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