This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
re.pm: Add /aa support
[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)
e6226b18
KW
25/*
26=for apidoc ibcmp_utf8
27
28This is a synonym for (! foldEQ_utf8())
29
30=cut
31*/
32#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
33 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
34
1d72bdf6
NIS
35#ifdef EBCDIC
36/* The equivalent of these macros but implementing UTF-EBCDIC
37 are in the following header file:
38 */
39
40#include "utfebcdic.h"
fd7cb289 41
d06134e5 42#else /* ! EBCDIC */
73c4f7a1
GS
43START_EXTERN_C
44
a0ed51b3 45#ifdef DOINIT
6f06b55f 46EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3
LW
471,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
481,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
491,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
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, /* bogus */
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, /* bogus */
532,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* scripts */
3c77ea2b
GS
543,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* cjk etc. */
557,13, /* Perl extended (not UTF-8). Up to 72bit allowed (64-bit + reserved). */
a0ed51b3
LW
56};
57#else
6f06b55f 58EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
59#endif
60
73c4f7a1 61END_EXTERN_C
7e2040f0 62
1d72bdf6
NIS
63/* Native character to iso-8859-1 */
64#define NATIVE_TO_ASCII(ch) (ch)
65#define ASCII_TO_NATIVE(ch) (ch)
66/* Transform after encoding */
67#define NATIVE_TO_UTF(ch) (ch)
68#define UTF_TO_NATIVE(ch) (ch)
69/* Transforms in wide UV chars */
70#define UNI_TO_NATIVE(ch) (ch)
71#define NATIVE_TO_UNI(ch) (ch)
72/* Transforms in invariant space */
73#define NATIVE_TO_NEED(enc,ch) (ch)
74#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 75
d06134e5 76/* As there are no translations, avoid the function wrapper */
1754c1a1
NC
77#define utf8n_to_uvchr utf8n_to_uvuni
78#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 79
877d9f0d 80/*
9041c2e3 81
8c007b5a 82 The following table is from Unicode 3.2.
877d9f0d
JH
83
84 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
85
375122d7 86 U+0000..U+007F 00..7F
e1b711da 87 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 88 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 89 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 90 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 91 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 92 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 93 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
94 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
95 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
96
e1b711da 97Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
98caused by legal UTF-8 avoiding non-shortest encodings: it is technically
99possible to UTF-8-encode a single code point in different ways, but that is
100explicitly forbidden, and the shortest possible encoding should always be used
101(and that is what Perl does).
8c007b5a 102
877d9f0d
JH
103 */
104
8c007b5a
JH
105/*
106 Another way to look at it, as bits:
107
108 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
109
110 0aaaaaaa 0aaaaaaa
111 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
112 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
113 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
114
115As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 116leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
117encoded character.
118
65ab9279
TC
119Perl's extended UTF-8 means we can have start bytes up to FF.
120
8c007b5a
JH
121*/
122
c4d5f83a 123#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
d06134e5
KW
124/* Note that C0 and C1 are invalid in legal UTF8, so the lower bound of the
125 * below might ought to be C2 */
65ab9279 126#define UTF8_IS_START(c) (((U8)c) >= 0xc0)
c512ce4f
JH
127#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
128#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 129#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 130
22901f30
RGS
131#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
132#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6
NIS
133
134#define UTF_CONTINUATION_MARK 0x80
135#define UTF_ACCUMULATION_SHIFT 6
136#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 137
1d68d6cd 138#ifdef HAS_QUAD
5bbb0b5a 139#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
140 (uv) < 0x800 ? 2 : \
141 (uv) < 0x10000 ? 3 : \
142 (uv) < 0x200000 ? 4 : \
143 (uv) < 0x4000000 ? 5 : \
144 (uv) < 0x80000000 ? 6 : \
9041c2e3 145 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
146#else
147/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 148#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
149 (uv) < 0x800 ? 2 : \
150 (uv) < 0x10000 ? 3 : \
151 (uv) < 0x200000 ? 4 : \
152 (uv) < 0x4000000 ? 5 : \
153 (uv) < 0x80000000 ? 6 : 7 )
154#endif
155
d06134e5
KW
156#endif /* EBCDIC vs ASCII */
157
158/* Rest of these are attributes of Unicode and perl's internals rather than the
159 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
160 * this level; the macros that some of these call may have different
161 * definitions in the two encodings */
162
163#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
164
165#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
166
2950f2a7
KW
167/* Convert a two (not one) byte utf8 character to a unicode code point value.
168 * Needs just one iteration of accumulate. Should not be used unless it is
169 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
170 * Note that the result can be larger than 255 if the input character is not
171 * downgradable */
172#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
173 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
174 NATIVE_TO_UTF(LO))
175
d06134e5
KW
176#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
177
178#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
179#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
180
181#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
182
183/* The macros in the next sets are used to generate the two utf8 or utfebcdic
184 * bytes from an ordinal that is known to fit into two bytes; it must be less
185 * than 0x3FF to work across both encodings. */
186/* Nocast allows these to be used in the case label of a switch statement */
0072151e 187#define UTF8_TWO_BYTE_HI_nocast(c) UTF_TO_NATIVE(((c) >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
14c28784 188#define UTF8_TWO_BYTE_LO_nocast(c) UTF_TO_NATIVE(((c) & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK)
d06134e5
KW
189
190#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
191#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
192
193/* This name is used when the source is a single byte */
194#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
195#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
196
7e2040f0
GS
197/*
198 * Note: we try to be careful never to call the isXXX_utf8() functions
d06134e5
KW
199 * unless we're pretty sure we've seen the beginning of a UTF-8 or UTFEBCDIC
200 * character. Otherwise we risk loading in the heavy-duty swash_init and
201 * swash_fetch routines unnecessarily.
7e2040f0 202 */
d06134e5 203#define isIDFIRST_lazy_if(p,c) ((IN_BYTES || (!c || ! UTF8_IS_START(*((const U8*)p)))) \
7e2040f0 204 ? isIDFIRST(*(p)) \
e1ec3a88 205 : isIDFIRST_utf8((const U8*)p))
d06134e5 206#define isALNUM_lazy_if(p,c) ((IN_BYTES || (!c || ! UTF8_IS_START(*((const U8*)p)))) \
7e2040f0 207 ? isALNUM(*(p)) \
e1ec3a88 208 : isALNUM_utf8((const U8*)p))
1d72bdf6 209
7e2040f0
GS
210#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
211#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 212
89ebb4a3
JH
213#define UTF8_MAXBYTES 13
214/* How wide can a single UTF-8 encoded character become in bytes.
215 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
216 * since UTF-8 is an encoding of Unicode and given Unicode's current
217 * upper limit only four bytes is possible. Perl thinks of UTF-8
218 * as a way to encode non-negative integers in a binary format. */
219#define UTF8_MAXLEN UTF8_MAXBYTES
220
89ebb4a3
JH
221/* The maximum number of UTF-8 bytes a single Unicode character can
222 * uppercase/lowercase/fold into; this number depends on the Unicode
223 * version. An example of maximal expansion is the U+03B0 which
224 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
d06134e5 225 * tell these things are UnicodeData.txt, CaseFolding.txt, and
89ebb4a3
JH
226 * SpecialCasing.txt. */
227#define UTF8_MAXBYTES_CASE 6
3bd709b1 228
8cb75cc8
KW
229/* A Unicode character can fold to up to 3 characters */
230#define UTF8_MAX_FOLD_CHAR_EXPAND 3
231
a98fe34d 232#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 233#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1863b879 234#define IN_UNI_8_BIT ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT) \
61fc5122 235 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
1d72bdf6 236
c76687c5
KW
237#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
238
239/* Allow first byte to be a continuation byte */
1d72bdf6 240#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
241
242/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 243#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
244
245/* expecting more bytes than were available in the string */
246#define UTF8_ALLOW_SHORT 0x0008
247
248/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
249#define UTF8_ALLOW_LONG 0x0010
250
251#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
252#define UTF8_WARN_SURROGATE 0x0040
253
254#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
255#define UTF8_WARN_NONCHAR 0x0100 /* code points */
256
257#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
258#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
259
260/* Code points which never were part of the original UTF-8 standard, the first
261 * byte of which is a FE or FF on ASCII platforms. */
262#define UTF8_DISALLOW_FE_FF 0x0800
263#define UTF8_WARN_FE_FF 0x1000
264
265#define UTF8_CHECK_ONLY 0x2000
266
267/* For backwards source compatibility. They do nothing, as the default now
268 * includes what they used to mean. The first one's meaning was to allow the
269 * just the single non-character 0xFFFF */
270#define UTF8_ALLOW_FFFF 0
271#define UTF8_ALLOW_SURROGATE 0
272
33d9abfb 273#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 274#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 275 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
276#define UTF8_ALLOW_ANY \
277 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
278#define UTF8_ALLOW_ANYUV \
279 (UTF8_ALLOW_EMPTY \
280 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
281#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
282 UTF8_ALLOW_ANYUV)
1d72bdf6 283
7131f24d
KW
284/* Surrogates, non-character code points and above-Unicode code points are
285 * problematic in some contexts. This allows code that needs to check for
286 * those to to quickly exclude the vast majority of code points it will
287 * encounter */
288#ifdef EBCDIC
289# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE UTF_TO_NATIVE(0xF1)
290#else
291# define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE 0xED
292#endif
293
294/* ASCII EBCDIC I8
295 * U+D7FF: \xED\x9F\xBF \xF1\xB5\xBF\xBF last before surrogates
296 * U+D800: \xED\xA0\x80 \xF1\xB6\xA0\xA0 1st surrogate
297 * U+DFFF: \xED\xBF\xBF \xF1\xB7\xBF\xBF final surrogate
298 * U+E000: \xEE\x80\x80 \xF1\xB8\xA0\xA0 next after surrogates
299 */
300#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
301# define UTF8_IS_SURROGATE(s) (*(s) == UTF_TO_NATIVE(0xF1) \
302 && (*((s) +1) == UTF_TO_NATIVE(0xB6)) || *((s) + 1) == UTF_TO_NATIVE(0xB7))
303#else
304# define UTF8_IS_SURROGATE(s) (*(s) == 0xED && *((s) + 1) >= 0xA0)
305#endif
306
307/* ASCII EBCDIC I8
308 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
309 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
310 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
311 */
312#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
313# define UTF8_IS_SUPER(s) (*(s) >= UTF_TO_NATIVE(0xF9) \
314 && (*(s) > UTF_TO_NATIVE(0xF9) || (*((s) + 1) >= UTF_TO_NATIVE(0xA2))))
315#else
316# define UTF8_IS_SUPER(s) (*(s) >= 0xF4 \
317 && (*(s) > 0xF4 || (*((s) + 1) >= 0x90)))
318#endif
319
320/* ASCII EBCDIC I8
321 * U+FDCF: \xEF\xB7\x8F \xF1\xBF\xAE\xAF last before non-char block
322 * U+FDD0: \xEF\xB7\x90 \xF1\xBF\xAE\xB0 first non-char in block
323 * U+FDEF: \xEF\xB7\xAF \xF1\xBF\xAF\xAF last non-char in block
324 * U+FDF0: \xEF\xB7\xB0 \xF1\xBF\xAF\xB0 first after non-char block
325 * U+FFFF: \xEF\xBF\xBF \xF1\xBF\xBF\xBF
326 * U+1FFFF: \xF0\x9F\xBF\xBF \xF3\xBF\xBF\xBF
327 * U+2FFFF: \xF0\xAF\xBF\xBF \xF5\xBF\xBF\xBF
328 * U+3FFFF: \xF0\xBF\xBF\xBF \xF7\xBF\xBF\xBF
329 * U+4FFFF: \xF1\x8F\xBF\xBF \xF8\xA9\xBF\xBF\xBF
330 * U+5FFFF: \xF1\x9F\xBF\xBF \xF8\xAB\xBF\xBF\xBF
331 * U+6FFFF: \xF1\xAF\xBF\xBF \xF8\xAD\xBF\xBF\xBF
332 * U+7FFFF: \xF1\xBF\xBF\xBF \xF8\xAF\xBF\xBF\xBF
333 * U+8FFFF: \xF2\x8F\xBF\xBF \xF8\xB1\xBF\xBF\xBF
334 * U+9FFFF: \xF2\x9F\xBF\xBF \xF8\xB3\xBF\xBF\xBF
335 * U+AFFFF: \xF2\xAF\xBF\xBF \xF8\xB5\xBF\xBF\xBF
336 * U+BFFFF: \xF2\xBF\xBF\xBF \xF8\xB7\xBF\xBF\xBF
337 * U+CFFFF: \xF3\x8F\xBF\xBF \xF8\xB9\xBF\xBF\xBF
338 * U+DFFFF: \xF3\x9F\xBF\xBF \xF8\xBB\xBF\xBF\xBF
339 * U+EFFFF: \xF3\xAF\xBF\xBF \xF8\xBD\xBF\xBF\xBF
340 * U+FFFFF: \xF3\xBF\xBF\xBF \xF8\xBF\xBF\xBF\xBF
341 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF
342 */
343#define UTF8_IS_NONCHAR_(s) ( \
344 *(s) >= UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
345 && ! UTF8_IS_SUPER(s) \
346 && UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_FIRST_PROBLEMATIC(s) \
347
348#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
349# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
350 ((*(s) == UTF_TO_NATIVE(0xF1) \
351 && (*((s) + 1) == UTF_TO_NATIVE(0xBF) \
352 && ((*((s) + 2) == UTF_TO_NATIVE(0xAE) \
353 && *((s) + 3) >= UTF_TO_NATIVE(0xB0)) \
354 || (*((s) + 2) == UTF_TO_NATIVE(0xAF) \
355 && *((s) + 3) <= UTF_TO_NATIVE(0xAF))))) \
356 || (UTF8SKIP(*(s)) > 3 \
357 /* (These were all derived by inspection and experimentation with an */ \
358 /* editor) The next line checks the next to final byte in the char */ \
359 && *((s) + UTF8SKIP(*(s)) - 2) == UTF_TO_NATIVE(0xBF) \
360 && *((s) + UTF8SKIP(*(s)) - 3) == UTF_TO_NATIVE(0xBF) \
361 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 4)) & 0x81) == 0x81 \
362 && (NATIVE_TO_UTF(*((s) + UTF8SKIP(*(s)) - 1)) & 0xBE) == 0XBE))
363#else
364# define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
365 ((*(s) == 0xEF \
366 && ((*((s) + 1) == 0xB7 && (*((s) + 2) >= 0x90 && (*((s) + 2) <= 0xAF)))\
367 /* Gets U+FFF[EF] */ \
368 || (*((s) + 1) == 0xBF && ((*((s) + 2) & 0xBE) == 0xBE)))) \
369 || ((*((s) + 2) == 0xBF \
370 && (*((s) + 3) & 0xBE) == 0xBE \
371 /* Excludes things like U+10FFE = \xF0\x90\xBF\xBE */ \
372 && (*((s) + 1) & 0x8F) == 0x8F)))
373#endif
374
c867b360
JH
375#define UNICODE_SURROGATE_FIRST 0xD800
376#define UNICODE_SURROGATE_LAST 0xDFFF
377#define UNICODE_REPLACEMENT 0xFFFD
378#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 379
b851fbc1 380/* Though our UTF-8 encoding can go beyond this,
c76687c5 381 * let's be conservative and do as Unicode says. */
b851fbc1
JH
382#define PERL_UNICODE_MAX 0x10FFFF
383
949cf498
KW
384#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
385#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
386#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
387#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
388#define UNICODE_DISALLOW_SURROGATE 0x0010
389#define UNICODE_DISALLOW_NONCHAR 0x0020
390#define UNICODE_DISALLOW_SUPER 0x0040
391#define UNICODE_DISALLOW_FE_FF 0x0080
392#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
393 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
394#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
395 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
396
397/* For backward source compatibility, as are now the default */
398#define UNICODE_ALLOW_SURROGATE 0
399#define UNICODE_ALLOW_SUPER 0
400#define UNICODE_ALLOW_ANY 0
b851fbc1 401
1d72bdf6
NIS
402#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
403 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 404#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 405#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
406#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
407 /* The other noncharacters end in FFFE or FFFF, which \
408 * the mask below catches both of, but beyond the last \
409 * official unicode code point, they aren't \
410 * noncharacters, since those aren't Unicode \
411 * characters at all */ \
412 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
413#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
414#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 415
f067b878
NA
416#ifdef HAS_QUAD
417# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
418#endif
3bd709b1 419
09091399
JH
420#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
421#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
422#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 423#define GREEK_SMALL_LETTER_MU 0x03BC
09091399 424
9e55ce06 425#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
426#define UNI_DISPLAY_BACKSLASH 0x0002
427#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
428#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 429
5cd46e1f
KW
430#ifndef EBCDIC
431# define LATIN_SMALL_LETTER_SHARP_S 0x00DF
432# define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS 0x00FF
433# define MICRO_SIGN 0x00B5
434#endif
435
436#define ANYOF_FOLD_SHARP_S(node, input, end) \
437 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
3ff7ceb3 438 (ANYOF_FLAGS(node) & ANYOF_NONBITMAP) && \
39065660 439 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
440 ((end) > (input) + 1) && \
441 toLOWER((input)[0]) == 's' && \
442 toLOWER((input)[1]) == 's')
ebc501f0 443#define SHARP_S_SKIP 2
3b0fc154 444
77263263
ST
445#ifdef EBCDIC
446/* IS_UTF8_CHAR() is not ported to EBCDIC */
447#else
3b0fc154
JH
448#define IS_UTF8_CHAR_1(p) \
449 ((p)[0] <= 0x7F)
450#define IS_UTF8_CHAR_2(p) \
451 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
452 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
453#define IS_UTF8_CHAR_3a(p) \
454 ((p)[0] == 0xE0 && \
455 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
456 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
457#define IS_UTF8_CHAR_3b(p) \
458 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
459 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
460 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
461#define IS_UTF8_CHAR_3c(p) \
462 ((p)[0] == 0xED && \
463 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
464 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
465/* In IS_UTF8_CHAR_3c(p) one could use
466 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
467 * if one wanted to exclude surrogates. */
468#define IS_UTF8_CHAR_3d(p) \
469 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
470 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
471 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
472#define IS_UTF8_CHAR_4a(p) \
473 ((p)[0] == 0xF0 && \
474 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
475 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
476 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
477#define IS_UTF8_CHAR_4b(p) \
478 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
479 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
480 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
481 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
482/* In IS_UTF8_CHAR_4c(p) one could use
483 * (p)[0] == 0xF4
484 * if one wanted to stop at the Unicode limit U+10FFFF.
485 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
486 * require five bytes). Not doing any further code points
487 * since that is not needed (and that would not be strict
488 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
489 * will take care of the "extended UTF-8". */
490#define IS_UTF8_CHAR_4c(p) \
a18d6e6e 491 ((p)[0] >= 0xF4 && (p)[0] <= 0xF7 && \
3b0fc154
JH
492 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
493 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
494 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
495
496#define IS_UTF8_CHAR_3(p) \
497 (IS_UTF8_CHAR_3a(p) || \
498 IS_UTF8_CHAR_3b(p) || \
499 IS_UTF8_CHAR_3c(p) || \
500 IS_UTF8_CHAR_3d(p))
501#define IS_UTF8_CHAR_4(p) \
502 (IS_UTF8_CHAR_4a(p) || \
503 IS_UTF8_CHAR_4b(p) || \
504 IS_UTF8_CHAR_4c(p))
505
506/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
507 * (1) allows UTF-8 encoded UTF-16 surrogates
508 * (2) it allows code points past U+10FFFF.
509 * The Perl_is_utf8_char() full "slow" code will handle the Perl
510 * "extended UTF-8". */
511#define IS_UTF8_CHAR(p, n) \
512 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
513 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
514 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
515 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
516
768c67ee
JH
517#define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
518
77263263 519#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099
MHM
520
521/*
522 * Local variables:
523 * c-indentation-style: bsd
524 * c-basic-offset: 4
525 * indent-tabs-mode: t
526 * End:
527 *
528 * ex: set ts=8 sts=4 sw=4 noet:
529 */