1 /* utf8.c 2 * 3 * Copyright (C) 2000, 2001, 2002, 2003, by Larry Wall and others 4 * 5 * You may distribute under the terms of either the GNU General Public 6 * License or the Artistic License, as specified in the README file. 7 * 8 */ 9 10 /* 11 * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever 12 * heard of that we don't want to see any closer; and that's the one place 13 * we're trying to get to! And that's just where we can't get, nohow.' 14 * 15 * 'Well do I understand your speech,' he answered in the same language; 16 * 'yet few strangers do so. Why then do you not speak in the Common Tongue, 17 * as is the custom in the West, if you wish to be answered?' 18 * 19 * ...the travellers perceived that the floor was paved with stones of many 20 * hues; branching runes and strange devices intertwined beneath their feet. 21 */ 22 23 #include "EXTERN.h" 24 #define PERL_IN_UTF8_C 25 #include "perl.h" 26 27 static char unees[] = "Malformed UTF-8 character (unexpected end of string)"; 28 29 /* 30 =head1 Unicode Support 31 32 =for apidoc A|U8 *|uvuni_to_utf8_flags|U8 *d|UV uv|UV flags 33 34 Adds the UTF-8 representation of the Unicode codepoint C<uv> to the end 35 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free 36 bytes available. The return value is the pointer to the byte after the 37 end of the new character. In other words, 38 39 d = uvuni_to_utf8_flags(d, uv, flags); 40 41 or, in most cases, 42 43 d = uvuni_to_utf8(d, uv); 44 45 (which is equivalent to) 46 47 d = uvuni_to_utf8_flags(d, uv, 0); 48 49 is the recommended Unicode-aware way of saying 50 51 *(d++) = uv; 52 53 =cut 54 */ 55 56 U8 * 57 Perl_uvuni_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags) 58 { 59 if (ckWARN(WARN_UTF8)) { 60 if (UNICODE_IS_SURROGATE(uv) && 61 !(flags & UNICODE_ALLOW_SURROGATE)) 62 Perl_warner(aTHX_ packWARN(WARN_UTF8), "UTF-16 surrogate 0x%04"UVxf, uv); 63 else if ( 64 ((uv >= 0xFDD0 && uv <= 0xFDEF && 65 !(flags & UNICODE_ALLOW_FDD0)) 66 || 67 ((uv & 0xFFFE) == 0xFFFE && /* Either FFFE or FFFF. */ 68 !(flags & UNICODE_ALLOW_FFFF))) && 69 /* UNICODE_ALLOW_SUPER includes 70 * FFFEs and FFFFs beyond 0x10FFFF. */ 71 ((uv <= PERL_UNICODE_MAX) || 72 !(flags & UNICODE_ALLOW_SUPER)) 73 ) 74 Perl_warner(aTHX_ packWARN(WARN_UTF8), 75 "Unicode character 0x%04"UVxf" is illegal", uv); 76 } 77 if (UNI_IS_INVARIANT(uv)) { 78 *d++ = (U8)UTF_TO_NATIVE(uv); 79 return d; 80 } 81 #if defined(EBCDIC) 82 else { 83 STRLEN len = UNISKIP(uv); 84 U8 *p = d+len-1; 85 while (p > d) { 86 *p-- = (U8)UTF_TO_NATIVE((uv & UTF_CONTINUATION_MASK) | UTF_CONTINUATION_MARK); 87 uv >>= UTF_ACCUMULATION_SHIFT; 88 } 89 *p = (U8)UTF_TO_NATIVE((uv & UTF_START_MASK(len)) | UTF_START_MARK(len)); 90 return d+len; 91 } 92 #else /* Non loop style */ 93 if (uv < 0x800) { 94 *d++ = (U8)(( uv >> 6) | 0xc0); 95 *d++ = (U8)(( uv & 0x3f) | 0x80); 96 return d; 97 } 98 if (uv < 0x10000) { 99 *d++ = (U8)(( uv >> 12) | 0xe0); 100 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 101 *d++ = (U8)(( uv & 0x3f) | 0x80); 102 return d; 103 } 104 if (uv < 0x200000) { 105 *d++ = (U8)(( uv >> 18) | 0xf0); 106 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80); 107 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 108 *d++ = (U8)(( uv & 0x3f) | 0x80); 109 return d; 110 } 111 if (uv < 0x4000000) { 112 *d++ = (U8)(( uv >> 24) | 0xf8); 113 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80); 114 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80); 115 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 116 *d++ = (U8)(( uv & 0x3f) | 0x80); 117 return d; 118 } 119 if (uv < 0x80000000) { 120 *d++ = (U8)(( uv >> 30) | 0xfc); 121 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80); 122 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80); 123 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80); 124 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 125 *d++ = (U8)(( uv & 0x3f) | 0x80); 126 return d; 127 } 128 #ifdef HAS_QUAD 129 if (uv < UTF8_QUAD_MAX) 130 #endif 131 { 132 *d++ = 0xfe; /* Can't match U+FEFF! */ 133 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80); 134 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80); 135 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80); 136 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80); 137 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 138 *d++ = (U8)(( uv & 0x3f) | 0x80); 139 return d; 140 } 141 #ifdef HAS_QUAD 142 { 143 *d++ = 0xff; /* Can't match U+FFFE! */ 144 *d++ = 0x80; /* 6 Reserved bits */ 145 *d++ = (U8)(((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */ 146 *d++ = (U8)(((uv >> 54) & 0x3f) | 0x80); 147 *d++ = (U8)(((uv >> 48) & 0x3f) | 0x80); 148 *d++ = (U8)(((uv >> 42) & 0x3f) | 0x80); 149 *d++ = (U8)(((uv >> 36) & 0x3f) | 0x80); 150 *d++ = (U8)(((uv >> 30) & 0x3f) | 0x80); 151 *d++ = (U8)(((uv >> 24) & 0x3f) | 0x80); 152 *d++ = (U8)(((uv >> 18) & 0x3f) | 0x80); 153 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80); 154 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 155 *d++ = (U8)(( uv & 0x3f) | 0x80); 156 return d; 157 } 158 #endif 159 #endif /* Loop style */ 160 } 161 162 U8 * 163 Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv) 164 { 165 return Perl_uvuni_to_utf8_flags(aTHX_ d, uv, 0); 166 } 167 168 169 /* 170 =for apidoc A|STRLEN|is_utf8_char|U8 *s 171 172 Tests if some arbitrary number of bytes begins in a valid UTF-8 173 character. Note that an INVARIANT (i.e. ASCII) character is a valid 174 UTF-8 character. The actual number of bytes in the UTF-8 character 175 will be returned if it is valid, otherwise 0. 176 177 =cut */ 178 STRLEN 179 Perl_is_utf8_char(pTHX_ U8 *s) 180 { 181 U8 u = *s; 182 STRLEN slen, len; 183 UV uv, ouv; 184 185 if (UTF8_IS_INVARIANT(u)) 186 return 1; 187 188 if (!UTF8_IS_START(u)) 189 return 0; 190 191 len = UTF8SKIP(s); 192 193 if (len < 2 || !UTF8_IS_CONTINUATION(s[1])) 194 return 0; 195 196 slen = len - 1; 197 s++; 198 u &= UTF_START_MASK(len); 199 uv = u; 200 ouv = uv; 201 while (slen--) { 202 if (!UTF8_IS_CONTINUATION(*s)) 203 return 0; 204 uv = UTF8_ACCUMULATE(uv, *s); 205 if (uv < ouv) 206 return 0; 207 ouv = uv; 208 s++; 209 } 210 211 if ((STRLEN)UNISKIP(uv) < len) 212 return 0; 213 214 return len; 215 } 216 217 /* 218 =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len 219 220 Returns true if first C<len> bytes of the given string form a valid 221 UTF-8 string, false otherwise. Note that 'a valid UTF-8 string' does 222 not mean 'a string that contains code points above 0x7F encoded in UTF-8' 223 because a valid ASCII string is a valid UTF-8 string. 224 225 =cut 226 */ 227 228 bool 229 Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len) 230 { 231 U8* x = s; 232 U8* send; 233 STRLEN c; 234 235 if (!len) 236 len = strlen((char *)s); 237 send = s + len; 238 239 while (x < send) { 240 /* Inline the easy bits of is_utf8_char() here for speed... */ 241 if (UTF8_IS_INVARIANT(*x)) 242 c = 1; 243 else if (!UTF8_IS_START(*x)) 244 return FALSE; 245 else { 246 /* ... and call is_utf8_char() only if really needed. */ 247 c = is_utf8_char(x); 248 if (!c) 249 return FALSE; 250 } 251 x += c; 252 } 253 if (x != send) 254 return FALSE; 255 256 return TRUE; 257 } 258 259 /* 260 =for apidoc A|bool|is_utf8_string_loc|U8 *s|STRLEN len|U8 **p 261 262 Like is_ut8_string but store the location of the failure in 263 the last argument. 264 265 =cut 266 */ 267 268 bool 269 Perl_is_utf8_string_loc(pTHX_ U8 *s, STRLEN len, U8 **p) 270 { 271 U8* x = s; 272 U8* send; 273 STRLEN c; 274 275 if (!len) 276 len = strlen((char *)s); 277 send = s + len; 278 279 while (x < send) { 280 /* Inline the easy bits of is_utf8_char() here for speed... */ 281 if (UTF8_IS_INVARIANT(*x)) 282 c = 1; 283 else if (!UTF8_IS_START(*x)) { 284 if (p) 285 *p = x; 286 return FALSE; 287 } 288 else { 289 /* ... and call is_utf8_char() only if really needed. */ 290 c = is_utf8_char(x); 291 if (!c) { 292 if (p) 293 *p = x; 294 return FALSE; 295 } 296 } 297 x += c; 298 } 299 if (x != send) { 300 if (p) 301 *p = x; 302 return FALSE; 303 } 304 305 return TRUE; 306 } 307 308 /* 309 =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags 310 311 Bottom level UTF-8 decode routine. 312 Returns the unicode code point value of the first character in the string C<s> 313 which is assumed to be in UTF-8 encoding and no longer than C<curlen>; 314 C<retlen> will be set to the length, in bytes, of that character. 315 316 If C<s> does not point to a well-formed UTF-8 character, the behaviour 317 is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY, 318 it is assumed that the caller will raise a warning, and this function 319 will silently just set C<retlen> to C<-1> and return zero. If the 320 C<flags> does not contain UTF8_CHECK_ONLY, warnings about 321 malformations will be given, C<retlen> will be set to the expected 322 length of the UTF-8 character in bytes, and zero will be returned. 323 324 The C<flags> can also contain various flags to allow deviations from 325 the strict UTF-8 encoding (see F<utf8.h>). 326 327 Most code should use utf8_to_uvchr() rather than call this directly. 328 329 =cut 330 */ 331 332 UV 333 Perl_utf8n_to_uvuni(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags) 334 { 335 U8 *s0 = s; 336 UV uv = *s, ouv = 0; 337 STRLEN len = 1; 338 bool dowarn = ckWARN_d(WARN_UTF8); 339 UV startbyte = *s; 340 STRLEN expectlen = 0; 341 U32 warning = 0; 342 343 /* This list is a superset of the UTF8_ALLOW_XXX. */ 344 345 #define UTF8_WARN_EMPTY 1 346 #define UTF8_WARN_CONTINUATION 2 347 #define UTF8_WARN_NON_CONTINUATION 3 348 #define UTF8_WARN_FE_FF 4 349 #define UTF8_WARN_SHORT 5 350 #define UTF8_WARN_OVERFLOW 6 351 #define UTF8_WARN_SURROGATE 7 352 #define UTF8_WARN_LONG 8 353 #define UTF8_WARN_FFFF 9 /* Also FFFE. */ 354 355 if (curlen == 0 && 356 !(flags & UTF8_ALLOW_EMPTY)) { 357 warning = UTF8_WARN_EMPTY; 358 goto malformed; 359 } 360 361 if (UTF8_IS_INVARIANT(uv)) { 362 if (retlen) 363 *retlen = 1; 364 return (UV) (NATIVE_TO_UTF(*s)); 365 } 366 367 if (UTF8_IS_CONTINUATION(uv) && 368 !(flags & UTF8_ALLOW_CONTINUATION)) { 369 warning = UTF8_WARN_CONTINUATION; 370 goto malformed; 371 } 372 373 if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) && 374 !(flags & UTF8_ALLOW_NON_CONTINUATION)) { 375 warning = UTF8_WARN_NON_CONTINUATION; 376 goto malformed; 377 } 378 379 #ifdef EBCDIC 380 uv = NATIVE_TO_UTF(uv); 381 #else 382 if ((uv == 0xfe || uv == 0xff) && 383 !(flags & UTF8_ALLOW_FE_FF)) { 384 warning = UTF8_WARN_FE_FF; 385 goto malformed; 386 } 387 #endif 388 389 if (!(uv & 0x20)) { len = 2; uv &= 0x1f; } 390 else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; } 391 else if (!(uv & 0x08)) { len = 4; uv &= 0x07; } 392 else if (!(uv & 0x04)) { len = 5; uv &= 0x03; } 393 #ifdef EBCDIC 394 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } 395 else { len = 7; uv &= 0x01; } 396 #else 397 else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } 398 else if (!(uv & 0x01)) { len = 7; uv = 0; } 399 else { len = 13; uv = 0; } /* whoa! */ 400 #endif 401 402 if (retlen) 403 *retlen = len; 404 405 expectlen = len; 406 407 if ((curlen < expectlen) && 408 !(flags & UTF8_ALLOW_SHORT)) { 409 warning = UTF8_WARN_SHORT; 410 goto malformed; 411 } 412 413 len--; 414 s++; 415 ouv = uv; 416 417 while (len--) { 418 if (!UTF8_IS_CONTINUATION(*s) && 419 !(flags & UTF8_ALLOW_NON_CONTINUATION)) { 420 s--; 421 warning = UTF8_WARN_NON_CONTINUATION; 422 goto malformed; 423 } 424 else 425 uv = UTF8_ACCUMULATE(uv, *s); 426 if (!(uv > ouv)) { 427 /* These cannot be allowed. */ 428 if (uv == ouv) { 429 if (!(flags & UTF8_ALLOW_LONG)) { 430 warning = UTF8_WARN_LONG; 431 goto malformed; 432 } 433 } 434 else { /* uv < ouv */ 435 /* This cannot be allowed. */ 436 warning = UTF8_WARN_OVERFLOW; 437 goto malformed; 438 } 439 } 440 s++; 441 ouv = uv; 442 } 443 444 if (UNICODE_IS_SURROGATE(uv) && 445 !(flags & UTF8_ALLOW_SURROGATE)) { 446 warning = UTF8_WARN_SURROGATE; 447 goto malformed; 448 } else if ((expectlen > (STRLEN)UNISKIP(uv)) && 449 !(flags & UTF8_ALLOW_LONG)) { 450 warning = UTF8_WARN_LONG; 451 goto malformed; 452 } else if (UNICODE_IS_ILLEGAL(uv) && 453 !(flags & UTF8_ALLOW_FFFF)) { 454 warning = UTF8_WARN_FFFF; 455 goto malformed; 456 } 457 458 return uv; 459 460 malformed: 461 462 if (flags & UTF8_CHECK_ONLY) { 463 if (retlen) 464 *retlen = -1; 465 return 0; 466 } 467 468 if (dowarn) { 469 SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0)); 470 471 switch (warning) { 472 case 0: /* Intentionally empty. */ break; 473 case UTF8_WARN_EMPTY: 474 Perl_sv_catpvf(aTHX_ sv, "(empty string)"); 475 break; 476 case UTF8_WARN_CONTINUATION: 477 Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf", with no preceding start byte)", uv); 478 break; 479 case UTF8_WARN_NON_CONTINUATION: 480 if (s == s0) 481 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", immediately after start byte 0x%02"UVxf")", 482 (UV)s[1], startbyte); 483 else 484 Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf", %d byte%s after start byte 0x%02"UVxf", expected %d bytes)", 485 (UV)s[1], s - s0, s - s0 > 1 ? "s" : "", startbyte, expectlen); 486 487 break; 488 case UTF8_WARN_FE_FF: 489 Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv); 490 break; 491 case UTF8_WARN_SHORT: 492 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")", 493 curlen, curlen == 1 ? "" : "s", expectlen, startbyte); 494 expectlen = curlen; /* distance for caller to skip */ 495 break; 496 case UTF8_WARN_OVERFLOW: 497 Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x, after start byte 0x%02"UVxf")", 498 ouv, *s, startbyte); 499 break; 500 case UTF8_WARN_SURROGATE: 501 Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv); 502 break; 503 case UTF8_WARN_LONG: 504 Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d, after start byte 0x%02"UVxf")", 505 expectlen, expectlen == 1 ? "": "s", UNISKIP(uv), startbyte); 506 break; 507 case UTF8_WARN_FFFF: 508 Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv); 509 break; 510 default: 511 Perl_sv_catpvf(aTHX_ sv, "(unknown reason)"); 512 break; 513 } 514 515 if (warning) { 516 char *s = SvPVX(sv); 517 518 if (PL_op) 519 Perl_warner(aTHX_ packWARN(WARN_UTF8), 520 "%s in %s", s, OP_DESC(PL_op)); 521 else 522 Perl_warner(aTHX_ packWARN(WARN_UTF8), "%s", s); 523 } 524 } 525 526 if (retlen) 527 *retlen = expectlen ? expectlen : len; 528 529 return 0; 530 } 531 532 /* 533 =for apidoc A|UV|utf8_to_uvchr|U8 *s|STRLEN *retlen 534 535 Returns the native character value of the first character in the string C<s> 536 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the 537 length, in bytes, of that character. 538 539 If C<s> does not point to a well-formed UTF-8 character, zero is 540 returned and retlen is set, if possible, to -1. 541 542 =cut 543 */ 544 545 UV 546 Perl_utf8_to_uvchr(pTHX_ U8 *s, STRLEN *retlen) 547 { 548 return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 549 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); 550 } 551 552 /* 553 =for apidoc A|UV|utf8_to_uvuni|U8 *s|STRLEN *retlen 554 555 Returns the Unicode code point of the first character in the string C<s> 556 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the 557 length, in bytes, of that character. 558 559 This function should only be used when returned UV is considered 560 an index into the Unicode semantic tables (e.g. swashes). 561 562 If C<s> does not point to a well-formed UTF-8 character, zero is 563 returned and retlen is set, if possible, to -1. 564 565 =cut 566 */ 567 568 UV 569 Perl_utf8_to_uvuni(pTHX_ U8 *s, STRLEN *retlen) 570 { 571 /* Call the low level routine asking for checks */ 572 return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 573 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); 574 } 575 576 /* 577 =for apidoc A|STRLEN|utf8_length|U8 *s|U8 *e 578 579 Return the length of the UTF-8 char encoded string C<s> in characters. 580 Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end 581 up past C<e>, croaks. 582 583 =cut 584 */ 585 586 STRLEN 587 Perl_utf8_length(pTHX_ U8 *s, U8 *e) 588 { 589 STRLEN len = 0; 590 591 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g. 592 * the bitops (especially ~) can create illegal UTF-8. 593 * In other words: in Perl UTF-8 is not just for Unicode. */ 594 595 if (e < s) { 596 if (ckWARN_d(WARN_UTF8)) { 597 if (PL_op) 598 Perl_warner(aTHX_ packWARN(WARN_UTF8), 599 "%s in %s", unees, OP_DESC(PL_op)); 600 else 601 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees); 602 } 603 return 0; 604 } 605 while (s < e) { 606 U8 t = UTF8SKIP(s); 607 608 if (e - s < t) { 609 if (ckWARN_d(WARN_UTF8)) { 610 if (PL_op) 611 Perl_warner(aTHX_ packWARN(WARN_UTF8), 612 unees, OP_DESC(PL_op)); 613 else 614 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees); 615 } 616 return len; 617 } 618 s += t; 619 len++; 620 } 621 622 return len; 623 } 624 625 /* 626 =for apidoc A|IV|utf8_distance|U8 *a|U8 *b 627 628 Returns the number of UTF-8 characters between the UTF-8 pointers C<a> 629 and C<b>. 630 631 WARNING: use only if you *know* that the pointers point inside the 632 same UTF-8 buffer. 633 634 =cut 635 */ 636 637 IV 638 Perl_utf8_distance(pTHX_ U8 *a, U8 *b) 639 { 640 IV off = 0; 641 642 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g. 643 * the bitops (especially ~) can create illegal UTF-8. 644 * In other words: in Perl UTF-8 is not just for Unicode. */ 645 646 if (a < b) { 647 while (a < b) { 648 U8 c = UTF8SKIP(a); 649 650 if (b - a < c) { 651 if (ckWARN_d(WARN_UTF8)) { 652 if (PL_op) 653 Perl_warner(aTHX_ packWARN(WARN_UTF8), 654 "%s in %s", unees, OP_DESC(PL_op)); 655 else 656 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees); 657 } 658 return off; 659 } 660 a += c; 661 off--; 662 } 663 } 664 else { 665 while (b < a) { 666 U8 c = UTF8SKIP(b); 667 668 if (a - b < c) { 669 if (ckWARN_d(WARN_UTF8)) { 670 if (PL_op) 671 Perl_warner(aTHX_ packWARN(WARN_UTF8), 672 "%s in %s", unees, OP_DESC(PL_op)); 673 else 674 Perl_warner(aTHX_ packWARN(WARN_UTF8), unees); 675 } 676 return off; 677 } 678 b += c; 679 off++; 680 } 681 } 682 683 return off; 684 } 685 686 /* 687 =for apidoc A|U8 *|utf8_hop|U8 *s|I32 off 688 689 Return the UTF-8 pointer C<s> displaced by C<off> characters, either 690 forward or backward. 691 692 WARNING: do not use the following unless you *know* C<off> is within 693 the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned 694 on the first byte of character or just after the last byte of a character. 695 696 =cut 697 */ 698 699 U8 * 700 Perl_utf8_hop(pTHX_ U8 *s, I32 off) 701 { 702 /* Note: cannot use UTF8_IS_...() too eagerly here since e.g 703 * the bitops (especially ~) can create illegal UTF-8. 704 * In other words: in Perl UTF-8 is not just for Unicode. */ 705 706 if (off >= 0) { 707 while (off--) 708 s += UTF8SKIP(s); 709 } 710 else { 711 while (off++) { 712 s--; 713 while (UTF8_IS_CONTINUATION(*s)) 714 s--; 715 } 716 } 717 return s; 718 } 719 720 /* 721 =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len 722 723 Converts a string C<s> of length C<len> from UTF-8 into byte encoding. 724 Unlike C<bytes_to_utf8>, this over-writes the original string, and 725 updates len to contain the new length. 726 Returns zero on failure, setting C<len> to -1. 727 728 =cut 729 */ 730 731 U8 * 732 Perl_utf8_to_bytes(pTHX_ U8 *s, STRLEN *len) 733 { 734 U8 *send; 735 U8 *d; 736 U8 *save = s; 737 738 /* ensure valid UTF-8 and chars < 256 before updating string */ 739 for (send = s + *len; s < send; ) { 740 U8 c = *s++; 741 742 if (!UTF8_IS_INVARIANT(c) && 743 (!UTF8_IS_DOWNGRADEABLE_START(c) || (s >= send) 744 || !(c = *s++) || !UTF8_IS_CONTINUATION(c))) { 745 *len = -1; 746 return 0; 747 } 748 } 749 750 d = s = save; 751 while (s < send) { 752 STRLEN ulen; 753 *d++ = (U8)utf8_to_uvchr(s, &ulen); 754 s += ulen; 755 } 756 *d = '\0'; 757 *len = d - save; 758 return save; 759 } 760 761 /* 762 =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8 763 764 Converts a string C<s> of length C<len> from UTF-8 into byte encoding. 765 Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to 766 the newly-created string, and updates C<len> to contain the new 767 length. Returns the original string if no conversion occurs, C<len> 768 is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to 769 0 if C<s> is converted or contains all 7bit characters. 770 771 =cut 772 */ 773 774 U8 * 775 Perl_bytes_from_utf8(pTHX_ U8 *s, STRLEN *len, bool *is_utf8) 776 { 777 U8 *d; 778 U8 *start = s; 779 U8 *send; 780 I32 count = 0; 781 782 if (!*is_utf8) 783 return start; 784 785 /* ensure valid UTF-8 and chars < 256 before converting string */ 786 for (send = s + *len; s < send;) { 787 U8 c = *s++; 788 if (!UTF8_IS_INVARIANT(c)) { 789 if (UTF8_IS_DOWNGRADEABLE_START(c) && s < send && 790 (c = *s++) && UTF8_IS_CONTINUATION(c)) 791 count++; 792 else 793 return start; 794 } 795 } 796 797 *is_utf8 = 0; 798 799 Newz(801, d, (*len) - count + 1, U8); 800 s = start; start = d; 801 while (s < send) { 802 U8 c = *s++; 803 if (!UTF8_IS_INVARIANT(c)) { 804 /* Then it is two-byte encoded */ 805 c = UTF8_ACCUMULATE(NATIVE_TO_UTF(c), *s++); 806 c = ASCII_TO_NATIVE(c); 807 } 808 *d++ = c; 809 } 810 *d = '\0'; 811 *len = d - start; 812 return start; 813 } 814 815 /* 816 =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len 817 818 Converts a string C<s> of length C<len> from ASCII into UTF-8 encoding. 819 Returns a pointer to the newly-created string, and sets C<len> to 820 reflect the new length. 821 822 If you want to convert to UTF-8 from other encodings than ASCII, 823 see sv_recode_to_utf8(). 824 825 =cut 826 */ 827 828 U8* 829 Perl_bytes_to_utf8(pTHX_ U8 *s, STRLEN *len) 830 { 831 U8 *send; 832 U8 *d; 833 U8 *dst; 834 send = s + (*len); 835 836 Newz(801, d, (*len) * 2 + 1, U8); 837 dst = d; 838 839 while (s < send) { 840 UV uv = NATIVE_TO_ASCII(*s++); 841 if (UNI_IS_INVARIANT(uv)) 842 *d++ = (U8)UTF_TO_NATIVE(uv); 843 else { 844 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); 845 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); 846 } 847 } 848 *d = '\0'; 849 *len = d-dst; 850 return dst; 851 } 852 853 /* 854 * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8. 855 * 856 * Destination must be pre-extended to 3/2 source. Do not use in-place. 857 * We optimize for native, for obvious reasons. */ 858 859 U8* 860 Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) 861 { 862 U8* pend; 863 U8* dstart = d; 864 865 if (bytelen & 1) 866 Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen"); 867 868 pend = p + bytelen; 869 870 while (p < pend) { 871 UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */ 872 p += 2; 873 if (uv < 0x80) { 874 *d++ = (U8)uv; 875 continue; 876 } 877 if (uv < 0x800) { 878 *d++ = (U8)(( uv >> 6) | 0xc0); 879 *d++ = (U8)(( uv & 0x3f) | 0x80); 880 continue; 881 } 882 if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */ 883 UV low = (p[0] << 8) + p[1]; 884 p += 2; 885 if (low < 0xdc00 || low >= 0xdfff) 886 Perl_croak(aTHX_ "Malformed UTF-16 surrogate"); 887 uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000; 888 } 889 if (uv < 0x10000) { 890 *d++ = (U8)(( uv >> 12) | 0xe0); 891 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 892 *d++ = (U8)(( uv & 0x3f) | 0x80); 893 continue; 894 } 895 else { 896 *d++ = (U8)(( uv >> 18) | 0xf0); 897 *d++ = (U8)(((uv >> 12) & 0x3f) | 0x80); 898 *d++ = (U8)(((uv >> 6) & 0x3f) | 0x80); 899 *d++ = (U8)(( uv & 0x3f) | 0x80); 900 continue; 901 } 902 } 903 *newlen = d - dstart; 904 return d; 905 } 906 907 /* Note: this one is slightly destructive of the source. */ 908 909 U8* 910 Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) 911 { 912 U8* s = (U8*)p; 913 U8* send = s + bytelen; 914 while (s < send) { 915 U8 tmp = s[0]; 916 s[0] = s[1]; 917 s[1] = tmp; 918 s += 2; 919 } 920 return utf16_to_utf8(p, d, bytelen, newlen); 921 } 922 923 /* for now these are all defined (inefficiently) in terms of the utf8 versions */ 924 925 bool 926 Perl_is_uni_alnum(pTHX_ UV c) 927 { 928 U8 tmpbuf[UTF8_MAXLEN+1]; 929 uvchr_to_utf8(tmpbuf, c); 930 return is_utf8_alnum(tmpbuf); 931 } 932 933 bool 934 Perl_is_uni_alnumc(pTHX_ UV c) 935 { 936 U8 tmpbuf[UTF8_MAXLEN+1]; 937 uvchr_to_utf8(tmpbuf, c); 938 return is_utf8_alnumc(tmpbuf); 939 } 940 941 bool 942 Perl_is_uni_idfirst(pTHX_ UV c) 943 { 944 U8 tmpbuf[UTF8_MAXLEN+1]; 945 uvchr_to_utf8(tmpbuf, c); 946 return is_utf8_idfirst(tmpbuf); 947 } 948 949 bool 950 Perl_is_uni_alpha(pTHX_ UV c) 951 { 952 U8 tmpbuf[UTF8_MAXLEN+1]; 953 uvchr_to_utf8(tmpbuf, c); 954 return is_utf8_alpha(tmpbuf); 955 } 956 957 bool 958 Perl_is_uni_ascii(pTHX_ UV c) 959 { 960 U8 tmpbuf[UTF8_MAXLEN+1]; 961 uvchr_to_utf8(tmpbuf, c); 962 return is_utf8_ascii(tmpbuf); 963 } 964 965 bool 966 Perl_is_uni_space(pTHX_ UV c) 967 { 968 U8 tmpbuf[UTF8_MAXLEN+1]; 969 uvchr_to_utf8(tmpbuf, c); 970 return is_utf8_space(tmpbuf); 971 } 972 973 bool 974 Perl_is_uni_digit(pTHX_ UV c) 975 { 976 U8 tmpbuf[UTF8_MAXLEN+1]; 977 uvchr_to_utf8(tmpbuf, c); 978 return is_utf8_digit(tmpbuf); 979 } 980 981 bool 982 Perl_is_uni_upper(pTHX_ UV c) 983 { 984 U8 tmpbuf[UTF8_MAXLEN+1]; 985 uvchr_to_utf8(tmpbuf, c); 986 return is_utf8_upper(tmpbuf); 987 } 988 989 bool 990 Perl_is_uni_lower(pTHX_ UV c) 991 { 992 U8 tmpbuf[UTF8_MAXLEN+1]; 993 uvchr_to_utf8(tmpbuf, c); 994 return is_utf8_lower(tmpbuf); 995 } 996 997 bool 998 Perl_is_uni_cntrl(pTHX_ UV c) 999 { 1000 U8 tmpbuf[UTF8_MAXLEN+1]; 1001 uvchr_to_utf8(tmpbuf, c); 1002 return is_utf8_cntrl(tmpbuf); 1003 } 1004 1005 bool 1006 Perl_is_uni_graph(pTHX_ UV c) 1007 { 1008 U8 tmpbuf[UTF8_MAXLEN+1]; 1009 uvchr_to_utf8(tmpbuf, c); 1010 return is_utf8_graph(tmpbuf); 1011 } 1012 1013 bool 1014 Perl_is_uni_print(pTHX_ UV c) 1015 { 1016 U8 tmpbuf[UTF8_MAXLEN+1]; 1017 uvchr_to_utf8(tmpbuf, c); 1018 return is_utf8_print(tmpbuf); 1019 } 1020 1021 bool 1022 Perl_is_uni_punct(pTHX_ UV c) 1023 { 1024 U8 tmpbuf[UTF8_MAXLEN+1]; 1025 uvchr_to_utf8(tmpbuf, c); 1026 return is_utf8_punct(tmpbuf); 1027 } 1028 1029 bool 1030 Perl_is_uni_xdigit(pTHX_ UV c) 1031 { 1032 U8 tmpbuf[UTF8_MAXLEN_UCLC+1]; 1033 uvchr_to_utf8(tmpbuf, c); 1034 return is_utf8_xdigit(tmpbuf); 1035 } 1036 1037 UV 1038 Perl_to_uni_upper(pTHX_ UV c, U8* p, STRLEN *lenp) 1039 { 1040 uvchr_to_utf8(p, c); 1041 return to_utf8_upper(p, p, lenp); 1042 } 1043 1044 UV 1045 Perl_to_uni_title(pTHX_ UV c, U8* p, STRLEN *lenp) 1046 { 1047 uvchr_to_utf8(p, c); 1048 return to_utf8_title(p, p, lenp); 1049 } 1050 1051 UV 1052 Perl_to_uni_lower(pTHX_ UV c, U8* p, STRLEN *lenp) 1053 { 1054 uvchr_to_utf8(p, c); 1055 return to_utf8_lower(p, p, lenp); 1056 } 1057 1058 UV 1059 Perl_to_uni_fold(pTHX_ UV c, U8* p, STRLEN *lenp) 1060 { 1061 uvchr_to_utf8(p, c); 1062 return to_utf8_fold(p, p, lenp); 1063 } 1064 1065 /* for now these all assume no locale info available for Unicode > 255 */ 1066 1067 bool 1068 Perl_is_uni_alnum_lc(pTHX_ UV c) 1069 { 1070 return is_uni_alnum(c); /* XXX no locale support yet */ 1071 } 1072 1073 bool 1074 Perl_is_uni_alnumc_lc(pTHX_ UV c) 1075 { 1076 return is_uni_alnumc(c); /* XXX no locale support yet */ 1077 } 1078 1079 bool 1080 Perl_is_uni_idfirst_lc(pTHX_ UV c) 1081 { 1082 return is_uni_idfirst(c); /* XXX no locale support yet */ 1083 } 1084 1085 bool 1086 Perl_is_uni_alpha_lc(pTHX_ UV c) 1087 { 1088 return is_uni_alpha(c); /* XXX no locale support yet */ 1089 } 1090 1091 bool 1092 Perl_is_uni_ascii_lc(pTHX_ UV c) 1093 { 1094 return is_uni_ascii(c); /* XXX no locale support yet */ 1095 } 1096 1097 bool 1098 Perl_is_uni_space_lc(pTHX_ UV c) 1099 { 1100 return is_uni_space(c); /* XXX no locale support yet */ 1101 } 1102 1103 bool 1104 Perl_is_uni_digit_lc(pTHX_ UV c) 1105 { 1106 return is_uni_digit(c); /* XXX no locale support yet */ 1107 } 1108 1109 bool 1110 Perl_is_uni_upper_lc(pTHX_ UV c) 1111 { 1112 return is_uni_upper(c); /* XXX no locale support yet */ 1113 } 1114 1115 bool 1116 Perl_is_uni_lower_lc(pTHX_ UV c) 1117 { 1118 return is_uni_lower(c); /* XXX no locale support yet */ 1119 } 1120 1121 bool 1122 Perl_is_uni_cntrl_lc(pTHX_ UV c) 1123 { 1124 return is_uni_cntrl(c); /* XXX no locale support yet */ 1125 } 1126 1127 bool 1128 Perl_is_uni_graph_lc(pTHX_ UV c) 1129 { 1130 return is_uni_graph(c); /* XXX no locale support yet */ 1131 } 1132 1133 bool 1134 Perl_is_uni_print_lc(pTHX_ UV c) 1135 { 1136 return is_uni_print(c); /* XXX no locale support yet */ 1137 } 1138 1139 bool 1140 Perl_is_uni_punct_lc(pTHX_ UV c) 1141 { 1142 return is_uni_punct(c); /* XXX no locale support yet */ 1143 } 1144 1145 bool 1146 Perl_is_uni_xdigit_lc(pTHX_ UV c) 1147 { 1148 return is_uni_xdigit(c); /* XXX no locale support yet */ 1149 } 1150 1151 U32 1152 Perl_to_uni_upper_lc(pTHX_ U32 c) 1153 { 1154 /* XXX returns only the first character -- do not use XXX */ 1155 /* XXX no locale support yet */ 1156 STRLEN len; 1157 U8 tmpbuf[UTF8_MAXLEN_UCLC+1]; 1158 return (U32)to_uni_upper(c, tmpbuf, &len); 1159 } 1160 1161 U32 1162 Perl_to_uni_title_lc(pTHX_ U32 c) 1163 { 1164 /* XXX returns only the first character XXX -- do not use XXX */ 1165 /* XXX no locale support yet */ 1166 STRLEN len; 1167 U8 tmpbuf[UTF8_MAXLEN_UCLC+1]; 1168 return (U32)to_uni_title(c, tmpbuf, &len); 1169 } 1170 1171 U32 1172 Perl_to_uni_lower_lc(pTHX_ U32 c) 1173 { 1174 /* XXX returns only the first character -- do not use XXX */ 1175 /* XXX no locale support yet */ 1176 STRLEN len; 1177 U8 tmpbuf[UTF8_MAXLEN_UCLC+1]; 1178 return (U32)to_uni_lower(c, tmpbuf, &len); 1179 } 1180 1181 bool 1182 Perl_is_utf8_alnum(pTHX_ U8 *p) 1183 { 1184 if (!is_utf8_char(p)) 1185 return FALSE; 1186 if (!PL_utf8_alnum) 1187 /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true 1188 * descendant of isalnum(3), in other words, it doesn't 1189 * contain the '_'. --jhi */ 1190 PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0); 1191 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0; 1192 /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */ 1193 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ 1194 if (!PL_utf8_alnum) 1195 PL_utf8_alnum = swash_init("utf8", "", 1196 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); 1197 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0; 1198 #endif 1199 } 1200 1201 bool 1202 Perl_is_utf8_alnumc(pTHX_ U8 *p) 1203 { 1204 if (!is_utf8_char(p)) 1205 return FALSE; 1206 if (!PL_utf8_alnum) 1207 PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0); 1208 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0; 1209 /* return is_utf8_alpha(p) || is_utf8_digit(p); */ 1210 #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ 1211 if (!PL_utf8_alnum) 1212 PL_utf8_alnum = swash_init("utf8", "", 1213 sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); 1214 return swash_fetch(PL_utf8_alnum, p, TRUE) != 0; 1215 #endif 1216 } 1217 1218 bool 1219 Perl_is_utf8_idfirst(pTHX_ U8 *p) /* The naming is historical. */ 1220 { 1221 if (*p == '_') 1222 return TRUE; 1223 if (!is_utf8_char(p)) 1224 return FALSE; 1225 if (!PL_utf8_idstart) /* is_utf8_idstart would be more logical. */ 1226 PL_utf8_idstart = swash_init("utf8", "IdStart", &PL_sv_undef, 0, 0); 1227 return swash_fetch(PL_utf8_idstart, p, TRUE) != 0; 1228 } 1229 1230 bool 1231 Perl_is_utf8_idcont(pTHX_ U8 *p) 1232 { 1233 if (*p == '_') 1234 return TRUE; 1235 if (!is_utf8_char(p)) 1236 return FALSE; 1237 if (!PL_utf8_idcont) 1238 PL_utf8_idcont = swash_init("utf8", "IdContinue", &PL_sv_undef, 0, 0); 1239 return swash_fetch(PL_utf8_idcont, p, TRUE) != 0; 1240 } 1241 1242 bool 1243 Perl_is_utf8_alpha(pTHX_ U8 *p) 1244 { 1245 if (!is_utf8_char(p)) 1246 return FALSE; 1247 if (!PL_utf8_alpha) 1248 PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0); 1249 return swash_fetch(PL_utf8_alpha, p, TRUE) != 0; 1250 } 1251 1252 bool 1253 Perl_is_utf8_ascii(pTHX_ U8 *p) 1254 { 1255 if (!is_utf8_char(p)) 1256 return FALSE; 1257 if (!PL_utf8_ascii) 1258 PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0); 1259 return swash_fetch(PL_utf8_ascii, p, TRUE) != 0; 1260 } 1261 1262 bool 1263 Perl_is_utf8_space(pTHX_ U8 *p) 1264 { 1265 if (!is_utf8_char(p)) 1266 return FALSE; 1267 if (!PL_utf8_space) 1268 PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0); 1269 return swash_fetch(PL_utf8_space, p, TRUE) != 0; 1270 } 1271 1272 bool 1273 Perl_is_utf8_digit(pTHX_ U8 *p) 1274 { 1275 if (!is_utf8_char(p)) 1276 return FALSE; 1277 if (!PL_utf8_digit) 1278 PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0); 1279 return swash_fetch(PL_utf8_digit, p, TRUE) != 0; 1280 } 1281 1282 bool 1283 Perl_is_utf8_upper(pTHX_ U8 *p) 1284 { 1285 if (!is_utf8_char(p)) 1286 return FALSE; 1287 if (!PL_utf8_upper) 1288 PL_utf8_upper = swash_init("utf8", "IsUppercase", &PL_sv_undef, 0, 0); 1289 return swash_fetch(PL_utf8_upper, p, TRUE) != 0; 1290 } 1291 1292 bool 1293 Perl_is_utf8_lower(pTHX_ U8 *p) 1294 { 1295 if (!is_utf8_char(p)) 1296 return FALSE; 1297 if (!PL_utf8_lower) 1298 PL_utf8_lower = swash_init("utf8", "IsLowercase", &PL_sv_undef, 0, 0); 1299 return swash_fetch(PL_utf8_lower, p, TRUE) != 0; 1300 } 1301 1302 bool 1303 Perl_is_utf8_cntrl(pTHX_ U8 *p) 1304 { 1305 if (!is_utf8_char(p)) 1306 return FALSE; 1307 if (!PL_utf8_cntrl) 1308 PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0); 1309 return swash_fetch(PL_utf8_cntrl, p, TRUE) != 0; 1310 } 1311 1312 bool 1313 Perl_is_utf8_graph(pTHX_ U8 *p) 1314 { 1315 if (!is_utf8_char(p)) 1316 return FALSE; 1317 if (!PL_utf8_graph) 1318 PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0); 1319 return swash_fetch(PL_utf8_graph, p, TRUE) != 0; 1320 } 1321 1322 bool 1323 Perl_is_utf8_print(pTHX_ U8 *p) 1324 { 1325 if (!is_utf8_char(p)) 1326 return FALSE; 1327 if (!PL_utf8_print) 1328 PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0); 1329 return swash_fetch(PL_utf8_print, p, TRUE) != 0; 1330 } 1331 1332 bool 1333 Perl_is_utf8_punct(pTHX_ U8 *p) 1334 { 1335 if (!is_utf8_char(p)) 1336 return FALSE; 1337 if (!PL_utf8_punct) 1338 PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0); 1339 return swash_fetch(PL_utf8_punct, p, TRUE) != 0; 1340 } 1341 1342 bool 1343 Perl_is_utf8_xdigit(pTHX_ U8 *p) 1344 { 1345 if (!is_utf8_char(p)) 1346 return FALSE; 1347 if (!PL_utf8_xdigit) 1348 PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0); 1349 return swash_fetch(PL_utf8_xdigit, p, TRUE) != 0; 1350 } 1351 1352 bool 1353 Perl_is_utf8_mark(pTHX_ U8 *p) 1354 { 1355 if (!is_utf8_char(p)) 1356 return FALSE; 1357 if (!PL_utf8_mark) 1358 PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0); 1359 return swash_fetch(PL_utf8_mark, p, TRUE) != 0; 1360 } 1361 1362 /* 1363 =for apidoc A|UV|to_utf8_case|U8 *p|U8* ustrp|STRLEN *lenp|SV **swash|char *normal|char *special 1364 1365 The "p" contains the pointer to the UTF-8 string encoding 1366 the character that is being converted. 1367 1368 The "ustrp" is a pointer to the character buffer to put the 1369 conversion result to. The "lenp" is a pointer to the length 1370 of the result. 1371 1372 The "swashp" is a pointer to the swash to use. 1373 1374 Both the special and normal mappings are stored lib/unicore/To/Foo.pl, 1375 and loaded by SWASHGET, using lib/utf8_heavy.pl. The special (usually, 1376 but not always, a multicharacter mapping), is tried first. 1377 1378 The "special" is a string like "utf8::ToSpecLower", which means the 1379 hash %utf8::ToSpecLower. The access to the hash is through 1380 Perl_to_utf8_case(). 1381 1382 The "normal" is a string like "ToLower" which means the swash 1383 %utf8::ToLower. 1384 1385 =cut */ 1386 1387 UV 1388 Perl_to_utf8_case(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp, SV **swashp, char *normal, char *special) 1389 { 1390 UV uv0, uv1; 1391 U8 tmpbuf[UTF8_MAXLEN_FOLD+1]; 1392 STRLEN len = 0; 1393 1394 uv0 = utf8_to_uvchr(p, 0); 1395 /* The NATIVE_TO_UNI() and UNI_TO_NATIVE() mappings 1396 * are necessary in EBCDIC, they are redundant no-ops 1397 * in ASCII-ish platforms, and hopefully optimized away. */ 1398 uv1 = NATIVE_TO_UNI(uv0); 1399 uvuni_to_utf8(tmpbuf, uv1); 1400 1401 if (!*swashp) /* load on-demand */ 1402 *swashp = swash_init("utf8", normal, &PL_sv_undef, 4, 0); 1403 1404 if (special) { 1405 /* It might be "special" (sometimes, but not always, 1406 * a multicharacter mapping) */ 1407 HV *hv; 1408 SV *keysv; 1409 HE *he; 1410 SV *val; 1411 1412 if ((hv = get_hv(special, FALSE)) && 1413 (keysv = sv_2mortal(Perl_newSVpvf(aTHX_ "%04"UVXf, uv1))) && 1414 (he = hv_fetch_ent(hv, keysv, FALSE, 0)) && 1415 (val = HeVAL(he))) { 1416 char *s; 1417 1418 s = SvPV(val, len); 1419 if (len == 1) 1420 len = uvuni_to_utf8(ustrp, NATIVE_TO_UNI(*(U8*)s)) - ustrp; 1421 else { 1422 #ifdef EBCDIC 1423 /* If we have EBCDIC we need to remap the characters 1424 * since any characters in the low 256 are Unicode 1425 * code points, not EBCDIC. */ 1426 U8 *t = (U8*)s, *tend = t + len, *d; 1427 1428 d = tmpbuf; 1429 if (SvUTF8(val)) { 1430 STRLEN tlen = 0; 1431 1432 while (t < tend) { 1433 UV c = utf8_to_uvchr(t, &tlen); 1434 if (tlen > 0) { 1435 d = uvchr_to_utf8(d, UNI_TO_NATIVE(c)); 1436 t += tlen; 1437 } 1438 else 1439 break; 1440 } 1441 } 1442 else { 1443 while (t < tend) { 1444 d = uvchr_to_utf8(d, UNI_TO_NATIVE(*t)); 1445 t++; 1446 } 1447 } 1448 len = d - tmpbuf; 1449 Copy(tmpbuf, ustrp, len, U8); 1450 #else 1451 Copy(s, ustrp, len, U8); 1452 #endif 1453 } 1454 } 1455 } 1456 1457 if (!len && *swashp) { 1458 UV uv2 = swash_fetch(*swashp, tmpbuf, TRUE); 1459 1460 if (uv2) { 1461 /* It was "normal" (a single character mapping). */ 1462 UV uv3 = UNI_TO_NATIVE(uv2); 1463 1464 len = uvchr_to_utf8(ustrp, uv3) - ustrp; 1465 } 1466 } 1467 1468 if (!len) /* Neither: just copy. */ 1469 len = uvchr_to_utf8(ustrp, uv0) - ustrp; 1470 1471 if (lenp) 1472 *lenp = len; 1473 1474 return len ? utf8_to_uvchr(ustrp, 0) : 0; 1475 } 1476 1477 /* 1478 =for apidoc A|UV|to_utf8_upper|U8 *p|U8 *ustrp|STRLEN *lenp 1479 1480 Convert the UTF-8 encoded character at p to its uppercase version and 1481 store that in UTF-8 in ustrp and its length in bytes in lenp. Note 1482 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the 1483 uppercase version may be longer than the original character (up to two 1484 characters). 1485 1486 The first character of the uppercased version is returned 1487 (but note, as explained above, that there may be more.) 1488 1489 =cut */ 1490 1491 UV 1492 Perl_to_utf8_upper(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp) 1493 { 1494 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp, 1495 &PL_utf8_toupper, "ToUpper", "utf8::ToSpecUpper"); 1496 } 1497 1498 /* 1499 =for apidoc A|UV|to_utf8_title|U8 *p|U8 *ustrp|STRLEN *lenp 1500 1501 Convert the UTF-8 encoded character at p to its titlecase version and 1502 store that in UTF-8 in ustrp and its length in bytes in lenp. Note 1503 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the 1504 titlecase version may be longer than the original character (up to two 1505 characters). 1506 1507 The first character of the titlecased version is returned 1508 (but note, as explained above, that there may be more.) 1509 1510 =cut */ 1511 1512 UV 1513 Perl_to_utf8_title(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp) 1514 { 1515 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp, 1516 &PL_utf8_totitle, "ToTitle", "utf8::ToSpecTitle"); 1517 } 1518 1519 /* 1520 =for apidoc A|UV|to_utf8_lower|U8 *p|U8 *ustrp|STRLEN *lenp 1521 1522 Convert the UTF-8 encoded character at p to its lowercase version and 1523 store that in UTF-8 in ustrp and its length in bytes in lenp. Note 1524 that the ustrp needs to be at least UTF8_MAXLEN_UCLC+1 bytes since the 1525 lowercase version may be longer than the original character (up to two 1526 characters). 1527 1528 The first character of the lowercased version is returned 1529 (but note, as explained above, that there may be more.) 1530 1531 =cut */ 1532 1533 UV 1534 Perl_to_utf8_lower(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp) 1535 { 1536 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp, 1537 &PL_utf8_tolower, "ToLower", "utf8::ToSpecLower"); 1538 } 1539 1540 /* 1541 =for apidoc A|UV|to_utf8_fold|U8 *p|U8 *ustrp|STRLEN *lenp 1542 1543 Convert the UTF-8 encoded character at p to its foldcase version and 1544 store that in UTF-8 in ustrp and its length in bytes in lenp. Note 1545 that the ustrp needs to be at least UTF8_MAXLEN_FOLD+1 bytes since the 1546 foldcase version may be longer than the original character (up to 1547 three characters). 1548 1549 The first character of the foldcased version is returned 1550 (but note, as explained above, that there may be more.) 1551 1552 =cut */ 1553 1554 UV 1555 Perl_to_utf8_fold(pTHX_ U8 *p, U8* ustrp, STRLEN *lenp) 1556 { 1557 return Perl_to_utf8_case(aTHX_ p, ustrp, lenp, 1558 &PL_utf8_tofold, "ToFold", "utf8::ToSpecFold"); 1559 } 1560 1561 /* a "swash" is a swatch hash */ 1562 1563 SV* 1564 Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none) 1565 { 1566 SV* retval; 1567 SV* tokenbufsv = sv_2mortal(NEWSV(0,0)); 1568 dSP; 1569 HV *stash = gv_stashpvn(pkg, strlen(pkg), FALSE); 1570 SV* errsv_save; 1571 1572 if (!gv_fetchmeth(stash, "SWASHNEW", 8, -1)) { /* demand load utf8 */ 1573 ENTER; 1574 errsv_save = newSVsv(ERRSV); 1575 Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv); 1576 if (!SvTRUE(ERRSV)) 1577 sv_setsv(ERRSV, errsv_save); 1578 SvREFCNT_dec(errsv_save); 1579 LEAVE; 1580 } 1581 SPAGAIN; 1582 PUSHSTACKi(PERLSI_MAGIC); 1583 PUSHMARK(SP); 1584 EXTEND(SP,5); 1585 PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg)))); 1586 PUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); 1587 PUSHs(listsv); 1588 PUSHs(sv_2mortal(newSViv(minbits))); 1589 PUSHs(sv_2mortal(newSViv(none))); 1590 PUTBACK; 1591 ENTER; 1592 SAVEI32(PL_hints); 1593 PL_hints = 0; 1594 save_re_context(); 1595 if (IN_PERL_COMPILETIME) { 1596 /* XXX ought to be handled by lex_start */ 1597 SAVEI32(PL_in_my); 1598 PL_in_my = 0; 1599 sv_setpv(tokenbufsv, PL_tokenbuf); 1600 } 1601 errsv_save = newSVsv(ERRSV); 1602 if (call_method("SWASHNEW", G_SCALAR)) 1603 retval = newSVsv(*PL_stack_sp--); 1604 else 1605 retval = &PL_sv_undef; 1606 if (!SvTRUE(ERRSV)) 1607 sv_setsv(ERRSV, errsv_save); 1608 SvREFCNT_dec(errsv_save); 1609 LEAVE; 1610 POPSTACK; 1611 if (IN_PERL_COMPILETIME) { 1612 STRLEN len; 1613 char* pv = SvPV(tokenbufsv, len); 1614 1615 Copy(pv, PL_tokenbuf, len+1, char); 1616 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK); 1617 } 1618 if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) { 1619 if (SvPOK(retval)) 1620 Perl_croak(aTHX_ "Can't find Unicode property definition \"%"SVf"\"", 1621 retval); 1622 Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref"); 1623 } 1624 return retval; 1625 } 1626 1627 1628 /* This API is wrong for special case conversions since we may need to 1629 * return several Unicode characters for a single Unicode character 1630 * (see lib/unicore/SpecCase.txt) The SWASHGET in lib/utf8_heavy.pl is 1631 * the lower-level routine, and it is similarly broken for returning 1632 * multiple values. --jhi */ 1633 UV 1634 Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr, bool do_utf8) 1635 { 1636 HV* hv = (HV*)SvRV(sv); 1637 U32 klen; 1638 U32 off; 1639 STRLEN slen; 1640 STRLEN needents; 1641 U8 *tmps = NULL; 1642 U32 bit; 1643 SV *retval; 1644 U8 tmputf8[2]; 1645 UV c = NATIVE_TO_ASCII(*ptr); 1646 1647 if (!do_utf8 && !UNI_IS_INVARIANT(c)) { 1648 tmputf8[0] = (U8)UTF8_EIGHT_BIT_HI(c); 1649 tmputf8[1] = (U8)UTF8_EIGHT_BIT_LO(c); 1650 ptr = tmputf8; 1651 } 1652 /* Given a UTF-X encoded char 0xAA..0xYY,0xZZ 1653 * then the "swatch" is a vec() for al the chars which start 1654 * with 0xAA..0xYY 1655 * So the key in the hash (klen) is length of encoded char -1 1656 */ 1657 klen = UTF8SKIP(ptr) - 1; 1658 off = ptr[klen]; 1659 1660 if (klen == 0) 1661 { 1662 /* If char in invariant then swatch is for all the invariant chars 1663 * In both UTF-8 and UTF-8-MOD that happens to be UTF_CONTINUATION_MARK 1664 */ 1665 needents = UTF_CONTINUATION_MARK; 1666 off = NATIVE_TO_UTF(ptr[klen]); 1667 } 1668 else 1669 { 1670 /* If char is encoded then swatch is for the prefix */ 1671 needents = (1 << UTF_ACCUMULATION_SHIFT); 1672 off = NATIVE_TO_UTF(ptr[klen]) & UTF_CONTINUATION_MASK; 1673 } 1674 1675 /* 1676 * This single-entry cache saves about 1/3 of the utf8 overhead in test 1677 * suite. (That is, only 7-8% overall over just a hash cache. Still, 1678 * it's nothing to sniff at.) Pity we usually come through at least 1679 * two function calls to get here... 1680 * 1681 * NB: this code assumes that swatches are never modified, once generated! 1682 */ 1683 1684 if (hv == PL_last_swash_hv && 1685 klen == PL_last_swash_klen && 1686 (!klen || memEQ((char *)ptr, (char *)PL_last_swash_key, klen)) ) 1687 { 1688 tmps = PL_last_swash_tmps; 1689 slen = PL_last_swash_slen; 1690 } 1691 else { 1692 /* Try our second-level swatch cache, kept in a hash. */ 1693 SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE); 1694 1695 /* If not cached, generate it via utf8::SWASHGET */ 1696 if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) { 1697 dSP; 1698 /* We use utf8n_to_uvuni() as we want an index into 1699 Unicode tables, not a native character number. 1700 */ 1701 UV code_point = utf8n_to_uvuni(ptr, UTF8_MAXLEN, 0, 1702 ckWARN(WARN_UTF8) ? 1703 0 : UTF8_ALLOW_ANY); 1704 SV *errsv_save; 1705 ENTER; 1706 SAVETMPS; 1707 save_re_context(); 1708 PUSHSTACKi(PERLSI_MAGIC); 1709 PUSHMARK(SP); 1710 EXTEND(SP,3); 1711 PUSHs((SV*)sv); 1712 /* On EBCDIC & ~(0xA0-1) isn't a useful thing to do */ 1713 PUSHs(sv_2mortal(newSViv((klen) ? 1714 (code_point & ~(needents - 1)) : 0))); 1715 PUSHs(sv_2mortal(newSViv(needents))); 1716 PUTBACK; 1717 errsv_save = newSVsv(ERRSV); 1718 if (call_method("SWASHGET", G_SCALAR)) 1719 retval = newSVsv(*PL_stack_sp--); 1720 else 1721 retval = &PL_sv_undef; 1722 if (!SvTRUE(ERRSV)) 1723 sv_setsv(ERRSV, errsv_save); 1724 SvREFCNT_dec(errsv_save); 1725 POPSTACK; 1726 FREETMPS; 1727 LEAVE; 1728 if (IN_PERL_COMPILETIME) 1729 PL_curcop->op_private = (U8)(PL_hints & HINT_PRIVATE_MASK); 1730 1731 svp = hv_store(hv, (char*)ptr, klen, retval, 0); 1732 1733 if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || (slen << 3) < needents) 1734 Perl_croak(aTHX_ "SWASHGET didn't return result of proper length"); 1735 } 1736 1737 PL_last_swash_hv = hv; 1738 PL_last_swash_klen = klen; 1739 PL_last_swash_tmps = tmps; 1740 PL_last_swash_slen = slen; 1741 if (klen) 1742 Copy(ptr, PL_last_swash_key, klen, U8); 1743 } 1744 1745 switch ((int)((slen << 3) / needents)) { 1746 case 1: 1747 bit = 1 << (off & 7); 1748 off >>= 3; 1749 return (tmps[off] & bit) != 0; 1750 case 8: 1751 return tmps[off]; 1752 case 16: 1753 off <<= 1; 1754 return (tmps[off] << 8) + tmps[off + 1] ; 1755 case 32: 1756 off <<= 2; 1757 return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ; 1758 } 1759 Perl_croak(aTHX_ "panic: swash_fetch"); 1760 return 0; 1761 } 1762 1763 1764 /* 1765 =for apidoc A|U8 *|uvchr_to_utf8|U8 *d|UV uv 1766 1767 Adds the UTF-8 representation of the Native codepoint C<uv> to the end 1768 of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free 1769 bytes available. The return value is the pointer to the byte after the 1770 end of the new character. In other words, 1771 1772 d = uvchr_to_utf8(d, uv); 1773 1774 is the recommended wide native character-aware way of saying 1775 1776 *(d++) = uv; 1777 1778 =cut 1779 */ 1780 1781 /* On ASCII machines this is normally a macro but we want a 1782 real function in case XS code wants it 1783 */ 1784 #undef Perl_uvchr_to_utf8 1785 U8 * 1786 Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv) 1787 { 1788 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), 0); 1789 } 1790 1791 U8 * 1792 Perl_uvchr_to_utf8_flags(pTHX_ U8 *d, UV uv, UV flags) 1793 { 1794 return Perl_uvuni_to_utf8_flags(aTHX_ d, NATIVE_TO_UNI(uv), flags); 1795 } 1796 1797 /* 1798 =for apidoc A|UV|utf8n_to_uvchr|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags 1799 1800 Returns the native character value of the first character in the string C<s> 1801 which is assumed to be in UTF-8 encoding; C<retlen> will be set to the 1802 length, in bytes, of that character. 1803 1804 Allows length and flags to be passed to low level routine. 1805 1806 =cut 1807 */ 1808 /* On ASCII machines this is normally a macro but we want 1809 a real function in case XS code wants it 1810 */ 1811 #undef Perl_utf8n_to_uvchr 1812 UV 1813 Perl_utf8n_to_uvchr(pTHX_ U8 *s, STRLEN curlen, STRLEN *retlen, U32 flags) 1814 { 1815 UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags); 1816 return UNI_TO_NATIVE(uv); 1817 } 1818 1819 /* 1820 =for apidoc A|char *|pv_uni_display|SV *dsv|U8 *spv|STRLEN len|STRLEN pvlim|UV flags 1821 1822 Build to the scalar dsv a displayable version of the string spv, 1823 length len, the displayable version being at most pvlim bytes long 1824 (if longer, the rest is truncated and "..." will be appended). 1825 1826 The flags argument can have UNI_DISPLAY_ISPRINT set to display 1827 isPRINT()able characters as themselves, UNI_DISPLAY_BACKSLASH 1828 to display the \\[nrfta\\] as the backslashed versions (like '\n') 1829 (UNI_DISPLAY_BACKSLASH is preferred over UNI_DISPLAY_ISPRINT for \\). 1830 UNI_DISPLAY_QQ (and its alias UNI_DISPLAY_REGEX) have both 1831 UNI_DISPLAY_BACKSLASH and UNI_DISPLAY_ISPRINT turned on. 1832 1833 The pointer to the PV of the dsv is returned. 1834 1835 =cut */ 1836 char * 1837 Perl_pv_uni_display(pTHX_ SV *dsv, U8 *spv, STRLEN len, STRLEN pvlim, UV flags) 1838 { 1839 int truncated = 0; 1840 char *s, *e; 1841 1842 sv_setpvn(dsv, "", 0); 1843 for (s = (char *)spv, e = s + len; s < e; s += UTF8SKIP(s)) { 1844 UV u; 1845 bool ok = FALSE; 1846 1847 if (pvlim && SvCUR(dsv) >= pvlim) { 1848 truncated++; 1849 break; 1850 } 1851 u = utf8_to_uvchr((U8*)s, 0); 1852 if (u < 256) { 1853 if (!ok && (flags & UNI_DISPLAY_BACKSLASH)) { 1854 switch (u & 0xFF) { 1855 case '\n': 1856 Perl_sv_catpvf(aTHX_ dsv, "\\n"); ok = TRUE; break; 1857 case '\r': 1858 Perl_sv_catpvf(aTHX_ dsv, "\\r"); ok = TRUE; break; 1859 case '\t': 1860 Perl_sv_catpvf(aTHX_ dsv, "\\t"); ok = TRUE; break; 1861 case '\f': 1862 Perl_sv_catpvf(aTHX_ dsv, "\\f"); ok = TRUE; break; 1863 case '\a': 1864 Perl_sv_catpvf(aTHX_ dsv, "\\a"); ok = TRUE; break; 1865 case '\\': 1866 Perl_sv_catpvf(aTHX_ dsv, "\\\\" ); ok = TRUE; break; 1867 default: break; 1868 } 1869 } 1870 /* isPRINT() is the locale-blind version. */ 1871 if (!ok && (flags & UNI_DISPLAY_ISPRINT) && isPRINT(u & 0xFF)) { 1872 Perl_sv_catpvf(aTHX_ dsv, "%c", (char)(u & 0xFF)); 1873 ok = TRUE; 1874 } 1875 } 1876 if (!ok) 1877 Perl_sv_catpvf(aTHX_ dsv, "\\x{%"UVxf"}", u); 1878 } 1879 if (truncated) 1880 sv_catpvn(dsv, "...", 3); 1881 1882 return SvPVX(dsv); 1883 } 1884 1885 /* 1886 =for apidoc A|char *|sv_uni_display|SV *dsv|SV *ssv|STRLEN pvlim|UV flags 1887 1888 Build to the scalar dsv a displayable version of the scalar sv, 1889 the displayable version being at most pvlim bytes long 1890 (if longer, the rest is truncated and "..." will be appended). 1891 1892 The flags argument is as in pv_uni_display(). 1893 1894 The pointer to the PV of the dsv is returned. 1895 1896 =cut */ 1897 char * 1898 Perl_sv_uni_display(pTHX_ SV *dsv, SV *ssv, STRLEN pvlim, UV flags) 1899 { 1900 return Perl_pv_uni_display(aTHX_ dsv, (U8*)SvPVX(ssv), SvCUR(ssv), 1901 pvlim, flags); 1902 } 1903 1904 /* 1905 =for apidoc A|I32|ibcmp_utf8|const char *s1|char **pe1|register UV l1|bool u1|const char *s2|char **pe2|register UV l2|bool u2 1906 1907 Return true if the strings s1 and s2 differ case-insensitively, false 1908 if not (if they are equal case-insensitively). If u1 is true, the 1909 string s1 is assumed to be in UTF-8-encoded Unicode. If u2 is true, 1910 the string s2 is assumed to be in UTF-8-encoded Unicode. If u1 or u2 1911 are false, the respective string is assumed to be in native 8-bit 1912 encoding. 1913 1914 If the pe1 and pe2 are non-NULL, the scanning pointers will be copied 1915 in there (they will point at the beginning of the I<next> character). 1916 If the pointers behind pe1 or pe2 are non-NULL, they are the end 1917 pointers beyond which scanning will not continue under any 1918 circustances. If the byte lengths l1 and l2 are non-zero, s1+l1 and 1919 s2+l2 will be used as goal end pointers that will also stop the scan, 1920 and which qualify towards defining a successful match: all the scans 1921 that define an explicit length must reach their goal pointers for 1922 a match to succeed). 1923 1924 For case-insensitiveness, the "casefolding" of Unicode is used 1925 instead of upper/lowercasing both the characters, see 1926 http://www.unicode.org/unicode/reports/tr21/ (Case Mappings). 1927 1928 =cut */ 1929 I32 1930 Perl_ibcmp_utf8(pTHX_ const char *s1, char **pe1, register UV l1, bool u1, const char *s2, char **pe2, register UV l2, bool u2) 1931 { 1932 register U8 *p1 = (U8*)s1; 1933 register U8 *p2 = (U8*)s2; 1934 register U8 *e1 = 0, *f1 = 0, *q1 = 0; 1935 register U8 *e2 = 0, *f2 = 0, *q2 = 0; 1936 STRLEN n1 = 0, n2 = 0; 1937 U8 foldbuf1[UTF8_MAXLEN_FOLD+1]; 1938 U8 foldbuf2[UTF8_MAXLEN_FOLD+1]; 1939 U8 natbuf[1+1]; 1940 STRLEN foldlen1, foldlen2; 1941 bool match; 1942 1943 if (pe1) 1944 e1 = *(U8**)pe1; 1945 if (e1 == 0 || (l1 && l1 < (UV)(e1 - (U8*)s1))) 1946 f1 = (U8*)s1 + l1; 1947 if (pe2) 1948 e2 = *(U8**)pe2; 1949 if (e2 == 0 || (l2 && l2 < (UV)(e2 - (U8*)s2))) 1950 f2 = (U8*)s2 + l2; 1951 1952 if ((e1 == 0 && f1 == 0) || (e2 == 0 && f2 == 0) || (f1 == 0 && f2 == 0)) 1953 return 1; /* mismatch; possible infinite loop or false positive */ 1954 1955 if (!u1 || !u2) 1956 natbuf[1] = 0; /* Need to terminate the buffer. */ 1957 1958 while ((e1 == 0 || p1 < e1) && 1959 (f1 == 0 || p1 < f1) && 1960 (e2 == 0 || p2 < e2) && 1961 (f2 == 0 || p2 < f2)) { 1962 if (n1 == 0) { 1963 if (u1) 1964 to_utf8_fold(p1, foldbuf1, &foldlen1); 1965 else { 1966 natbuf[0] = *p1; 1967 to_utf8_fold(natbuf, foldbuf1, &foldlen1); 1968 } 1969 q1 = foldbuf1; 1970 n1 = foldlen1; 1971 } 1972 if (n2 == 0) { 1973 if (u2) 1974 to_utf8_fold(p2, foldbuf2, &foldlen2); 1975 else { 1976 natbuf[0] = *p2; 1977 to_utf8_fold(natbuf, foldbuf2, &foldlen2); 1978 } 1979 q2 = foldbuf2; 1980 n2 = foldlen2; 1981 } 1982 while (n1 && n2) { 1983 if ( UTF8SKIP(q1) != UTF8SKIP(q2) || 1984 (UTF8SKIP(q1) == 1 && *q1 != *q2) || 1985 memNE((char*)q1, (char*)q2, UTF8SKIP(q1)) ) 1986 return 1; /* mismatch */ 1987 n1 -= UTF8SKIP(q1); 1988 q1 += UTF8SKIP(q1); 1989 n2 -= UTF8SKIP(q2); 1990 q2 += UTF8SKIP(q2); 1991 } 1992 if (n1 == 0) 1993 p1 += u1 ? UTF8SKIP(p1) : 1; 1994 if (n2 == 0) 1995 p2 += u2 ? UTF8SKIP(p2) : 1; 1996 1997 } 1998 1999 /* A match is defined by all the scans that specified 2000 * an explicit length reaching their final goals. */ 2001 match = (f1 == 0 || p1 == f1) && (f2 == 0 || p2 == f2); 2002 2003 if (match) { 2004 if (pe1) 2005 *pe1 = (char*)p1; 2006 if (pe2) 2007 *pe2 = (char*)p2; 2008 } 2009 2010 return match ? 0 : 1; /* 0 match, 1 mismatch */ 2011 } 2012 2013