This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
embed.fnc: Allow NULL arg to to_utf8_case()
[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
fd7cb289
RGS
19/* Source backward compatibility. */
20#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
21#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
22
eda9cac1
KW
23#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
24 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 25#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 26#define FOLDEQ_UTF8_LOCALE (1 << 1)
a33c29bc 27
e6226b18
KW
28/*
29=for apidoc ibcmp_utf8
30
31This is a synonym for (! foldEQ_utf8())
32
33=cut
34*/
35#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
36 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
37
1d72bdf6
NIS
38#ifdef EBCDIC
39/* The equivalent of these macros but implementing UTF-EBCDIC
40 are in the following header file:
41 */
42
43#include "utfebcdic.h"
fd7cb289 44
d06134e5 45#else /* ! EBCDIC */
73c4f7a1
GS
46START_EXTERN_C
47
a0ed51b3 48#ifdef DOINIT
6f06b55f 49EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3
LW
501,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
511,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
521,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
531,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
541,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
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, /* bogus */
562,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* scripts */
3c77ea2b
GS
573,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* cjk etc. */
587,13, /* Perl extended (not UTF-8). Up to 72bit allowed (64-bit + reserved). */
a0ed51b3
LW
59};
60#else
6f06b55f 61EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
62#endif
63
73c4f7a1 64END_EXTERN_C
7e2040f0 65
1d72bdf6
NIS
66/* Native character to iso-8859-1 */
67#define NATIVE_TO_ASCII(ch) (ch)
68#define ASCII_TO_NATIVE(ch) (ch)
69/* Transform after encoding */
70#define NATIVE_TO_UTF(ch) (ch)
71#define UTF_TO_NATIVE(ch) (ch)
72/* Transforms in wide UV chars */
73#define UNI_TO_NATIVE(ch) (ch)
74#define NATIVE_TO_UNI(ch) (ch)
75/* Transforms in invariant space */
76#define NATIVE_TO_NEED(enc,ch) (ch)
77#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 78
d06134e5 79/* As there are no translations, avoid the function wrapper */
1754c1a1
NC
80#define utf8n_to_uvchr utf8n_to_uvuni
81#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 82
877d9f0d 83/*
9041c2e3 84
8c007b5a 85 The following table is from Unicode 3.2.
877d9f0d
JH
86
87 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
88
375122d7 89 U+0000..U+007F 00..7F
e1b711da 90 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 91 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 92 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 93 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 94 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 95 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 96 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
97 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
98 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
99
e1b711da 100Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
101caused by legal UTF-8 avoiding non-shortest encodings: it is technically
102possible to UTF-8-encode a single code point in different ways, but that is
103explicitly forbidden, and the shortest possible encoding should always be used
104(and that is what Perl does).
8c007b5a 105
877d9f0d
JH
106 */
107
8c007b5a
JH
108/*
109 Another way to look at it, as bits:
110
111 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
112
113 0aaaaaaa 0aaaaaaa
114 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
115 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
116 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
117
118As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 119leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
120encoded character.
121
65ab9279
TC
122Perl's extended UTF-8 means we can have start bytes up to FF.
123
8c007b5a
JH
124*/
125
c4d5f83a 126#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
d06134e5
KW
127/* Note that C0 and C1 are invalid in legal UTF8, so the lower bound of the
128 * below might ought to be C2 */
65ab9279 129#define UTF8_IS_START(c) (((U8)c) >= 0xc0)
c512ce4f
JH
130#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
131#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 132#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 133
22901f30
RGS
134#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
135#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
136
137#define UTF_CONTINUATION_MARK 0x80
138#define UTF_ACCUMULATION_SHIFT 6
139#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 140
1d68d6cd 141#ifdef HAS_QUAD
5bbb0b5a 142#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
143 (uv) < 0x800 ? 2 : \
144 (uv) < 0x10000 ? 3 : \
145 (uv) < 0x200000 ? 4 : \
146 (uv) < 0x4000000 ? 5 : \
147 (uv) < 0x80000000 ? 6 : \
9041c2e3 148 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
149#else
150/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 151#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
152 (uv) < 0x800 ? 2 : \
153 (uv) < 0x10000 ? 3 : \
154 (uv) < 0x200000 ? 4 : \
155 (uv) < 0x4000000 ? 5 : \
156 (uv) < 0x80000000 ? 6 : 7 )
157#endif
158
d06134e5
KW
159#endif /* EBCDIC vs ASCII */
160
161/* Rest of these are attributes of Unicode and perl's internals rather than the
162 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
163 * this level; the macros that some of these call may have different
164 * definitions in the two encodings */
165
166#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
167
168#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
169
2950f2a7
KW
170/* Convert a two (not one) byte utf8 character to a unicode code point value.
171 * Needs just one iteration of accumulate. Should not be used unless it is
172 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
173 * Note that the result can be larger than 255 if the input character is not
174 * downgradable */
175#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
176 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
177 NATIVE_TO_UTF(LO))
178
d06134e5
KW
179#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
180
181#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
182#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
183
184#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
185
186/* The macros in the next sets are used to generate the two utf8 or utfebcdic
187 * bytes from an ordinal that is known to fit into two bytes; it must be less
188 * than 0x3FF to work across both encodings. */
189/* Nocast allows these to be used in the case label of a switch statement */
0072151e 190#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 191#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW
192
193#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
194#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
195
196/* This name is used when the source is a single byte */
197#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
198#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
199
7e2040f0
GS
200/*
201 * Note: we try to be careful never to call the isXXX_utf8() functions
d06134e5
KW
202 * unless we're pretty sure we've seen the beginning of a UTF-8 or UTFEBCDIC
203 * character. Otherwise we risk loading in the heavy-duty swash_init and
204 * swash_fetch routines unnecessarily.
7e2040f0 205 */
d06134e5 206#define isIDFIRST_lazy_if(p,c) ((IN_BYTES || (!c || ! UTF8_IS_START(*((const U8*)p)))) \
7e2040f0 207 ? isIDFIRST(*(p)) \
e1ec3a88 208 : isIDFIRST_utf8((const U8*)p))
d06134e5 209#define isALNUM_lazy_if(p,c) ((IN_BYTES || (!c || ! UTF8_IS_START(*((const U8*)p)))) \
7e2040f0 210 ? isALNUM(*(p)) \
e1ec3a88 211 : isALNUM_utf8((const U8*)p))
1d72bdf6 212
7e2040f0
GS
213#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
214#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 215
89ebb4a3
JH
216#define UTF8_MAXBYTES 13
217/* How wide can a single UTF-8 encoded character become in bytes.
218 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
219 * since UTF-8 is an encoding of Unicode and given Unicode's current
220 * upper limit only four bytes is possible. Perl thinks of UTF-8
221 * as a way to encode non-negative integers in a binary format. */
222#define UTF8_MAXLEN UTF8_MAXBYTES
223
89ebb4a3
JH
224/* The maximum number of UTF-8 bytes a single Unicode character can
225 * uppercase/lowercase/fold into; this number depends on the Unicode
226 * version. An example of maximal expansion is the U+03B0 which
227 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 228 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW
229 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
230 * to be as big as Perl allows for a single character */
231#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 232
8cb75cc8
KW
233/* A Unicode character can fold to up to 3 characters */
234#define UTF8_MAX_FOLD_CHAR_EXPAND 3
235
a98fe34d 236#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 237#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1863b879 238#define IN_UNI_8_BIT ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT) \
61fc5122 239 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
1d72bdf6 240
c76687c5
KW
241#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
242
243/* Allow first byte to be a continuation byte */
1d72bdf6 244#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
245
246/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 247#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
248
249/* expecting more bytes than were available in the string */
250#define UTF8_ALLOW_SHORT 0x0008
251
252/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
253#define UTF8_ALLOW_LONG 0x0010
254
255#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
256#define UTF8_WARN_SURROGATE 0x0040
257
258#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
259#define UTF8_WARN_NONCHAR 0x0100 /* code points */
260
261#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
262#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
263
264/* Code points which never were part of the original UTF-8 standard, the first
265 * byte of which is a FE or FF on ASCII platforms. */
266#define UTF8_DISALLOW_FE_FF 0x0800
267#define UTF8_WARN_FE_FF 0x1000
268
269#define UTF8_CHECK_ONLY 0x2000
270
271/* For backwards source compatibility. They do nothing, as the default now
272 * includes what they used to mean. The first one's meaning was to allow the
273 * just the single non-character 0xFFFF */
274#define UTF8_ALLOW_FFFF 0
275#define UTF8_ALLOW_SURROGATE 0
276
33d9abfb 277#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 278#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 279 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
280#define UTF8_ALLOW_ANY \
281 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
282#define UTF8_ALLOW_ANYUV \
283 (UTF8_ALLOW_EMPTY \
284 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
285#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
286 UTF8_ALLOW_ANYUV)
1d72bdf6 287
7131f24d
KW
288/* Surrogates, non-character code points and above-Unicode code points are
289 * problematic in some contexts. This allows code that needs to check for
290 * those to to quickly exclude the vast majority of code points it will
291 * encounter */
292#ifdef EBCDIC
293# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
294#else
295# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
296#endif
297
298/* ASCII EBCDIC I8
299 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
300 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
301 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
302 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
303 */
304#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
305# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
306 && (*((s) +1) == UTF_TO_NATIVE(0xB6)) || *((s) + 1) == UTF_TO_NATIVE(0xB7))
307#else
308# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
309#endif
310
311/* ASCII EBCDIC I8
312 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
313 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
314 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
315 */
316#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
317# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
318 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
319#else
320# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
321 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
322#endif
323
324/* ASCII EBCDIC I8
325 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
326 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
327 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
328 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
329 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
330 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
331 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
332 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
333 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
334 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
335 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
336 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
337 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
338 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
339 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
340 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
341 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
342 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
343 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
344 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
345 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
346 */
347#define UTF8_IS_NONCHAR_(s) ( \
348 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
349 && ! UTF8_IS_SUPER(s) \
350 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
351
352#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
353# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
354 ((*(s) == UTF_TO_NATIVE(0xF1) \
355 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
356 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
357 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
358 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
359 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
360 || (UTF8SKIP(*(s)) > 3 \
361 /* (These were all derived by inspection and experimentation with an */ \
362 /* editor) The next line checks the next to final byte in the char */ \
363 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
364 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
365 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
366 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
367#else
368# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
369 ((*(s) == 0xEF \
370 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
371 /* Gets U+FFF[EF] */ \
372 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
373 || ((*((s) + 2) == 0xBF \
374 && (*((s) + 3) & 0xBE) == 0xBE \
375 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
376 && (*((s) + 1) & 0x8F) == 0x8F)))
377#endif
378
c867b360
JH
379#define UNICODE_SURROGATE_FIRST 0xD800
380#define UNICODE_SURROGATE_LAST 0xDFFF
381#define UNICODE_REPLACEMENT 0xFFFD
382#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 383
b851fbc1 384/* Though our UTF-8 encoding can go beyond this,
c76687c5 385 * let's be conservative and do as Unicode says. */
b851fbc1
JH
386#define PERL_UNICODE_MAX 0x10FFFF
387
949cf498
KW
388#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
389#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
390#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
391#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
392#define UNICODE_DISALLOW_SURROGATE 0x0010
393#define UNICODE_DISALLOW_NONCHAR 0x0020
394#define UNICODE_DISALLOW_SUPER 0x0040
395#define UNICODE_DISALLOW_FE_FF 0x0080
396#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
397 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
398#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
399 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
400
401/* For backward source compatibility, as are now the default */
402#define UNICODE_ALLOW_SURROGATE 0
403#define UNICODE_ALLOW_SUPER 0
404#define UNICODE_ALLOW_ANY 0
b851fbc1 405
1d72bdf6
NIS
406#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
407 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 408#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 409#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
410#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
411 /* The other noncharacters end in FFFE or FFFF, which \
412 * the mask below catches both of, but beyond the last \
413 * official unicode code point, they aren't \
414 * noncharacters, since those aren't Unicode \
415 * characters at all */ \
416 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
417#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
418#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 419
f067b878
NA
420#ifdef HAS_QUAD
421# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
422#endif
3bd709b1 423
09091399
JH
424#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
425#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
426#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 427#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
428#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
429#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 430#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
09091399 431
9e55ce06 432#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
433#define UNI_DISPLAY_BACKSLASH 0x0002
434#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
435#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 436
5cd46e1f
KW
437#ifndef EBCDIC
438# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
439# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
440# define MICRO_SIGN 0x00B5
f508a607
KW
441# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5
442# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5
5cd46e1f
KW
443#endif
444
445#define ANYOF_FOLD_SHARP_S(node, input, end) \
446 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 447 (ANYOF_NONBITMAP(node)) && \
39065660 448 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
449 ((end) > (input) + 1) && \
450 toLOWER((input)[0]) == 's' && \
451 toLOWER((input)[1]) == 's')
ebc501f0 452#define SHARP_S_SKIP 2
3b0fc154 453
77263263
ST
454#ifdef EBCDIC
455/* IS_UTF8_CHAR() is not ported to EBCDIC */
456#else
3b0fc154
JH
457#define IS_UTF8_CHAR_1(p) \
458 ((p)[0] <= 0x7F)
459#define IS_UTF8_CHAR_2(p) \
460 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
461 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
462#define IS_UTF8_CHAR_3a(p) \
463 ((p)[0] == 0xE0 && \
464 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
465 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
466#define IS_UTF8_CHAR_3b(p) \
467 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
468 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
469 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
470#define IS_UTF8_CHAR_3c(p) \
471 ((p)[0] == 0xED && \
472 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
473 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
474/* In IS_UTF8_CHAR_3c(p) one could use
475 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
476 * if one wanted to exclude surrogates. */
477#define IS_UTF8_CHAR_3d(p) \
478 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
479 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
480 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
481#define IS_UTF8_CHAR_4a(p) \
482 ((p)[0] == 0xF0 && \
483 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
484 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
485 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
486#define IS_UTF8_CHAR_4b(p) \
487 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
488 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
489 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
490 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
491/* In IS_UTF8_CHAR_4c(p) one could use
492 * (p)[0] == 0xF4
493 * if one wanted to stop at the Unicode limit U+10FFFF.
494 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
495 * require five bytes). Not doing any further code points
496 * since that is not needed (and that would not be strict
497 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
498 * will take care of the "extended UTF-8". */
499#define IS_UTF8_CHAR_4c(p) \
a18d6e6e 500 ((p)[0] >= 0xF4 && (p)[0] <= 0xF7 && \
3b0fc154
JH
501 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
502 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
503 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
504
505#define IS_UTF8_CHAR_3(p) \
506 (IS_UTF8_CHAR_3a(p) || \
507 IS_UTF8_CHAR_3b(p) || \
508 IS_UTF8_CHAR_3c(p) || \
509 IS_UTF8_CHAR_3d(p))
510#define IS_UTF8_CHAR_4(p) \
511 (IS_UTF8_CHAR_4a(p) || \
512 IS_UTF8_CHAR_4b(p) || \
513 IS_UTF8_CHAR_4c(p))
514
515/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
516 * (1) allows UTF-8 encoded UTF-16 surrogates
517 * (2) it allows code points past U+10FFFF.
518 * The Perl_is_utf8_char() full "slow" code will handle the Perl
519 * "extended UTF-8". */
520#define IS_UTF8_CHAR(p, n) \
521 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
522 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
523 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
524 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
525
768c67ee
JH
526#define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
527
77263263 528#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
529
530/*
531 * Local variables:
532 * c-indentation-style: bsd
533 * c-basic-offset: 4
534 * indent-tabs-mode: t
535 * End:
536 *
537 * ex: set ts=8 sts=4 sw=4 noet:
538 */