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 | ||
3cd96634 KW |
29 | #include "regcharclass.h" |
30 | #include "unicode_constants.h" | |
31 | ||
051a06d4 KW |
32 | /* For to_utf8_fold_flags, q.v. */ |
33 | #define FOLD_FLAGS_LOCALE 0x1 | |
34 | #define FOLD_FLAGS_FULL 0x2 | |
a0270393 | 35 | #define FOLD_FLAGS_NOMIX_ASCII 0x4 |
051a06d4 | 36 | |
83199d38 KW |
37 | /* For _core_swash_init(), internal core use only */ |
38 | #define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1 | |
5d3d13d1 | 39 | #define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2 |
87367d5f | 40 | #define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4 |
83199d38 | 41 | |
7bbfa158 KW |
42 | /* |
43 | =head1 Unicode Support | |
8cca77bc KW |
44 | L<perlguts/Unicode Support> has an introduction to this API. |
45 | ||
46 | See also L</Character classification>, | |
47 | and L</Character case changing>. | |
48 | Various functions outside this section also work specially with Unicode. | |
49 | Search for the string "utf8" in this document. | |
7bbfa158 KW |
50 | |
51 | =for apidoc is_ascii_string | |
52 | ||
53 | This is a misleadingly-named synonym for L</is_invariant_string>. | |
54 | On ASCII-ish platforms, the name isn't misleading: the ASCII-range characters | |
55 | are exactly the UTF-8 invariants. But EBCDIC machines have more invariants | |
56 | than just the ASCII characters, so C<is_invariant_string> is preferred. | |
57 | ||
58 | =cut | |
59 | */ | |
60 | #define is_ascii_string(s, len) is_invariant_string(s, len) | |
61 | ||
de69f3af KW |
62 | #define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0) |
63 | #define uvchr_to_utf8_flags(d,uv,flags) \ | |
64 | uvoffuni_to_utf8_flags(d,NATIVE_TO_UNI(uv),flags) | |
65 | #define utf8_to_uvchr_buf(s, e, lenp) \ | |
842991ae | 66 | utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \ |
de69f3af KW |
67 | ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY) |
68 | ||
a0270393 | 69 | #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) |
445bf929 KW |
70 | #define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) |
71 | #define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0) | |
72 | #define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0) | |
73 | #define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0) | |
36bb2ab6 | 74 | |
fd7cb289 | 75 | /* Source backward compatibility. */ |
fd7cb289 RGS |
76 | #define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0) |
77 | ||
eda9cac1 KW |
78 | #define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ |
79 | foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0) | |
baa60164 | 80 | #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0) |
cea315b6 | 81 | #define FOLDEQ_LOCALE (1 << 1) |
18f762c3 KW |
82 | #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2) |
83 | #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3) | |
d635b710 KW |
84 | #define FOLDEQ_S1_FOLDS_SANE (1 << 4) |
85 | #define FOLDEQ_S2_FOLDS_SANE (1 << 5) | |
a33c29bc | 86 | |
e6226b18 KW |
87 | #define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ |
88 | cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2)) | |
89 | ||
1d72bdf6 NIS |
90 | #ifdef EBCDIC |
91 | /* The equivalent of these macros but implementing UTF-EBCDIC | |
92 | are in the following header file: | |
93 | */ | |
94 | ||
95 | #include "utfebcdic.h" | |
fd7cb289 | 96 | |
d06134e5 | 97 | #else /* ! EBCDIC */ |
73c4f7a1 GS |
98 | START_EXTERN_C |
99 | ||
a0ed51b3 | 100 | #ifdef DOINIT |
6f06b55f | 101 | EXTCONST unsigned char PL_utf8skip[] = { |
b2635aa8 KW |
102 | /* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ |
103 | /* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
104 | /* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
105 | /* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
106 | /* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
107 | /* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
108 | /* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
109 | /* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ | |
110 | /* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
111 | /* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
112 | /* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
113 | /* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ | |
114 | /* 0xC0 */ 2,2, /* overlong */ | |
115 | /* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */ | |
116 | /* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */ | |
117 | /* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */ | |
118 | /* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */ | |
119 | /* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit | |
120 | allowed (64-bit + reserved). */ | |
a0ed51b3 LW |
121 | }; |
122 | #else | |
6f06b55f | 123 | EXTCONST unsigned char PL_utf8skip[]; |
a0ed51b3 LW |
124 | #endif |
125 | ||
73c4f7a1 | 126 | END_EXTERN_C |
7e2040f0 | 127 | |
59a449d5 KW |
128 | /* Native character to/from iso-8859-1. Are the identity functions on ASCII |
129 | * platforms */ | |
130 | #define NATIVE_TO_LATIN1(ch) (ch) | |
131 | #define LATIN1_TO_NATIVE(ch) (ch) | |
132 | ||
133 | /* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus | |
134 | * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking | |
135 | * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them | |
136 | * because they are 8-bit encodings that serve the same purpose in Perl, and | |
137 | * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to | |
138 | * whichever one is applicable on the current platform */ | |
139 | #define NATIVE_UTF8_TO_I8(ch) (ch) | |
140 | #define I8_TO_NATIVE_UTF8(ch) (ch) | |
141 | ||
1d72bdf6 NIS |
142 | /* Transforms in wide UV chars */ |
143 | #define UNI_TO_NATIVE(ch) (ch) | |
144 | #define NATIVE_TO_UNI(ch) (ch) | |
d7578b48 | 145 | |
877d9f0d | 146 | /* |
9041c2e3 | 147 | |
8c007b5a | 148 | The following table is from Unicode 3.2. |
877d9f0d JH |
149 | |
150 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte | |
151 | ||
375122d7 | 152 | U+0000..U+007F 00..7F |
e1b711da | 153 | U+0080..U+07FF * C2..DF 80..BF |
37e2e78e | 154 | U+0800..U+0FFF E0 * A0..BF 80..BF |
375122d7 | 155 | U+1000..U+CFFF E1..EC 80..BF 80..BF |
e1b711da | 156 | U+D000..U+D7FF ED 80..9F 80..BF |
537124e4 | 157 | U+D800..U+DFFF ED A0..BF 80..BF (surrogates) |
375122d7 | 158 | U+E000..U+FFFF EE..EF 80..BF 80..BF |
37e2e78e | 159 | U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF |
877d9f0d JH |
160 | U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF |
161 | U+100000..U+10FFFF F4 80..8F 80..BF 80..BF | |
b2635aa8 KW |
162 | Below are non-Unicode code points |
163 | U+110000..U+13FFFF F4 90..BF 80..BF 80..BF | |
164 | U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF | |
537124e4 | 165 | U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF |
877d9f0d | 166 | |
e1b711da | 167 | Note the gaps before several of the byte entries above marked by '*'. These are |
37e2e78e KW |
168 | caused by legal UTF-8 avoiding non-shortest encodings: it is technically |
169 | possible to UTF-8-encode a single code point in different ways, but that is | |
170 | explicitly forbidden, and the shortest possible encoding should always be used | |
15824458 | 171 | (and that is what Perl does). The non-shortest ones are called 'overlongs'. |
8c007b5a | 172 | |
877d9f0d JH |
173 | */ |
174 | ||
8c007b5a JH |
175 | /* |
176 | Another way to look at it, as bits: | |
177 | ||
b2635aa8 | 178 | Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte |
8c007b5a | 179 | |
b2635aa8 KW |
180 | 0aaa aaaa 0aaa aaaa |
181 | 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa | |
182 | cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa | |
183 | 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa | |
8c007b5a JH |
184 | |
185 | As you can see, the continuation bytes all begin with C<10>, and the | |
e1b711da | 186 | leading bits of the start byte tell how many bytes there are in the |
8c007b5a JH |
187 | encoded character. |
188 | ||
65ab9279 TC |
189 | Perl's extended UTF-8 means we can have start bytes up to FF. |
190 | ||
8c007b5a JH |
191 | */ |
192 | ||
15824458 KW |
193 | /* Is the representation of the Unicode code point 'c' the same regardless of |
194 | * being encoded in UTF-8 or not? */ | |
c4d5f83a | 195 | #define UNI_IS_INVARIANT(c) (((UV)c) < 0x80) |
15824458 KW |
196 | |
197 | /* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is | |
198 | * the inverse of UTF8_IS_INVARIANT */ | |
199 | #define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80) | |
200 | ||
201 | /* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence? | |
202 | * This doesn't catch invariants (they are single-byte). It also excludes the | |
203 | * illegal overlong sequences that begin with C0 and C1. */ | |
0447e8df | 204 | #define UTF8_IS_START(c) (((U8)c) >= 0xc2) |
15824458 KW |
205 | |
206 | /* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the | |
207 | * first byte thereof? */ | |
e021c6e6 | 208 | #define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80) |
0ae1fa71 | 209 | |
15824458 KW |
210 | /* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use |
211 | * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to | |
212 | * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus | |
213 | * this matches 0xc[23]. */ | |
559c7f10 | 214 | #define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2) |
4ab10950 | 215 | |
15824458 KW |
216 | /* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that |
217 | * represent a code point > 255? */ | |
7b4252f4 | 218 | #define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4) |
8850bf83 | 219 | |
15824458 KW |
220 | /* This defines the 1-bits that are to be in the first byte of a multi-byte |
221 | * UTF-8 encoded character that give the number of bytes that comprise the | |
f710bf0c | 222 | * character. 'len' is the number of bytes in the multi-byte sequence. */ |
ee372ee9 | 223 | #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len))))) |
b2635aa8 KW |
224 | |
225 | /* Masks out the initial one bits in a start byte, leaving the real data ones. | |
f710bf0c KW |
226 | * Doesn't work on an invariant byte. 'len' is the number of bytes in the |
227 | * multi-byte sequence that comprises the character. */ | |
22901f30 | 228 | #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2))) |
1d72bdf6 | 229 | |
15824458 KW |
230 | /* This defines the bits that are to be in the continuation bytes of a multi-byte |
231 | * UTF-8 encoded character that indicate it is a continuation byte. */ | |
1d72bdf6 | 232 | #define UTF_CONTINUATION_MARK 0x80 |
15824458 KW |
233 | |
234 | /* This is the number of low-order bits a continuation byte in a UTF-8 encoded | |
235 | * sequence contributes to the specification of the code point. In the bit | |
236 | * maps above, you see that the first 2 bits are a constant '10', leaving 6 of | |
237 | * real information */ | |
1d72bdf6 | 238 | #define UTF_ACCUMULATION_SHIFT 6 |
b2635aa8 KW |
239 | |
240 | /* 2**UTF_ACCUMULATION_SHIFT - 1 */ | |
1d72bdf6 | 241 | #define UTF_CONTINUATION_MASK ((U8)0x3f) |
c512ce4f | 242 | |
15824458 KW |
243 | /* If a value is anded with this, and the result is non-zero, then using the |
244 | * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left | |
245 | * */ | |
eb83ed87 | 246 | #define UTF_ACCUMULATION_OVERFLOW_MASK \ |
bb88be5f KW |
247 | (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \ |
248 | - UTF_ACCUMULATION_SHIFT)) | |
eb83ed87 | 249 | |
6588300d NC |
250 | #if UVSIZE >= 8 |
251 | # define UTF8_QUAD_MAX UINT64_C(0x1000000000) | |
252 | ||
5aaebcb3 KW |
253 | /* Input is a true Unicode (not-native) code point */ |
254 | #define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \ | |
1d68d6cd SC |
255 | (uv) < 0x800 ? 2 : \ |
256 | (uv) < 0x10000 ? 3 : \ | |
257 | (uv) < 0x200000 ? 4 : \ | |
258 | (uv) < 0x4000000 ? 5 : \ | |
259 | (uv) < 0x80000000 ? 6 : \ | |
9041c2e3 | 260 | (uv) < UTF8_QUAD_MAX ? 7 : 13 ) |
1d68d6cd SC |
261 | #else |
262 | /* No, I'm not even going to *TRY* putting #ifdef inside a #define */ | |
5aaebcb3 | 263 | #define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \ |
1d68d6cd SC |
264 | (uv) < 0x800 ? 2 : \ |
265 | (uv) < 0x10000 ? 3 : \ | |
266 | (uv) < 0x200000 ? 4 : \ | |
267 | (uv) < 0x4000000 ? 5 : \ | |
268 | (uv) < 0x80000000 ? 6 : 7 ) | |
269 | #endif | |
270 | ||
03c76984 KW |
271 | /* How wide can a single UTF-8 encoded character become in bytes. */ |
272 | /* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8 | |
273 | * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be | |
274 | * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode | |
275 | * non-negative integers in a binary format, even those above Unicode */ | |
276 | #define UTF8_MAXBYTES 13 | |
277 | ||
278 | /* The maximum number of UTF-8 bytes a single Unicode character can | |
279 | * uppercase/lowercase/fold into. Unicode guarantees that the maximum | |
280 | * expansion is 3 characters. On ASCIIish platforms, the highest Unicode | |
281 | * character occupies 4 bytes, therefore this number would be 12, but this is | |
282 | * smaller than the maximum width a single above-Unicode character can occupy, | |
283 | * so use that instead */ | |
284 | #if UTF8_MAXBYTES < 12 | |
285 | #error UTF8_MAXBYTES must be at least 12 | |
286 | #endif | |
287 | ||
0ed2b00b KW |
288 | /* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL() |
289 | * for more */ | |
290 | #define QUESTION_MARK_CTRL DEL_NATIVE | |
291 | ||
843a4590 KW |
292 | #define MAX_UTF8_TWO_BYTE 0x7FF |
293 | ||
03c76984 KW |
294 | #define UTF8_MAXBYTES_CASE UTF8_MAXBYTES |
295 | ||
d06134e5 KW |
296 | #endif /* EBCDIC vs ASCII */ |
297 | ||
298 | /* Rest of these are attributes of Unicode and perl's internals rather than the | |
299 | * encoding, or happen to be the same in both ASCII and EBCDIC (at least at | |
300 | * this level; the macros that some of these call may have different | |
301 | * definitions in the two encodings */ | |
302 | ||
59a449d5 KW |
303 | /* In domain restricted to ASCII, these may make more sense to the reader than |
304 | * the ones with Latin1 in the name */ | |
305 | #define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch) | |
306 | #define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch) | |
307 | ||
308 | /* More or less misleadingly-named defines, retained for back compat */ | |
309 | #define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch) | |
310 | #define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch) | |
311 | #define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch) | |
312 | #define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch) | |
313 | #define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch) | |
d06134e5 | 314 | |
537124e4 KW |
315 | /* Adds a UTF8 continuation byte 'new' of information to a running total code |
316 | * point 'old' of all the continuation bytes so far. This is designed to be | |
155d2738 KW |
317 | * used in a loop to convert from UTF-8 to the code point represented. Note |
318 | * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is | |
319 | * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC) | |
320 | * code point in process of being generated */ | |
321 | #define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \ | |
322 | | ((NATIVE_UTF8_TO_I8((U8)new)) \ | |
323 | & UTF_CONTINUATION_MASK)) | |
d06134e5 | 324 | |
4ab10950 KW |
325 | /* This works in the face of malformed UTF-8. */ |
326 | #define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \ | |
327 | && ( (e) - (s) > 1) \ | |
328 | && UTF8_IS_CONTINUATION(*((s)+1))) | |
329 | ||
5aaebcb3 KW |
330 | /* Number of bytes a code point occupies in UTF-8. */ |
331 | #define NATIVE_SKIP(uv) OFFUNISKIP(NATIVE_TO_UNI(uv)) | |
332 | ||
bd18bd40 KW |
333 | /* |
334 | ||
335 | =for apidoc Am|STRLEN|UVCHR_SKIP|UV cp | |
336 | returns the number of bytes required to represent the code point C<cp> when | |
337 | encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than | |
338 | 255; a Unicode code point otherwise. | |
339 | ||
340 | =cut | |
341 | */ | |
342 | ||
5aaebcb3 KW |
343 | /* Most code which says UNISKIP is really thinking in terms of native code |
344 | * points (0-255) plus all those beyond. This is an imprecise term, but having | |
2accb712 KW |
345 | * it means existing code continues to work. For precision, use UVCHR_SKIP, |
346 | * NATIVE_SKIP, and OFFUNISKIP */ | |
5aaebcb3 | 347 | #define UNISKIP(uv) NATIVE_SKIP(uv) |
2accb712 | 348 | #define UVCHR_SKIP(uv) NATIVE_SKIP(uv) |
5aaebcb3 | 349 | |
94bb8c36 | 350 | /* Convert a two (not one) byte utf8 character to a native code point value. |
2950f2a7 KW |
351 | * Needs just one iteration of accumulate. Should not be used unless it is |
352 | * known that the two bytes are legal: 1) two-byte start, and 2) continuation. | |
353 | * Note that the result can be larger than 255 if the input character is not | |
354 | * downgradable */ | |
94bb8c36 KW |
355 | #define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \ |
356 | UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \ | |
155d2738 | 357 | (LO))) |
94bb8c36 KW |
358 | |
359 | /* Should never be used, and be deprecated */ | |
360 | #define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO)) | |
2950f2a7 | 361 | |
bd18bd40 KW |
362 | /* |
363 | ||
364 | =for apidoc Am|STRLEN|UTF8SKIP|char* s | |
365 | returns the number of bytes in the UTF-8 encoded character whose first (perhaps | |
366 | only) byte is pointed to by C<s>. | |
367 | ||
368 | =cut | |
369 | */ | |
d06134e5 KW |
370 | #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)] |
371 | ||
15824458 KW |
372 | /* Is the byte 'c' the same character when encoded in UTF-8 as when not. This |
373 | * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in | |
374 | * each for the exact same set of bit patterns. (And it works on any byte in a | |
375 | * UTF-8 encoded string) */ | |
bc3632a8 | 376 | #define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_UTF8_TO_I8(c)) |
15824458 | 377 | |
6f2d5cbc KW |
378 | /* Like the above, but its name implies a non-UTF8 input, and is implemented |
379 | * differently (for no particular reason) */ | |
380 | #define NATIVE_BYTE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_LATIN1(c)) | |
381 | ||
382 | /* Like the above, but accepts any UV as input */ | |
383 | #define UVCHR_IS_INVARIANT(uv) UNI_IS_INVARIANT(NATIVE_TO_UNI(uv)) | |
d06134e5 KW |
384 | |
385 | #define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */ | |
386 | ||
48ccf5e1 KW |
387 | /* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic |
388 | * bytes from an ordinal that is known to fit into exactly two (not one) bytes; | |
389 | * it must be less than 0x3FF to work across both encodings. */ | |
390 | ||
391 | /* These two are helper macros for the other three sets, and should not be used | |
392 | * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1 | |
393 | * (which works for code points up to 0xFF) or NATIVE_TO_UNI which works for any | |
394 | * code point */ | |
395 | #define __BASE_TWO_BYTE_HI(c, translate_function) \ | |
396 | I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \ | |
397 | | UTF_START_MARK(2)) | |
398 | #define __BASE_TWO_BYTE_LO(c, translate_function) \ | |
399 | I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \ | |
400 | | UTF_CONTINUATION_MARK) | |
401 | ||
48ccf5e1 KW |
402 | /* The next two macros should not be used. They were designed to be usable as |
403 | * the case label of a switch statement, but this doesn't work for EBCDIC. Use | |
9d0d3a03 | 404 | * regen/unicode_constants.pl instead */ |
48ccf5e1 KW |
405 | #define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI) |
406 | #define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI) | |
407 | ||
408 | /* The next two macros are used when the source should be a single byte | |
409 | * character; checked for under DEBUGGING */ | |
410 | #define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ | |
411 | ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1))) | |
412 | #define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ | |
413 | ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1))) | |
414 | ||
415 | /* These final two macros in the series are used when the source can be any | |
416 | * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient | |
417 | * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~' | |
418 | * operator instead of "<=" to avoid getting compiler warnings. | |
d52b8576 | 419 | * MAX_UTF8_TWO_BYTE should be exactly all one bits in the lower few |
48ccf5e1 KW |
420 | * places, so the ~ works */ |
421 | #define UTF8_TWO_BYTE_HI(c) \ | |
422 | (__ASSERT_((sizeof(c) == 1) \ | |
d52b8576 | 423 | || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \ |
d9759938 | 424 | ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI))) |
48ccf5e1 KW |
425 | #define UTF8_TWO_BYTE_LO(c) \ |
426 | (__ASSERT_((sizeof(c) == 1) \ | |
d52b8576 | 427 | || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \ |
d9759938 | 428 | ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI))) |
d06134e5 | 429 | |
e7214ce8 KW |
430 | /* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII |
431 | * as it is only in overlongs. */ | |
432 | #define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1) | |
433 | ||
7e2040f0 | 434 | /* |
e3036cf4 | 435 | * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem |
20df05f4 KW |
436 | * from an earlier version of these macros in which they didn't call the |
437 | * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the | |
438 | * beginning of a utf8 character. Now that foo_utf8() determines that itself, | |
439 | * no need to do it again here | |
7e2040f0 | 440 | */ |
e3036cf4 KW |
441 | #define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \ |
442 | ? isIDFIRST(*(p)) \ | |
443 | : isIDFIRST_utf8((const U8*)p)) | |
32636478 KW |
444 | #define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \ |
445 | ? isWORDCHAR(*(p)) \ | |
446 | : isWORDCHAR_utf8((const U8*)p)) | |
447 | #define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF) | |
1d72bdf6 | 448 | |
89ebb4a3 JH |
449 | #define UTF8_MAXLEN UTF8_MAXBYTES |
450 | ||
8cb75cc8 KW |
451 | /* A Unicode character can fold to up to 3 characters */ |
452 | #define UTF8_MAX_FOLD_CHAR_EXPAND 3 | |
453 | ||
a98fe34d | 454 | #define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES) |
bd18bd40 KW |
455 | |
456 | /* | |
457 | ||
458 | =for apidoc Am|bool|DO_UTF8|SV* sv | |
459 | Returns a bool giving whether or not the PV in C<sv> is to be treated as being | |
460 | encoded in UTF-8. | |
461 | ||
462 | You should use this I<after> a call to C<SvPV()> or one of its variants, in | |
463 | case any call to string overloading updates the internal UTF-8 encoding flag. | |
464 | ||
465 | =cut | |
466 | */ | |
0064a8a9 | 467 | #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES) |
66cbab2c | 468 | #define IN_UNI_8_BIT \ |
d6ded950 KW |
469 | (((CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT)) \ |
470 | || (CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \ | |
471 | /* -1 below is for :not_characters */ \ | |
472 | && _is_in_locale_category(FALSE, -1))) \ | |
473 | && ! IN_BYTES) | |
b36bf33f | 474 | |
1d72bdf6 | 475 | |
c76687c5 KW |
476 | #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */ |
477 | ||
478 | /* Allow first byte to be a continuation byte */ | |
1d72bdf6 | 479 | #define UTF8_ALLOW_CONTINUATION 0x0002 |
c76687c5 KW |
480 | |
481 | /* Allow second... bytes to be non-continuation bytes */ | |
1d72bdf6 | 482 | #define UTF8_ALLOW_NON_CONTINUATION 0x0004 |
949cf498 KW |
483 | |
484 | /* expecting more bytes than were available in the string */ | |
485 | #define UTF8_ALLOW_SHORT 0x0008 | |
486 | ||
487 | /* Overlong sequence; i.e., the code point can be specified in fewer bytes. */ | |
488 | #define UTF8_ALLOW_LONG 0x0010 | |
489 | ||
490 | #define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */ | |
491 | #define UTF8_WARN_SURROGATE 0x0040 | |
492 | ||
493 | #define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */ | |
494 | #define UTF8_WARN_NONCHAR 0x0100 /* code points */ | |
495 | ||
496 | #define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */ | |
497 | #define UTF8_WARN_SUPER 0x0400 /* points above the legal max */ | |
498 | ||
499 | /* Code points which never were part of the original UTF-8 standard, the first | |
ea5ced44 KW |
500 | * byte of which is a FE or FF on ASCII platforms. If the first byte is FF, it |
501 | * will overflow a 32-bit word. If the first byte is FE, it will overflow a | |
502 | * signed 32-bit word. */ | |
949cf498 KW |
503 | #define UTF8_DISALLOW_FE_FF 0x0800 |
504 | #define UTF8_WARN_FE_FF 0x1000 | |
505 | ||
506 | #define UTF8_CHECK_ONLY 0x2000 | |
507 | ||
508 | /* For backwards source compatibility. They do nothing, as the default now | |
509 | * includes what they used to mean. The first one's meaning was to allow the | |
510 | * just the single non-character 0xFFFF */ | |
511 | #define UTF8_ALLOW_FFFF 0 | |
512 | #define UTF8_ALLOW_SURROGATE 0 | |
513 | ||
33d9abfb | 514 | #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF) |
949cf498 | 515 | #define UTF8_WARN_ILLEGAL_INTERCHANGE \ |
33d9abfb | 516 | (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF) |
949cf498 KW |
517 | #define UTF8_ALLOW_ANY \ |
518 | (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) | |
519 | #define UTF8_ALLOW_ANYUV \ | |
520 | (UTF8_ALLOW_EMPTY \ | |
521 | & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) | |
9f7f3913 TS |
522 | #define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \ |
523 | UTF8_ALLOW_ANYUV) | |
1d72bdf6 | 524 | |
7131f24d KW |
525 | /* Surrogates, non-character code points and above-Unicode code points are |
526 | * problematic in some contexts. This allows code that needs to check for | |
527 | * those to to quickly exclude the vast majority of code points it will | |
528 | * encounter */ | |
b96a92fb KW |
529 | #define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \ |
530 | FIRST_SURROGATE_UTF8_FIRST_BYTE | |
7131f24d | 531 | |
b96a92fb KW |
532 | #define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s)) |
533 | #define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send)) | |
7131f24d KW |
534 | |
535 | /* ASCII EBCDIC I8 | |
536 | * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode | |
537 | * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0 | |
538 | * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1 | |
a1776718 KW |
539 | * |
540 | * BE AWARE that this test doesn't rule out malformed code points, in | |
541 | * particular overlongs */ | |
7131f24d | 542 | #ifdef EBCDIC /* Both versions assume well-formed UTF8 */ |
bc3632a8 KW |
543 | # define UTF8_IS_SUPER(s) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \ |
544 | && (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \ | |
a3481822 | 545 | || (NATIVE_UTF8_TO_I8(* ((U8*) (s) + 1)) >= 0xA2))) |
7131f24d | 546 | #else |
0404c92a KW |
547 | # define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \ |
548 | && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))) | |
7131f24d KW |
549 | #endif |
550 | ||
b96a92fb KW |
551 | /* These are now machine generated, and the 'given' clause is no longer |
552 | * applicable */ | |
553 | #define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \ | |
554 | cBOOL(is_NONCHAR_utf8(s)) | |
555 | #define UTF8_IS_NONCHAR_(s) \ | |
556 | UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) | |
7131f24d | 557 | |
c867b360 JH |
558 | #define UNICODE_SURROGATE_FIRST 0xD800 |
559 | #define UNICODE_SURROGATE_LAST 0xDFFF | |
560 | #define UNICODE_REPLACEMENT 0xFFFD | |
561 | #define UNICODE_BYTE_ORDER_MARK 0xFEFF | |
1d72bdf6 | 562 | |
b851fbc1 | 563 | /* Though our UTF-8 encoding can go beyond this, |
c76687c5 | 564 | * let's be conservative and do as Unicode says. */ |
b851fbc1 JH |
565 | #define PERL_UNICODE_MAX 0x10FFFF |
566 | ||
949cf498 KW |
567 | #define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */ |
568 | #define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */ | |
569 | #define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */ | |
570 | #define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */ | |
571 | #define UNICODE_DISALLOW_SURROGATE 0x0010 | |
572 | #define UNICODE_DISALLOW_NONCHAR 0x0020 | |
573 | #define UNICODE_DISALLOW_SUPER 0x0040 | |
574 | #define UNICODE_DISALLOW_FE_FF 0x0080 | |
bb88be5f KW |
575 | #define UNICODE_WARN_ILLEGAL_INTERCHANGE \ |
576 | (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER) | |
577 | #define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \ | |
578 | (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER) | |
949cf498 KW |
579 | |
580 | /* For backward source compatibility, as are now the default */ | |
581 | #define UNICODE_ALLOW_SURROGATE 0 | |
582 | #define UNICODE_ALLOW_SUPER 0 | |
583 | #define UNICODE_ALLOW_ANY 0 | |
b851fbc1 | 584 | |
1d72bdf6 NIS |
585 | #define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \ |
586 | (c) <= UNICODE_SURROGATE_LAST) | |
a10ec373 | 587 | #define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT) |
872c91ae | 588 | #define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK) |
7131f24d KW |
589 | #define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \ |
590 | /* The other noncharacters end in FFFE or FFFF, which \ | |
591 | * the mask below catches both of, but beyond the last \ | |
592 | * official unicode code point, they aren't \ | |
593 | * noncharacters, since those aren't Unicode \ | |
594 | * characters at all */ \ | |
595 | || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c))) | |
596 | #define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX) | |
597 | #define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF) | |
1d72bdf6 | 598 | |
ec34087a KW |
599 | #define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE |
600 | #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \ | |
601 | LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE | |
602 | #define MICRO_SIGN MICRO_SIGN_NATIVE | |
603 | #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \ | |
604 | LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE | |
605 | #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \ | |
606 | LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE | |
09091399 JH |
607 | #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3 |
608 | #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2 | |
609 | #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3 | |
9dcbe121 | 610 | #define GREEK_SMALL_LETTER_MU 0x03BC |
9e682c18 KW |
611 | #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case |
612 | of MICRON */ | |
613 | #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */ | |
0766489e KW |
614 | #ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8 |
615 | # define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E | |
616 | #endif | |
74894415 KW |
617 | #define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130 |
618 | #define LATIN_SMALL_LETTER_DOTLESS_I 0x131 | |
9e682c18 | 619 | #define LATIN_SMALL_LETTER_LONG_S 0x017F |
a9f50d33 KW |
620 | #define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05 |
621 | #define LATIN_SMALL_LIGATURE_ST 0xFB06 | |
9e682c18 KW |
622 | #define KELVIN_SIGN 0x212A |
623 | #define ANGSTROM_SIGN 0x212B | |
09091399 | 624 | |
9e55ce06 | 625 | #define UNI_DISPLAY_ISPRINT 0x0001 |
c728cb41 JH |
626 | #define UNI_DISPLAY_BACKSLASH 0x0002 |
627 | #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) | |
628 | #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) | |
9e55ce06 | 629 | |
5cd46e1f KW |
630 | #define ANYOF_FOLD_SHARP_S(node, input, end) \ |
631 | (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \ | |
137165a6 | 632 | (ANYOF_NONBITMAP(node)) && \ |
39065660 | 633 | (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \ |
07b6858f | 634 | ((end) > (input) + 1) && \ |
305b8651 | 635 | isALPHA_FOLD_EQ((input)[0], 's')) |
6302f837 | 636 | |
ebc501f0 | 637 | #define SHARP_S_SKIP 2 |
3b0fc154 | 638 | |
a4f7a67c KW |
639 | /* If you want to exclude surrogates, and beyond legal Unicode, see the blame |
640 | * log for earlier versions which gave details for these */ | |
4d646140 | 641 | |
6302f837 KW |
642 | /* A helper macro for isUTF8_CHAR, so use that one, and not this one. This is |
643 | * retained solely for backwards compatibility and may be deprecated and | |
644 | * removed in a future Perl version. | |
645 | * | |
646 | * regen/regcharclass.pl generates is_UTF8_CHAR_utf8() macros for up to these | |
d9f92374 KW |
647 | * number of bytes. So this has to be coordinated with that file */ |
648 | #ifdef EBCDIC | |
649 | # define IS_UTF8_CHAR_FAST(n) ((n) <= 3) | |
650 | #else | |
651 | # define IS_UTF8_CHAR_FAST(n) ((n) <= 4) | |
652 | #endif | |
653 | ||
4d646140 | 654 | #ifndef EBCDIC |
6302f837 KW |
655 | /* A helper macro for isUTF8_CHAR, so use that one instead of this. This was |
656 | * generated by regen/regcharclass.pl, and then moved here. The lines that | |
657 | * generated it were then commented out. This was done solely because it takes | |
658 | * on the order of 10 minutes to generate, and is never going to change, unless | |
659 | * the generated code is improved. | |
660 | * | |
661 | * The EBCDIC versions have been cut to not cover all of legal Unicode, so | |
662 | * don't take too long to generate, and there is a separate one for each code | |
663 | * page, so they are in regcharclass.h instead of here */ | |
39a0f513 | 664 | /* |
5dca9278 | 665 | UTF8_CHAR: Matches legal UTF-8 encoded characters from 2 through 4 bytes |
39a0f513 | 666 | |
5dca9278 | 667 | 0x80 - 0x1FFFFF |
39a0f513 | 668 | */ |
4d646140 | 669 | /*** GENERATED CODE ***/ |
5dca9278 KW |
670 | #define is_UTF8_CHAR_utf8_no_length_checks(s) \ |
671 | ( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \ | |
672 | ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \ | |
673 | : ( 0xE0 == ((U8*)s)[0] ) ? \ | |
674 | ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ | |
675 | : ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \ | |
676 | ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ | |
677 | : ( 0xF0 == ((U8*)s)[0] ) ? \ | |
678 | ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\ | |
679 | : ( ( ( ( 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 | 680 | #endif |
3b0fc154 | 681 | |
6302f837 | 682 | /* |
5dca9278 KW |
683 | |
684 | =for apidoc Am|STRLEN|isUTF8_CHAR|const U8 *s|const U8 *e | |
685 | ||
686 | Returns the number of bytes beginning at C<s> which form a legal UTF-8 (or | |
687 | UTF-EBCDIC) encoded character, looking no further than C<e - s> bytes into | |
688 | C<s>. Returns 0 if the sequence starting at C<s> through C<e - 1> is not | |
689 | well-formed UTF-8 | |
6302f837 KW |
690 | |
691 | Note that an INVARIANT character (i.e. ASCII on non-EBCDIC | |
5dca9278 KW |
692 | machines) is a valid UTF-8 character. |
693 | ||
694 | =cut | |
695 | */ | |
6302f837 | 696 | |
dd9bc2b0 | 697 | #define isUTF8_CHAR(s, e) (UNLIKELY((e) <= (s)) \ |
6302f837 KW |
698 | ? 0 \ |
699 | : (UTF8_IS_INVARIANT(*s)) \ | |
700 | ? 1 \ | |
dd9bc2b0 | 701 | : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \ |
6302f837 | 702 | ? 0 \ |
dd9bc2b0 | 703 | : LIKELY(IS_UTF8_CHAR_FAST(UTF8SKIP(s))) \ |
5dca9278 | 704 | ? is_UTF8_CHAR_utf8_no_length_checks(s) \ |
6302f837 KW |
705 | : _is_utf8_char_slow(s, e)) |
706 | ||
3cedd9d9 KW |
707 | #define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end) |
708 | ||
6302f837 KW |
709 | /* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is |
710 | * retained solely for backwards compatibility */ | |
711 | #define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n) | |
e9a8c099 | 712 | |
57f0e7e2 KW |
713 | #endif /* H_UTF8 */ |
714 | ||
e9a8c099 | 715 | /* |
14d04a33 | 716 | * ex: set ts=8 sts=4 sw=4 et: |
e9a8c099 | 717 | */ |