1 /* $FreeBSD$ */ 2 /* $NetBSD: citrus_iso2022.c,v 1.19 2008/06/14 16:01:07 tnozaki Exp $ */ 3 4 /*- 5 * Copyright (c)1999, 2002 Citrus Project, 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * $Citrus: xpg4dl/FreeBSD/lib/libc/locale/iso2022.c,v 1.23 2001/06/21 01:51:44 yamt Exp $ 30 */ 31 32 #include <sys/cdefs.h> 33 #include <sys/types.h> 34 35 #include <assert.h> 36 #include <errno.h> 37 #include <limits.h> 38 #include <stdbool.h> 39 #include <stddef.h> 40 #include <stdio.h> 41 #include <stdlib.h> 42 #include <string.h> 43 #include <wchar.h> 44 45 #include "citrus_namespace.h" 46 #include "citrus_types.h" 47 #include "citrus_module.h" 48 #include "citrus_stdenc.h" 49 #include "citrus_iso2022.h" 50 51 52 /* ---------------------------------------------------------------------- 53 * private stuffs used by templates 54 */ 55 56 57 /* 58 * wchar_t mappings: 59 * ASCII (ESC ( B) 00000000 00000000 00000000 0xxxxxxx 60 * iso-8859-1 (ESC , A) 00000000 00000000 00000000 1xxxxxxx 61 * 94 charset (ESC ( F) 0fffffff 00000000 00000000 0xxxxxxx 62 * 94 charset (ESC ( M F) 0fffffff 1mmmmmmm 00000000 0xxxxxxx 63 * 96 charset (ESC , F) 0fffffff 00000000 00000000 1xxxxxxx 64 * 96 charset (ESC , M F) 0fffffff 1mmmmmmm 00000000 1xxxxxxx 65 * 94x94 charset (ESC $ ( F) 0fffffff 00000000 0xxxxxxx 0xxxxxxx 66 * 96x96 charset (ESC $ , F) 0fffffff 00000000 0xxxxxxx 1xxxxxxx 67 * 94x94 charset (ESC & V ESC $ ( F) 68 * 0fffffff 1vvvvvvv 0xxxxxxx 0xxxxxxx 69 * 94x94x94 charset (ESC $ ( F) 0fffffff 0xxxxxxx 0xxxxxxx 0xxxxxxx 70 * 96x96x96 charset (ESC $ , F) 0fffffff 0xxxxxxx 0xxxxxxx 1xxxxxxx 71 * reserved for UCS4 co-existence (UCS4 is 31bit encoding thanks to mohta bit) 72 * 1xxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx 73 */ 74 75 #define CS94 (0U) 76 #define CS96 (1U) 77 #define CS94MULTI (2U) 78 #define CS96MULTI (3U) 79 80 typedef struct { 81 unsigned char interm; 82 unsigned char final; 83 unsigned char type; 84 unsigned char vers; 85 } _ISO2022Charset; 86 87 static const _ISO2022Charset ascii = { CS94, 'B', '\0', '\0' }; 88 static const _ISO2022Charset iso88591 = { CS96, 'A', '\0', '\0' }; 89 90 typedef struct { 91 _ISO2022Charset g[4]; 92 /* need 3 bits to hold -1, 0, ..., 3 */ 93 int gl:3, 94 gr:3, 95 singlegl:3, 96 singlegr:3; 97 char ch[7]; /* longest escape sequence (ESC & V ESC $ ( F) */ 98 size_t chlen; 99 int flags; 100 #define _ISO2022STATE_FLAG_INITIALIZED 1 101 } _ISO2022State; 102 103 typedef struct { 104 _ISO2022Charset *recommend[4]; 105 size_t recommendsize[4]; 106 _ISO2022Charset initg[4]; 107 int maxcharset; 108 int flags; 109 #define F_8BIT 0x0001 110 #define F_NOOLD 0x0002 111 #define F_SI 0x0010 /*0F*/ 112 #define F_SO 0x0020 /*0E*/ 113 #define F_LS0 0x0010 /*0F*/ 114 #define F_LS1 0x0020 /*0E*/ 115 #define F_LS2 0x0040 /*ESC n*/ 116 #define F_LS3 0x0080 /*ESC o*/ 117 #define F_LS1R 0x0100 /*ESC ~*/ 118 #define F_LS2R 0x0200 /*ESC }*/ 119 #define F_LS3R 0x0400 /*ESC |*/ 120 #define F_SS2 0x0800 /*ESC N*/ 121 #define F_SS3 0x1000 /*ESC O*/ 122 #define F_SS2R 0x2000 /*8E*/ 123 #define F_SS3R 0x4000 /*8F*/ 124 } _ISO2022EncodingInfo; 125 126 #define _CEI_TO_EI(_cei_) (&(_cei_)->ei) 127 #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_ 128 129 #define _FUNCNAME(m) _citrus_ISO2022_##m 130 #define _ENCODING_INFO _ISO2022EncodingInfo 131 #define _ENCODING_STATE _ISO2022State 132 #define _ENCODING_MB_CUR_MAX(_ei_) MB_LEN_MAX 133 #define _ENCODING_IS_STATE_DEPENDENT 1 134 #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) \ 135 (!((_ps_)->flags & _ISO2022STATE_FLAG_INITIALIZED)) 136 137 138 #define _ISO2022INVALID (wchar_t)-1 139 140 static __inline bool isc0(__uint8_t x) 141 { 142 143 return ((x & 0x1f) == x); 144 } 145 146 static __inline bool isc1(__uint8_t x) 147 { 148 149 return (0x80 <= x && x <= 0x9f); 150 } 151 152 static __inline bool iscntl(__uint8_t x) 153 { 154 155 return (isc0(x) || isc1(x) || x == 0x7f); 156 } 157 158 static __inline bool is94(__uint8_t x) 159 { 160 161 return (0x21 <= x && x <= 0x7e); 162 } 163 164 static __inline bool is96(__uint8_t x) 165 { 166 167 return (0x20 <= x && x <= 0x7f); 168 } 169 170 static __inline bool isecma(__uint8_t x) 171 { 172 173 return (0x30 <= x && x <= 0x7f); 174 } 175 176 static __inline bool isinterm(__uint8_t x) 177 { 178 179 return (0x20 <= x && x <= 0x2f); 180 } 181 182 static __inline bool isthree(__uint8_t x) 183 { 184 185 return (0x60 <= x && x <= 0x6f); 186 } 187 188 static __inline int 189 getcs(const char * __restrict p, _ISO2022Charset * __restrict cs) 190 { 191 192 if (!strncmp(p, "94$", 3) && p[3] && !p[4]) { 193 cs->final = (unsigned char)(p[3] & 0xff); 194 cs->interm = '\0'; 195 cs->vers = '\0'; 196 cs->type = CS94MULTI; 197 } else if (!strncmp(p, "96$", 3) && p[3] && !p[4]) { 198 cs->final = (unsigned char)(p[3] & 0xff); 199 cs->interm = '\0'; 200 cs->vers = '\0'; 201 cs->type = CS96MULTI; 202 } else if (!strncmp(p, "94", 2) && p[2] && !p[3]) { 203 cs->final = (unsigned char)(p[2] & 0xff); 204 cs->interm = '\0'; 205 cs->vers = '\0'; 206 cs->type = CS94; 207 } else if (!strncmp(p, "96", 2) && p[2] && !p[3]) { 208 cs->final = (unsigned char )(p[2] & 0xff); 209 cs->interm = '\0'; 210 cs->vers = '\0'; 211 cs->type = CS96; 212 } else 213 return (1); 214 215 return (0); 216 } 217 218 219 #define _NOTMATCH 0 220 #define _MATCH 1 221 #define _PARSEFAIL 2 222 223 static __inline int 224 get_recommend(_ISO2022EncodingInfo * __restrict ei, 225 const char * __restrict token) 226 { 227 _ISO2022Charset cs, *p; 228 int i; 229 230 if (!strchr("0123", token[0]) || token[1] != '=') 231 return (_NOTMATCH); 232 233 if (getcs(&token[2], &cs) == 0) 234 ; 235 else if (!strcmp(&token[2], "94")) { 236 cs.final = (unsigned char)(token[4]); 237 cs.interm = '\0'; 238 cs.vers = '\0'; 239 cs.type = CS94; 240 } else if (!strcmp(&token[2], "96")) { 241 cs.final = (unsigned char)(token[4]); 242 cs.interm = '\0'; 243 cs.vers = '\0'; 244 cs.type = CS96; 245 } else if (!strcmp(&token[2], "94$")) { 246 cs.final = (unsigned char)(token[5]); 247 cs.interm = '\0'; 248 cs.vers = '\0'; 249 cs.type = CS94MULTI; 250 } else if (!strcmp(&token[2], "96$")) { 251 cs.final = (unsigned char)(token[5]); 252 cs.interm = '\0'; 253 cs.vers = '\0'; 254 cs.type = CS96MULTI; 255 } else 256 return (_PARSEFAIL); 257 258 i = token[0] - '0'; 259 if (!ei->recommend[i]) 260 ei->recommend[i] = malloc(sizeof(_ISO2022Charset)); 261 else { 262 p = realloc(ei->recommend[i], 263 sizeof(_ISO2022Charset) * (ei->recommendsize[i] + 1)); 264 if (!p) 265 return (_PARSEFAIL); 266 ei->recommend[i] = p; 267 } 268 if (!ei->recommend[i]) 269 return (_PARSEFAIL); 270 ei->recommendsize[i]++; 271 272 (ei->recommend[i] + (ei->recommendsize[i] - 1))->final = cs.final; 273 (ei->recommend[i] + (ei->recommendsize[i] - 1))->interm = cs.interm; 274 (ei->recommend[i] + (ei->recommendsize[i] - 1))->vers = cs.vers; 275 (ei->recommend[i] + (ei->recommendsize[i] - 1))->type = cs.type; 276 277 return (_MATCH); 278 } 279 280 static __inline int 281 get_initg(_ISO2022EncodingInfo * __restrict ei, 282 const char * __restrict token) 283 { 284 _ISO2022Charset cs; 285 286 if (strncmp("INIT", &token[0], 4) || 287 !strchr("0123", token[4]) || 288 token[5] != '=') 289 return (_NOTMATCH); 290 291 if (getcs(&token[6], &cs) != 0) 292 return (_PARSEFAIL); 293 294 ei->initg[token[4] - '0'].type = cs.type; 295 ei->initg[token[4] - '0'].final = cs.final; 296 ei->initg[token[4] - '0'].interm = cs.interm; 297 ei->initg[token[4] - '0'].vers = cs.vers; 298 299 return (_MATCH); 300 } 301 302 static __inline int 303 get_max(_ISO2022EncodingInfo * __restrict ei, 304 const char * __restrict token) 305 { 306 if (!strcmp(token, "MAX1")) 307 ei->maxcharset = 1; 308 else if (!strcmp(token, "MAX2")) 309 ei->maxcharset = 2; 310 else if (!strcmp(token, "MAX3")) 311 ei->maxcharset = 3; 312 else 313 return (_NOTMATCH); 314 315 return (_MATCH); 316 } 317 318 319 static __inline int 320 get_flags(_ISO2022EncodingInfo * __restrict ei, 321 const char * __restrict token) 322 { 323 static struct { 324 const char *tag; 325 int flag; 326 } const tags[] = { 327 { "DUMMY", 0 }, 328 { "8BIT", F_8BIT }, 329 { "NOOLD", F_NOOLD }, 330 { "SI", F_SI }, 331 { "SO", F_SO }, 332 { "LS0", F_LS0 }, 333 { "LS1", F_LS1 }, 334 { "LS2", F_LS2 }, 335 { "LS3", F_LS3 }, 336 { "LS1R", F_LS1R }, 337 { "LS2R", F_LS2R }, 338 { "LS3R", F_LS3R }, 339 { "SS2", F_SS2 }, 340 { "SS3", F_SS3 }, 341 { "SS2R", F_SS2R }, 342 { "SS3R", F_SS3R }, 343 { NULL, 0 } 344 }; 345 int i; 346 347 for (i = 0; tags[i].tag; i++) 348 if (!strcmp(token, tags[i].tag)) { 349 ei->flags |= tags[i].flag; 350 return (_MATCH); 351 } 352 353 return (_NOTMATCH); 354 } 355 356 357 static __inline int 358 _citrus_ISO2022_parse_variable(_ISO2022EncodingInfo * __restrict ei, 359 const void * __restrict var, size_t lenvar __unused) 360 { 361 char const *e, *v; 362 char buf[20]; 363 size_t len; 364 int i, ret; 365 366 /* 367 * parse VARIABLE section. 368 */ 369 370 if (!var) 371 return (EFTYPE); 372 373 v = (const char *) var; 374 375 /* initialize structure */ 376 ei->maxcharset = 0; 377 for (i = 0; i < 4; i++) { 378 ei->recommend[i] = NULL; 379 ei->recommendsize[i] = 0; 380 } 381 ei->flags = 0; 382 383 while (*v) { 384 while (*v == ' ' || *v == '\t') 385 ++v; 386 387 /* find the token */ 388 e = v; 389 while (*e && *e != ' ' && *e != '\t') 390 ++e; 391 392 len = e - v; 393 if (len == 0) 394 break; 395 if (len >= sizeof(buf)) 396 goto parsefail; 397 snprintf(buf, sizeof(buf), "%.*s", (int)len, v); 398 399 if ((ret = get_recommend(ei, buf)) != _NOTMATCH) 400 ; 401 else if ((ret = get_initg(ei, buf)) != _NOTMATCH) 402 ; 403 else if ((ret = get_max(ei, buf)) != _NOTMATCH) 404 ; 405 else if ((ret = get_flags(ei, buf)) != _NOTMATCH) 406 ; 407 else 408 ret = _PARSEFAIL; 409 if (ret == _PARSEFAIL) 410 goto parsefail; 411 v = e; 412 413 } 414 415 return (0); 416 417 parsefail: 418 free(ei->recommend[0]); 419 free(ei->recommend[1]); 420 free(ei->recommend[2]); 421 free(ei->recommend[3]); 422 423 return (EFTYPE); 424 } 425 426 static __inline void 427 /*ARGSUSED*/ 428 _citrus_ISO2022_init_state(_ISO2022EncodingInfo * __restrict ei, 429 _ISO2022State * __restrict s) 430 { 431 int i; 432 433 memset(s, 0, sizeof(*s)); 434 s->gl = 0; 435 s->gr = (ei->flags & F_8BIT) ? 1 : -1; 436 437 for (i = 0; i < 4; i++) 438 if (ei->initg[i].final) { 439 s->g[i].type = ei->initg[i].type; 440 s->g[i].final = ei->initg[i].final; 441 s->g[i].interm = ei->initg[i].interm; 442 } 443 s->singlegl = s->singlegr = -1; 444 s->flags |= _ISO2022STATE_FLAG_INITIALIZED; 445 } 446 447 static __inline void 448 /*ARGSUSED*/ 449 _citrus_ISO2022_pack_state(_ISO2022EncodingInfo * __restrict ei __unused, 450 void * __restrict pspriv, const _ISO2022State * __restrict s) 451 { 452 453 memcpy(pspriv, (const void *)s, sizeof(*s)); 454 } 455 456 static __inline void 457 /*ARGSUSED*/ 458 _citrus_ISO2022_unpack_state(_ISO2022EncodingInfo * __restrict ei __unused, 459 _ISO2022State * __restrict s, const void * __restrict pspriv) 460 { 461 462 memcpy((void *)s, pspriv, sizeof(*s)); 463 } 464 465 static int 466 /*ARGSUSED*/ 467 _citrus_ISO2022_encoding_module_init(_ISO2022EncodingInfo * __restrict ei, 468 const void * __restrict var, size_t lenvar) 469 { 470 471 return (_citrus_ISO2022_parse_variable(ei, var, lenvar)); 472 } 473 474 static void 475 /*ARGSUSED*/ 476 _citrus_ISO2022_encoding_module_uninit(_ISO2022EncodingInfo *ei __unused) 477 { 478 479 } 480 481 #define ESC '\033' 482 #define ECMA -1 483 #define INTERM -2 484 #define OECMA -3 485 static const struct seqtable { 486 int type; 487 int csoff; 488 int finaloff; 489 int intermoff; 490 int versoff; 491 int len; 492 int chars[10]; 493 } seqtable[] = { 494 /* G0 94MULTI special */ 495 { CS94MULTI, -1, 2, -1, -1, 3, { ESC, '$', OECMA }, }, 496 /* G0 94MULTI special with version identification */ 497 { CS94MULTI, -1, 5, -1, 2, 6, { ESC, '&', ECMA, ESC, '$', OECMA }, }, 498 /* G? 94 */ 499 { CS94, 1, 2, -1, -1, 3, { ESC, CS94, ECMA, }, }, 500 /* G? 94 with 2nd intermediate char */ 501 { CS94, 1, 3, 2, -1, 4, { ESC, CS94, INTERM, ECMA, }, }, 502 /* G? 96 */ 503 { CS96, 1, 2, -1, -1, 3, { ESC, CS96, ECMA, }, }, 504 /* G? 96 with 2nd intermediate char */ 505 { CS96, 1, 3, 2, -1, 4, { ESC, CS96, INTERM, ECMA, }, }, 506 /* G? 94MULTI */ 507 { CS94MULTI, 2, 3, -1, -1, 4, { ESC, '$', CS94, ECMA, }, }, 508 /* G? 96MULTI */ 509 { CS96MULTI, 2, 3, -1, -1, 4, { ESC, '$', CS96, ECMA, }, }, 510 /* G? 94MULTI with version specification */ 511 { CS94MULTI, 5, 6, -1, 2, 7, { ESC, '&', ECMA, ESC, '$', CS94, ECMA, }, }, 512 /* LS2/3 */ 513 { -1, -1, -1, -1, -1, 2, { ESC, 'n', }, }, 514 { -1, -1, -1, -1, -1, 2, { ESC, 'o', }, }, 515 /* LS1/2/3R */ 516 { -1, -1, -1, -1, -1, 2, { ESC, '~', }, }, 517 { -1, -1, -1, -1, -1, 2, { ESC, /*{*/ '}', }, }, 518 { -1, -1, -1, -1, -1, 2, { ESC, '|', }, }, 519 /* SS2/3 */ 520 { -1, -1, -1, -1, -1, 2, { ESC, 'N', }, }, 521 { -1, -1, -1, -1, -1, 2, { ESC, 'O', }, }, 522 /* end of records */ 523 // { 0, } 524 { 0, 0, 0, 0, 0, 0, { ESC, 0, }, } 525 }; 526 527 static int 528 seqmatch(const char * __restrict s, size_t n, 529 const struct seqtable * __restrict sp) 530 { 531 const int *p; 532 533 p = sp->chars; 534 while ((size_t)(p - sp->chars) < n && p - sp->chars < sp->len) { 535 switch (*p) { 536 case ECMA: 537 if (!isecma(*s)) 538 goto terminate; 539 break; 540 case OECMA: 541 if (*s && strchr("@AB", *s)) 542 break; 543 else 544 goto terminate; 545 case INTERM: 546 if (!isinterm(*s)) 547 goto terminate; 548 break; 549 case CS94: 550 if (*s && strchr("()*+", *s)) 551 break; 552 else 553 goto terminate; 554 case CS96: 555 if (*s && strchr(",-./", *s)) 556 break; 557 else 558 goto terminate; 559 default: 560 if (*s != *p) 561 goto terminate; 562 break; 563 } 564 565 p++; 566 s++; 567 } 568 569 terminate: 570 return (p - sp->chars); 571 } 572 573 static wchar_t 574 _ISO2022_sgetwchar(_ISO2022EncodingInfo * __restrict ei __unused, 575 const char * __restrict string, size_t n, const char ** __restrict result, 576 _ISO2022State * __restrict psenc) 577 { 578 const struct seqtable *sp; 579 wchar_t wchar = 0; 580 int i, cur, nmatch; 581 582 while (1) { 583 /* SI/SO */ 584 if (1 <= n && string[0] == '\017') { 585 psenc->gl = 0; 586 string++; 587 n--; 588 continue; 589 } 590 if (1 <= n && string[0] == '\016') { 591 psenc->gl = 1; 592 string++; 593 n--; 594 continue; 595 } 596 597 /* SS2/3R */ 598 if (1 <= n && string[0] && strchr("\217\216", string[0])) { 599 psenc->singlegl = psenc->singlegr = 600 (string[0] - '\216') + 2; 601 string++; 602 n--; 603 continue; 604 } 605 606 /* eat the letter if this is not ESC */ 607 if (1 <= n && string[0] != '\033') 608 break; 609 610 /* look for a perfect match from escape sequences */ 611 for (sp = &seqtable[0]; sp->len; sp++) { 612 nmatch = seqmatch(string, n, sp); 613 if (sp->len == nmatch && n >= (size_t)(sp->len)) 614 break; 615 } 616 617 if (!sp->len) 618 goto notseq; 619 620 if (sp->type != -1) { 621 if (sp->csoff == -1) 622 i = 0; 623 else { 624 switch (sp->type) { 625 case CS94: 626 case CS94MULTI: 627 i = string[sp->csoff] - '('; 628 break; 629 case CS96: 630 case CS96MULTI: 631 i = string[sp->csoff] - ','; 632 break; 633 default: 634 return (_ISO2022INVALID); 635 } 636 } 637 psenc->g[i].type = sp->type; 638 psenc->g[i].final = '\0'; 639 psenc->g[i].interm = '\0'; 640 psenc->g[i].vers = '\0'; 641 /* sp->finaloff must not be -1 */ 642 if (sp->finaloff != -1) 643 psenc->g[i].final = string[sp->finaloff]; 644 if (sp->intermoff != -1) 645 psenc->g[i].interm = string[sp->intermoff]; 646 if (sp->versoff != -1) 647 psenc->g[i].vers = string[sp->versoff]; 648 649 string += sp->len; 650 n -= sp->len; 651 continue; 652 } 653 654 /* LS2/3 */ 655 if (2 <= n && string[0] == '\033' && 656 string[1] && strchr("no", string[1])) { 657 psenc->gl = string[1] - 'n' + 2; 658 string += 2; 659 n -= 2; 660 continue; 661 } 662 663 /* LS1/2/3R */ 664 /* XXX: { for vi showmatch */ 665 if (2 <= n && string[0] == '\033' && 666 string[1] && strchr("~}|", string[1])) { 667 psenc->gr = 3 - (string[1] - '|'); 668 string += 2; 669 n -= 2; 670 continue; 671 } 672 673 /* SS2/3 */ 674 if (2 <= n && string[0] == '\033' && string[1] && 675 strchr("NO", string[1])) { 676 psenc->singlegl = (string[1] - 'N') + 2; 677 string += 2; 678 n -= 2; 679 continue; 680 } 681 682 notseq: 683 /* 684 * if we've got an unknown escape sequence, eat the ESC at the 685 * head. otherwise, wait till full escape sequence comes. 686 */ 687 for (sp = &seqtable[0]; sp->len; sp++) { 688 nmatch = seqmatch(string, n, sp); 689 if (!nmatch) 690 continue; 691 692 /* 693 * if we are in the middle of escape sequence, 694 * we still need to wait for more characters to come 695 */ 696 if (n < (size_t)(sp->len)) { 697 if ((size_t)(nmatch) == n) { 698 if (result) 699 *result = string; 700 return (_ISO2022INVALID); 701 } 702 } else { 703 if (nmatch == sp->len) { 704 /* this case should not happen */ 705 goto eat; 706 } 707 } 708 } 709 710 break; 711 } 712 713 eat: 714 /* no letter to eat */ 715 if (n < 1) { 716 if (result) 717 *result = string; 718 return (_ISO2022INVALID); 719 } 720 721 /* normal chars. always eat C0/C1 as is. */ 722 if (iscntl(*string & 0xff)) 723 cur = -1; 724 else if (*string & 0x80) 725 cur = (psenc->singlegr == -1) ? psenc->gr : psenc->singlegr; 726 else 727 cur = (psenc->singlegl == -1) ? psenc->gl : psenc->singlegl; 728 729 if (cur == -1) { 730 asis: 731 wchar = *string++ & 0xff; 732 if (result) 733 *result = string; 734 /* reset single shift state */ 735 psenc->singlegr = psenc->singlegl = -1; 736 return (wchar); 737 } 738 739 /* length error check */ 740 switch (psenc->g[cur].type) { 741 case CS94MULTI: 742 case CS96MULTI: 743 if (!isthree(psenc->g[cur].final)) { 744 if (2 <= n && 745 (string[0] & 0x80) == (string[1] & 0x80)) 746 break; 747 } else { 748 if (3 <= n && 749 (string[0] & 0x80) == (string[1] & 0x80) && 750 (string[0] & 0x80) == (string[2] & 0x80)) 751 break; 752 } 753 754 /* we still need to wait for more characters to come */ 755 if (result) 756 *result = string; 757 return (_ISO2022INVALID); 758 759 case CS94: 760 case CS96: 761 if (1 <= n) 762 break; 763 764 /* we still need to wait for more characters to come */ 765 if (result) 766 *result = string; 767 return (_ISO2022INVALID); 768 } 769 770 /* range check */ 771 switch (psenc->g[cur].type) { 772 case CS94: 773 if (!(is94(string[0] & 0x7f))) 774 goto asis; 775 case CS96: 776 if (!(is96(string[0] & 0x7f))) 777 goto asis; 778 break; 779 case CS94MULTI: 780 if (!(is94(string[0] & 0x7f) && is94(string[1] & 0x7f))) 781 goto asis; 782 break; 783 case CS96MULTI: 784 if (!(is96(string[0] & 0x7f) && is96(string[1] & 0x7f))) 785 goto asis; 786 break; 787 } 788 789 /* extract the character. */ 790 switch (psenc->g[cur].type) { 791 case CS94: 792 /* special case for ASCII. */ 793 if (psenc->g[cur].final == 'B' && !psenc->g[cur].interm) { 794 wchar = *string++; 795 wchar &= 0x7f; 796 break; 797 } 798 wchar = psenc->g[cur].final; 799 wchar = (wchar << 8); 800 wchar |= (psenc->g[cur].interm ? (0x80 | psenc->g[cur].interm) : 0); 801 wchar = (wchar << 8); 802 wchar = (wchar << 8) | (*string++ & 0x7f); 803 break; 804 case CS96: 805 /* special case for ISO-8859-1. */ 806 if (psenc->g[cur].final == 'A' && !psenc->g[cur].interm) { 807 wchar = *string++; 808 wchar &= 0x7f; 809 wchar |= 0x80; 810 break; 811 } 812 wchar = psenc->g[cur].final; 813 wchar = (wchar << 8); 814 wchar |= (psenc->g[cur].interm ? (0x80 | psenc->g[cur].interm) : 0); 815 wchar = (wchar << 8); 816 wchar = (wchar << 8) | (*string++ & 0x7f); 817 wchar |= 0x80; 818 break; 819 case CS94MULTI: 820 case CS96MULTI: 821 wchar = psenc->g[cur].final; 822 wchar = (wchar << 8); 823 if (isthree(psenc->g[cur].final)) 824 wchar |= (*string++ & 0x7f); 825 wchar = (wchar << 8) | (*string++ & 0x7f); 826 wchar = (wchar << 8) | (*string++ & 0x7f); 827 if (psenc->g[cur].type == CS96MULTI) 828 wchar |= 0x80; 829 break; 830 } 831 832 if (result) 833 *result = string; 834 /* reset single shift state */ 835 psenc->singlegr = psenc->singlegl = -1; 836 return (wchar); 837 } 838 839 840 841 static int 842 _citrus_ISO2022_mbrtowc_priv(_ISO2022EncodingInfo * __restrict ei, 843 wchar_t * __restrict pwc, const char ** __restrict s, 844 size_t n, _ISO2022State * __restrict psenc, size_t * __restrict nresult) 845 { 846 const char *p, *result, *s0; 847 wchar_t wchar; 848 int c, chlenbak; 849 850 if (*s == NULL) { 851 _citrus_ISO2022_init_state(ei, psenc); 852 *nresult = _ENCODING_IS_STATE_DEPENDENT; 853 return (0); 854 } 855 s0 = *s; 856 c = 0; 857 chlenbak = psenc->chlen; 858 859 /* 860 * if we have something in buffer, use that. 861 * otherwise, skip here 862 */ 863 if (psenc->chlen > sizeof(psenc->ch)) { 864 /* illgeal state */ 865 _citrus_ISO2022_init_state(ei, psenc); 866 goto encoding_error; 867 } 868 if (psenc->chlen == 0) 869 goto emptybuf; 870 871 /* buffer is not empty */ 872 p = psenc->ch; 873 while (psenc->chlen < sizeof(psenc->ch)) { 874 if (n > 0) { 875 psenc->ch[psenc->chlen++] = *s0++; 876 n--; 877 } 878 879 wchar = _ISO2022_sgetwchar(ei, p, psenc->chlen - (p-psenc->ch), 880 &result, psenc); 881 c += result - p; 882 if (wchar != _ISO2022INVALID) { 883 if (psenc->chlen > (size_t)c) 884 memmove(psenc->ch, result, psenc->chlen - c); 885 if (psenc->chlen < (size_t)c) 886 psenc->chlen = 0; 887 else 888 psenc->chlen -= c; 889 goto output; 890 } 891 892 if (n == 0) { 893 if ((size_t)(result - p) == psenc->chlen) 894 /* complete shift sequence. */ 895 psenc->chlen = 0; 896 goto restart; 897 } 898 899 p = result; 900 } 901 902 /* escape sequence too long? */ 903 goto encoding_error; 904 905 emptybuf: 906 wchar = _ISO2022_sgetwchar(ei, s0, n, &result, psenc); 907 if (wchar != _ISO2022INVALID) { 908 c += result - s0; 909 psenc->chlen = 0; 910 s0 = result; 911 goto output; 912 } 913 if (result > s0) { 914 c += (result - s0); 915 n -= (result - s0); 916 s0 = result; 917 if (n > 0) 918 goto emptybuf; 919 /* complete shift sequence. */ 920 goto restart; 921 } 922 n += c; 923 if (n < sizeof(psenc->ch)) { 924 memcpy(psenc->ch, s0 - c, n); 925 psenc->chlen = n; 926 s0 = result; 927 goto restart; 928 } 929 930 /* escape sequence too long? */ 931 932 encoding_error: 933 psenc->chlen = 0; 934 *nresult = (size_t)-1; 935 return (EILSEQ); 936 937 output: 938 *s = s0; 939 if (pwc) 940 *pwc = wchar; 941 *nresult = wchar ? c - chlenbak : 0; 942 return (0); 943 944 restart: 945 *s = s0; 946 *nresult = (size_t)-2; 947 948 return (0); 949 } 950 951 static int 952 recommendation(_ISO2022EncodingInfo * __restrict ei, 953 _ISO2022Charset * __restrict cs) 954 { 955 _ISO2022Charset *recommend; 956 size_t j; 957 int i; 958 959 /* first, try a exact match. */ 960 for (i = 0; i < 4; i++) { 961 recommend = ei->recommend[i]; 962 for (j = 0; j < ei->recommendsize[i]; j++) { 963 if (cs->type != recommend[j].type) 964 continue; 965 if (cs->final != recommend[j].final) 966 continue; 967 if (cs->interm != recommend[j].interm) 968 continue; 969 970 return (i); 971 } 972 } 973 974 /* then, try a wildcard match over final char. */ 975 for (i = 0; i < 4; i++) { 976 recommend = ei->recommend[i]; 977 for (j = 0; j < ei->recommendsize[i]; j++) { 978 if (cs->type != recommend[j].type) 979 continue; 980 if (cs->final && (cs->final != recommend[j].final)) 981 continue; 982 if (cs->interm && (cs->interm != recommend[j].interm)) 983 continue; 984 985 return (i); 986 } 987 } 988 989 /* there's no recommendation. make a guess. */ 990 if (ei->maxcharset == 0) { 991 return (0); 992 } else { 993 switch (cs->type) { 994 case CS94: 995 case CS94MULTI: 996 return (0); 997 case CS96: 998 case CS96MULTI: 999 return (1); 1000 } 1001 } 1002 return (0); 1003 } 1004 1005 static int 1006 _ISO2022_sputwchar(_ISO2022EncodingInfo * __restrict ei, wchar_t wc, 1007 char * __restrict string, size_t n, char ** __restrict result, 1008 _ISO2022State * __restrict psenc, size_t * __restrict nresult) 1009 { 1010 _ISO2022Charset cs; 1011 char *p; 1012 char tmp[MB_LEN_MAX]; 1013 size_t len; 1014 int bit8, i = 0, target; 1015 unsigned char mask; 1016 1017 if (isc0(wc & 0xff)) { 1018 /* go back to INIT0 or ASCII on control chars */ 1019 cs = ei->initg[0].final ? ei->initg[0] : ascii; 1020 } else if (isc1(wc & 0xff)) { 1021 /* go back to INIT1 or ISO-8859-1 on control chars */ 1022 cs = ei->initg[1].final ? ei->initg[1] : iso88591; 1023 } else if (!(wc & ~0xff)) { 1024 if (wc & 0x80) { 1025 /* special treatment for ISO-8859-1 */ 1026 cs = iso88591; 1027 } else { 1028 /* special treatment for ASCII */ 1029 cs = ascii; 1030 } 1031 } else { 1032 cs.final = (wc >> 24) & 0x7f; 1033 if ((wc >> 16) & 0x80) 1034 cs.interm = (wc >> 16) & 0x7f; 1035 else 1036 cs.interm = '\0'; 1037 if (wc & 0x80) 1038 cs.type = (wc & 0x00007f00) ? CS96MULTI : CS96; 1039 else 1040 cs.type = (wc & 0x00007f00) ? CS94MULTI : CS94; 1041 } 1042 target = recommendation(ei, &cs); 1043 p = tmp; 1044 bit8 = ei->flags & F_8BIT; 1045 1046 /* designate the charset onto the target plane(G0/1/2/3). */ 1047 if (psenc->g[target].type == cs.type && 1048 psenc->g[target].final == cs.final && 1049 psenc->g[target].interm == cs.interm) 1050 goto planeok; 1051 1052 *p++ = '\033'; 1053 if (cs.type == CS94MULTI || cs.type == CS96MULTI) 1054 *p++ = '$'; 1055 if (target == 0 && cs.type == CS94MULTI && strchr("@AB", cs.final) && 1056 !cs.interm && !(ei->flags & F_NOOLD)) 1057 ; 1058 else if (cs.type == CS94 || cs.type == CS94MULTI) 1059 *p++ = "()*+"[target]; 1060 else 1061 *p++ = ",-./"[target]; 1062 if (cs.interm) 1063 *p++ = cs.interm; 1064 *p++ = cs.final; 1065 1066 psenc->g[target].type = cs.type; 1067 psenc->g[target].final = cs.final; 1068 psenc->g[target].interm = cs.interm; 1069 1070 planeok: 1071 /* invoke the plane onto GL or GR. */ 1072 if (psenc->gl == target) 1073 goto sideok; 1074 if (bit8 && psenc->gr == target) 1075 goto sideok; 1076 1077 if (target == 0 && (ei->flags & F_LS0)) { 1078 *p++ = '\017'; 1079 psenc->gl = 0; 1080 } else if (target == 1 && (ei->flags & F_LS1)) { 1081 *p++ = '\016'; 1082 psenc->gl = 1; 1083 } else if (target == 2 && (ei->flags & F_LS2)) { 1084 *p++ = '\033'; 1085 *p++ = 'n'; 1086 psenc->gl = 2; 1087 } else if (target == 3 && (ei->flags & F_LS3)) { 1088 *p++ = '\033'; 1089 *p++ = 'o'; 1090 psenc->gl = 3; 1091 } else if (bit8 && target == 1 && (ei->flags & F_LS1R)) { 1092 *p++ = '\033'; 1093 *p++ = '~'; 1094 psenc->gr = 1; 1095 } else if (bit8 && target == 2 && (ei->flags & F_LS2R)) { 1096 *p++ = '\033'; 1097 /*{*/ 1098 *p++ = '}'; 1099 psenc->gr = 2; 1100 } else if (bit8 && target == 3 && (ei->flags & F_LS3R)) { 1101 *p++ = '\033'; 1102 *p++ = '|'; 1103 psenc->gr = 3; 1104 } else if (target == 2 && (ei->flags & F_SS2)) { 1105 *p++ = '\033'; 1106 *p++ = 'N'; 1107 psenc->singlegl = 2; 1108 } else if (target == 3 && (ei->flags & F_SS3)) { 1109 *p++ = '\033'; 1110 *p++ = 'O'; 1111 psenc->singlegl = 3; 1112 } else if (bit8 && target == 2 && (ei->flags & F_SS2R)) { 1113 *p++ = '\216'; 1114 *p++ = 'N'; 1115 psenc->singlegl = psenc->singlegr = 2; 1116 } else if (bit8 && target == 3 && (ei->flags & F_SS3R)) { 1117 *p++ = '\217'; 1118 *p++ = 'O'; 1119 psenc->singlegl = psenc->singlegr = 3; 1120 } else 1121 goto ilseq; 1122 1123 sideok: 1124 if (psenc->singlegl == target) 1125 mask = 0x00; 1126 else if (psenc->singlegr == target) 1127 mask = 0x80; 1128 else if (psenc->gl == target) 1129 mask = 0x00; 1130 else if ((ei->flags & F_8BIT) && psenc->gr == target) 1131 mask = 0x80; 1132 else 1133 goto ilseq; 1134 1135 switch (cs.type) { 1136 case CS94: 1137 case CS96: 1138 i = 1; 1139 break; 1140 case CS94MULTI: 1141 case CS96MULTI: 1142 i = !iscntl(wc & 0xff) ? 1143 (isthree(cs.final) ? 3 : 2) : 1; 1144 break; 1145 } 1146 while (i-- > 0) 1147 *p++ = ((wc >> (i << 3)) & 0x7f) | mask; 1148 1149 /* reset single shift state */ 1150 psenc->singlegl = psenc->singlegr = -1; 1151 1152 len = (size_t)(p - tmp); 1153 if (n < len) { 1154 if (result) 1155 *result = (char *)0; 1156 *nresult = (size_t)-1; 1157 return (E2BIG); 1158 } 1159 if (result) 1160 *result = string + len; 1161 memcpy(string, tmp, len); 1162 *nresult = len; 1163 1164 return (0); 1165 1166 ilseq: 1167 *nresult = (size_t)-1; 1168 return (EILSEQ); 1169 } 1170 1171 static int 1172 _citrus_ISO2022_put_state_reset(_ISO2022EncodingInfo * __restrict ei, 1173 char * __restrict s, size_t n, _ISO2022State * __restrict psenc, 1174 size_t * __restrict nresult) 1175 { 1176 char *result; 1177 char buf[MB_LEN_MAX]; 1178 size_t len; 1179 int ret; 1180 1181 /* XXX state will be modified after this operation... */ 1182 ret = _ISO2022_sputwchar(ei, L'\0', buf, sizeof(buf), &result, psenc, 1183 &len); 1184 if (ret) { 1185 *nresult = len; 1186 return (ret); 1187 } 1188 1189 if (sizeof(buf) < len || n < len-1) { 1190 /* XXX should recover state? */ 1191 *nresult = (size_t)-1; 1192 return (E2BIG); 1193 } 1194 1195 memcpy(s, buf, len - 1); 1196 *nresult = len - 1; 1197 return (0); 1198 } 1199 1200 static int 1201 _citrus_ISO2022_wcrtomb_priv(_ISO2022EncodingInfo * __restrict ei, 1202 char * __restrict s, size_t n, wchar_t wc, 1203 _ISO2022State * __restrict psenc, size_t * __restrict nresult) 1204 { 1205 char *result; 1206 char buf[MB_LEN_MAX]; 1207 size_t len; 1208 int ret; 1209 1210 /* XXX state will be modified after this operation... */ 1211 ret = _ISO2022_sputwchar(ei, wc, buf, sizeof(buf), &result, psenc, 1212 &len); 1213 if (ret) { 1214 *nresult = len; 1215 return (ret); 1216 } 1217 1218 if (sizeof(buf) < len || n < len) { 1219 /* XXX should recover state? */ 1220 *nresult = (size_t)-1; 1221 return (E2BIG); 1222 } 1223 1224 memcpy(s, buf, len); 1225 *nresult = len; 1226 return (0); 1227 } 1228 1229 static __inline int 1230 /*ARGSUSED*/ 1231 _citrus_ISO2022_stdenc_wctocs(_ISO2022EncodingInfo * __restrict ei __unused, 1232 _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc) 1233 { 1234 wchar_t m, nm; 1235 1236 m = wc & 0x7FFF8080; 1237 nm = wc & 0x007F7F7F; 1238 if (m & 0x00800000) 1239 nm &= 0x00007F7F; 1240 else 1241 m &= 0x7F008080; 1242 if (nm & 0x007F0000) { 1243 /* ^3 mark */ 1244 m |= 0x007F0000; 1245 } else if (nm & 0x00007F00) { 1246 /* ^2 mark */ 1247 m |= 0x00007F00; 1248 } 1249 *csid = (_csid_t)m; 1250 *idx = (_index_t)nm; 1251 1252 return (0); 1253 } 1254 1255 static __inline int 1256 /*ARGSUSED*/ 1257 _citrus_ISO2022_stdenc_cstowc(_ISO2022EncodingInfo * __restrict ei __unused, 1258 wchar_t * __restrict wc, _csid_t csid, _index_t idx) 1259 { 1260 1261 *wc = (wchar_t)(csid & 0x7F808080) | (wchar_t)idx; 1262 1263 return (0); 1264 } 1265 1266 static __inline int 1267 /*ARGSUSED*/ 1268 _citrus_ISO2022_stdenc_get_state_desc_generic(_ISO2022EncodingInfo * __restrict ei __unused, 1269 _ISO2022State * __restrict psenc, int * __restrict rstate) 1270 { 1271 1272 if (psenc->chlen == 0) { 1273 /* XXX: it should distinguish initial and stable. */ 1274 *rstate = _STDENC_SDGEN_STABLE; 1275 } else 1276 *rstate = (psenc->ch[0] == '\033') ? 1277 _STDENC_SDGEN_INCOMPLETE_SHIFT : 1278 _STDENC_SDGEN_INCOMPLETE_CHAR; 1279 return (0); 1280 } 1281 1282 /* ---------------------------------------------------------------------- 1283 * public interface for stdenc 1284 */ 1285 1286 _CITRUS_STDENC_DECLS(ISO2022); 1287 _CITRUS_STDENC_DEF_OPS(ISO2022); 1288 1289 #include "citrus_stdenc_template.h" 1290