Commit | Line | Data |
---|---|---|
a0ed51b3 LW |
1 | /* utf8.h |
2 | * | |
f3cb6f94 KW |
3 | * This file contains definitions for use with the UTF-8 encoding. It |
4 | * actually also works with the variant UTF-8 encoding called UTF-EBCDIC, and | |
5 | * hides almost all of the differences between these from the caller. In other | |
6 | * words, someone should #include this file, and if the code is being compiled | |
7 | * on an EBCDIC platform, things should mostly just work. | |
8 | * | |
2eee27d7 SS |
9 | * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009, |
10 | * 2010, 2011 by Larry Wall and others | |
a0ed51b3 LW |
11 | * |
12 | * You may distribute under the terms of either the GNU General Public | |
13 | * License or the Artistic License, as specified in the README file. | |
14 | * | |
15 | */ | |
16 | ||
57f0e7e2 KW |
17 | #ifndef H_UTF8 /* Guard against recursive inclusion */ |
18 | #define H_UTF8 1 | |
19 | ||
39e02b42 | 20 | /* Use UTF-8 as the default script encoding? |
1e54db1a | 21 | * Turning this on will break scripts having non-UTF-8 binary |
39e02b42 JH |
22 | * data (such as Latin-1) in string literals. */ |
23 | #ifdef USE_UTF8_SCRIPTS | |
24 | # define USE_UTF8_IN_NAMES (!IN_BYTES) | |
25 | #else | |
26 | # define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8) | |
27 | #endif | |
28 | ||
051a06d4 KW |
29 | /* For to_utf8_fold_flags, q.v. */ |
30 | #define FOLD_FLAGS_LOCALE 0x1 | |
31 | #define FOLD_FLAGS_FULL 0x2 | |
a0270393 | 32 | #define FOLD_FLAGS_NOMIX_ASCII 0x4 |
051a06d4 | 33 | |
83199d38 KW |
34 | /* For _core_swash_init(), internal core use only */ |
35 | #define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1 | |
5d3d13d1 | 36 | #define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2 |
87367d5f | 37 | #define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4 |
83199d38 | 38 | |
a0270393 | 39 | #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) |
051a06d4 KW |
40 | #define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \ |
41 | FOLD_FLAGS_FULL, NULL) | |
42 | #define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL) | |
43 | #define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL) | |
44 | #define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL) | |
36bb2ab6 | 45 | |
fd7cb289 RGS |
46 | /* Source backward compatibility. */ |
47 | #define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0) | |
48 | #define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0) | |
49 | ||
eda9cac1 KW |
50 | #define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ |
51 | foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0) | |
a33c29bc | 52 | #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0) |
5e64d0fa | 53 | #define FOLDEQ_UTF8_LOCALE (1 << 1) |
18f762c3 KW |
54 | #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2) |
55 | #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3) | |
a33c29bc | 56 | |
e6226b18 KW |
57 | /* |
58 | =for apidoc ibcmp_utf8 | |
59 | ||
60 | This is a synonym for (! foldEQ_utf8()) | |
61 | ||
62 | =cut | |
63 | */ | |
64 | #define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ | |
65 | cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2)) | |
66 | ||
1d72bdf6 NIS |
67 | #ifdef EBCDIC |
68 | /* The equivalent of these macros but implementing UTF-EBCDIC | |
69 | are in the following header file: | |
70 | */ | |
71 | ||
72 | #include "utfebcdic.h" | |
fd7cb289 | 73 | |
d06134e5 | 74 | #else /* ! EBCDIC */ |
73c4f7a1 GS |
75 | START_EXTERN_C |
76 | ||
a0ed51b3 | 77 | #ifdef DOINIT |
6f06b55f | 78 | EXTCONST unsigned char PL_utf8skip[] = { |
b2635aa8 KW |
79 | /* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ |
80 | /* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
81 | /* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
82 | /* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
83 | /* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
84 | /* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
85 | /* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
86 | /* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
87 | /* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
88 | /* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
89 | /* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
90 | /* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
91 | /* 0xC0 */ 2,2, /* overlong */ | |
92 | /* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */ | |
93 | /* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */ | |
94 | /* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */ | |
95 | /* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */ | |
96 | /* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit | |
97 | allowed (64-bit + reserved). */ | |
a0ed51b3 LW |
98 | }; |
99 | #else | |
6f06b55f | 100 | EXTCONST unsigned char PL_utf8skip[]; |
a0ed51b3 LW |
101 | #endif |
102 | ||
73c4f7a1 | 103 | END_EXTERN_C |
7e2040f0 | 104 | |
b96a92fb | 105 | #include "regcharclass.h" |
ec34087a KW |
106 | #include "unicode_constants.h" |
107 | ||
1d72bdf6 NIS |
108 | /* Native character to iso-8859-1 */ |
109 | #define NATIVE_TO_ASCII(ch) (ch) | |
110 | #define ASCII_TO_NATIVE(ch) (ch) | |
111 | /* Transform after encoding */ | |
112 | #define NATIVE_TO_UTF(ch) (ch) | |
428921e5 | 113 | #define NATIVE_TO_I8(ch) NATIVE_TO_UTF(ch) /* a clearer synonym */ |
1d72bdf6 | 114 | #define UTF_TO_NATIVE(ch) (ch) |
428921e5 | 115 | #define I8_TO_NATIVE(ch) UTF_TO_NATIVE(ch) |
1d72bdf6 NIS |
116 | /* Transforms in wide UV chars */ |
117 | #define UNI_TO_NATIVE(ch) (ch) | |
118 | #define NATIVE_TO_UNI(ch) (ch) | |
119 | /* Transforms in invariant space */ | |
120 | #define NATIVE_TO_NEED(enc,ch) (ch) | |
121 | #define ASCII_TO_NEED(enc,ch) (ch) | |
d7578b48 | 122 | |
d06134e5 | 123 | /* As there are no translations, avoid the function wrapper */ |
1754c1a1 | 124 | #define utf8n_to_uvchr utf8n_to_uvuni |
6dd9dce9 | 125 | #define valid_utf8_to_uvchr valid_utf8_to_uvuni |
1754c1a1 | 126 | #define uvchr_to_utf8 uvuni_to_utf8 |
2b9d42f0 | 127 | |
877d9f0d | 128 | /* |
9041c2e3 | 129 | |
8c007b5a | 130 | The following table is from Unicode 3.2. |
877d9f0d JH |
131 | |
132 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte | |
133 | ||
375122d7 | 134 | U+0000..U+007F 00..7F |
e1b711da | 135 | U+0080..U+07FF * C2..DF 80..BF |
37e2e78e | 136 | U+0800..U+0FFF E0 * A0..BF 80..BF |
375122d7 | 137 | U+1000..U+CFFF E1..EC 80..BF 80..BF |
e1b711da | 138 | U+D000..U+D7FF ED 80..9F 80..BF |
537124e4 | 139 | U+D800..U+DFFF ED A0..BF 80..BF (surrogates) |
375122d7 | 140 | U+E000..U+FFFF EE..EF 80..BF 80..BF |
37e2e78e | 141 | U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF |
877d9f0d JH |
142 | U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF |
143 | U+100000..U+10FFFF F4 80..8F 80..BF 80..BF | |
b2635aa8 KW |
144 | Below are non-Unicode code points |
145 | U+110000..U+13FFFF F4 90..BF 80..BF 80..BF | |
146 | U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF | |
537124e4 | 147 | U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF |
877d9f0d | 148 | |
e1b711da | 149 | Note the gaps before several of the byte entries above marked by '*'. These are |
37e2e78e KW |
150 | caused by legal UTF-8 avoiding non-shortest encodings: it is technically |
151 | possible to UTF-8-encode a single code point in different ways, but that is | |
152 | explicitly forbidden, and the shortest possible encoding should always be used | |
15824458 | 153 | (and that is what Perl does). The non-shortest ones are called 'overlongs'. |
8c007b5a | 154 | |
877d9f0d JH |
155 | */ |
156 | ||
8c007b5a JH |
157 | /* |
158 | Another way to look at it, as bits: | |
159 | ||
b2635aa8 | 160 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte |
8c007b5a | 161 | |
b2635aa8 KW |
162 | 0aaa aaaa 0aaa aaaa |
163 | 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa | |
164 | cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa | |
165 | 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa | |
8c007b5a JH |
166 | |
167 | As you can see, the continuation bytes all begin with C<10>, and the | |
e1b711da | 168 | leading bits of the start byte tell how many bytes there are in the |
8c007b5a JH |
169 | encoded character. |
170 | ||
65ab9279 TC |
171 | Perl's extended UTF-8 means we can have start bytes up to FF. |
172 | ||
8c007b5a JH |
173 | */ |
174 | ||
15824458 KW |
175 | /* Is the representation of the Unicode code point 'c' the same regardless of |
176 | * being encoded in UTF-8 or not? */ | |
c4d5f83a | 177 | #define UNI_IS_INVARIANT(c) (((UV)c) < 0x80) |
15824458 KW |
178 | |
179 | /* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is | |
180 | * the inverse of UTF8_IS_INVARIANT */ | |
181 | #define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80) | |
182 | ||
183 | /* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence? | |
184 | * This doesn't catch invariants (they are single-byte). It also excludes the | |
185 | * illegal overlong sequences that begin with C0 and C1. */ | |
0447e8df | 186 | #define UTF8_IS_START(c) (((U8)c) >= 0xc2) |
15824458 KW |
187 | |
188 | /* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the | |
189 | * first byte thereof? */ | |
e021c6e6 | 190 | #define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80) |
0ae1fa71 | 191 | |
15824458 KW |
192 | /* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use |
193 | * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to | |
194 | * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus | |
195 | * this matches 0xc[23]. */ | |
559c7f10 | 196 | #define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2) |
4ab10950 | 197 | |
15824458 KW |
198 | /* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that |
199 | * represent a code point > 255? */ | |
7b4252f4 | 200 | #define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4) |
8850bf83 | 201 | |
15824458 KW |
202 | /* This defines the 1-bits that are to be in the first byte of a multi-byte |
203 | * UTF-8 encoded character that give the number of bytes that comprise the | |
204 | * character. | |
205 | * */ | |
22901f30 | 206 | #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len)))) |
b2635aa8 KW |
207 | |
208 | /* Masks out the initial one bits in a start byte, leaving the real data ones. | |
209 | * Doesn't work on an invariant byte */ | |
22901f30 | 210 | #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2))) |
1d72bdf6 | 211 | |
15824458 KW |
212 | /* This defines the bits that are to be in the continuation bytes of a multi-byte |
213 | * UTF-8 encoded character that indicate it is a continuation byte. */ | |
1d72bdf6 | 214 | #define UTF_CONTINUATION_MARK 0x80 |
15824458 KW |
215 | |
216 | /* This is the number of low-order bits a continuation byte in a UTF-8 encoded | |
217 | * sequence contributes to the specification of the code point. In the bit | |
218 | * maps above, you see that the first 2 bits are a constant '10', leaving 6 of | |
219 | * real information */ | |
1d72bdf6 | 220 | #define UTF_ACCUMULATION_SHIFT 6 |
b2635aa8 KW |
221 | |
222 | /* 2**UTF_ACCUMULATION_SHIFT - 1 */ | |
1d72bdf6 | 223 | #define UTF_CONTINUATION_MASK ((U8)0x3f) |
c512ce4f | 224 | |
15824458 KW |
225 | /* If a value is anded with this, and the result is non-zero, then using the |
226 | * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left | |
227 | * */ | |
eb83ed87 | 228 | #define UTF_ACCUMULATION_OVERFLOW_MASK \ |
bb88be5f KW |
229 | (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \ |
230 | - UTF_ACCUMULATION_SHIFT)) | |
eb83ed87 | 231 | |
1d68d6cd | 232 | #ifdef HAS_QUAD |
5bbb0b5a | 233 | #define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \ |
1d68d6cd SC |
234 | (uv) < 0x800 ? 2 : \ |
235 | (uv) < 0x10000 ? 3 : \ | |
236 | (uv) < 0x200000 ? 4 : \ | |
237 | (uv) < 0x4000000 ? 5 : \ | |
238 | (uv) < 0x80000000 ? 6 : \ | |
9041c2e3 | 239 | (uv) < UTF8_QUAD_MAX ? 7 : 13 ) |
1d68d6cd SC |
240 | #else |
241 | /* No, I'm not even going to *TRY* putting #ifdef inside a #define */ | |
5bbb0b5a | 242 | #define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \ |
1d68d6cd SC |
243 | (uv) < 0x800 ? 2 : \ |
244 | (uv) < 0x10000 ? 3 : \ | |
245 | (uv) < 0x200000 ? 4 : \ | |
246 | (uv) < 0x4000000 ? 5 : \ | |
247 | (uv) < 0x80000000 ? 6 : 7 ) | |
248 | #endif | |
249 | ||
03c76984 KW |
250 | /* How wide can a single UTF-8 encoded character become in bytes. */ |
251 | /* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8 | |
252 | * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be | |
253 | * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode | |
254 | * non-negative integers in a binary format, even those above Unicode */ | |
255 | #define UTF8_MAXBYTES 13 | |
256 | ||
257 | /* The maximum number of UTF-8 bytes a single Unicode character can | |
258 | * uppercase/lowercase/fold into. Unicode guarantees that the maximum | |
259 | * expansion is 3 characters. On ASCIIish platforms, the highest Unicode | |
260 | * character occupies 4 bytes, therefore this number would be 12, but this is | |
261 | * smaller than the maximum width a single above-Unicode character can occupy, | |
262 | * so use that instead */ | |
263 | #if UTF8_MAXBYTES < 12 | |
264 | #error UTF8_MAXBYTES must be at least 12 | |
265 | #endif | |
266 | ||
267 | #define UTF8_MAXBYTES_CASE UTF8_MAXBYTES | |
268 | ||
d06134e5 KW |
269 | #endif /* EBCDIC vs ASCII */ |
270 | ||
271 | /* Rest of these are attributes of Unicode and perl's internals rather than the | |
272 | * encoding, or happen to be the same in both ASCII and EBCDIC (at least at | |
273 | * this level; the macros that some of these call may have different | |
274 | * definitions in the two encodings */ | |
275 | ||
276 | #define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */ | |
277 | ||
537124e4 KW |
278 | /* Adds a UTF8 continuation byte 'new' of information to a running total code |
279 | * point 'old' of all the continuation bytes so far. This is designed to be | |
280 | * used in a loop to convert from UTF-8 to the code point represented */ | |
bb88be5f KW |
281 | #define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \ |
282 | | (((U8)new) & UTF_CONTINUATION_MASK)) | |
d06134e5 | 283 | |
4ab10950 KW |
284 | /* This works in the face of malformed UTF-8. */ |
285 | #define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \ | |
286 | && ( (e) - (s) > 1) \ | |
287 | && UTF8_IS_CONTINUATION(*((s)+1))) | |
288 | ||
2950f2a7 KW |
289 | /* Convert a two (not one) byte utf8 character to a unicode code point value. |
290 | * Needs just one iteration of accumulate. Should not be used unless it is | |
291 | * known that the two bytes are legal: 1) two-byte start, and 2) continuation. | |
292 | * Note that the result can be larger than 255 if the input character is not | |
293 | * downgradable */ | |
294 | #define TWO_BYTE_UTF8_TO_UNI(HI, LO) \ | |
295 | UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \ | |
296 | NATIVE_TO_UTF(LO)) | |
297 | ||
15824458 KW |
298 | /* How many bytes in the UTF-8 encoded character whose first (perhaps only) |
299 | * byte is pointed to by 's' */ | |
d06134e5 KW |
300 | #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)] |
301 | ||
15824458 KW |
302 | /* Is the byte 'c' the same character when encoded in UTF-8 as when not. This |
303 | * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in | |
304 | * each for the exact same set of bit patterns. (And it works on any byte in a | |
305 | * UTF-8 encoded string) */ | |
d06134e5 | 306 | #define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c)) |
15824458 | 307 | |
d06134e5 KW |
308 | #define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c)) |
309 | ||
310 | #define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */ | |
311 | ||
312 | /* The macros in the next sets are used to generate the two utf8 or utfebcdic | |
313 | * bytes from an ordinal that is known to fit into two bytes; it must be less | |
314 | * than 0x3FF to work across both encodings. */ | |
315 | /* Nocast allows these to be used in the case label of a switch statement */ | |
428921e5 KW |
316 | #define UTF8_TWO_BYTE_HI_nocast(c) NATIVE_TO_I8(((c) \ |
317 | >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2))) | |
318 | #define UTF8_TWO_BYTE_LO_nocast(c) NATIVE_TO_I8(((c) & UTF_CONTINUATION_MASK) \ | |
319 | | UTF_CONTINUATION_MARK) | |
d06134e5 KW |
320 | |
321 | #define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c))) | |
322 | #define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c))) | |
323 | ||
324 | /* This name is used when the source is a single byte */ | |
325 | #define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c)) | |
326 | #define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c)) | |
327 | ||
7e2040f0 | 328 | /* |
e3036cf4 | 329 | * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem |
20df05f4 KW |
330 | * from an earlier version of these macros in which they didn't call the |
331 | * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the | |
332 | * beginning of a utf8 character. Now that foo_utf8() determines that itself, | |
333 | * no need to do it again here | |
7e2040f0 | 334 | */ |
e3036cf4 KW |
335 | #define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \ |
336 | ? isIDFIRST(*(p)) \ | |
337 | : isIDFIRST_utf8((const U8*)p)) | |
32636478 KW |
338 | #define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \ |
339 | ? isWORDCHAR(*(p)) \ | |
340 | : isWORDCHAR_utf8((const U8*)p)) | |
341 | #define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF) | |
1d72bdf6 | 342 | |
89ebb4a3 JH |
343 | #define UTF8_MAXLEN UTF8_MAXBYTES |
344 | ||
8cb75cc8 KW |
345 | /* A Unicode character can fold to up to 3 characters */ |
346 | #define UTF8_MAX_FOLD_CHAR_EXPAND 3 | |
347 | ||
a98fe34d | 348 | #define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES) |
0064a8a9 | 349 | #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES) |
66cbab2c | 350 | #define IN_UNI_8_BIT \ |
b36bf33f KW |
351 | (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \ |
352 | && ! IN_LOCALE_RUNTIME && ! IN_BYTES) | |
353 | ||
1d72bdf6 | 354 | |
c76687c5 KW |
355 | #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */ |
356 | ||
357 | /* Allow first byte to be a continuation byte */ | |
1d72bdf6 | 358 | #define UTF8_ALLOW_CONTINUATION 0x0002 |
c76687c5 KW |
359 | |
360 | /* Allow second... bytes to be non-continuation bytes */ | |
1d72bdf6 | 361 | #define UTF8_ALLOW_NON_CONTINUATION 0x0004 |
949cf498 KW |
362 | |
363 | /* expecting more bytes than were available in the string */ | |
364 | #define UTF8_ALLOW_SHORT 0x0008 | |
365 | ||
366 | /* Overlong sequence; i.e., the code point can be specified in fewer bytes. */ | |
367 | #define UTF8_ALLOW_LONG 0x0010 | |
368 | ||
369 | #define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */ | |
370 | #define UTF8_WARN_SURROGATE 0x0040 | |
371 | ||
372 | #define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */ | |
373 | #define UTF8_WARN_NONCHAR 0x0100 /* code points */ | |
374 | ||
375 | #define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */ | |
376 | #define UTF8_WARN_SUPER 0x0400 /* points above the legal max */ | |
377 | ||
378 | /* Code points which never were part of the original UTF-8 standard, the first | |
379 | * byte of which is a FE or FF on ASCII platforms. */ | |
380 | #define UTF8_DISALLOW_FE_FF 0x0800 | |
381 | #define UTF8_WARN_FE_FF 0x1000 | |
382 | ||
383 | #define UTF8_CHECK_ONLY 0x2000 | |
384 | ||
385 | /* For backwards source compatibility. They do nothing, as the default now | |
386 | * includes what they used to mean. The first one's meaning was to allow the | |
387 | * just the single non-character 0xFFFF */ | |
388 | #define UTF8_ALLOW_FFFF 0 | |
389 | #define UTF8_ALLOW_SURROGATE 0 | |
390 | ||
33d9abfb | 391 | #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF) |
949cf498 | 392 | #define UTF8_WARN_ILLEGAL_INTERCHANGE \ |
33d9abfb | 393 | (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF) |
949cf498 KW |
394 | #define UTF8_ALLOW_ANY \ |
395 | (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) | |
396 | #define UTF8_ALLOW_ANYUV \ | |
397 | (UTF8_ALLOW_EMPTY \ | |
398 | & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) | |
9f7f3913 TS |
399 | #define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \ |
400 | UTF8_ALLOW_ANYUV) | |
1d72bdf6 | 401 | |
7131f24d KW |
402 | /* Surrogates, non-character code points and above-Unicode code points are |
403 | * problematic in some contexts. This allows code that needs to check for | |
404 | * those to to quickly exclude the vast majority of code points it will | |
405 | * encounter */ | |
b96a92fb KW |
406 | #define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \ |
407 | FIRST_SURROGATE_UTF8_FIRST_BYTE | |
7131f24d | 408 | |
b96a92fb KW |
409 | #define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s)) |
410 | #define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send)) | |
7131f24d KW |
411 | |
412 | /* ASCII EBCDIC I8 | |
413 | * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode | |
414 | * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0 | |
415 | * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1 | |
416 | */ | |
417 | #ifdef EBCDIC /* Both versions assume well-formed UTF8 */ | |
0404c92a KW |
418 | # define UTF8_IS_SUPER(s) (NATIVE_TO_I8(* (U8*) (s)) >= 0xF9 \ |
419 | && (NATIVE_TO_I8(* (U8*) (s)) > 0xF9 \ | |
420 | || (NATIVE_TO_I8(* (U8*) ((s)) + 1 >= 0xA2)))) | |
7131f24d | 421 | #else |
0404c92a KW |
422 | # define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \ |
423 | && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))) | |
7131f24d KW |
424 | #endif |
425 | ||
b96a92fb KW |
426 | /* These are now machine generated, and the 'given' clause is no longer |
427 | * applicable */ | |
428 | #define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \ | |
429 | cBOOL(is_NONCHAR_utf8(s)) | |
430 | #define UTF8_IS_NONCHAR_(s) \ | |
431 | UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) | |
7131f24d | 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 | |
bb88be5f KW |
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) | |
949cf498 KW |
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 | |
ec34087a KW |
478 | #define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE |
479 | #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \ | |
480 | LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE | |
481 | #define MICRO_SIGN MICRO_SIGN_NATIVE | |
482 | #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \ | |
483 | LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE | |
484 | #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \ | |
485 | LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE | |
09091399 JH |
486 | #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3 |
487 | #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2 | |
488 | #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3 | |
9dcbe121 | 489 | #define GREEK_SMALL_LETTER_MU 0x03BC |
78a0d3cc KW |
490 | #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */ |
491 | #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */ | |
97298f37 | 492 | #define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E |
8d64d87f KW |
493 | #define LATIN_SMALL_LETTER_LONG_S 0x017F |
494 | #define KELVIN_SIGN 0x212A | |
495 | #define ANGSTROM_SIGN 0x212B | |
09091399 | 496 | |
9e55ce06 | 497 | #define UNI_DISPLAY_ISPRINT 0x0001 |
c728cb41 JH |
498 | #define UNI_DISPLAY_BACKSLASH 0x0002 |
499 | #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) | |
500 | #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) | |
9e55ce06 | 501 | |
5cd46e1f KW |
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 | 506 | ((end) > (input) + 1) && \ |
d22b930b KW |
507 | toFOLD((input)[0]) == 's' && \ |
508 | toFOLD((input)[1]) == 's') | |
ebc501f0 | 509 | #define SHARP_S_SKIP 2 |
3b0fc154 | 510 | |
a4f7a67c KW |
511 | /* If you want to exclude surrogates, and beyond legal Unicode, see the blame |
512 | * log for earlier versions which gave details for these */ | |
4d646140 KW |
513 | |
514 | #ifndef EBCDIC | |
515 | /* This was generated by regen/regcharclass.pl, and then moved here. The lines | |
516 | * that generated it were then commented out. This was done solely because it | |
517 | * takes on the order of 10 minutes to generate, and is never going to change. | |
518 | * The EBCDIC equivalent hasn't been commented out in regcharclass.pl, so it | |
519 | * should generate and run the correct stuff */ | |
39a0f513 YO |
520 | /* |
521 | UTF8_CHAR: Matches utf8 from 1 to 4 bytes | |
522 | ||
523 | 0x0 - 0x1FFFFF | |
524 | */ | |
4d646140 KW |
525 | /*** GENERATED CODE ***/ |
526 | #define is_UTF8_CHAR_utf8_safe(s,e) \ | |
527 | ( ((e)-(s) > 3) ? \ | |
528 | ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \ | |
529 | : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \ | |
530 | ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \ | |
531 | : ( 0xE0 == ((U8*)s)[0] ) ? \ | |
532 | ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ | |
533 | : ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \ | |
534 | ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ | |
535 | : ( 0xF0 == ((U8*)s)[0] ) ? \ | |
536 | ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\ | |
39a0f513 | 537 | : ( ( ( ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF7 ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\ |
4d646140 KW |
538 | : ((e)-(s) > 2) ? \ |
539 | ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \ | |
540 | : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \ | |
541 | ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \ | |
542 | : ( 0xE0 == ((U8*)s)[0] ) ? \ | |
543 | ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ | |
39a0f513 | 544 | : ( ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ |
4d646140 KW |
545 | : ((e)-(s) > 1) ? \ |
546 | ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \ | |
39a0f513 | 547 | : ( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) ? 2 : 0 )\ |
4d646140 KW |
548 | : ((e)-(s) > 0) ? \ |
549 | ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) \ | |
550 | : 0 ) | |
551 | #endif | |
3b0fc154 JH |
552 | |
553 | /* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it | |
554 | * (1) allows UTF-8 encoded UTF-16 surrogates | |
555 | * (2) it allows code points past U+10FFFF. | |
556 | * The Perl_is_utf8_char() full "slow" code will handle the Perl | |
557 | * "extended UTF-8". */ | |
4d646140 | 558 | #define IS_UTF8_CHAR(p, n) (is_UTF8_CHAR_utf8_safe(p, (p) + (n)) == n) |
2f454f11 | 559 | |
4d646140 KW |
560 | /* regen/regcharclass.pl generates is_UTF8_CHAR_utf8_safe() macros for up to |
561 | * these number of bytes. So this has to be coordinated with it */ | |
562 | #ifdef EBCDIC | |
2f454f11 | 563 | # define IS_UTF8_CHAR_FAST(n) ((n) <= 5) |
4d646140 KW |
564 | #else |
565 | # define IS_UTF8_CHAR_FAST(n) ((n) <= 4) | |
566 | #endif | |
e9a8c099 | 567 | |
57f0e7e2 KW |
568 | #endif /* H_UTF8 */ |
569 | ||
e9a8c099 MHM |
570 | /* |
571 | * Local variables: | |
572 | * c-indentation-style: bsd | |
573 | * c-basic-offset: 4 | |
14d04a33 | 574 | * indent-tabs-mode: nil |
e9a8c099 MHM |
575 | * End: |
576 | * | |
14d04a33 | 577 | * ex: set ts=8 sts=4 sw=4 et: |
e9a8c099 | 578 | */ |