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 H_UTF8 /* Guard against recursive inclusion */ 18 #define H_UTF8 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 /* For _core_swash_init(), internal core use only */ 38 #define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1 39 #define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2 40 #define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4 41 42 #define uvchr_to_utf8(a,b) uvchr_to_utf8_flags(a,b,0) 43 #define uvchr_to_utf8_flags(d,uv,flags) \ 44 uvoffuni_to_utf8_flags(d,NATIVE_TO_UNI(uv),flags) 45 #define utf8_to_uvchr_buf(s, e, lenp) \ 46 utf8n_to_uvchr(s, (U8*)(e) - (U8*)(s), lenp, \ 47 ckWARN_d(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY) 48 49 #define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) 50 #define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, FOLD_FLAGS_FULL) 51 #define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0) 52 #define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0) 53 #define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0) 54 55 /* Source backward compatibility. */ 56 #define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0) 57 58 #define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ 59 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0) 60 #define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0) 61 #define FOLDEQ_LOCALE (1 << 1) 62 #define FOLDEQ_S1_ALREADY_FOLDED (1 << 2) 63 #define FOLDEQ_S2_ALREADY_FOLDED (1 << 3) 64 65 /* 66 =for apidoc ibcmp_utf8 67 68 This is a synonym for (! foldEQ_utf8()) 69 70 =cut 71 */ 72 #define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \ 73 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2)) 74 75 #ifdef EBCDIC 76 /* The equivalent of these macros but implementing UTF-EBCDIC 77 are in the following header file: 78 */ 79 80 #include "utfebcdic.h" 81 82 #else /* ! EBCDIC */ 83 START_EXTERN_C 84 85 #ifdef DOINIT 86 EXTCONST unsigned char PL_utf8skip[] = { 87 /* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 88 /* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 89 /* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 90 /* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 91 /* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 92 /* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 93 /* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 94 /* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */ 95 /* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 96 /* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 97 /* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 98 /* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */ 99 /* 0xC0 */ 2,2, /* overlong */ 100 /* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */ 101 /* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */ 102 /* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */ 103 /* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */ 104 /* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit 105 allowed (64-bit + reserved). */ 106 }; 107 #else 108 EXTCONST unsigned char PL_utf8skip[]; 109 #endif 110 111 END_EXTERN_C 112 113 /* Native character to/from iso-8859-1. Are the identity functions on ASCII 114 * platforms */ 115 #define NATIVE_TO_LATIN1(ch) (ch) 116 #define LATIN1_TO_NATIVE(ch) (ch) 117 118 /* I8 is an intermediate version of UTF-8 used only in UTF-EBCDIC. We thus 119 * consider it to be identical to UTF-8 on ASCII platforms. Strictly speaking 120 * UTF-8 and UTF-EBCDIC are two different things, but we often conflate them 121 * because they are 8-bit encodings that serve the same purpose in Perl, and 122 * rarely do we need to distinguish them. The term "NATIVE_UTF8" applies to 123 * whichever one is applicable on the current platform */ 124 #define NATIVE_UTF8_TO_I8(ch) (ch) 125 #define I8_TO_NATIVE_UTF8(ch) (ch) 126 127 /* Transforms in wide UV chars */ 128 #define UNI_TO_NATIVE(ch) (ch) 129 #define NATIVE_TO_UNI(ch) (ch) 130 131 /* 132 133 The following table is from Unicode 3.2. 134 135 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte 136 137 U+0000..U+007F 00..7F 138 U+0080..U+07FF * C2..DF 80..BF 139 U+0800..U+0FFF E0 * A0..BF 80..BF 140 U+1000..U+CFFF E1..EC 80..BF 80..BF 141 U+D000..U+D7FF ED 80..9F 80..BF 142 U+D800..U+DFFF ED A0..BF 80..BF (surrogates) 143 U+E000..U+FFFF EE..EF 80..BF 80..BF 144 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF 145 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF 146 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF 147 Below are non-Unicode code points 148 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF 149 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF 150 U+200000..: F8.. * 88..BF 80..BF 80..BF 80..BF 151 152 Note the gaps before several of the byte entries above marked by '*'. These are 153 caused by legal UTF-8 avoiding non-shortest encodings: it is technically 154 possible to UTF-8-encode a single code point in different ways, but that is 155 explicitly forbidden, and the shortest possible encoding should always be used 156 (and that is what Perl does). The non-shortest ones are called 'overlongs'. 157 158 */ 159 160 /* 161 Another way to look at it, as bits: 162 163 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte 164 165 0aaa aaaa 0aaa aaaa 166 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa 167 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa 168 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa 169 170 As you can see, the continuation bytes all begin with C<10>, and the 171 leading bits of the start byte tell how many bytes there are in the 172 encoded character. 173 174 Perl's extended UTF-8 means we can have start bytes up to FF. 175 176 */ 177 178 /* Is the representation of the Unicode code point 'c' the same regardless of 179 * being encoded in UTF-8 or not? */ 180 #define UNI_IS_INVARIANT(c) (((UV)c) < 0x80) 181 182 /* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is 183 * the inverse of UTF8_IS_INVARIANT */ 184 #define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80) 185 186 /* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence? 187 * This doesn't catch invariants (they are single-byte). It also excludes the 188 * illegal overlong sequences that begin with C0 and C1. */ 189 #define UTF8_IS_START(c) (((U8)c) >= 0xc2) 190 191 /* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the 192 * first byte thereof? */ 193 #define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80) 194 195 /* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use 196 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to 197 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus 198 * this matches 0xc[23]. */ 199 #define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2) 200 201 /* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that 202 * represent a code point > 255? */ 203 #define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4) 204 205 /* This defines the 1-bits that are to be in the first byte of a multi-byte 206 * UTF-8 encoded character that give the number of bytes that comprise the 207 * character. 'len' is the number of bytes in the multi-byte sequence. */ 208 #define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFF & (0xFE << (7-(len))))) 209 210 /* Masks out the initial one bits in a start byte, leaving the real data ones. 211 * Doesn't work on an invariant byte. 'len' is the number of bytes in the 212 * multi-byte sequence that comprises the character. */ 213 #define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2))) 214 215 /* This defines the bits that are to be in the continuation bytes of a multi-byte 216 * UTF-8 encoded character that indicate it is a continuation byte. */ 217 #define UTF_CONTINUATION_MARK 0x80 218 219 /* This is the number of low-order bits a continuation byte in a UTF-8 encoded 220 * sequence contributes to the specification of the code point. In the bit 221 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of 222 * real information */ 223 #define UTF_ACCUMULATION_SHIFT 6 224 225 /* 2**UTF_ACCUMULATION_SHIFT - 1 */ 226 #define UTF_CONTINUATION_MASK ((U8)0x3f) 227 228 /* If a value is anded with this, and the result is non-zero, then using the 229 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left 230 * */ 231 #define UTF_ACCUMULATION_OVERFLOW_MASK \ 232 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \ 233 - UTF_ACCUMULATION_SHIFT)) 234 235 #if UVSIZE >= 8 236 # define UTF8_QUAD_MAX UINT64_C(0x1000000000) 237 238 /* Input is a true Unicode (not-native) code point */ 239 #define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \ 240 (uv) < 0x800 ? 2 : \ 241 (uv) < 0x10000 ? 3 : \ 242 (uv) < 0x200000 ? 4 : \ 243 (uv) < 0x4000000 ? 5 : \ 244 (uv) < 0x80000000 ? 6 : \ 245 (uv) < UTF8_QUAD_MAX ? 7 : 13 ) 246 #else 247 /* No, I'm not even going to *TRY* putting #ifdef inside a #define */ 248 #define OFFUNISKIP(uv) ( (uv) < 0x80 ? 1 : \ 249 (uv) < 0x800 ? 2 : \ 250 (uv) < 0x10000 ? 3 : \ 251 (uv) < 0x200000 ? 4 : \ 252 (uv) < 0x4000000 ? 5 : \ 253 (uv) < 0x80000000 ? 6 : 7 ) 254 #endif 255 256 /* How wide can a single UTF-8 encoded character become in bytes. */ 257 /* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8 258 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be 259 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode 260 * non-negative integers in a binary format, even those above Unicode */ 261 #define UTF8_MAXBYTES 13 262 263 /* The maximum number of UTF-8 bytes a single Unicode character can 264 * uppercase/lowercase/fold into. Unicode guarantees that the maximum 265 * expansion is 3 characters. On ASCIIish platforms, the highest Unicode 266 * character occupies 4 bytes, therefore this number would be 12, but this is 267 * smaller than the maximum width a single above-Unicode character can occupy, 268 * so use that instead */ 269 #if UTF8_MAXBYTES < 12 270 #error UTF8_MAXBYTES must be at least 12 271 #endif 272 273 #define MAX_UTF8_TWO_BYTE 0x7FF 274 275 #define UTF8_MAXBYTES_CASE UTF8_MAXBYTES 276 277 #endif /* EBCDIC vs ASCII */ 278 279 /* Rest of these are attributes of Unicode and perl's internals rather than the 280 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at 281 * this level; the macros that some of these call may have different 282 * definitions in the two encodings */ 283 284 /* In domain restricted to ASCII, these may make more sense to the reader than 285 * the ones with Latin1 in the name */ 286 #define NATIVE_TO_ASCII(ch) NATIVE_TO_LATIN1(ch) 287 #define ASCII_TO_NATIVE(ch) LATIN1_TO_NATIVE(ch) 288 289 /* More or less misleadingly-named defines, retained for back compat */ 290 #define NATIVE_TO_UTF(ch) NATIVE_UTF8_TO_I8(ch) 291 #define NATIVE_TO_I8(ch) NATIVE_UTF8_TO_I8(ch) 292 #define UTF_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch) 293 #define I8_TO_NATIVE(ch) I8_TO_NATIVE_UTF8(ch) 294 #define NATIVE8_TO_UNI(ch) NATIVE_TO_LATIN1(ch) 295 296 /* Adds a UTF8 continuation byte 'new' of information to a running total code 297 * point 'old' of all the continuation bytes so far. This is designed to be 298 * used in a loop to convert from UTF-8 to the code point represented. Note 299 * that this is asymmetric on EBCDIC platforms, in that the 'new' parameter is 300 * the UTF-EBCDIC byte, whereas the 'old' parameter is a Unicode (not EBCDIC) 301 * code point in process of being generated */ 302 #define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \ 303 | ((NATIVE_UTF8_TO_I8((U8)new)) \ 304 & UTF_CONTINUATION_MASK)) 305 306 /* This works in the face of malformed UTF-8. */ 307 #define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \ 308 && ( (e) - (s) > 1) \ 309 && UTF8_IS_CONTINUATION(*((s)+1))) 310 311 /* Number of bytes a code point occupies in UTF-8. */ 312 #define NATIVE_SKIP(uv) OFFUNISKIP(NATIVE_TO_UNI(uv)) 313 314 /* Most code which says UNISKIP is really thinking in terms of native code 315 * points (0-255) plus all those beyond. This is an imprecise term, but having 316 * it means existing code continues to work. For precision, use NATIVE_SKIP 317 * and OFFUNISKIP */ 318 #define UNISKIP(uv) NATIVE_SKIP(uv) 319 320 /* Convert a two (not one) byte utf8 character to a native code point value. 321 * Needs just one iteration of accumulate. Should not be used unless it is 322 * known that the two bytes are legal: 1) two-byte start, and 2) continuation. 323 * Note that the result can be larger than 255 if the input character is not 324 * downgradable */ 325 #define TWO_BYTE_UTF8_TO_NATIVE(HI, LO) \ 326 UNI_TO_NATIVE(UTF8_ACCUMULATE((NATIVE_UTF8_TO_I8(HI) & UTF_START_MASK(2)), \ 327 (LO))) 328 329 /* Should never be used, and be deprecated */ 330 #define TWO_BYTE_UTF8_TO_UNI(HI, LO) NATIVE_TO_UNI(TWO_BYTE_UTF8_TO_NATIVE(HI, LO)) 331 332 /* How many bytes in the UTF-8 encoded character whose first (perhaps only) 333 * byte is pointed to by 's' */ 334 #define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)] 335 336 /* Is the byte 'c' the same character when encoded in UTF-8 as when not. This 337 * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in 338 * each for the exact same set of bit patterns. (And it works on any byte in a 339 * UTF-8 encoded string) */ 340 #define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_UTF8_TO_I8(c)) 341 342 /* Like the above, but its name implies a non-UTF8 input, and is implemented 343 * differently (for no particular reason) */ 344 #define NATIVE_BYTE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_LATIN1(c)) 345 346 /* Like the above, but accepts any UV as input */ 347 #define UVCHR_IS_INVARIANT(uv) UNI_IS_INVARIANT(NATIVE_TO_UNI(uv)) 348 349 #define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */ 350 351 /* The macros in the next 4 sets are used to generate the two utf8 or utfebcdic 352 * bytes from an ordinal that is known to fit into exactly two (not one) bytes; 353 * it must be less than 0x3FF to work across both encodings. */ 354 355 /* These two are helper macros for the other three sets, and should not be used 356 * directly anywhere else. 'translate_function' is either NATIVE_TO_LATIN1 357 * (which works for code points up to 0xFF) or NATIVE_TO_UNI which works for any 358 * code point */ 359 #define __BASE_TWO_BYTE_HI(c, translate_function) \ 360 I8_TO_NATIVE_UTF8((translate_function(c) >> UTF_ACCUMULATION_SHIFT) \ 361 | UTF_START_MARK(2)) 362 #define __BASE_TWO_BYTE_LO(c, translate_function) \ 363 I8_TO_NATIVE_UTF8((translate_function(c) & UTF_CONTINUATION_MASK) \ 364 | UTF_CONTINUATION_MARK) 365 366 /* The next two macros should not be used. They were designed to be usable as 367 * the case label of a switch statement, but this doesn't work for EBCDIC. Use 368 * regen/unicode_constants.pl instead */ 369 #define UTF8_TWO_BYTE_HI_nocast(c) __BASE_TWO_BYTE_HI(c, NATIVE_TO_UNI) 370 #define UTF8_TWO_BYTE_LO_nocast(c) __BASE_TWO_BYTE_LO(c, NATIVE_TO_UNI) 371 372 /* The next two macros are used when the source should be a single byte 373 * character; checked for under DEBUGGING */ 374 #define UTF8_EIGHT_BIT_HI(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 375 ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1))) 376 #define UTF8_EIGHT_BIT_LO(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 377 ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1))) 378 379 /* These final two macros in the series are used when the source can be any 380 * code point whose UTF-8 is known to occupy 2 bytes; they are less efficient 381 * than the EIGHT_BIT versions on EBCDIC platforms. We use the logical '~' 382 * operator instead of "<=" to avoid getting compiler warnings. 383 * MAX_PORTABLE_UTF8_TWO_BYTE should be exactly all one bits in the lower few 384 * places, so the ~ works */ 385 #define UTF8_TWO_BYTE_HI(c) \ 386 (__ASSERT_((sizeof(c) == 1) \ 387 || !(((WIDEST_UTYPE)(c)) & ~MAX_PORTABLE_UTF8_TWO_BYTE)) \ 388 ((U8) __BASE_TWO_BYTE_HI(c, NATIVE_TO_LATIN1))) 389 #define UTF8_TWO_BYTE_LO(c) \ 390 (__ASSERT_((sizeof(c) == 1) \ 391 || !(((WIDEST_UTYPE)(c)) & ~MAX_PORTABLE_UTF8_TWO_BYTE)) \ 392 ((U8) __BASE_TWO_BYTE_LO(c, NATIVE_TO_LATIN1))) 393 394 /* This is illegal in any well-formed UTF-8 in both EBCDIC and ASCII 395 * as it is only in overlongs. */ 396 #define ILLEGAL_UTF8_BYTE I8_TO_NATIVE_UTF8(0xC1) 397 398 /* 399 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem 400 * from an earlier version of these macros in which they didn't call the 401 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the 402 * beginning of a utf8 character. Now that foo_utf8() determines that itself, 403 * no need to do it again here 404 */ 405 #define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \ 406 ? isIDFIRST(*(p)) \ 407 : isIDFIRST_utf8((const U8*)p)) 408 #define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \ 409 ? isWORDCHAR(*(p)) \ 410 : isWORDCHAR_utf8((const U8*)p)) 411 #define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF) 412 413 #define UTF8_MAXLEN UTF8_MAXBYTES 414 415 /* A Unicode character can fold to up to 3 characters */ 416 #define UTF8_MAX_FOLD_CHAR_EXPAND 3 417 418 #define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES) 419 #define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES) 420 #define IN_UNI_8_BIT \ 421 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \ 422 && ! IN_LOCALE_RUNTIME && ! IN_BYTES) 423 424 425 #define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */ 426 427 /* Allow first byte to be a continuation byte */ 428 #define UTF8_ALLOW_CONTINUATION 0x0002 429 430 /* Allow second... bytes to be non-continuation bytes */ 431 #define UTF8_ALLOW_NON_CONTINUATION 0x0004 432 433 /* expecting more bytes than were available in the string */ 434 #define UTF8_ALLOW_SHORT 0x0008 435 436 /* Overlong sequence; i.e., the code point can be specified in fewer bytes. */ 437 #define UTF8_ALLOW_LONG 0x0010 438 439 #define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */ 440 #define UTF8_WARN_SURROGATE 0x0040 441 442 #define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */ 443 #define UTF8_WARN_NONCHAR 0x0100 /* code points */ 444 445 #define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */ 446 #define UTF8_WARN_SUPER 0x0400 /* points above the legal max */ 447 448 /* Code points which never were part of the original UTF-8 standard, the first 449 * byte of which is a FE or FF on ASCII platforms. If the first byte is FF, it 450 * will overflow a 32-bit word. If the first byte is FE, it will overflow a 451 * signed 32-bit word. */ 452 #define UTF8_DISALLOW_FE_FF 0x0800 453 #define UTF8_WARN_FE_FF 0x1000 454 455 #define UTF8_CHECK_ONLY 0x2000 456 457 /* For backwards source compatibility. They do nothing, as the default now 458 * includes what they used to mean. The first one's meaning was to allow the 459 * just the single non-character 0xFFFF */ 460 #define UTF8_ALLOW_FFFF 0 461 #define UTF8_ALLOW_SURROGATE 0 462 463 #define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF) 464 #define UTF8_WARN_ILLEGAL_INTERCHANGE \ 465 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF) 466 #define UTF8_ALLOW_ANY \ 467 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) 468 #define UTF8_ALLOW_ANYUV \ 469 (UTF8_ALLOW_EMPTY \ 470 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE)) 471 #define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \ 472 UTF8_ALLOW_ANYUV) 473 474 /* Surrogates, non-character code points and above-Unicode code points are 475 * problematic in some contexts. This allows code that needs to check for 476 * those to to quickly exclude the vast majority of code points it will 477 * encounter */ 478 #define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \ 479 FIRST_SURROGATE_UTF8_FIRST_BYTE 480 481 #define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s)) 482 #define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send)) 483 484 /* ASCII EBCDIC I8 485 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode 486 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0 487 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1 488 */ 489 #ifdef EBCDIC /* Both versions assume well-formed UTF8 */ 490 # define UTF8_IS_SUPER(s) (NATIVE_UTF8_TO_I8(* (U8*) (s)) >= 0xF9 \ 491 && (NATIVE_UTF8_TO_I8(* (U8*) (s)) > 0xF9 \ 492 || (NATIVE_UTF8_TO_I8(* ((U8*) (s) + 1)) >= 0xA2))) 493 #else 494 # define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \ 495 && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90))) 496 #endif 497 498 /* These are now machine generated, and the 'given' clause is no longer 499 * applicable */ 500 #define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \ 501 cBOOL(is_NONCHAR_utf8(s)) 502 #define UTF8_IS_NONCHAR_(s) \ 503 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) 504 505 #define UNICODE_SURROGATE_FIRST 0xD800 506 #define UNICODE_SURROGATE_LAST 0xDFFF 507 #define UNICODE_REPLACEMENT 0xFFFD 508 #define UNICODE_BYTE_ORDER_MARK 0xFEFF 509 510 /* Though our UTF-8 encoding can go beyond this, 511 * let's be conservative and do as Unicode says. */ 512 #define PERL_UNICODE_MAX 0x10FFFF 513 514 #define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */ 515 #define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */ 516 #define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */ 517 #define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */ 518 #define UNICODE_DISALLOW_SURROGATE 0x0010 519 #define UNICODE_DISALLOW_NONCHAR 0x0020 520 #define UNICODE_DISALLOW_SUPER 0x0040 521 #define UNICODE_DISALLOW_FE_FF 0x0080 522 #define UNICODE_WARN_ILLEGAL_INTERCHANGE \ 523 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER) 524 #define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \ 525 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER) 526 527 /* For backward source compatibility, as are now the default */ 528 #define UNICODE_ALLOW_SURROGATE 0 529 #define UNICODE_ALLOW_SUPER 0 530 #define UNICODE_ALLOW_ANY 0 531 532 #define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \ 533 (c) <= UNICODE_SURROGATE_LAST) 534 #define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT) 535 #define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK) 536 #define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \ 537 /* The other noncharacters end in FFFE or FFFF, which \ 538 * the mask below catches both of, but beyond the last \ 539 * official unicode code point, they aren't \ 540 * noncharacters, since those aren't Unicode \ 541 * characters at all */ \ 542 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c))) 543 #define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX) 544 #define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF) 545 546 #define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE 547 #define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \ 548 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE 549 #define MICRO_SIGN MICRO_SIGN_NATIVE 550 #define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \ 551 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE 552 #define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \ 553 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE 554 #define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3 555 #define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2 556 #define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3 557 #define GREEK_SMALL_LETTER_MU 0x03BC 558 #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case 559 of MICRON */ 560 #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */ 561 #define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E 562 #define LATIN_SMALL_LETTER_LONG_S 0x017F 563 #define LATIN_SMALL_LIGATURE_LONG_S_T 0xFB05 564 #define LATIN_SMALL_LIGATURE_ST 0xFB06 565 #define KELVIN_SIGN 0x212A 566 #define ANGSTROM_SIGN 0x212B 567 568 #define UNI_DISPLAY_ISPRINT 0x0001 569 #define UNI_DISPLAY_BACKSLASH 0x0002 570 #define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) 571 #define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH) 572 573 #define ANYOF_FOLD_SHARP_S(node, input, end) \ 574 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \ 575 (ANYOF_NONBITMAP(node)) && \ 576 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \ 577 ((end) > (input) + 1) && \ 578 toFOLD((input)[0]) == 's' && \ 579 toFOLD((input)[1]) == 's') 580 #define SHARP_S_SKIP 2 581 582 /* If you want to exclude surrogates, and beyond legal Unicode, see the blame 583 * log for earlier versions which gave details for these */ 584 585 #ifndef EBCDIC 586 /* This was generated by regen/regcharclass.pl, and then moved here. The lines 587 * that generated it were then commented out. This was done solely because it 588 * takes on the order of 10 minutes to generate, and is never going to change. 589 * The EBCDIC equivalent hasn't been commented out in regcharclass.pl, so it 590 * should generate and run the correct stuff */ 591 /* 592 UTF8_CHAR: Matches utf8 from 1 to 4 bytes 593 594 0x0 - 0x1FFFFF 595 */ 596 /*** GENERATED CODE ***/ 597 #define is_UTF8_CHAR_utf8_safe(s,e) \ 598 ( ((e)-(s) > 3) ? \ 599 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \ 600 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \ 601 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \ 602 : ( 0xE0 == ((U8*)s)[0] ) ? \ 603 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ 604 : ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \ 605 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ 606 : ( 0xF0 == ((U8*)s)[0] ) ? \ 607 ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\ 608 : ( ( ( ( 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 )\ 609 : ((e)-(s) > 2) ? \ 610 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \ 611 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \ 612 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \ 613 : ( 0xE0 == ((U8*)s)[0] ) ? \ 614 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ 615 : ( ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\ 616 : ((e)-(s) > 1) ? \ 617 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \ 618 : ( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) ? 2 : 0 )\ 619 : ((e)-(s) > 0) ? \ 620 ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) \ 621 : 0 ) 622 #endif 623 624 /* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it 625 * (1) allows UTF-8 encoded UTF-16 surrogates 626 * (2) it allows code points past U+10FFFF. 627 * The Perl_is_utf8_char() full "slow" code will handle the Perl 628 * "extended UTF-8". */ 629 #define IS_UTF8_CHAR(p, n) (is_UTF8_CHAR_utf8_safe(p, (p) + (n)) == n) 630 631 /* regen/regcharclass.pl generates is_UTF8_CHAR_utf8_safe() macros for up to 632 * these number of bytes. So this has to be coordinated with it */ 633 #ifdef EBCDIC 634 # define IS_UTF8_CHAR_FAST(n) ((n) <= 5) 635 #else 636 # define IS_UTF8_CHAR_FAST(n) ((n) <= 4) 637 #endif 638 639 #endif /* H_UTF8 */ 640 641 /* 642 * Local variables: 643 * c-indentation-style: bsd 644 * c-basic-offset: 4 645 * indent-tabs-mode: nil 646 * End: 647 * 648 * ex: set ts=8 sts=4 sw=4 et: 649 */ 650