This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
charnames: Make work 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
051a06d4
KW
29/* For to_utf8_fold_flags, q.v. */
30#define FOLD_FLAGS_LOCALE 0x1
31#define FOLD_FLAGS_FULL 0x2
a0270393 32#define FOLD_FLAGS_NOMIX_ASCII 0x4
051a06d4 33
83199d38
KW
34/* For _core_swash_init(), internal core use only */
35#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
5d3d13d1 36#define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2
87367d5f 37#define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4
83199d38 38
a0270393 39#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
051a06d4
KW
40#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
41 FOLD_FLAGS_FULL, NULL)
42#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
43#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
44#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 45
fd7cb289
RGS
46/* Source backward compatibility. */
47#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
48#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
49
eda9cac1
KW
50#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
51 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 52#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 53#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
54#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
55#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 56
e6226b18
KW
57/*
58=for apidoc ibcmp_utf8
59
60This is a synonym for (! foldEQ_utf8())
61
62=cut
63*/
64#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
65 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
66
1d72bdf6
NIS
67#ifdef EBCDIC
68/* The equivalent of these macros but implementing UTF-EBCDIC
69 are in the following header file:
70 */
71
72#include "utfebcdic.h"
fd7cb289 73
d06134e5 74#else /* ! EBCDIC */
73c4f7a1
GS
75START_EXTERN_C
76
a0ed51b3 77#ifdef DOINIT
6f06b55f 78EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
79/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
80/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
81/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
82/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
83/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
84/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
85/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
86/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
87/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
88/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
89/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
90/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
91/* 0xC0 */ 2,2, /* overlong */
92/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
93/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
94/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
95/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
96/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
97 allowed (64-bit + reserved). */
a0ed51b3
LW
98};
99#else
6f06b55f 100EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
101#endif
102
73c4f7a1 103END_EXTERN_C
7e2040f0 104
b96a92fb 105#include "regcharclass.h"
ec34087a
KW
106#include "unicode_constants.h"
107
59a449d5
KW
108/* Native character to/from iso-8859-1. Are the identity functions on ASCII
109 * platforms */
110#define NATIVE_TO_LATIN1(ch) (ch)
111#define LATIN1_TO_NATIVE(ch) (ch)
112
113/* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus
114 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking
115 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them
116 * because they are 8-bit encodings that serve the same purpose in Perl, and
117 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to
118 * whichever one is applicable on the current platform */
119#define NATIVE_UTF8_TO_I8(ch) (ch)
120#define I8_TO_NATIVE_UTF8(ch) (ch)
121
1d72bdf6
NIS
122/* Transforms in wide UV chars */
123#define UNI_TO_NATIVE(ch) (ch)
124#define NATIVE_TO_UNI(ch) (ch)
125/* Transforms in invariant space */
126#define NATIVE_TO_NEED(enc,ch) (ch)
127#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 128
d06134e5 129/* As there are no translations, avoid the function wrapper */
1754c1a1 130#define utf8n_to_uvchr utf8n_to_uvuni
6dd9dce9 131#define valid_utf8_to_uvchr valid_utf8_to_uvuni
1754c1a1 132#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 133
877d9f0d 134/*
9041c2e3 135
8c007b5a 136 The following table is from Unicode 3.2.
877d9f0d
JH
137
138 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
139
375122d7 140 U+0000..U+007F 00..7F
e1b711da 141 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 142 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 143 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 144 U+D000..U+D7FF ED 80..9F 80..BF
537124e4 145 U+D800..U+DFFF ED A0..BF 80..BF (surrogates)
375122d7 146 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 147 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
148 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
149 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
150 Below are non-Unicode code points
151 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
152 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
537124e4 153 U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 154
e1b711da 155Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
156caused by legal UTF-8 avoiding non-shortest encodings: it is technically
157possible to UTF-8-encode a single code point in different ways, but that is
158explicitly forbidden, and the shortest possible encoding should always be used
15824458 159(and that is what Perl does). The non-shortest ones are called 'overlongs'.
8c007b5a 160
877d9f0d
JH
161 */
162
8c007b5a
JH
163/*
164 Another way to look at it, as bits:
165
b2635aa8 166 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 167
b2635aa8
KW
168 0aaa aaaa 0aaa aaaa
169 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
170 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
171 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
172
173As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 174leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
175encoded character.
176
65ab9279
TC
177Perl's extended UTF-8 means we can have start bytes up to FF.
178
8c007b5a
JH
179*/
180
15824458
KW
181/* Is the representation of the Unicode code point 'c' the same regardless of
182 * being encoded in UTF-8 or not? */
c4d5f83a 183#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
15824458
KW
184
185/* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is
186 * the inverse of UTF8_IS_INVARIANT */
187#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
188
189/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
190 * This doesn't catch invariants (they are single-byte). It also excludes the
191 * illegal overlong sequences that begin with C0 and C1. */
0447e8df 192#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
15824458
KW
193
194/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
195 * first byte thereof? */
e021c6e6 196#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80)
0ae1fa71 197
15824458
KW
198/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
199 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
200 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
201 * this matches 0xc[23]. */
559c7f10 202#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2)
4ab10950 203
15824458
KW
204/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
205 * represent a code point > 255? */
7b4252f4 206#define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4)
8850bf83 207
15824458
KW
208/* This defines the 1-bits that are to be in the first byte of a multi-byte
209 * UTF-8 encoded character that give the number of bytes that comprise the
210 * character.
211 * */
22901f30 212#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
b2635aa8
KW
213
214/* Masks out the initial one bits in a start byte, leaving the real data ones.
215 * Doesn't work on an invariant byte */
22901f30 216#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6 217
15824458
KW
218/* This defines the bits that are to be in the continuation bytes of a multi-byte
219 * UTF-8 encoded character that indicate it is a continuation byte. */
1d72bdf6 220#define UTF_CONTINUATION_MARK 0x80
15824458
KW
221
222/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
223 * sequence contributes to the specification of the code point. In the bit
224 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
225 * real information */
1d72bdf6 226#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW
227
228/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 229#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 230
15824458
KW
231/* If a value is anded with this, and the result is non-zero, then using the
232 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
233 * */
eb83ed87 234#define UTF_ACCUMULATION_OVERFLOW_MASK \
bb88be5f
KW
235 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
236 - UTF_ACCUMULATION_SHIFT))
eb83ed87 237
1d68d6cd 238#ifdef HAS_QUAD
5bbb0b5a 239#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
240 (uv) < 0x800 ? 2 : \
241 (uv) < 0x10000 ? 3 : \
242 (uv) < 0x200000 ? 4 : \
243 (uv) < 0x4000000 ? 5 : \
244 (uv) < 0x80000000 ? 6 : \
9041c2e3 245 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
246#else
247/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 248#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
249 (uv) < 0x800 ? 2 : \
250 (uv) < 0x10000 ? 3 : \
251 (uv) < 0x200000 ? 4 : \
252 (uv) < 0x4000000 ? 5 : \
253 (uv) < 0x80000000 ? 6 : 7 )
254#endif
255
03c76984
KW
256/* How wide can a single UTF-8 encoded character become in bytes. */
257/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
258 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
259 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
260 * non-negative integers in a binary format, even those above Unicode */
261#define UTF8_MAXBYTES 13
262
263/* The maximum number of UTF-8 bytes a single Unicode character can
264 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
265 * expansion is 3 characters. On ASCIIish platforms, the highest Unicode
266 * character occupies 4 bytes, therefore this number would be 12, but this is
267 * smaller than the maximum width a single above-Unicode character can occupy,
268 * so use that instead */
269#if UTF8_MAXBYTES < 12
270#error UTF8_MAXBYTES must be at least 12
271#endif
272
273#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
274
d06134e5
KW
275#endif /* EBCDIC vs ASCII */
276
277/* Rest of these are attributes of Unicode and perl's internals rather than the
278 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
279 * this level; the macros that some of these call may have different
280 * definitions in the two encodings */
281
59a449d5
KW
282/* In domain restricted to ASCII, these may make more sense to the reader than
283 * the ones with Latin1 in the name */
284#define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch)
285#define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch)
286
287/* More or less misleadingly-named defines, retained for back compat */
288#define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch)
289#define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch)
290#define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
291#define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch)
292#define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch)
d06134e5 293
537124e4
KW
294/* Adds a UTF8 continuation byte 'new' of information to a running total code
295 * point 'old' of all the continuation bytes so far. This is designed to be
296 * used in a loop to convert from UTF-8 to the code point represented */
bb88be5f
KW
297#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
298 | (((U8)new) & UTF_CONTINUATION_MASK))
d06134e5 299
4ab10950
KW
300/* This works in the face of malformed UTF-8. */
301#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
302 && ( (e) - (s) > 1) \
303 && UTF8_IS_CONTINUATION(*((s)+1)))
304
2950f2a7
KW
305/* Convert a two (not one) byte utf8 character to a unicode code point value.
306 * Needs just one iteration of accumulate. Should not be used unless it is
307 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
308 * Note that the result can be larger than 255 if the input character is not
309 * downgradable */
310#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
311 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
312 NATIVE_TO_UTF(LO))
313
15824458
KW
314/* How many bytes in the UTF-8 encoded character whose first (perhaps only)
315 * byte is pointed to by 's' */
d06134e5
KW
316#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
317
15824458
KW
318/* Is the byte 'c' the same character when encoded in UTF-8 as when not. This
319 * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in
320 * each for the exact same set of bit patterns. (And it works on any byte in a
321 * UTF-8 encoded string) */
bc3632a8 322#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_UTF8_TO_I8(c))
15824458 323
bc3632a8 324#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_LATIN1(c))
d06134e5
KW
325
326#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
327
328/* The macros in the next sets are used to generate the two utf8 or utfebcdic
329 * bytes from an ordinal that is known to fit into two bytes; it must be less
330 * than 0x3FF to work across both encodings. */
67af0a71
KW
331/* Nocast allows these to be used in the case label of a switch statement;
332 * however this doesn't won't work for ebcdic, and should be avoided. Use
333 * regen/unicode_constants instead */
334#define UTF8_TWO_BYTE_HI_nocast(c) I8_TO_NATIVE_UTF8((NATIVE_TO_UNI(c) \
428921e5 335 >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
67af0a71
KW
336#define UTF8_TWO_BYTE_LO_nocast(c) I8_TO_NATIVE_UTF8((NATIVE_TO_UNI(c) \
337 & UTF_CONTINUATION_MASK) \
338 | UTF_CONTINUATION_MARK)
d06134e5
KW
339
340#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
341#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
342
67af0a71
KW
343/* This name is used when the source is a single byte. For EBCDIC these could
344 * be more efficiently written; the reason is that things above 0xFF have to be
345 * special-cased, which is done by the EBCDIC version of NATIVE_TO_UNI() */
d06134e5
KW
346#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
347#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
348
7e2040f0 349/*
e3036cf4 350 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
351 * from an earlier version of these macros in which they didn't call the
352 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
353 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
354 * no need to do it again here
7e2040f0 355 */
e3036cf4
KW
356#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
357 ? isIDFIRST(*(p)) \
358 : isIDFIRST_utf8((const U8*)p))
32636478
KW
359#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
360 ? isWORDCHAR(*(p)) \
361 : isWORDCHAR_utf8((const U8*)p))
362#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
1d72bdf6 363
89ebb4a3
JH
364#define UTF8_MAXLEN UTF8_MAXBYTES
365
8cb75cc8
KW
366/* A Unicode character can fold to up to 3 characters */
367#define UTF8_MAX_FOLD_CHAR_EXPAND 3
368
a98fe34d 369#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 370#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 371#define IN_UNI_8_BIT \
b36bf33f
KW
372 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
373 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
374
1d72bdf6 375
c76687c5
KW
376#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
377
378/* Allow first byte to be a continuation byte */
1d72bdf6 379#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
380
381/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 382#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
383
384/* expecting more bytes than were available in the string */
385#define UTF8_ALLOW_SHORT 0x0008
386
387/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
388#define UTF8_ALLOW_LONG 0x0010
389
390#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
391#define UTF8_WARN_SURROGATE 0x0040
392
393#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
394#define UTF8_WARN_NONCHAR 0x0100 /* code points */
395
396#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
397#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
398
399/* Code points which never were part of the original UTF-8 standard, the first
400 * byte of which is a FE or FF on ASCII platforms. */
401#define UTF8_DISALLOW_FE_FF 0x0800
402#define UTF8_WARN_FE_FF 0x1000
403
404#define UTF8_CHECK_ONLY 0x2000
405
406/* For backwards source compatibility. They do nothing, as the default now
407 * includes what they used to mean. The first one's meaning was to allow the
408 * just the single non-character 0xFFFF */
409#define UTF8_ALLOW_FFFF 0
410#define UTF8_ALLOW_SURROGATE 0
411
33d9abfb 412#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 413#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 414 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
415#define UTF8_ALLOW_ANY \
416 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
417#define UTF8_ALLOW_ANYUV \
418 (UTF8_ALLOW_EMPTY \
419 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
420#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
421 UTF8_ALLOW_ANYUV)
1d72bdf6 422
7131f24d
KW
423/* Surrogates, non-character code points and above-Unicode code points are
424 * problematic in some contexts. This allows code that needs to check for
425 * those to to quickly exclude the vast majority of code points it will
426 * encounter */
b96a92fb
KW
427#define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
428 FIRST_SURROGATE_UTF8_FIRST_BYTE
7131f24d 429
b96a92fb
KW
430#define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s))
431#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
7131f24d
KW
432
433/* ASCII EBCDIC I8
434 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
435 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
436 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
437 */
438#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
bc3632a8
KW
439# define UTF8_IS_SUPER(s) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \
440 && (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \
441 || (NATIVE_UTF8_TO_I8(* (U8*) ((s)) + 1 >= 0xA2))))
7131f24d 442#else
0404c92a
KW
443# define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \
444 && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
7131f24d
KW
445#endif
446
b96a92fb
KW
447/* These are now machine generated, and the 'given' clause is no longer
448 * applicable */
449#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
450 cBOOL(is_NONCHAR_utf8(s))
451#define UTF8_IS_NONCHAR_(s) \
452 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s)
7131f24d 453
c867b360
JH
454#define UNICODE_SURROGATE_FIRST 0xD800
455#define UNICODE_SURROGATE_LAST 0xDFFF
456#define UNICODE_REPLACEMENT 0xFFFD
457#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 458
b851fbc1 459/* Though our UTF-8 encoding can go beyond this,
c76687c5 460 * let's be conservative and do as Unicode says. */
b851fbc1
JH
461#define PERL_UNICODE_MAX 0x10FFFF
462
949cf498
KW
463#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
464#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
465#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
466#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
467#define UNICODE_DISALLOW_SURROGATE 0x0010
468#define UNICODE_DISALLOW_NONCHAR 0x0020
469#define UNICODE_DISALLOW_SUPER 0x0040
470#define UNICODE_DISALLOW_FE_FF 0x0080
bb88be5f
KW
471#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
472 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
473#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
474 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
949cf498
KW
475
476/* For backward source compatibility, as are now the default */
477#define UNICODE_ALLOW_SURROGATE 0
478#define UNICODE_ALLOW_SUPER 0
479#define UNICODE_ALLOW_ANY 0
b851fbc1 480
1d72bdf6
NIS
481#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
482 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 483#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 484#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
485#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
486 /* The other noncharacters end in FFFE or FFFF, which \
487 * the mask below catches both of, but beyond the last \
488 * official unicode code point, they aren't \
489 * noncharacters, since those aren't Unicode \
490 * characters at all */ \
491 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
492#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
493#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 494
f067b878
NA
495#ifdef HAS_QUAD
496# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
497#endif
3bd709b1 498
ec34087a
KW
499#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
500#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
501 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
502#define MICRO_SIGN MICRO_SIGN_NATIVE
503#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
504 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
505#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
506 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
507#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
508#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
509#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 510#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
511#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
512#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 513#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
8d64d87f
KW
514#define LATIN_SMALL_LETTER_LONG_S 0x017F
515#define KELVIN_SIGN 0x212A
516#define ANGSTROM_SIGN 0x212B
09091399 517
9e55ce06 518#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
519#define UNI_DISPLAY_BACKSLASH 0x0002
520#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
521#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 522
5cd46e1f
KW
523#define ANYOF_FOLD_SHARP_S(node, input, end) \
524 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 525 (ANYOF_NONBITMAP(node)) && \
39065660 526 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f 527 ((end) > (input) + 1) && \
d22b930b
KW
528 toFOLD((input)[0]) == 's' && \
529 toFOLD((input)[1]) == 's')
ebc501f0 530#define SHARP_S_SKIP 2
3b0fc154 531
a4f7a67c
KW
532/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
533 * log for earlier versions which gave details for these */
4d646140
KW
534
535#ifndef EBCDIC
536/* This was generated by regen/regcharclass.pl, and then moved here. The lines
537 * that generated it were then commented out. This was done solely because it
538 * takes on the order of 10 minutes to generate, and is never going to change.
539 * The EBCDIC equivalent hasn't been commented out in regcharclass.pl, so it
540 * should generate and run the correct stuff */
39a0f513
YO
541/*
542 UTF8_CHAR: Matches utf8 from 1 to 4 bytes
543
544 0x0 - 0x1FFFFF
545*/
4d646140
KW
546/*** GENERATED CODE ***/
547#define is_UTF8_CHAR_utf8_safe(s,e) \
548( ((e)-(s) > 3) ? \
549 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
550 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
551 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
552 : ( 0xE0 == ((U8*)s)[0] ) ? \
553 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
554 : ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \
555 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
556 : ( 0xF0 == ((U8*)s)[0] ) ? \
557 ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
39a0f513 558 : ( ( ( ( 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
559: ((e)-(s) > 2) ? \
560 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
561 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
562 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
563 : ( 0xE0 == ((U8*)s)[0] ) ? \
564 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
39a0f513 565 : ( ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
4d646140
KW
566: ((e)-(s) > 1) ? \
567 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
39a0f513 568 : ( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) ? 2 : 0 )\
4d646140
KW
569: ((e)-(s) > 0) ? \
570 ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) \
571: 0 )
572#endif
3b0fc154
JH
573
574/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
575 * (1) allows UTF-8 encoded UTF-16 surrogates
576 * (2) it allows code points past U+10FFFF.
577 * The Perl_is_utf8_char() full "slow" code will handle the Perl
578 * "extended UTF-8". */
4d646140 579#define IS_UTF8_CHAR(p, n) (is_UTF8_CHAR_utf8_safe(p, (p) + (n)) == n)
2f454f11 580
4d646140
KW
581/* regen/regcharclass.pl generates is_UTF8_CHAR_utf8_safe() macros for up to
582 * these number of bytes. So this has to be coordinated with it */
583#ifdef EBCDIC
2f454f11 584# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
4d646140
KW
585#else
586# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
587#endif
e9a8c099 588
57f0e7e2
KW
589#endif /* H_UTF8 */
590
e9a8c099
MHM
591/*
592 * Local variables:
593 * c-indentation-style: bsd
594 * c-basic-offset: 4
14d04a33 595 * indent-tabs-mode: nil
e9a8c099
MHM
596 * End:
597 *
14d04a33 598 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 599 */