1 /* utf8.h 2 * 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 * 9 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009, 10 * 2010, 2011 by Larry Wall and others 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 17 #ifndef PERL_UTF8_H_ /* Guard against recursive inclusion */ 18 #define PERL_UTF8_H_ 1 19 20 /* Use UTF-8 as the default script encoding? 21 * Turning this on will break scripts having non-UTF-8 binary 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 29 #include "regcharclass.h" 30 #include "unicode_constants.h" 31 32 /* For to_utf8_fold_flags, q.v. */ 33 #define FOLD_FLAGS_LOCALE 0x1 34 #define FOLD_FLAGS_FULL 0x2 35 #define FOLD_FLAGS_NOMIX_ASCII 0x4 36 37 /* 38 =head1 Unicode Support 39 L<perlguts/Unicode Support> has an introduction to this API. 40 41 See also L</Character classification>, 42 and L</Character case changing>. 43 Various functions outside this section also work specially with Unicode. 44 Search for the string "utf8" in this document. 45 46 =for apidoc is_ascii_string 47 48 This is a misleadingly-named synonym for L</is_utf8_invariant_string>. 49 On ASCII-ish platforms, the name isn't misleading: the ASCII-range characters 50 are exactly the UTF-8 invariants. But EBCDIC machines have more invariants 51 than just the ASCII characters, so C<is_utf8_invariant_string> is preferred. 52 53 =for apidoc is_invariant_string 54 55 This is a somewhat misleadingly-named synonym for L</is_utf8_invariant_string>. 56 C<is_utf8_invariant_string> is preferred, as it indicates under what conditions 57 the string is invariant. 58 59 =cut 60 */ 61 #define is_ascii_string(s, len) is_utf8_invariant_string(s, len) 62 #define is_invariant_string(s, len) is_utf8_invariant_string(s, len) 63 64 #define uvoffuni_to_utf8_flags(d,uv,flags) \ 65 uvoffuni_to_utf8_flags_msgs(d, uv, flags, 0) 66 #define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0) 67 #define uvchr_to_utf8_flags(d,uv,flags) \ 68 uvchr_to_utf8_flags_msgs(d,uv,flags, 0) 69 #define uvchr_to_utf8_flags_msgs(d,uv,flags,msgs) \ 70 uvoffuni_to_utf8_flags_msgs(d,NATIVE_TO_UNI(uv),flags, msgs) 71 #define utf8_to_uvchr_buf(s, e, lenp) \ 72 (__ASSERT_((U8*) (e) > (U8*) (s)) \ 73 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \ 74 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY)) 75 #define utf8n_to_uvchr(s, len, lenp, flags) \ 76 utf8n_to_uvchr_error(s, len, lenp, flags, 0) 77 #define utf8n_to_uvchr_error(s, len, lenp, flags, errors) \ 78 utf8n_to_uvchr_msgs(s, len, lenp, flags, errors, 0) 79 80 #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) 81 82 #define to_utf8_fold(s, r, lenr) \ 83 _to_utf8_fold_flags (s, NULL, r, lenr, FOLD_FLAGS_FULL, __FILE__, __LINE__) 84 #define to_utf8_lower(s, r, lenr) \ 85 _to_utf8_lower_flags(s, NULL, r ,lenr, 0, __FILE__, __LINE__) 86 #define to_utf8_upper(s, r, lenr) \ 87 _to_utf8_upper_flags(s, NULL, r, lenr, 0, __FILE__, __LINE__) 88 #define to_utf8_title(s, r, lenr) \ 89 _to_utf8_title_flags(s, NULL, r, lenr ,0, __FILE__, __LINE__) 90 91 #define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ 92 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0) 93 #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0) 94 #define FOLDEQ_LOCALE (1 << 1) 95 #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2) 96 #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3) 97 #define FOLDEQ_S1_FOLDS_SANE (1 << 4) 98 #define FOLDEQ_S2_FOLDS_SANE (1 << 5) 99 100 #define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ 101 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2)) 102 103 #ifdef EBCDIC 104 /* The equivalent of these macros but implementing UTF-EBCDIC 105 are in the following header file: 106 */ 107 108 #include "utfebcdic.h" 109 110 #else /* ! EBCDIC */ 111 START_EXTERN_C 112 113 /* How wide can a single UTF-8 encoded character become in bytes. */ 114 /* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8 115 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be 116 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode 117 * non-negative integers in a binary format, even those above Unicode */ 118 #define UTF8_MAXBYTES 13 119 120 #ifdef DOINIT 121 EXTCONST unsigned char PL_utf8skip[] = { 122 /* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 123 /* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 124 /* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 125 /* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 126 /* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 127 /* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 128 /* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 129 /* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 130 /* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 131 /* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 132 /* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 133 /* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 134 /* 0xC0 */ 2,2, /* overlong */ 135 /* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */ 136 /* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */ 137 /* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */ 138 /* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */ 139 /* Perl extended (never was official UTF-8). Up to 36 bit */ 140 /* 0xFE */ 7, 141 /* More extended, Up to 72 bits (64-bit + reserved) */ 142 /* 0xFF */ UTF8_MAXBYTES 143 }; 144 #else 145 EXTCONST unsigned char PL_utf8skip[]; 146 #endif 147 148 END_EXTERN_C 149 150 #if defined(_MSC_VER) && _MSC_VER < 1400 151 /* older MSVC versions have a smallish macro buffer */ 152 #define PERL_SMALL_MACRO_BUFFER 153 #endif 154 155 /* Native character to/from iso-8859-1. Are the identity functions on ASCII 156 * platforms */ 157 #ifdef PERL_SMALL_MACRO_BUFFER 158 #define NATIVE_TO_LATIN1(ch) ((U8)(ch)) 159 #define LATIN1_TO_NATIVE(ch) ((U8)(ch)) 160 #else 161 #define NATIVE_TO_LATIN1(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch))) 162 #define LATIN1_TO_NATIVE(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch))) 163 #endif 164 165 /* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus 166 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking 167 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them 168 * because they are 8-bit encodings that serve the same purpose in Perl, and 169 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to 170 * whichever one is applicable on the current platform */ 171 #ifdef PERL_SMALL_MACRO_BUFFER 172 #define NATIVE_UTF8_TO_I8(ch) ((U8) (ch)) 173 #define I8_TO_NATIVE_UTF8(ch) ((U8) (ch)) 174 #else 175 #define NATIVE_UTF8_TO_I8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch))) 176 #define I8_TO_NATIVE_UTF8(ch) (__ASSERT_(FITS_IN_8_BITS(ch)) ((U8) (ch))) 177 #endif 178 179 /* Transforms in wide UV chars */ 180 #define UNI_TO_NATIVE(ch) ((UV) (ch)) 181 #define NATIVE_TO_UNI(ch) ((UV) (ch)) 182 183 /* 184 185 The following table is from Unicode 3.2, plus the Perl extensions for above 186 U+10FFFF 187 188 Code Points 1st Byte 2nd Byte 3rd 4th 5th 6th 7th 8th-13th 189 190 U+0000..U+007F 00..7F 191 U+0080..U+07FF * C2..DF 80..BF 192 U+0800..U+0FFF E0 * A0..BF 80..BF 193 U+1000..U+CFFF E1..EC 80..BF 80..BF 194 U+D000..U+D7FF ED 80..9F 80..BF 195 U+D800..U+DFFF ED A0..BF 80..BF (surrogates) 196 U+E000..U+FFFF EE..EF 80..BF 80..BF 197 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF 198 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF 199 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF 200 Below are above-Unicode code points 201 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF 202 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF 203 U+200000..U+FFFFFF F8 * 88..BF 80..BF 80..BF 80..BF 204 U+1000000..U+3FFFFFF F9..FB 80..BF 80..BF 80..BF 80..BF 205 U+4000000..U+3FFFFFFF FC * 84..BF 80..BF 80..BF 80..BF 80..BF 206 U+40000000..U+7FFFFFFF FD 80..BF 80..BF 80..BF 80..BF 80..BF 207 U+80000000..U+FFFFFFFFF FE * 82..BF 80..BF 80..BF 80..BF 80..BF 80..BF 208 U+1000000000.. FF 80..BF 80..BF 80..BF 80..BF 80..BF * 81..BF 80..BF 209 210 Note the gaps before several of the byte entries above marked by '*'. These are 211 caused by legal UTF-8 avoiding non-shortest encodings: it is technically 212 possible to UTF-8-encode a single code point in different ways, but that is 213 explicitly forbidden, and the shortest possible encoding should always be used 214 (and that is what Perl does). The non-shortest ones are called 'overlongs'. 215 216 */ 217 218 /* 219 Another way to look at it, as bits: 220 221 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte 222 223 0aaa aaaa 0aaa aaaa 224 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa 225 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa 226 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa 227 228 As you can see, the continuation bytes all begin with C<10>, and the 229 leading bits of the start byte tell how many bytes there are in the 230 encoded character. 231 232 Perl's extended UTF-8 means we can have start bytes up through FF, though any 233 beginning with FF yields a code point that is too large for 32-bit ASCII 234 platforms. FF signals to use 13 bytes for the encoded character. This breaks 235 the paradigm that the number of leading bits gives how many total bytes there 236 are in the character. 237 238 */ 239 240 /* Is the representation of the Unicode code point 'cp' the same regardless of 241 * being encoded in UTF-8 or not? */ 242 #define OFFUNI_IS_INVARIANT(cp) isASCII(cp) 243 244 /* 245 =for apidoc Am|bool|UVCHR_IS_INVARIANT|UV cp 246 247 Evaluates to 1 if the representation of code point C<cp> is the same whether or 248 not it is encoded in UTF-8; otherwise evaluates to 0. UTF-8 invariant 249 characters can be copied as-is when converting to/from UTF-8, saving time. 250 C<cp> is Unicode if above 255; otherwise is platform-native. 251 252 =cut 253 */ 254 255 #define UVCHR_IS_INVARIANT(cp) OFFUNI_IS_INVARIANT(cp) 256 257 /* This defines the bits that are to be in the continuation bytes of a multi-byte 258 * UTF-8 encoded character that mark it is a continuation byte. */ 259 #define UTF_CONTINUATION_MARK 0x80 260 261 /* Misleadingly named: is the UTF8-encoded byte 'c' part of a variant sequence 262 * in UTF-8? This is the inverse of UTF8_IS_INVARIANT. The |0 makes sure this 263 * isn't mistakenly called with a ptr argument */ 264 #define UTF8_IS_CONTINUED(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 265 ((U8)((c) | 0)) & UTF_CONTINUATION_MARK) 266 267 /* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence? 268 * This doesn't catch invariants (they are single-byte). It also excludes the 269 * illegal overlong sequences that begin with C0 and C1. The |0 makes sure 270 * this isn't mistakenly called with a ptr argument */ 271 #define UTF8_IS_START(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 272 ((U8)((c) | 0)) >= 0xc2) 273 274 /* For use in UTF8_IS_CONTINUATION() below */ 275 #define UTF_IS_CONTINUATION_MASK 0xC0 276 277 /* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the 278 * first byte thereof? The |0 makes sure this isn't mistakenly called with a 279 * ptr argument */ 280 #define UTF8_IS_CONTINUATION(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 281 (((U8)((c) | 0)) & UTF_IS_CONTINUATION_MASK) == UTF_CONTINUATION_MARK) 282 283 /* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use 284 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to 285 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus 286 * this matches 0xc[23]. The |0 makes sure this isn't mistakenly called with a 287 * ptr argument */ 288 #define UTF8_IS_DOWNGRADEABLE_START(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 289 (((U8)((c) | 0)) & 0xfe) == 0xc2) 290 291 /* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that 292 * represent a code point > 255? The |0 makes sure this isn't mistakenly 293 * called with a ptr argument */ 294 #define UTF8_IS_ABOVE_LATIN1(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 295 ((U8)((c) | 0)) >= 0xc4) 296 297 /* This is the number of low-order bits a continuation byte in a UTF-8 encoded 298 * sequence contributes to the specification of the code point. In the bit 299 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of 300 * real information */ 301 #define UTF_ACCUMULATION_SHIFT 6 302 303 /* ^? is defined to be DEL on ASCII systems. See the definition of toCTRL() 304 * for more */ 305 #define QUESTION_MARK_CTRL DEL_NATIVE 306 307 /* Surrogates, non-character code points and above-Unicode code points are 308 * problematic in some contexts. This allows code that needs to check for 309 * those to to quickly exclude the vast majority of code points it will 310 * encounter */ 311 #define isUTF8_POSSIBLY_PROBLEMATIC(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 312 (U8) c >= 0xED) 313 314 #define UNICODE_IS_PERL_EXTENDED(uv) UNLIKELY((UV) (uv) > 0x7FFFFFFF) 315 316 #endif /* EBCDIC vs ASCII */ 317 318 /* 2**UTF_ACCUMULATION_SHIFT - 1 */ 319 #define UTF_CONTINUATION_MASK ((U8) ((1U << UTF_ACCUMULATION_SHIFT) - 1)) 320 321 /* Internal macro to be used only in this file to aid in constructing other 322 * publicly accessible macros. 323 * The number of bytes required to express this uv in UTF-8, for just those 324 * uv's requiring 2 through 6 bytes, as these are common to all platforms and 325 * word sizes. The number of bytes needed is given by the number of leading 1 326 * bits in the start byte. There are 32 start bytes that have 2 initial 1 bits 327 * (C0-DF); there are 16 that have 3 initial 1 bits (E0-EF); 8 that have 4 328 * initial 1 bits (F0-F8); 4 that have 5 initial 1 bits (F9-FB), and 2 that 329 * have 6 initial 1 bits (FC-FD). The largest number a string of n bytes can 330 * represent is (the number of possible start bytes for 'n') 331 * * (the number of possiblities for each start byte 332 * The latter in turn is 333 * 2 ** ( (how many continuation bytes there are) 334 * * (the number of bits of information each 335 * continuation byte holds)) 336 * 337 * If we were on a platform where we could use a fast find first set bit 338 * instruction (or count leading zeros instruction) this could be replaced by 339 * using that to find the log2 of the uv, and divide that by the number of bits 340 * of information in each continuation byte, adjusting for large cases and how 341 * much information is in a start byte for that length */ 342 #define __COMMON_UNI_SKIP(uv) \ 343 (UV) (uv) < (32 * (1U << ( UTF_ACCUMULATION_SHIFT))) ? 2 : \ 344 (UV) (uv) < (16 * (1U << (2 * UTF_ACCUMULATION_SHIFT))) ? 3 : \ 345 (UV) (uv) < ( 8 * (1U << (3 * UTF_ACCUMULATION_SHIFT))) ? 4 : \ 346 (UV) (uv) < ( 4 * (1U << (4 * UTF_ACCUMULATION_SHIFT))) ? 5 : \ 347 (UV) (uv) < ( 2 * (1U << (5 * UTF_ACCUMULATION_SHIFT))) ? 6 : 348 349 /* Internal macro to be used only in this file. 350 * This adds to __COMMON_UNI_SKIP the details at this platform's upper range. 351 * For any-sized EBCDIC platforms, or 64-bit ASCII ones, we need one more test 352 * to see if just 7 bytes is needed, or if the maximum is needed. For 32-bit 353 * ASCII platforms, everything is representable by 7 bytes */ 354 #if defined(UV_IS_QUAD) || defined(EBCDIC) 355 # define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) \ 356 (UV) (uv) < ((UV) 1U << (6 * UTF_ACCUMULATION_SHIFT)) ? 7 : UTF8_MAXBYTES) 357 #else 358 # define __BASE_UNI_SKIP(uv) (__COMMON_UNI_SKIP(uv) 7) 359 #endif 360 361 /* The next two macros use the base macro defined above, and add in the tests 362 * at the low-end of the range, for just 1 byte, yielding complete macros, 363 * publicly accessible. */ 364 365 /* Input is a true Unicode (not-native) code point */ 366 #define OFFUNISKIP(uv) (OFFUNI_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv)) 367 368 /* 369 370 =for apidoc Am|STRLEN|UVCHR_SKIP|UV cp 371 returns the number of bytes required to represent the code point C<cp> when 372 encoded as UTF-8. C<cp> is a native (ASCII or EBCDIC) code point if less than 373 255; a Unicode code point otherwise. 374 375 =cut 376 */ 377 #define UVCHR_SKIP(uv) ( UVCHR_IS_INVARIANT(uv) ? 1 : __BASE_UNI_SKIP(uv)) 378 379 /* The largest code point representable by two UTF-8 bytes on this platform. 380 * As explained in the comments for __COMMON_UNI_SKIP, 32 start bytes with 381 * UTF_ACCUMULATION_SHIFT bits of information each */ 382 #define MAX_UTF8_TWO_BYTE (32 * (1U << UTF_ACCUMULATION_SHIFT) - 1) 383 384 /* The largest code point representable by two UTF-8 bytes on any platform that 385 * Perl runs on. This value is constrained by EBCDIC which has 5 bits per 386 * continuation byte */ 387 #define MAX_PORTABLE_UTF8_TWO_BYTE (32 * (1U << 5) - 1) 388 389 /* The maximum number of UTF-8 bytes a single Unicode character can 390 * uppercase/lowercase/fold into. Unicode guarantees that the maximum 391 * expansion is UTF8_MAX_FOLD_CHAR_EXPAND characters, but any above-Unicode 392 * code point will fold to itself, so we only have to look at the expansion of 393 * the maximum Unicode code point. But this number may be less than the space 394 * occupied by a very large code point under Perl's extended UTF-8. We have to 395 * make it large enough to fit any single character. (It turns out that ASCII 396 * and EBCDIC differ in which is larger) */ 397 #define UTF8_MAXBYTES_CASE \ 398 (UTF8_MAXBYTES >= (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF)) \ 399 ? UTF8_MAXBYTES \ 400 : (UTF8_MAX_FOLD_CHAR_EXPAND * OFFUNISKIP(0x10FFFF))) 401 402 /* Rest of these are attributes of Unicode and perl's internals rather than the 403 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at 404 * this level; the macros that some of these call may have different 405 * definitions in the two encodings */ 406 407 /* In domain restricted to ASCII, these may make more sense to the reader than 408 * the ones with Latin1 in the name */ 409 #define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch) 410 #define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch) 411 412 /* More or less misleadingly-named defines, retained for back compat */ 413 #define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch) 414 #define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch) 415 #define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch) 416 #define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch) 417 #define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch) 418 419 /* This defines the 1-bits that are to be in the first byte of a multi-byte 420 * UTF-8 encoded character that mark it as a start byte and give the number of 421 * bytes that comprise the character. 'len' is the number of bytes in the 422 * multi-byte sequence. */ 423 #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len))))) 424 425 /* Masks out the initial one bits in a start byte, leaving the real data ones. 426 * Doesn't work on an invariant byte. 'len' is the number of bytes in the 427 * multi-byte sequence that comprises the character. */ 428 #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2))) 429 430 /* Adds a UTF8 continuation byte 'new' of information to a running total code 431 * point 'old' of all the continuation bytes so far. This is designed to be 432 * used in a loop to convert from UTF-8 to the code point represented. Note 433 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is 434 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC) 435 * code point in process of being generated */ 436 #define UTF8_ACCUMULATE(old, new) (__ASSERT_(FITS_IN_8_BITS(new)) \ 437 ((old) << UTF_ACCUMULATION_SHIFT) \ 438 | ((NATIVE_UTF8_TO_I8((U8)new)) \ 439 & UTF_CONTINUATION_MASK)) 440 441 /* This works in the face of malformed UTF-8. */ 442 #define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) \ 443 ( UTF8_IS_DOWNGRADEABLE_START(*(s)) \ 444 && ( (e) - (s) > 1) \ 445 && UTF8_IS_CONTINUATION(*((s)+1))) 446 447 /* Number of bytes a code point occupies in UTF-8. */ 448 #define NATIVE_SKIP(uv) UVCHR_SKIP(uv) 449 450 /* Most code which says UNISKIP is really thinking in terms of native code 451 * points (0-255) plus all those beyond. This is an imprecise term, but having 452 * it means existing code continues to work. For precision, use UVCHR_SKIP, 453 * NATIVE_SKIP, or OFFUNISKIP */ 454 #define UNISKIP(uv) UVCHR_SKIP(uv) 455 456 /* Longer, but more accurate name */ 457 #define UTF8_IS_ABOVE_LATIN1_START(c) UTF8_IS_ABOVE_LATIN1(c) 458 459 /* Convert a UTF-8 variant Latin1 character to a native code point value. 460 * Needs just one iteration of accumulate. Should be used only if it is known 461 * that the code point is < 256, and is not UTF-8 invariant. Use the slower 462 * but more general TWO_BYTE_UTF8_TO_NATIVE() which handles any code point 463 * representable by two bytes (which turns out to be up through 464 * MAX_PORTABLE_UTF8_TWO_BYTE). The two parameters are: 465 * HI: a downgradable start byte; 466 * LO: continuation. 467 * */ 468 #define EIGHT_BIT_UTF8_TO_NATIVE(HI, LO) \ 469 ( __ASSERT_(UTF8_IS_DOWNGRADEABLE_START(HI)) \ 470 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \ 471 LATIN1_TO_NATIVE(UTF8_ACCUMULATE(( \ 472 NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), (LO)))) 473 474 /* Convert a two (not one) byte utf8 character to a native code point value. 475 * Needs just one iteration of accumulate. Should not be used unless it is 476 * known that the two bytes are legal: 1) two-byte start, and 2) continuation. 477 * Note that the result can be larger than 255 if the input character is not 478 * downgradable */ 479 #define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \ 480 (__ASSERT_(FITS_IN_8_BITS(HI)) \ 481 __ASSERT_(FITS_IN_8_BITS(LO)) \ 482 __ASSERT_(PL_utf8skip[HI] == 2) \ 483 __ASSERT_(UTF8_IS_CONTINUATION(LO)) \ 484 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \ 485 (LO)))) 486 487 /* Should never be used, and be deprecated */ 488 #define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO)) 489 490 /* 491 492 =for apidoc Am|STRLEN|UTF8SKIP|char* s 493 returns the number of bytes in the UTF-8 encoded character whose first (perhaps 494 only) byte is pointed to by C<s>. 495 496 =cut 497 */ 498 #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)] 499 #define UTF8_SKIP(s) UTF8SKIP(s) 500 501 /* 502 503 =for apidoc Am|STRLEN|UTF8_SAFE_SKIP|char* s|char* e 504 returns 0 if S<C<s E<gt>= e>>; otherwise returns the number of bytes in the 505 UTF-8 encoded character whose first byte is pointed to by C<s>. But it never 506 returns beyond C<e>. On DEBUGGING builds, it asserts that S<C<s E<lt>= e>>. 507 508 =cut 509 */ 510 #define UTF8_SAFE_SKIP(s, e) (__ASSERT_((e) >= (s)) \ 511 ((e) - (s)) <= 0 \ 512 ? 0 \ 513 : MIN(((e) - (s)), UTF8_SKIP(s))) 514 515 /* Most code that says 'UNI_' really means the native value for code points up 516 * through 255 */ 517 #define UNI_IS_INVARIANT(cp) UVCHR_IS_INVARIANT(cp) 518 519 /* 520 =for apidoc Am|bool|UTF8_IS_INVARIANT|char c 521 522 Evaluates to 1 if the byte C<c> represents the same character when encoded in 523 UTF-8 as when not; otherwise evaluates to 0. UTF-8 invariant characters can be 524 copied as-is when converting to/from UTF-8, saving time. 525 526 In spite of the name, this macro gives the correct result if the input string 527 from which C<c> comes is not encoded in UTF-8. 528 529 See C<L</UVCHR_IS_INVARIANT>> for checking if a UV is invariant. 530 531 =cut 532 533 The reason it works on both UTF-8 encoded strings and non-UTF-8 encoded, is 534 that it returns TRUE in each for the exact same set of bit patterns. It is 535 valid on a subset of what UVCHR_IS_INVARIANT is valid on, so can just use that; 536 and the compiler should optimize out anything extraneous given the 537 implementation of the latter. The |0 makes sure this isn't mistakenly called 538 with a ptr argument. 539 */ 540 #define UTF8_IS_INVARIANT(c) UVCHR_IS_INVARIANT((c) | 0) 541 542 /* Like the above, but its name implies a non-UTF8 input, which as the comments 543 * above show, doesn't matter as to its implementation */ 544 #define NATIVE_BYTE_IS_INVARIANT(c) UVCHR_IS_INVARIANT(c) 545 546 /* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic 547 * bytes from an ordinal that is known to fit into exactly two (not one) bytes; 548 * it must be less than 0x3FF to work across both encodings. */ 549 550 /* These two are helper macros for the other three sets, and should not be used 551 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1 552 * (which works for code points up through 0xFF) or NATIVE_TO_UNI which works 553 * for any code point */ 554 #define __BASE_TWO_BYTE_HI(c, translate_function) \ 555 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \ 556 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \ 557 | UTF_START_MARK(2))) 558 #define __BASE_TWO_BYTE_LO(c, translate_function) \ 559 (__ASSERT_(! UVCHR_IS_INVARIANT(c)) \ 560 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \ 561 | UTF_CONTINUATION_MARK)) 562 563 /* The next two macros should not be used. They were designed to be usable as 564 * the case label of a switch statement, but this doesn't work for EBCDIC. Use 565 * regen/unicode_constants.pl instead */ 566 #define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI) 567 #define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI) 568 569 /* The next two macros are used when the source should be a single byte 570 * character; checked for under DEBUGGING */ 571 #define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 572 ( __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1))) 573 #define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 574 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1))) 575 576 /* These final two macros in the series are used when the source can be any 577 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient 578 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~' 579 * operator instead of "<=" to avoid getting compiler warnings. 580 * MAX_UTF8_TWO_BYTE should be exactly all one bits in the lower few 581 * places, so the ~ works */ 582 #define UTF8_TWO_BYTE_HI(c) \ 583 (__ASSERT_((sizeof(c) == 1) \ 584 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \ 585 (__BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI))) 586 #define UTF8_TWO_BYTE_LO(c) \ 587 (__ASSERT_((sizeof(c) == 1) \ 588 || !(((WIDEST_UTYPE)(c)) & ~MAX_UTF8_TWO_BYTE)) \ 589 (__BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI))) 590 591 /* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII 592 * as it is only in overlongs. */ 593 #define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1) 594 595 /* 596 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem 597 * from an earlier version of these macros in which they didn't call the 598 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the 599 * beginning of a utf8 character. Now that foo_utf8() determines that itself, 600 * no need to do it again here 601 */ 602 #define isIDFIRST_lazy_if(p,UTF) \ 603 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isIDFIRST_lazy_if", \ 604 "isIDFIRST_lazy_if_safe", \ 605 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__) 606 607 #define isIDFIRST_lazy_if_safe(p, e, UTF) \ 608 ((IN_BYTES || !UTF) \ 609 ? isIDFIRST(*(p)) \ 610 : isIDFIRST_utf8_safe(p, e)) 611 612 #define isWORDCHAR_lazy_if(p,UTF) \ 613 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isWORDCHAR_lazy_if", \ 614 "isWORDCHAR_lazy_if_safe", \ 615 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__) 616 617 #define isWORDCHAR_lazy_if_safe(p, e, UTF) \ 618 ((IN_BYTES || !UTF) \ 619 ? isWORDCHAR(*(p)) \ 620 : isWORDCHAR_utf8_safe((U8 *) p, (U8 *) e)) 621 622 #define isALNUM_lazy_if(p,UTF) \ 623 _is_utf8_FOO(_CC_IDFIRST, (const U8 *) p, "isALNUM_lazy_if", \ 624 "isWORDCHAR_lazy_if_safe", \ 625 cBOOL(UTF && ! IN_BYTES), 0, __FILE__,__LINE__) 626 627 #define UTF8_MAXLEN UTF8_MAXBYTES 628 629 /* A Unicode character can fold to up to 3 characters */ 630 #define UTF8_MAX_FOLD_CHAR_EXPAND 3 631 632 #define IN_BYTES UNLIKELY(CopHINTS_get(PL_curcop) & HINT_BYTES) 633 634 /* 635 636 =for apidoc Am|bool|DO_UTF8|SV* sv 637 Returns a bool giving whether or not the PV in C<sv> is to be treated as being 638 encoded in UTF-8. 639 640 You should use this I<after> a call to C<SvPV()> or one of its variants, in 641 case any call to string overloading updates the internal UTF-8 encoding flag. 642 643 =cut 644 */ 645 #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES) 646 647 /* Should all strings be treated as Unicode, and not just UTF-8 encoded ones? 648 * Is so within 'feature unicode_strings' or 'locale :not_characters', and not 649 * within 'use bytes'. UTF-8 locales are not tested for here, but perhaps 650 * could be */ 651 #define IN_UNI_8_BIT \ 652 (( ( (CopHINTS_get(PL_curcop) & HINT_UNI_8_BIT)) \ 653 || ( CopHINTS_get(PL_curcop) & HINT_LOCALE_PARTIAL \ 654 /* -1 below is for :not_characters */ \ 655 && _is_in_locale_category(FALSE, -1))) \ 656 && (! IN_BYTES)) 657 658 659 #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */ 660 #define UTF8_GOT_EMPTY UTF8_ALLOW_EMPTY 661 662 /* Allow first byte to be a continuation byte */ 663 #define UTF8_ALLOW_CONTINUATION 0x0002 664 #define UTF8_GOT_CONTINUATION UTF8_ALLOW_CONTINUATION 665 666 /* Unexpected non-continuation byte */ 667 #define UTF8_ALLOW_NON_CONTINUATION 0x0004 668 #define UTF8_GOT_NON_CONTINUATION UTF8_ALLOW_NON_CONTINUATION 669 670 /* expecting more bytes than were available in the string */ 671 #define UTF8_ALLOW_SHORT 0x0008 672 #define UTF8_GOT_SHORT UTF8_ALLOW_SHORT 673 674 /* Overlong sequence; i.e., the code point can be specified in fewer bytes. 675 * First one will convert the overlong to the REPLACEMENT CHARACTER; second 676 * will return what the overlong evaluates to */ 677 #define UTF8_ALLOW_LONG 0x0010 678 #define UTF8_ALLOW_LONG_AND_ITS_VALUE (UTF8_ALLOW_LONG|0x0020) 679 #define UTF8_GOT_LONG UTF8_ALLOW_LONG 680 681 #define UTF8_ALLOW_OVERFLOW 0x0080 682 #define UTF8_GOT_OVERFLOW UTF8_ALLOW_OVERFLOW 683 684 #define UTF8_DISALLOW_SURROGATE 0x0100 /* Unicode surrogates */ 685 #define UTF8_GOT_SURROGATE UTF8_DISALLOW_SURROGATE 686 #define UTF8_WARN_SURROGATE 0x0200 687 688 /* Unicode non-character code points */ 689 #define UTF8_DISALLOW_NONCHAR 0x0400 690 #define UTF8_GOT_NONCHAR UTF8_DISALLOW_NONCHAR 691 #define UTF8_WARN_NONCHAR 0x0800 692 693 /* Super-set of Unicode: code points above the legal max */ 694 #define UTF8_DISALLOW_SUPER 0x1000 695 #define UTF8_GOT_SUPER UTF8_DISALLOW_SUPER 696 #define UTF8_WARN_SUPER 0x2000 697 698 /* The original UTF-8 standard did not define UTF-8 with start bytes of 0xFE or 699 * 0xFF, though UTF-EBCDIC did. This allowed both versions to represent code 700 * points up to 2 ** 31 - 1. Perl extends UTF-8 so that 0xFE and 0xFF are 701 * usable on ASCII platforms, and 0xFF means something different than 702 * UTF-EBCDIC defines. These changes allow code points of 64 bits (actually 703 * somewhat more) to be represented on both platforms. But these are Perl 704 * extensions, and not likely to be interchangeable with other languages. Note 705 * that on ASCII platforms, FE overflows a signed 32-bit word, and FF an 706 * unsigned one. */ 707 #define UTF8_DISALLOW_PERL_EXTENDED 0x4000 708 #define UTF8_GOT_PERL_EXTENDED UTF8_DISALLOW_PERL_EXTENDED 709 #define UTF8_WARN_PERL_EXTENDED 0x8000 710 711 /* For back compat, these old names are misleading for overlongs and 712 * UTF_EBCDIC. */ 713 #define UTF8_DISALLOW_ABOVE_31_BIT UTF8_DISALLOW_PERL_EXTENDED 714 #define UTF8_GOT_ABOVE_31_BIT UTF8_GOT_PERL_EXTENDED 715 #define UTF8_WARN_ABOVE_31_BIT UTF8_WARN_PERL_EXTENDED 716 #define UTF8_DISALLOW_FE_FF UTF8_DISALLOW_PERL_EXTENDED 717 #define UTF8_WARN_FE_FF UTF8_WARN_PERL_EXTENDED 718 719 #define UTF8_CHECK_ONLY 0x10000 720 #define _UTF8_NO_CONFIDENCE_IN_CURLEN 0x20000 /* Internal core use only */ 721 722 /* For backwards source compatibility. They do nothing, as the default now 723 * includes what they used to mean. The first one's meaning was to allow the 724 * just the single non-character 0xFFFF */ 725 #define UTF8_ALLOW_FFFF 0 726 #define UTF8_ALLOW_FE_FF 0 727 #define UTF8_ALLOW_SURROGATE 0 728 729 /* C9 refers to Unicode Corrigendum #9: allows but discourages non-chars */ 730 #define UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE \ 731 (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_SURROGATE) 732 #define UTF8_WARN_ILLEGAL_C9_INTERCHANGE (UTF8_WARN_SUPER|UTF8_WARN_SURROGATE) 733 734 #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE \ 735 (UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE|UTF8_DISALLOW_NONCHAR) 736 #define UTF8_WARN_ILLEGAL_INTERCHANGE \ 737 (UTF8_WARN_ILLEGAL_C9_INTERCHANGE|UTF8_WARN_NONCHAR) 738 739 /* This is typically used for code that processes UTF-8 input and doesn't want 740 * to have to deal with any malformations that might be present. All such will 741 * be safely replaced by the REPLACEMENT CHARACTER, unless other flags 742 * overriding this are also present. */ 743 #define UTF8_ALLOW_ANY ( UTF8_ALLOW_CONTINUATION \ 744 |UTF8_ALLOW_NON_CONTINUATION \ 745 |UTF8_ALLOW_SHORT \ 746 |UTF8_ALLOW_LONG \ 747 |UTF8_ALLOW_OVERFLOW) 748 749 /* Accept any Perl-extended UTF-8 that evaluates to any UV on the platform, but 750 * not any malformed. This is the default. */ 751 #define UTF8_ALLOW_ANYUV 0 752 #define UTF8_ALLOW_DEFAULT UTF8_ALLOW_ANYUV 753 754 /* 755 =for apidoc Am|bool|UTF8_IS_SURROGATE|const U8 *s|const U8 *e 756 757 Evaluates to non-zero if the first few bytes of the string starting at C<s> and 758 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one 759 of the Unicode surrogate code points; otherwise it evaluates to 0. If 760 non-zero, the value gives how many bytes starting at C<s> comprise the code 761 point's representation. 762 763 =cut 764 */ 765 #define UTF8_IS_SURROGATE(s, e) is_SURROGATE_utf8_safe(s, e) 766 767 768 #define UTF8_IS_REPLACEMENT(s, send) is_REPLACEMENT_utf8_safe(s,send) 769 770 #define MAX_LEGAL_CP ((UV)IV_MAX) 771 772 /* 773 =for apidoc Am|bool|UTF8_IS_SUPER|const U8 *s|const U8 *e 774 775 Recall that Perl recognizes an extension to UTF-8 that can encode code 776 points larger than the ones defined by Unicode, which are 0..0x10FFFF. 777 778 This macro evaluates to non-zero if the first few bytes of the string starting 779 at C<s> and looking no further than S<C<e - 1>> are from this UTF-8 extension; 780 otherwise it evaluates to 0. If non-zero, the value gives how many bytes 781 starting at C<s> comprise the code point's representation. 782 783 0 is returned if the bytes are not well-formed extended UTF-8, or if they 784 represent a code point that cannot fit in a UV on the current platform. Hence 785 this macro can give different results when run on a 64-bit word machine than on 786 one with a 32-bit word size. 787 788 Note that it is illegal to have code points that are larger than what can 789 fit in an IV on the current machine. 790 791 =cut 792 793 * ASCII EBCDIC I8 794 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode 795 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0 796 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1 797 */ 798 #ifdef EBCDIC 799 # define UTF8_IS_SUPER(s, e) \ 800 (( LIKELY((e) > (s) + 4) \ 801 && NATIVE_UTF8_TO_I8(*(s)) >= 0xF9 \ 802 && ( NATIVE_UTF8_TO_I8(*(s)) > 0xF9 \ 803 || (NATIVE_UTF8_TO_I8(*((s) + 1)) >= 0xA2)) \ 804 && LIKELY((s) + UTF8SKIP(s) <= (e))) \ 805 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0) 806 #else 807 # define UTF8_IS_SUPER(s, e) \ 808 (( LIKELY((e) > (s) + 3) \ 809 && (*(U8*) (s)) >= 0xF4 \ 810 && ((*(U8*) (s)) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))\ 811 && LIKELY((s) + UTF8SKIP(s) <= (e))) \ 812 ? _is_utf8_char_helper(s, s + UTF8SKIP(s), 0) : 0) 813 #endif 814 815 /* These are now machine generated, and the 'given' clause is no longer 816 * applicable */ 817 #define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) \ 818 cBOOL(is_NONCHAR_utf8_safe(s,e)) 819 820 /* 821 =for apidoc Am|bool|UTF8_IS_NONCHAR|const U8 *s|const U8 *e 822 823 Evaluates to non-zero if the first few bytes of the string starting at C<s> and 824 looking no further than S<C<e - 1>> are well-formed UTF-8 that represents one 825 of the Unicode non-character code points; otherwise it evaluates to 0. If 826 non-zero, the value gives how many bytes starting at C<s> comprise the code 827 point's representation. 828 829 =cut 830 */ 831 #define UTF8_IS_NONCHAR(s, e) \ 832 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s, e) 833 834 #define UNICODE_SURROGATE_FIRST 0xD800 835 #define UNICODE_SURROGATE_LAST 0xDFFF 836 #define UNICODE_REPLACEMENT 0xFFFD 837 #define UNICODE_BYTE_ORDER_MARK 0xFEFF 838 839 /* Though our UTF-8 encoding can go beyond this, 840 * let's be conservative and do as Unicode says. */ 841 #define PERL_UNICODE_MAX 0x10FFFF 842 843 #define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */ 844 #define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */ 845 #define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */ 846 #define UNICODE_WARN_PERL_EXTENDED 0x0008 /* Above 0x7FFF_FFFF */ 847 #define UNICODE_WARN_ABOVE_31_BIT UNICODE_WARN_PERL_EXTENDED 848 #define UNICODE_DISALLOW_SURROGATE 0x0010 849 #define UNICODE_DISALLOW_NONCHAR 0x0020 850 #define UNICODE_DISALLOW_SUPER 0x0040 851 #define UNICODE_DISALLOW_PERL_EXTENDED 0x0080 852 #define UNICODE_DISALLOW_ABOVE_31_BIT UNICODE_DISALLOW_PERL_EXTENDED 853 854 #define UNICODE_GOT_SURROGATE UNICODE_DISALLOW_SURROGATE 855 #define UNICODE_GOT_NONCHAR UNICODE_DISALLOW_NONCHAR 856 #define UNICODE_GOT_SUPER UNICODE_DISALLOW_SUPER 857 #define UNICODE_GOT_PERL_EXTENDED UNICODE_DISALLOW_PERL_EXTENDED 858 859 #define UNICODE_WARN_ILLEGAL_C9_INTERCHANGE \ 860 (UNICODE_WARN_SURROGATE|UNICODE_WARN_SUPER) 861 #define UNICODE_WARN_ILLEGAL_INTERCHANGE \ 862 (UNICODE_WARN_ILLEGAL_C9_INTERCHANGE|UNICODE_WARN_NONCHAR) 863 #define UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE \ 864 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_SUPER) 865 #define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \ 866 (UNICODE_DISALLOW_ILLEGAL_C9_INTERCHANGE|UNICODE_DISALLOW_NONCHAR) 867 868 /* For backward source compatibility, as are now the default */ 869 #define UNICODE_ALLOW_SURROGATE 0 870 #define UNICODE_ALLOW_SUPER 0 871 #define UNICODE_ALLOW_ANY 0 872 873 /* This matches the 2048 code points between UNICODE_SURROGATE_FIRST (0xD800) and 874 * UNICODE_SURROGATE_LAST (0xDFFF) */ 875 #define UNICODE_IS_SURROGATE(uv) (((UV) (uv) & (~0xFFFF | 0xF800)) \ 876 == 0xD800) 877 878 #define UNICODE_IS_REPLACEMENT(uv) ((UV) (uv) == UNICODE_REPLACEMENT) 879 #define UNICODE_IS_BYTE_ORDER_MARK(uv) ((UV) (uv) == UNICODE_BYTE_ORDER_MARK) 880 881 /* Is 'uv' one of the 32 contiguous-range noncharacters? */ 882 #define UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) ((UV) (uv) >= 0xFDD0 \ 883 && (UV) (uv) <= 0xFDEF) 884 885 /* Is 'uv' one of the 34 plane-ending noncharacters 0xFFFE, 0xFFFF, 0x1FFFE, 886 * 0x1FFFF, ... 0x10FFFE, 0x10FFFF, given that we know that 'uv' is not above 887 * the Unicode legal max */ 888 #define UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv) \ 889 (((UV) (uv) & 0xFFFE) == 0xFFFE) 890 891 #define UNICODE_IS_NONCHAR(uv) \ 892 ( UNICODE_IS_32_CONTIGUOUS_NONCHARS(uv) \ 893 || ( LIKELY( ! UNICODE_IS_SUPER(uv)) \ 894 && UNICODE_IS_END_PLANE_NONCHAR_GIVEN_NOT_SUPER(uv))) 895 896 #define UNICODE_IS_SUPER(uv) ((UV) (uv) > PERL_UNICODE_MAX) 897 898 #define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE 899 #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \ 900 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 901 #define MICRO_SIGN MICRO_SIGN_NATIVE 902 #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \ 903 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 904 #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \ 905 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 906 #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3 907 #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2 908 #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3 909 #define GREEK_SMALL_LETTER_MU 0x03BC 910 #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case 911 of MICRON */ 912 #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */ 913 #ifdef LATIN_CAPITAL_LETTER_SHARP_S_UTF8 914 # define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E 915 #endif 916 #define LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE 0x130 917 #define LATIN_SMALL_LETTER_DOTLESS_I 0x131 918 #define LATIN_SMALL_LETTER_LONG_S 0x017F 919 #define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05 920 #define LATIN_SMALL_LIGATURE_ST 0xFB06 921 #define KELVIN_SIGN 0x212A 922 #define ANGSTROM_SIGN 0x212B 923 924 #define UNI_DISPLAY_ISPRINT 0x0001 925 #define UNI_DISPLAY_BACKSLASH 0x0002 926 #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) 927 #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) 928 929 #define ANYOF_FOLD_SHARP_S(node, input, end) \ 930 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \ 931 (ANYOF_NONBITMAP(node)) && \ 932 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \ 933 ((end) > (input) + 1) && \ 934 isALPHA_FOLD_EQ((input)[0], 's')) 935 936 #define SHARP_S_SKIP 2 937 938 #define is_utf8_char_buf(buf, buf_end) isUTF8_CHAR(buf, buf_end) 939 #define bytes_from_utf8(s, lenp, is_utf8p) \ 940 bytes_from_utf8_loc(s, lenp, is_utf8p, 0) 941 942 /* 943 944 =for apidoc Am|STRLEN|isUTF8_CHAR_flags|const U8 *s|const U8 *e| const U32 flags 945 946 Evaluates to non-zero if the first few bytes of the string starting at C<s> and 947 looking no further than S<C<e - 1>> are well-formed UTF-8, as extended by Perl, 948 that represents some code point, subject to the restrictions given by C<flags>; 949 otherwise it evaluates to 0. If non-zero, the value gives how many bytes 950 starting at C<s> comprise the code point's representation. Any bytes remaining 951 before C<e>, but beyond the ones needed to form the first code point in C<s>, 952 are not examined. 953 954 If C<flags> is 0, this gives the same results as C<L</isUTF8_CHAR>>; 955 if C<flags> is C<UTF8_DISALLOW_ILLEGAL_INTERCHANGE>, this gives the same results 956 as C<L</isSTRICT_UTF8_CHAR>>; 957 and if C<flags> is C<UTF8_DISALLOW_ILLEGAL_C9_INTERCHANGE>, this gives 958 the same results as C<L</isC9_STRICT_UTF8_CHAR>>. 959 Otherwise C<flags> may be any combination of the C<UTF8_DISALLOW_I<foo>> flags 960 understood by C<L</utf8n_to_uvchr>>, with the same meanings. 961 962 The three alternative macros are for the most commonly needed validations; they 963 are likely to run somewhat faster than this more general one, as they can be 964 inlined into your code. 965 966 Use L</is_utf8_string_flags>, L</is_utf8_string_loc_flags>, and 967 L</is_utf8_string_loclen_flags> to check entire strings. 968 969 =cut 970 */ 971 972 #define isUTF8_CHAR_flags(s, e, flags) \ 973 (UNLIKELY((e) <= (s)) \ 974 ? 0 \ 975 : (UTF8_IS_INVARIANT(*s)) \ 976 ? 1 \ 977 : UNLIKELY(((e) - (s)) < UTF8SKIP(s)) \ 978 ? 0 \ 979 : _is_utf8_char_helper(s, e, flags)) 980 981 /* Do not use; should be deprecated. Use isUTF8_CHAR() instead; this is 982 * retained solely for backwards compatibility */ 983 #define IS_UTF8_CHAR(p, n) (isUTF8_CHAR(p, (p) + (n)) == n) 984 985 #endif /* PERL_UTF8_H_ */ 986 987 /* 988 * ex: set ts=8 sts=4 sw=4 et: 989 */ 990