This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
refine the skip test for the 32-bit x86 ABI brokeness
[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
9d1c51c1
KW
39Also, the use of Unicode may present security issues that aren't obvious.
40Read L<Unicode Security Considerations|http://www.unicode.org/reports/tr36>.
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
a6a7eedc
KW
63=item You should convert your non-ASCII, non-UTF-8 Perl scripts to be
64UTF-8.
21bad921 65
a6a7eedc 66See L<encoding>.
21bad921 67
a6a7eedc 68=item C<use utf8> still needed to enable L<UTF-8|/Unicode Encodings> in scripts
21bad921 69
a6a7eedc
KW
70If your Perl script is itself encoded in L<UTF-8|/Unicode Encodings>,
71the S<C<use utf8>> pragma must be explicitly included to enable
72recognition of that (in string or regular expression literals, or in
73identifier names). B<This is the only time when an explicit S<C<use
74utf8>> is needed.> (See L<utf8>).
7aa207d6 75
a6a7eedc 76=item C<BOM>-marked scripts and L<UTF-16|/Unicode Encodings> scripts autodetected
7aa207d6 77
a6a7eedc
KW
78However, if a Perl script begins with the Unicode C<BOM> (UTF-16LE,
79UTF16-BE, or UTF-8), or if the script looks like non-C<BOM>-marked
80UTF-16 of either endianness, Perl will correctly read in the script as
81the appropriate Unicode encoding. (C<BOM>-less UTF-8 cannot be
82effectively recognized or differentiated from ISO 8859-1 or other
83eight-bit encodings.)
990e18f7 84
21bad921
GS
85=back
86
376d9008 87=head2 Byte and Character Semantics
393fec97 88
a6a7eedc
KW
89Before Unicode, most encodings used 8 bits (a single byte) to encode
90each character. Thus a character was a byte, and a byte was a
91character, and there could be only 256 or fewer possible characters.
92"Byte Semantics" in the title of this section refers to
93this behavior. There was no need to distinguish between "Byte" and
94"Character".
95
96Then along comes Unicode which has room for over a million characters
97(and Perl allows for even more). This means that a character may
98require more than a single byte to represent it, and so the two terms
99are no longer equivalent. What matter are the characters as whole
100entities, and not usually the bytes that comprise them. That's what the
101term "Character Semantics" in the title of this section refers to.
102
103Perl had to change internally to decouple "bytes" from "characters".
104It is important that you too change your ideas, if you haven't already,
105so that "byte" and "character" no longer mean the same thing in your
106mind.
107
108The basic building block of Perl strings has always been a "character".
109The changes basically come down to that the implementation no longer
110thinks that a character is always just a single byte.
111
112There are various things to note:
393fec97
GS
113
114=over 4
115
116=item *
117
a6a7eedc
KW
118String handling functions, for the most part, continue to operate in
119terms of characters. C<length()>, for example, returns the number of
120characters in a string, just as before. But that number no longer is
121necessarily the same as the number of bytes in the string (there may be
122more bytes than characters). The other such functions include
123C<chop()>, C<chomp()>, C<substr()>, C<pos()>, C<index()>, C<rindex()>,
124C<sort()>, C<sprintf()>, and C<write()>.
125
126The exceptions are:
127
128=over 4
129
130=item *
131
132the bit-oriented C<vec>
133
134E<nbsp>
135
136=item *
137
138the byte-oriented C<pack>/C<unpack> C<"C"> format
139
140However, the C<W> specifier does operate on whole characters, as does the
141C<U> specifier.
142
143=item *
144
145some operators that interact with the platform's operating system
146
147Operators dealing with filenames are examples.
148
149=item *
150
151when the functions are called from within the scope of the
152S<C<L<use bytes|bytes>>> pragma
153
154Likely, you should use this only for debugging anyway.
155
156=back
157
158=item *
159
376d9008 160Strings--including hash keys--and regular expression patterns may
b65e6125 161contain characters that have ordinal values larger than 255.
393fec97 162
2575c402
JW
163If you use a Unicode editor to edit your program, Unicode characters may
164occur directly within the literal strings in UTF-8 encoding, or UTF-16.
a9130ea9 165(The former requires a C<BOM> or C<use utf8>, the latter requires a C<BOM>.)
3e4dbfed 166
a6a7eedc
KW
167L<perluniintro/Creating Unicode> gives other ways to place non-ASCII
168characters in your strings.
6f335b04 169
a6a7eedc 170=item *
fbb93542 171
a6a7eedc 172The C<chr()> and C<ord()> functions work on whole characters.
376d9008 173
393fec97
GS
174=item *
175
a6a7eedc
KW
176Regular expressions match whole characters. For example, C<"."> matches
177a whole character instead of only a single byte.
393fec97 178
393fec97
GS
179=item *
180
a6a7eedc
KW
181The C<tr///> operator translates whole characters. (Note that the
182C<tr///CU> functionality has been removed. For similar functionality to
183that, see C<pack('U0', ...)> and C<pack('C0', ...)>).
393fec97 184
393fec97
GS
185=item *
186
a6a7eedc 187C<scalar reverse()> reverses by character rather than by byte.
393fec97 188
393fec97
GS
189=item *
190
a6a7eedc
KW
191The bit string operators, C<& | ^ ~> and (starting in v5.22)
192C<&. |. ^. ~.> can operate on characters that don't fit into a byte.
193However, the current behavior is likely to change. You should not use
194these operators on strings that are encoded in UTF-8. If you're not
195sure about the encoding of a string, downgrade it before using any of
196these operators; you can use
197L<C<utf8::utf8_downgrade()>|utf8/Utility functions>.
822502e5 198
a6a7eedc 199=back
822502e5 200
a6a7eedc
KW
201The bottom line is that Perl has always practiced "Character Semantics",
202but with the advent of Unicode, that is now different than "Byte
203Semantics".
204
205=head2 ASCII Rules versus Unicode Rules
206
207Before Unicode, when a character was a byte was a character,
208Perl knew only about the 128 characters defined by ASCII, code points 0
209through 127 (except for under S<C<use locale>>). That left the code
210points 128 to 255 as unassigned, and available for whatever use a
211program might want. The only semantics they have is their ordinal
212numbers, and that they are members of none of the non-negative character
213classes. None are considered to match C<\w> for example, but all match
214C<\W>.
822502e5 215
a6a7eedc
KW
216Unicode, of course, assigns each of those code points a particular
217meaning (along with ones above 255). To preserve backward
218compatibility, Perl only uses the Unicode meanings when there is some
219indication that Unicode is what is intended; otherwise the non-ASCII
220code points remain treated as if they are unassigned.
221
222Here are the ways that Perl knows that a string should be treated as
223Unicode:
224
225=over
822502e5
ST
226
227=item *
228
a6a7eedc
KW
229Within the scope of S<C<use utf8>>
230
231If the whole program is Unicode (signified by using 8-bit B<U>nicode
232B<T>ransformation B<F>ormat), then all strings within it must be
233Unicode.
822502e5
ST
234
235=item *
236
a6a7eedc
KW
237Within the scope of
238L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
239
240This pragma was created so you can explicitly tell Perl that operations
241executed within its scope are to use Unicode rules. More operations are
242affected with newer perls. See L</The "Unicode Bug">.
822502e5
ST
243
244=item *
245
a6a7eedc
KW
246Within the scope of S<C<use 5.012>> or higher
247
248This implicitly turns on S<C<use feature 'unicode_strings'>>.
822502e5
ST
249
250=item *
251
a6a7eedc
KW
252Within the scope of
253L<S<C<use locale 'not_characters'>>|perllocale/Unicode and UTF-8>,
254or L<S<C<use locale>>|perllocale> and the current
255locale is a UTF-8 locale.
822502e5 256
a6a7eedc
KW
257The former is defined to imply Unicode handling; and the latter
258indicates a Unicode locale, hence a Unicode interpretation of all
259strings within it.
822502e5
ST
260
261=item *
262
a6a7eedc
KW
263When the string contains a Unicode-only code point
264
265Perl has never accepted code points above 255 without them being
266Unicode, so their use implies Unicode for the whole string.
822502e5
ST
267
268=item *
269
a6a7eedc
KW
270When the string contains a Unicode named code point C<\N{...}>
271
272The C<\N{...}> construct explicitly refers to a Unicode code point,
273even if it is one that is also in ASCII. Therefore the string
274containing it must be Unicode.
822502e5
ST
275
276=item *
277
a6a7eedc
KW
278When the string has come from an external source marked as
279Unicode
280
281The L<C<-C>|perlrun/-C [numberE<sol>list]> command line option can
282specify that certain inputs to the program are Unicode, and the values
283of this can be read by your Perl code, see L<perlvar/"${^UNICODE}">.
284
285=item * When the string has been upgraded to UTF-8
286
287The function L<C<utf8::utf8_upgrade()>|utf8/Utility functions>
288can be explicitly used to permanently (unless a subsequent
289C<utf8::utf8_downgrade()> is called) cause a string to be treated as
290Unicode.
291
292=item * There are additional methods for regular expression patterns
293
f6cf4627
KW
294A pattern that is compiled with the C<< /u >> or C<< /a >> modifiers is
295treated as Unicode (though there are some restrictions with C<< /a >>).
296Under the C<< /d >> and C<< /l >> modifiers, there are several other
297indications for Unicode; see L<perlre/Character set modifiers>.
822502e5
ST
298
299=back
300
a6a7eedc
KW
301Note that all of the above are overridden within the scope of
302C<L<use bytes|bytes>>; but you should be using this pragma only for
303debugging.
304
305Note also that some interactions with the platform's operating system
306never use Unicode rules.
307
308When Unicode rules are in effect:
309
822502e5
ST
310=over 4
311
312=item *
313
a6a7eedc
KW
314Case translation operators use the Unicode case translation tables.
315
316Note that C<uc()>, or C<\U> in interpolated strings, translates to
317uppercase, while C<ucfirst>, or C<\u> in interpolated strings,
318translates to titlecase in languages that make the distinction (which is
319equivalent to uppercase in languages without the distinction).
320
321There is a CPAN module, C<L<Unicode::Casing>>, which allows you to
322define your own mappings to be used in C<lc()>, C<lcfirst()>, C<uc()>,
323C<ucfirst()>, and C<fc> (or their double-quoted string inlined versions
324such as C<\U>). (Prior to Perl 5.16, this functionality was partially
325provided in the Perl core, but suffered from a number of insurmountable
326drawbacks, so the CPAN module was written instead.)
327
328=item *
329
330Character classes in regular expressions match based on the character
331properties specified in the Unicode properties database.
332
333C<\w> can be used to match a Japanese ideograph, for instance; and
334C<[[:digit:]]> a Bengali number.
335
336=item *
337
338Named Unicode properties, scripts, and block ranges may be used (like
339bracketed character classes) by using the C<\p{}> "matches property"
340construct and the C<\P{}> negation, "doesn't match property".
341
342See L</"Unicode Character Properties"> for more details.
343
344You can define your own character properties and use them
345in the regular expression with the C<\p{}> or C<\P{}> construct.
346See L</"User-Defined Character Properties"> for more details.
822502e5
ST
347
348=back
349
a6a7eedc
KW
350=head2 Extended Grapheme Clusters (Logical characters)
351
352Consider a character, say C<H>. It could appear with various marks around it,
353such as an acute accent, or a circumflex, or various hooks, circles, arrows,
354I<etc.>, above, below, to one side or the other, I<etc>. There are many
355possibilities among the world's languages. The number of combinations is
356astronomical, and if there were a character for each combination, it would
357soon exhaust Unicode's more than a million possible characters. So Unicode
358took a different approach: there is a character for the base C<H>, and a
359character for each of the possible marks, and these can be variously combined
360to get a final logical character. So a logical character--what appears to be a
361single character--can be a sequence of more than one individual characters.
362The Unicode standard calls these "extended grapheme clusters" (which
363is an improved version of the no-longer much used "grapheme cluster");
364Perl furnishes the C<\X> regular expression construct to match such
365sequences in their entirety.
366
367But Unicode's intent is to unify the existing character set standards and
368practices, and several pre-existing standards have single characters that
369mean the same thing as some of these combinations, like ISO-8859-1,
370which has quite a few of them. For example, C<"LATIN CAPITAL LETTER E
371WITH ACUTE"> was already in this standard when Unicode came along.
372Unicode therefore added it to its repertoire as that single character.
373But this character is considered by Unicode to be equivalent to the
374sequence consisting of the character C<"LATIN CAPITAL LETTER E">
375followed by the character C<"COMBINING ACUTE ACCENT">.
376
377C<"LATIN CAPITAL LETTER E WITH ACUTE"> is called a "pre-composed"
378character, and its equivalence with the "E" and the "COMBINING ACCENT"
379sequence is called canonical equivalence. All pre-composed characters
380are said to have a decomposition (into the equivalent sequence), and the
381decomposition type is also called canonical. A string may be comprised
382as much as possible of precomposed characters, or it may be comprised of
383entirely decomposed characters. Unicode calls these respectively,
384"Normalization Form Composed" (NFC) and "Normalization Form Decomposed".
385The C<L<Unicode::Normalize>> module contains functions that convert
386between the two. A string may also have both composed characters and
387decomposed characters; this module can be used to make it all one or the
388other.
389
390You may be presented with strings in any of these equivalent forms.
391There is currently nothing in Perl 5 that ignores the differences. So
392you'll have to specially hanlde it. The usual advice is to convert your
393inputs to C<NFD> before processing further.
394
395For more detailed information, see L<http://unicode.org/reports/tr15/>.
396
822502e5
ST
397=head2 Unicode Character Properties
398
ee88f7b6 399(The only time that Perl considers a sequence of individual code
9d1c51c1
KW
400points as a single logical character is in the C<\X> construct, already
401mentioned above. Therefore "character" in this discussion means a single
ee88f7b6
KW
402Unicode code point.)
403
404Very nearly all Unicode character properties are accessible through
405regular expressions by using the C<\p{}> "matches property" construct
406and the C<\P{}> "doesn't match property" for its negation.
51f494cc 407
9d1c51c1 408For instance, C<\p{Uppercase}> matches any single character with the Unicode
a9130ea9
KW
409C<"Uppercase"> property, while C<\p{L}> matches any character with a
410C<General_Category> of C<"L"> (letter) property (see
411L</General_Category> below). Brackets are not
9d1c51c1 412required for single letter property names, so C<\p{L}> is equivalent to C<\pL>.
51f494cc 413
9d1c51c1 414More formally, C<\p{Uppercase}> matches any single character whose Unicode
a9130ea9
KW
415C<Uppercase> property value is C<True>, and C<\P{Uppercase}> matches any character
416whose C<Uppercase> property value is C<False>, and they could have been written as
9d1c51c1 417C<\p{Uppercase=True}> and C<\p{Uppercase=False}>, respectively.
51f494cc 418
b19eb496 419This formality is needed when properties are not binary; that is, if they can
a9130ea9
KW
420take on more values than just C<True> and C<False>. For example, the
421C<Bidi_Class> property (see L</"Bidirectional Character Types"> below),
422can take on several different
423values, such as C<Left>, C<Right>, C<Whitespace>, and others. To match these, one needs
424to specify both the property name (C<Bidi_Class>), AND the value being
5bff2035 425matched against
b65e6125 426(C<Left>, C<Right>, I<etc.>). This is done, as in the examples above, by having the
9f815e24 427two components separated by an equal sign (or interchangeably, a colon), like
51f494cc
KW
428C<\p{Bidi_Class: Left}>.
429
430All Unicode-defined character properties may be written in these compound forms
a9130ea9 431of C<\p{I<property>=I<value>}> or C<\p{I<property>:I<value>}>, but Perl provides some
51f494cc
KW
432additional properties that are written only in the single form, as well as
433single-form short-cuts for all binary properties and certain others described
434below, in which you may omit the property name and the equals or colon
435separator.
436
437Most Unicode character properties have at least two synonyms (or aliases if you
b19eb496 438prefer): a short one that is easier to type and a longer one that is more
a9130ea9
KW
439descriptive and hence easier to understand. Thus the C<"L"> and
440C<"Letter"> properties above are equivalent and can be used
441interchangeably. Likewise, C<"Upper"> is a synonym for C<"Uppercase">,
442and we could have written C<\p{Uppercase}> equivalently as C<\p{Upper}>.
443Also, there are typically various synonyms for the values the property
444can be. For binary properties, C<"True"> has 3 synonyms: C<"T">,
445C<"Yes">, and C<"Y">; and C<"False"> has correspondingly C<"F">,
446C<"No">, and C<"N">. But be careful. A short form of a value for one
447property may not mean the same thing as the same short form for another.
448Thus, for the C<L</General_Category>> property, C<"L"> means
449C<"Letter">, but for the L<C<Bidi_Class>|/Bidirectional Character Types>
450property, C<"L"> means C<"Left">. A complete list of properties and
451synonyms is in L<perluniprops>.
51f494cc 452
b19eb496 453Upper/lower case differences in property names and values are irrelevant;
51f494cc
KW
454thus C<\p{Upper}> means the same thing as C<\p{upper}> or even C<\p{UpPeR}>.
455Similarly, you can add or subtract underscores anywhere in the middle of a
456word, so that these are also equivalent to C<\p{U_p_p_e_r}>. And white space
457is irrelevant adjacent to non-word characters, such as the braces and the equals
b19eb496
TC
458or colon separators, so C<\p{ Upper }> and C<\p{ Upper_case : Y }> are
459equivalent to these as well. In fact, white space and even
460hyphens can usually be added or deleted anywhere. So even C<\p{ Up-per case = Yes}> is
51f494cc 461equivalent. All this is called "loose-matching" by Unicode. The few places
b19eb496 462where stricter matching is used is in the middle of numbers, and in the Perl
51f494cc 463extension properties that begin or end with an underscore. Stricter matching
b19eb496 464cares about white space (except adjacent to non-word characters),
51f494cc 465hyphens, and non-interior underscores.
4193bef7 466
376d9008 467You can also use negation in both C<\p{}> and C<\P{}> by introducing a caret
a9130ea9 468(C<^>) between the first brace and the property name: C<\p{^Tamil}> is
eb0cc9e3 469equal to C<\P{Tamil}>.
4193bef7 470
56ca34ca
KW
471Almost all properties are immune to case-insensitive matching. That is,
472adding a C</i> regular expression modifier does not change what they
473match. There are two sets that are affected.
474The first set is
475C<Uppercase_Letter>,
476C<Lowercase_Letter>,
477and C<Titlecase_Letter>,
478all of which match C<Cased_Letter> under C</i> matching.
479And the second set is
480C<Uppercase>,
481C<Lowercase>,
482and C<Titlecase>,
483all of which match C<Cased> under C</i> matching.
484This set also includes its subsets C<PosixUpper> and C<PosixLower> both
a9130ea9 485of which under C</i> match C<PosixAlpha>.
56ca34ca 486(The difference between these sets is that some things, such as Roman
b65e6125
KW
487numerals, come in both upper and lower case so they are C<Cased>, but
488aren't considered letters, so they aren't C<Cased_Letter>'s.)
56ca34ca 489
2d88a86a
KW
490See L</Beyond Unicode code points> for special considerations when
491matching Unicode properties against non-Unicode code points.
94b42e47 492
51f494cc 493=head3 B<General_Category>
14bb0a9a 494
51f494cc
KW
495Every Unicode character is assigned a general category, which is the "most
496usual categorization of a character" (from
497L<http://www.unicode.org/reports/tr44>).
822502e5 498
9f815e24 499The compound way of writing these is like C<\p{General_Category=Number}>
b65e6125 500(short: C<\p{gc:n}>). But Perl furnishes shortcuts in which everything up
51f494cc
KW
501through the equal or colon separator is omitted. So you can instead just write
502C<\pN>.
822502e5 503
a9130ea9
KW
504Here are the short and long forms of the values the C<General Category> property
505can have:
393fec97 506
d73e5302
JH
507 Short Long
508
509 L Letter
51f494cc
KW
510 LC, L& Cased_Letter (that is: [\p{Ll}\p{Lu}\p{Lt}])
511 Lu Uppercase_Letter
512 Ll Lowercase_Letter
513 Lt Titlecase_Letter
514 Lm Modifier_Letter
515 Lo Other_Letter
d73e5302
JH
516
517 M Mark
51f494cc
KW
518 Mn Nonspacing_Mark
519 Mc Spacing_Mark
520 Me Enclosing_Mark
d73e5302
JH
521
522 N Number
51f494cc
KW
523 Nd Decimal_Number (also Digit)
524 Nl Letter_Number
525 No Other_Number
526
527 P Punctuation (also Punct)
528 Pc Connector_Punctuation
529 Pd Dash_Punctuation
530 Ps Open_Punctuation
531 Pe Close_Punctuation
532 Pi Initial_Punctuation
d73e5302 533 (may behave like Ps or Pe depending on usage)
51f494cc 534 Pf Final_Punctuation
d73e5302 535 (may behave like Ps or Pe depending on usage)
51f494cc 536 Po Other_Punctuation
d73e5302
JH
537
538 S Symbol
51f494cc
KW
539 Sm Math_Symbol
540 Sc Currency_Symbol
541 Sk Modifier_Symbol
542 So Other_Symbol
d73e5302
JH
543
544 Z Separator
51f494cc
KW
545 Zs Space_Separator
546 Zl Line_Separator
547 Zp Paragraph_Separator
d73e5302
JH
548
549 C Other
d88362ca 550 Cc Control (also Cntrl)
e150c829 551 Cf Format
6d4f9cf2 552 Cs Surrogate
51f494cc 553 Co Private_Use
e150c829 554 Cn Unassigned
1ac13f9a 555
376d9008 556Single-letter properties match all characters in any of the
3e4dbfed 557two-letter sub-properties starting with the same letter.
b19eb496 558C<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 559
51f494cc 560=head3 B<Bidirectional Character Types>
822502e5 561
b19eb496 562Because scripts differ in their directionality (Hebrew and Arabic are
a9130ea9 563written right to left, for example) Unicode supplies a C<Bidi_Class> property.
1850f57f 564Some of the values this property can have are:
32293815 565
88af3b93 566 Value Meaning
92e830a9 567
12ac2576
JP
568 L Left-to-Right
569 LRE Left-to-Right Embedding
570 LRO Left-to-Right Override
571 R Right-to-Left
51f494cc 572 AL Arabic Letter
12ac2576
JP
573 RLE Right-to-Left Embedding
574 RLO Right-to-Left Override
575 PDF Pop Directional Format
576 EN European Number
51f494cc
KW
577 ES European Separator
578 ET European Terminator
12ac2576 579 AN Arabic Number
51f494cc 580 CS Common Separator
12ac2576
JP
581 NSM Non-Spacing Mark
582 BN Boundary Neutral
583 B Paragraph Separator
584 S Segment Separator
585 WS Whitespace
586 ON Other Neutrals
587
51f494cc
KW
588This property is always written in the compound form.
589For example, C<\p{Bidi_Class:R}> matches characters that are normally
1850f57f 590written right to left. Unlike the
a9130ea9 591C<L</General_Category>> property, this
1850f57f
KW
592property can have more values added in a future Unicode release. Those
593listed above comprised the complete set for many Unicode releases, but
594others were added in Unicode 6.3; you can always find what the
20ada7da 595current ones are in L<perluniprops>. And
1850f57f 596L<http://www.unicode.org/reports/tr9/> describes how to use them.
eb0cc9e3 597
51f494cc
KW
598=head3 B<Scripts>
599
b19eb496 600The world's languages are written in many different scripts. This sentence
e1b711da 601(unless you're reading it in translation) is written in Latin, while Russian is
c69ca1d4 602written in Cyrillic, and Greek is written in, well, Greek; Japanese mainly in
e1b711da 603Hiragana or Katakana. There are many more.
51f494cc 604
b65e6125 605The Unicode C<Script> and C<Script_Extensions> properties give what script a
82aed44a
KW
606given character is in. Either property can be specified with the
607compound form like
608C<\p{Script=Hebrew}> (short: C<\p{sc=hebr}>), or
609C<\p{Script_Extensions=Javanese}> (short: C<\p{scx=java}>).
610In addition, Perl furnishes shortcuts for all
611C<Script> property names. You can omit everything up through the equals
612(or colon), and simply write C<\p{Latin}> or C<\P{Cyrillic}>.
613(This is not true for C<Script_Extensions>, which is required to be
614written in the compound form.)
615
616The difference between these two properties involves characters that are
617used in multiple scripts. For example the digits '0' through '9' are
618used in many parts of the world. These are placed in a script named
619C<Common>. Other characters are used in just a few scripts. For
a9130ea9 620example, the C<"KATAKANA-HIRAGANA DOUBLE HYPHEN"> is used in both Japanese
82aed44a
KW
621scripts, Katakana and Hiragana, but nowhere else. The C<Script>
622property places all characters that are used in multiple scripts in the
623C<Common> script, while the C<Script_Extensions> property places those
624that are used in only a few scripts into each of those scripts; while
625still using C<Common> for those used in many scripts. Thus both these
626match:
627
628 "0" =~ /\p{sc=Common}/ # Matches
629 "0" =~ /\p{scx=Common}/ # Matches
630
631and only the first of these match:
632
633 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Common} # Matches
634 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Common} # No match
635
636And only the last two of these match:
637
638 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Hiragana} # No match
639 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{sc=Katakana} # No match
640 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Hiragana} # Matches
641 "\N{KATAKANA-HIRAGANA DOUBLE HYPHEN}" =~ /\p{scx=Katakana} # Matches
642
643C<Script_Extensions> is thus an improved C<Script>, in which there are
644fewer characters in the C<Common> script, and correspondingly more in
645other scripts. It is new in Unicode version 6.0, and its data are likely
646to change significantly in later releases, as things get sorted out.
b65e6125
KW
647New code should probably be using C<Script_Extensions> and not plain
648C<Script>.
82aed44a
KW
649
650(Actually, besides C<Common>, the C<Inherited> script, contains
651characters that are used in multiple scripts. These are modifier
b65e6125 652characters which inherit the script value
82aed44a
KW
653of the controlling character. Some of these are used in many scripts,
654and so go into C<Inherited> in both C<Script> and C<Script_Extensions>.
655Others are used in just a few scripts, so are in C<Inherited> in
656C<Script>, but not in C<Script_Extensions>.)
657
658It is worth stressing that there are several different sets of digits in
659Unicode that are equivalent to 0-9 and are matchable by C<\d> in a
660regular expression. If they are used in a single language only, they
661are in that language's C<Script> and C<Script_Extension>. If they are
662used in more than one script, they will be in C<sc=Common>, but only
663if they are used in many scripts should they be in C<scx=Common>.
51f494cc
KW
664
665A complete list of scripts and their shortcuts is in L<perluniprops>.
666
a9130ea9 667=head3 B<Use of the C<"Is"> Prefix>
822502e5 668
b65e6125
KW
669For backward compatibility (with Perl 5.6), all properties writable
670without using the compound form mentioned
51f494cc
KW
671so far may have C<Is> or C<Is_> prepended to their name, so C<\P{Is_Lu}>, for
672example, is equal to C<\P{Lu}>, and C<\p{IsScript:Arabic}> is equal to
673C<\p{Arabic}>.
eb0cc9e3 674
51f494cc 675=head3 B<Blocks>
2796c109 676
1bfb14c4
JH
677In addition to B<scripts>, Unicode also defines B<blocks> of
678characters. The difference between scripts and blocks is that the
679concept of scripts is closer to natural languages, while the concept
51f494cc 680of blocks is more of an artificial grouping based on groups of Unicode
a9130ea9 681characters with consecutive ordinal values. For example, the C<"Basic Latin">
b65e6125 682block is all the characters whose ordinals are between 0 and 127, inclusive; in
a9130ea9
KW
683other words, the ASCII characters. The C<"Latin"> script contains some letters
684from this as well as several other blocks, like C<"Latin-1 Supplement">,
b65e6125 685C<"Latin Extended-A">, I<etc.>, but it does not contain all the characters from
7be67b37 686those blocks. It does not, for example, contain the digits 0-9, because
82aed44a
KW
687those digits are shared across many scripts, and hence are in the
688C<Common> script.
51f494cc
KW
689
690For more about scripts versus blocks, see UAX#24 "Unicode Script Property":
691L<http://www.unicode.org/reports/tr24>
692
82aed44a
KW
693The C<Script> or C<Script_Extensions> properties are likely to be the
694ones you want to use when processing
a9130ea9 695natural language; the C<Block> property may occasionally be useful in working
b19eb496 696with the nuts and bolts of Unicode.
51f494cc
KW
697
698Block names are matched in the compound form, like C<\p{Block: Arrows}> or
b19eb496 699C<\p{Blk=Hebrew}>. Unlike most other properties, only a few block names have a
a6a7eedc
KW
700Unicode-defined short name. But Perl does provide a (slight, no longer
701recommended) shortcut: You can say, for example C<\p{In_Arrows}> or
702C<\p{In_Hebrew}>.
703
704For backwards compatibility, the C<In> prefix may be
705omitted if there is no naming conflict with a script or any other
706property, and you can even use an C<Is> prefix instead in those cases.
707But don't do this for new code because your code could break in new
708releases, and this has already happened: There was a time in very
709early Unicode releases when C<\p{Hebrew}> would have matched the
710I<block> Hebrew; now it doesn't.
711
712Using the C<In> prefix avoids this ambiguity, so far. But new versions
713of Unicode continue to add new properties whose names begin with C<In>.
714There is a possibility that one of them someday will conflict with your
715usage. Since this is just a Perl extension, Unicode's name will take
716precedence and your code will become broken. Also, Unicode is free to
717add a script whose name begins with C<In>; that would cause problems.
718
719So it's clearer and best to use the compound form when specifying
720blocks. And be sure that is what you really really want to do. In most
721cases scripts are what you want instead.
51f494cc
KW
722
723A complete list of blocks and their shortcuts is in L<perluniprops>.
724
9f815e24
KW
725=head3 B<Other Properties>
726
727There are many more properties than the very basic ones described here.
728A complete list is in L<perluniprops>.
729
730Unicode defines all its properties in the compound form, so all single-form
b19eb496
TC
731properties are Perl extensions. Most of these are just synonyms for the
732Unicode ones, but some are genuine extensions, including several that are in
9f815e24
KW
733the compound form. And quite a few of these are actually recommended by Unicode
734(in L<http://www.unicode.org/reports/tr18>).
735
5bff2035
KW
736This section gives some details on all extensions that aren't just
737synonyms for compound-form Unicode properties
738(for those properties, you'll have to refer to the
9f815e24
KW
739L<Unicode Standard|http://www.unicode.org/reports/tr44>.
740
741=over
742
743=item B<C<\p{All}>>
744
2d88a86a
KW
745This matches every possible code point. It is equivalent to C<qr/./s>.
746Unlike all the other non-user-defined C<\p{}> property matches, no
747warning is ever generated if this is property is matched against a
748non-Unicode code point (see L</Beyond Unicode code points> below).
9f815e24
KW
749
750=item B<C<\p{Alnum}>>
751
752This matches any C<\p{Alphabetic}> or C<\p{Decimal_Number}> character.
753
754=item B<C<\p{Any}>>
755
2d88a86a
KW
756This matches any of the 1_114_112 Unicode code points. It is a synonym
757for C<\p{Unicode}>.
9f815e24 758
42581d5d
KW
759=item B<C<\p{ASCII}>>
760
761This matches any of the 128 characters in the US-ASCII character set,
762which is a subset of Unicode.
763
9f815e24
KW
764=item B<C<\p{Assigned}>>
765
a9130ea9
KW
766This matches any assigned code point; that is, any code point whose L<general
767category|/General_Category> is not C<Unassigned> (or equivalently, not C<Cn>).
9f815e24
KW
768
769=item B<C<\p{Blank}>>
770
771This is the same as C<\h> and C<\p{HorizSpace}>: A character that changes the
772spacing horizontally.
773
774=item B<C<\p{Decomposition_Type: Non_Canonical}>> (Short: C<\p{Dt=NonCanon}>)
775
776Matches a character that has a non-canonical decomposition.
777
a6a7eedc
KW
778The L</Extended Grapheme Clusters (Logical characters)> section above
779talked about canonical decompositions. However, many more characters
780have a different type of decomposition, a "compatible" or
781"non-canonical" decomposition. The sequences that form these
782decompositions are not considered canonically equivalent to the
783pre-composed character. An example is the C<"SUPERSCRIPT ONE">. It is
784somewhat like a regular digit 1, but not exactly; its decomposition into
785the digit 1 is called a "compatible" decomposition, specifically a
9f815e24 786"super" decomposition. There are several such compatibility
b65e6125
KW
787decompositions (see L<http://www.unicode.org/reports/tr44>), including
788one called "compat", which means some miscellaneous type of
789decomposition that doesn't fit into the other decomposition categories
790that Unicode has chosen.
9f815e24
KW
791
792Note that most Unicode characters don't have a decomposition, so their
a9130ea9 793decomposition type is C<"None">.
9f815e24 794
b19eb496
TC
795For your convenience, Perl has added the C<Non_Canonical> decomposition
796type to mean any of the several compatibility decompositions.
9f815e24
KW
797
798=item B<C<\p{Graph}>>
799
800Matches any character that is graphic. Theoretically, this means a character
801that on a printer would cause ink to be used.
802
803=item B<C<\p{HorizSpace}>>
804
b19eb496 805This is the same as C<\h> and C<\p{Blank}>: a character that changes the
9f815e24
KW
806spacing horizontally.
807
42581d5d 808=item B<C<\p{In=*}>>
9f815e24
KW
809
810This is a synonym for C<\p{Present_In=*}>
811
812=item B<C<\p{PerlSpace}>>
813
d28d8023 814This is the same as C<\s>, restricted to ASCII, namely C<S<[ \f\n\r\t]>>
779cf272 815and starting in Perl v5.18, a vertical tab.
9f815e24
KW
816
817Mnemonic: Perl's (original) space
818
819=item B<C<\p{PerlWord}>>
820
821This is the same as C<\w>, restricted to ASCII, namely C<[A-Za-z0-9_]>
822
823Mnemonic: Perl's (original) word.
824
42581d5d 825=item B<C<\p{Posix...}>>
9f815e24 826
b65e6125
KW
827There are several of these, which are equivalents, using the C<\p{}>
828notation, for Posix classes and are described in
42581d5d 829L<perlrecharclass/POSIX Character Classes>.
9f815e24
KW
830
831=item B<C<\p{Present_In: *}>> (Short: C<\p{In=*}>)
832
833This property is used when you need to know in what Unicode version(s) a
834character is.
835
836The "*" above stands for some two digit Unicode version number, such as
837C<1.1> or C<4.0>; or the "*" can also be C<Unassigned>. This property will
838match the code points whose final disposition has been settled as of the
839Unicode release given by the version number; C<\p{Present_In: Unassigned}>
840will match those code points whose meaning has yet to be assigned.
841
a9130ea9 842For example, C<U+0041> C<"LATIN CAPITAL LETTER A"> was present in the very first
9f815e24
KW
843Unicode release available, which is C<1.1>, so this property is true for all
844valid "*" versions. On the other hand, C<U+1EFF> was not assigned until version
a9130ea9 8455.1 when it became C<"LATIN SMALL LETTER Y WITH LOOP">, so the only "*" that
9f815e24
KW
846would match it are 5.1, 5.2, and later.
847
848Unicode furnishes the C<Age> property from which this is derived. The problem
849with Age is that a strict interpretation of it (which Perl takes) has it
850matching the precise release a code point's meaning is introduced in. Thus
851C<U+0041> would match only 1.1; and C<U+1EFF> only 5.1. This is not usually what
852you want.
853
854Some non-Perl implementations of the Age property may change its meaning to be
a9130ea9 855the same as the Perl C<Present_In> property; just be aware of that.
9f815e24
KW
856
857Another confusion with both these properties is that the definition is not
b19eb496
TC
858that the code point has been I<assigned>, but that the meaning of the code point
859has been I<determined>. This is because 66 code points will always be
a9130ea9 860unassigned, and so the C<Age> for them is the Unicode version in which the decision
b19eb496 861to make them so was made. For example, C<U+FDD0> is to be permanently
9f815e24 862unassigned to a character, and the decision to do that was made in version 3.1,
b19eb496 863so C<\p{Age=3.1}> matches this character, as also does C<\p{Present_In: 3.1}> and up.
9f815e24
KW
864
865=item B<C<\p{Print}>>
866
ae5b72c8 867This matches any character that is graphical or blank, except controls.
9f815e24
KW
868
869=item B<C<\p{SpacePerl}>>
870
871This is the same as C<\s>, including beyond ASCII.
872
4d4acfba 873Mnemonic: Space, as modified by Perl. (It doesn't include the vertical tab
779cf272 874until v5.18, which both the Posix standard and Unicode consider white space.)
9f815e24 875
4364919a
KW
876=item B<C<\p{Title}>> and B<C<\p{Titlecase}>>
877
878Under case-sensitive matching, these both match the same code points as
879C<\p{General Category=Titlecase_Letter}> (C<\p{gc=lt}>). The difference
880is that under C</i> caseless matching, these match the same as
881C<\p{Cased}>, whereas C<\p{gc=lt}> matches C<\p{Cased_Letter>).
882
2d88a86a
KW
883=item B<C<\p{Unicode}>>
884
885This matches any of the 1_114_112 Unicode code points.
886C<\p{Any}>.
887
9f815e24
KW
888=item B<C<\p{VertSpace}>>
889
890This is the same as C<\v>: A character that changes the spacing vertically.
891
892=item B<C<\p{Word}>>
893
b19eb496 894This is the same as C<\w>, including over 100_000 characters beyond ASCII.
9f815e24 895
42581d5d
KW
896=item B<C<\p{XPosix...}>>
897
b19eb496 898There are several of these, which are the standard Posix classes
42581d5d
KW
899extended to the full Unicode range. They are described in
900L<perlrecharclass/POSIX Character Classes>.
901
9f815e24
KW
902=back
903
a9130ea9 904
376d9008 905=head2 User-Defined Character Properties
491fd90a 906
51f494cc 907You can define your own binary character properties by defining subroutines
a9130ea9 908whose names begin with C<"In"> or C<"Is">. (The experimental feature
9d1a5160
KW
909L<perlre/(?[ ])> provides an alternative which allows more complex
910definitions.) The subroutines can be defined in any
51f494cc 911package. The user-defined properties can be used in the regular expression
a9130ea9 912C<\p{}> and C<\P{}> constructs; if you are using a user-defined property from a
51f494cc 913package other than the one you are in, you must specify its package in the
a9130ea9 914C<\p{}> or C<\P{}> construct.
bac0b425 915
51f494cc 916 # assuming property Is_Foreign defined in Lang::
bac0b425
JP
917 package main; # property package name required
918 if ($txt =~ /\p{Lang::IsForeign}+/) { ... }
919
920 package Lang; # property package name not required
921 if ($txt =~ /\p{IsForeign}+/) { ... }
922
923
924Note that the effect is compile-time and immutable once defined.
b19eb496
TC
925However, the subroutines are passed a single parameter, which is 0 if
926case-sensitive matching is in effect and non-zero if caseless matching
56ca34ca
KW
927is in effect. The subroutine may return different values depending on
928the value of the flag, and one set of values will immutably be in effect
b19eb496 929for all case-sensitive matches, and the other set for all case-insensitive
56ca34ca 930matches.
491fd90a 931
b19eb496 932Note that if the regular expression is tainted, then Perl will die rather
a9130ea9 933than calling the subroutine when the name of the subroutine is
0e9be77f
DM
934determined by the tainted data.
935
376d9008
JB
936The subroutines must return a specially-formatted string, with one
937or more newline-separated lines. Each line must be one of the following:
491fd90a
JH
938
939=over 4
940
941=item *
942
df9e1087 943A single hexadecimal number denoting a code point to include.
510254c9
A
944
945=item *
946
99a6b1f0 947Two hexadecimal numbers separated by horizontal whitespace (space or
df9e1087 948tabular characters) denoting a range of code points to include.
491fd90a
JH
949
950=item *
951
a9130ea9
KW
952Something to include, prefixed by C<"+">: a built-in character
953property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 954name) user-defined character property,
bac0b425
JP
955to represent all the characters in that property; two hexadecimal code
956points for a range; or a single hexadecimal code point.
491fd90a
JH
957
958=item *
959
a9130ea9
KW
960Something to exclude, prefixed by C<"-">: an existing character
961property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 962name) user-defined character property,
bac0b425
JP
963to represent all the characters in that property; two hexadecimal code
964points for a range; or a single hexadecimal code point.
491fd90a
JH
965
966=item *
967
a9130ea9
KW
968Something to negate, prefixed C<"!">: an existing character
969property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 970name) user-defined character property,
bac0b425
JP
971to represent all the characters in that property; two hexadecimal code
972points for a range; or a single hexadecimal code point.
973
974=item *
975
a9130ea9
KW
976Something to intersect with, prefixed by C<"&">: an existing character
977property (prefixed by C<"utf8::">) or a fully qualified (including package
830137a2 978name) user-defined character property,
bac0b425
JP
979for all the characters except the characters in the property; two
980hexadecimal code points for a range; or a single hexadecimal code point.
491fd90a
JH
981
982=back
983
984For example, to define a property that covers both the Japanese
985syllabaries (hiragana and katakana), you can define
986
987 sub InKana {
d88362ca 988 return <<END;
d5822f25
A
989 3040\t309F
990 30A0\t30FF
491fd90a
JH
991 END
992 }
993
d5822f25
A
994Imagine that the here-doc end marker is at the beginning of the line.
995Now you can use C<\p{InKana}> and C<\P{InKana}>.
491fd90a
JH
996
997You could also have used the existing block property names:
998
999 sub InKana {
d88362ca 1000 return <<'END';
491fd90a
JH
1001 +utf8::InHiragana
1002 +utf8::InKatakana
1003 END
1004 }
1005
1006Suppose you wanted to match only the allocated characters,
d5822f25 1007not the raw block ranges: in other words, you want to remove
b65e6125 1008the unassigned characters:
491fd90a
JH
1009
1010 sub InKana {
d88362ca 1011 return <<'END';
491fd90a
JH
1012 +utf8::InHiragana
1013 +utf8::InKatakana
1014 -utf8::IsCn
1015 END
1016 }
1017
1018The negation is useful for defining (surprise!) negated classes.
1019
1020 sub InNotKana {
d88362ca 1021 return <<'END';
491fd90a
JH
1022 !utf8::InHiragana
1023 -utf8::InKatakana
1024 +utf8::IsCn
1025 END
1026 }
1027
461020ad
KW
1028This will match all non-Unicode code points, since every one of them is
1029not in Kana. You can use intersection to exclude these, if desired, as
1030this modified example shows:
bac0b425 1031
461020ad 1032 sub InNotKana {
bac0b425 1033 return <<'END';
461020ad
KW
1034 !utf8::InHiragana
1035 -utf8::InKatakana
1036 +utf8::IsCn
1037 &utf8::Any
bac0b425
JP
1038 END
1039 }
1040
461020ad
KW
1041C<&utf8::Any> must be the last line in the definition.
1042
1043Intersection is used generally for getting the common characters matched
a9130ea9 1044by two (or more) classes. It's important to remember not to use C<"&"> for
461020ad
KW
1045the first set; that would be intersecting with nothing, resulting in an
1046empty set.
1047
2d88a86a
KW
1048Unlike non-user-defined C<\p{}> property matches, no warning is ever
1049generated if these properties are matched against a non-Unicode code
1050point (see L</Beyond Unicode code points> below).
bac0b425 1051
68585b5e 1052=head2 User-Defined Case Mappings (for serious hackers only)
822502e5 1053
5d1892be 1054B<This feature has been removed as of Perl 5.16.>
a9130ea9 1055The CPAN module C<L<Unicode::Casing>> provides better functionality without
5d1892be
KW
1056the drawbacks that this feature had. If you are using a Perl earlier
1057than 5.16, this feature was most fully documented in the 5.14 version of
1058this pod:
1059L<http://perldoc.perl.org/5.14.0/perlunicode.html#User-Defined-Case-Mappings-%28for-serious-hackers-only%29>
3a2263fe 1060
376d9008 1061=head2 Character Encodings for Input and Output
8cbd9a7a 1062
7221edc9 1063See L<Encode>.
8cbd9a7a 1064
c29a771d 1065=head2 Unicode Regular Expression Support Level
776f8809 1066
b19eb496
TC
1067The following list of Unicode supported features for regular expressions describes
1068all features currently directly supported by core Perl. The references to "Level N"
8158862b 1069and the section numbers refer to the Unicode Technical Standard #18,
b19eb496 1070"Unicode Regular Expressions", version 13, from August 2008.
776f8809
JH
1071
1072=over 4
1073
1074=item *
1075
1076Level 1 - Basic Unicode Support
1077
755789c0
KW
1078 RL1.1 Hex Notation - done [1]
1079 RL1.2 Properties - done [2][3]
1080 RL1.2a Compatibility Properties - done [4]
9d1a5160 1081 RL1.3 Subtraction and Intersection - experimental [5]
755789c0
KW
1082 RL1.4 Simple Word Boundaries - done [6]
1083 RL1.5 Simple Loose Matches - done [7]
1084 RL1.6 Line Boundaries - MISSING [8][9]
1085 RL1.7 Supplementary Code Points - done [10]
1086
6f33e417
KW
1087=over 4
1088
a6a7eedc 1089=item [1] C<\N{U+...}> and C<\x{...}>
6f33e417 1090
a6a7eedc 1091=item [2] C<\p{...}> C<\P{...}>
6f33e417 1092
a6a7eedc
KW
1093=item [3] supports not only minimal list, but all Unicode character
1094properties (see Unicode Character Properties above)
6f33e417 1095
a6a7eedc
KW
1096=item [4] C<\d> C<\D> C<\s> C<\S> C<\w> C<\W> C<\X> C<[:I<prop>:]>
1097C<[:^I<prop>:]>
6f33e417 1098
a6a7eedc
KW
1099=item [5] The experimental feature starting in v5.18 C<"(?[...])"> accomplishes
1100this.
6f33e417 1101
a6a7eedc
KW
1102See L<perlre/(?[ ])>. If you don't want to use an experimental
1103feature, you can use one of the following:
6f33e417
KW
1104
1105=over 4
1106
a6a7eedc
KW
1107=item *
1108
1109Regular expression look-ahead
6f33e417
KW
1110
1111You can mimic class subtraction using lookahead.
8158862b 1112For example, what UTS#18 might write as
29bdacb8 1113
209c9685 1114 [{Block=Greek}-[{UNASSIGNED}]]
dbe420b4
JH
1115
1116in Perl can be written as:
1117
209c9685
KW
1118 (?!\p{Unassigned})\p{Block=Greek}
1119 (?=\p{Assigned})\p{Block=Greek}
dbe420b4
JH
1120
1121But in this particular example, you probably really want
1122
209c9685 1123 \p{Greek}
dbe420b4
JH
1124
1125which will match assigned characters known to be part of the Greek script.
29bdacb8 1126
a6a7eedc
KW
1127=item *
1128
1129CPAN module C<L<Unicode::Regex::Set>>
8158862b 1130
6f33e417
KW
1131It does implement the full UTS#18 grouping, intersection, union, and
1132removal (subtraction) syntax.
8158862b 1133
a6a7eedc
KW
1134=item *
1135
1136L</"User-Defined Character Properties">
6f33e417 1137
a9130ea9 1138C<"+"> for union, C<"-"> for removal (set-difference), C<"&"> for intersection
6f33e417
KW
1139
1140=back
1141
a6a7eedc 1142=item [6] C<\b> C<\B>
6f33e417
KW
1143
1144=item [7]
a6a7eedc 1145Note that Perl does Full case-folding in matching, not Simple:
6f33e417 1146
a6a7eedc
KW
1147For example C<U+1F88> is equivalent to C<U+1F00 U+03B9>, instead of just
1148C<U+1F80>. This difference matters mainly for certain Greek capital
a9130ea9
KW
1149letters with certain modifiers: the Full case-folding decomposes the
1150letter, while the Simple case-folding would map it to a single
1151character.
6f33e417
KW
1152
1153=item [8]
a6a7eedc
KW
1154Perl treats C<\n> as the start- and end-line delimiter. Unicode
1155specifies more characters that should be so-interpreted.
6f33e417 1156
a6a7eedc 1157These are:
6f33e417 1158
a6a7eedc
KW
1159 VT U+000B (\v in C)
1160 FF U+000C (\f)
1161 CR U+000D (\r)
1162 NEL U+0085
1163 LS U+2028
1164 PS U+2029
6f33e417 1165
a6a7eedc
KW
1166C<^> and C<$> in regular expression patterns are supposed to match all
1167these, but don't.
1168These characters also don't, but should, affect C<< <> >> C<$.>, and
1169script line numbers.
6f33e417 1170
a6a7eedc
KW
1171Also, lines should not be split within C<CRLF> (i.e. there is no
1172empty line between C<\r> and C<\n>). For C<CRLF>, try the C<:crlf>
1173layer (see L<PerlIO>).
1174
1175=item [9] But C<L<Unicode::LineBreak>> is available.
6f33e417 1176
a6a7eedc
KW
1177This module supplies line breaking conformant with
1178L<UAX#14 "Unicode Line Breaking Algorithm"|http://www.unicode.org/reports/tr14>.
1179
1180=item [10]
a9130ea9
KW
1181UTF-8/UTF-EBDDIC used in Perl allows not only C<U+10000> to
1182C<U+10FFFF> but also beyond C<U+10FFFF>
6f33e417
KW
1183
1184=back
5ca1ac52 1185
776f8809
JH
1186=item *
1187
1188Level 2 - Extended Unicode Support
1189
755789c0
KW
1190 RL2.1 Canonical Equivalents - MISSING [10][11]
1191 RL2.2 Default Grapheme Clusters - MISSING [12]
ae3bb8ea 1192 RL2.3 Default Word Boundaries - DONE [14]
755789c0
KW
1193 RL2.4 Default Loose Matches - MISSING [15]
1194 RL2.5 Name Properties - DONE
1195 RL2.6 Wildcard Properties - MISSING
8158862b 1196
755789c0
KW
1197 [10] see UAX#15 "Unicode Normalization Forms"
1198 [11] have Unicode::Normalize but not integrated to regexes
64935bc6
KW
1199 [12] have \X and \b{gcb} but we don't have a "Grapheme Cluster
1200 Mode"
755789c0 1201 [14] see UAX#29, Word Boundaries
902b08d0 1202 [15] This is covered in Chapter 3.13 (in Unicode 6.0)
776f8809
JH
1203
1204=item *
1205
8158862b
ST
1206Level 3 - Tailored Support
1207
755789c0
KW
1208 RL3.1 Tailored Punctuation - MISSING
1209 RL3.2 Tailored Grapheme Clusters - MISSING [17][18]
1210 RL3.3 Tailored Word Boundaries - MISSING
1211 RL3.4 Tailored Loose Matches - MISSING
1212 RL3.5 Tailored Ranges - MISSING
1213 RL3.6 Context Matching - MISSING [19]
1214 RL3.7 Incremental Matches - MISSING
8158862b 1215 ( RL3.8 Unicode Set Sharing )
755789c0
KW
1216 RL3.9 Possible Match Sets - MISSING
1217 RL3.10 Folded Matching - MISSING [20]
1218 RL3.11 Submatchers - MISSING
1219
1220 [17] see UAX#10 "Unicode Collation Algorithms"
1221 [18] have Unicode::Collate but not integrated to regexes
1222 [19] have (?<=x) and (?=x), but look-aheads or look-behinds
1223 should see outside of the target substring
1224 [20] need insensitive matching for linguistic features other
1225 than case; for example, hiragana to katakana, wide and
1226 narrow, simplified Han to traditional Han (see UTR#30
1227 "Character Foldings")
776f8809
JH
1228
1229=back
1230
c349b1b9
JH
1231=head2 Unicode Encodings
1232
376d9008
JB
1233Unicode characters are assigned to I<code points>, which are abstract
1234numbers. To use these numbers, various encodings are needed.
c349b1b9
JH
1235
1236=over 4
1237
c29a771d 1238=item *
5cb3728c
RB
1239
1240UTF-8
c349b1b9 1241
6d4f9cf2 1242UTF-8 is a variable-length (1 to 4 bytes), byte-order independent
a6a7eedc
KW
1243encoding. In most of Perl's documentation, including elsewhere in this
1244document, the term "UTF-8" means also "UTF-EBCDIC". But in this section,
1245"UTF-8" refers only to the encoding used on ASCII platforms. It is a
1246superset of 7-bit US-ASCII, so anything encoded in ASCII has the
1247identical representation when encoded in UTF-8.
c349b1b9 1248
8c007b5a 1249The following table is from Unicode 3.2.
05632f9a 1250
755789c0 1251 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
05632f9a 1252
d88362ca 1253 U+0000..U+007F 00..7F
e1b711da 1254 U+0080..U+07FF * C2..DF 80..BF
d88362ca 1255 U+0800..U+0FFF E0 * A0..BF 80..BF
ec90690f
ST
1256 U+1000..U+CFFF E1..EC 80..BF 80..BF
1257 U+D000..U+D7FF ED 80..9F 80..BF
755789c0 1258 U+D800..U+DFFF +++++ utf16 surrogates, not legal utf8 +++++
ec90690f 1259 U+E000..U+FFFF EE..EF 80..BF 80..BF
d88362ca
KW
1260 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
1261 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
1262 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
e1b711da 1263
b19eb496 1264Note the gaps marked by "*" before several of the byte entries above. These are
e1b711da
KW
1265caused by legal UTF-8 avoiding non-shortest encodings: it is technically
1266possible to UTF-8-encode a single code point in different ways, but that is
1267explicitly forbidden, and the shortest possible encoding should always be used
1268(and that is what Perl does).
37361303 1269
376d9008 1270Another way to look at it is via bits:
05632f9a 1271
755789c0 1272 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
05632f9a 1273
755789c0
KW
1274 0aaaaaaa 0aaaaaaa
1275 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
1276 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
1277 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
05632f9a 1278
a9130ea9 1279As you can see, the continuation bytes all begin with C<"10">, and the
e1b711da 1280leading bits of the start byte tell how many bytes there are in the
05632f9a
JH
1281encoded character.
1282
6d4f9cf2 1283The original UTF-8 specification allowed up to 6 bytes, to allow
a9130ea9 1284encoding of numbers up to C<0x7FFF_FFFF>. Perl continues to allow those,
6d4f9cf2
KW
1285and has extended that up to 13 bytes to encode code points up to what
1286can fit in a 64-bit word. However, Perl will warn if you output any of
b19eb496 1287these as being non-portable; and under strict UTF-8 input protocols,
6d4f9cf2
KW
1288they are forbidden.
1289
c29a771d 1290=item *
5cb3728c
RB
1291
1292UTF-EBCDIC
dbe420b4 1293
b65e6125 1294Like UTF-8, but EBCDIC-safe, in the way that UTF-8 is ASCII-safe.
a6a7eedc
KW
1295This means that all the basic characters (which includes all
1296those that have ASCII equivalents (like C<"A">, C<"0">, C<"%">, I<etc.>)
1297are the same in both EBCDIC and UTF-EBCDIC.)
1298
1299UTF-EBCDIC is used on EBCDIC platforms. The largest Unicode code points
1300take 5 bytes to represent (instead of 4 in UTF-8), and Perl extends it
1301to a maximum of 7 bytes to encode pode points up to what can fit in a
130232-bit word (instead of 13 bytes and a 64-bit word in UTF-8).
dbe420b4 1303
c29a771d 1304=item *
5cb3728c 1305
b65e6125 1306UTF-16, UTF-16BE, UTF-16LE, Surrogates, and C<BOM>'s (Byte Order Marks)
c349b1b9 1307
1bfb14c4
JH
1308The followings items are mostly for reference and general Unicode
1309knowledge, Perl doesn't use these constructs internally.
dbe420b4 1310
b19eb496
TC
1311Like UTF-8, UTF-16 is a variable-width encoding, but where
1312UTF-8 uses 8-bit code units, UTF-16 uses 16-bit code units.
1313All code points occupy either 2 or 4 bytes in UTF-16: code points
1314C<U+0000..U+FFFF> are stored in a single 16-bit unit, and code
1bfb14c4 1315points C<U+10000..U+10FFFF> in two 16-bit units. The latter case is
c349b1b9
JH
1316using I<surrogates>, the first 16-bit unit being the I<high
1317surrogate>, and the second being the I<low surrogate>.
1318
376d9008 1319Surrogates are code points set aside to encode the C<U+10000..U+10FFFF>
c349b1b9 1320range of Unicode code points in pairs of 16-bit units. The I<high
9f815e24 1321surrogates> are the range C<U+D800..U+DBFF> and the I<low surrogates>
376d9008 1322are the range C<U+DC00..U+DFFF>. The surrogate encoding is
c349b1b9 1323
d88362ca
KW
1324 $hi = ($uni - 0x10000) / 0x400 + 0xD800;
1325 $lo = ($uni - 0x10000) % 0x400 + 0xDC00;
c349b1b9
JH
1326
1327and the decoding is
1328
d88362ca 1329 $uni = 0x10000 + ($hi - 0xD800) * 0x400 + ($lo - 0xDC00);
c349b1b9 1330
376d9008 1331Because of the 16-bitness, UTF-16 is byte-order dependent. UTF-16
c349b1b9 1332itself can be used for in-memory computations, but if storage or
376d9008
JB
1333transfer is required either UTF-16BE (big-endian) or UTF-16LE
1334(little-endian) encodings must be chosen.
c349b1b9
JH
1335
1336This introduces another problem: what if you just know that your data
376d9008 1337is UTF-16, but you don't know which endianness? Byte Order Marks, or
b65e6125 1338C<BOM>'s, are a solution to this. A special character has been reserved
86bbd6d1 1339in Unicode to function as a byte order marker: the character with the
a9130ea9 1340code point C<U+FEFF> is the C<BOM>.
042da322 1341
a9130ea9 1342The trick is that if you read a C<BOM>, you will know the byte order,
376d9008
JB
1343since if it was written on a big-endian platform, you will read the
1344bytes C<0xFE 0xFF>, but if it was written on a little-endian platform,
1345you will read the bytes C<0xFF 0xFE>. (And if the originating platform
b65e6125
KW
1346was writing in ASCII platform UTF-8, you will read the bytes
1347C<0xEF 0xBB 0xBF>.)
042da322 1348
86bbd6d1 1349The way this trick works is that the character with the code point
6d4f9cf2 1350C<U+FFFE> is not supposed to be in input streams, so the
a9130ea9 1351sequence of bytes C<0xFF 0xFE> is unambiguously "C<BOM>, represented in
1bfb14c4 1352little-endian format" and cannot be C<U+FFFE>, represented in big-endian
6d4f9cf2
KW
1353format".
1354
1355Surrogates have no meaning in Unicode outside their use in pairs to
1356represent other code points. However, Perl allows them to be
1357represented individually internally, for example by saying
f651977e
TC
1358C<chr(0xD801)>, so that all code points, not just those valid for open
1359interchange, are
6d4f9cf2 1360representable. Unicode does define semantics for them, such as their
a9130ea9
KW
1361C<L</General_Category>> is C<"Cs">. But because their use is somewhat dangerous,
1362Perl will warn (using the warning category C<"surrogate">, which is a
1363sub-category of C<"utf8">) if an attempt is made
6d4f9cf2
KW
1364to do things like take the lower case of one, or match
1365case-insensitively, or to output them. (But don't try this on Perls
1366before 5.14.)
c349b1b9 1367
c29a771d 1368=item *
5cb3728c 1369
1e54db1a 1370UTF-32, UTF-32BE, UTF-32LE
c349b1b9 1371
b65e6125 1372The UTF-32 family is pretty much like the UTF-16 family, except that
042da322 1373the units are 32-bit, and therefore the surrogate scheme is not
a9130ea9 1374needed. UTF-32 is a fixed-width encoding. The C<BOM> signatures are
b19eb496 1375C<0x00 0x00 0xFE 0xFF> for BE and C<0xFF 0xFE 0x00 0x00> for LE.
c349b1b9 1376
c29a771d 1377=item *
5cb3728c
RB
1378
1379UCS-2, UCS-4
c349b1b9 1380
b19eb496 1381Legacy, fixed-width encodings defined by the ISO 10646 standard. UCS-2 is a 16-bit
376d9008 1382encoding. Unlike UTF-16, UCS-2 is not extensible beyond C<U+FFFF>,
339cfa0e 1383because it does not use surrogates. UCS-4 is a 32-bit encoding,
b19eb496 1384functionally identical to UTF-32 (the difference being that
a9130ea9 1385UCS-4 forbids neither surrogates nor code points larger than C<0x10_FFFF>).
c349b1b9 1386
c29a771d 1387=item *
5cb3728c
RB
1388
1389UTF-7
c349b1b9 1390
376d9008
JB
1391A seven-bit safe (non-eight-bit) encoding, which is useful if the
1392transport or storage is not eight-bit safe. Defined by RFC 2152.
c349b1b9 1393
95a1a48b
JH
1394=back
1395
57e88091 1396=head2 Noncharacter code points
6d4f9cf2 1397
57e88091 139866 code points are set aside in Unicode as "noncharacter code points".
a9130ea9 1399These all have the C<Unassigned> (C<Cn>) C<L</General_Category>>, and
57e88091
KW
1400no character will ever be assigned to any of them. They are the 32 code
1401points between C<U+FDD0> and C<U+FDEF> inclusive, and the 34 code
1402points:
1403
1404 U+FFFE U+FFFF
1405 U+1FFFE U+1FFFF
1406 U+2FFFE U+2FFFF
1407 ...
1408 U+EFFFE U+EFFFF
1409 U+FFFFE U+FFFFF
1410 U+10FFFE U+10FFFF
1411
1412Until Unicode 7.0, the noncharacters were "B<forbidden> for use in open
1413interchange of Unicode text data", so that code that processed those
1414streams could use these code points as sentinels that could be mixed in
1415with character data, and would always be distinguishable from that data.
1416(Emphasis above and in the next paragraph are added in this document.)
1417
1418Unicode 7.0 changed the wording so that they are "B<not recommended> for
1419use in open interchange of Unicode text data". The 7.0 Standard goes on
1420to say:
1421
1422=over 4
1423
1424"If a noncharacter is received in open interchange, an application is
1425not required to interpret it in any way. It is good practice, however,
1426to recognize it as a noncharacter and to take appropriate action, such
1427as replacing it with C<U+FFFD> replacement character, to indicate the
1428problem in the text. It is not recommended to simply delete
1429noncharacter code points from such text, because of the potential
1430security issues caused by deleting uninterpreted characters. (See
1431conformance clause C7 in Section 3.2, Conformance Requirements, and
1432L<Unicode Technical Report #36, "Unicode Security
1433Considerations"|http://www.unicode.org/reports/tr36/#Substituting_for_Ill_Formed_Subsequences>)."
1434
1435=back
1436
1437This change was made because it was found that various commercial tools
1438like editors, or for things like source code control, had been written
1439so that they would not handle program files that used these code points,
1440effectively precluding their use almost entirely! And that was never
1441the intent. They've always been meant to be usable within an
1442application, or cooperating set of applications, at will.
1443
1444If you're writing code, such as an editor, that is supposed to be able
1445to handle any Unicode text data, then you shouldn't be using these code
1446points yourself, and instead allow them in the input. If you need
1447sentinels, they should instead be something that isn't legal Unicode.
1448For UTF-8 data, you can use the bytes 0xC1 and 0xC2 as sentinels, as
1449they never appear in well-formed UTF-8. (There are equivalents for
1450UTF-EBCDIC). You can also store your Unicode code points in integer
1451variables and use negative values as sentinels.
1452
1453If you're not writing such a tool, then whether you accept noncharacters
1454as input is up to you (though the Standard recommends that you not). If
1455you do strict input stream checking with Perl, these code points
1456continue to be forbidden. This is to maintain backward compatibility
1457(otherwise potential security holes could open up, as an unsuspecting
1458application that was written assuming the noncharacters would be
1459filtered out before getting to it, could now, without warning, start
1460getting them). To do strict checking, you can use the layer
1461C<:encoding('UTF-8')>.
1462
1463Perl continues to warn (using the warning category C<"nonchar">, which
1464is a sub-category of C<"utf8">) if an attempt is made to output
1465noncharacters.
42581d5d
KW
1466
1467=head2 Beyond Unicode code points
1468
a9130ea9
KW
1469The maximum Unicode code point is C<U+10FFFF>, and Unicode only defines
1470operations on code points up through that. But Perl works on code
42581d5d
KW
1471points up to the maximum permissible unsigned number available on the
1472platform. However, Perl will not accept these from input streams unless
1473lax rules are being used, and will warn (using the warning category
2d88a86a
KW
1474C<"non_unicode">, which is a sub-category of C<"utf8">) if any are output.
1475
1476Since Unicode rules are not defined on these code points, if a
1477Unicode-defined operation is done on them, Perl uses what we believe are
1478sensible rules, while generally warning, using the C<"non_unicode">
1479category. For example, C<uc("\x{11_0000}")> will generate such a
1480warning, returning the input parameter as its result, since Perl defines
1481the uppercase of every non-Unicode code point to be the code point
b65e6125
KW
1482itself. (All the case changing operations, not just uppercasing, work
1483this way.)
2d88a86a
KW
1484
1485The situation with matching Unicode properties in regular expressions,
1486the C<\p{}> and C<\P{}> constructs, against these code points is not as
1487clear cut, and how these are handled has changed as we've gained
1488experience.
1489
1490One possibility is to treat any match against these code points as
1491undefined. But since Perl doesn't have the concept of a match being
1492undefined, it converts this to failing or C<FALSE>. This is almost, but
1493not quite, what Perl did from v5.14 (when use of these code points
1494became generally reliable) through v5.18. The difference is that Perl
1495treated all C<\p{}> matches as failing, but all C<\P{}> matches as
1496succeeding.
1497
1498One problem with this is that it leads to unexpected, and confusting
1499results in some cases:
1500
1501 chr(0x110000) =~ \p{ASCII_Hex_Digit=True} # Failed on <= v5.18
1502 chr(0x110000) =~ \p{ASCII_Hex_Digit=False} # Failed! on <= v5.18
1503
1504That is, it treated both matches as undefined, and converted that to
1505false (raising a warning on each). The first case is the expected
1506result, but the second is likely counterintuitive: "How could both be
1507false when they are complements?" Another problem was that the
1508implementation optimized many Unicode property matches down to already
1509existing simpler, faster operations, which don't raise the warning. We
1510chose to not forgo those optimizations, which help the vast majority of
1511matches, just to generate a warning for the unlikely event that an
1512above-Unicode code point is being matched against.
1513
1514As a result of these problems, starting in v5.20, what Perl does is
1515to treat non-Unicode code points as just typical unassigned Unicode
1516characters, and matches accordingly. (Note: Unicode has atypical
57e88091 1517unassigned code points. For example, it has noncharacter code points,
2d88a86a
KW
1518and ones that, when they do get assigned, are destined to be written
1519Right-to-left, as Arabic and Hebrew are. Perl assumes that no
1520non-Unicode code point has any atypical properties.)
1521
1522Perl, in most cases, will raise a warning when matching an above-Unicode
1523code point against a Unicode property when the result is C<TRUE> for
1524C<\p{}>, and C<FALSE> for C<\P{}>. For example:
1525
1526 chr(0x110000) =~ \p{ASCII_Hex_Digit=True} # Fails, no warning
1527 chr(0x110000) =~ \p{ASCII_Hex_Digit=False} # Succeeds, with warning
1528
1529In both these examples, the character being matched is non-Unicode, so
1530Unicode doesn't define how it should match. It clearly isn't an ASCII
1531hex digit, so the first example clearly should fail, and so it does,
1532with no warning. But it is arguable that the second example should have
1533an undefined, hence C<FALSE>, result. So a warning is raised for it.
1534
1535Thus the warning is raised for many fewer cases than in earlier Perls,
1536and only when what the result is could be arguable. It turns out that
1537none of the optimizations made by Perl (or are ever likely to be made)
1538cause the warning to be skipped, so it solves both problems of Perl's
1539earlier approach. The most commonly used property that is affected by
1540this change is C<\p{Unassigned}> which is a short form for
1541C<\p{General_Category=Unassigned}>. Starting in v5.20, all non-Unicode
1542code points are considered C<Unassigned>. In earlier releases the
1543matches failed because the result was considered undefined.
1544
1545The only place where the warning is not raised when it might ought to
1546have been is if optimizations cause the whole pattern match to not even
1547be attempted. For example, Perl may figure out that for a string to
1548match a certain regular expression pattern, the string has to contain
1549the substring C<"foobar">. Before attempting the match, Perl may look
1550for that substring, and if not found, immediately fail the match without
1551actually trying it; so no warning gets generated even if the string
1552contains an above-Unicode code point.
1553
1554This behavior is more "Do what I mean" than in earlier Perls for most
1555applications. But it catches fewer issues for code that needs to be
1556strictly Unicode compliant. Therefore there is an additional mode of
1557operation available to accommodate such code. This mode is enabled if a
1558regular expression pattern is compiled within the lexical scope where
1559the C<"non_unicode"> warning class has been made fatal, say by:
1560
1561 use warnings FATAL => "non_unicode"
1562
44ecbbd8 1563(see L<warnings>). In this mode of operation, Perl will raise the
2d88a86a
KW
1564warning for all matches against a non-Unicode code point (not just the
1565arguable ones), and it skips the optimizations that might cause the
1566warning to not be output. (It currently still won't warn if the match
1567isn't even attempted, like in the C<"foobar"> example above.)
1568
1569In summary, Perl now normally treats non-Unicode code points as typical
1570Unicode unassigned code points for regular expression matches, raising a
1571warning only when it is arguable what the result should be. However, if
1572this warning has been made fatal, it isn't skipped.
1573
1574There is one exception to all this. C<\p{All}> looks like a Unicode
1575property, but it is a Perl extension that is defined to be true for all
1576possible code points, Unicode or not, so no warning is ever generated
1577when matching this against a non-Unicode code point. (Prior to v5.20,
1578it was an exact synonym for C<\p{Any}>, matching code points C<0>
1579through C<0x10FFFF>.)
6d4f9cf2 1580
0d7c09bb
JH
1581=head2 Security Implications of Unicode
1582
b65e6125
KW
1583First, read
1584L<Unicode Security Considerations|http://www.unicode.org/reports/tr36>.
1585
e1b711da
KW
1586Also, note the following:
1587
0d7c09bb
JH
1588=over 4
1589
1590=item *
1591
1592Malformed UTF-8
bf0fa0b2 1593
42581d5d 1594Unfortunately, the original specification of UTF-8 leaves some room for
bf0fa0b2 1595interpretation of how many bytes of encoded output one should generate
376d9008
JB
1596from one input Unicode character. Strictly speaking, the shortest
1597possible sequence of UTF-8 bytes should be generated,
1598because otherwise there is potential for an input buffer overflow at
feda178f 1599the receiving end of a UTF-8 connection. Perl always generates the
e1b711da 1600shortest length UTF-8, and with warnings on, Perl will warn about
376d9008 1601non-shortest length UTF-8 along with other malformations, such as the
b19eb496 1602surrogates, which are not Unicode code points valid for interchange.
bf0fa0b2 1603
0d7c09bb
JH
1604=item *
1605
68693f9e 1606Regular expression pattern matching may surprise you if you're not
b19eb496
TC
1607accustomed to Unicode. Starting in Perl 5.14, several pattern
1608modifiers are available to control this, called the character set
42581d5d
KW
1609modifiers. Details are given in L<perlre/Character set modifiers>.
1610
1611=back
0d7c09bb 1612
376d9008 1613As discussed elsewhere, Perl has one foot (two hooves?) planted in
a6a7eedc
KW
1614each of two worlds: the old world of ASCII and single-byte locales, and
1615the new world of Unicode, upgrading when necessary.
376d9008 1616If your legacy code does not explicitly use Unicode, no automatic
a6a7eedc 1617switch-over to Unicode should happen.
0d7c09bb 1618
c349b1b9
JH
1619=head2 Unicode in Perl on EBCDIC
1620
a6a7eedc
KW
1621Unicode is supported on EBCDIC platforms. See L<perlebcdic>.
1622
1623Unless ASCII vs. EBCDIC issues are specifically being discussed,
1624references to UTF-8 encoding in this document and elsewhere should be
1625read as meaning UTF-EBCDIC on EBCDIC platforms.
1626See L<perlebcdic/Unicode and UTF>.
1627
1628Because UTF-EBCDIC is so similar to UTF-8, the differences are mostly
1629hidden from you; S<C<use utf8>> (and NOT something like
1630S<C<use utfebcdic>>) declares the the script is in the platform's
1631"native" 8-bit encoding of Unicode. (Similarly for the C<":utf8">
1632layer.)
c349b1b9 1633
b310b053
JH
1634=head2 Locales
1635
42581d5d 1636See L<perllocale/Unicode and UTF-8>
b310b053 1637
1aad1664
JH
1638=head2 When Unicode Does Not Happen
1639
b65e6125
KW
1640There are still many places where Unicode (in some encoding or
1641another) could be given as arguments or received as results, or both in
1642Perl, but it is not, in spite of Perl having extensive ways to input and
1643output in Unicode, and a few other "entry points" like the C<@ARGV>
1644array (which can sometimes be interpreted as UTF-8).
1aad1664 1645
e1b711da
KW
1646The following are such interfaces. Also, see L</The "Unicode Bug">.
1647For all of these interfaces Perl
b9cedb1b 1648currently (as of v5.16.0) simply assumes byte strings both as arguments
b65e6125 1649and results, or UTF-8 strings if the (deprecated) C<encoding> pragma has been used.
1aad1664 1650
b19eb496
TC
1651One reason that Perl does not attempt to resolve the role of Unicode in
1652these situations is that the answers are highly dependent on the operating
1aad1664 1653system and the file system(s). For example, whether filenames can be
b19eb496
TC
1654in Unicode and in exactly what kind of encoding, is not exactly a
1655portable concept. Similarly for C<qx> and C<system>: how well will the
1656"command-line interface" (and which of them?) handle Unicode?
1aad1664
JH
1657
1658=over 4
1659
557a2462
RB
1660=item *
1661
a9130ea9
KW
1662C<chdir>, C<chmod>, C<chown>, C<chroot>, C<exec>, C<link>, C<lstat>, C<mkdir>,
1663C<rename>, C<rmdir>, C<stat>, C<symlink>, C<truncate>, C<unlink>, C<utime>, C<-X>
557a2462
RB
1664
1665=item *
1666
a9130ea9 1667C<%ENV>
557a2462
RB
1668
1669=item *
1670
a9130ea9 1671C<glob> (aka the C<E<lt>*E<gt>>)
557a2462
RB
1672
1673=item *
1aad1664 1674
a9130ea9 1675C<open>, C<opendir>, C<sysopen>
1aad1664 1676
557a2462 1677=item *
1aad1664 1678
a9130ea9 1679C<qx> (aka the backtick operator), C<system>
1aad1664 1680
557a2462 1681=item *
1aad1664 1682
a9130ea9 1683C<readdir>, C<readlink>
1aad1664
JH
1684
1685=back
1686
e1b711da
KW
1687=head2 The "Unicode Bug"
1688
a6a7eedc
KW
1689The term, "Unicode bug" has been applied to an inconsistency with the
1690code points in the C<Latin-1 Supplement> block, that is, between
1691128 and 255. Without a locale specified, unlike all other characters or
1692code points, these characters can have very different semantics
1693depending on the rules in effect. (Characters whose code points are
1694above 255 force Unicode rules; whereas the rules for ASCII characters
1695are the same under both ASCII and Unicode rules.)
1696
1697Under Unicode rules, these upper-Latin1 characters are interpreted as
1698Unicode code points, which means they have the same semantics as Latin-1
1699(ISO-8859-1) and C1 controls.
1700
1701As explained in L</ASCII Rules versus Unicode Rules>, under ASCII rules,
1702they are considered to be unassigned characters.
1703
1704This can lead to unexpected results. For example, a string's
1705semantics can suddenly change if a code point above 255 is appended to
1706it, which changes the rules from ASCII to Unicode. As an
1707example, consider the following program and its output:
1708
1709 $ perl -le'
1710 no feature 'unicode_strings';
1711 $s1 = "\xC2";
1712 $s2 = "\x{2660}";
1713 for ($s1, $s2, $s1.$s2) {
1714 print /\w/ || 0;
1715 }
1716 '
1717 0
1718 0
1719 1
1720
1721If there's no C<\w> in C<s1> nor in C<s2>, why does their concatenation
1722have one?
1723
1724This anomaly stems from Perl's attempt to not disturb older programs that
1725didn't use Unicode, along with Perl's desire to add Unicode support
1726seamlessly. But the result turned out to not be seamless. (By the way,
1727you can choose to be warned when things like this happen. See
1728C<L<encoding::warnings>>.)
1729
1730L<S<C<use feature 'unicode_strings'>>|feature/The 'unicode_strings' feature>
1731was added, starting in Perl v5.12, to address this problem. It affects
1732these things:
e1b711da
KW
1733
1734=over 4
1735
1736=item *
1737
1738Changing the case of a scalar, that is, using C<uc()>, C<ucfirst()>, C<lc()>,
2e2b2571
KW
1739and C<lcfirst()>, or C<\L>, C<\U>, C<\u> and C<\l> in double-quotish
1740contexts, such as regular expression substitutions.
a6a7eedc
KW
1741
1742Under C<unicode_strings> starting in Perl 5.12.0, Unicode rules are
2e2b2571
KW
1743generally used. See L<perlfunc/lc> for details on how this works
1744in combination with various other pragmas.
e1b711da
KW
1745
1746=item *
1747
2e2b2571 1748Using caseless (C</i>) regular expression matching.
a6a7eedc 1749
2e2b2571 1750Starting in Perl 5.14.0, regular expressions compiled within
a6a7eedc 1751the scope of C<unicode_strings> use Unicode rules
2e2b2571
KW
1752even when executed or compiled into larger
1753regular expressions outside the scope.
e1b711da
KW
1754
1755=item *
1756
a6a7eedc
KW
1757Matching any of several properties in regular expressions.
1758
1759These properties are C<\b> (without braces), C<\B> (without braces),
1760C<\s>, C<\S>, C<\w>, C<\W>, and all the Posix character classes
630d17dc 1761I<except> C<[[:ascii:]]>.
a6a7eedc 1762
2e2b2571 1763Starting in Perl 5.14.0, regular expressions compiled within
a6a7eedc 1764the scope of C<unicode_strings> use Unicode rules
2e2b2571
KW
1765even when executed or compiled into larger
1766regular expressions outside the scope.
e1b711da
KW
1767
1768=item *
1769
a6a7eedc
KW
1770In C<quotemeta> or its inline equivalent C<\Q>.
1771
2e2b2571
KW
1772Starting in Perl 5.16.0, consistent quoting rules are used within the
1773scope of C<unicode_strings>, as described in L<perlfunc/quotemeta>.
a6a7eedc
KW
1774Prior to that, or outside its scope, no code points above 127 are quoted
1775in UTF-8 encoded strings, but in byte encoded strings, code points
1776between 128-255 are always quoted.
eb88ed9e 1777
e1b711da
KW
1778=back
1779
a6a7eedc
KW
1780You can see from the above that the effect of C<unicode_strings>
1781increased over several Perl releases. (And Perl's support for Unicode
1782continues to improve; it's best to use the latest available release in
1783order to get the most complete and accurate results possible.) Note that
1784C<unicode_strings> is automatically chosen if you S<C<use 5.012>> or
1785higher.
e1b711da 1786
2e2b2571 1787For Perls earlier than those described above, or when a string is passed
a6a7eedc 1788to a function outside the scope of C<unicode_strings>, see the next section.
e1b711da 1789
1aad1664
JH
1790=head2 Forcing Unicode in Perl (Or Unforcing Unicode in Perl)
1791
e1b711da
KW
1792Sometimes (see L</"When Unicode Does Not Happen"> or L</The "Unicode Bug">)
1793there are situations where you simply need to force a byte
a6a7eedc
KW
1794string into UTF-8, or vice versa. The standard module L<Encode> can be
1795used for this, or the low-level calls
a9130ea9 1796L<C<utf8::upgrade($bytestring)>|utf8/Utility functions> and
a6a7eedc 1797L<C<utf8::downgrade($utf8string[, FAIL_OK])>|utf8/Utility functions>.
1aad1664 1798
a9130ea9 1799Note that C<utf8::downgrade()> can fail if the string contains characters
2bbc8d55 1800that don't fit into a byte.
1aad1664 1801
e1b711da
KW
1802Calling either function on a string that already is in the desired state is a
1803no-op.
1804
a6a7eedc
KW
1805L</ASCII Rules versus Unicode Rules> gives all the ways that a string is
1806made to use Unicode rules.
95a1a48b 1807
37b3b608 1808=head2 Using Unicode in XS
c349b1b9 1809
37b3b608
KW
1810See L<perlguts/"Unicode Support"> for an introduction to Unicode at
1811the XS level, and L<perlapi/Unicode Support> for the API details.
95a1a48b 1812
e1b711da
KW
1813=head2 Hacking Perl to work on earlier Unicode versions (for very serious hackers only)
1814
a6a7eedc
KW
1815Perl by default comes with the latest supported Unicode version built-in, but
1816the goal is to allow you to change to use any earlier one. In Perls
1817v5.20 and v5.22, however, the earliest usable version is Unicode 5.1.
1818Perl v5.18 is able to handle all earlier versions.
e1b711da 1819
42581d5d 1820Download the files in the desired version of Unicode from the Unicode web
e1b711da 1821site L<http://www.unicode.org>). These should replace the existing files in
b19eb496 1822F<lib/unicore> in the Perl source tree. Follow the instructions in
116693e8 1823F<README.perl> in that directory to change some of their names, and then build
26e391dd 1824perl (see L<INSTALL>).
116693e8 1825
c8d992ba
A
1826=head2 Porting code from perl-5.6.X
1827
a6a7eedc
KW
1828Perls starting in 5.8 have a different Unicode model from 5.6. In 5.6 the
1829programmer was required to use the C<utf8> pragma to declare that a
1830given scope expected to deal with Unicode data and had to make sure that
1831only Unicode data were reaching that scope. If you have code that is
c8d992ba 1832working with 5.6, you will need some of the following adjustments to
a6a7eedc
KW
1833your code. The examples are written such that the code will continue to
1834work under 5.6, so you should be safe to try them out.
c8d992ba 1835
755789c0 1836=over 3
c8d992ba
A
1837
1838=item *
1839
1840A filehandle that should read or write UTF-8
1841
b9cedb1b 1842 if ($] > 5.008) {
740d4bb2 1843 binmode $fh, ":encoding(utf8)";
c8d992ba
A
1844 }
1845
1846=item *
1847
1848A scalar that is going to be passed to some extension
1849
a9130ea9 1850Be it C<Compress::Zlib>, C<Apache::Request> or any extension that has no
c8d992ba 1851mention of Unicode in the manpage, you need to make sure that the
2575c402 1852UTF8 flag is stripped off. Note that at the time of this writing
b9cedb1b 1853(January 2012) the mentioned modules are not UTF-8-aware. Please
c8d992ba
A
1854check the documentation to verify if this is still true.
1855
b9cedb1b 1856 if ($] > 5.008) {
c8d992ba
A
1857 require Encode;
1858 $val = Encode::encode_utf8($val); # make octets
1859 }
1860
1861=item *
1862
1863A scalar we got back from an extension
1864
1865If you believe the scalar comes back as UTF-8, you will most likely
2575c402 1866want the UTF8 flag restored:
c8d992ba 1867
b9cedb1b 1868 if ($] > 5.008) {
c8d992ba
A
1869 require Encode;
1870 $val = Encode::decode_utf8($val);
1871 }
1872
1873=item *
1874
1875Same thing, if you are really sure it is UTF-8
1876
b9cedb1b 1877 if ($] > 5.008) {
c8d992ba
A
1878 require Encode;
1879 Encode::_utf8_on($val);
1880 }
1881
1882=item *
1883
a9130ea9 1884A wrapper for L<DBI> C<fetchrow_array> and C<fetchrow_hashref>
c8d992ba
A
1885
1886When the database contains only UTF-8, a wrapper function or method is
a9130ea9
KW
1887a convenient way to replace all your C<fetchrow_array> and
1888C<fetchrow_hashref> calls. A wrapper function will also make it easier to
c8d992ba 1889adapt to future enhancements in your database driver. Note that at the
b9cedb1b 1890time of this writing (January 2012), the DBI has no standardized way
a9130ea9 1891to deal with UTF-8 data. Please check the L<DBI documentation|DBI> to verify if
c8d992ba
A
1892that is still true.
1893
1894 sub fetchrow {
d88362ca
KW
1895 # $what is one of fetchrow_{array,hashref}
1896 my($self, $sth, $what) = @_;
b9cedb1b 1897 if ($] < 5.008) {
c8d992ba
A
1898 return $sth->$what;
1899 } else {
1900 require Encode;
1901 if (wantarray) {
1902 my @arr = $sth->$what;
1903 for (@arr) {
1904 defined && /[^\000-\177]/ && Encode::_utf8_on($_);
1905 }
1906 return @arr;
1907 } else {
1908 my $ret = $sth->$what;
1909 if (ref $ret) {
1910 for my $k (keys %$ret) {
d88362ca
KW
1911 defined
1912 && /[^\000-\177]/
1913 && Encode::_utf8_on($_) for $ret->{$k};
c8d992ba
A
1914 }
1915 return $ret;
1916 } else {
1917 defined && /[^\000-\177]/ && Encode::_utf8_on($_) for $ret;
1918 return $ret;
1919 }
1920 }
1921 }
1922 }
1923
1924
1925=item *
1926
1927A large scalar that you know can only contain ASCII
1928
1929Scalars that contain only ASCII and are marked as UTF-8 are sometimes
1930a drag to your program. If you recognize such a situation, just remove
2575c402 1931the UTF8 flag:
c8d992ba 1932
b9cedb1b 1933 utf8::downgrade($val) if $] > 5.008;
c8d992ba
A
1934
1935=back
1936
a6a7eedc
KW
1937=head1 BUGS
1938
1939See also L</The "Unicode Bug"> above.
1940
1941=head2 Interaction with Extensions
1942
1943When Perl exchanges data with an extension, the extension should be
1944able to understand the UTF8 flag and act accordingly. If the
1945extension doesn't recognize that flag, it's likely that the extension
1946will return incorrectly-flagged data.
1947
1948So if you're working with Unicode data, consult the documentation of
1949every module you're using if there are any issues with Unicode data
1950exchange. If the documentation does not talk about Unicode at all,
1951suspect the worst and probably look at the source to learn how the
1952module is implemented. Modules written completely in Perl shouldn't
1953cause problems. Modules that directly or indirectly access code written
1954in other programming languages are at risk.
1955
1956For affected functions, the simple strategy to avoid data corruption is
1957to always make the encoding of the exchanged data explicit. Choose an
1958encoding that you know the extension can handle. Convert arguments passed
1959to the extensions to that encoding and convert results back from that
1960encoding. Write wrapper functions that do the conversions for you, so
1961you can later change the functions when the extension catches up.
1962
1963To provide an example, let's say the popular C<Foo::Bar::escape_html>
1964function doesn't deal with Unicode data yet. The wrapper function
1965would convert the argument to raw UTF-8 and convert the result back to
1966Perl's internal representation like so:
1967
1968 sub my_escape_html ($) {
1969 my($what) = shift;
1970 return unless defined $what;
1971 Encode::decode_utf8(Foo::Bar::escape_html(
1972 Encode::encode_utf8($what)));
1973 }
1974
1975Sometimes, when the extension does not convert data but just stores
1976and retrieves it, you will be able to use the otherwise
1977dangerous L<C<Encode::_utf8_on()>|Encode/_utf8_on> function. Let's say
1978the popular C<Foo::Bar> extension, written in C, provides a C<param>
1979method that lets you store and retrieve data according to these prototypes:
1980
1981 $self->param($name, $value); # set a scalar
1982 $value = $self->param($name); # retrieve a scalar
1983
1984If it does not yet provide support for any encoding, one could write a
1985derived class with such a C<param> method:
1986
1987 sub param {
1988 my($self,$name,$value) = @_;
1989 utf8::upgrade($name); # make sure it is UTF-8 encoded
1990 if (defined $value) {
1991 utf8::upgrade($value); # make sure it is UTF-8 encoded
1992 return $self->SUPER::param($name,$value);
1993 } else {
1994 my $ret = $self->SUPER::param($name);
1995 Encode::_utf8_on($ret); # we know, it is UTF-8 encoded
1996 return $ret;
1997 }
1998 }
1999
2000Some extensions provide filters on data entry/exit points, such as
2001C<DB_File::filter_store_key> and family. Look out for such filters in
2002the documentation of your extensions; they can make the transition to
2003Unicode data much easier.
2004
2005=head2 Speed
2006
2007Some functions are slower when working on UTF-8 encoded strings than
2008on byte encoded strings. All functions that need to hop over
2009characters such as C<length()>, C<substr()> or C<index()>, or matching
2010regular expressions can work B<much> faster when the underlying data are
2011byte-encoded.
2012
2013In Perl 5.8.0 the slowness was often quite spectacular; in Perl 5.8.1
2014a caching scheme was introduced which improved the situation. In general,
2015operations with UTF-8 encoded strings are still slower. As an example,
2016the Unicode properties (character classes) like C<\p{Nd}> are known to
2017be quite a bit slower (5-20 times) than their simpler counterparts
2018like C<[0-9]> (then again, there are hundreds of Unicode characters matching
2019C<Nd> compared with the 10 ASCII characters matching C<[0-9]>).
2020
393fec97
GS
2021=head1 SEE ALSO
2022
51f494cc 2023L<perlunitut>, L<perluniintro>, L<perluniprops>, L<Encode>, L<open>, L<utf8>, L<bytes>,
b65e6125 2024L<perlretut>, L<perlvar/"${^UNICODE}">,
51f494cc 2025L<http://www.unicode.org/reports/tr44>).
393fec97
GS
2026
2027=cut