This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.h: Refactor a macro
[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
53This is a misleadingly-named synonym for L</is_invariant_string>.
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
56than just the ASCII characters, so C<is_invariant_string> is preferred.
57
58=cut
59*/
60#define is_ascii_string(s, len) is_invariant_string(s, len)
61
de69f3af
KW
62#define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0)
63#define uvchr_to_utf8_flags(d,uv,flags) \
64 uvoffuni_to_utf8_flags(d,NATIVE_TO_UNI(uv),flags)
65#define utf8_to_uvchr_buf(s, e, lenp) \
842991ae 66 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \
de69f3af
KW
67 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY)
68
a0270393 69#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
445bf929
KW
70#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
71#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0)
72#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0)
73#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0)
36bb2ab6 74
fd7cb289 75/* Source backward compatibility. */
fd7cb289
RGS
76#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
77
eda9cac1
KW
78#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
79 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
baa60164 80#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
cea315b6 81#define FOLDEQ_LOCALE (1 << 1)
18f762c3
KW
82#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
83#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
d635b710
KW
84#define FOLDEQ_S1_FOLDS_SANE (1 << 4)
85#define FOLDEQ_S2_FOLDS_SANE (1 << 5)
a33c29bc 86
e6226b18
KW
87#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
88 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
89
1d72bdf6
NIS
90#ifdef EBCDIC
91/* The equivalent of these macros but implementing UTF-EBCDIC
92 are in the following header file:
93 */
94
95#include "utfebcdic.h"
fd7cb289 96
d06134e5 97#else /* ! EBCDIC */
73c4f7a1
GS
98START_EXTERN_C
99
111e8ed9
KW
100/* How wide can a single UTF-8 encoded character become in bytes. */
101/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
102 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
103 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
104 * non-negative integers in a binary format, even those above Unicode */
105#define UTF8_MAXBYTES 13
106
a0ed51b3 107#ifdef DOINIT
6f06b55f 108EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
109/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
110/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
111/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
112/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
113/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
114/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
115/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
116/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
117/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
118/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
119/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
120/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
121/* 0xC0 */ 2,2, /* overlong */
1ff3baa2 122/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
b2635aa8
KW
123/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
124/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
125/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
6937f885
KW
126 /* Perl extended (never was official UTF-8). Up to 36 bit */
127/* 0xFE */ 7,
128 /* More extended, Up to 72 bits (64-bit + reserved) */
111e8ed9 129/* 0xFF */ UTF8_MAXBYTES
a0ed51b3
LW
130};
131#else
6f06b55f 132EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
133#endif
134
73c4f7a1 135END_EXTERN_C
7e2040f0 136
59a449d5
KW
137/* Native character to/from iso-8859-1. Are the identity functions on ASCII
138 * platforms */
4c8cd605
KW
139#define NATIVE_TO_LATIN1(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
140#define LATIN1_TO_NATIVE(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
59a449d5
KW
141
142/* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
143 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
144 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
145 * because they are 8-bit encodings that serve the same purpose in Perl, and
146 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
147 * whichever one is applicable on the current platform */
4c8cd605
KW
148#define NATIVE_UTF8_TO_I8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
149#define I8_TO_NATIVE_UTF8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
59a449d5 150
1d72bdf6 151/* Transforms in wide UV chars */
4c8cd605
KW
152#define UNI_TO_NATIVE(ch) ((UV) (ch))
153#define NATIVE_TO_UNI(ch) ((UV) (ch))
d7578b48 154
877d9f0d 155/*
9041c2e3 156
8c007b5a 157 The following table is from Unicode 3.2.
877d9f0d
JH
158
159 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
160
375122d7 161 U+0000..U+007F 00..7F
e1b711da 162 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 163 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 164 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 165 U+D000..U+D7FF ED 80..9F 80..BF
537124e4 166 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
375122d7 167 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 168 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
169 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
170 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
171 Below are non-Unicode code points
172 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
173 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
537124e4 174 U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 175
e1b711da 176Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
177caused by legal UTF-8 avoiding non-shortest encodings: it is technically
178possible to UTF-8-encode a single code point in different ways, but that is
179explicitly forbidden, and the shortest possible encoding should always be used
15824458 180(and that is what Perl does). The non-shortest ones are called 'overlongs'.
8c007b5a 181
877d9f0d
JH
182 */
183
8c007b5a
JH
184/*
185 Another way to look at it, as bits:
186
b2635aa8 187 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 188
b2635aa8
KW
189 0aaa aaaa 0aaa aaaa
190 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
191 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
192 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
193
194As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 195leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
196encoded character.
197
65ab9279
TC
198Perl's extended UTF-8 means we can have start bytes up to FF.
199
8c007b5a
JH
200*/
201
6c88483e 202/* Is the representation of the Unicode code point 'cp' the same regardless of
15824458 203 * being encoded in UTF-8 or not? */
2d1545e5 204#define OFFUNI_IS_INVARIANT(cp) isASCII(cp)
15824458 205
38953e5a
KW
206/* Is the representation of the code point 'cp' the same regardless of
207 * being encoded in UTF-8 or not? 'cp' is native if < 256; Unicode otherwise
208 * */
cf1be84e 209#define UVCHR_IS_INVARIANT(cp) OFFUNI_IS_INVARIANT(cp)
38953e5a 210
c9264833
KW
211/* This defines the bits that are to be in the continuation bytes of a multi-byte
212 * UTF-8 encoded character that mark it is a continuation byte. */
213#define UTF_CONTINUATION_MARK 0x80
214
a95ec4fb
KW
215/* Misleadingly named: is the UTF8-encoded byte 'c' part of a variant sequence
216 * in UTF-8? This is the inverse of UTF8_IS_INVARIANT */
c9264833 217#define UTF8_IS_CONTINUED(c) (((U8)c) & UTF_CONTINUATION_MARK)
15824458
KW
218
219/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
220 * This doesn't catch invariants (they are single-byte). It also excludes the
221 * illegal overlong sequences that begin with C0 and C1. */
e4f4ef45 222#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
15824458
KW
223
224/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
225 * first byte thereof? */
c9264833 226#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == UTF_CONTINUATION_MARK)
0ae1fa71 227
15824458
KW
228/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
229 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
230 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
231 * this matches 0xc[23]. */
559c7f10 232#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2)
4ab10950 233
15824458
KW
234/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
235 * represent a code point > 255? */
e4f4ef45 236#define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4)
8850bf83 237
15824458
KW
238/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
239 * sequence contributes to the specification of the code point. In the bit
240 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
241 * real information */
1d72bdf6 242#define UTF_ACCUMULATION_SHIFT 6
b2635aa8 243
6588300d
NC
244#if UVSIZE >= 8
245# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
72164d3a 246#endif
6588300d 247
5aaebcb3 248/* Input is a true Unicode (not-native) code point */
72164d3a 249#define __COMMON_UNI_SKIP(uv) \
1d68d6cd
SC
250 (uv) < 0x800 ? 2 : \
251 (uv) < 0x10000 ? 3 : \
252 (uv) < 0x200000 ? 4 : \
253 (uv) < 0x4000000 ? 5 : \
72164d3a
KW
254 (uv) < 0x80000000 ? 6 :
255
256/* Internal macro to be used only in this file.
257 * This adds to __COMMON_UNI_SKIP the details at this platform's upper range.
258 * For 64-bit ASCII platforms, we need one more test
259 * to see if just 7 bytes is needed, or if the maximum is needed. For 32-bit
260 * ASCII platforms, everything is representable by 7 bytes */
261#ifdef UV_IS_QUAD
262# define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) \
263 (uv) < UTF8_QUAD_MAX ? 7 : UTF8_MAXBYTES )
1d68d6cd 264#else
72164d3a 265# define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) 7)
1d68d6cd
SC
266#endif
267
72164d3a 268#define OFFUNISKIP(uv) ( OFFUNI_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
5352a763
KW
269/*
270
271=for apidoc Am|STRLEN|UVCHR_SKIP|UV cp
272returns the number of bytes required to represent the code point C<cp> when
273encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
274255; a Unicode code point otherwise.
275
276=cut
277 */
fdb6583d 278#define UVCHR_SKIP(uv) ( UVCHR_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
5352a763 279
4ab16ae1
KW
280/* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
281 * for more */
282#define QUESTION_MARK_CTRL DEL_NATIVE
283
5d5376e2
KW
284/* Surrogates, non-character code points and above-Unicode code points are
285 * problematic in some contexts. This allows code that needs to check for
286 * those to to quickly exclude the vast majority of code points it will
287 * encounter */
288#define isUTF8_POSSIBLY_PROBLEMATIC(c) ((U8) c >= 0xED)
289
d06134e5
KW
290#endif /* EBCDIC vs ASCII */
291
6bff35ff
KW
292/* 2**UTF_ACCUMULATION_SHIFT - 1 */
293#define UTF_CONTINUATION_MASK ((U8) ((1U << UTF_ACCUMULATION_SHIFT) - 1))
294
aa206fb7
KW
295/* 32 start bytes with UTF_ACCUMULATION_SHIFT bits of information each */
296#define MAX_UTF8_TWO_BYTE (32 * (1U << UTF_ACCUMULATION_SHIFT) - 1)
297
298/* constrained by EBCDIC which has 5 bits per continuation byte */
299#define MAX_PORTABLE_UTF8_TWO_BYTE (32 * (1U << 5) - 1)
300
c03c0950
KW
301/* The maximum number of UTF-8 bytes a single Unicode character can
302 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
303 * expansion is UTF8_MAX_FOLD_CHAR_EXPAND characters, but any above-Unicode
304 * code point will fold to itself, so we only have to look at the expansion of
305 * the maximum Unicode code point. But this number may be less than the space
306 * occupied by a very large code point under Perl's extended UTF-8. We have to
307 * make it large enough to fit any single character. (It turns out that ASCII
308 * and EBCDIC differ in which is larger) */
309#define UTF8_MAXBYTES_CASE \
310 (UTF8_MAXBYTES >= (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)) \
311 ? UTF8_MAXBYTES \
312 : (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)))
313
d06134e5
KW
314/* Rest of these are attributes of Unicode and perl's internals rather than the
315 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
316 * this level; the macros that some of these call may have different
317 * definitions in the two encodings */
318
59a449d5
KW
319/* In domain restricted to ASCII, these may make more sense to the reader than
320 * the ones with Latin1 in the name */
321#define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
322#define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
323
324/* More or less misleadingly-named defines, retained for back compat */
325#define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
326#define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
327#define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
328#define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
329#define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
d06134e5 330
c0236afe
KW
331/* This defines the 1-bits that are to be in the first byte of a multi-byte
332 * UTF-8 encoded character that give the number of bytes that comprise the
333 * character. 'len' is the number of bytes in the multi-byte sequence. */
334#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
335
336/* Masks out the initial one bits in a start byte, leaving the real data ones.
337 * Doesn't work on an invariant byte. 'len' is the number of bytes in the
338 * multi-byte sequence that comprises the character. */
339#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
340
537124e4
KW
341/* Adds a UTF8 continuation byte 'new' of information to a running total code
342 * point 'old' of all the continuation bytes so far. This is designed to be
155d2738
KW
343 * used in a loop to convert from UTF-8 to the code point represented. Note
344 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is
345 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC)
346 * code point in process of being generated */
347#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
348 | ((NATIVE_UTF8_TO_I8((U8)new)) \
349 & UTF_CONTINUATION_MASK))
d06134e5 350
7c560c3b
KW
351/* If a value is anded with this, and the result is non-zero, then using the
352 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
353 * */
354#define UTF_ACCUMULATION_OVERFLOW_MASK \
355 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
356 - UTF_ACCUMULATION_SHIFT))
357
4ab10950
KW
358/* This works in the face of malformed UTF-8. */
359#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
360 && ( (e) - (s) > 1) \
361 && UTF8_IS_CONTINUATION(*((s)+1)))
362
5aaebcb3 363/* Number of bytes a code point occupies in UTF-8. */
5352a763 364#define NATIVE_SKIP(uv) UVCHR_SKIP(uv)
bd18bd40 365
5aaebcb3
KW
366/* Most code which says UNISKIP is really thinking in terms of native code
367 * points (0-255) plus all those beyond. This is an imprecise term, but having
2accb712 368 * it means existing code continues to work. For precision, use UVCHR_SKIP,
5352a763
KW
369 * NATIVE_SKIP, or OFFUNISKIP */
370#define UNISKIP(uv) UVCHR_SKIP(uv)
5aaebcb3 371
3c0792e4
KW
372/* Longer, but more accurate name */
373#define UTF8_IS_ABOVE_LATIN1_START(c) UTF8_IS_ABOVE_LATIN1(c)
374
a62b247b
KW
375/* Convert a UTF-8 variant Latin1 character to a native code point value.
376 * Needs just one iteration of accumulate. Should be used only if it is known
377 * that the code point is < 256, and is not UTF-8 invariant. Use the slower
378 * but more general TWO_BYTE_UTF8_TO_NATIVE() which handles any code point
379 * representable by two bytes (which turns out to be up through
380 * MAX_PORTABLE_UTF8_TWO_BYTE). The two parameters are:
381 * HI: a downgradable start byte;
382 * LO: continuation.
383 * */
384#define EIGHT_BIT_UTF8_TO_NATIVE(HI, LO) \
385 ( __ASSERT_(UTF8_IS_DOWNGRADEABLE_START(HI)) \
386 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
387 LATIN1_TO_NATIVE(UTF8_ACCUMULATE(( \
388 NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), (LO))))
389
94bb8c36 390/* Convert a two (not one) byte utf8 character to a native code point value.
2950f2a7
KW
391 * Needs just one iteration of accumulate. Should not be used unless it is
392 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
393 * Note that the result can be larger than 255 if the input character is not
394 * downgradable */
94bb8c36 395#define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
635e76f5
KW
396 ( __ASSERT_(UTF8SKIP(HI) == 2) \
397 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
94bb8c36 398 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
635e76f5 399 (LO))))
94bb8c36
KW
400
401/* Should never be used, and be deprecated */
402#define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
2950f2a7 403
bd18bd40
KW
404/*
405
406=for apidoc Am|STRLEN|UTF8SKIP|char* s
407returns the number of bytes in the UTF-8 encoded character whose first (perhaps
408only) byte is pointed to by C<s>.
409
410=cut
411 */
2a70536e
KW
412#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
413#define UTF8_SKIP(s) UTF8SKIP(s)
d06134e5 414
2d1545e5
KW
415/* Most code that says 'UNI_' really means the native value for code points up
416 * through 255 */
417#define UNI_IS_INVARIANT(cp) UVCHR_IS_INVARIANT(cp)
418
15824458
KW
419/* Is the byte 'c' the same character when encoded in UTF-8 as when not. This
420 * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in
5fc230f1
KW
421 * each for the exact same set of bit patterns. It is valid on a subset of
422 * what UVCHR_IS_INVARIANT is valid on, so can just use that; and the compiler
423 * should optimize out anything extraneous given the implementation of the
424 * latter */
425#define UTF8_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c)
426
427/* Like the above, but its name implies a non-UTF8 input, which as the comments
428 * above show, doesn't matter as to its implementation */
38953e5a 429#define NATIVE_BYTE_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c)
d06134e5 430
48ccf5e1
KW
431/* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic
432 * bytes from an ordinal that is known to fit into exactly two (not one) bytes;
433 * it must be less than 0x3FF to work across both encodings. */
434
435/* These two are helper macros for the other three sets, and should not be used
436 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1
1ff3baa2
KW
437 * (which works for code points up through 0xFF) or NATIVE_TO_UNI which works
438 * for any code point */
48ccf5e1
KW
439#define __BASE_TWO_BYTE_HI(c, translate_function) \
440 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \
441 | UTF_START_MARK(2))
442#define __BASE_TWO_BYTE_LO(c, translate_function) \
443 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \
444 | UTF_CONTINUATION_MARK)
445
48ccf5e1
KW
446/* The next two macros should not be used. They were designed to be usable as
447 * the case label of a switch statement, but this doesn't work for EBCDIC. Use
9d0d3a03 448 * regen/unicode_constants.pl instead */
48ccf5e1
KW
449#define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)
450#define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)
451
452/* The next two macros are used when the source should be a single byte
453 * character; checked for under DEBUGGING */
454#define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
4c8cd605 455 ( __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
48ccf5e1 456#define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
4c8cd605 457 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
48ccf5e1
KW
458
459/* These final two macros in the series are used when the source can be any
460 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient
461 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~'
462 * operator instead of "<=" to avoid getting compiler warnings.
d52b8576 463 * MAX_UTF8_TWO_BYTE should be exactly all one bits in the lower few
48ccf5e1
KW
464 * places, so the ~ works */
465#define UTF8_TWO_BYTE_HI(c) \
466 (__ASSERT_((sizeof(c) == 1) \
d52b8576 467 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
4c8cd605 468 (__BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
48ccf5e1
KW
469#define UTF8_TWO_BYTE_LO(c) \
470 (__ASSERT_((sizeof(c) == 1) \
d52b8576 471 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
4c8cd605 472 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
d06134e5 473
e7214ce8
KW
474/* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
475 * as it is only in overlongs. */
476#define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1)
477
7e2040f0 478/*
e3036cf4 479 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
480 * from an earlier version of these macros in which they didn't call the
481 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
482 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
483 * no need to do it again here
7e2040f0 484 */
e3036cf4
KW
485#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
486 ? isIDFIRST(*(p)) \
487 : isIDFIRST_utf8((const U8*)p))
32636478
KW
488#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
489 ? isWORDCHAR(*(p)) \
490 : isWORDCHAR_utf8((const U8*)p))
491#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
1d72bdf6 492
89ebb4a3
JH
493#define UTF8_MAXLEN UTF8_MAXBYTES
494
8cb75cc8
KW
495/* A Unicode character can fold to up to 3 characters */
496#define UTF8_MAX_FOLD_CHAR_EXPAND 3
497
a98fe34d 498#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
bd18bd40
KW
499
500/*
501
502=for apidoc Am|bool|DO_UTF8|SV* sv
503Returns a bool giving whether or not the PV in C<sv> is to be treated as being
504encoded in UTF-8.
505
506You should use this I<after> a call to C<SvPV()> or one of its variants, in
507case any call to string overloading updates the internal UTF-8 encoding flag.
508
509=cut
510*/
0064a8a9 511#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1ff3baa2
KW
512
513/* Should all strings be treated as Unicode, and not just UTF-8 encoded ones?
514 * Is so within 'feature unicode_strings' or 'locale :not_characters', and not
515 * within 'use bytes'. UTF-8 locales are not tested for here, but perhaps
516 * could be */
517#define IN_UNI_8_BIT \
d6ded950
KW
518 (((CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT)) \
519 || (CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \
520 /* -1 below is for :not_characters */ \
521 && _is_in_locale_category(FALSE, -1))) \
522 && ! IN_BYTES)
b36bf33f 523
1d72bdf6 524
c76687c5
KW
525#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
526
527/* Allow first byte to be a continuation byte */
1d72bdf6 528#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
529
530/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 531#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
532
533/* expecting more bytes than were available in the string */
534#define UTF8_ALLOW_SHORT 0x0008
535
536/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
537#define UTF8_ALLOW_LONG 0x0010
538
539#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
540#define UTF8_WARN_SURROGATE 0x0040
541
542#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
543#define UTF8_WARN_NONCHAR 0x0100 /* code points */
544
545#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
546#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
547
d35f2ca5
KW
548/* Code points which never were part of the original UTF-8 standard, which only
549 * went up to 2 ** 31 - 1. Note that these all overflow a signed 32-bit word,
550 * The first byte of these code points is FE or FF on ASCII platforms. If the
551 * first byte is FF, it will overflow a 32-bit word. */
552#define UTF8_DISALLOW_ABOVE_31_BIT 0x0800
553#define UTF8_WARN_ABOVE_31_BIT 0x1000
554
555/* For back compat, these old names are misleading for UTF_EBCDIC */
556#define UTF8_DISALLOW_FE_FF UTF8_DISALLOW_ABOVE_31_BIT
557#define UTF8_WARN_FE_FF UTF8_WARN_ABOVE_31_BIT
949cf498
KW
558
559#define UTF8_CHECK_ONLY 0x2000
560
561/* For backwards source compatibility. They do nothing, as the default now
562 * includes what they used to mean. The first one's meaning was to allow the
563 * just the single non-character 0xFFFF */
564#define UTF8_ALLOW_FFFF 0
565#define UTF8_ALLOW_SURROGATE 0
566
d35f2ca5 567#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE \
93e6dbd6
KW
568 ( UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR \
569 |UTF8_DISALLOW_SURROGATE)
949cf498 570#define UTF8_WARN_ILLEGAL_INTERCHANGE \
93e6dbd6
KW
571 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE)
572#define UTF8_ALLOW_ANY \
573 (~( UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_DISALLOW_ABOVE_31_BIT \
574 |UTF8_WARN_ILLEGAL_INTERCHANGE|UTF8_WARN_ABOVE_31_BIT))
949cf498
KW
575#define UTF8_ALLOW_ANYUV \
576 (UTF8_ALLOW_EMPTY \
577 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
578#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
579 UTF8_ALLOW_ANYUV)
1d72bdf6 580
0c58a72b
KW
581/* Several of the macros below have a second parameter that is currently
582 * unused; but could be used in the future to make sure that the input is
583 * well-formed. */
584
585#define UTF8_IS_SURROGATE(s, e) cBOOL(is_SURROGATE_utf8(s))
b96a92fb 586#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
7131f24d
KW
587
588/* ASCII EBCDIC I8
589 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
590 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
591 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
a1776718
KW
592 *
593 * BE AWARE that this test doesn't rule out malformed code points, in
594 * particular overlongs */
7131f24d 595#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
0c58a72b 596# define UTF8_IS_SUPER(s, e) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \
bc3632a8 597 && (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \
a3481822 598 || (NATIVE_UTF8_TO_I8(* ((U8*) (s) + 1)) >= 0xA2)))
7131f24d 599#else
0c58a72b
KW
600# define UTF8_IS_SUPER(s, e) (*(U8*) (s) >= 0xF4 \
601 && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
7131f24d
KW
602#endif
603
b96a92fb
KW
604/* These are now machine generated, and the 'given' clause is no longer
605 * applicable */
0c58a72b 606#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) \
b96a92fb 607 cBOOL(is_NONCHAR_utf8(s))
0c58a72b
KW
608#define UTF8_IS_NONCHAR(s, e) \
609 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e)
7131f24d 610
c867b360
JH
611#define UNICODE_SURROGATE_FIRST 0xD800
612#define UNICODE_SURROGATE_LAST 0xDFFF
613#define UNICODE_REPLACEMENT 0xFFFD
614#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 615
b851fbc1 616/* Though our UTF-8 encoding can go beyond this,
c76687c5 617 * let's be conservative and do as Unicode says. */
b851fbc1
JH
618#define PERL_UNICODE_MAX 0x10FFFF
619
d35f2ca5
KW
620#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
621#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
622#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
623#define UNICODE_WARN_ABOVE_31_BIT 0x0008 /* Above 0x7FFF_FFFF */
624#define UNICODE_DISALLOW_SURROGATE 0x0010
625#define UNICODE_DISALLOW_NONCHAR 0x0020
626#define UNICODE_DISALLOW_SUPER 0x0040
627#define UNICODE_DISALLOW_ABOVE_31_BIT 0x0080
bb88be5f
KW
628#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
629 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
630#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
631 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
949cf498
KW
632
633/* For backward source compatibility, as are now the default */
634#define UNICODE_ALLOW_SURROGATE 0
635#define UNICODE_ALLOW_SUPER 0
636#define UNICODE_ALLOW_ANY 0
b851fbc1 637
1d72bdf6
NIS
638#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
639 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 640#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 641#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
642#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
643 /* The other noncharacters end in FFFE or FFFF, which \
644 * the mask below catches both of, but beyond the last \
645 * official unicode code point, they aren't \
646 * noncharacters, since those aren't Unicode \
647 * characters at all */ \
648 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
649#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
d35f2ca5 650#define UNICODE_IS_ABOVE_31_BIT(uv) ((UV) (uv) > 0x7FFFFFFF)
1d72bdf6 651
ec34087a
KW
652#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
653#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
654 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
655#define MICRO_SIGN MICRO_SIGN_NATIVE
656#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
657 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
658#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
659 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
660#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
661#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
662#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 663#define GREEK_SMALL_LETTER_MU 0x03BC
9e682c18
KW
664#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case
665 of MICRON */
666#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
0766489e
KW
667#ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8
668# define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
669#endif
74894415
KW
670#define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130
671#define LATIN_SMALL_LETTER_DOTLESS_I 0x131
9e682c18 672#define LATIN_SMALL_LETTER_LONG_S 0x017F
a9f50d33
KW
673#define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05
674#define LATIN_SMALL_LIGATURE_ST 0xFB06
9e682c18
KW
675#define KELVIN_SIGN 0x212A
676#define ANGSTROM_SIGN 0x212B
09091399 677
9e55ce06 678#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
679#define UNI_DISPLAY_BACKSLASH 0x0002
680#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
681#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 682
5cd46e1f
KW
683#define ANYOF_FOLD_SHARP_S(node, input, end) \
684 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 685 (ANYOF_NONBITMAP(node)) && \
39065660 686 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f 687 ((end) > (input) + 1) && \
305b8651 688 isALPHA_FOLD_EQ((input)[0], 's'))
6302f837 689
ebc501f0 690#define SHARP_S_SKIP 2
3b0fc154 691
a4f7a67c
KW
692/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
693 * log for earlier versions which gave details for these */
4d646140 694
6302f837
KW
695/* A helper macro for isUTF8_CHAR, so use that one, and not this one. This is
696 * retained solely for backwards compatibility and may be deprecated and
697 * removed in a future Perl version.
698 *
699 * regen/regcharclass.pl generates is_UTF8_CHAR_utf8() macros for up to these
d9f92374
KW
700 * number of bytes. So this has to be coordinated with that file */
701#ifdef EBCDIC
702# define IS_UTF8_CHAR_FAST(n) ((n) <= 3)
703#else
704# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
705#endif
706
4d646140 707#ifndef EBCDIC
6302f837
KW
708/* A helper macro for isUTF8_CHAR, so use that one instead of this. This was
709 * generated by regen/regcharclass.pl, and then moved here. The lines that
710 * generated it were then commented out. This was done solely because it takes
711 * on the order of 10 minutes to generate, and is never going to change, unless
712 * the generated code is improved.
713 *
1ff3baa2
KW
714 * The EBCDIC versions have been cut to not cover all of legal Unicode,
715 * otherwise they take too long to generate; besides there is a separate one
716 * for each code page, so they are in regcharclass.h instead of here */
39a0f513 717/*
5dca9278 718 UTF8_CHAR: Matches legal UTF-8 encoded characters from 2 through 4 bytes
39a0f513 719
5dca9278 720 0x80 - 0x1FFFFF
39a0f513 721*/
4d646140 722/*** GENERATED CODE ***/
5dca9278
KW
723#define is_UTF8_CHAR_utf8_no_length_checks(s) \
724( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
725 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
726: ( 0xE0 == ((U8*)s)[0] ) ? \
727 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
728: ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \
729 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
730: ( 0xF0 == ((U8*)s)[0] ) ? \
731 ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
732: ( ( ( ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF7 ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )
4d646140 733#endif
3b0fc154 734
6302f837 735/*
5dca9278
KW
736
737=for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e
738
739Returns the number of bytes beginning at C<s> which form a legal UTF-8 (or
61b16eb9
KW
740UTF-EBCDIC) encoded character, looking no further than S<C<e - s>> bytes into
741C<s>. Returns 0 if the sequence starting at C<s> through S<C<e - 1>> is not
742well-formed UTF-8.
6302f837
KW
743
744Note that an INVARIANT character (i.e. ASCII on non-EBCDIC
5dca9278
KW
745machines) is a valid UTF-8 character.
746
747=cut
748*/
6302f837 749
dd9bc2b0 750#define isUTF8_CHAR(s, e) (UNLIKELY((e) <= (s)) \
6302f837
KW
751 ? 0 \
752 : (UTF8_IS_INVARIANT(*s)) \
753 ? 1 \
dd9bc2b0 754 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
6302f837 755 ? 0 \
dd9bc2b0 756 : LIKELY(IS_UTF8_CHAR_FAST(UTF8SKIP(s))) \
5dca9278 757 ? is_UTF8_CHAR_utf8_no_length_checks(s) \
6302f837
KW
758 : _is_utf8_char_slow(s, e))
759
3cedd9d9
KW
760#define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end)
761
6302f837
KW
762/* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is
763 * retained solely for backwards compatibility */
764#define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n)
e9a8c099 765
57f0e7e2
KW
766#endif /* H_UTF8 */
767
e9a8c099 768/*
14d04a33 769 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 770 */