This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Make unicode label tests use unicode_eval.
[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
23
36bb2ab6 24#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, 1)
051a06d4
KW
25#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
26 FOLD_FLAGS_FULL, NULL)
27#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
28#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
29#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 30
fd7cb289
RGS
31/* Source backward compatibility. */
32#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
33#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
34
eda9cac1
KW
35#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
36 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 37#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 38#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
39#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
40#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 41
e6226b18
KW
42/*
43=for apidoc ibcmp_utf8
44
45This is a synonym for (! foldEQ_utf8())
46
47=cut
48*/
49#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
50 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
51
1d72bdf6
NIS
52#ifdef EBCDIC
53/* The equivalent of these macros but implementing UTF-EBCDIC
54 are in the following header file:
55 */
56
57#include "utfebcdic.h"
fd7cb289 58
d06134e5 59#else /* ! EBCDIC */
73c4f7a1
GS
60START_EXTERN_C
61
a0ed51b3 62#ifdef DOINIT
6f06b55f 63EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3
LW
641,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
651,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
661,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
671,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
681,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
691,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
702,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* scripts */
3c77ea2b 713,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* cjk etc. */
7eb168e2
KW
727,13, /* Perl extended (not official UTF-8). Up to 72bit allowed (64-bit +
73 reserved). */
a0ed51b3
LW
74};
75#else
6f06b55f 76EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
77#endif
78
73c4f7a1 79END_EXTERN_C
7e2040f0 80
1d72bdf6
NIS
81/* Native character to iso-8859-1 */
82#define NATIVE_TO_ASCII(ch) (ch)
83#define ASCII_TO_NATIVE(ch) (ch)
84/* Transform after encoding */
85#define NATIVE_TO_UTF(ch) (ch)
86#define UTF_TO_NATIVE(ch) (ch)
87/* Transforms in wide UV chars */
88#define UNI_TO_NATIVE(ch) (ch)
89#define NATIVE_TO_UNI(ch) (ch)
90/* Transforms in invariant space */
91#define NATIVE_TO_NEED(enc,ch) (ch)
92#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 93
d06134e5 94/* As there are no translations, avoid the function wrapper */
1754c1a1
NC
95#define utf8n_to_uvchr utf8n_to_uvuni
96#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 97
877d9f0d 98/*
9041c2e3 99
8c007b5a 100 The following table is from Unicode 3.2.
877d9f0d
JH
101
102 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
103
375122d7 104 U+0000..U+007F 00..7F
e1b711da 105 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 106 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 107 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 108 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 109 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 110 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 111 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
112 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
113 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
114
e1b711da 115Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
116caused by legal UTF-8 avoiding non-shortest encodings: it is technically
117possible to UTF-8-encode a single code point in different ways, but that is
118explicitly forbidden, and the shortest possible encoding should always be used
119(and that is what Perl does).
8c007b5a 120
877d9f0d
JH
121 */
122
8c007b5a
JH
123/*
124 Another way to look at it, as bits:
125
126 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
127
128 0aaaaaaa 0aaaaaaa
129 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
130 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
131 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
132
133As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 134leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
135encoded character.
136
65ab9279
TC
137Perl's extended UTF-8 means we can have start bytes up to FF.
138
8c007b5a
JH
139*/
140
c4d5f83a 141#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
d06134e5
KW
142/* Note that C0 and C1 are invalid in legal UTF8, so the lower bound of the
143 * below might ought to be C2 */
65ab9279 144#define UTF8_IS_START(c) (((U8)c) >= 0xc0)
c512ce4f
JH
145#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
146#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 147#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 148
22901f30
RGS
149#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
150#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
151
152#define UTF_CONTINUATION_MARK 0x80
153#define UTF_ACCUMULATION_SHIFT 6
154#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 155
1d68d6cd 156#ifdef HAS_QUAD
5bbb0b5a 157#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
158 (uv) < 0x800 ? 2 : \
159 (uv) < 0x10000 ? 3 : \
160 (uv) < 0x200000 ? 4 : \
161 (uv) < 0x4000000 ? 5 : \
162 (uv) < 0x80000000 ? 6 : \
9041c2e3 163 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
164#else
165/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 166#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
167 (uv) < 0x800 ? 2 : \
168 (uv) < 0x10000 ? 3 : \
169 (uv) < 0x200000 ? 4 : \
170 (uv) < 0x4000000 ? 5 : \
171 (uv) < 0x80000000 ? 6 : 7 )
172#endif
173
d06134e5
KW
174#endif /* EBCDIC vs ASCII */
175
176/* Rest of these are attributes of Unicode and perl's internals rather than the
177 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
178 * this level; the macros that some of these call may have different
179 * definitions in the two encodings */
180
181#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
182
183#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
184
2950f2a7
KW
185/* Convert a two (not one) byte utf8 character to a unicode code point value.
186 * Needs just one iteration of accumulate. Should not be used unless it is
187 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
188 * Note that the result can be larger than 255 if the input character is not
189 * downgradable */
190#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
191 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
192 NATIVE_TO_UTF(LO))
193
d06134e5
KW
194#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
195
196#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
197#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
198
199#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
200
201/* The macros in the next sets are used to generate the two utf8 or utfebcdic
202 * bytes from an ordinal that is known to fit into two bytes; it must be less
203 * than 0x3FF to work across both encodings. */
204/* Nocast allows these to be used in the case label of a switch statement */
0072151e 205#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 206#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW
207
208#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
209#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
210
211/* This name is used when the source is a single byte */
212#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
213#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
214
7e2040f0 215/*
e3036cf4 216 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
217 * from an earlier version of these macros in which they didn't call the
218 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
219 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
220 * no need to do it again here
7e2040f0 221 */
e3036cf4
KW
222#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
223 ? isIDFIRST(*(p)) \
224 : isIDFIRST_utf8((const U8*)p))
225#define isALNUM_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
226 ? isALNUM(*(p)) \
227 : isALNUM_utf8((const U8*)p))
1d72bdf6 228
7e2040f0
GS
229#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
230#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 231
89ebb4a3
JH
232#define UTF8_MAXBYTES 13
233/* How wide can a single UTF-8 encoded character become in bytes.
234 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
235 * since UTF-8 is an encoding of Unicode and given Unicode's current
236 * upper limit only four bytes is possible. Perl thinks of UTF-8
237 * as a way to encode non-negative integers in a binary format. */
238#define UTF8_MAXLEN UTF8_MAXBYTES
239
89ebb4a3
JH
240/* The maximum number of UTF-8 bytes a single Unicode character can
241 * uppercase/lowercase/fold into; this number depends on the Unicode
242 * version. An example of maximal expansion is the U+03B0 which
243 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 244 * tell these things are UnicodeData.txt, CaseFolding.txt, and
88d45ddf
KW
245 * SpecialCasing.txt. The value is 6 for strict Unicode characters, but it has
246 * to be as big as Perl allows for a single character */
247#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
3bd709b1 248
8cb75cc8
KW
249/* A Unicode character can fold to up to 3 characters */
250#define UTF8_MAX_FOLD_CHAR_EXPAND 3
251
a98fe34d 252#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 253#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 254#define IN_UNI_8_BIT \
b36bf33f
KW
255 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
256 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
257
1d72bdf6 258
c76687c5
KW
259#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
260
261/* Allow first byte to be a continuation byte */
1d72bdf6 262#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
263
264/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 265#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
266
267/* expecting more bytes than were available in the string */
268#define UTF8_ALLOW_SHORT 0x0008
269
270/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
271#define UTF8_ALLOW_LONG 0x0010
272
273#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
274#define UTF8_WARN_SURROGATE 0x0040
275
276#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
277#define UTF8_WARN_NONCHAR 0x0100 /* code points */
278
279#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
280#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
281
282/* Code points which never were part of the original UTF-8 standard, the first
283 * byte of which is a FE or FF on ASCII platforms. */
284#define UTF8_DISALLOW_FE_FF 0x0800
285#define UTF8_WARN_FE_FF 0x1000
286
287#define UTF8_CHECK_ONLY 0x2000
288
289/* For backwards source compatibility. They do nothing, as the default now
290 * includes what they used to mean. The first one's meaning was to allow the
291 * just the single non-character 0xFFFF */
292#define UTF8_ALLOW_FFFF 0
293#define UTF8_ALLOW_SURROGATE 0
294
33d9abfb 295#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 296#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 297 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
298#define UTF8_ALLOW_ANY \
299 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
300#define UTF8_ALLOW_ANYUV \
301 (UTF8_ALLOW_EMPTY \
302 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
303#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
304 UTF8_ALLOW_ANYUV)
1d72bdf6 305
7131f24d
KW
306/* Surrogates, non-character code points and above-Unicode code points are
307 * problematic in some contexts. This allows code that needs to check for
308 * those to to quickly exclude the vast majority of code points it will
309 * encounter */
310#ifdef EBCDIC
311# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
312#else
313# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
314#endif
315
316/* ASCII EBCDIC I8
317 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
318 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
319 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
320 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
321 */
322#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
323# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
880d3d38 324 && ((*((s) +1) == UTF_TO_NATIVE(0xB6)) || *((s) + 1) == UTF_TO_NATIVE(0xB7)))
7131f24d
KW
325#else
326# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
327#endif
328
329/* ASCII EBCDIC I8
330 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
331 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
332 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
333 */
334#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
335# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
336 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
337#else
338# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
339 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
340#endif
341
342/* ASCII EBCDIC I8
343 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
344 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
345 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
346 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
347 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
348 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
349 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
350 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
351 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
352 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
353 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
354 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
355 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
356 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
357 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
358 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
359 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
360 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
361 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
362 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
363 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
364 */
365#define UTF8_IS_NONCHAR_(s) ( \
366 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
367 && ! UTF8_IS_SUPER(s) \
368 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
369
370#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
371# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
372 ((*(s) == UTF_TO_NATIVE(0xF1) \
373 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
374 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
375 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
376 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
377 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
378 || (UTF8SKIP(*(s)) > 3 \
379 /* (These were all derived by inspection and experimentation with an */ \
380 /* editor) The next line checks the next to final byte in the char */ \
381 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
382 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
383 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
384 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
385#else
386# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
387 ((*(s) == 0xEF \
388 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
389 /* Gets U+FFF[EF] */ \
390 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
391 || ((*((s) + 2) == 0xBF \
392 && (*((s) + 3) & 0xBE) == 0xBE \
393 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
394 && (*((s) + 1) & 0x8F) == 0x8F)))
395#endif
396
c867b360
JH
397#define UNICODE_SURROGATE_FIRST 0xD800
398#define UNICODE_SURROGATE_LAST 0xDFFF
399#define UNICODE_REPLACEMENT 0xFFFD
400#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 401
b851fbc1 402/* Though our UTF-8 encoding can go beyond this,
c76687c5 403 * let's be conservative and do as Unicode says. */
b851fbc1
JH
404#define PERL_UNICODE_MAX 0x10FFFF
405
949cf498
KW
406#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
407#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
408#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
409#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
410#define UNICODE_DISALLOW_SURROGATE 0x0010
411#define UNICODE_DISALLOW_NONCHAR 0x0020
412#define UNICODE_DISALLOW_SUPER 0x0040
413#define UNICODE_DISALLOW_FE_FF 0x0080
414#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
415 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
416#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
417 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
418
419/* For backward source compatibility, as are now the default */
420#define UNICODE_ALLOW_SURROGATE 0
421#define UNICODE_ALLOW_SUPER 0
422#define UNICODE_ALLOW_ANY 0
b851fbc1 423
1d72bdf6
NIS
424#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
425 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 426#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 427#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
428#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
429 /* The other noncharacters end in FFFE or FFFF, which \
430 * the mask below catches both of, but beyond the last \
431 * official unicode code point, they aren't \
432 * noncharacters, since those aren't Unicode \
433 * characters at all */ \
434 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
435#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
436#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 437
f067b878
NA
438#ifdef HAS_QUAD
439# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
440#endif
3bd709b1 441
09091399
JH
442#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
443#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
444#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 445#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
446#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
447#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 448#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
09091399 449
9e55ce06 450#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
451#define UNI_DISPLAY_BACKSLASH 0x0002
452#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
453#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 454
5cd46e1f
KW
455#ifndef EBCDIC
456# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
457# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
458# define MICRO_SIGN 0x00B5
f508a607
KW
459# define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE 0x00C5
460# define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE 0x00E5
5cd46e1f
KW
461#endif
462
463#define ANYOF_FOLD_SHARP_S(node, input, end) \
464 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 465 (ANYOF_NONBITMAP(node)) && \
39065660 466 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
467 ((end) > (input) + 1) && \
468 toLOWER((input)[0]) == 's' && \
469 toLOWER((input)[1]) == 's')
ebc501f0 470#define SHARP_S_SKIP 2
3b0fc154 471
2f454f11 472#ifndef EBCDIC
c78f6c49 473# define IS_UTF8_CHAR_1(p) \
3b0fc154 474 ((p)[0] <= 0x7F)
c78f6c49 475# define IS_UTF8_CHAR_2(p) \
3b0fc154
JH
476 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
477 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
c78f6c49 478# define IS_UTF8_CHAR_3a(p) \
3b0fc154
JH
479 ((p)[0] == 0xE0 && \
480 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
481 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 482# define IS_UTF8_CHAR_3b(p) \
3b0fc154
JH
483 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
484 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
485 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 486# define IS_UTF8_CHAR_3c(p) \
3b0fc154
JH
487 ((p)[0] == 0xED && \
488 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
489 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49
KW
490 /* In IS_UTF8_CHAR_3c(p) one could use
491 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
492 * if one wanted to exclude surrogates. */
493# define IS_UTF8_CHAR_3d(p) \
3b0fc154
JH
494 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
495 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
496 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
c78f6c49 497# define IS_UTF8_CHAR_4a(p) \
3b0fc154
JH
498 ((p)[0] == 0xF0 && \
499 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
500 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
501 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
c78f6c49 502# define IS_UTF8_CHAR_4b(p) \
3b0fc154
JH
503 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
504 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
505 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
506 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
507/* In IS_UTF8_CHAR_4c(p) one could use
508 * (p)[0] == 0xF4
509 * if one wanted to stop at the Unicode limit U+10FFFF.
510 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
511 * require five bytes). Not doing any further code points
512 * since that is not needed (and that would not be strict
513 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
514 * will take care of the "extended UTF-8". */
c78f6c49 515# define IS_UTF8_CHAR_4c(p) \
a18d6e6e 516 ((p)[0] >= 0xF4 && (p)[0] <= 0xF7 && \
3b0fc154
JH
517 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
518 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
519 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
520
c78f6c49 521# define IS_UTF8_CHAR_3(p) \
3b0fc154
JH
522 (IS_UTF8_CHAR_3a(p) || \
523 IS_UTF8_CHAR_3b(p) || \
524 IS_UTF8_CHAR_3c(p) || \
525 IS_UTF8_CHAR_3d(p))
c78f6c49 526# define IS_UTF8_CHAR_4(p) \
3b0fc154
JH
527 (IS_UTF8_CHAR_4a(p) || \
528 IS_UTF8_CHAR_4b(p) || \
529 IS_UTF8_CHAR_4c(p))
530
531/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
532 * (1) allows UTF-8 encoded UTF-16 surrogates
533 * (2) it allows code points past U+10FFFF.
534 * The Perl_is_utf8_char() full "slow" code will handle the Perl
535 * "extended UTF-8". */
c78f6c49 536# define IS_UTF8_CHAR(p, n) \
3b0fc154
JH
537 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
538 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
539 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
540 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
541
c78f6c49 542# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
768c67ee 543
2f454f11
KW
544#else /* EBCDIC */
545
546/* This is an attempt to port IS_UTF8_CHAR to EBCDIC based on eyeballing.
547 * untested. If want to exclude surrogates and above-Unicode, see the
548 * definitions for UTF8_IS_SURROGATE and UTF8_IS_SUPER */
549# define IS_UTF8_CHAR_1(p) \
550 (NATIVE_TO_ASCII((p)[0]) <= 0x9F)
551# define IS_UTF8_CHAR_2(p) \
552 (NATIVE_TO_I8((p)[0]) >= 0xC5 && NATIVE_TO_I8((p)[0]) <= 0xDF && \
553 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF)
554# define IS_UTF8_CHAR_3(p) \
555 (NATIVE_TO_I8((p)[0]) == 0xE1 && NATIVE_TO_I8((p)[1]) <= 0xEF && \
556 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
557 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF)
558# define IS_UTF8_CHAR_4a(p) \
559 (NATIVE_TO_I8((p)[0]) == 0xF0 && \
560 NATIVE_TO_I8((p)[1]) >= 0xB0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
561 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
562 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
563# define IS_UTF8_CHAR_4b(p) \
564 (NATIVE_TO_I8((p)[0]) >= 0xF1 && NATIVE_TO_I8((p)[0]) <= 0xF7 && \
565 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
566 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
567 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
568# define IS_UTF8_CHAR_5a(p) \
569 (NATIVE_TO_I8((p)[0]) == 0xF8 && \
570 NATIVE_TO_I8((p)[1]) >= 0xA8 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
571 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
572 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
573 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
574# define IS_UTF8_CHAR_5b(p) \
575 (NATIVE_TO_I8((p)[0]) >= 0xF9 && NATIVE_TO_I8((p)[1]) <= 0xFB && \
576 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
577 NATIVE_TO_I8((p)[1]) >= 0xA0 && NATIVE_TO_I8((p)[1]) <= 0xBF && \
578 NATIVE_TO_I8((p)[2]) >= 0xA0 && NATIVE_TO_I8((p)[2]) <= 0xBF && \
579 NATIVE_TO_I8((p)[3]) >= 0xA0 && NATIVE_TO_I8((p)[3]) <= 0xBF)
580
581# define IS_UTF8_CHAR_4(p) \
582 (IS_UTF8_CHAR_4a(p) || \
583 IS_UTF8_CHAR_4b(p))
584# define IS_UTF8_CHAR_5(p) \
585 (IS_UTF8_CHAR_5a(p) || \
586 IS_UTF8_CHAR_5b(p))
587# define IS_UTF8_CHAR(p, n) \
588 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
589 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
590 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
591 (n) == 4 ? IS_UTF8_CHAR_4(p) : \
592 (n) == 5 ? IS_UTF8_CHAR_5(p) : 0)
593
594# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
595
77263263 596#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
597
598/*
599 * Local variables:
600 * c-indentation-style: bsd
601 * c-basic-offset: 4
602 * indent-tabs-mode: t
603 * End:
604 *
605 * ex: set ts=8 sts=4 sw=4 noet:
606 */