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
38 =head1 Unicode Support
39 L<perlguts/Unicode Support> has an introduction to this API.
41 See also L</Character classification>,
42 and L</Character case changing>.
43 Various functions outside this section also work specially with Unicode.
44 Search for the string "utf8" in this document.
46 =for apidoc is_ascii_string
48 This is a misleadingly-named synonym for L</is_utf8_invariant_string>.
49 On ASCII-ish platforms, the name isn't misleading: the ASCII-range characters
50 are exactly the UTF-8 invariants. But EBCDIC machines have more invariants
51 than just the ASCII characters, so C<is_utf8_invariant_string> is preferred.
53 =for apidoc is_invariant_string
55 This is a somewhat misleadingly-named synonym for L</is_utf8_invariant_string>.
56 C<is_utf8_invariant_string> is preferred, as it indicates under what conditions
57 the string is invariant.
61 #define is_ascii_string(s, len) is_utf8_invariant_string(s, len)
62 #define is_invariant_string(s, len) is_utf8_invariant_string(s, len)
64 #define uvoffuni_to_utf8_flags(d,uv,flags) \
65 uvoffuni_to_utf8_flags_msgs(d, uv, flags, 0)
66 #define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0)
67 #define uvchr_to_utf8_flags(d,uv,flags) \
68 uvchr_to_utf8_flags_msgs(d,uv,flags, 0)
69 #define uvchr_to_utf8_flags_msgs(d,uv,flags,msgs) \
70 uvoffuni_to_utf8_flags_msgs(d,NATIVE_TO_UNI(uv),flags, msgs)
71 #define utf8_to_uvchr_buf(s, e, lenp) \
72 (__ASSERT_((U8*) (e) > (U8*) (s)) \
73 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \
74 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY))
75 #define utf8n_to_uvchr(s, len, lenp, flags) \
76 utf8n_to_uvchr_error(s, len, lenp, flags, 0)
77 #define utf8n_to_uvchr_error(s, len, lenp, flags, errors) \
78 utf8n_to_uvchr_msgs(s, len, lenp, flags, errors, 0)
80 #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
82 #define to_utf8_fold(s, r, lenr) \
83 _to_utf8_fold_flags (s, NULL, r, lenr, FOLD_FLAGS_FULL, __FILE__, __LINE__)
84 #define to_utf8_lower(s, r, lenr) \
85 _to_utf8_lower_flags(s, NULL, r ,lenr, 0, __FILE__, __LINE__)
86 #define to_utf8_upper(s, r, lenr) \
87 _to_utf8_upper_flags(s, NULL, r, lenr, 0, __FILE__, __LINE__)
88 #define to_utf8_title(s, r, lenr) \
89 _to_utf8_title_flags(s, NULL, r, lenr ,0, __FILE__, __LINE__)
91 #define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
92 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
93 #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
94 #define FOLDEQ_LOCALE (1 << 1)
95 #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
96 #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
97 #define FOLDEQ_S1_FOLDS_SANE (1 << 4)
98 #define FOLDEQ_S2_FOLDS_SANE (1 << 5)
100 #define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
101 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
104 /* The equivalent of these macros but implementing UTF-EBCDIC
105 are in the following header file:
108 #include "utfebcdic.h"
113 /* How wide can a single UTF-8 encoded character become in bytes. */
114 /* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
115 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
116 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
117 * non-negative integers in a binary format, even those above Unicode */
118 #define UTF8_MAXBYTES 13
121 EXTCONST unsigned char PL_utf8skip[] = {
122 /* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
123 /* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
124 /* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
125 /* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
126 /* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
127 /* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
128 /* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
129 /* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
130 /* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
131 /* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
132 /* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
133 /* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
134 /* 0xC0 */ 2,2, /* overlong */
135 /* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
136 /* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
137 /* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
138 /* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
139 /* Perl extended (never was official UTF-8). Up to 36 bit */
141 /* More extended, Up to 72 bits (64-bit + reserved) */
142 /* 0xFF */ UTF8_MAXBYTES
145 EXTCONST unsigned char PL_utf8skip[];
150 #if defined(_MSC_VER) && _MSC_VER < 1400
151 /* older MSVC versions have a smallish macro buffer */
152 #define PERL_SMALL_MACRO_BUFFER
155 /* Native character to/from iso-8859-1. Are the identity functions on ASCII
157 #ifdef PERL_SMALL_MACRO_BUFFER
158 #define NATIVE_TO_LATIN1(ch) ((U8)(ch))
159 #define LATIN1_TO_NATIVE(ch) ((U8)(ch))
161 #define NATIVE_TO_LATIN1(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
162 #define LATIN1_TO_NATIVE(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
165 /* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
166 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
167 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
168 * because they are 8-bit encodings that serve the same purpose in Perl, and
169 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
170 * whichever one is applicable on the current platform */
171 #ifdef PERL_SMALL_MACRO_BUFFER
172 #define NATIVE_UTF8_TO_I8(ch) ((U8) (ch))
173 #define I8_TO_NATIVE_UTF8(ch) ((U8) (ch))
175 #define NATIVE_UTF8_TO_I8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
176 #define I8_TO_NATIVE_UTF8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch)))
179 /* Transforms in wide UV chars */
180 #define UNI_TO_NATIVE(ch) ((UV) (ch))
181 #define NATIVE_TO_UNI(ch) ((UV) (ch))
185 The following table is from Unicode 3.2, plus the Perl extensions for above
188 Code Points 1st Byte 2nd Byte 3rd 4th 5th 6th 7th 8th-13th
190 U+0000..U+007F 00..7F
191 U+0080..U+07FF * C2..DF 80..BF
192 U+0800..U+0FFF E0 * A0..BF 80..BF
193 U+1000..U+CFFF E1..EC 80..BF 80..BF
194 U+D000..U+D7FF ED 80..9F 80..BF
195 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
196 U+E000..U+FFFF EE..EF 80..BF 80..BF
197 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
198 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
199 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
200 Below are above-Unicode code points
201 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
202 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
203 U+200000..U+FFFFFF F8 * 88..BF 80..BF 80..BF 80..BF
204 U+1000000..U+3FFFFFF F9..FB 80..BF 80..BF 80..BF 80..BF
205 U+4000000..U+3FFFFFFF FC * 84..BF 80..BF 80..BF 80..BF 80..BF
206 U+40000000..U+7FFFFFFF FD 80..BF 80..BF 80..BF 80..BF 80..BF
207 U+80000000..U+FFFFFFFFF FE * 82..BF 80..BF 80..BF 80..BF 80..BF 80..BF
208 U+1000000000.. FF 80..BF 80..BF 80..BF 80..BF 80..BF * 81..BF 80..BF
210 Note the gaps before several of the byte entries above marked by '*'. These are
211 caused by legal UTF-8 avoiding non-shortest encodings: it is technically
212 possible to UTF-8-encode a single code point in different ways, but that is
213 explicitly forbidden, and the shortest possible encoding should always be used
214 (and that is what Perl does). The non-shortest ones are called 'overlongs'.
219 Another way to look at it, as bits:
221 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
224 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
225 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
226 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
228 As you can see, the continuation bytes all begin with C<10>, and the
229 leading bits of the start byte tell how many bytes there are in the
232 Perl's extended UTF-8 means we can have start bytes up through FF, though any
233 beginning with FF yields a code point that is too large for 32-bit ASCII
234 platforms. FF signals to use 13 bytes for the encoded character. This breaks
235 the paradigm that the number of leading bits gives how many total bytes there
236 are in the character.
240 /* Is the representation of the Unicode code point 'cp' the same regardless of
241 * being encoded in UTF-8 or not? */
242 #define OFFUNI_IS_INVARIANT(cp) isASCII(cp)
245 =for apidoc Am|bool|UVCHR_IS_INVARIANT|UV cp
247 Evaluates to 1 if the representation of code point C<cp> is the same whether or
248 not it is encoded in UTF-8; otherwise evaluates to 0. UTF-8 invariant
249 characters can be copied as-is when converting to/from UTF-8, saving time.
250 C<cp> is Unicode if above 255; otherwise is platform-native.
255 #define UVCHR_IS_INVARIANT(cp) OFFUNI_IS_INVARIANT(cp)
257 /* This defines the bits that are to be in the continuation bytes of a multi-byte
258 * UTF-8 encoded character that mark it is a continuation byte. */
259 #define UTF_CONTINUATION_MARK 0x80
261 /* Misleadingly named: is the UTF8-encoded byte 'c' part of a variant sequence
262 * in UTF-8? This is the inverse of UTF8_IS_INVARIANT. The |0 makes sure this
263 * isn't mistakenly called with a ptr argument */
264 #define UTF8_IS_CONTINUED(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
265 ((U8)((c) | 0)) & UTF_CONTINUATION_MARK)
267 /* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
268 * This doesn't catch invariants (they are single-byte). It also excludes the
269 * illegal overlong sequences that begin with C0 and C1. The |0 makes sure
270 * this isn't mistakenly called with a ptr argument */
271 #define UTF8_IS_START(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
272 ((U8)((c) | 0)) >= 0xc2)
274 /* For use in UTF8_IS_CONTINUATION() below */
275 #define UTF_IS_CONTINUATION_MASK 0xC0
277 /* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
278 * first byte thereof? The |0 makes sure this isn't mistakenly called with a
280 #define UTF8_IS_CONTINUATION(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
281 (((U8)((c) | 0)) & UTF_IS_CONTINUATION_MASK) == UTF_CONTINUATION_MARK)
283 /* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
284 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
285 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
286 * this matches 0xc[23]. The |0 makes sure this isn't mistakenly called with a
288 #define UTF8_IS_DOWNGRADEABLE_START(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
289 (((U8)((c) | 0)) & 0xfe) == 0xc2)
291 /* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
292 * represent a code point > 255? The |0 makes sure this isn't mistakenly
293 * called with a ptr argument */
294 #define UTF8_IS_ABOVE_LATIN1(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
295 ((U8)((c) | 0)) >= 0xc4)
297 /* This is the number of low-order bits a continuation byte in a UTF-8 encoded
298 * sequence contributes to the specification of the code point. In the bit
299 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
300 * real information */
301 #define UTF_ACCUMULATION_SHIFT 6
303 /* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL()
305 #define QUESTION_MARK_CTRL DEL_NATIVE
307 /* Surrogates, non-character code points and above-Unicode code points are
308 * problematic in some contexts. This allows code that needs to check for
309 * those to to quickly exclude the vast majority of code points it will
311 #define isUTF8_POSSIBLY_PROBLEMATIC(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
314 #define UNICODE_IS_PERL_EXTENDED(uv) UNLIKELY((UV) (uv) > 0x7FFFFFFF)
316 #endif /* EBCDIC vs ASCII */
318 /* 2**UTF_ACCUMULATION_SHIFT - 1 */
319 #define UTF_CONTINUATION_MASK ((U8) ((1U << UTF_ACCUMULATION_SHIFT) - 1))
321 /* Internal macro to be used only in this file to aid in constructing other
322 * publicly accessible macros.
323 * The number of bytes required to express this uv in UTF-8, for just those
324 * uv's requiring 2 through 6 bytes, as these are common to all platforms and
325 * word sizes. The number of bytes needed is given by the number of leading 1
326 * bits in the start byte. There are 32 start bytes that have 2 initial 1 bits
327 * (C0-DF); there are 16 that have 3 initial 1 bits (E0-EF); 8 that have 4
328 * initial 1 bits (F0-F8); 4 that have 5 initial 1 bits (F9-FB), and 2 that
329 * have 6 initial 1 bits (FC-FD). The largest number a string of n bytes can
330 * represent is (the number of possible start bytes for 'n')
331 * * (the number of possiblities for each start byte
332 * The latter in turn is
333 * 2 ** ( (how many continuation bytes there are)
334 * * (the number of bits of information each
335 * continuation byte holds))
337 * If we were on a platform where we could use a fast find first set bit
338 * instruction (or count leading zeros instruction) this could be replaced by
339 * using that to find the log2 of the uv, and divide that by the number of bits
340 * of information in each continuation byte, adjusting for large cases and how
341 * much information is in a start byte for that length */
342 #define __COMMON_UNI_SKIP(uv) \
343 (UV) (uv) < (32 * (1U << ( UTF_ACCUMULATION_SHIFT))) ? 2 : \
344 (UV) (uv) < (16 * (1U << (2 * UTF_ACCUMULATION_SHIFT))) ? 3 : \
345 (UV) (uv) < ( 8 * (1U << (3 * UTF_ACCUMULATION_SHIFT))) ? 4 : \
346 (UV) (uv) < ( 4 * (1U << (4 * UTF_ACCUMULATION_SHIFT))) ? 5 : \
347 (UV) (uv) < ( 2 * (1U << (5 * UTF_ACCUMULATION_SHIFT))) ? 6 :
349 /* Internal macro to be used only in this file.
350 * This adds to __COMMON_UNI_SKIP the details at this platform's upper range.
351 * For any-sized EBCDIC platforms, or 64-bit ASCII ones, we need one more test
352 * to see if just 7 bytes is needed, or if the maximum is needed. For 32-bit
353 * ASCII platforms, everything is representable by 7 bytes */
354 #if defined(UV_IS_QUAD) || defined(EBCDIC)
355 # define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) \
356 (UV) (uv) < ((UV) 1U << (6 * UTF_ACCUMULATION_SHIFT)) ? 7 : UTF8_MAXBYTES)
358 # define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) 7)
361 /* The next two macros use the base macro defined above, and add in the tests
362 * at the low-end of the range, for just 1 byte, yielding complete macros,
363 * publicly accessible. */
365 /* Input is a true Unicode (not-native) code point */
366 #define OFFUNISKIP(uv) (OFFUNI_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
370 =for apidoc Am|STRLEN|UVCHR_SKIP|UV cp
371 returns the number of bytes required to represent the code point C<cp> when
372 encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than
373 255; a Unicode code point otherwise.
377 #define UVCHR_SKIP(uv) ( UVCHR_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv))
379 /* The largest code point representable by two UTF-8 bytes on this platform.
380 * As explained in the comments for __COMMON_UNI_SKIP, 32 start bytes with
381 * UTF_ACCUMULATION_SHIFT bits of information each */
382 #define MAX_UTF8_TWO_BYTE (32 * (1U << UTF_ACCUMULATION_SHIFT) - 1)
384 /* The largest code point representable by two UTF-8 bytes on any platform that
385 * Perl runs on. This value is constrained by EBCDIC which has 5 bits per
386 * continuation byte */
387 #define MAX_PORTABLE_UTF8_TWO_BYTE (32 * (1U << 5) - 1)
389 /* The maximum number of UTF-8 bytes a single Unicode character can
390 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
391 * expansion is UTF8_MAX_FOLD_CHAR_EXPAND characters, but any above-Unicode
392 * code point will fold to itself, so we only have to look at the expansion of
393 * the maximum Unicode code point. But this number may be less than the space
394 * occupied by a very large code point under Perl's extended UTF-8. We have to
395 * make it large enough to fit any single character. (It turns out that ASCII
396 * and EBCDIC differ in which is larger) */
397 #define UTF8_MAXBYTES_CASE \
398 (UTF8_MAXBYTES >= (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)) \
400 : (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)))
402 /* Rest of these are attributes of Unicode and perl's internals rather than the
403 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
404 * this level; the macros that some of these call may have different
405 * definitions in the two encodings */
407 /* In domain restricted to ASCII, these may make more sense to the reader than
408 * the ones with Latin1 in the name */
409 #define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
410 #define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
412 /* More or less misleadingly-named defines, retained for back compat */
413 #define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
414 #define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
415 #define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
416 #define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
417 #define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
419 /* This defines the 1-bits that are to be in the first byte of a multi-byte
420 * UTF-8 encoded character that mark it as a start byte and give the number of
421 * bytes that comprise the character. 'len' is the number of bytes in the
422 * multi-byte sequence. */
423 #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
425 /* Masks out the initial one bits in a start byte, leaving the real data ones.
426 * Doesn't work on an invariant byte. 'len' is the number of bytes in the
427 * multi-byte sequence that comprises the character. */
428 #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
430 /* Adds a UTF8 continuation byte 'new' of information to a running total code
431 * point 'old' of all the continuation bytes so far. This is designed to be
432 * used in a loop to convert from UTF-8 to the code point represented. Note
433 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is
434 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC)
435 * code point in process of being generated */
436 #define UTF8_ACCUMULATE(old, new) (__ASSERT_(FITS_IN_8_BITS(new)) \
437 ((old) << UTF_ACCUMULATION_SHIFT) \
438 | ((NATIVE_UTF8_TO_I8((U8)new)) \
439 & UTF_CONTINUATION_MASK))
441 /* This works in the face of malformed UTF-8. */
442 #define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) \
443 ( UTF8_IS_DOWNGRADEABLE_START(*(s)) \
444 && ( (e) - (s) > 1) \
445 && UTF8_IS_CONTINUATION(*((s)+1)))
447 /* Number of bytes a code point occupies in UTF-8. */
448 #define NATIVE_SKIP(uv) UVCHR_SKIP(uv)
450 /* Most code which says UNISKIP is really thinking in terms of native code
451 * points (0-255) plus all those beyond. This is an imprecise term, but having
452 * it means existing code continues to work. For precision, use UVCHR_SKIP,
453 * NATIVE_SKIP, or OFFUNISKIP */
454 #define UNISKIP(uv) UVCHR_SKIP(uv)
456 /* Longer, but more accurate name */
457 #define UTF8_IS_ABOVE_LATIN1_START(c) UTF8_IS_ABOVE_LATIN1(c)
459 /* Convert a UTF-8 variant Latin1 character to a native code point value.
460 * Needs just one iteration of accumulate. Should be used only if it is known
461 * that the code point is < 256, and is not UTF-8 invariant. Use the slower
462 * but more general TWO_BYTE_UTF8_TO_NATIVE() which handles any code point
463 * representable by two bytes (which turns out to be up through
464 * MAX_PORTABLE_UTF8_TWO_BYTE). The two parameters are:
465 * HI: a downgradable start byte;
468 #define EIGHT_BIT_UTF8_TO_NATIVE(HI, LO) \
469 ( __ASSERT_(UTF8_IS_DOWNGRADEABLE_START(HI)) \
470 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
471 LATIN1_TO_NATIVE(UTF8_ACCUMULATE(( \
472 NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), (LO))))
474 /* Convert a two (not one) byte utf8 character to a native code point value.
475 * Needs just one iteration of accumulate. Should not be used unless it is
476 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
477 * Note that the result can be larger than 255 if the input character is not
479 #define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
480 (__ASSERT_(FITS_IN_8_BITS(HI)) \
481 __ASSERT_(FITS_IN_8_BITS(LO)) \
482 __ASSERT_(PL_utf8skip[HI] == 2) \
483 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \
484 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
487 /* Should never be used, and be deprecated */
488 #define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
492 =for apidoc Am|STRLEN|UTF8SKIP|char* s
493 returns the number of bytes in the UTF-8 encoded character whose first (perhaps
494 only) byte is pointed to by C<s>.
498 #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
499 #define UTF8_SKIP(s) UTF8SKIP(s)
503 =for apidoc Am|STRLEN|UTF8_SAFE_SKIP|char* s|char* e
504 returns the number of bytes in the UTF-8 encoded character whose first (perhaps
505 only) byte is pointed to by C<s>. But never returns beyond C<e>.
509 #define UTF8_SAFE_SKIP(s, e) (__ASSERT_((e) > (s)) \
510 MIN(((e) - (s)), UTF8_SKIP(s)))
512 /* Most code that says 'UNI_' really means the native value for code points up
514 #define UNI_IS_INVARIANT(cp) UVCHR_IS_INVARIANT(cp)
517 =for apidoc Am|bool|UTF8_IS_INVARIANT|char c
519 Evaluates to 1 if the byte C<c> represents the same character when encoded in
520 UTF-8 as when not; otherwise evaluates to 0. UTF-8 invariant characters can be
521 copied as-is when converting to/from UTF-8, saving time.
523 In spite of the name, this macro gives the correct result if the input string
524 from which C<c> comes is not encoded in UTF-8.
526 See C<L</UVCHR_IS_INVARIANT>> for checking if a UV is invariant.
530 The reason it works on both UTF-8 encoded strings and non-UTF-8 encoded, is
531 that it returns TRUE in each for the exact same set of bit patterns. It is
532 valid on a subset of what UVCHR_IS_INVARIANT is valid on, so can just use that;
533 and the compiler should optimize out anything extraneous given the
534 implementation of the latter. The |0 makes sure this isn't mistakenly called
537 #define UTF8_IS_INVARIANT(c) UVCHR_IS_INVARIANT((c) | 0)
539 /* Like the above, but its name implies a non-UTF8 input, which as the comments
540 * above show, doesn't matter as to its implementation */
541 #define NATIVE_BYTE_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c)
543 /* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic
544 * bytes from an ordinal that is known to fit into exactly two (not one) bytes;
545 * it must be less than 0x3FF to work across both encodings. */
547 /* These two are helper macros for the other three sets, and should not be used
548 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1
549 * (which works for code points up through 0xFF) or NATIVE_TO_UNI which works
550 * for any code point */
551 #define __BASE_TWO_BYTE_HI(c, translate_function) \
552 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
553 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \
554 | UTF_START_MARK(2)))
555 #define __BASE_TWO_BYTE_LO(c, translate_function) \
556 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \
557 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \
558 | UTF_CONTINUATION_MARK))
560 /* The next two macros should not be used. They were designed to be usable as
561 * the case label of a switch statement, but this doesn't work for EBCDIC. Use
562 * regen/unicode_constants.pl instead */
563 #define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)
564 #define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)
566 /* The next two macros are used when the source should be a single byte
567 * character; checked for under DEBUGGING */
568 #define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
569 ( __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1)))
570 #define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
571 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1)))
573 /* These final two macros in the series are used when the source can be any
574 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient
575 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~'
576 * operator instead of "<=" to avoid getting compiler warnings.
577 * MAX_UTF8_TWO_BYTE should be exactly all one bits in the lower few
578 * places, so the ~ works */
579 #define UTF8_TWO_BYTE_HI(c) \
580 (__ASSERT_((sizeof(c) == 1) \
581 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
582 (__BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI)))
583 #define UTF8_TWO_BYTE_LO(c) \
584 (__ASSERT_((sizeof(c) == 1) \
585 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \
586 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI)))
588 /* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
589 * as it is only in overlongs. */
590 #define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1)
593 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
594 * from an earlier version of these macros in which they didn't call the
595 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
596 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
597 * no need to do it again here
599 #define isIDFIRST_lazy_if(p,UTF) \
600 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isIDFIRST_lazy_if", \
601 "isIDFIRST_lazy_if_safe", \
602 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__)
604 #define isIDFIRST_lazy_if_safe(p, e, UTF) \
605 ((IN_BYTES || !UTF) \
607 : isIDFIRST_utf8_safe(p, e))
609 #define isWORDCHAR_lazy_if(p,UTF) \
610 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isWORDCHAR_lazy_if", \
611 "isWORDCHAR_lazy_if_safe", \
612 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__)
614 #define isWORDCHAR_lazy_if_safe(p, e, UTF) \
615 ((IN_BYTES || !UTF) \
617 : isWORDCHAR_utf8_safe((U8 *) p, (U8 *) e))
619 #define isALNUM_lazy_if(p,UTF) \
620 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isALNUM_lazy_if", \
621 "isWORDCHAR_lazy_if_safe", \
622 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__)
624 #define UTF8_MAXLEN UTF8_MAXBYTES
626 /* A Unicode character can fold to up to 3 characters */
627 #define UTF8_MAX_FOLD_CHAR_EXPAND 3
629 #define IN_BYTES UNLIKELY(CopHINTS_get(PL_curcop) & HINT_BYTES)
633 =for apidoc Am|bool|DO_UTF8|SV* sv
634 Returns a bool giving whether or not the PV in C<sv> is to be treated as being
637 You should use this I<after> a call to C<SvPV()> or one of its variants, in
638 case any call to string overloading updates the internal UTF-8 encoding flag.
642 #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
644 /* Should all strings be treated as Unicode, and not just UTF-8 encoded ones?
645 * Is so within 'feature unicode_strings' or 'locale :not_characters', and not
646 * within 'use bytes'. UTF-8 locales are not tested for here, but perhaps
648 #define IN_UNI_8_BIT \
649 (( ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT)) \
650 || ( CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \
651 /* -1 below is for :not_characters */ \
652 && _is_in_locale_category(FALSE, -1))) \
656 #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
657 #define UTF8_GOT_EMPTY UTF8_ALLOW_EMPTY
659 /* Allow first byte to be a continuation byte */
660 #define UTF8_ALLOW_CONTINUATION 0x0002
661 #define UTF8_GOT_CONTINUATION UTF8_ALLOW_CONTINUATION
663 /* Unexpected non-continuation byte */
664 #define UTF8_ALLOW_NON_CONTINUATION 0x0004
665 #define UTF8_GOT_NON_CONTINUATION UTF8_ALLOW_NON_CONTINUATION
667 /* expecting more bytes than were available in the string */
668 #define UTF8_ALLOW_SHORT 0x0008
669 #define UTF8_GOT_SHORT UTF8_ALLOW_SHORT
671 /* Overlong sequence; i.e., the code point can be specified in fewer bytes.
672 * First one will convert the overlong to the REPLACEMENT CHARACTER; second
673 * will return what the overlong evaluates to */
674 #define UTF8_ALLOW_LONG 0x0010
675 #define UTF8_ALLOW_LONG_AND_ITS_VALUE (UTF8_ALLOW_LONG|0x0020)
676 #define UTF8_GOT_LONG UTF8_ALLOW_LONG
678 #define UTF8_ALLOW_OVERFLOW 0x0080
679 #define UTF8_GOT_OVERFLOW UTF8_ALLOW_OVERFLOW
681 #define UTF8_DISALLOW_SURROGATE 0x0100 /* Unicode surrogates */
682 #define UTF8_GOT_SURROGATE UTF8_DISALLOW_SURROGATE
683 #define UTF8_WARN_SURROGATE 0x0200
685 /* Unicode non-character code points */
686 #define UTF8_DISALLOW_NONCHAR 0x0400
687 #define UTF8_GOT_NONCHAR UTF8_DISALLOW_NONCHAR
688 #define UTF8_WARN_NONCHAR 0x0800
690 /* Super-set of Unicode: code points above the legal max */
691 #define UTF8_DISALLOW_SUPER 0x1000
692 #define UTF8_GOT_SUPER UTF8_DISALLOW_SUPER
693 #define UTF8_WARN_SUPER 0x2000
695 /* The original UTF-8 standard did not define UTF-8 with start bytes of 0xFE or
696 * 0xFF, though UTF-EBCDIC did. This allowed both versions to represent code
697 * points up to 2 ** 31 - 1. Perl extends UTF-8 so that 0xFE and 0xFF are
698 * usable on ASCII platforms, and 0xFF means something different than
699 * UTF-EBCDIC defines. These changes allow code points of 64 bits (actually
700 * somewhat more) to be represented on both platforms. But these are Perl
701 * extensions, and not likely to be interchangeable with other languages. Note
702 * that on ASCII platforms, FE overflows a signed 32-bit word, and FF an
704 #define UTF8_DISALLOW_PERL_EXTENDED 0x4000
705 #define UTF8_GOT_PERL_EXTENDED UTF8_DISALLOW_PERL_EXTENDED
706 #define UTF8_WARN_PERL_EXTENDED 0x8000
708 /* For back compat, these old names are misleading for overlongs and
710 #define UTF8_DISALLOW_ABOVE_31_BIT UTF8_DISALLOW_PERL_EXTENDED
711 #define UTF8_GOT_ABOVE_31_BIT UTF8_GOT_PERL_EXTENDED
712 #define UTF8_WARN_ABOVE_31_BIT UTF8_WARN_PERL_EXTENDED
713 #define UTF8_DISALLOW_FE_FF UTF8_DISALLOW_PERL_EXTENDED
714 #define UTF8_WARN_FE_FF UTF8_WARN_PERL_EXTENDED
716 #define UTF8_CHECK_ONLY 0x10000
717 #define _UTF8_NO_CONFIDENCE_IN_CURLEN 0x20000 /* Internal core use only */
719 /* For backwards source compatibility. They do nothing, as the default now
720 * includes what they used to mean. The first one's meaning was to allow the
721 * just the single non-character 0xFFFF */
722 #define UTF8_ALLOW_FFFF 0
723 #define UTF8_ALLOW_FE_FF 0
724 #define UTF8_ALLOW_SURROGATE 0
726 /* C9 refers to Unicode Corrigendum #9: allows but discourages non-chars */
727 #define UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE \
728 (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_SURROGATE)
729 #define UTF8_WARN_ILLEGAL_C9_INTERCHANGE (UTF8_WARN_SUPER|UTF8_WARN_SURROGATE)
731 #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE \
732 (UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE|UTF8_DISALLOW_NONCHAR)
733 #define UTF8_WARN_ILLEGAL_INTERCHANGE \
734 (UTF8_WARN_ILLEGAL_C9_INTERCHANGE|UTF8_WARN_NONCHAR)
736 /* This is typically used for code that processes UTF-8 input and doesn't want
737 * to have to deal with any malformations that might be present. All such will
738 * be safely replaced by the REPLACEMENT CHARACTER, unless other flags
739 * overriding this are also present. */
740 #define UTF8_ALLOW_ANY ( UTF8_ALLOW_CONTINUATION \
741 |UTF8_ALLOW_NON_CONTINUATION \
744 |UTF8_ALLOW_OVERFLOW)
746 /* Accept any Perl-extended UTF-8 that evaluates to any UV on the platform, but
747 * not any malformed. This is the default. */
748 #define UTF8_ALLOW_ANYUV 0
749 #define UTF8_ALLOW_DEFAULT UTF8_ALLOW_ANYUV
752 =for apidoc Am|bool|UTF8_IS_SURROGATE|const U8 *s|const U8 *e
754 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
755 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one
756 of the Unicode surrogate code points; otherwise it evaluates to 0. If
757 non-zero, the value gives how many bytes starting at C<s> comprise the code
758 point's representation.
762 #define UTF8_IS_SURROGATE(s, e) is_SURROGATE_utf8_safe(s, e)
765 #define UTF8_IS_REPLACEMENT(s, send) is_REPLACEMENT_utf8_safe(s,send)
767 #define MAX_LEGAL_CP ((UV)IV_MAX)
770 =for apidoc Am|bool|UTF8_IS_SUPER|const U8 *s|const U8 *e
772 Recall that Perl recognizes an extension to UTF-8 that can encode code
773 points larger than the ones defined by Unicode, which are 0..0x10FFFF.
775 This macro evaluates to non-zero if the first few bytes of the string starting
776 at C<s> and looking no further than S<C<e - 1>> are from this UTF-8 extension;
777 otherwise it evaluates to 0. If non-zero, the value gives how many bytes
778 starting at C<s> comprise the code point's representation.
780 0 is returned if the bytes are not well-formed extended UTF-8, or if they
781 represent a code point that cannot fit in a UV on the current platform. Hence
782 this macro can give different results when run on a 64-bit word machine than on
783 one with a 32-bit word size.
785 Note that it is illegal to have code points that are larger than what can
786 fit in an IV on the current machine.
791 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
792 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
793 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
796 # define UTF8_IS_SUPER(s, e) \
797 (( LIKELY((e) > (s) + 4) \
798 && NATIVE_UTF8_TO_I8(*(s)) >= 0xF9 \
799 && ( NATIVE_UTF8_TO_I8(*(s)) > 0xF9 \
800 || (NATIVE_UTF8_TO_I8(*((s) + 1)) >= 0xA2)) \
801 && LIKELY((s) + UTF8SKIP(s) <= (e))) \
802 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0)
804 # define UTF8_IS_SUPER(s, e) \
805 (( LIKELY((e) > (s) + 3) \
806 && (*(U8*) (s)) >= 0xF4 \
807 && ((*(U8*) (s)) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))\
808 && LIKELY((s) + UTF8SKIP(s) <= (e))) \
809 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0)
812 /* These are now machine generated, and the 'given' clause is no longer
814 #define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) \
815 cBOOL(is_NONCHAR_utf8_safe(s,e))
818 =for apidoc Am|bool|UTF8_IS_NONCHAR|const U8 *s|const U8 *e
820 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
821 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one
822 of the Unicode non-character code points; otherwise it evaluates to 0. If
823 non-zero, the value gives how many bytes starting at C<s> comprise the code
824 point's representation.
828 #define UTF8_IS_NONCHAR(s, e) \
829 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e)
831 #define UNICODE_SURROGATE_FIRST 0xD800
832 #define UNICODE_SURROGATE_LAST 0xDFFF
833 #define UNICODE_REPLACEMENT 0xFFFD
834 #define UNICODE_BYTE_ORDER_MARK 0xFEFF
836 /* Though our UTF-8 encoding can go beyond this,
837 * let's be conservative and do as Unicode says. */
838 #define PERL_UNICODE_MAX 0x10FFFF
840 #define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
841 #define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
842 #define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
843 #define UNICODE_WARN_PERL_EXTENDED 0x0008 /* Above 0x7FFF_FFFF */
844 #define UNICODE_WARN_ABOVE_31_BIT UNICODE_WARN_PERL_EXTENDED
845 #define UNICODE_DISALLOW_SURROGATE 0x0010
846 #define UNICODE_DISALLOW_NONCHAR 0x0020
847 #define UNICODE_DISALLOW_SUPER 0x0040
848 #define UNICODE_DISALLOW_PERL_EXTENDED 0x0080
849 #define UNICODE_DISALLOW_ABOVE_31_BIT UNICODE_DISALLOW_PERL_EXTENDED
851 #define UNICODE_GOT_SURROGATE UNICODE_DISALLOW_SURROGATE
852 #define UNICODE_GOT_NONCHAR UNICODE_DISALLOW_NONCHAR
853 #define UNICODE_GOT_SUPER UNICODE_DISALLOW_SUPER
854 #define UNICODE_GOT_PERL_EXTENDED UNICODE_DISALLOW_PERL_EXTENDED
856 #define UNICODE_WARN_ILLEGAL_C9_INTERCHANGE \
857 (UNICODE_WARN_SURROGATE|UNICODE_WARN_SUPER)
858 #define UNICODE_WARN_ILLEGAL_INTERCHANGE \
859 (UNICODE_WARN_ILLEGAL_C9_INTERCHANGE|UNICODE_WARN_NONCHAR)
860 #define UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE \
861 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_SUPER)
862 #define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
863 (UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE|UNICODE_DISALLOW_NONCHAR)
865 /* For backward source compatibility, as are now the default */
866 #define UNICODE_ALLOW_SURROGATE 0
867 #define UNICODE_ALLOW_SUPER 0
868 #define UNICODE_ALLOW_ANY 0
870 /* This matches the 2048 code points between UNICODE_SURROGATE_FIRST (0xD800) and
871 * UNICODE_SURROGATE_LAST (0xDFFF) */
872 #define UNICODE_IS_SURROGATE(uv) (((UV) (uv) & (~0xFFFF | 0xF800)) \
875 #define UNICODE_IS_REPLACEMENT(uv) ((UV) (uv) == UNICODE_REPLACEMENT)
876 #define UNICODE_IS_BYTE_ORDER_MARK(uv) ((UV) (uv) == UNICODE_BYTE_ORDER_MARK)
878 /* Is 'uv' one of the 32 contiguous-range noncharacters? */
879 #define UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) ((UV) (uv) >= 0xFDD0 \
880 && (UV) (uv) <= 0xFDEF)
882 /* Is 'uv' one of the 34 plane-ending noncharacters 0xFFFE, 0xFFFF, 0x1FFFE,
883 * 0x1FFFF, ... 0x10FFFE, 0x10FFFF, given that we know that 'uv' is not above
884 * the Unicode legal max */
885 #define UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv) \
886 (((UV) (uv) & 0xFFFE) == 0xFFFE)
888 #define UNICODE_IS_NONCHAR(uv) \
889 ( UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) \
890 || ( LIKELY( ! UNICODE_IS_SUPER(uv)) \
891 && UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv)))
893 #define UNICODE_IS_SUPER(uv) ((UV) (uv) > PERL_UNICODE_MAX)
895 #define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
896 #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
897 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
898 #define MICRO_SIGN MICRO_SIGN_NATIVE
899 #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
900 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
901 #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
902 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
903 #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
904 #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
905 #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
906 #define GREEK_SMALL_LETTER_MU 0x03BC
907 #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case
909 #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
910 #ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8
911 # define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
913 #define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130
914 #define LATIN_SMALL_LETTER_DOTLESS_I 0x131
915 #define LATIN_SMALL_LETTER_LONG_S 0x017F
916 #define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05
917 #define LATIN_SMALL_LIGATURE_ST 0xFB06
918 #define KELVIN_SIGN 0x212A
919 #define ANGSTROM_SIGN 0x212B
921 #define UNI_DISPLAY_ISPRINT 0x0001
922 #define UNI_DISPLAY_BACKSLASH 0x0002
923 #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
924 #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
926 #define ANYOF_FOLD_SHARP_S(node, input, end) \
927 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
928 (ANYOF_NONBITMAP(node)) && \
929 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
930 ((end) > (input) + 1) && \
931 isALPHA_FOLD_EQ((input)[0], 's'))
933 #define SHARP_S_SKIP 2
935 #define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end)
936 #define bytes_from_utf8(s, lenp, is_utf8p) \
937 bytes_from_utf8_loc(s, lenp, is_utf8p, 0)
941 =for apidoc Am|STRLEN|isUTF8_CHAR_flags|const U8 *s|const U8 *e| const U32 flags
943 Evaluates to non-zero if the first few bytes of the string starting at C<s> and
944 looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl,
945 that represents some code point, subject to the restrictions given by C<flags>;
946 otherwise it evaluates to 0. If non-zero, the value gives how many bytes
947 starting at C<s> comprise the code point's representation. Any bytes remaining
948 before C<e>, but beyond the ones needed to form the first code point in C<s>,
951 If C<flags> is 0, this gives the same results as C<L</isUTF8_CHAR>>;
952 if C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results
953 as C<L</isSTRICT_UTF8_CHAR>>;
954 and if C<flags> is C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives
955 the same results as C<L</isC9_STRICT_UTF8_CHAR>>.
956 Otherwise C<flags> may be any combination of the C<UTF8_DISALLOW_I<foo>> flags
957 understood by C<L</utf8n_to_uvchr>>, with the same meanings.
959 The three alternative macros are for the most commonly needed validations; they
960 are likely to run somewhat faster than this more general one, as they can be
961 inlined into your code.
963 Use L</is_utf8_string_flags>, L</is_utf8_string_loc_flags>, and
964 L</is_utf8_string_loclen_flags> to check entire strings.
969 #define isUTF8_CHAR_flags(s, e, flags) \
970 (UNLIKELY((e) <= (s)) \
972 : (UTF8_IS_INVARIANT(*s)) \
974 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \
976 : _is_utf8_char_helper(s, e, flags))
978 /* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is
979 * retained solely for backwards compatibility */
980 #define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n)
982 #endif /* PERL_UTF8_H_ */
985 * ex: set ts=8 sts=4 sw=4 et: