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.
9 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009,
10 * 2010, 2011 by Larry Wall and others
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.
17 #ifndef PERL_UTF8_H_ /* Guard against recursive inclusion */
18 #define PERL_UTF8_H_ 1
20 /* Use UTF-8 as the default script encoding?
21 * Turning this on will break scripts having non-UTF-8 binary
22 * data (such as Latin-1) in string literals. */
23 #ifdef USE_UTF8_SCRIPTS
24 # define USE_UTF8_IN_NAMES (!IN_BYTES)
26 # define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
29 #include "regcharclass.h"
30 #include "unicode_constants.h"
32 /* For to_utf8_fold_flags, q.v. */
33 #define FOLD_FLAGS_LOCALE 0x1
34 #define FOLD_FLAGS_FULL 0x2
35 #define FOLD_FLAGS_NOMIX_ASCII 0x4
37 /* For _core_swash_init(), internal core use only */
38 #define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
39 #define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2
40 #define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4
43 =head1 Unicode Support
44 L<perlguts/Unicode Support> has an introduction to this API.
46 See also L</Character classification>,
47 and L</Character case changing>.
48 Various functions outside this section also work specially with Unicode.
49 Search for the string "utf8" in this document.
51 =for apidoc is_ascii_string
53 This is a misleadingly-named synonym for L</is_utf8_invariant_string>.
54 On ASCII-ish platforms, the name isn't misleading: the ASCII-range characters
55 are exactly the UTF-8 invariants. But EBCDIC machines have more invariants
56 than just the ASCII characters, so C<is_utf8_invariant_string> is preferred.
58 =for apidoc is_invariant_string
60 This is a somewhat misleadingly-named synonym for L</is_utf8_invariant_string>.
61 C<is_utf8_invariant_string> is preferred, as it indicates under what conditions
62 the string is invariant.
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)
69 #define uvoffuni_to_utf8_flags(d,uv,flags) \
70 uvoffuni_to_utf8_flags_msgs(d, uv, flags, 0)
71 #define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0)
72 #define uvchr_to_utf8_flags(d,uv,flags) \
73 uvchr_to_utf8_flags_msgs(d,uv,flags, 0)
74 #define uvchr_to_utf8_flags_msgs(d,uv,flags,msgs) \
75 uvoffuni_to_utf8_flags_msgs(d,NATIVE_TO_UNI(uv),flags, msgs)
76 #define utf8_to_uvchr_buf(s, e, lenp) \
77 (__ASSERT_((U8*) (e) > (U8*) (s)) \
78 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \
79 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY))
80 #define utf8n_to_uvchr(s, len, lenp, flags) \
81 utf8n_to_uvchr_error(s, len, lenp, flags, 0)
82 #define utf8n_to_uvchr_error(s, len, lenp, flags, errors) \
83 utf8n_to_uvchr_msgs(s, len, lenp, flags, errors, 0)
85 #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
87 #define to_utf8_fold(s, r, lenr) \
88 _to_utf8_fold_flags (s, NULL, r, lenr, FOLD_FLAGS_FULL, __FILE__, __LINE__)
89 #define to_utf8_lower(s, r, lenr) \
90 _to_utf8_lower_flags(s, NULL, r ,lenr, 0, __FILE__, __LINE__)
91 #define to_utf8_upper(s, r, lenr) \
92 _to_utf8_upper_flags(s, NULL, r, lenr, 0, __FILE__, __LINE__)
93 #define to_utf8_title(s, r, lenr) \
94 _to_utf8_title_flags(s, NULL, r, lenr ,0, __FILE__, __LINE__)
96 #define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
97 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
98 #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
99 #define FOLDEQ_LOCALE (1 << 1)
100 #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
101 #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
102 #define FOLDEQ_S1_FOLDS_SANE (1 << 4)
103 #define FOLDEQ_S2_FOLDS_SANE (1 << 5)
105 #define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
106 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
109 /* The equivalent of these macros but implementing UTF-EBCDIC
110 are in the following header file:
113 #include "utfebcdic.h"
118 /* How wide can a single UTF-8 encoded character become in bytes. */
119 /* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
120 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
121 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
122 * non-negative integers in a binary format, even those above Unicode */
123 #define UTF8_MAXBYTES 13
126 EXTCONST unsigned char PL_utf8skip[] = {
127 /* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
128 /* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
129 /* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
130 /* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
131 /* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
132 /* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
133 /* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
134 /* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
135 /* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
136 /* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
137 /* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
138 /* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
139 /* 0xC0 */ 2,2, /* overlong */
140 /* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
141 /* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
142 /* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
143 /* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
144 /* Perl extended (never was official UTF-8). Up to 36 bit */
146 /* More extended, Up to 72 bits (64-bit + reserved) */
147 /* 0xFF */ UTF8_MAXBYTES
150 EXTCONST unsigned char PL_utf8skip[];
155 #if defined(_MSC_VER) && _MSC_VER < 1400
156 /* older MSVC versions have a smallish macro buffer */
157 #define PERL_SMALL_MACRO_BUFFER
160 /* Native character to/from iso-8859-1. Are the identity functions on ASCII
162 #ifdef PERL_SMALL_MACRO_BUFFER
163 #define NATIVE_TO_LATIN1(ch) ((U8)(ch))
164 #define LATIN1_TO_NATIVE(ch) ((U8)(ch))
166 #define NATIVE_TO_LATIN1(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
167 #define LATIN1_TO_NATIVE(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
170 /* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
171 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
172 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
173 * because they are 8-bit encodings that serve the same purpose in Perl, and
174 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
175 * whichever one is applicable on the current platform */
176 #ifdef PERL_SMALL_MACRO_BUFFER
177 #define NATIVE_UTF8_TO_I8(ch) ((U8) (ch))
178 #define I8_TO_NATIVE_UTF8(ch) ((U8) (ch))
180 #define NATIVE_UTF8_TO_I8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
181 #define I8_TO_NATIVE_UTF8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
184 /* Transforms in wide UV chars */
185 #define UNI_TO_NATIVE(ch) ((UV) (ch))
186 #define NATIVE_TO_UNI(ch) ((UV) (ch))
190 The following table is from Unicode 3.2, plus the Perl extensions for above
193 Code Points 1st Byte 2nd Byte 3rd 4th 5th 6th 7th 8th-13th
195 U+0000..U+007F 00..7F
196 U+0080..U+07FF * C2..DF 80..BF
197 U+0800..U+0FFF E0 * A0..BF 80..BF
198 U+1000..U+CFFF E1..EC 80..BF 80..BF
199 U+D000..U+D7FF ED 80..9F 80..BF
200 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
201 U+E000..U+FFFF EE..EF 80..BF 80..BF
202 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
203 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
204 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
205 Below are above-Unicode code points
206 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
207 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
208 U+200000..U+FFFFFF F8 * 88..BF 80..BF 80..BF 80..BF
209 U+1000000..U+3FFFFFF F9..FB 80..BF 80..BF 80..BF 80..BF
210 U+4000000..U+3FFFFFFF FC * 84..BF 80..BF 80..BF 80..BF 80..BF
211 U+40000000..U+7FFFFFFF FD 80..BF 80..BF 80..BF 80..BF 80..BF
212 U+80000000..U+FFFFFFFFF FE * 82..BF 80..BF 80..BF 80..BF 80..BF 80..BF
213 U+1000000000.. FF 80..BF 80..BF 80..BF 80..BF 80..BF * 81..BF 80..BF
215 Note the gaps before several of the byte entries above marked by '*'. These are
216 caused by legal UTF-8 avoiding non-shortest encodings: it is technically
217 possible to UTF-8-encode a single code point in different ways, but that is
218 explicitly forbidden, and the shortest possible encoding should always be used
219 (and that is what Perl does). The non-shortest ones are called 'overlongs'.
224 Another way to look at it, as bits:
226 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
229 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
230 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
231 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
233 As you can see, the continuation bytes all begin with C<10>, and the
234 leading bits of the start byte tell how many bytes there are in the
237 Perl's extended UTF-8 means we can have start bytes up through FF, though any
238 beginning with FF yields a code point that is too large for 32-bit ASCII
239 platforms. FF signals to use 13 bytes for the encoded character. This breaks
240 the paradigm that the number of leading bits gives how many total bytes there
241 are in the character.
245 /* Is the representation of the Unicode code point 'cp' the same regardless of
246 * being encoded in UTF-8 or not? */
247 #define OFFUNI_IS_INVARIANT(cp) isASCII(cp)
250 =for apidoc Am|bool|UVCHR_IS_INVARIANT|UV cp
252 Evaluates to 1 if the representation of code point C<cp> is the same whether or
253 not it is encoded in UTF-8; otherwise evaluates to 0. UTF-8 invariant
254 characters can be copied as-is when converting to/from UTF-8, saving time.
255 C<cp> is Unicode if above 255; otherwise is platform-native.
260 #define UVCHR_IS_INVARIANT(cp) OFFUNI_IS_INVARIANT(cp)
262 /* This defines the bits that are to be in the continuation bytes of a multi-byte
263 * UTF-8 encoded character that mark it is a continuation byte. */
264 #define UTF_CONTINUATION_MARK 0x80
266 /* Misleadingly named: is the UTF8-encoded byte 'c' part of a variant sequence
267 * in UTF-8? This is the inverse of UTF8_IS_INVARIANT. The |0 makes sure this
268 * isn't mistakenly called with a ptr argument */
269 #define UTF8_IS_CONTINUED(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
270 ((U8)((c) | 0)) & UTF_CONTINUATION_MARK)
272 /* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
273 * This doesn't catch invariants (they are single-byte). It also excludes the
274 * illegal overlong sequences that begin with C0 and C1. The |0 makes sure
275 * this isn't mistakenly called with a ptr argument */
276 #define UTF8_IS_START(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
277 ((U8)((c) | 0)) >= 0xc2)
279 /* For use in UTF8_IS_CONTINUATION() below */
280 #define UTF_IS_CONTINUATION_MASK 0xC0
282 /* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
283 * first byte thereof? The |0 makes sure this isn't mistakenly called with a
285 #define UTF8_IS_CONTINUATION(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
286 (((U8)((c) | 0)) & UTF_IS_CONTINUATION_MASK) == UTF_CONTINUATION_MARK)
288 /* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
289 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
290 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
291 * this matches 0xc[23]. The |0 makes sure this isn't mistakenly called with a
293 #define UTF8_IS_DOWNGRADEABLE_START(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
294 (((U8)((c) | 0)) & 0xfe) == 0xc2)
296 /* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
297 * represent a code point > 255? The |0 makes sure this isn't mistakenly
298 * called with a ptr argument */
299 #define UTF8_IS_ABOVE_LATIN1(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
300 ((U8)((c) | 0)) >= 0xc4)
302 /* This is the number of low-order bits a continuation byte in a UTF-8 encoded
303 * sequence contributes to the specification of the code point. In the bit
304 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
305 * real information */
306 #define UTF_ACCUMULATION_SHIFT 6
308 /* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
310 #define QUESTION_MARK_CTRL DEL_NATIVE
312 /* Surrogates, non-character code points and above-Unicode code points are
313 * problematic in some contexts. This allows code that needs to check for
314 * those to to quickly exclude the vast majority of code points it will
316 #define isUTF8_POSSIBLY_PROBLEMATIC(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
319 #define UNICODE_IS_PERL_EXTENDED(uv) UNLIKELY((UV) (uv) > 0x7FFFFFFF)
321 #endif /* EBCDIC vs ASCII */
323 /* 2**UTF_ACCUMULATION_SHIFT - 1 */
324 #define UTF_CONTINUATION_MASK ((U8) ((1U << UTF_ACCUMULATION_SHIFT) - 1))
326 /* Internal macro to be used only in this file to aid in constructing other
327 * publicly accessible macros.
328 * The number of bytes required to express this uv in UTF-8, for just those
329 * uv's requiring 2 through 6 bytes, as these are common to all platforms and
330 * word sizes. The number of bytes needed is given by the number of leading 1
331 * bits in the start byte. There are 32 start bytes that have 2 initial 1 bits
332 * (C0-DF); there are 16 that have 3 initial 1 bits (E0-EF); 8 that have 4
333 * initial 1 bits (F0-F8); 4 that have 5 initial 1 bits (F9-FB), and 2 that
334 * have 6 initial 1 bits (FC-FD). The largest number a string of n bytes can
335 * represent is (the number of possible start bytes for 'n')
336 * * (the number of possiblities for each start byte
337 * The latter in turn is
338 * 2 ** ( (how many continuation bytes there are)
339 * * (the number of bits of information each
340 * continuation byte holds))
342 * If we were on a platform where we could use a fast find first set bit
343 * instruction (or count leading zeros instruction) this could be replaced by
344 * using that to find the log2 of the uv, and divide that by the number of bits
345 * of information in each continuation byte, adjusting for large cases and how
346 * much information is in a start byte for that length */
347 #define __COMMON_UNI_SKIP(uv) \
348 (UV) (uv) < (32 * (1U << ( UTF_ACCUMULATION_SHIFT))) ? 2 : \
349 (UV) (uv) < (16 * (1U << (2 * UTF_ACCUMULATION_SHIFT))) ? 3 : \
350 (UV) (uv) < ( 8 * (1U << (3 * UTF_ACCUMULATION_SHIFT))) ? 4 : \
351 (UV) (uv) < ( 4 * (1U << (4 * UTF_ACCUMULATION_SHIFT))) ? 5 : \
352 (UV) (uv) < ( 2 * (1U << (5 * UTF_ACCUMULATION_SHIFT))) ? 6 :
354 /* Internal macro to be used only in this file.
355 * This adds to __COMMON_UNI_SKIP the details at this platform's upper range.
356 * For any-sized EBCDIC platforms, or 64-bit ASCII ones, we need one more test
357 * to see if just 7 bytes is needed, or if the maximum is needed. For 32-bit
358 * ASCII platforms, everything is representable by 7 bytes */
359 #if defined(UV_IS_QUAD) || defined(EBCDIC)
360 # define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) \
361 (UV) (uv) < ((UV) 1U << (6 * UTF_ACCUMULATION_SHIFT)) ? 7 : UTF8_MAXBYTES)
363 # define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) 7)
366 /* The next two macros use the base macro defined above, and add in the tests
367 * at the low-end of the range, for just 1 byte, yielding complete macros,
368 * publicly accessible. */
370 /* Input is a true Unicode (not-native) code point */
371 #define OFFUNISKIP(uv) (OFFUNI_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
375 =for apidoc Am|STRLEN|UVCHR_SKIP|UV cp
376 returns the number of bytes required to represent the code point C<cp> when
377 encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
378 255; a Unicode code point otherwise.
382 #define UVCHR_SKIP(uv) ( UVCHR_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
384 /* The largest code point representable by two UTF-8 bytes on this platform.
385 * As explained in the comments for __COMMON_UNI_SKIP, 32 start bytes with
386 * UTF_ACCUMULATION_SHIFT bits of information each */
387 #define MAX_UTF8_TWO_BYTE (32 * (1U << UTF_ACCUMULATION_SHIFT) - 1)
389 /* The largest code point representable by two UTF-8 bytes on any platform that
390 * Perl runs on. This value is constrained by EBCDIC which has 5 bits per
391 * continuation byte */
392 #define MAX_PORTABLE_UTF8_TWO_BYTE (32 * (1U << 5) - 1)
394 /* The maximum number of UTF-8 bytes a single Unicode character can
395 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
396 * expansion is UTF8_MAX_FOLD_CHAR_EXPAND characters, but any above-Unicode
397 * code point will fold to itself, so we only have to look at the expansion of
398 * the maximum Unicode code point. But this number may be less than the space
399 * occupied by a very large code point under Perl's extended UTF-8. We have to
400 * make it large enough to fit any single character. (It turns out that ASCII
401 * and EBCDIC differ in which is larger) */
402 #define UTF8_MAXBYTES_CASE \
403 (UTF8_MAXBYTES >= (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)) \
405 : (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)))
407 /* Rest of these are attributes of Unicode and perl's internals rather than the
408 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
409 * this level; the macros that some of these call may have different
410 * definitions in the two encodings */
412 /* In domain restricted to ASCII, these may make more sense to the reader than
413 * the ones with Latin1 in the name */
414 #define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
415 #define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
417 /* More or less misleadingly-named defines, retained for back compat */
418 #define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
419 #define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
420 #define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
421 #define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
422 #define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
424 /* This defines the 1-bits that are to be in the first byte of a multi-byte
425 * UTF-8 encoded character that mark it as a start byte and give the number of
426 * bytes that comprise the character. 'len' is the number of bytes in the
427 * multi-byte sequence. */
428 #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
430 /* Masks out the initial one bits in a start byte, leaving the real data ones.
431 * Doesn't work on an invariant byte. 'len' is the number of bytes in the
432 * multi-byte sequence that comprises the character. */
433 #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
435 /* Adds a UTF8 continuation byte 'new' of information to a running total code
436 * point 'old' of all the continuation bytes so far. This is designed to be
437 * used in a loop to convert from UTF-8 to the code point represented. Note
438 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is
439 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC)
440 * code point in process of being generated */
441 #define UTF8_ACCUMULATE(old, new) (__ASSERT_(FITS_IN_8_BITS(new)) \
442 ((old) << UTF_ACCUMULATION_SHIFT) \
443 | ((NATIVE_UTF8_TO_I8((U8)new)) \
444 & UTF_CONTINUATION_MASK))
446 /* This works in the face of malformed UTF-8. */
447 #define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) \
448 ( UTF8_IS_DOWNGRADEABLE_START(*(s)) \
449 && ( (e) - (s) > 1) \
450 && UTF8_IS_CONTINUATION(*((s)+1)))
452 /* Number of bytes a code point occupies in UTF-8. */
453 #define NATIVE_SKIP(uv) UVCHR_SKIP(uv)
455 /* Most code which says UNISKIP is really thinking in terms of native code
456 * points (0-255) plus all those beyond. This is an imprecise term, but having
457 * it means existing code continues to work. For precision, use UVCHR_SKIP,
458 * NATIVE_SKIP, or OFFUNISKIP */
459 #define UNISKIP(uv) UVCHR_SKIP(uv)
461 /* Longer, but more accurate name */
462 #define UTF8_IS_ABOVE_LATIN1_START(c) UTF8_IS_ABOVE_LATIN1(c)
464 /* Convert a UTF-8 variant Latin1 character to a native code point value.
465 * Needs just one iteration of accumulate. Should be used only if it is known
466 * that the code point is < 256, and is not UTF-8 invariant. Use the slower
467 * but more general TWO_BYTE_UTF8_TO_NATIVE() which handles any code point
468 * representable by two bytes (which turns out to be up through
469 * MAX_PORTABLE_UTF8_TWO_BYTE). The two parameters are:
470 * HI: a downgradable start byte;
473 #define EIGHT_BIT_UTF8_TO_NATIVE(HI, LO) \
474 ( __ASSERT_(UTF8_IS_DOWNGRADEABLE_START(HI)) \
475 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
476 LATIN1_TO_NATIVE(UTF8_ACCUMULATE(( \
477 NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), (LO))))
479 /* Convert a two (not one) byte utf8 character to a native code point value.
480 * Needs just one iteration of accumulate. Should not be used unless it is
481 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
482 * Note that the result can be larger than 255 if the input character is not
484 #define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
485 (__ASSERT_(FITS_IN_8_BITS(HI)) \
486 __ASSERT_(FITS_IN_8_BITS(LO)) \
487 __ASSERT_(PL_utf8skip[HI] == 2) \
488 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
489 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
492 /* Should never be used, and be deprecated */
493 #define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
497 =for apidoc Am|STRLEN|UTF8SKIP|char* s
498 returns the number of bytes in the UTF-8 encoded character whose first (perhaps
499 only) byte is pointed to by C<s>.
503 #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
504 #define UTF8_SKIP(s) UTF8SKIP(s)
506 /* Most code that says 'UNI_' really means the native value for code points up
508 #define UNI_IS_INVARIANT(cp) UVCHR_IS_INVARIANT(cp)
511 =for apidoc Am|bool|UTF8_IS_INVARIANT|char c
513 Evaluates to 1 if the byte C<c> represents the same character when encoded in
514 UTF-8 as when not; otherwise evaluates to 0. UTF-8 invariant characters can be
515 copied as-is when converting to/from UTF-8, saving time.
517 In spite of the name, this macro gives the correct result if the input string
518 from which C<c> comes is not encoded in UTF-8.
520 See C<L</UVCHR_IS_INVARIANT>> for checking if a UV is invariant.
524 The reason it works on both UTF-8 encoded strings and non-UTF-8 encoded, is
525 that it returns TRUE in each for the exact same set of bit patterns. It is
526 valid on a subset of what UVCHR_IS_INVARIANT is valid on, so can just use that;
527 and the compiler should optimize out anything extraneous given the
528 implementation of the latter. The |0 makes sure this isn't mistakenly called
531 #define UTF8_IS_INVARIANT(c) UVCHR_IS_INVARIANT((c) | 0)
533 /* Like the above, but its name implies a non-UTF8 input, which as the comments
534 * above show, doesn't matter as to its implementation */
535 #define NATIVE_BYTE_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c)
537 /* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic
538 * bytes from an ordinal that is known to fit into exactly two (not one) bytes;
539 * it must be less than 0x3FF to work across both encodings. */
541 /* These two are helper macros for the other three sets, and should not be used
542 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1
543 * (which works for code points up through 0xFF) or NATIVE_TO_UNI which works
544 * for any code point */
545 #define __BASE_TWO_BYTE_HI(c, translate_function) \
546 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
547 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \
548 | UTF_START_MARK(2)))
549 #define __BASE_TWO_BYTE_LO(c, translate_function) \
550 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
551 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \
552 | UTF_CONTINUATION_MARK))
554 /* The next two macros should not be used. They were designed to be usable as
555 * the case label of a switch statement, but this doesn't work for EBCDIC. Use
556 * regen/unicode_constants.pl instead */
557 #define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)
558 #define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)
560 /* The next two macros are used when the source should be a single byte
561 * character; checked for under DEBUGGING */
562 #define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
563 ( __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
564 #define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
565 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
567 /* These final two macros in the series are used when the source can be any
568 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient
569 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~'
570 * operator instead of "<=" to avoid getting compiler warnings.
571 * MAX_UTF8_TWO_BYTE should be exactly all one bits in the lower few
572 * places, so the ~ works */
573 #define UTF8_TWO_BYTE_HI(c) \
574 (__ASSERT_((sizeof(c) == 1) \
575 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
576 (__BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
577 #define UTF8_TWO_BYTE_LO(c) \
578 (__ASSERT_((sizeof(c) == 1) \
579 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
580 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
582 /* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
583 * as it is only in overlongs. */
584 #define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1)
587 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
588 * from an earlier version of these macros in which they didn't call the
589 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
590 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
591 * no need to do it again here
593 #define isIDFIRST_lazy_if(p,UTF) \
594 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isIDFIRST_lazy_if", \
595 "isIDFIRST_lazy_if_safe", \
596 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__)
598 #define isIDFIRST_lazy_if_safe(p, e, UTF) \
599 ((IN_BYTES || !UTF) \
601 : isIDFIRST_utf8_safe(p, e))
603 #define isWORDCHAR_lazy_if(p,UTF) \
604 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isWORDCHAR_lazy_if", \
605 "isWORDCHAR_lazy_if_safe", \
606 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__)
608 #define isWORDCHAR_lazy_if_safe(p, e, UTF) \
609 ((IN_BYTES || !UTF) \
611 : isWORDCHAR_utf8_safe((U8 *) p, (U8 *) e))
613 #define isALNUM_lazy_if(p,UTF) \
614 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isALNUM_lazy_if", \
615 "isWORDCHAR_lazy_if_safe", \
616 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__)
618 #define UTF8_MAXLEN UTF8_MAXBYTES
620 /* A Unicode character can fold to up to 3 characters */
621 #define UTF8_MAX_FOLD_CHAR_EXPAND 3
623 #define IN_BYTES UNLIKELY(CopHINTS_get(PL_curcop) & HINT_BYTES)
627 =for apidoc Am|bool|DO_UTF8|SV* sv
628 Returns a bool giving whether or not the PV in C<sv> is to be treated as being
631 You should use this I<after> a call to C<SvPV()> or one of its variants, in
632 case any call to string overloading updates the internal UTF-8 encoding flag.
636 #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
638 /* Should all strings be treated as Unicode, and not just UTF-8 encoded ones?
639 * Is so within 'feature unicode_strings' or 'locale :not_characters', and not
640 * within 'use bytes'. UTF-8 locales are not tested for here, but perhaps
642 #define IN_UNI_8_BIT \
643 (( ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT)) \
644 || ( CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \
645 /* -1 below is for :not_characters */ \
646 && _is_in_locale_category(FALSE, -1))) \
650 #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
651 #define UTF8_GOT_EMPTY UTF8_ALLOW_EMPTY
653 /* Allow first byte to be a continuation byte */
654 #define UTF8_ALLOW_CONTINUATION 0x0002
655 #define UTF8_GOT_CONTINUATION UTF8_ALLOW_CONTINUATION
657 /* Unexpected non-continuation byte */
658 #define UTF8_ALLOW_NON_CONTINUATION 0x0004
659 #define UTF8_GOT_NON_CONTINUATION UTF8_ALLOW_NON_CONTINUATION
661 /* expecting more bytes than were available in the string */
662 #define UTF8_ALLOW_SHORT 0x0008
663 #define UTF8_GOT_SHORT UTF8_ALLOW_SHORT
665 /* Overlong sequence; i.e., the code point can be specified in fewer bytes.
666 * First one will convert the overlong to the REPLACEMENT CHARACTER; second
667 * will return what the overlong evaluates to */
668 #define UTF8_ALLOW_LONG 0x0010
669 #define UTF8_ALLOW_LONG_AND_ITS_VALUE (UTF8_ALLOW_LONG|0x0020)
670 #define UTF8_GOT_LONG UTF8_ALLOW_LONG
672 #define UTF8_ALLOW_OVERFLOW 0x0080
673 #define UTF8_GOT_OVERFLOW UTF8_ALLOW_OVERFLOW
675 #define UTF8_DISALLOW_SURROGATE 0x0100 /* Unicode surrogates */
676 #define UTF8_GOT_SURROGATE UTF8_DISALLOW_SURROGATE
677 #define UTF8_WARN_SURROGATE 0x0200
679 /* Unicode non-character code points */
680 #define UTF8_DISALLOW_NONCHAR 0x0400
681 #define UTF8_GOT_NONCHAR UTF8_DISALLOW_NONCHAR
682 #define UTF8_WARN_NONCHAR 0x0800
684 /* Super-set of Unicode: code points above the legal max */
685 #define UTF8_DISALLOW_SUPER 0x1000
686 #define UTF8_GOT_SUPER UTF8_DISALLOW_SUPER
687 #define UTF8_WARN_SUPER 0x2000
689 /* The original UTF-8 standard did not define UTF-8 with start bytes of 0xFE or
690 * 0xFF, though UTF-EBCDIC did. This allowed both versions to represent code
691 * points up to 2 ** 31 - 1. Perl extends UTF-8 so that 0xFE and 0xFF are
692 * usable on ASCII platforms, and 0xFF means something different than
693 * UTF-EBCDIC defines. These changes allow code points of 64 bits (actually
694 * somewhat more) to be represented on both platforms. But these are Perl
695 * extensions, and not likely to be interchangeable with other languages. Note
696 * that on ASCII platforms, FE overflows a signed 32-bit word, and FF an
698 #define UTF8_DISALLOW_PERL_EXTENDED 0x4000
699 #define UTF8_GOT_PERL_EXTENDED UTF8_DISALLOW_PERL_EXTENDED
700 #define UTF8_WARN_PERL_EXTENDED 0x8000
702 /* For back compat, these old names are misleading for overlongs and
704 #define UTF8_DISALLOW_ABOVE_31_BIT UTF8_DISALLOW_PERL_EXTENDED
705 #define UTF8_GOT_ABOVE_31_BIT UTF8_GOT_PERL_EXTENDED
706 #define UTF8_WARN_ABOVE_31_BIT UTF8_WARN_PERL_EXTENDED
707 #define UTF8_DISALLOW_FE_FF UTF8_DISALLOW_PERL_EXTENDED
708 #define UTF8_WARN_FE_FF UTF8_WARN_PERL_EXTENDED
710 #define UTF8_CHECK_ONLY 0x10000
711 #define _UTF8_NO_CONFIDENCE_IN_CURLEN 0x20000 /* Internal core use only */
713 /* For backwards source compatibility. They do nothing, as the default now
714 * includes what they used to mean. The first one's meaning was to allow the
715 * just the single non-character 0xFFFF */
716 #define UTF8_ALLOW_FFFF 0
717 #define UTF8_ALLOW_FE_FF 0
718 #define UTF8_ALLOW_SURROGATE 0
720 /* C9 refers to Unicode Corrigendum #9: allows but discourages non-chars */
721 #define UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE \
722 (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_SURROGATE)
723 #define UTF8_WARN_ILLEGAL_C9_INTERCHANGE (UTF8_WARN_SUPER|UTF8_WARN_SURROGATE)
725 #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE \
726 (UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE|UTF8_DISALLOW_NONCHAR)
727 #define UTF8_WARN_ILLEGAL_INTERCHANGE \
728 (UTF8_WARN_ILLEGAL_C9_INTERCHANGE|UTF8_WARN_NONCHAR)
730 /* This is typically used for code that processes UTF-8 input and doesn't want
731 * to have to deal with any malformations that might be present. All such will
732 * be safely replaced by the REPLACEMENT CHARACTER, unless other flags
733 * overriding this are also present. */
734 #define UTF8_ALLOW_ANY ( UTF8_ALLOW_CONTINUATION \
735 |UTF8_ALLOW_NON_CONTINUATION \
738 |UTF8_ALLOW_OVERFLOW)
740 /* Accept any Perl-extended UTF-8 that evaluates to any UV on the platform, but
741 * not any malformed. This is the default. */
742 #define UTF8_ALLOW_ANYUV 0
743 #define UTF8_ALLOW_DEFAULT UTF8_ALLOW_ANYUV
746 =for apidoc Am|bool|UTF8_IS_SURROGATE|const U8 *s|const U8 *e
748 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
749 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one
750 of the Unicode surrogate code points; otherwise it evaluates to 0. If
751 non-zero, the value gives how many bytes starting at C<s> comprise the code
752 point's representation.
756 #define UTF8_IS_SURROGATE(s, e) is_SURROGATE_utf8_safe(s, e)
759 #define UTF8_IS_REPLACEMENT(s, send) is_REPLACEMENT_utf8_safe(s,send)
761 #define MAX_LEGAL_CP IV_MAX
764 =for apidoc Am|bool|UTF8_IS_SUPER|const U8 *s|const U8 *e
766 Recall that Perl recognizes an extension to UTF-8 that can encode code
767 points larger than the ones defined by Unicode, which are 0..0x10FFFF.
769 This macro evaluates to non-zero if the first few bytes of the string starting
770 at C<s> and looking no further than S<C<e - 1>> are from this UTF-8 extension;
771 otherwise it evaluates to 0. If non-zero, the value gives how many bytes
772 starting at C<s> comprise the code point's representation.
774 0 is returned if the bytes are not well-formed extended UTF-8, or if they
775 represent a code point that cannot fit in a UV on the current platform. Hence
776 this macro can give different results when run on a 64-bit word machine than on
777 one with a 32-bit word size.
779 Note that it is illegal to have code points that are larger than what can
780 fit in an IV on the current machine.
785 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
786 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
787 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
790 # define UTF8_IS_SUPER(s, e) \
791 (( LIKELY((e) > (s) + 4) \
792 && NATIVE_UTF8_TO_I8(*(s)) >= 0xF9 \
793 && ( NATIVE_UTF8_TO_I8(*(s)) > 0xF9 \
794 || (NATIVE_UTF8_TO_I8(*((s) + 1)) >= 0xA2)) \
795 && LIKELY((s) + UTF8SKIP(s) <= (e))) \
796 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0)
798 # define UTF8_IS_SUPER(s, e) \
799 (( LIKELY((e) > (s) + 3) \
800 && (*(U8*) (s)) >= 0xF4 \
801 && ((*(U8*) (s)) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))\
802 && LIKELY((s) + UTF8SKIP(s) <= (e))) \
803 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0)
806 /* These are now machine generated, and the 'given' clause is no longer
808 #define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) \
809 cBOOL(is_NONCHAR_utf8_safe(s,e))
812 =for apidoc Am|bool|UTF8_IS_NONCHAR|const U8 *s|const U8 *e
814 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
815 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one
816 of the Unicode non-character code points; otherwise it evaluates to 0. If
817 non-zero, the value gives how many bytes starting at C<s> comprise the code
818 point's representation.
822 #define UTF8_IS_NONCHAR(s, e) \
823 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e)
825 #define UNICODE_SURROGATE_FIRST 0xD800
826 #define UNICODE_SURROGATE_LAST 0xDFFF
827 #define UNICODE_REPLACEMENT 0xFFFD
828 #define UNICODE_BYTE_ORDER_MARK 0xFEFF
830 /* Though our UTF-8 encoding can go beyond this,
831 * let's be conservative and do as Unicode says. */
832 #define PERL_UNICODE_MAX 0x10FFFF
834 #define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
835 #define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
836 #define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
837 #define UNICODE_WARN_PERL_EXTENDED 0x0008 /* Above 0x7FFF_FFFF */
838 #define UNICODE_WARN_ABOVE_31_BIT UNICODE_WARN_PERL_EXTENDED
839 #define UNICODE_DISALLOW_SURROGATE 0x0010
840 #define UNICODE_DISALLOW_NONCHAR 0x0020
841 #define UNICODE_DISALLOW_SUPER 0x0040
842 #define UNICODE_DISALLOW_PERL_EXTENDED 0x0080
843 #define UNICODE_DISALLOW_ABOVE_31_BIT UNICODE_DISALLOW_PERL_EXTENDED
845 #define UNICODE_GOT_SURROGATE UNICODE_DISALLOW_SURROGATE
846 #define UNICODE_GOT_NONCHAR UNICODE_DISALLOW_NONCHAR
847 #define UNICODE_GOT_SUPER UNICODE_DISALLOW_SUPER
848 #define UNICODE_GOT_PERL_EXTENDED UNICODE_DISALLOW_PERL_EXTENDED
850 #define UNICODE_WARN_ILLEGAL_C9_INTERCHANGE \
851 (UNICODE_WARN_SURROGATE|UNICODE_WARN_SUPER)
852 #define UNICODE_WARN_ILLEGAL_INTERCHANGE \
853 (UNICODE_WARN_ILLEGAL_C9_INTERCHANGE|UNICODE_WARN_NONCHAR)
854 #define UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE \
855 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_SUPER)
856 #define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
857 (UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE|UNICODE_DISALLOW_NONCHAR)
859 /* For backward source compatibility, as are now the default */
860 #define UNICODE_ALLOW_SURROGATE 0
861 #define UNICODE_ALLOW_SUPER 0
862 #define UNICODE_ALLOW_ANY 0
864 /* This matches the 2048 code points between UNICODE_SURROGATE_FIRST (0xD800) and
865 * UNICODE_SURROGATE_LAST (0xDFFF) */
866 #define UNICODE_IS_SURROGATE(uv) (((UV) (uv) & (~0xFFFF | 0xF800)) \
869 #define UNICODE_IS_REPLACEMENT(uv) ((UV) (uv) == UNICODE_REPLACEMENT)
870 #define UNICODE_IS_BYTE_ORDER_MARK(uv) ((UV) (uv) == UNICODE_BYTE_ORDER_MARK)
872 /* Is 'uv' one of the 32 contiguous-range noncharacters? */
873 #define UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) ((UV) (uv) >= 0xFDD0 \
874 && (UV) (uv) <= 0xFDEF)
876 /* Is 'uv' one of the 34 plane-ending noncharacters 0xFFFE, 0xFFFF, 0x1FFFE,
877 * 0x1FFFF, ... 0x10FFFE, 0x10FFFF, given that we know that 'uv' is not above
878 * the Unicode legal max */
879 #define UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv) \
880 (((UV) (uv) & 0xFFFE) == 0xFFFE)
882 #define UNICODE_IS_NONCHAR(uv) \
883 ( UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) \
884 || ( LIKELY( ! UNICODE_IS_SUPER(uv)) \
885 && UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv)))
887 #define UNICODE_IS_SUPER(uv) ((UV) (uv) > PERL_UNICODE_MAX)
889 #define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
890 #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
891 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
892 #define MICRO_SIGN MICRO_SIGN_NATIVE
893 #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
894 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
895 #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
896 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
897 #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
898 #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
899 #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
900 #define GREEK_SMALL_LETTER_MU 0x03BC
901 #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case
903 #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
904 #ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8
905 # define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
907 #define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130
908 #define LATIN_SMALL_LETTER_DOTLESS_I 0x131
909 #define LATIN_SMALL_LETTER_LONG_S 0x017F
910 #define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05
911 #define LATIN_SMALL_LIGATURE_ST 0xFB06
912 #define KELVIN_SIGN 0x212A
913 #define ANGSTROM_SIGN 0x212B
915 #define UNI_DISPLAY_ISPRINT 0x0001
916 #define UNI_DISPLAY_BACKSLASH 0x0002
917 #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
918 #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
920 #define ANYOF_FOLD_SHARP_S(node, input, end) \
921 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
922 (ANYOF_NONBITMAP(node)) && \
923 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
924 ((end) > (input) + 1) && \
925 isALPHA_FOLD_EQ((input)[0], 's'))
927 #define SHARP_S_SKIP 2
929 #define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end)
930 #define bytes_from_utf8(s, lenp, is_utf8p) \
931 bytes_from_utf8_loc(s, lenp, is_utf8p, 0)
935 =for apidoc Am|STRLEN|isUTF8_CHAR_flags|const U8 *s|const U8 *e| const U32 flags
937 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
938 looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl,
939 that represents some code point, subject to the restrictions given by C<flags>;
940 otherwise it evaluates to 0. If non-zero, the value gives how many bytes
941 starting at C<s> comprise the code point's representation. Any bytes remaining
942 before C<e>, but beyond the ones needed to form the first code point in C<s>,
945 If C<flags> is 0, this gives the same results as C<L</isUTF8_CHAR>>;
946 if C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results
947 as C<L</isSTRICT_UTF8_CHAR>>;
948 and if C<flags> is C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives
949 the same results as C<L</isC9_STRICT_UTF8_CHAR>>.
950 Otherwise C<flags> may be any combination of the C<UTF8_DISALLOW_I<foo>> flags
951 understood by C<L</utf8n_to_uvchr>>, with the same meanings.
953 The three alternative macros are for the most commonly needed validations; they
954 are likely to run somewhat faster than this more general one, as they can be
955 inlined into your code.
957 Use L</is_utf8_string_flags>, L</is_utf8_string_loc_flags>, and
958 L</is_utf8_string_loclen_flags> to check entire strings.
963 #define isUTF8_CHAR_flags(s, e, flags) \
964 (UNLIKELY((e) <= (s)) \
966 : (UTF8_IS_INVARIANT(*s)) \
968 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
970 : _is_utf8_char_helper(s, e, flags))
972 /* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is
973 * retained solely for backwards compatibility */
974 #define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n)
976 #endif /* PERL_UTF8_H_ */
979 * ex: set ts=8 sts=4 sw=4 et: