This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexec.c: Reword comment
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
2eee27d7
SS
3 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009,
4 * 2010, 2011 by Larry Wall and others
a0ed51b3
LW
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
9 */
10
39e02b42 11/* Use UTF-8 as the default script encoding?
1e54db1a 12 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
13 * data (such as Latin-1) in string literals. */
14#ifdef USE_UTF8_SCRIPTS
15# define USE_UTF8_IN_NAMES (!IN_BYTES)
16#else
17# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
18#endif
19
051a06d4
KW
20/* For to_utf8_fold_flags, q.v. */
21#define FOLD_FLAGS_LOCALE 0x1
22#define FOLD_FLAGS_FULL 0x2
a0270393 23#define FOLD_FLAGS_NOMIX_ASCII 0x4
051a06d4 24
83199d38
KW
25/* For _core_swash_init(), internal core use only */
26#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
27
a0270393 28#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
051a06d4
KW
29#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
30 FOLD_FLAGS_FULL, NULL)
31#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
32#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
33#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 34
fd7cb289
RGS
35/* Source backward compatibility. */
36#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
37#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
38
eda9cac1
KW
39#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
40 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 41#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 42#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
43#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
44#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 45
e6226b18
KW
46/*
47=for apidoc ibcmp_utf8
48
49This is a synonym for (! foldEQ_utf8())
50
51=cut
52*/
53#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
54 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
55
1d72bdf6
NIS
56#ifdef EBCDIC
57/* The equivalent of these macros but implementing UTF-EBCDIC
58 are in the following header file:
59 */
60
61#include "utfebcdic.h"
fd7cb289 62
d06134e5 63#else /* ! EBCDIC */
73c4f7a1
GS
64START_EXTERN_C
65
a0ed51b3 66#ifdef DOINIT
6f06b55f 67EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
68/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
69/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
70/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
71/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
72/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
73/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
74/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
75/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
76/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
77/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
78/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
79/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
80/* 0xC0 */ 2,2, /* overlong */
81/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
82/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
83/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
84/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
85/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
86 allowed (64-bit + reserved). */
a0ed51b3
LW
87};
88#else
6f06b55f 89EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
90#endif
91
73c4f7a1 92END_EXTERN_C
7e2040f0 93
1d72bdf6
NIS
94/* Native character to iso-8859-1 */
95#define NATIVE_TO_ASCII(ch) (ch)
96#define ASCII_TO_NATIVE(ch) (ch)
97/* Transform after encoding */
98#define NATIVE_TO_UTF(ch) (ch)
99#define UTF_TO_NATIVE(ch) (ch)
100/* Transforms in wide UV chars */
101#define UNI_TO_NATIVE(ch) (ch)
102#define NATIVE_TO_UNI(ch) (ch)
103/* Transforms in invariant space */
104#define NATIVE_TO_NEED(enc,ch) (ch)
105#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 106
d06134e5 107/* As there are no translations, avoid the function wrapper */
1754c1a1 108#define utf8n_to_uvchr utf8n_to_uvuni
6dd9dce9 109#define valid_utf8_to_uvchr valid_utf8_to_uvuni
1754c1a1 110#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 111
877d9f0d 112/*
9041c2e3 113
8c007b5a 114 The following table is from Unicode 3.2.
877d9f0d
JH
115
116 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
117
375122d7 118 U+0000..U+007F 00..7F
e1b711da 119 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 120 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 121 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 122 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 123 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 124 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 125 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
126 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
127 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
128 Below are non-Unicode code points
129 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
130 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
131 U+200000: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 132
e1b711da 133Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
134caused by legal UTF-8 avoiding non-shortest encodings: it is technically
135possible to UTF-8-encode a single code point in different ways, but that is
136explicitly forbidden, and the shortest possible encoding should always be used
137(and that is what Perl does).
8c007b5a 138
877d9f0d
JH
139 */
140
8c007b5a
JH
141/*
142 Another way to look at it, as bits:
143
b2635aa8 144 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 145
b2635aa8
KW
146 0aaa aaaa 0aaa aaaa
147 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
148 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
149 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
150
151As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 152leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
153encoded character.
154
65ab9279
TC
155Perl's extended UTF-8 means we can have start bytes up to FF.
156
8c007b5a
JH
157*/
158
c4d5f83a 159#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
0447e8df 160#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
c512ce4f
JH
161#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
162#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
0ae1fa71
CH
163
164/* Masking with 0xfe allows low bit to be 0 or 1; thus this matches 0xc[23] */
165#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfe) == 0xc2)
8850bf83 166
22901f30 167#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
b2635aa8
KW
168
169/* Masks out the initial one bits in a start byte, leaving the real data ones.
170 * Doesn't work on an invariant byte */
22901f30 171#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
172
173#define UTF_CONTINUATION_MARK 0x80
174#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW
175
176/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 177#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 178
eb83ed87
KW
179/* This sets the UTF_CONTINUATION_MASK in the upper bits of a word. If a value
180 * is anded with it, and the result is non-zero, then using the original value
181 * in UTF8_ACCUMULATE will overflow, shifting bits off the left */
182#define UTF_ACCUMULATION_OVERFLOW_MASK \
183 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) - UTF_ACCUMULATION_SHIFT))
184
1d68d6cd 185#ifdef HAS_QUAD
5bbb0b5a 186#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
187 (uv) < 0x800 ? 2 : \
188 (uv) < 0x10000 ? 3 : \
189 (uv) < 0x200000 ? 4 : \
190 (uv) < 0x4000000 ? 5 : \
191 (uv) < 0x80000000 ? 6 : \
9041c2e3 192 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
193#else
194/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 195#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
196 (uv) < 0x800 ? 2 : \
197 (uv) < 0x10000 ? 3 : \
198 (uv) < 0x200000 ? 4 : \
199 (uv) < 0x4000000 ? 5 : \
200 (uv) < 0x80000000 ? 6 : 7 )
201#endif
202
d06134e5
KW
203#endif /* EBCDIC vs ASCII */
204
205/* Rest of these are attributes of Unicode and perl's internals rather than the
206 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
207 * this level; the macros that some of these call may have different
208 * definitions in the two encodings */
209
210#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
211
212#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
213
2950f2a7
KW
214/* Convert a two (not one) byte utf8 character to a unicode code point value.
215 * Needs just one iteration of accumulate. Should not be used unless it is
216 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
217 * Note that the result can be larger than 255 if the input character is not
218 * downgradable */
219#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
220 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
221 NATIVE_TO_UTF(LO))
222
d06134e5
KW
223#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
224
225#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
226#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
227
228#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
229
230/* The macros in the next sets are used to generate the two utf8 or utfebcdic
231 * bytes from an ordinal that is known to fit into two bytes; it must be less
232 * than 0x3FF to work across both encodings. */
233/* Nocast allows these to be used in the case label of a switch statement */
0072151e 234#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 235#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW
236
237#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
238#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
239
240/* This name is used when the source is a single byte */
241#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
242#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
243
7e2040f0 244/*
e3036cf4 245 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
246 * from an earlier version of these macros in which they didn't call the
247 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
248 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
249 * no need to do it again here
7e2040f0 250 */
e3036cf4
KW
251#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
252 ? isIDFIRST(*(p)) \
253 : isIDFIRST_utf8((const U8*)p))
254#define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
255 ? isALNUM(*(p)) \
256 : isALNUM_utf8((const U8*)p))
1d72bdf6 257
7e2040f0
GS
258#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
259#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 260
89ebb4a3
JH
261#define UTF8_MAXBYTES 13
262/* How wide can a single UTF-8 encoded character become in bytes.
263 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
264 * since UTF-8 is an encoding of Unicode and given Unicode's current
265 * upper limit only four bytes is possible. Perl thinks of UTF-8
266 * as a way to encode non-negative integers in a binary format. */
267#define UTF8_MAXLEN UTF8_MAXBYTES
268
89ebb4a3
JH
269/* The maximum number of UTF-8 bytes a single Unicode character can
270 * uppercase/lowercase/fold into; this number depends on the Unicode
271 * version. An example of maximal expansion is the U+03B0 which
272 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 273 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW
274 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
275 * to be as big as Perl allows for a single character */
276#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 277
8cb75cc8
KW
278/* A Unicode character can fold to up to 3 characters */
279#define UTF8_MAX_FOLD_CHAR_EXPAND 3
280
a98fe34d 281#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 282#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 283#define IN_UNI_8_BIT \
b36bf33f
KW
284 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
285 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
286
1d72bdf6 287
c76687c5
KW
288#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
289
290/* Allow first byte to be a continuation byte */
1d72bdf6 291#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
292
293/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 294#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
295
296/* expecting more bytes than were available in the string */
297#define UTF8_ALLOW_SHORT 0x0008
298
299/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
300#define UTF8_ALLOW_LONG 0x0010
301
302#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
303#define UTF8_WARN_SURROGATE 0x0040
304
305#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
306#define UTF8_WARN_NONCHAR 0x0100 /* code points */
307
308#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
309#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
310
311/* Code points which never were part of the original UTF-8 standard, the first
312 * byte of which is a FE or FF on ASCII platforms. */
313#define UTF8_DISALLOW_FE_FF 0x0800
314#define UTF8_WARN_FE_FF 0x1000
315
316#define UTF8_CHECK_ONLY 0x2000
317
318/* For backwards source compatibility. They do nothing, as the default now
319 * includes what they used to mean. The first one's meaning was to allow the
320 * just the single non-character 0xFFFF */
321#define UTF8_ALLOW_FFFF 0
322#define UTF8_ALLOW_SURROGATE 0
323
33d9abfb 324#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 325#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 326 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
327#define UTF8_ALLOW_ANY \
328 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
329#define UTF8_ALLOW_ANYUV \
330 (UTF8_ALLOW_EMPTY \
331 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
332#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
333 UTF8_ALLOW_ANYUV)
1d72bdf6 334
7131f24d
KW
335/* Surrogates, non-character code points and above-Unicode code points are
336 * problematic in some contexts. This allows code that needs to check for
337 * those to to quickly exclude the vast majority of code points it will
338 * encounter */
339#ifdef EBCDIC
340# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
341#else
342# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
343#endif
344
345/* ASCII EBCDIC I8
346 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
347 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
348 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
349 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
350 */
351#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
b2635aa8
KW
352# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
353 && ((*((s) +1) == UTF_TO_NATIVE(0xB6)) \
354 || *((s) + 1) == UTF_TO_NATIVE(0xB7)))
28936164
KW
355 /* <send> points to one beyond the end of the string that starts at <s> */
356# define UTF8_IS_REPLACEMENT(s, send) (*(s) == UTF_TO_NATIVE(0xEF) \
357 && (send - s) >= 4 \
358 && *((s) + 1) == UTF_TO_NATIVE(0xBF) \
359 && *((s) + 2) == UTF_TO_NATIVE(0xBF) \
360 && *((s) + 3) == UTF_TO_NATIVE(0xBD)
7131f24d
KW
361#else
362# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
28936164
KW
363# define UTF8_IS_REPLACEMENT(s, send) (*(s) == 0xEF \
364 && (send - s) >= 3 \
365 && *((s) + 1) == 0xBF \
366 && *((s) + 2) == 0xBD)
7131f24d
KW
367#endif
368
369/* ASCII EBCDIC I8
370 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
371 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
372 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
373 */
374#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
375# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
376 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
377#else
378# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
379 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
380#endif
381
382/* ASCII EBCDIC I8
383 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
384 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
385 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
386 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
387 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
388 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
389 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
390 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
391 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
392 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
393 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
394 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
395 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
396 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
397 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
398 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
399 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
400 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
401 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
402 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
403 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
404 */
405#define UTF8_IS_NONCHAR_(s) ( \
406 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
407 && ! UTF8_IS_SUPER(s) \
408 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
409
410#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
411# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
412 ((*(s) == UTF_TO_NATIVE(0xF1) \
413 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
414 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
415 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
416 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
417 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
418 || (UTF8SKIP(*(s)) > 3 \
419 /* (These were all derived by inspection and experimentation with an */ \
420 /* editor) The next line checks the next to final byte in the char */ \
421 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
422 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
423 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
424 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
425#else
426# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
427 ((*(s) == 0xEF \
428 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
429 /* Gets U+FFF[EF] */ \
430 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
431 || ((*((s) + 2) == 0xBF \
432 && (*((s) + 3) & 0xBE) == 0xBE \
433 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
434 && (*((s) + 1) & 0x8F) == 0x8F)))
435#endif
436
c867b360
JH
437#define UNICODE_SURROGATE_FIRST 0xD800
438#define UNICODE_SURROGATE_LAST 0xDFFF
439#define UNICODE_REPLACEMENT 0xFFFD
440#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 441
b851fbc1 442/* Though our UTF-8 encoding can go beyond this,
c76687c5 443 * let's be conservative and do as Unicode says. */
b851fbc1
JH
444#define PERL_UNICODE_MAX 0x10FFFF
445
949cf498
KW
446#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
447#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
448#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
449#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
450#define UNICODE_DISALLOW_SURROGATE 0x0010
451#define UNICODE_DISALLOW_NONCHAR 0x0020
452#define UNICODE_DISALLOW_SUPER 0x0040
453#define UNICODE_DISALLOW_FE_FF 0x0080
454#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
455 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
456#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
457 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
458
459/* For backward source compatibility, as are now the default */
460#define UNICODE_ALLOW_SURROGATE 0
461#define UNICODE_ALLOW_SUPER 0
462#define UNICODE_ALLOW_ANY 0
b851fbc1 463
1d72bdf6
NIS
464#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
465 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 466#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 467#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
468#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
469 /* The other noncharacters end in FFFE or FFFF, which \
470 * the mask below catches both of, but beyond the last \
471 * official unicode code point, they aren't \
472 * noncharacters, since those aren't Unicode \
473 * characters at all */ \
474 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
475#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
476#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 477
f067b878
NA
478#ifdef HAS_QUAD
479# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
480#endif
3bd709b1 481
09091399
JH
482#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
483#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
484#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 485#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
486#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
487#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 488#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
8d64d87f
KW
489#define LATIN_SMALL_LETTER_LONG_S 0x017F
490#define KELVIN_SIGN 0x212A
491#define ANGSTROM_SIGN 0x212B
09091399 492
9e55ce06 493#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
494#define UNI_DISPLAY_BACKSLASH 0x0002
495#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
496#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 497
5cd46e1f
KW
498#ifndef EBCDIC
499# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
500# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
501# define MICRO_SIGN 0x00B5
f508a607
KW
502# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5
503# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5
5cd46e1f
KW
504#endif
505
506#define ANYOF_FOLD_SHARP_S(node, input, end) \
507 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 508 (ANYOF_NONBITMAP(node)) && \
39065660 509 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
510 ((end) > (input) + 1) && \
511 toLOWER((input)[0]) == 's' && \
512 toLOWER((input)[1]) == 's')
ebc501f0 513#define SHARP_S_SKIP 2
3b0fc154 514
2f454f11 515#ifndef EBCDIC
a4f7a67c
KW
516/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
517 * log for earlier versions which gave details for these */
c78f6c49 518# define IS_UTF8_CHAR_1(p) \
3b0fc154 519 ((p)[0] <= 0x7F)
c78f6c49 520# define IS_UTF8_CHAR_2(p) \
3b0fc154
JH
521 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
522 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
c78f6c49 523# define IS_UTF8_CHAR_3a(p) \
3b0fc154
JH
524 ((p)[0] == 0xE0 && \
525 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
526 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 527# define IS_UTF8_CHAR_3b(p) \
a4f7a67c 528 ((p)[0] >= 0xE1 && (p)[0] <= 0xEF && \
3b0fc154
JH
529 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
530 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 531# define IS_UTF8_CHAR_4a(p) \
3b0fc154
JH
532 ((p)[0] == 0xF0 && \
533 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
534 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
535 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
a4f7a67c 536/* The 0xF7 allows us to go to 0x1fffff (0x200000 would
3b0fc154
JH
537 * require five bytes). Not doing any further code points
538 * since that is not needed (and that would not be strict
539 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
540 * will take care of the "extended UTF-8". */
a4f7a67c
KW
541# define IS_UTF8_CHAR_4b(p) \
542 ((p)[0] >= 0xF1 && (p)[0] <= 0xF7 && \
3b0fc154
JH
543 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
544 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
545 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
546
c78f6c49 547# define IS_UTF8_CHAR_3(p) \
3b0fc154 548 (IS_UTF8_CHAR_3a(p) || \
a4f7a67c 549 IS_UTF8_CHAR_3b(p))
c78f6c49 550# define IS_UTF8_CHAR_4(p) \
3b0fc154 551 (IS_UTF8_CHAR_4a(p) || \
a4f7a67c 552 IS_UTF8_CHAR_4b(p))
3b0fc154
JH
553
554/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
555 * (1) allows UTF-8 encoded UTF-16 surrogates
556 * (2) it allows code points past U+10FFFF.
557 * The Perl_is_utf8_char() full "slow" code will handle the Perl
558 * "extended UTF-8". */
c78f6c49 559# define IS_UTF8_CHAR(p, n) \
3b0fc154
JH
560 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
561 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
562 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
563 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
564
c78f6c49 565# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
768c67ee 566
2f454f11
KW
567#else /* EBCDIC */
568
569/* This is an attempt to port IS_UTF8_CHAR to EBCDIC based on eyeballing.
570 * untested. If want to exclude surrogates and above-Unicode, see the
571 * definitions for UTF8_IS_SURROGATE and UTF8_IS_SUPER */
572# define IS_UTF8_CHAR_1(p) \
573 (NATIVE_TO_ASCII((p)[0]) <= 0x9F)
574# define IS_UTF8_CHAR_2(p) \
575 (NATIVE_TO_I8((p)[0]) >= 0xC5 && NATIVE_TO_I8((p)[0]) <= 0xDF && \
576 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF)
577# define IS_UTF8_CHAR_3(p) \
578 (NATIVE_TO_I8((p)[0]) == 0xE1 && NATIVE_TO_I8((p)[1]) <= 0xEF && \
579 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
580 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF)
581# define IS_UTF8_CHAR_4a(p) \
582 (NATIVE_TO_I8((p)[0]) == 0xF0 && \
583 NATIVE_TO_I8((p)[1]) >= 0xB0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
584 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
585 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
586# define IS_UTF8_CHAR_4b(p) \
587 (NATIVE_TO_I8((p)[0]) >= 0xF1 && NATIVE_TO_I8((p)[0]) <= 0xF7 && \
588 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
589 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
590 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
591# define IS_UTF8_CHAR_5a(p) \
592 (NATIVE_TO_I8((p)[0]) == 0xF8 && \
593 NATIVE_TO_I8((p)[1]) >= 0xA8 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
594 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
595 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
596 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
597# define IS_UTF8_CHAR_5b(p) \
598 (NATIVE_TO_I8((p)[0]) >= 0xF9 && NATIVE_TO_I8((p)[1]) <= 0xFB && \
599 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
600 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
601 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
602 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
603
604# define IS_UTF8_CHAR_4(p) \
605 (IS_UTF8_CHAR_4a(p) || \
606 IS_UTF8_CHAR_4b(p))
607# define IS_UTF8_CHAR_5(p) \
608 (IS_UTF8_CHAR_5a(p) || \
609 IS_UTF8_CHAR_5b(p))
610# define IS_UTF8_CHAR(p, n) \
611 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
612 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
613 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
614 (n) == 4 ? IS_UTF8_CHAR_4(p) : \
615 (n) == 5 ? IS_UTF8_CHAR_5(p) : 0)
616
617# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
618
77263263 619#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
620
621/*
622 * Local variables:
623 * c-indentation-style: bsd
624 * c-basic-offset: 4
14d04a33 625 * indent-tabs-mode: nil
e9a8c099
MHM
626 * End:
627 *
14d04a33 628 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 629 */