RE: [PATCH] compress 2.018
[p5sagit/p5-mst-13.2.git] / utf8.h
CommitLineData
a0ed51b3 1/* utf8.h
2 *
83706693 3 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009 by Larry Wall and others
a0ed51b3 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 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 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 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 31START_EXTERN_C
32
a0ed51b3 33#ifdef DOINIT
6f06b55f 34EXTCONST unsigned char PL_utf8skip[] = {
a0ed51b3 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 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 44};
45#else
6f06b55f 46EXTCONST unsigned char PL_utf8skip[];
a0ed51b3 47#endif
48
73c4f7a1 49END_EXTERN_C
27da23d5 50#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
7e2040f0 51
1d72bdf6 52/* Native character to iso-8859-1 */
53#define NATIVE_TO_ASCII(ch) (ch)
b3ab6785 54#define NATIVE8_TO_UNI(ch) (ch)
1d72bdf6 55#define ASCII_TO_NATIVE(ch) (ch)
56/* Transform after encoding */
57#define NATIVE_TO_UTF(ch) (ch)
58#define UTF_TO_NATIVE(ch) (ch)
59/* Transforms in wide UV chars */
60#define UNI_TO_NATIVE(ch) (ch)
61#define NATIVE_TO_UNI(ch) (ch)
62/* Transforms in invariant space */
63#define NATIVE_TO_NEED(enc,ch) (ch)
64#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 65
2b9d42f0 66/* As there are no translations avoid the function wrapper */
1754c1a1 67#define utf8n_to_uvchr utf8n_to_uvuni
68#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 69
877d9f0d 70/*
9041c2e3 71
8c007b5a 72 The following table is from Unicode 3.2.
877d9f0d 73
74 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
75
375122d7 76 U+0000..U+007F 00..7F
77 U+0080..U+07FF C2..DF 80..BF
78 U+0800..U+0FFF E0 A0..BF 80..BF
79 U+1000..U+CFFF E1..EC 80..BF 80..BF
80 U+D000..U+D7FF ED 80..9F 80..BF
8c007b5a 81 U+D800..U+DFFF ******* ill-formed *******
375122d7 82 U+E000..U+FFFF EE..EF 80..BF 80..BF
877d9f0d 83 U+10000..U+3FFFF F0 90..BF 80..BF 80..BF
84 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
85 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
86
8c007b5a 87Note the A0..BF in U+0800..U+0FFF, the 80..9F in U+D000...U+D7FF,
88the 90..BF in U+10000..U+3FFFF, and the 80...8F in U+100000..U+10FFFF.
37361303 89The "gaps" are caused by legal UTF-8 avoiding non-shortest encodings:
90it is technically possible to UTF-8-encode a single code point in different
91ways, but that is explicitly forbidden, and the shortest possible encoding
92should always be used (and that is what Perl does).
8c007b5a 93
877d9f0d 94 */
95
8c007b5a 96/*
97 Another way to look at it, as bits:
98
99 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
100
101 0aaaaaaa 0aaaaaaa
102 00000bbbbbaaaaaa 110bbbbb 10aaaaaa
103 ccccbbbbbbaaaaaa 1110cccc 10bbbbbb 10aaaaaa
104 00000dddccccccbbbbbbaaaaaa 11110ddd 10cccccc 10bbbbbb 10aaaaaa
105
106As you can see, the continuation bytes all begin with C<10>, and the
107leading bits of the start byte tell how many bytes the are in the
108encoded character.
109
110*/
111
112
c4d5f83a 113#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
114#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
115#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_ASCII(c))
c512ce4f 116#define UTF8_IS_START(c) (((U8)c) >= 0xc0 && (((U8)c) <= 0xfd))
117#define UTF8_IS_CONTINUATION(c) (((U8)c) >= 0x80 && (((U8)c) <= 0xbf))
118#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
db42d148 119#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)c & 0xfc) == 0xc0)
8850bf83 120
22901f30 121#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
122#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6 123
124#define UTF_CONTINUATION_MARK 0x80
125#define UTF_ACCUMULATION_SHIFT 6
126#define UTF_CONTINUATION_MASK ((U8)0x3f)
cb359b41 127#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) | (((U8)new) & UTF_CONTINUATION_MASK))
421a8bf2 128
1d72bdf6 129#define UTF8_EIGHT_BIT_HI(c) ((((U8)(c))>>UTF_ACCUMULATION_SHIFT)|UTF_START_MARK(2))
130#define UTF8_EIGHT_BIT_LO(c) (((((U8)(c)))&UTF_CONTINUATION_MASK)|UTF_CONTINUATION_MARK)
c512ce4f 131
1d68d6cd 132#ifdef HAS_QUAD
5bbb0b5a 133#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd 134 (uv) < 0x800 ? 2 : \
135 (uv) < 0x10000 ? 3 : \
136 (uv) < 0x200000 ? 4 : \
137 (uv) < 0x4000000 ? 5 : \
138 (uv) < 0x80000000 ? 6 : \
9041c2e3 139 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd 140#else
141/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 142#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd 143 (uv) < 0x800 ? 2 : \
144 (uv) < 0x10000 ? 3 : \
145 (uv) < 0x200000 ? 4 : \
146 (uv) < 0x4000000 ? 5 : \
147 (uv) < 0x80000000 ? 6 : 7 )
148#endif
149
7e2040f0 150/*
151 * Note: we try to be careful never to call the isXXX_utf8() functions
152 * unless we're pretty sure we've seen the beginning of a UTF-8 character
153 * (that is, the two high bits are set). Otherwise we risk loading in the
979f2922 154 * heavy-duty swash_init and swash_fetch routines unnecessarily.
7e2040f0 155 */
e1ec3a88 156#define isIDFIRST_lazy_if(p,c) ((IN_BYTES || (!c || (*((const U8*)p) < 0xc0))) \
7e2040f0 157 ? isIDFIRST(*(p)) \
e1ec3a88 158 : isIDFIRST_utf8((const U8*)p))
159#define isALNUM_lazy_if(p,c) ((IN_BYTES || (!c || (*((const U8*)p) < 0xc0))) \
7e2040f0 160 ? isALNUM(*(p)) \
e1ec3a88 161 : isALNUM_utf8((const U8*)p))
1d72bdf6 162
163
164#endif /* EBCDIC vs ASCII */
165
166/* Rest of these are attributes of Unicode and perl's internals rather than the encoding */
167
7e2040f0 168#define isIDFIRST_lazy(p) isIDFIRST_lazy_if(p,1)
169#define isALNUM_lazy(p) isALNUM_lazy_if(p,1)
3bd709b1 170
89ebb4a3 171#define UTF8_MAXBYTES 13
172/* How wide can a single UTF-8 encoded character become in bytes.
173 * NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8
174 * since UTF-8 is an encoding of Unicode and given Unicode's current
175 * upper limit only four bytes is possible. Perl thinks of UTF-8
176 * as a way to encode non-negative integers in a binary format. */
177#define UTF8_MAXLEN UTF8_MAXBYTES
178
179#define UTF8_MAXLEN_UCLC 3 /* Obsolete, do not use. */
180#define UTF8_MAXLEN_UCLC_MULT 39 /* Obsolete, do not use. */
181#define UTF8_MAXLEN_FOLD 3 /* Obsolete, do not use. */
182#define UTF8_MAXLEN_FOLD_MULT 39 /* Obsolete, do not use. */
183
184/* The maximum number of UTF-8 bytes a single Unicode character can
185 * uppercase/lowercase/fold into; this number depends on the Unicode
186 * version. An example of maximal expansion is the U+03B0 which
187 * uppercases to U+03C5 U+0308 U+0301. The Unicode databases that
188 * tell these things are UnicodeDatabase.txt, CaseFolding.txt, and
189 * SpecialCasing.txt. */
190#define UTF8_MAXBYTES_CASE 6
3bd709b1 191
a98fe34d 192#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 193#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
1d72bdf6 194
195#define UTF8_ALLOW_EMPTY 0x0001
196#define UTF8_ALLOW_CONTINUATION 0x0002
197#define UTF8_ALLOW_NON_CONTINUATION 0x0004
9f7f3913 198#define UTF8_ALLOW_FE_FF 0x0008 /* Allow above 0x7fffFFFF */
1d72bdf6 199#define UTF8_ALLOW_SHORT 0x0010
200#define UTF8_ALLOW_SURROGATE 0x0020
9f7f3913 201#define UTF8_ALLOW_FFFF 0x0040 /* Allow UNICODE_ILLEGAL */
c867b360 202#define UTF8_ALLOW_LONG 0x0080
1d72bdf6 203#define UTF8_ALLOW_ANYUV (UTF8_ALLOW_EMPTY|UTF8_ALLOW_FE_FF|\
61843245 204 UTF8_ALLOW_SURROGATE|UTF8_ALLOW_FFFF)
c867b360 205#define UTF8_ALLOW_ANY 0x00FF
e83d50c9 206#define UTF8_CHECK_ONLY 0x0200
9f7f3913 207#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
208 UTF8_ALLOW_ANYUV)
1d72bdf6 209
c867b360 210#define UNICODE_SURROGATE_FIRST 0xD800
211#define UNICODE_SURROGATE_LAST 0xDFFF
212#define UNICODE_REPLACEMENT 0xFFFD
213#define UNICODE_BYTE_ORDER_MARK 0xFEFF
214#define UNICODE_ILLEGAL 0xFFFF
1d72bdf6 215
b851fbc1 216/* Though our UTF-8 encoding can go beyond this,
b3ab6785 217 * let's be conservative and do as Unicode 5.1 says. */
b851fbc1 218#define PERL_UNICODE_MAX 0x10FFFF
219
220#define UNICODE_ALLOW_SURROGATE 0x0001 /* Allow UTF-16 surrogates (EVIL) */
221#define UNICODE_ALLOW_FDD0 0x0002 /* Allow the U+FDD0...U+FDEF */
9f7f3913 222#define UNICODE_ALLOW_FFFF 0x0004 /* Allow U+FFF[EF], U+1FFF[EF], ... */
223#define UNICODE_ALLOW_SUPER 0x0008 /* Allow past 0x10FFFF */
c867b360 224#define UNICODE_ALLOW_ANY 0x000F
b851fbc1 225
1d72bdf6 226#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
227 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 228#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 229#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
1d72bdf6 230#define UNICODE_IS_ILLEGAL(c) ((c) == UNICODE_ILLEGAL)
231
f067b878 232#ifdef HAS_QUAD
233# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
234#endif
3bd709b1 235
1d72bdf6 236#define UTF8_IS_ASCII(c) UTF8_IS_INVARIANT(c)
09091399 237
e0f9d4a8 238#define UNICODE_LATIN_SMALL_LETTER_SHARP_S 0x00DF
09091399 239#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
240#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
241#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
242
ebc501f0 243#define EBCDIC_LATIN_SMALL_LETTER_SHARP_S 0x0059
244
9e55ce06 245#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41 246#define UNI_DISPLAY_BACKSLASH 0x0002
247#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
248#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 249
ebc501f0 250#ifdef EBCDIC
251# define ANYOF_FOLD_SHARP_S(node, input, end) \
252 (ANYOF_BITMAP_TEST(node, EBCDIC_LATIN_SMALL_LETTER_SHARP_S) && \
253 (ANYOF_FLAGS(node) & ANYOF_UNICODE) && \
254 (ANYOF_FLAGS(node) & ANYOF_FOLD) && \
255 ((end) > (input) + 1) && \
256 toLOWER((input)[0]) == 's' && \
257 toLOWER((input)[1]) == 's')
258#else
259# define ANYOF_FOLD_SHARP_S(node, input, end) \
07b6858f 260 (ANYOF_BITMAP_TEST(node, UNICODE_LATIN_SMALL_LETTER_SHARP_S) && \
261 (ANYOF_FLAGS(node) & ANYOF_UNICODE) && \
262 (ANYOF_FLAGS(node) & ANYOF_FOLD) && \
263 ((end) > (input) + 1) && \
264 toLOWER((input)[0]) == 's' && \
265 toLOWER((input)[1]) == 's')
ebc501f0 266#endif
267#define SHARP_S_SKIP 2
3b0fc154 268
77263263 269#ifdef EBCDIC
270/* IS_UTF8_CHAR() is not ported to EBCDIC */
271#else
3b0fc154 272#define IS_UTF8_CHAR_1(p) \
273 ((p)[0] <= 0x7F)
274#define IS_UTF8_CHAR_2(p) \
275 ((p)[0] >= 0xC2 && (p)[0] <= 0xDF && \
276 (p)[1] >= 0x80 && (p)[1] <= 0xBF)
277#define IS_UTF8_CHAR_3a(p) \
278 ((p)[0] == 0xE0 && \
279 (p)[1] >= 0xA0 && (p)[1] <= 0xBF && \
280 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
281#define IS_UTF8_CHAR_3b(p) \
282 ((p)[0] >= 0xE1 && (p)[0] <= 0xEC && \
283 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
284 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
285#define IS_UTF8_CHAR_3c(p) \
286 ((p)[0] == 0xED && \
287 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
288 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
289/* In IS_UTF8_CHAR_3c(p) one could use
290 * (p)[1] >= 0x80 && (p)[1] <= 0x9F
291 * if one wanted to exclude surrogates. */
292#define IS_UTF8_CHAR_3d(p) \
293 ((p)[0] >= 0xEE && (p)[0] <= 0xEF && \
294 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
295 (p)[2] >= 0x80 && (p)[2] <= 0xBF)
296#define IS_UTF8_CHAR_4a(p) \
297 ((p)[0] == 0xF0 && \
298 (p)[1] >= 0x90 && (p)[1] <= 0xBF && \
299 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
300 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
301#define IS_UTF8_CHAR_4b(p) \
302 ((p)[0] >= 0xF1 && (p)[0] <= 0xF3 && \
303 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
304 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
305 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
306/* In IS_UTF8_CHAR_4c(p) one could use
307 * (p)[0] == 0xF4
308 * if one wanted to stop at the Unicode limit U+10FFFF.
309 * The 0xF7 allows us to go to 0x1fffff (0x200000 would
310 * require five bytes). Not doing any further code points
311 * since that is not needed (and that would not be strict
312 * UTF-8, anyway). The "slow path" in Perl_is_utf8_char()
313 * will take care of the "extended UTF-8". */
314#define IS_UTF8_CHAR_4c(p) \
315 ((p)[0] == 0xF4 && (p)[0] <= 0xF7 && \
316 (p)[1] >= 0x80 && (p)[1] <= 0xBF && \
317 (p)[2] >= 0x80 && (p)[2] <= 0xBF && \
318 (p)[3] >= 0x80 && (p)[3] <= 0xBF)
319
320#define IS_UTF8_CHAR_3(p) \
321 (IS_UTF8_CHAR_3a(p) || \
322 IS_UTF8_CHAR_3b(p) || \
323 IS_UTF8_CHAR_3c(p) || \
324 IS_UTF8_CHAR_3d(p))
325#define IS_UTF8_CHAR_4(p) \
326 (IS_UTF8_CHAR_4a(p) || \
327 IS_UTF8_CHAR_4b(p) || \
328 IS_UTF8_CHAR_4c(p))
329
330/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
331 * (1) allows UTF-8 encoded UTF-16 surrogates
332 * (2) it allows code points past U+10FFFF.
333 * The Perl_is_utf8_char() full "slow" code will handle the Perl
334 * "extended UTF-8". */
335#define IS_UTF8_CHAR(p, n) \
336 ((n) == 1 ? IS_UTF8_CHAR_1(p) : \
337 (n) == 2 ? IS_UTF8_CHAR_2(p) : \
338 (n) == 3 ? IS_UTF8_CHAR_3(p) : \
339 (n) == 4 ? IS_UTF8_CHAR_4(p) : 0)
340
768c67ee 341#define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
342
77263263 343#endif /* IS_UTF8_CHAR() for UTF-8 */
e9a8c099 344
345/*
346 * Local variables:
347 * c-indentation-style: bsd
348 * c-basic-offset: 4
349 * indent-tabs-mode: t
350 * End:
351 *
352 * ex: set ts=8 sts=4 sw=4 noet:
353 */