::decode("UTF-8", Foo::Bar::escape_html(
2046 Encode::encode("UTF-8", $what)));
a6a7eedc
KW
2047 }
2048
2049Sometimes, when the extension does not convert data but just stores
2050and retrieves it, you will be able to use the otherwise
2051dangerous L<C<Encode::_utf8_on()>|Encode/_utf8_on> function. Let's say
2052the popular C<Foo::Bar> extension, written in C, provides a C<param>
2053method that lets you store and retrieve data according to these prototypes:
2054
2055 $self->param($name, $value); # set a scalar
2056 $value = $self->param($name); # retrieve a scalar
2057
2058If it does not yet provide support for any encoding, one could write a
2059derived class with such a C<param> method:
2060
2061 sub param {
2062 my($self,$name,$value) = @_;
2063 utf8::upgrade($name); # make sure it is UTF-8 encoded
2064 if (defined $value) {
2065 utf8::upgrade($value); # make sure it is UTF-8 encoded
2066 return $self->SUPER::param($name,$value);
2067 } else {
2068 my $ret = $self->SUPER::param($name);
2069 Encode::_utf8_on($ret); # we know, it is UTF-8 encoded
2070 return $ret;
2071 }
2072 }
2073
2074Some extensions provide filters on data entry/exit points, such as
2075C<DB_File::filter_store_key> and family. Look out for such filters in
2076the documentation of your extensions; they can make the transition to
2077Unicode data much easier.
2078
2079=head2 Speed
2080
2081Some functions are slower when working on UTF-8 encoded strings than
2082on byte encoded strings. All functions that need to hop over
2083characters such as C<length()>, C<substr()> or C<index()>, or matching
2084regular expressions can work B<much> faster when the underlying data are
2085byte-encoded.
2086
2087In Perl 5.8.0 the slowness was often quite spectacular; in Perl 5.8.1
2088a caching scheme was introduced which improved the situation. In general,
2089operations with UTF-8 encoded strings are still slower. As an example,
2090the Unicode properties (character classes) like C<\p{Nd}> are known to
2091be quite a bit slower (5-20 times) than their simpler counterparts
2092like C<[0-9]> (then again, there are hundreds of Unicode characters matching
2093C<Nd> compared with the 10 ASCII characters matching C<[0-9]>).
2094
393fec97
GS
2095=head1 SEE ALSO
2096
51f494cc 2097L<perlunitut>, L<perluniintro>, L<perluniprops>, L<Encode>, L<open>, L<utf8>, L<bytes>,
b65e6125 2098L<perlretut>, L<perlvar/"${^UNICODE}">,
51f494cc 2099L<http://www.unicode.org/reports/tr44>).
393fec97
GS
2100
2101=cut