This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.c: Add 'input pre-folded' flags to foldEQ_utf8_flags
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
83706693 3 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009 by Larry Wall and others
a0ed51b3
LW
4 *
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
7 *
8 */
9
39e02b42 10/* Use UTF-8 as the default script encoding?
1e54db1a 11 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
12 * data (such as Latin-1) in string literals. */
13#ifdef USE_UTF8_SCRIPTS
14# define USE_UTF8_IN_NAMES (!IN_BYTES)
15#else
16# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
17#endif
18
36bb2ab6
KW
19#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, 1)
20#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, 1)
21
fd7cb289
RGS
22/* Source backward compatibility. */
23#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
24#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
25
eda9cac1
KW
26#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
27 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 28#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 29#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
30#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
31#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 32
e6226b18
KW
33/*
34=for apidoc ibcmp_utf8
35
36This is a synonym for (! foldEQ_utf8())
37
38=cut
39*/
40#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
41 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
42
1d72bdf6
NIS
43#ifdef EBCDIC
44/* The equivalent of these macros but implementing UTF-EBCDIC
45 are in the following header file:
46 */
47
48#include "utfebcdic.h"
fd7cb289 49
d06134e5 50#else /* ! EBCDIC */
73c4f7a1
GS
51START_EXTERN_C
52
a0ed51b3 53#ifdef DOINIT
6f06b55f 54EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3
LW
551,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
561,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
571,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
581,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
591,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
601,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
612,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* scripts */
3c77ea2b
GS
623,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* cjk etc. */
637,13, /* Perl extended (not UTF-8). Up to 72bit allowed (64-bit + reserved). */
a0ed51b3
LW
64};
65#else
6f06b55f 66EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
67#endif
68
73c4f7a1 69END_EXTERN_C
7e2040f0 70
1d72bdf6
NIS
71/* Native character to iso-8859-1 */
72#define NATIVE_TO_ASCII(ch) (ch)
73#define ASCII_TO_NATIVE(ch) (ch)
74/* Transform after encoding */
75#define NATIVE_TO_UTF(ch) (ch)
76#define UTF_TO_NATIVE(ch) (ch)
77/* Transforms in wide UV chars */
78#define UNI_TO_NATIVE(ch) (ch)
79#define NATIVE_TO_UNI(ch) (ch)
80/* Transforms in invariant space */
81#define NATIVE_TO_NEED(enc,ch) (ch)
82#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 83
d06134e5 84/* As there are no translations, avoid the function wrapper */
1754c1a1
NC
85#define utf8n_to_uvchr utf8n_to_uvuni
86#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 87
877d9f0d 88/*
9041c2e3 89
8c007b5a 90 The following table is from Unicode 3.2.
877d9f0d
JH
91
92 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
93
375122d7 94 U+0000..U+007F 00..7F
e1b711da 95 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 96 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 97 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 98 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 99 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 100 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 101 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
102 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
103 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
104
e1b711da 105Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
106caused by legal UTF-8 avoiding non-shortest encodings: it is technically
107possible to UTF-8-encode a single code point in different ways, but that is
108explicitly forbidden, and the shortest possible encoding should always be used
109(and that is what Perl does).
8c007b5a 110
877d9f0d
JH
111 */
112
8c007b5a
JH
113/*
114 Another way to look at it, as bits:
115
116 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
117
118 0aaaaaaa 0aaaaaaa
119 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
120 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
121 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
122
123As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 124leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
125encoded character.
126
65ab9279
TC
127Perl's extended UTF-8 means we can have start bytes up to FF.
128
8c007b5a
JH
129*/
130
c4d5f83a 131#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
d06134e5
KW
132/* Note that C0 and C1 are invalid in legal UTF8, so the lower bound of the
133 * below might ought to be C2 */
65ab9279 134#define UTF8_IS_START(c) (((U8)c) >= 0xc0)
c512ce4f
JH
135#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
136#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 137#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 138
22901f30
RGS
139#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
140#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
141
142#define UTF_CONTINUATION_MARK 0x80
143#define UTF_ACCUMULATION_SHIFT 6
144#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 145
1d68d6cd 146#ifdef HAS_QUAD
5bbb0b5a 147#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
148 (uv) < 0x800 ? 2 : \
149 (uv) < 0x10000 ? 3 : \
150 (uv) < 0x200000 ? 4 : \
151 (uv) < 0x4000000 ? 5 : \
152 (uv) < 0x80000000 ? 6 : \
9041c2e3 153 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
154#else
155/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 156#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
157 (uv) < 0x800 ? 2 : \
158 (uv) < 0x10000 ? 3 : \
159 (uv) < 0x200000 ? 4 : \
160 (uv) < 0x4000000 ? 5 : \
161 (uv) < 0x80000000 ? 6 : 7 )
162#endif
163
d06134e5
KW
164#endif /* EBCDIC vs ASCII */
165
166/* Rest of these are attributes of Unicode and perl's internals rather than the
167 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
168 * this level; the macros that some of these call may have different
169 * definitions in the two encodings */
170
171#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
172
173#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
174
2950f2a7
KW
175/* Convert a two (not one) byte utf8 character to a unicode code point value.
176 * Needs just one iteration of accumulate. Should not be used unless it is
177 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
178 * Note that the result can be larger than 255 if the input character is not
179 * downgradable */
180#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
181 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
182 NATIVE_TO_UTF(LO))
183
d06134e5
KW
184#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
185
186#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
187#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
188
189#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
190
191/* The macros in the next sets are used to generate the two utf8 or utfebcdic
192 * bytes from an ordinal that is known to fit into two bytes; it must be less
193 * than 0x3FF to work across both encodings. */
194/* Nocast allows these to be used in the case label of a switch statement */
0072151e 195#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 196#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW
197
198#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
199#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
200
201/* This name is used when the source is a single byte */
202#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
203#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
204
7e2040f0 205/*
e3036cf4 206 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
207 * from an earlier version of these macros in which they didn't call the
208 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
209 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
210 * no need to do it again here
7e2040f0 211 */
e3036cf4
KW
212#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
213 ? isIDFIRST(*(p)) \
214 : isIDFIRST_utf8((const U8*)p))
215#define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
216 ? isALNUM(*(p)) \
217 : isALNUM_utf8((const U8*)p))
1d72bdf6 218
7e2040f0
GS
219#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
220#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 221
89ebb4a3
JH
222#define UTF8_MAXBYTES 13
223/* How wide can a single UTF-8 encoded character become in bytes.
224 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
225 * since UTF-8 is an encoding of Unicode and given Unicode's current
226 * upper limit only four bytes is possible. Perl thinks of UTF-8
227 * as a way to encode non-negative integers in a binary format. */
228#define UTF8_MAXLEN UTF8_MAXBYTES
229
89ebb4a3
JH
230/* The maximum number of UTF-8 bytes a single Unicode character can
231 * uppercase/lowercase/fold into; this number depends on the Unicode
232 * version. An example of maximal expansion is the U+03B0 which
233 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 234 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW
235 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
236 * to be as big as Perl allows for a single character */
237#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 238
8cb75cc8
KW
239/* A Unicode character can fold to up to 3 characters */
240#define UTF8_MAX_FOLD_CHAR_EXPAND 3
241
a98fe34d 242#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 243#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1863b879 244#define IN_UNI_8_BIT ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT) \
61fc5122 245 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
1d72bdf6 246
c76687c5
KW
247#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
248
249/* Allow first byte to be a continuation byte */
1d72bdf6 250#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
251
252/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 253#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
254
255/* expecting more bytes than were available in the string */
256#define UTF8_ALLOW_SHORT 0x0008
257
258/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
259#define UTF8_ALLOW_LONG 0x0010
260
261#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
262#define UTF8_WARN_SURROGATE 0x0040
263
264#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
265#define UTF8_WARN_NONCHAR 0x0100 /* code points */
266
267#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
268#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
269
270/* Code points which never were part of the original UTF-8 standard, the first
271 * byte of which is a FE or FF on ASCII platforms. */
272#define UTF8_DISALLOW_FE_FF 0x0800
273#define UTF8_WARN_FE_FF 0x1000
274
275#define UTF8_CHECK_ONLY 0x2000
276
277/* For backwards source compatibility. They do nothing, as the default now
278 * includes what they used to mean. The first one's meaning was to allow the
279 * just the single non-character 0xFFFF */
280#define UTF8_ALLOW_FFFF 0
281#define UTF8_ALLOW_SURROGATE 0
282
33d9abfb 283#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 284#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 285 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
286#define UTF8_ALLOW_ANY \
287 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
288#define UTF8_ALLOW_ANYUV \
289 (UTF8_ALLOW_EMPTY \
290 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
291#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
292 UTF8_ALLOW_ANYUV)
1d72bdf6 293
7131f24d
KW
294/* Surrogates, non-character code points and above-Unicode code points are
295 * problematic in some contexts. This allows code that needs to check for
296 * those to to quickly exclude the vast majority of code points it will
297 * encounter */
298#ifdef EBCDIC
299# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
300#else
301# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
302#endif
303
304/* ASCII EBCDIC I8
305 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
306 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
307 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
308 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
309 */
310#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
311# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
312 && (*((s) +1) == UTF_TO_NATIVE(0xB6)) || *((s) + 1) == UTF_TO_NATIVE(0xB7))
313#else
314# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
315#endif
316
317/* ASCII EBCDIC I8
318 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
319 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
320 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
321 */
322#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
323# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
324 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
325#else
326# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
327 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
328#endif
329
330/* ASCII EBCDIC I8
331 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
332 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
333 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
334 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
335 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
336 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
337 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
338 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
339 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
340 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
341 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
342 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
343 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
344 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
345 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
346 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
347 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
348 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
349 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
350 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
351 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
352 */
353#define UTF8_IS_NONCHAR_(s) ( \
354 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
355 && ! UTF8_IS_SUPER(s) \
356 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
357
358#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
359# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
360 ((*(s) == UTF_TO_NATIVE(0xF1) \
361 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
362 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
363 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
364 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
365 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
366 || (UTF8SKIP(*(s)) > 3 \
367 /* (These were all derived by inspection and experimentation with an */ \
368 /* editor) The next line checks the next to final byte in the char */ \
369 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
370 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
371 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
372 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
373#else
374# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
375 ((*(s) == 0xEF \
376 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
377 /* Gets U+FFF[EF] */ \
378 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
379 || ((*((s) + 2) == 0xBF \
380 && (*((s) + 3) & 0xBE) == 0xBE \
381 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
382 && (*((s) + 1) & 0x8F) == 0x8F)))
383#endif
384
c867b360
JH
385#define UNICODE_SURROGATE_FIRST 0xD800
386#define UNICODE_SURROGATE_LAST 0xDFFF
387#define UNICODE_REPLACEMENT 0xFFFD
388#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 389
b851fbc1 390/* Though our UTF-8 encoding can go beyond this,
c76687c5 391 * let's be conservative and do as Unicode says. */
b851fbc1
JH
392#define PERL_UNICODE_MAX 0x10FFFF
393
949cf498
KW
394#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
395#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
396#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
397#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
398#define UNICODE_DISALLOW_SURROGATE 0x0010
399#define UNICODE_DISALLOW_NONCHAR 0x0020
400#define UNICODE_DISALLOW_SUPER 0x0040
401#define UNICODE_DISALLOW_FE_FF 0x0080
402#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
403 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
404#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
405 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
406
407/* For backward source compatibility, as are now the default */
408#define UNICODE_ALLOW_SURROGATE 0
409#define UNICODE_ALLOW_SUPER 0
410#define UNICODE_ALLOW_ANY 0
b851fbc1 411
1d72bdf6
NIS
412#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
413 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 414#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 415#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
416#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
417 /* The other noncharacters end in FFFE or FFFF, which \
418 * the mask below catches both of, but beyond the last \
419 * official unicode code point, they aren't \
420 * noncharacters, since those aren't Unicode \
421 * characters at all */ \
422 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
423#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
424#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 425
f067b878
NA
426#ifdef HAS_QUAD
427# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
428#endif
3bd709b1 429
09091399
JH
430#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
431#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
432#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 433#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
434#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
435#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 436#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
09091399 437
9e55ce06 438#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
439#define UNI_DISPLAY_BACKSLASH 0x0002
440#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
441#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 442
5cd46e1f
KW
443#ifndef EBCDIC
444# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
445# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
446# define MICRO_SIGN 0x00B5
f508a607
KW
447# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5
448# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5
5cd46e1f
KW
449#endif
450
451#define ANYOF_FOLD_SHARP_S(node, input, end) \
452 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 453 (ANYOF_NONBITMAP(node)) && \
39065660 454 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
455 ((end) > (input) + 1) && \
456 toLOWER((input)[0]) == 's' && \
457 toLOWER((input)[1]) == 's')
ebc501f0 458#define SHARP_S_SKIP 2
3b0fc154 459
77263263
ST
460#ifdef EBCDIC
461/* IS_UTF8_CHAR() is not ported to EBCDIC */
462#else
3b0fc154
JH
463#define IS_UTF8_CHAR_1(p) \
464 ((p)[0] <= 0x7F)
465#define IS_UTF8_CHAR_2(p) \
466 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
467 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
468#define IS_UTF8_CHAR_3a(p) \
469 ((p)[0] == 0xE0 && \
470 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
471 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
472#define IS_UTF8_CHAR_3b(p) \
473 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
474 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
475 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
476#define IS_UTF8_CHAR_3c(p) \
477 ((p)[0] == 0xED && \
478 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
479 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
480/* In IS_UTF8_CHAR_3c(p) one could use
481 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
482 * if one wanted to exclude surrogates. */
483#define IS_UTF8_CHAR_3d(p) \
484 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
485 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
486 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
487#define IS_UTF8_CHAR_4a(p) \
488 ((p)[0] == 0xF0 && \
489 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
490 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
491 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
492#define IS_UTF8_CHAR_4b(p) \
493 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
494 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
495 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
496 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
497/* In IS_UTF8_CHAR_4c(p) one could use
498 * (p)[0] == 0xF4
499 * if one wanted to stop at the Unicode limit U+10FFFF.
500 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
501 * require five bytes). Not doing any further code points
502 * since that is not needed (and that would not be strict
503 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
504 * will take care of the "extended UTF-8". */
505#define IS_UTF8_CHAR_4c(p) \
a18d6e6e 506 ((p)[0] >= 0xF4 && (p)[0] <= 0xF7 && \
3b0fc154
JH
507 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
508 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
509 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
510
511#define IS_UTF8_CHAR_3(p) \
512 (IS_UTF8_CHAR_3a(p) || \
513 IS_UTF8_CHAR_3b(p) || \
514 IS_UTF8_CHAR_3c(p) || \
515 IS_UTF8_CHAR_3d(p))
516#define IS_UTF8_CHAR_4(p) \
517 (IS_UTF8_CHAR_4a(p) || \
518 IS_UTF8_CHAR_4b(p) || \
519 IS_UTF8_CHAR_4c(p))
520
521/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
522 * (1) allows UTF-8 encoded UTF-16 surrogates
523 * (2) it allows code points past U+10FFFF.
524 * The Perl_is_utf8_char() full "slow" code will handle the Perl
525 * "extended UTF-8". */
526#define IS_UTF8_CHAR(p, n) \
527 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
528 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
529 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
530 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
531
768c67ee
JH
532#define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
533
77263263 534#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
535
536/*
537 * Local variables:
538 * c-indentation-style: bsd
539 * c-basic-offset: 4
540 * indent-tabs-mode: t
541 * End:
542 *
543 * ex: set ts=8 sts=4 sw=4 noet:
544 */