This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
First stab at a more reliable Glossary. This now is (again)
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
814fafa7 3 * Copyright (C) 2000, 2001, 2002, 2005 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
1d72bdf6
NIS
23#ifdef EBCDIC
24/* The equivalent of these macros but implementing UTF-EBCDIC
25 are in the following header file:
26 */
27
28#include "utfebcdic.h"
fd7cb289 29
1d72bdf6 30#else
73c4f7a1
GS
31START_EXTERN_C
32
a0ed51b3 33#ifdef DOINIT
6f06b55f 34EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3
LW
351,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
361,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
371,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
381,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
391,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
401,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus */
412,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* scripts */
3c77ea2b
GS
423,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* cjk etc. */
437,13, /* Perl extended (not UTF-8). Up to 72bit allowed (64-bit + reserved). */
a0ed51b3
LW
44};
45#else
6f06b55f 46EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
47#endif
48
73c4f7a1 49END_EXTERN_C
27da23d5 50#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
7e2040f0 51
1d72bdf6
NIS
52/* Native character to iso-8859-1 */
53#define NATIVE_TO_ASCII(ch) (ch)
54#define ASCII_TO_NATIVE(ch) (ch)
55/* Transform after encoding */
56#define NATIVE_TO_UTF(ch) (ch)
57#define UTF_TO_NATIVE(ch) (ch)
58/* Transforms in wide UV chars */
59#define UNI_TO_NATIVE(ch) (ch)
60#define NATIVE_TO_UNI(ch) (ch)
61/* Transforms in invariant space */
62#define NATIVE_TO_NEED(enc,ch) (ch)
63#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 64
2b9d42f0 65/* As there are no translations avoid the function wrapper */
1754c1a1
NC
66#define utf8n_to_uvchr utf8n_to_uvuni
67#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 68
877d9f0d 69/*
9041c2e3 70
8c007b5a 71 The following table is from Unicode 3.2.
877d9f0d
JH
72
73 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
74
375122d7
JH
75 U+0000..U+007F 00..7F
76 U+0080..U+07FF C2..DF 80..BF
77 U+0800..U+0FFF E0 A0..BF 80..BF
78 U+1000..U+CFFF E1..EC 80..BF 80..BF
79 U+D000..U+D7FF ED 80..9F 80..BF
8c007b5a 80 U+D800..U+DFFF ******* ill-formed *******
375122d7 81 U+E000..U+FFFF EE..EF 80..BF 80..BF
877d9f0d
JH
82 U+10000..U+3FFFF F0 90..BF 80..BF 80..BF
83 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
84 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
85
8c007b5a
JH
86Note the A0..BF in U+0800..U+0FFF, the 80..9F in U+D000...U+D7FF,
87the 90..BF in U+10000..U+3FFFF, and the 80...8F in U+100000..U+10FFFF.
37361303
JH
88The "gaps" are caused by legal UTF-8 avoiding non-shortest encodings:
89it is technically possible to UTF-8-encode a single code point in different
90ways, but that is explicitly forbidden, and the shortest possible encoding
91should always be used (and that is what Perl does).
8c007b5a 92
877d9f0d
JH
93 */
94
8c007b5a
JH
95/*
96 Another way to look at it, as bits:
97
98 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
99
100 0aaaaaaa 0aaaaaaa
101 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
102 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
103 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
104
105As you can see, the continuation bytes all begin with C<10>, and the
106leading bits of the start byte tell how many bytes the are in the
107encoded character.
108
109*/
110
111
c4d5f83a
NIS
112#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
113#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
114#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_ASCII(c))
c512ce4f
JH
115#define UTF8_IS_START(c) (((U8)c) >= 0xc0 && (((U8)c) <= 0xfd))
116#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
117#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 118#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 119
1d72bdf6
NIS
120#define UTF_START_MARK(len) ((len > 7) ? 0xFF : (0xFE << (7-len)))
121#define UTF_START_MASK(len) ((len >= 7) ? 0x00 : (0x1F >> (len-2)))
122
123#define UTF_CONTINUATION_MARK 0x80
124#define UTF_ACCUMULATION_SHIFT 6
125#define UTF_CONTINUATION_MASK ((U8)0x3f)
cb359b41 126#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
421a8bf2 127
1d72bdf6
NIS
128#define UTF8_EIGHT_BIT_HI(c) ((((U8)(c))>>UTF_ACCUMULATION_SHIFT)|UTF_START_MARK(2))
129#define UTF8_EIGHT_BIT_LO(c) (((((U8)(c)))&UTF_CONTINUATION_MASK)|UTF_CONTINUATION_MARK)
c512ce4f 130
1d68d6cd 131#ifdef HAS_QUAD
5bbb0b5a 132#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
133 (uv) < 0x800 ? 2 : \
134 (uv) < 0x10000 ? 3 : \
135 (uv) < 0x200000 ? 4 : \
136 (uv) < 0x4000000 ? 5 : \
137 (uv) < 0x80000000 ? 6 : \
9041c2e3 138 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
139#else
140/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 141#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
142 (uv) < 0x800 ? 2 : \
143 (uv) < 0x10000 ? 3 : \
144 (uv) < 0x200000 ? 4 : \
145 (uv) < 0x4000000 ? 5 : \
146 (uv) < 0x80000000 ? 6 : 7 )
147#endif
148
7e2040f0
GS
149/*
150 * Note: we try to be careful never to call the isXXX_utf8() functions
151 * unless we're pretty sure we've seen the beginning of a UTF-8 character
152 * (that is, the two high bits are set). Otherwise we risk loading in the
979f2922 153 * heavy-duty swash_init and swash_fetch routines unnecessarily.
7e2040f0 154 */
e1ec3a88 155#define isIDFIRST_lazy_if(p,c) ((IN_BYTES || (!c || (*((const U8*)p) < 0xc0))) \
7e2040f0 156 ? isIDFIRST(*(p)) \
e1ec3a88
AL
157 : isIDFIRST_utf8((const U8*)p))
158#define isALNUM_lazy_if(p,c) ((IN_BYTES || (!c || (*((const U8*)p) < 0xc0))) \
7e2040f0 159 ? isALNUM(*(p)) \
e1ec3a88 160 : isALNUM_utf8((const U8*)p))
1d72bdf6
NIS
161
162
163#endif /* EBCDIC vs ASCII */
164
165/* Rest of these are attributes of Unicode and perl's internals rather than the encoding */
166
7e2040f0
GS
167#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
168#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 169
89ebb4a3
JH
170#define UTF8_MAXBYTES 13
171/* How wide can a single UTF-8 encoded character become in bytes.
172 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
173 * since UTF-8 is an encoding of Unicode and given Unicode's current
174 * upper limit only four bytes is possible. Perl thinks of UTF-8
175 * as a way to encode non-negative integers in a binary format. */
176#define UTF8_MAXLEN UTF8_MAXBYTES
177
178#define UTF8_MAXLEN_UCLC 3 /* Obsolete, do not use. */
179#define UTF8_MAXLEN_UCLC_MULT 39 /* Obsolete, do not use. */
180#define UTF8_MAXLEN_FOLD 3 /* Obsolete, do not use. */
181#define UTF8_MAXLEN_FOLD_MULT 39 /* Obsolete, do not use. */
182
183/* The maximum number of UTF-8 bytes a single Unicode character can
184 * uppercase/lowercase/fold into; this number depends on the Unicode
185 * version. An example of maximal expansion is the U+03B0 which
186 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
187 * tell these things are UnicodeDatabase.txt, CaseFolding.txt, and
188 * SpecialCasing.txt. */
189#define UTF8_MAXBYTES_CASE 6
3bd709b1 190
0064a8a9
JH
191#define IN_BYTES (PL_curcop->op_private & HINT_BYTES)
192#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1d72bdf6
NIS
193
194#define UTF8_ALLOW_EMPTY 0x0001
195#define UTF8_ALLOW_CONTINUATION 0x0002
196#define UTF8_ALLOW_NON_CONTINUATION 0x0004
197#define UTF8_ALLOW_FE_FF 0x0008
198#define UTF8_ALLOW_SHORT 0x0010
199#define UTF8_ALLOW_SURROGATE 0x0020
c867b360
JH
200#define UTF8_ALLOW_FFFF 0x0040 /* Allows also FFFE. */
201#define UTF8_ALLOW_LONG 0x0080
1d72bdf6 202#define UTF8_ALLOW_ANYUV (UTF8_ALLOW_EMPTY|UTF8_ALLOW_FE_FF|\
61843245 203 UTF8_ALLOW_SURROGATE|UTF8_ALLOW_FFFF)
c867b360 204#define UTF8_ALLOW_ANY 0x00FF
e83d50c9 205#define UTF8_CHECK_ONLY 0x0200
1d72bdf6 206
c867b360
JH
207#define UNICODE_SURROGATE_FIRST 0xD800
208#define UNICODE_SURROGATE_LAST 0xDFFF
209#define UNICODE_REPLACEMENT 0xFFFD
210#define UNICODE_BYTE_ORDER_MARK 0xFEFF
211#define UNICODE_ILLEGAL 0xFFFF
1d72bdf6 212
b851fbc1 213/* Though our UTF-8 encoding can go beyond this,
872c91ae 214 * let's be conservative and do as Unicode 3.2 says. */
b851fbc1
JH
215#define PERL_UNICODE_MAX 0x10FFFF
216
217#define UNICODE_ALLOW_SURROGATE 0x0001 /* Allow UTF-16 surrogates (EVIL) */
218#define UNICODE_ALLOW_FDD0 0x0002 /* Allow the U+FDD0...U+FDEF */
c867b360
JH
219#define UNICODE_ALLOW_FFFF 0x0004 /* Allow 0xFFF[EF], 0x1FFF[EF], ... */
220#define UNICODE_ALLOW_SUPER 0x0008 /* Allow past 10xFFFF */
221#define UNICODE_ALLOW_ANY 0x000F
b851fbc1 222
1d72bdf6
NIS
223#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
224 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 225#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 226#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
1d72bdf6
NIS
227#define UNICODE_IS_ILLEGAL(c) ((c) == UNICODE_ILLEGAL)
228
f067b878
NA
229#ifdef HAS_QUAD
230# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
231#endif
3bd709b1 232
1d72bdf6 233#define UTF8_IS_ASCII(c) UTF8_IS_INVARIANT(c)
09091399 234
e0f9d4a8 235#define UNICODE_LATIN_SMALL_LETTER_SHARP_S 0x00DF
09091399
JH
236#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
237#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
238#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
239
ebc501f0
JH
240#define EBCDIC_LATIN_SMALL_LETTER_SHARP_S 0x0059
241
9e55ce06 242#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
243#define UNI_DISPLAY_BACKSLASH 0x0002
244#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
245#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 246
ebc501f0
JH
247#ifdef EBCDIC
248# define ANYOF_FOLD_SHARP_S(node, input, end) \
249 (ANYOF_BITMAP_TEST(node, EBCDIC_LATIN_SMALL_LETTER_SHARP_S) && \
250 (ANYOF_FLAGS(node) & ANYOF_UNICODE) && \
251 (ANYOF_FLAGS(node) & ANYOF_FOLD) && \
252 ((end) > (input) + 1) && \
253 toLOWER((input)[0]) == 's' && \
254 toLOWER((input)[1]) == 's')
255#else
256# define ANYOF_FOLD_SHARP_S(node, input, end) \
07b6858f
JH
257 (ANYOF_BITMAP_TEST(node, UNICODE_LATIN_SMALL_LETTER_SHARP_S) && \
258 (ANYOF_FLAGS(node) & ANYOF_UNICODE) && \
259 (ANYOF_FLAGS(node) & ANYOF_FOLD) && \
260 ((end) > (input) + 1) && \
261 toLOWER((input)[0]) == 's' && \
262 toLOWER((input)[1]) == 's')
ebc501f0
JH
263#endif
264#define SHARP_S_SKIP 2
3b0fc154 265
77263263
ST
266#ifdef EBCDIC
267/* IS_UTF8_CHAR() is not ported to EBCDIC */
268#else
3b0fc154
JH
269#define IS_UTF8_CHAR_1(p) \
270 ((p)[0] <= 0x7F)
271#define IS_UTF8_CHAR_2(p) \
272 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
273 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
274#define IS_UTF8_CHAR_3a(p) \
275 ((p)[0] == 0xE0 && \
276 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
277 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
278#define IS_UTF8_CHAR_3b(p) \
279 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
280 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
281 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
282#define IS_UTF8_CHAR_3c(p) \
283 ((p)[0] == 0xED && \
284 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
285 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
286/* In IS_UTF8_CHAR_3c(p) one could use
287 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
288 * if one wanted to exclude surrogates. */
289#define IS_UTF8_CHAR_3d(p) \
290 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
291 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
292 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
293#define IS_UTF8_CHAR_4a(p) \
294 ((p)[0] == 0xF0 && \
295 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
296 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
297 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
298#define IS_UTF8_CHAR_4b(p) \
299 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
300 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
301 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
302 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
303/* In IS_UTF8_CHAR_4c(p) one could use
304 * (p)[0] == 0xF4
305 * if one wanted to stop at the Unicode limit U+10FFFF.
306 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
307 * require five bytes). Not doing any further code points
308 * since that is not needed (and that would not be strict
309 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
310 * will take care of the "extended UTF-8". */
311#define IS_UTF8_CHAR_4c(p) \
312 ((p)[0] == 0xF4 && (p)[0] <= 0xF7 && \
313 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
314 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
315 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
316
317#define IS_UTF8_CHAR_3(p) \
318 (IS_UTF8_CHAR_3a(p) || \
319 IS_UTF8_CHAR_3b(p) || \
320 IS_UTF8_CHAR_3c(p) || \
321 IS_UTF8_CHAR_3d(p))
322#define IS_UTF8_CHAR_4(p) \
323 (IS_UTF8_CHAR_4a(p) || \
324 IS_UTF8_CHAR_4b(p) || \
325 IS_UTF8_CHAR_4c(p))
326
327/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
328 * (1) allows UTF-8 encoded UTF-16 surrogates
329 * (2) it allows code points past U+10FFFF.
330 * The Perl_is_utf8_char() full "slow" code will handle the Perl
331 * "extended UTF-8". */
332#define IS_UTF8_CHAR(p, n) \
333 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
334 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
335 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
336 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
337
768c67ee
JH
338#define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
339
77263263 340#endif /* IS_UTF8_CHAR() for UTF-8 */