This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
APItest/t/utf8.t: Extract code to common function
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
f3cb6f94
KW
3 * This file contains definitions for use with the UTF-8 encoding. It
4 * actually also works with the variant UTF-8 encoding called UTF-EBCDIC, and
5 * hides almost all of the differences between these from the caller. In other
6 * words, someone should #include this file, and if the code is being compiled
7 * on an EBCDIC platform, things should mostly just work.
8 *
2eee27d7
SS
9 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009,
10 * 2010, 2011 by Larry Wall and others
a0ed51b3
LW
11 *
12 * You may distribute under the terms of either the GNU General Public
13 * License or the Artistic License, as specified in the README file.
14 *
15 */
16
57f0e7e2
KW
17#ifndef H_UTF8 /* Guard against recursive inclusion */
18#define H_UTF8 1
19
39e02b42 20/* Use UTF-8 as the default script encoding?
1e54db1a 21 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
22 * data (such as Latin-1) in string literals. */
23#ifdef USE_UTF8_SCRIPTS
24# define USE_UTF8_IN_NAMES (!IN_BYTES)
25#else
26# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
27#endif
28
3cd96634
KW
29#include "regcharclass.h"
30#include "unicode_constants.h"
31
051a06d4 32/* For to_utf8_fold_flags, q.v. */
e4f4ef45
KW
33#define FOLD_FLAGS_LOCALE 0x1
34#define FOLD_FLAGS_FULL 0x2
35#define FOLD_FLAGS_NOMIX_ASCII 0x4
051a06d4 36
83199d38
KW
37/* For _core_swash_init(), internal core use only */
38#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
5d3d13d1 39#define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2
87367d5f 40#define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4
83199d38 41
7bbfa158
KW
42/*
43=head1 Unicode Support
8cca77bc
KW
44L<perlguts/Unicode Support> has an introduction to this API.
45
46See also L</Character classification>,
47and L</Character case changing>.
48Various functions outside this section also work specially with Unicode.
49Search for the string "utf8" in this document.
7bbfa158
KW
50
51=for apidoc is_ascii_string
52
8871a094 53This is a misleadingly-named synonym for L</is_utf8_invariant_string>.
7bbfa158
KW
54On ASCII-ish platforms, the name isn't misleading: the ASCII-range characters
55are exactly the UTF-8 invariants. But EBCDIC machines have more invariants
8871a094
KW
56than just the ASCII characters, so C<is_utf8_invariant_string> is preferred.
57
58=for apidoc is_invariant_string
59
60This is a somewhat misleadingly-named synonym for L</is_utf8_invariant_string>.
61C<is_utf8_invariant_string> is preferred, as it indicates under what conditions
62the string is invariant.
7bbfa158
KW
63
64=cut
65*/
8871a094
KW
66#define is_ascii_string(s, len) is_utf8_invariant_string(s, len)
67#define is_invariant_string(s, len) is_utf8_invariant_string(s, len)
7bbfa158 68
de69f3af
KW
69#define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0)
70#define uvchr_to_utf8_flags(d,uv,flags) \
71 uvoffuni_to_utf8_flags(d,NATIVE_TO_UNI(uv),flags)
72#define utf8_to_uvchr_buf(s, e, lenp) \
842991ae 73 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \
de69f3af
KW
74 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY)
75
a0270393 76#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
445bf929
KW
77#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
78#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0)
79#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0)
80#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0)
36bb2ab6 81
eda9cac1
KW
82#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
83 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
baa60164 84#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
cea315b6 85#define FOLDEQ_LOCALE (1 << 1)
18f762c3
KW
86#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
87#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
d635b710
KW
88#define FOLDEQ_S1_FOLDS_SANE (1 << 4)
89#define FOLDEQ_S2_FOLDS_SANE (1 << 5)
a33c29bc 90
e6226b18
KW
91#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
92 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
93
1d72bdf6
NIS
94#ifdef EBCDIC
95/* The equivalent of these macros but implementing UTF-EBCDIC
96 are in the following header file:
97 */
98
99#include "utfebcdic.h"
fd7cb289 100
d06134e5 101#else /* ! EBCDIC */
73c4f7a1
GS
102START_EXTERN_C
103
111e8ed9
KW
104/* How wide can a single UTF-8 encoded character become in bytes. */
105/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
106 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
107 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
108 * non-negative integers in a binary format, even those above Unicode */
109#define UTF8_MAXBYTES 13
110
a0ed51b3 111#ifdef DOINIT
6f06b55f 112EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
113/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
114/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
115/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
116/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
117/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
118/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
119/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
120/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
121/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
122/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
123/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
124/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
125/* 0xC0 */ 2,2, /* overlong */
1ff3baa2 126/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
b2635aa8
KW
127/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
128/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
129/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
6937f885
KW
130 /* Perl extended (never was official UTF-8). Up to 36 bit */
131/* 0xFE */ 7,
132 /* More extended, Up to 72 bits (64-bit + reserved) */
111e8ed9 133/* 0xFF */ UTF8_MAXBYTES
a0ed51b3
LW
134};
135#else
6f06b55f 136EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
137#endif
138
73c4f7a1 139END_EXTERN_C
7e2040f0 140
1a3756de 141#if defined(_MSC_VER) && _MSC_VER < 1400
6f6d1bab
TC
142/* older MSVC versions have a smallish macro buffer */
143#define PERL_SMALL_MACRO_BUFFER
144#endif
145
59a449d5
KW
146/* Native character to/from iso-8859-1. Are the identity functions on ASCII
147 * platforms */
6f6d1bab 148#ifdef PERL_SMALL_MACRO_BUFFER
1a3756de
TC
149#define NATIVE_TO_LATIN1(ch) ((U8)(ch))
150#define LATIN1_TO_NATIVE(ch) ((U8)(ch))
6f6d1bab 151#else
4c8cd605
KW
152#define NATIVE_TO_LATIN1(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
153#define LATIN1_TO_NATIVE(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
6f6d1bab 154#endif
59a449d5
KW
155
156/* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
157 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
158 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
159 * because they are 8-bit encodings that serve the same purpose in Perl, and
160 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
161 * whichever one is applicable on the current platform */
6f6d1bab 162#ifdef PERL_SMALL_MACRO_BUFFER
1d4ea287
DD
163#define NATIVE_UTF8_TO_I8(ch) ((U8) (ch))
164#define I8_TO_NATIVE_UTF8(ch) ((U8) (ch))
6f6d1bab 165#else
4c8cd605
KW
166#define NATIVE_UTF8_TO_I8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
167#define I8_TO_NATIVE_UTF8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
6f6d1bab 168#endif
59a449d5 169
1d72bdf6 170/* Transforms in wide UV chars */
4c8cd605
KW
171#define UNI_TO_NATIVE(ch) ((UV) (ch))
172#define NATIVE_TO_UNI(ch) ((UV) (ch))
d7578b48 173
877d9f0d 174/*
9041c2e3 175
a14e0a36
KW
176 The following table is from Unicode 3.2, plus the Perl extensions for above
177 U+10FFFF
877d9f0d 178
a14e0a36 179 Code Points 1st Byte 2nd Byte 3rd 4th 5th 6th 7th 8th-13th
877d9f0d 180
375122d7 181 U+0000..U+007F 00..7F
e1b711da 182 U+0080..U+07FF * C2..DF 80..BF
a14e0a36
KW
183 U+0800..U+0FFF E0 * A0..BF 80..BF
184 U+1000..U+CFFF E1..EC 80..BF 80..BF
185 U+D000..U+D7FF ED 80..9F 80..BF
186 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
187 U+E000..U+FFFF EE..EF 80..BF 80..BF
188 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
189 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
190 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
191 Below are above-Unicode code points
192 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
193 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
194 U+200000..U+FFFFFF F8 * 88..BF 80..BF 80..BF 80..BF
195U+1000000..U+3FFFFFF F9..FB 80..BF 80..BF 80..BF 80..BF
196U+4000000..U+3FFFFFFF FC * 84..BF 80..BF 80..BF 80..BF 80..BF
197U+40000000..U+7FFFFFFF FD 80..BF 80..BF 80..BF 80..BF 80..BF
198U+80000000..U+FFFFFFFFF FE * 82..BF 80..BF 80..BF 80..BF 80..BF 80..BF
199U+1000000000.. FF 80..BF 80..BF 80..BF 80..BF 80..BF * 81..BF 80..BF
877d9f0d 200
e1b711da 201Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
202caused by legal UTF-8 avoiding non-shortest encodings: it is technically
203possible to UTF-8-encode a single code point in different ways, but that is
204explicitly forbidden, and the shortest possible encoding should always be used
15824458 205(and that is what Perl does). The non-shortest ones are called 'overlongs'.
8c007b5a 206
877d9f0d
JH
207 */
208
8c007b5a
JH
209/*
210 Another way to look at it, as bits:
211
b2635aa8 212 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 213
b2635aa8
KW
214 0aaa aaaa 0aaa aaaa
215 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
216 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
217 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
218
219As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 220leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
221encoded character.
222
df863e43
KW
223Perl's extended UTF-8 means we can have start bytes up through FF, though any
224beginning with FF yields a code point that is too large for 32-bit ASCII
225platforms. FF signals to use 13 bytes for the encoded character. This breaks
226the paradigm that the number of leading bits gives how many total bytes there
227are in the character.
65ab9279 228
8c007b5a
JH
229*/
230
eda91ad7
KW
231/* Anything larger than this will overflow the word if it were converted into a UV */
232#if defined(UV_IS_QUAD)
233# define HIGHEST_REPRESENTABLE_UTF8 "\xFF\x80\x8F\xBF\xBF\xBF\xBF\xBF\xBF\xBF\xBF\xBF\xBF"
234#else
235# define HIGHEST_REPRESENTABLE_UTF8 "\xFE\x83\xBF\xBF\xBF\xBF\xBF"
236#endif
237
6c88483e 238/* Is the representation of the Unicode code point 'cp' the same regardless of
15824458 239 * being encoded in UTF-8 or not? */
2d1545e5 240#define OFFUNI_IS_INVARIANT(cp) isASCII(cp)
15824458 241
c2b32798
KW
242/*
243=for apidoc Am|bool|UVCHR_IS_INVARIANT|UV cp
244
245Evaluates to 1 if the representation of code point C<cp> is the same whether or
246not it is encoded in UTF-8; otherwise evaluates to 0. UTF-8 invariant
247characters can be copied as-is when converting to/from UTF-8, saving time.
248C<cp> is Unicode if above 255; otherwise is platform-native.
249
250=cut
251 */
252
cf1be84e 253#define UVCHR_IS_INVARIANT(cp) OFFUNI_IS_INVARIANT(cp)
38953e5a 254
c9264833
KW
255/* This defines the bits that are to be in the continuation bytes of a multi-byte
256 * UTF-8 encoded character that mark it is a continuation byte. */
257#define UTF_CONTINUATION_MARK 0x80
258
a95ec4fb 259/* Misleadingly named: is the UTF8-encoded byte 'c' part of a variant sequence
5c06326b
KW
260 * in UTF-8? This is the inverse of UTF8_IS_INVARIANT. The |0 makes sure this
261 * isn't mistakenly called with a ptr argument */
262#define UTF8_IS_CONTINUED(c) (((U8)((c) | 0)) & UTF_CONTINUATION_MARK)
15824458
KW
263
264/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
265 * This doesn't catch invariants (they are single-byte). It also excludes the
5c06326b
KW
266 * illegal overlong sequences that begin with C0 and C1. The |0 makes sure
267 * this isn't mistakenly called with a ptr argument */
268#define UTF8_IS_START(c) (((U8)((c) | 0)) >= 0xc2)
15824458 269
858cd8ab
KW
270/* For use in UTF8_IS_CONTINUATION() below */
271#define UTF_IS_CONTINUATION_MASK 0xC0
272
15824458 273/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
5c06326b
KW
274 * first byte thereof? The |0 makes sure this isn't mistakenly called with a
275 * ptr argument */
858cd8ab 276#define UTF8_IS_CONTINUATION(c) \
5c06326b 277 ((((U8)((c) | 0)) & UTF_IS_CONTINUATION_MASK) == UTF_CONTINUATION_MARK)
0ae1fa71 278
15824458
KW
279/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
280 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
281 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
5c06326b
KW
282 * this matches 0xc[23]. The |0 makes sure this isn't mistakenly called with a
283 * ptr argument */
284#define UTF8_IS_DOWNGRADEABLE_START(c) ((((U8)((c) | 0)) & 0xfe) == 0xc2)
4ab10950 285
15824458 286/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
5c06326b
KW
287 * represent a code point > 255? The |0 makes sure this isn't mistakenly
288 * called with a ptr argument */
289#define UTF8_IS_ABOVE_LATIN1(c) (((U8)((c) | 0)) >= 0xc4)
8850bf83 290
15824458
KW
291/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
292 * sequence contributes to the specification of the code point. In the bit
293 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
294 * real information */
1d72bdf6 295#define UTF_ACCUMULATION_SHIFT 6
b2635aa8 296
fed423a5
KW
297/* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
298 * for more */
299#define QUESTION_MARK_CTRL DEL_NATIVE
300
301/* Surrogates, non-character code points and above-Unicode code points are
302 * problematic in some contexts. This allows code that needs to check for
303 * those to to quickly exclude the vast majority of code points it will
304 * encounter */
305#define isUTF8_POSSIBLY_PROBLEMATIC(c) ((U8) c >= 0xED)
306
784d4f31
KW
307/* A helper macro for isUTF8_CHAR, so use that one instead of this. This was
308 * generated by regen/regcharclass.pl, and then moved here. Then it was
309 * hand-edited to add some LIKELY() calls, presuming that malformations are
310 * unlikely. The lines that generated it were then commented out. This was
311 * done because it takes on the order of 10 minutes to generate, and is never
312 * going to change, unless the generated code is improved, and figuring out
313 * the LIKELYs there would be hard.
314 *
315 UTF8_CHAR: Matches legal UTF-8 variant code points up through 0x1FFFFFF
316
317 0x80 - 0x1FFFFF
318*/
319/*** GENERATED CODE ***/
320#define is_UTF8_CHAR_utf8_no_length_checks(s) \
321( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
322 ( LIKELY( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
323: ( 0xE0 == ((U8*)s)[0] ) ? \
324 ( LIKELY( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
325: ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \
326 ( LIKELY( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
327: ( 0xF0 == ((U8*)s)[0] ) ? \
328 ( LIKELY( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
329: ( ( ( ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF7 ) && LIKELY( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && LIKELY( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && LIKELY( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )
330
331/* The above macro handles UTF-8 that has this start byte as the maximum */
332#define _IS_UTF8_CHAR_HIGHEST_START_BYTE 0xF7
333
e23e8bc1
KW
334/* A helper macro for isSTRICT_UTF8_CHAR, so use that one instead of this.
335 * Like is_UTF8_CHAR_utf8_no_length_checks(), this was moved here and LIKELYs
336 * added manually.
337 *
338 STRICT_UTF8_CHAR: Matches legal Unicode UTF-8 variant code points, no
339 surrrogates nor non-character code points
340*/
341/*** GENERATED CODE ***/
342#define is_STRICT_UTF8_CHAR_utf8_no_length_checks(s) \
343( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
344 ( LIKELY( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
345: ( 0xE0 == ((U8*)s)[0] ) ? \
346 ( LIKELY( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
347: ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEC ) || 0xEE == ((U8*)s)[0] ) ?\
348 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
349: ( 0xED == ((U8*)s)[0] ) ? \
350 ( LIKELY( ( ( ((U8*)s)[1] & 0xE0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
351: ( 0xEF == ((U8*)s)[0] ) ? \
352 ( ( ( 0x80 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xB6 ) || ( 0xB8 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBE ) ) ?\
353 ( LIKELY( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ? 3 : 0 ) \
354 : ( 0xB7 == ((U8*)s)[1] ) ? \
355 ( LIKELY( ( ((U8*)s)[2] & 0xF0 ) == 0x80 || ( ((U8*)s)[2] & 0xF0 ) == 0xB0 ) ? 3 : 0 )\
356 : ( ( 0xBF == ((U8*)s)[1] ) && ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBD ) ) ? 3 : 0 )\
357: ( 0xF0 == ((U8*)s)[0] ) ? \
358 ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0x9E ) || ( 0xA0 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xAE ) || ( 0xB0 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBE ) ) ?\
359 ( LIKELY( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
360 : ( ((U8*)s)[1] == 0x9F || ( ( ((U8*)s)[1] & 0xEF ) == 0xAF ) ) ? \
361 ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBE ) ? \
362 ( LIKELY( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ? 4 : 0 ) \
363 : LIKELY( ( 0xBF == ((U8*)s)[2] ) && ( 0x80 <= ((U8*)s)[3] && ((U8*)s)[3] <= 0xBD ) ) ? 4 : 0 )\
364 : 0 ) \
365: ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF3 ) ? \
366 ( ( ( ( ((U8*)s)[1] & 0xC8 ) == 0x80 ) || ( ( ((U8*)s)[1] & 0xCC ) == 0x88 ) || ( ( ((U8*)s)[1] & 0xCE ) == 0x8C ) || ( ( ((U8*)s)[1] & 0xCF ) == 0x8E ) ) ?\
367 ( LIKELY( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
368 : ( ( ((U8*)s)[1] & 0xCF ) == 0x8F ) ? \
369 ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBE ) ? \
370 ( LIKELY( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ? 4 : 0 ) \
371 : LIKELY( ( 0xBF == ((U8*)s)[2] ) && ( 0x80 <= ((U8*)s)[3] && ((U8*)s)[3] <= 0xBD ) ) ? 4 : 0 )\
372 : 0 ) \
373: ( 0xF4 == ((U8*)s)[0] ) ? \
374 ( ( 0x80 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0x8E ) ? \
375 ( LIKELY( ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
376 : ( 0x8F == ((U8*)s)[1] ) ? \
377 ( ( 0x80 <= ((U8*)s)[2] && ((U8*)s)[2] <= 0xBE ) ? \
378 ( LIKELY( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ? 4 : 0 ) \
379 : LIKELY( ( 0xBF == ((U8*)s)[2] ) && ( 0x80 <= ((U8*)s)[3] && ((U8*)s)[3] <= 0xBD ) ) ? 4 : 0 )\
380 : 0 ) \
381: 0 )
382
a82be82b
KW
383/* Similarly,
384 C9_STRICT_UTF8_CHAR: Matches legal Unicode UTF-8 variant code
385 points, no surrogates
386 0x0080 - 0xD7FF
387 0xE000 - 0x10FFFF
388*/
389/*** GENERATED CODE ***/
390#define is_C9_STRICT_UTF8_CHAR_utf8_no_length_checks(s) \
391( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
392 ( LIKELY( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
393: ( 0xE0 == ((U8*)s)[0] ) ? \
394 ( LIKELY( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
395: ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEC ) || ( ((U8*)s)[0] & 0xFE ) == 0xEE ) ?\
396 ( LIKELY( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
397: ( 0xED == ((U8*)s)[0] ) ? \
398 ( LIKELY( ( ( ((U8*)s)[1] & 0xE0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
399: ( 0xF0 == ((U8*)s)[0] ) ? \
400 ( LIKELY( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
401: ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF3 ) ? \
402 ( LIKELY( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
403: LIKELY( ( ( ( 0xF4 == ((U8*)s)[0] ) && ( ( ((U8*)s)[1] & 0xF0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )
404
fed423a5
KW
405#endif /* EBCDIC vs ASCII */
406
407/* 2**UTF_ACCUMULATION_SHIFT - 1 */
408#define UTF_CONTINUATION_MASK ((U8) ((1U << UTF_ACCUMULATION_SHIFT) - 1))
409
7028aeba
KW
410/* Internal macro to be used only in this file to aid in constructing other
411 * publicly accessible macros.
412 * The number of bytes required to express this uv in UTF-8, for just those
413 * uv's requiring 2 through 6 bytes, as these are common to all platforms and
414 * word sizes. The number of bytes needed is given by the number of leading 1
415 * bits in the start byte. There are 32 start bytes that have 2 initial 1 bits
416 * (C0-DF); there are 16 that have 3 initial 1 bits (E0-EF); 8 that have 4
417 * initial 1 bits (F0-F8); 4 that have 5 initial 1 bits (F9-FB), and 2 that
418 * have 6 initial 1 bits (FC-FD). The largest number a string of n bytes can
419 * represent is (the number of possible start bytes for 'n')
420 * * (the number of possiblities for each start byte
421 * The latter in turn is
422 * 2 ** ( (how many continuation bytes there are)
423 * * (the number of bits of information each
424 * continuation byte holds))
425 *
426 * If we were on a platform where we could use a fast find first set bit
427 * instruction (or count leading zeros instruction) this could be replaced by
428 * using that to find the log2 of the uv, and divide that by the number of bits
429 * of information in each continuation byte, adjusting for large cases and how
430 * much information is in a start byte for that length */
72164d3a 431#define __COMMON_UNI_SKIP(uv) \
7028aeba
KW
432 (UV) (uv) < (32 * (1U << ( UTF_ACCUMULATION_SHIFT))) ? 2 : \
433 (UV) (uv) < (16 * (1U << (2 * UTF_ACCUMULATION_SHIFT))) ? 3 : \
434 (UV) (uv) < ( 8 * (1U << (3 * UTF_ACCUMULATION_SHIFT))) ? 4 : \
435 (UV) (uv) < ( 4 * (1U << (4 * UTF_ACCUMULATION_SHIFT))) ? 5 : \
436 (UV) (uv) < ( 2 * (1U << (5 * UTF_ACCUMULATION_SHIFT))) ? 6 :
72164d3a
KW
437
438/* Internal macro to be used only in this file.
439 * This adds to __COMMON_UNI_SKIP the details at this platform's upper range.
fed423a5 440 * For any-sized EBCDIC platforms, or 64-bit ASCII ones, we need one more test
72164d3a
KW
441 * to see if just 7 bytes is needed, or if the maximum is needed. For 32-bit
442 * ASCII platforms, everything is representable by 7 bytes */
fed423a5 443#if defined(UV_IS_QUAD) || defined(EBCDIC)
72164d3a 444# define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) \
7028aeba 445 (UV) (uv) < ((UV) 1U << (6 * UTF_ACCUMULATION_SHIFT)) ? 7 : UTF8_MAXBYTES)
1d68d6cd 446#else
72164d3a 447# define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) 7)
1d68d6cd
SC
448#endif
449
fed423a5
KW
450/* The next two macros use the base macro defined above, and add in the tests
451 * at the low-end of the range, for just 1 byte, yielding complete macros,
452 * publicly accessible. */
453
454/* Input is a true Unicode (not-native) code point */
455#define OFFUNISKIP(uv) (OFFUNI_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
2084b489 456
5352a763
KW
457/*
458
459=for apidoc Am|STRLEN|UVCHR_SKIP|UV cp
460returns the number of bytes required to represent the code point C<cp> when
461encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
462255; a Unicode code point otherwise.
463
464=cut
465 */
fdb6583d 466#define UVCHR_SKIP(uv) ( UVCHR_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
5352a763 467
b651802e
KW
468/* The largest code point representable by two UTF-8 bytes on this platform.
469 * As explained in the comments for __COMMON_UNI_SKIP, 32 start bytes with
fed423a5 470 * UTF_ACCUMULATION_SHIFT bits of information each */
aa206fb7
KW
471#define MAX_UTF8_TWO_BYTE (32 * (1U << UTF_ACCUMULATION_SHIFT) - 1)
472
b651802e
KW
473/* The largest code point representable by two UTF-8 bytes on any platform that
474 * Perl runs on. This value is constrained by EBCDIC which has 5 bits per
475 * continuation byte */
aa206fb7
KW
476#define MAX_PORTABLE_UTF8_TWO_BYTE (32 * (1U << 5) - 1)
477
c03c0950
KW
478/* The maximum number of UTF-8 bytes a single Unicode character can
479 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
480 * expansion is UTF8_MAX_FOLD_CHAR_EXPAND characters, but any above-Unicode
481 * code point will fold to itself, so we only have to look at the expansion of
482 * the maximum Unicode code point. But this number may be less than the space
483 * occupied by a very large code point under Perl's extended UTF-8. We have to
484 * make it large enough to fit any single character. (It turns out that ASCII
485 * and EBCDIC differ in which is larger) */
486#define UTF8_MAXBYTES_CASE \
487 (UTF8_MAXBYTES >= (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)) \
488 ? UTF8_MAXBYTES \
489 : (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)))
490
d06134e5
KW
491/* Rest of these are attributes of Unicode and perl's internals rather than the
492 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
493 * this level; the macros that some of these call may have different
494 * definitions in the two encodings */
495
59a449d5
KW
496/* In domain restricted to ASCII, these may make more sense to the reader than
497 * the ones with Latin1 in the name */
498#define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
499#define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
500
501/* More or less misleadingly-named defines, retained for back compat */
502#define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
503#define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
504#define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
505#define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
506#define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
d06134e5 507
c0236afe 508/* This defines the 1-bits that are to be in the first byte of a multi-byte
97d0ceda
KW
509 * UTF-8 encoded character that mark it as a start byte and give the number of
510 * bytes that comprise the character. 'len' is the number of bytes in the
511 * multi-byte sequence. */
c0236afe
KW
512#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
513
514/* Masks out the initial one bits in a start byte, leaving the real data ones.
515 * Doesn't work on an invariant byte. 'len' is the number of bytes in the
516 * multi-byte sequence that comprises the character. */
517#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
518
537124e4
KW
519/* Adds a UTF8 continuation byte 'new' of information to a running total code
520 * point 'old' of all the continuation bytes so far. This is designed to be
155d2738
KW
521 * used in a loop to convert from UTF-8 to the code point represented. Note
522 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is
523 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC)
524 * code point in process of being generated */
525#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
526 | ((NATIVE_UTF8_TO_I8((U8)new)) \
527 & UTF_CONTINUATION_MASK))
d06134e5 528
7c560c3b
KW
529/* If a value is anded with this, and the result is non-zero, then using the
530 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
531 * */
532#define UTF_ACCUMULATION_OVERFLOW_MASK \
533 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
534 - UTF_ACCUMULATION_SHIFT))
535
4ab10950
KW
536/* This works in the face of malformed UTF-8. */
537#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
538 && ( (e) - (s) > 1) \
539 && UTF8_IS_CONTINUATION(*((s)+1)))
540
5aaebcb3 541/* Number of bytes a code point occupies in UTF-8. */
5352a763 542#define NATIVE_SKIP(uv) UVCHR_SKIP(uv)
bd18bd40 543
5aaebcb3
KW
544/* Most code which says UNISKIP is really thinking in terms of native code
545 * points (0-255) plus all those beyond. This is an imprecise term, but having
2accb712 546 * it means existing code continues to work. For precision, use UVCHR_SKIP,
5352a763
KW
547 * NATIVE_SKIP, or OFFUNISKIP */
548#define UNISKIP(uv) UVCHR_SKIP(uv)
5aaebcb3 549
3c0792e4
KW
550/* Longer, but more accurate name */
551#define UTF8_IS_ABOVE_LATIN1_START(c) UTF8_IS_ABOVE_LATIN1(c)
552
a62b247b
KW
553/* Convert a UTF-8 variant Latin1 character to a native code point value.
554 * Needs just one iteration of accumulate. Should be used only if it is known
555 * that the code point is < 256, and is not UTF-8 invariant. Use the slower
556 * but more general TWO_BYTE_UTF8_TO_NATIVE() which handles any code point
557 * representable by two bytes (which turns out to be up through
558 * MAX_PORTABLE_UTF8_TWO_BYTE). The two parameters are:
559 * HI: a downgradable start byte;
560 * LO: continuation.
561 * */
562#define EIGHT_BIT_UTF8_TO_NATIVE(HI, LO) \
563 ( __ASSERT_(UTF8_IS_DOWNGRADEABLE_START(HI)) \
564 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
565 LATIN1_TO_NATIVE(UTF8_ACCUMULATE(( \
566 NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), (LO))))
567
94bb8c36 568/* Convert a two (not one) byte utf8 character to a native code point value.
2950f2a7
KW
569 * Needs just one iteration of accumulate. Should not be used unless it is
570 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
571 * Note that the result can be larger than 255 if the input character is not
572 * downgradable */
94bb8c36 573#define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
b18b15ad 574 ( __ASSERT_(PL_utf8skip[HI] == 2) \
635e76f5 575 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
94bb8c36 576 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
635e76f5 577 (LO))))
94bb8c36
KW
578
579/* Should never be used, and be deprecated */
580#define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
2950f2a7 581
bd18bd40
KW
582/*
583
584=for apidoc Am|STRLEN|UTF8SKIP|char* s
585returns the number of bytes in the UTF-8 encoded character whose first (perhaps
586only) byte is pointed to by C<s>.
587
588=cut
589 */
2a70536e
KW
590#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
591#define UTF8_SKIP(s) UTF8SKIP(s)
d06134e5 592
2d1545e5
KW
593/* Most code that says 'UNI_' really means the native value for code points up
594 * through 255 */
595#define UNI_IS_INVARIANT(cp) UVCHR_IS_INVARIANT(cp)
596
c2b32798
KW
597/*
598=for apidoc Am|bool|UTF8_IS_INVARIANT|char c
599
600Evaluates to 1 if the byte C<c> represents the same character when encoded in
601UTF-8 as when not; otherwise evaluates to 0. UTF-8 invariant characters can be
602copied as-is when converting to/from UTF-8, saving time.
603
604In spite of the name, this macro gives the correct result if the input string
605from which C<c> comes is not encoded in UTF-8.
606
607See C<L</UVCHR_IS_INVARIANT>> for checking if a UV is invariant.
608
609=cut
610
611The reason it works on both UTF-8 encoded strings and non-UTF-8 encoded, is
612that it returns TRUE in each for the exact same set of bit patterns. It is
613valid on a subset of what UVCHR_IS_INVARIANT is valid on, so can just use that;
614and the compiler should optimize out anything extraneous given the
615implementation of the latter. The |0 makes sure this isn't mistakenly called
616with a ptr argument.
617*/
5c06326b 618#define UTF8_IS_INVARIANT(c) UVCHR_IS_INVARIANT((c) | 0)
5fc230f1
KW
619
620/* Like the above, but its name implies a non-UTF8 input, which as the comments
621 * above show, doesn't matter as to its implementation */
38953e5a 622#define NATIVE_BYTE_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c)
d06134e5 623
48ccf5e1
KW
624/* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic
625 * bytes from an ordinal that is known to fit into exactly two (not one) bytes;
626 * it must be less than 0x3FF to work across both encodings. */
627
628/* These two are helper macros for the other three sets, and should not be used
629 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1
1ff3baa2
KW
630 * (which works for code points up through 0xFF) or NATIVE_TO_UNI which works
631 * for any code point */
48ccf5e1 632#define __BASE_TWO_BYTE_HI(c, translate_function) \
2863dafa 633 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
48ccf5e1 634 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \
2863dafa 635 | UTF_START_MARK(2)))
48ccf5e1 636#define __BASE_TWO_BYTE_LO(c, translate_function) \
2863dafa 637 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
48ccf5e1 638 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \
2863dafa 639 | UTF_CONTINUATION_MARK))
48ccf5e1 640
48ccf5e1
KW
641/* The next two macros should not be used. They were designed to be usable as
642 * the case label of a switch statement, but this doesn't work for EBCDIC. Use
9d0d3a03 643 * regen/unicode_constants.pl instead */
48ccf5e1
KW
644#define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)
645#define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)
646
647/* The next two macros are used when the source should be a single byte
648 * character; checked for under DEBUGGING */
649#define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
4c8cd605 650 ( __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
48ccf5e1 651#define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
4c8cd605 652 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
48ccf5e1
KW
653
654/* These final two macros in the series are used when the source can be any
655 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient
656 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~'
657 * operator instead of "<=" to avoid getting compiler warnings.
d52b8576 658 * MAX_UTF8_TWO_BYTE should be exactly all one bits in the lower few
48ccf5e1
KW
659 * places, so the ~ works */
660#define UTF8_TWO_BYTE_HI(c) \
661 (__ASSERT_((sizeof(c) == 1) \
d52b8576 662 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
4c8cd605 663 (__BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
48ccf5e1
KW
664#define UTF8_TWO_BYTE_LO(c) \
665 (__ASSERT_((sizeof(c) == 1) \
d52b8576 666 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
4c8cd605 667 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
d06134e5 668
e7214ce8
KW
669/* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
670 * as it is only in overlongs. */
671#define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1)
672
7e2040f0 673/*
e3036cf4 674 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
675 * from an earlier version of these macros in which they didn't call the
676 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
677 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
678 * no need to do it again here
7e2040f0 679 */
97d0ceda
KW
680#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF) \
681 ? isIDFIRST(*(p)) \
e3036cf4 682 : isIDFIRST_utf8((const U8*)p))
97d0ceda
KW
683#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF)) \
684 ? isWORDCHAR(*(p)) \
32636478
KW
685 : isWORDCHAR_utf8((const U8*)p))
686#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
1d72bdf6 687
89ebb4a3
JH
688#define UTF8_MAXLEN UTF8_MAXBYTES
689
8cb75cc8
KW
690/* A Unicode character can fold to up to 3 characters */
691#define UTF8_MAX_FOLD_CHAR_EXPAND 3
692
a98fe34d 693#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
bd18bd40
KW
694
695/*
696
697=for apidoc Am|bool|DO_UTF8|SV* sv
698Returns a bool giving whether or not the PV in C<sv> is to be treated as being
699encoded in UTF-8.
700
701You should use this I<after> a call to C<SvPV()> or one of its variants, in
702case any call to string overloading updates the internal UTF-8 encoding flag.
703
704=cut
705*/
0064a8a9 706#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1ff3baa2
KW
707
708/* Should all strings be treated as Unicode, and not just UTF-8 encoded ones?
709 * Is so within 'feature unicode_strings' or 'locale :not_characters', and not
710 * within 'use bytes'. UTF-8 locales are not tested for here, but perhaps
711 * could be */
712#define IN_UNI_8_BIT \
d6ded950
KW
713 (((CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT)) \
714 || (CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \
715 /* -1 below is for :not_characters */ \
716 && _is_in_locale_category(FALSE, -1))) \
717 && ! IN_BYTES)
b36bf33f 718
1d72bdf6 719
c76687c5
KW
720#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
721
722/* Allow first byte to be a continuation byte */
1d72bdf6 723#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
724
725/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 726#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
727
728/* expecting more bytes than were available in the string */
729#define UTF8_ALLOW_SHORT 0x0008
730
731/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
732#define UTF8_ALLOW_LONG 0x0010
733
734#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
735#define UTF8_WARN_SURROGATE 0x0040
736
737#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
738#define UTF8_WARN_NONCHAR 0x0100 /* code points */
739
740#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
741#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
742
d35f2ca5
KW
743/* Code points which never were part of the original UTF-8 standard, which only
744 * went up to 2 ** 31 - 1. Note that these all overflow a signed 32-bit word,
745 * The first byte of these code points is FE or FF on ASCII platforms. If the
746 * first byte is FF, it will overflow a 32-bit word. */
747#define UTF8_DISALLOW_ABOVE_31_BIT 0x0800
748#define UTF8_WARN_ABOVE_31_BIT 0x1000
749
750/* For back compat, these old names are misleading for UTF_EBCDIC */
751#define UTF8_DISALLOW_FE_FF UTF8_DISALLOW_ABOVE_31_BIT
752#define UTF8_WARN_FE_FF UTF8_WARN_ABOVE_31_BIT
949cf498
KW
753
754#define UTF8_CHECK_ONLY 0x2000
755
756/* For backwards source compatibility. They do nothing, as the default now
757 * includes what they used to mean. The first one's meaning was to allow the
758 * just the single non-character 0xFFFF */
759#define UTF8_ALLOW_FFFF 0
760#define UTF8_ALLOW_SURROGATE 0
761
ecc1615f
KW
762/* C9 refers to Unicode Corrigendum #9: allows but discourages non-chars */
763#define UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE \
764 (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_SURROGATE)
765#define UTF8_WARN_ILLEGAL_C9_INTERCHANGE (UTF8_WARN_SUPER|UTF8_WARN_SURROGATE)
766
d35f2ca5 767#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE \
ecc1615f 768 (UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE|UTF8_DISALLOW_NONCHAR)
949cf498 769#define UTF8_WARN_ILLEGAL_INTERCHANGE \
ecc1615f
KW
770 (UTF8_WARN_ILLEGAL_C9_INTERCHANGE|UTF8_WARN_NONCHAR)
771
93e6dbd6
KW
772#define UTF8_ALLOW_ANY \
773 (~( UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_DISALLOW_ABOVE_31_BIT \
774 |UTF8_WARN_ILLEGAL_INTERCHANGE|UTF8_WARN_ABOVE_31_BIT))
949cf498
KW
775#define UTF8_ALLOW_ANYUV \
776 (UTF8_ALLOW_EMPTY \
777 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
778#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
779 UTF8_ALLOW_ANYUV)
1d72bdf6 780
89d986df
KW
781/*
782=for apidoc Am|bool|UTF8_IS_SURROGATE|const U8 *s|const U8 *e
783
784Evaluates to non-zero if the first few bytes of the string starting at C<s> and
785looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one
786of the Unicode surrogate code points; otherwise it evaluates to 0. If
787non-zero, the value gives how many bytes starting at C<s> comprise the code
788point's representation.
789
790=cut
791 */
792#define UTF8_IS_SURROGATE(s, e) is_SURROGATE_utf8_safe(s, e)
793
794
795#define UTF8_IS_REPLACEMENT(s, send) is_REPLACEMENT_utf8_safe(s,send)
796
797/*
798=for apidoc Am|bool|UTF8_IS_SUPER|const U8 *s|const U8 *e
799
800Recall that Perl recognizes an extension to UTF-8 that can encode code
801points larger than the ones defined by Unicode, which are 0..0x10FFFF.
802
803This macro evaluates to non-zero if the first few bytes of the string starting
804at C<s> and looking no further than S<C<e - 1>> are from this UTF-8 extension;
805otherwise it evaluates to 0. If non-zero, the value gives how many bytes
806starting at C<s> comprise the code point's representation.
807
8080 is returned if the bytes are not well-formed extended UTF-8, or if they
809represent a code point that cannot fit in a UV on the current platform. Hence
810this macro can give different results when run on a 64-bit word machine than on
811one with a 32-bit word size.
0c58a72b 812
89d986df
KW
813Note that it is deprecated to have code points that are larger than what can
814fit in an IV on the current machine.
7131f24d 815
89d986df
KW
816=cut
817
818 * ASCII EBCDIC I8
7131f24d
KW
819 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
820 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
821 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
89d986df
KW
822 */
823#ifdef EBCDIC
a14e0a36 824# define UTF8_IS_SUPER(s, e) \
89d986df
KW
825 (( LIKELY((e) > (s) + 4) \
826 && NATIVE_UTF8_TO_I8(*(s)) >= 0xF9 \
827 && ( NATIVE_UTF8_TO_I8(*(s)) > 0xF9 \
828 || (NATIVE_UTF8_TO_I8(*(s) + 1) >= 0xA2)) \
829 && LIKELY((s) + UTF8SKIP(s) <= (e))) \
2b479609 830 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0)
7131f24d 831#else
a14e0a36 832# define UTF8_IS_SUPER(s, e) \
89d986df
KW
833 (( LIKELY((e) > (s) + 3) \
834 && (*(U8*) (s)) >= 0xF4 \
835 && ((*(U8*) (s)) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))\
836 && LIKELY((s) + UTF8SKIP(s) <= (e))) \
2b479609 837 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0)
7131f24d
KW
838#endif
839
b96a92fb
KW
840/* These are now machine generated, and the 'given' clause is no longer
841 * applicable */
0c58a72b 842#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) \
89d986df
KW
843 cBOOL(is_NONCHAR_utf8_safe(s,e))
844
845/*
846=for apidoc Am|bool|UTF8_IS_NONCHAR|const U8 *s|const U8 *e
847
848Evaluates to non-zero if the first few bytes of the string starting at C<s> and
849looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one
850of the Unicode non-character code points; otherwise it evaluates to 0. If
851non-zero, the value gives how many bytes starting at C<s> comprise the code
852point's representation.
853
854=cut
855 */
0c58a72b
KW
856#define UTF8_IS_NONCHAR(s, e) \
857 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e)
7131f24d 858
c867b360
JH
859#define UNICODE_SURROGATE_FIRST 0xD800
860#define UNICODE_SURROGATE_LAST 0xDFFF
861#define UNICODE_REPLACEMENT 0xFFFD
862#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 863
b851fbc1 864/* Though our UTF-8 encoding can go beyond this,
c76687c5 865 * let's be conservative and do as Unicode says. */
b851fbc1
JH
866#define PERL_UNICODE_MAX 0x10FFFF
867
d35f2ca5
KW
868#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
869#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
870#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
871#define UNICODE_WARN_ABOVE_31_BIT 0x0008 /* Above 0x7FFF_FFFF */
872#define UNICODE_DISALLOW_SURROGATE 0x0010
873#define UNICODE_DISALLOW_NONCHAR 0x0020
874#define UNICODE_DISALLOW_SUPER 0x0040
875#define UNICODE_DISALLOW_ABOVE_31_BIT 0x0080
ecc1615f
KW
876#define UNICODE_WARN_ILLEGAL_C9_INTERCHANGE \
877 (UNICODE_WARN_SURROGATE|UNICODE_WARN_SUPER)
bb88be5f 878#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
ecc1615f
KW
879 (UNICODE_WARN_ILLEGAL_C9_INTERCHANGE|UNICODE_WARN_NONCHAR)
880#define UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE \
881 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_SUPER)
bb88be5f 882#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
ecc1615f 883 (UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE|UNICODE_DISALLOW_NONCHAR)
949cf498
KW
884
885/* For backward source compatibility, as are now the default */
886#define UNICODE_ALLOW_SURROGATE 0
887#define UNICODE_ALLOW_SUPER 0
888#define UNICODE_ALLOW_ANY 0
b851fbc1 889
2d6b3d38
KW
890/* This matches the 2048 code points between UNICODE_SURROGATE_FIRST (0xD800) and
891 * UNICODE_SURROGATE_LAST (0xDFFF) */
892#define UNICODE_IS_SURROGATE(uv) (((UV) (uv) & (~0xFFFF | 0xF800)) \
893 == 0xD800)
894
646d1759
KW
895#define UNICODE_IS_REPLACEMENT(uv) ((UV) (uv) == UNICODE_REPLACEMENT)
896#define UNICODE_IS_BYTE_ORDER_MARK(uv) ((UV) (uv) == UNICODE_BYTE_ORDER_MARK)
c149ab20
KW
897
898/* Is 'uv' one of the 32 contiguous-range noncharacters? */
899#define UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) ((UV) (uv) >= 0xFDD0 \
900 && (UV) (uv) <= 0xFDEF)
901
902/* Is 'uv' one of the 34 plane-ending noncharacters 0xFFFE, 0xFFFF, 0x1FFFE,
903 * 0x1FFFF, ... 0x10FFFE, 0x10FFFF, given that we know that 'uv' is not above
904 * the Unicode legal max */
905#define UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv) \
906 (((UV) (uv) & 0xFFFE) == 0xFFFE)
907
908#define UNICODE_IS_NONCHAR(uv) \
909 ( UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) \
910 || ( LIKELY( ! UNICODE_IS_SUPER(uv)) \
911 && UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv)))
912
913#define UNICODE_IS_SUPER(uv) ((UV) (uv) > PERL_UNICODE_MAX)
d35f2ca5 914#define UNICODE_IS_ABOVE_31_BIT(uv) ((UV) (uv) > 0x7FFFFFFF)
1d72bdf6 915
ec34087a
KW
916#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
917#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
918 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
919#define MICRO_SIGN MICRO_SIGN_NATIVE
920#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
921 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
922#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
923 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
924#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
925#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
926#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 927#define GREEK_SMALL_LETTER_MU 0x03BC
9e682c18
KW
928#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case
929 of MICRON */
930#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
0766489e
KW
931#ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8
932# define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
933#endif
74894415
KW
934#define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130
935#define LATIN_SMALL_LETTER_DOTLESS_I 0x131
9e682c18 936#define LATIN_SMALL_LETTER_LONG_S 0x017F
a9f50d33
KW
937#define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05
938#define LATIN_SMALL_LIGATURE_ST 0xFB06
9e682c18
KW
939#define KELVIN_SIGN 0x212A
940#define ANGSTROM_SIGN 0x212B
09091399 941
9e55ce06 942#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
943#define UNI_DISPLAY_BACKSLASH 0x0002
944#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
945#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 946
5cd46e1f
KW
947#define ANYOF_FOLD_SHARP_S(node, input, end) \
948 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 949 (ANYOF_NONBITMAP(node)) && \
39065660 950 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f 951 ((end) > (input) + 1) && \
305b8651 952 isALPHA_FOLD_EQ((input)[0], 's'))
6302f837 953
ebc501f0 954#define SHARP_S_SKIP 2
3b0fc154 955
6302f837 956/*
5dca9278
KW
957
958=for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e
959
35f8c9bd 960Evaluates to non-zero if the first few bytes of the string starting at C<s> and
2c6ed66c
KW
961looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl,
962that represents some code point; otherwise it evaluates to 0. If non-zero, the
963value gives how many bytes starting at C<s> comprise the code point's
2717076a
KW
964representation. Any bytes remaining before C<e>, but beyond the ones needed to
965form the first code point in C<s>, are not examined.
6302f837 966
35f8c9bd
KW
967The code point can be any that will fit in a UV on this machine, using Perl's
968extension to official UTF-8 to represent those higher than the Unicode maximum
969of 0x10FFFF. That means that this macro is used to efficiently decide if the
2717076a
KW
970next few bytes in C<s> is legal UTF-8 for a single character.
971
972Use C<L</isSTRICT_UTF8_CHAR>> to restrict the acceptable code points to those
973defined by Unicode to be fully interchangeable across applications;
974C<L</isC9_STRICT_UTF8_CHAR>> to use the L<Unicode Corrigendum
975#9|http://www.unicode.org/versions/corrigendum9.html> definition of allowable
976code points; and C<L</isUTF8_CHAR_flags>> for a more customized definition.
977
978Use C<L</is_utf8_string>>, C<L</is_utf8_string_loc>>, and
979C<L</is_utf8_string_loclen>> to check entire strings.
35f8c9bd
KW
980
981Note that it is deprecated to use code points higher than what will fit in an
982IV. This macro does not raise any warnings for such code points, treating them
983as valid.
984
985Note also that a UTF-8 INVARIANT character (i.e. ASCII on non-EBCDIC machines)
986is a valid UTF-8 character.
5dca9278
KW
987
988=cut
989*/
6302f837 990
784d4f31
KW
991#define isUTF8_CHAR(s, e) \
992 (UNLIKELY((e) <= (s)) \
993 ? 0 \
994 : (UTF8_IS_INVARIANT(*s)) \
df863e43
KW
995 ? 1 \
996 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
997 ? 0 \
998 : LIKELY(NATIVE_UTF8_TO_I8(*s) <= _IS_UTF8_CHAR_HIGHEST_START_BYTE) \
999 ? is_UTF8_CHAR_utf8_no_length_checks(s) \
1000 : _is_utf8_char_helper(s, e, 0))
6302f837 1001
3cedd9d9
KW
1002#define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end)
1003
e23e8bc1
KW
1004/*
1005
1006=for apidoc Am|STRLEN|isSTRICT_UTF8_CHAR|const U8 *s|const U8 *e
1007
1008Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1009looking no further than S<C<e - 1>> are well-formed UTF-8 that represents some
1010Unicode code point completely acceptable for open interchange between all
1011applications; otherwise it evaluates to 0. If non-zero, the value gives how
2717076a
KW
1012many bytes starting at C<s> comprise the code point's representation. Any
1013bytes remaining before C<e>, but beyond the ones needed to form the first code
1014point in C<s>, are not examined.
e23e8bc1
KW
1015
1016The largest acceptable code point is the Unicode maximum 0x10FFFF, and must not
1017be a surrogate nor a non-character code point. Thus this excludes any code
1018point from Perl's extended UTF-8.
1019
1020This is used to efficiently decide if the next few bytes in C<s> is
2717076a
KW
1021legal Unicode-acceptable UTF-8 for a single character.
1022
1023Use C<L</isC9_STRICT_UTF8_CHAR>> to use the L<Unicode Corrigendum
1024#9|http://www.unicode.org/versions/corrigendum9.html> definition of allowable
1025code points; C<L</isUTF8_CHAR>> to check for Perl's extended UTF-8;
1026and C<L</isUTF8_CHAR_flags>> for a more customized definition.
e23e8bc1 1027
9f2abfde
KW
1028Use C<L</is_strict_utf8_string>>, C<L</is_strict_utf8_string_loc>>, and
1029C<L</is_strict_utf8_string_loclen>> to check entire strings.
1030
e23e8bc1
KW
1031=cut
1032*/
1033
1034#define isSTRICT_UTF8_CHAR(s, e) \
1035 (UNLIKELY((e) <= (s)) \
1036 ? 0 \
1037 : (UTF8_IS_INVARIANT(*s)) \
1038 ? 1 \
1039 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
1040 ? 0 \
1041 : is_STRICT_UTF8_CHAR_utf8_no_length_checks(s))
1042
a82be82b
KW
1043/*
1044
1045=for apidoc Am|STRLEN|isC9_STRICT_UTF8_CHAR|const U8 *s|const U8 *e
1046
1047Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1048looking no further than S<C<e - 1>> are well-formed UTF-8 that represents some
1049Unicode non-surrogate code point; otherwise it evaluates to 0. If non-zero,
1050the value gives how many bytes starting at C<s> comprise the code point's
2717076a
KW
1051representation. Any bytes remaining before C<e>, but beyond the ones needed to
1052form the first code point in C<s>, are not examined.
a82be82b
KW
1053
1054The largest acceptable code point is the Unicode maximum 0x10FFFF. This
1055differs from C<L</isSTRICT_UTF8_CHAR>> only in that it accepts non-character
1056code points. This corresponds to
1057L<Unicode Corrigendum #9|http://www.unicode.org/versions/corrigendum9.html>.
1058which said that non-character code points are merely discouraged rather than
1059completely forbidden in open interchange. See
1060L<perlunicode/Noncharacter code points>.
1061
2717076a
KW
1062Use C<L</isUTF8_CHAR>> to check for Perl's extended UTF-8; and
1063C<L</isUTF8_CHAR_flags>> for a more customized definition.
1064
9f2abfde
KW
1065Use C<L</is_c9strict_utf8_string>>, C<L</is_c9strict_utf8_string_loc>>, and
1066C<L</is_c9strict_utf8_string_loclen>> to check entire strings.
1067
a82be82b
KW
1068=cut
1069*/
1070
1071#define isC9_STRICT_UTF8_CHAR(s, e) \
1072 (UNLIKELY((e) <= (s)) \
1073 ? 0 \
1074 : (UTF8_IS_INVARIANT(*s)) \
1075 ? 1 \
1076 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
1077 ? 0 \
1078 : is_C9_STRICT_UTF8_CHAR_utf8_no_length_checks(s))
1079
25e3a4e0
KW
1080/*
1081
1082=for apidoc Am|STRLEN|isUTF8_CHAR_flags|const U8 *s|const U8 *e| const U32 flags
1083
1084Evaluates to non-zero if the first few bytes of the string starting at C<s> and
1085looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl,
1086that represents some code point, subject to the restrictions given by C<flags>;
1087otherwise it evaluates to 0. If non-zero, the value gives how many bytes
2717076a
KW
1088starting at C<s> comprise the code point's representation. Any bytes remaining
1089before C<e>, but beyond the ones needed to form the first code point in C<s>,
1090are not examined.
25e3a4e0
KW
1091
1092If C<flags> is 0, this gives the same results as C<L</isUTF8_CHAR>>;
1093if C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results
1094as C<L</isSTRICT_UTF8_CHAR>>;
1095and if C<flags> is C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives
1096the same results as C<L</isC9_STRICT_UTF8_CHAR>>.
1097Otherwise C<flags> may be any combination of the C<UTF8_DISALLOW_I<foo>> flags
1098understood by C<L</utf8n_to_uvchr>>, with the same meanings.
1099
1100The three alternative macros are for the most commonly needed validations; they
1101are likely to run somewhat faster than this more general one, as they can be
1102inlined into your code.
1103
9f2abfde
KW
1104Use L</is_utf8_string_flags>, L</is_utf8_string_loc_flags>, and
1105L</is_utf8_string_loclen_flags> to check entire strings.
1106
25e3a4e0
KW
1107=cut
1108*/
1109
1110#define isUTF8_CHAR_flags(s, e, flags) \
1111 (UNLIKELY((e) <= (s)) \
1112 ? 0 \
1113 : (UTF8_IS_INVARIANT(*s)) \
1114 ? 1 \
1115 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
1116 ? 0 \
1117 : _is_utf8_char_helper(s, e, flags))
1118
6302f837
KW
1119/* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is
1120 * retained solely for backwards compatibility */
1121#define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n)
e9a8c099 1122
57f0e7e2
KW
1123#endif /* H_UTF8 */
1124
e9a8c099 1125/*
14d04a33 1126 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 1127 */