This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: Fix bug in EBCDIC
[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
KW
32/* For to_utf8_fold_flags, q.v. */
33#define FOLD_FLAGS_LOCALE 0x1
34#define FOLD_FLAGS_FULL 0x2
a0270393 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
a0270393 42#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
051a06d4
KW
43#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
44 FOLD_FLAGS_FULL, NULL)
45#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
46#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
47#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 48
fd7cb289 49/* Source backward compatibility. */
fd7cb289
RGS
50#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
51
eda9cac1
KW
52#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
53 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 54#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 55#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
56#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
57#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 58
e6226b18
KW
59/*
60=for apidoc ibcmp_utf8
61
62This is a synonym for (! foldEQ_utf8())
63
64=cut
65*/
66#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
67 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
68
1d72bdf6
NIS
69#ifdef EBCDIC
70/* The equivalent of these macros but implementing UTF-EBCDIC
71 are in the following header file:
72 */
73
74#include "utfebcdic.h"
fd7cb289 75
d06134e5 76#else /* ! EBCDIC */
73c4f7a1
GS
77START_EXTERN_C
78
a0ed51b3 79#ifdef DOINIT
6f06b55f 80EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
81/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
82/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
83/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
84/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
85/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
86/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
87/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
88/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
89/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
90/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
91/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
92/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
93/* 0xC0 */ 2,2, /* overlong */
94/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
95/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
96/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
97/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
98/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
99 allowed (64-bit + reserved). */
a0ed51b3
LW
100};
101#else
6f06b55f 102EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
103#endif
104
73c4f7a1 105END_EXTERN_C
7e2040f0 106
59a449d5
KW
107/* Native character to/from iso-8859-1. Are the identity functions on ASCII
108 * platforms */
109#define NATIVE_TO_LATIN1(ch) (ch)
110#define LATIN1_TO_NATIVE(ch) (ch)
111
112/* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
113 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
114 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
115 * because they are 8-bit encodings that serve the same purpose in Perl, and
116 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
117 * whichever one is applicable on the current platform */
118#define NATIVE_UTF8_TO_I8(ch) (ch)
119#define I8_TO_NATIVE_UTF8(ch) (ch)
120
1d72bdf6
NIS
121/* Transforms in wide UV chars */
122#define UNI_TO_NATIVE(ch) (ch)
123#define NATIVE_TO_UNI(ch) (ch)
d7578b48 124
d06134e5 125/* As there are no translations, avoid the function wrapper */
378516de
KW
126#define utf8n_to_uvchr utf8n_to_uvoffuni
127#define uvchr_to_utf8(a,b) uvoffuni_to_utf8_flags(a,b,0)
2b9d42f0 128
877d9f0d 129/*
9041c2e3 130
8c007b5a 131 The following table is from Unicode 3.2.
877d9f0d
JH
132
133 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
134
375122d7 135 U+0000..U+007F 00..7F
e1b711da 136 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 137 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 138 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 139 U+D000..U+D7FF ED 80..9F 80..BF
537124e4 140 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
375122d7 141 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 142 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
143 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
144 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
145 Below are non-Unicode code points
146 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
147 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
537124e4 148 U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 149
e1b711da 150Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
151caused by legal UTF-8 avoiding non-shortest encodings: it is technically
152possible to UTF-8-encode a single code point in different ways, but that is
153explicitly forbidden, and the shortest possible encoding should always be used
15824458 154(and that is what Perl does). The non-shortest ones are called 'overlongs'.
8c007b5a 155
877d9f0d
JH
156 */
157
8c007b5a
JH
158/*
159 Another way to look at it, as bits:
160
b2635aa8 161 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 162
b2635aa8
KW
163 0aaa aaaa 0aaa aaaa
164 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
165 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
166 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
167
168As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 169leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
170encoded character.
171
65ab9279
TC
172Perl's extended UTF-8 means we can have start bytes up to FF.
173
8c007b5a
JH
174*/
175
15824458
KW
176/* Is the representation of the Unicode code point 'c' the same regardless of
177 * being encoded in UTF-8 or not? */
c4d5f83a 178#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
15824458
KW
179
180/* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is
181 * the inverse of UTF8_IS_INVARIANT */
182#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
183
184/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
185 * This doesn't catch invariants (they are single-byte). It also excludes the
186 * illegal overlong sequences that begin with C0 and C1. */
0447e8df 187#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
15824458
KW
188
189/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
190 * first byte thereof? */
e021c6e6 191#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80)
0ae1fa71 192
15824458
KW
193/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
194 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
195 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
196 * this matches 0xc[23]. */
559c7f10 197#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2)
4ab10950 198
15824458
KW
199/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
200 * represent a code point > 255? */
7b4252f4 201#define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4)
8850bf83 202
15824458
KW
203/* This defines the 1-bits that are to be in the first byte of a multi-byte
204 * UTF-8 encoded character that give the number of bytes that comprise the
205 * character.
206 * */
ee372ee9 207#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len)))))
b2635aa8
KW
208
209/* Masks out the initial one bits in a start byte, leaving the real data ones.
210 * Doesn't work on an invariant byte */
22901f30 211#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6 212
15824458
KW
213/* This defines the bits that are to be in the continuation bytes of a multi-byte
214 * UTF-8 encoded character that indicate it is a continuation byte. */
1d72bdf6 215#define UTF_CONTINUATION_MARK 0x80
15824458
KW
216
217/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
218 * sequence contributes to the specification of the code point. In the bit
219 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
220 * real information */
1d72bdf6 221#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW
222
223/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 224#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 225
15824458
KW
226/* If a value is anded with this, and the result is non-zero, then using the
227 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
228 * */
eb83ed87 229#define UTF_ACCUMULATION_OVERFLOW_MASK \
bb88be5f
KW
230 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
231 - UTF_ACCUMULATION_SHIFT))
eb83ed87 232
1d68d6cd 233#ifdef HAS_QUAD
5aaebcb3
KW
234/* Input is a true Unicode (not-native) code point */
235#define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
236 (uv) < 0x800 ? 2 : \
237 (uv) < 0x10000 ? 3 : \
238 (uv) < 0x200000 ? 4 : \
239 (uv) < 0x4000000 ? 5 : \
240 (uv) < 0x80000000 ? 6 : \
9041c2e3 241 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
242#else
243/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5aaebcb3 244#define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
245 (uv) < 0x800 ? 2 : \
246 (uv) < 0x10000 ? 3 : \
247 (uv) < 0x200000 ? 4 : \
248 (uv) < 0x4000000 ? 5 : \
249 (uv) < 0x80000000 ? 6 : 7 )
250#endif
251
03c76984
KW
252/* How wide can a single UTF-8 encoded character become in bytes. */
253/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
254 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
255 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
256 * non-negative integers in a binary format, even those above Unicode */
257#define UTF8_MAXBYTES 13
258
259/* The maximum number of UTF-8 bytes a single Unicode character can
260 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
261 * expansion is 3 characters. On ASCIIish platforms, the highest Unicode
262 * character occupies 4 bytes, therefore this number would be 12, but this is
263 * smaller than the maximum width a single above-Unicode character can occupy,
264 * so use that instead */
265#if UTF8_MAXBYTES < 12
266#error UTF8_MAXBYTES must be at least 12
267#endif
268
269#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
270
d06134e5
KW
271#endif /* EBCDIC vs ASCII */
272
273/* Rest of these are attributes of Unicode and perl's internals rather than the
274 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
275 * this level; the macros that some of these call may have different
276 * definitions in the two encodings */
277
59a449d5
KW
278/* In domain restricted to ASCII, these may make more sense to the reader than
279 * the ones with Latin1 in the name */
280#define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
281#define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
282
283/* More or less misleadingly-named defines, retained for back compat */
284#define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
285#define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
286#define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
287#define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
288#define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
d06134e5 289
537124e4
KW
290/* Adds a UTF8 continuation byte 'new' of information to a running total code
291 * point 'old' of all the continuation bytes so far. This is designed to be
292 * used in a loop to convert from UTF-8 to the code point represented */
bb88be5f
KW
293#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
294 | (((U8)new) & UTF_CONTINUATION_MASK))
d06134e5 295
4ab10950
KW
296/* This works in the face of malformed UTF-8. */
297#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
298 && ( (e) - (s) > 1) \
299 && UTF8_IS_CONTINUATION(*((s)+1)))
300
5aaebcb3
KW
301/* Number of bytes a code point occupies in UTF-8. */
302#define NATIVE_SKIP(uv) OFFUNISKIP(NATIVE_TO_UNI(uv))
303
304/* Most code which says UNISKIP is really thinking in terms of native code
305 * points (0-255) plus all those beyond. This is an imprecise term, but having
306 * it means existing code continues to work. For precision, use NATIVE_SKIP
307 * and OFFUNISKIP */
308#define UNISKIP(uv) NATIVE_SKIP(uv)
309
94bb8c36 310/* Convert a two (not one) byte utf8 character to a native code point value.
2950f2a7
KW
311 * Needs just one iteration of accumulate. Should not be used unless it is
312 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
313 * Note that the result can be larger than 255 if the input character is not
314 * downgradable */
94bb8c36
KW
315#define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \
316 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \
317 NATIVE_UTF8_TO_I8(LO)))
318
319/* Should never be used, and be deprecated */
320#define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO))
2950f2a7 321
15824458
KW
322/* How many bytes in the UTF-8 encoded character whose first (perhaps only)
323 * byte is pointed to by 's' */
d06134e5
KW
324#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
325
15824458
KW
326/* Is the byte 'c' the same character when encoded in UTF-8 as when not. This
327 * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in
328 * each for the exact same set of bit patterns. (And it works on any byte in a
329 * UTF-8 encoded string) */
bc3632a8 330#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_UTF8_TO_I8(c))
15824458 331
bc3632a8 332#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_LATIN1(c))
d06134e5
KW
333
334#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
335
336/* The macros in the next sets are used to generate the two utf8 or utfebcdic
337 * bytes from an ordinal that is known to fit into two bytes; it must be less
338 * than 0x3FF to work across both encodings. */
67af0a71
KW
339/* Nocast allows these to be used in the case label of a switch statement;
340 * however this doesn't won't work for ebcdic, and should be avoided. Use
341 * regen/unicode_constants instead */
342#define UTF8_TWO_BYTE_HI_nocast(c) I8_TO_NATIVE_UTF8((NATIVE_TO_UNI(c) \
ee372ee9 343 >> UTF_ACCUMULATION_SHIFT) | UTF_START_MARK(2))
67af0a71
KW
344#define UTF8_TWO_BYTE_LO_nocast(c) I8_TO_NATIVE_UTF8((NATIVE_TO_UNI(c) \
345 & UTF_CONTINUATION_MASK) \
346 | UTF_CONTINUATION_MARK)
d06134e5
KW
347
348#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
349#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
350
fca3b696
KW
351/* This name is used when the source is a single byte (input not checked).
352 * These expand identically to the TWO_BYTE versions on ASCII platforms, but
353 * use to/from LATIN1 instead of UNI, which on EBCDIC eliminates tests */
354#define UTF8_EIGHT_BIT_HI(c) I8_TO_NATIVE_UTF8((NATIVE_TO_LATIN1(c) \
ee372ee9 355 >> UTF_ACCUMULATION_SHIFT) | UTF_START_MARK(2))
fca3b696
KW
356#define UTF8_EIGHT_BIT_LO(c) I8_TO_NATIVE_UTF8((NATIVE_TO_LATIN1(c) \
357 & UTF_CONTINUATION_MASK) \
358 | UTF_CONTINUATION_MARK)
d06134e5 359
e7214ce8
KW
360/* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII
361 * as it is only in overlongs. */
362#define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1)
363
7e2040f0 364/*
e3036cf4 365 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
366 * from an earlier version of these macros in which they didn't call the
367 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
368 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
369 * no need to do it again here
7e2040f0 370 */
e3036cf4
KW
371#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
372 ? isIDFIRST(*(p)) \
373 : isIDFIRST_utf8((const U8*)p))
32636478
KW
374#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
375 ? isWORDCHAR(*(p)) \
376 : isWORDCHAR_utf8((const U8*)p))
377#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
1d72bdf6 378
89ebb4a3
JH
379#define UTF8_MAXLEN UTF8_MAXBYTES
380
8cb75cc8
KW
381/* A Unicode character can fold to up to 3 characters */
382#define UTF8_MAX_FOLD_CHAR_EXPAND 3
383
a98fe34d 384#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 385#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 386#define IN_UNI_8_BIT \
b36bf33f
KW
387 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
388 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
389
1d72bdf6 390
c76687c5
KW
391#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
392
393/* Allow first byte to be a continuation byte */
1d72bdf6 394#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
395
396/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 397#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
398
399/* expecting more bytes than were available in the string */
400#define UTF8_ALLOW_SHORT 0x0008
401
402/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
403#define UTF8_ALLOW_LONG 0x0010
404
405#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
406#define UTF8_WARN_SURROGATE 0x0040
407
408#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
409#define UTF8_WARN_NONCHAR 0x0100 /* code points */
410
411#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
412#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
413
414/* Code points which never were part of the original UTF-8 standard, the first
415 * byte of which is a FE or FF on ASCII platforms. */
416#define UTF8_DISALLOW_FE_FF 0x0800
417#define UTF8_WARN_FE_FF 0x1000
418
419#define UTF8_CHECK_ONLY 0x2000
420
421/* For backwards source compatibility. They do nothing, as the default now
422 * includes what they used to mean. The first one's meaning was to allow the
423 * just the single non-character 0xFFFF */
424#define UTF8_ALLOW_FFFF 0
425#define UTF8_ALLOW_SURROGATE 0
426
33d9abfb 427#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 428#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 429 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
430#define UTF8_ALLOW_ANY \
431 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
432#define UTF8_ALLOW_ANYUV \
433 (UTF8_ALLOW_EMPTY \
434 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
435#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
436 UTF8_ALLOW_ANYUV)
1d72bdf6 437
7131f24d
KW
438/* Surrogates, non-character code points and above-Unicode code points are
439 * problematic in some contexts. This allows code that needs to check for
440 * those to to quickly exclude the vast majority of code points it will
441 * encounter */
b96a92fb
KW
442#define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
443 FIRST_SURROGATE_UTF8_FIRST_BYTE
7131f24d 444
b96a92fb
KW
445#define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s))
446#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
7131f24d
KW
447
448/* ASCII EBCDIC I8
449 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
450 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
451 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
452 */
453#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
bc3632a8
KW
454# define UTF8_IS_SUPER(s) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \
455 && (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \
456 || (NATIVE_UTF8_TO_I8(* (U8*) ((s)) + 1 >= 0xA2))))
7131f24d 457#else
0404c92a
KW
458# define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \
459 && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
7131f24d
KW
460#endif
461
b96a92fb
KW
462/* These are now machine generated, and the 'given' clause is no longer
463 * applicable */
464#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
465 cBOOL(is_NONCHAR_utf8(s))
466#define UTF8_IS_NONCHAR_(s) \
467 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s)
7131f24d 468
c867b360
JH
469#define UNICODE_SURROGATE_FIRST 0xD800
470#define UNICODE_SURROGATE_LAST 0xDFFF
471#define UNICODE_REPLACEMENT 0xFFFD
472#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 473
b851fbc1 474/* Though our UTF-8 encoding can go beyond this,
c76687c5 475 * let's be conservative and do as Unicode says. */
b851fbc1
JH
476#define PERL_UNICODE_MAX 0x10FFFF
477
949cf498
KW
478#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
479#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
480#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
481#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
482#define UNICODE_DISALLOW_SURROGATE 0x0010
483#define UNICODE_DISALLOW_NONCHAR 0x0020
484#define UNICODE_DISALLOW_SUPER 0x0040
485#define UNICODE_DISALLOW_FE_FF 0x0080
bb88be5f
KW
486#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
487 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
488#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
489 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
949cf498
KW
490
491/* For backward source compatibility, as are now the default */
492#define UNICODE_ALLOW_SURROGATE 0
493#define UNICODE_ALLOW_SUPER 0
494#define UNICODE_ALLOW_ANY 0
b851fbc1 495
1d72bdf6
NIS
496#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
497 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 498#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 499#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
500#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
501 /* The other noncharacters end in FFFE or FFFF, which \
502 * the mask below catches both of, but beyond the last \
503 * official unicode code point, they aren't \
504 * noncharacters, since those aren't Unicode \
505 * characters at all */ \
506 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
507#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
508#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 509
f067b878
NA
510#ifdef HAS_QUAD
511# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
512#endif
3bd709b1 513
ec34087a
KW
514#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
515#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
516 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
517#define MICRO_SIGN MICRO_SIGN_NATIVE
518#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
519 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
520#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
521 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
522#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
523#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
524#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 525#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
526#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
527#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 528#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
8d64d87f
KW
529#define LATIN_SMALL_LETTER_LONG_S 0x017F
530#define KELVIN_SIGN 0x212A
531#define ANGSTROM_SIGN 0x212B
09091399 532
9e55ce06 533#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
534#define UNI_DISPLAY_BACKSLASH 0x0002
535#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
536#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 537
5cd46e1f
KW
538#define ANYOF_FOLD_SHARP_S(node, input, end) \
539 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 540 (ANYOF_NONBITMAP(node)) && \
39065660 541 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f 542 ((end) > (input) + 1) && \
d22b930b
KW
543 toFOLD((input)[0]) == 's' && \
544 toFOLD((input)[1]) == 's')
ebc501f0 545#define SHARP_S_SKIP 2
3b0fc154 546
a4f7a67c
KW
547/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
548 * log for earlier versions which gave details for these */
4d646140
KW
549
550#ifndef EBCDIC
551/* This was generated by regen/regcharclass.pl, and then moved here. The lines
552 * that generated it were then commented out. This was done solely because it
553 * takes on the order of 10 minutes to generate, and is never going to change.
554 * The EBCDIC equivalent hasn't been commented out in regcharclass.pl, so it
555 * should generate and run the correct stuff */
39a0f513
YO
556/*
557 UTF8_CHAR: Matches utf8 from 1 to 4 bytes
558
559 0x0 - 0x1FFFFF
560*/
4d646140
KW
561/*** GENERATED CODE ***/
562#define is_UTF8_CHAR_utf8_safe(s,e) \
563( ((e)-(s) > 3) ? \
564 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
565 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
566 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
567 : ( 0xE0 == ((U8*)s)[0] ) ? \
568 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
569 : ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \
570 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
571 : ( 0xF0 == ((U8*)s)[0] ) ? \
572 ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
39a0f513 573 : ( ( ( ( 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
KW
574: ((e)-(s) > 2) ? \
575 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
576 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
577 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
578 : ( 0xE0 == ((U8*)s)[0] ) ? \
579 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
39a0f513 580 : ( ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
4d646140
KW
581: ((e)-(s) > 1) ? \
582 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
39a0f513 583 : ( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) ? 2 : 0 )\
4d646140
KW
584: ((e)-(s) > 0) ? \
585 ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) \
586: 0 )
587#endif
3b0fc154
JH
588
589/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
590 * (1) allows UTF-8 encoded UTF-16 surrogates
591 * (2) it allows code points past U+10FFFF.
592 * The Perl_is_utf8_char() full "slow" code will handle the Perl
593 * "extended UTF-8". */
4d646140 594#define IS_UTF8_CHAR(p, n) (is_UTF8_CHAR_utf8_safe(p, (p) + (n)) == n)
2f454f11 595
4d646140
KW
596/* regen/regcharclass.pl generates is_UTF8_CHAR_utf8_safe() macros for up to
597 * these number of bytes. So this has to be coordinated with it */
598#ifdef EBCDIC
2f454f11 599# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
4d646140
KW
600#else
601# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
602#endif
e9a8c099 603
57f0e7e2
KW
604#endif /* H_UTF8 */
605
e9a8c099
MHM
606/*
607 * Local variables:
608 * c-indentation-style: bsd
609 * c-basic-offset: 4
14d04a33 610 * indent-tabs-mode: nil
e9a8c099
MHM
611 * End:
612 *
14d04a33 613 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 614 */