1 /* Conversion module for ISO-2022-JP-3. 2 Copyright (C) 1998-2022 Free Software Foundation, Inc. 3 Copyright The GNU Toolchain Authors. 4 This file is part of the GNU C Library. 5 6 The GNU C Library is free software; you can redistribute it and/or 7 modify it under the terms of the GNU Lesser General Public 8 License as published by the Free Software Foundation; either 9 version 2.1 of the License, or (at your option) any later version. 10 11 The GNU C Library is distributed in the hope that it will be useful, 12 but WITHOUT ANY WARRANTY; without even the implied warranty of 13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 14 Lesser General Public License for more details. 15 16 You should have received a copy of the GNU Lesser General Public 17 License along with the GNU C Library; if not, see 18 <https://www.gnu.org/licenses/>. */ 19 20 #include <assert.h> 21 #include <dlfcn.h> 22 #include <gconv.h> 23 #include <stdint.h> 24 #include <string.h> 25 26 #include "jis0201.h" 27 #include "jis0208.h" 28 #include "jisx0213.h" 29 30 /* This makes obvious what everybody knows: 0x1b is the Esc character. */ 31 #define ESC 0x1b 32 33 /* Definitions used in the body of the `gconv' function. */ 34 #define CHARSET_NAME "ISO-2022-JP-3//" 35 #define FROM_LOOP from_iso2022jp3_loop 36 #define TO_LOOP to_iso2022jp3_loop 37 #define DEFINE_INIT 1 38 #define DEFINE_FINI 1 39 #define ONE_DIRECTION 0 40 #define FROM_LOOP_MIN_NEEDED_FROM 1 41 #define FROM_LOOP_MAX_NEEDED_FROM 4 42 #define FROM_LOOP_MIN_NEEDED_TO 4 43 #define FROM_LOOP_MAX_NEEDED_TO 8 44 #define TO_LOOP_MIN_NEEDED_FROM 4 45 #define TO_LOOP_MAX_NEEDED_FROM 4 46 #define TO_LOOP_MIN_NEEDED_TO 1 47 #define TO_LOOP_MAX_NEEDED_TO 6 48 #define PREPARE_LOOP \ 49 int saved_state; \ 50 int *statep = &data->__statep->__count; 51 #define EXTRA_LOOP_ARGS , statep 52 53 54 /* The COUNT element of the state keeps track of the currently selected 55 character set. The possible values are: */ 56 enum 57 { 58 ASCII_set = 0, /* Esc ( B */ 59 JISX0208_1978_set = 1 << 3, /* Esc $ @ */ 60 JISX0208_1983_set = 2 << 3, /* Esc $ B */ 61 JISX0201_Roman_set = 3 << 3, /* Esc ( J */ 62 JISX0201_Kana_set = 4 << 3, /* Esc ( I */ 63 JISX0213_1_2000_set = 5 << 3, /* Esc $ ( O */ 64 JISX0213_2_set = 6 << 3, /* Esc $ ( P */ 65 JISX0213_1_2004_set = 7 << 3, /* Esc $ ( Q */ 66 CURRENT_SEL_MASK = 7 << 3 67 }; 68 69 /* During UCS-4 to ISO-2022-JP-3 conversion, the COUNT element of the 70 state also contains the last two bytes to be output, shifted by 6 71 bits, and a one-bit indicator whether they must be preceded by the 72 shift sequence, in bit 22. During ISO-2022-JP-3 to UCS-4 73 conversion, COUNT may also contain a non-zero pending wide 74 character, shifted by six bits. This happens for certain inputs in 75 JISX0213_1_2004_set and JISX0213_2_set if the second wide character 76 in a combining sequence cannot be written because the buffer is 77 full. */ 78 79 /* Since this is a stateful encoding we have to provide code which resets 80 the output state to the initial state. This has to be done during the 81 flushing. */ 82 #define EMIT_SHIFT_TO_INIT \ 83 if ((data->__statep->__count & ~7) != ASCII_set) \ 84 { \ 85 if (FROM_DIRECTION) \ 86 { \ 87 uint32_t ch = data->__statep->__count >> 6; \ 88 \ 89 if (__glibc_unlikely (ch != 0)) \ 90 { \ 91 if (__glibc_likely (outbuf + 4 <= outend)) \ 92 { \ 93 /* Write out the last character. */ \ 94 put32u (outbuf, ch); \ 95 outbuf += 4; \ 96 data->__statep->__count &= 7; \ 97 data->__statep->__count |= ASCII_set; \ 98 } \ 99 else \ 100 /* We don't have enough room in the output buffer. */ \ 101 status = __GCONV_FULL_OUTPUT; \ 102 } \ 103 else \ 104 { \ 105 data->__statep->__count &= 7; \ 106 data->__statep->__count |= ASCII_set; \ 107 } \ 108 } \ 109 else \ 110 { \ 111 /* We are not in the initial state. To switch back we have \ 112 to write out the buffered character and/or emit the sequence \ 113 `Esc ( B'. */ \ 114 size_t need = \ 115 (data->__statep->__count >> 6 \ 116 ? (data->__statep->__count >> 22 ? 3 : 0) + 2 \ 117 : 0) \ 118 + ((data->__statep->__count & CURRENT_SEL_MASK) != ASCII_set \ 119 ? 3 : 0); \ 120 \ 121 if (__glibc_unlikely (outbuf + need > outend)) \ 122 /* We don't have enough room in the output buffer. */ \ 123 status = __GCONV_FULL_OUTPUT; \ 124 else \ 125 { \ 126 if (data->__statep->__count >> 6) \ 127 { \ 128 uint32_t lasttwo = data->__statep->__count >> 6; \ 129 \ 130 if (lasttwo >> 16) \ 131 { \ 132 /* Write out the shift sequence before the last \ 133 character. */ \ 134 assert ((data->__statep->__count & CURRENT_SEL_MASK) \ 135 == JISX0208_1983_set); \ 136 *outbuf++ = ESC; \ 137 *outbuf++ = '$'; \ 138 *outbuf++ = 'B'; \ 139 } \ 140 /* Write out the last character. */ \ 141 *outbuf++ = (lasttwo >> 8) & 0xff; \ 142 *outbuf++ = lasttwo & 0xff; \ 143 } \ 144 if ((data->__statep->__count & CURRENT_SEL_MASK) != ASCII_set) \ 145 { \ 146 /* Write out the shift sequence. */ \ 147 *outbuf++ = ESC; \ 148 *outbuf++ = '('; \ 149 *outbuf++ = 'B'; \ 150 } \ 151 data->__statep->__count &= 7; \ 152 data->__statep->__count |= ASCII_set; \ 153 } \ 154 } \ 155 } 156 157 158 /* Since we might have to reset input pointer we must be able to save 159 and retore the state. */ 160 #define SAVE_RESET_STATE(Save) \ 161 if (Save) \ 162 saved_state = *statep; \ 163 else \ 164 *statep = saved_state 165 166 167 /* First define the conversion function from ISO-2022-JP-3 to UCS-4. */ 168 #define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM 169 #define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM 170 #define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO 171 #define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO 172 #define LOOPFCT FROM_LOOP 173 #define BODY \ 174 { \ 175 uint32_t ch; \ 176 \ 177 /* Output any pending character. */ \ 178 ch = set >> 6; \ 179 if (__glibc_unlikely (ch != 0)) \ 180 { \ 181 put32 (outptr, ch); \ 182 outptr += 4; \ 183 /* Remove the pending character, but preserve state bits. */ \ 184 set &= (1 << 6) - 1; \ 185 continue; \ 186 } \ 187 \ 188 /* Otherwise read the next input byte. */ \ 189 ch = *inptr; \ 190 \ 191 /* Recognize escape sequences. */ \ 192 if (__glibc_unlikely (ch == ESC)) \ 193 { \ 194 /* We now must be prepared to read two to three more bytes. \ 195 If we have a match in the first byte but then the input buffer \ 196 ends we terminate with an error since we must not risk missing \ 197 an escape sequence just because it is not entirely in the \ 198 current input buffer. */ \ 199 if (__builtin_expect (inptr + 2 >= inend, 0) \ 200 || (inptr[1] == '$' && inptr[2] == '(' \ 201 && __builtin_expect (inptr + 3 >= inend, 0))) \ 202 { \ 203 /* Not enough input available. */ \ 204 result = __GCONV_INCOMPLETE_INPUT; \ 205 break; \ 206 } \ 207 \ 208 if (inptr[1] == '(') \ 209 { \ 210 if (inptr[2] == 'B') \ 211 { \ 212 /* ASCII selected. */ \ 213 set = ASCII_set; \ 214 inptr += 3; \ 215 continue; \ 216 } \ 217 else if (inptr[2] == 'J') \ 218 { \ 219 /* JIS X 0201 selected. */ \ 220 set = JISX0201_Roman_set; \ 221 inptr += 3; \ 222 continue; \ 223 } \ 224 else if (inptr[2] == 'I') \ 225 { \ 226 /* JIS X 0201 selected. */ \ 227 set = JISX0201_Kana_set; \ 228 inptr += 3; \ 229 continue; \ 230 } \ 231 } \ 232 else if (inptr[1] == '$') \ 233 { \ 234 if (inptr[2] == '@') \ 235 { \ 236 /* JIS X 0208-1978 selected. */ \ 237 set = JISX0208_1978_set; \ 238 inptr += 3; \ 239 continue; \ 240 } \ 241 else if (inptr[2] == 'B') \ 242 { \ 243 /* JIS X 0208-1983 selected. */ \ 244 set = JISX0208_1983_set; \ 245 inptr += 3; \ 246 continue; \ 247 } \ 248 else if (inptr[2] == '(') \ 249 { \ 250 if (inptr[3] == 'O' || inptr[3] == 'Q') \ 251 { \ 252 /* JIS X 0213 plane 1 selected. */ \ 253 /* In this direction we don't need to distinguish the \ 254 versions from 2000 and 2004. */ \ 255 set = JISX0213_1_2004_set; \ 256 inptr += 4; \ 257 continue; \ 258 } \ 259 else if (inptr[3] == 'P') \ 260 { \ 261 /* JIS X 0213 plane 2 selected. */ \ 262 set = JISX0213_2_set; \ 263 inptr += 4; \ 264 continue; \ 265 } \ 266 } \ 267 } \ 268 } \ 269 \ 270 if (ch >= 0x80) \ 271 { \ 272 STANDARD_FROM_LOOP_ERR_HANDLER (1); \ 273 } \ 274 else if (set == ASCII_set || (ch < 0x21 || ch == 0x7f)) \ 275 /* Almost done, just advance the input pointer. */ \ 276 ++inptr; \ 277 else if (set == JISX0201_Roman_set) \ 278 { \ 279 /* Use the JIS X 0201 table. */ \ 280 ch = jisx0201_to_ucs4 (ch); \ 281 if (__glibc_unlikely (ch == __UNKNOWN_10646_CHAR)) \ 282 { \ 283 STANDARD_FROM_LOOP_ERR_HANDLER (1); \ 284 } \ 285 ++inptr; \ 286 } \ 287 else if (set == JISX0201_Kana_set) \ 288 { \ 289 /* Use the JIS X 0201 table. */ \ 290 ch = jisx0201_to_ucs4 (ch + 0x80); \ 291 if (__glibc_unlikely (ch == __UNKNOWN_10646_CHAR)) \ 292 { \ 293 STANDARD_FROM_LOOP_ERR_HANDLER (1); \ 294 } \ 295 ++inptr; \ 296 } \ 297 else if (set == JISX0208_1978_set || set == JISX0208_1983_set) \ 298 { \ 299 /* XXX I don't have the tables for these two old variants of \ 300 JIS X 0208. Therefore I'm using the tables for JIS X \ 301 0208-1990. If somebody has problems with this please \ 302 provide the appropriate tables. */ \ 303 ch = jisx0208_to_ucs4 (&inptr, inend - inptr, 0); \ 304 \ 305 if (__glibc_unlikely (ch == 0)) \ 306 { \ 307 result = __GCONV_INCOMPLETE_INPUT; \ 308 break; \ 309 } \ 310 else if (__glibc_unlikely (ch == __UNKNOWN_10646_CHAR)) \ 311 { \ 312 STANDARD_FROM_LOOP_ERR_HANDLER (1); \ 313 } \ 314 } \ 315 else /* (set == JISX0213_1_2004_set || set == JISX0213_2_set) */ \ 316 { \ 317 if (__glibc_unlikely (inptr + 1 >= inend)) \ 318 { \ 319 result = __GCONV_INCOMPLETE_INPUT; \ 320 break; \ 321 } \ 322 \ 323 ch = jisx0213_to_ucs4 ( \ 324 ((JISX0213_1_2004_set - set + (1 << 3)) << 5) + ch, \ 325 inptr[1]); \ 326 if (ch == 0) \ 327 STANDARD_FROM_LOOP_ERR_HANDLER (1); \ 328 \ 329 if (ch < 0x80) \ 330 { \ 331 /* It's a combining character. */ \ 332 uint32_t u1 = __jisx0213_to_ucs_combining[ch - 1][0]; \ 333 uint32_t u2 = __jisx0213_to_ucs_combining[ch - 1][1]; \ 334 \ 335 inptr += 2; \ 336 \ 337 put32 (outptr, u1); \ 338 outptr += 4; \ 339 \ 340 /* See whether we have room for two characters. */ \ 341 if (outptr + 4 <= outend) \ 342 { \ 343 put32 (outptr, u2); \ 344 outptr += 4; \ 345 continue; \ 346 } \ 347 \ 348 /* Otherwise store only the first character now, and \ 349 put the second one into the queue. */ \ 350 set |= u2 << 6; \ 351 /* Tell the caller why we terminate the loop. */ \ 352 result = __GCONV_FULL_OUTPUT; \ 353 break; \ 354 } \ 355 \ 356 inptr += 2; \ 357 } \ 358 \ 359 put32 (outptr, ch); \ 360 outptr += 4; \ 361 } 362 #define LOOP_NEED_FLAGS 363 #define EXTRA_LOOP_DECLS , int *statep 364 #define INIT_PARAMS int set = *statep 365 #define UPDATE_PARAMS *statep = set 366 #include <iconv/loop.c> 367 368 369 /* Next, define the other direction, from UCS-4 to ISO-2022-JP-3. */ 370 371 /* Composition tables for each of the relevant combining characters. */ 372 static const struct 373 { 374 uint16_t base; 375 uint16_t composed; 376 } comp_table_data[] = 377 { 378 #define COMP_TABLE_IDX_02E5 0 379 #define COMP_TABLE_LEN_02E5 1 380 { 0x2b64, 0x2b65 }, /* 0x12B65 = 0x12B64 U+02E5 */ 381 #define COMP_TABLE_IDX_02E9 (COMP_TABLE_IDX_02E5 + COMP_TABLE_LEN_02E5) 382 #define COMP_TABLE_LEN_02E9 1 383 { 0x2b60, 0x2b66 }, /* 0x12B66 = 0x12B60 U+02E9 */ 384 #define COMP_TABLE_IDX_0300 (COMP_TABLE_IDX_02E9 + COMP_TABLE_LEN_02E9) 385 #define COMP_TABLE_LEN_0300 5 386 { 0x295c, 0x2b44 }, /* 0x12B44 = 0x1295C U+0300 */ 387 { 0x2b38, 0x2b48 }, /* 0x12B48 = 0x12B38 U+0300 */ 388 { 0x2b37, 0x2b4a }, /* 0x12B4A = 0x12B37 U+0300 */ 389 { 0x2b30, 0x2b4c }, /* 0x12B4C = 0x12B30 U+0300 */ 390 { 0x2b43, 0x2b4e }, /* 0x12B4E = 0x12B43 U+0300 */ 391 #define COMP_TABLE_IDX_0301 (COMP_TABLE_IDX_0300 + COMP_TABLE_LEN_0300) 392 #define COMP_TABLE_LEN_0301 4 393 { 0x2b38, 0x2b49 }, /* 0x12B49 = 0x12B38 U+0301 */ 394 { 0x2b37, 0x2b4b }, /* 0x12B4B = 0x12B37 U+0301 */ 395 { 0x2b30, 0x2b4d }, /* 0x12B4D = 0x12B30 U+0301 */ 396 { 0x2b43, 0x2b4f }, /* 0x12B4F = 0x12B43 U+0301 */ 397 #define COMP_TABLE_IDX_309A (COMP_TABLE_IDX_0301 + COMP_TABLE_LEN_0301) 398 #define COMP_TABLE_LEN_309A 14 399 { 0x242b, 0x2477 }, /* 0x12477 = 0x1242B U+309A */ 400 { 0x242d, 0x2478 }, /* 0x12478 = 0x1242D U+309A */ 401 { 0x242f, 0x2479 }, /* 0x12479 = 0x1242F U+309A */ 402 { 0x2431, 0x247a }, /* 0x1247A = 0x12431 U+309A */ 403 { 0x2433, 0x247b }, /* 0x1247B = 0x12433 U+309A */ 404 { 0x252b, 0x2577 }, /* 0x12577 = 0x1252B U+309A */ 405 { 0x252d, 0x2578 }, /* 0x12578 = 0x1252D U+309A */ 406 { 0x252f, 0x2579 }, /* 0x12579 = 0x1252F U+309A */ 407 { 0x2531, 0x257a }, /* 0x1257A = 0x12531 U+309A */ 408 { 0x2533, 0x257b }, /* 0x1257B = 0x12533 U+309A */ 409 { 0x253b, 0x257c }, /* 0x1257C = 0x1253B U+309A */ 410 { 0x2544, 0x257d }, /* 0x1257D = 0x12544 U+309A */ 411 { 0x2548, 0x257e }, /* 0x1257E = 0x12548 U+309A */ 412 { 0x2675, 0x2678 }, /* 0x12678 = 0x12675 U+309A */ 413 }; 414 415 #define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM 416 #define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM 417 #define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO 418 #define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO 419 #define LOOPFCT TO_LOOP 420 #define BODY \ 421 { \ 422 uint32_t ch = get32 (inptr); \ 423 \ 424 if (lasttwo != 0) \ 425 { \ 426 /* Attempt to combine the last character with this one. */ \ 427 unsigned int idx; \ 428 unsigned int len; \ 429 \ 430 if (ch == 0x02e5) \ 431 idx = COMP_TABLE_IDX_02E5, len = COMP_TABLE_LEN_02E5; \ 432 else if (ch == 0x02e9) \ 433 idx = COMP_TABLE_IDX_02E9, len = COMP_TABLE_LEN_02E9; \ 434 else if (ch == 0x0300) \ 435 idx = COMP_TABLE_IDX_0300, len = COMP_TABLE_LEN_0300; \ 436 else if (ch == 0x0301) \ 437 idx = COMP_TABLE_IDX_0301, len = COMP_TABLE_LEN_0301; \ 438 else if (ch == 0x309a) \ 439 idx = COMP_TABLE_IDX_309A, len = COMP_TABLE_LEN_309A; \ 440 else \ 441 goto not_combining; \ 442 \ 443 do \ 444 if (comp_table_data[idx].base == (uint16_t) lasttwo) \ 445 break; \ 446 while (++idx, --len > 0); \ 447 \ 448 if (len > 0) \ 449 { \ 450 /* Output the combined character. */ \ 451 /* We know the combined character is in JISX0213 plane 1, \ 452 but the buffered character may have been in JISX0208 or in \ 453 JISX0213 plane 1. */ \ 454 size_t need = \ 455 (lasttwo >> 16 \ 456 || (set != JISX0213_1_2000_set && set != JISX0213_1_2004_set) \ 457 ? 4 : 0); \ 458 \ 459 if (__glibc_unlikely (outptr + need + 2 > outend)) \ 460 { \ 461 result = __GCONV_FULL_OUTPUT; \ 462 break; \ 463 } \ 464 if (need) \ 465 { \ 466 /* But first, output the escape sequence. */ \ 467 *outptr++ = ESC; \ 468 *outptr++ = '$'; \ 469 *outptr++ = '('; \ 470 *outptr++ = 'O'; \ 471 set = JISX0213_1_2000_set; \ 472 } \ 473 lasttwo = comp_table_data[idx].composed; \ 474 *outptr++ = (lasttwo >> 8) & 0xff; \ 475 *outptr++ = lasttwo & 0xff; \ 476 lasttwo = 0; \ 477 inptr += 4; \ 478 continue; \ 479 } \ 480 \ 481 not_combining: \ 482 /* Output the buffered character. */ \ 483 /* We know it is in JISX0208 or in JISX0213 plane 1. */ \ 484 { \ 485 size_t need = (lasttwo >> 16 ? 3 : 0); \ 486 \ 487 if (__glibc_unlikely (outptr + need + 2 > outend)) \ 488 { \ 489 result = __GCONV_FULL_OUTPUT; \ 490 break; \ 491 } \ 492 if (need) \ 493 { \ 494 /* But first, output the escape sequence. */ \ 495 assert (set == JISX0208_1983_set); \ 496 *outptr++ = ESC; \ 497 *outptr++ = '$'; \ 498 *outptr++ = 'B'; \ 499 } \ 500 *outptr++ = (lasttwo >> 8) & 0xff; \ 501 *outptr++ = lasttwo & 0xff; \ 502 lasttwo = 0; \ 503 continue; \ 504 } \ 505 } \ 506 \ 507 /* First see whether we can write the character using the currently \ 508 selected character set. */ \ 509 if (set == ASCII_set) \ 510 { \ 511 /* Please note that the NUL byte is *not* matched if we are not \ 512 currently using the ASCII charset. This is because we must \ 513 switch to the initial state whenever a NUL byte is written. */ \ 514 if (ch <= 0x7f) \ 515 { \ 516 *outptr++ = ch; \ 517 inptr += 4; \ 518 continue; \ 519 } \ 520 } \ 521 /* ISO-2022-JP recommends to encode the newline character always in \ 522 ASCII since this allows a context-free interpretation of the \ 523 characters at the beginning of the next line. Otherwise it would \ 524 have to be known whether the last line ended using ASCII or \ 525 JIS X 0201. */ \ 526 else if (set == JISX0201_Roman_set) \ 527 { \ 528 unsigned char buf[1]; \ 529 if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ 530 && buf[0] > 0x20 && buf[0] < 0x80) \ 531 { \ 532 *outptr++ = buf[0]; \ 533 inptr += 4; \ 534 continue; \ 535 } \ 536 } \ 537 else if (set == JISX0201_Kana_set) \ 538 { \ 539 unsigned char buf[1]; \ 540 if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ 541 && buf[0] >= 0x80) \ 542 { \ 543 *outptr++ = buf[0] - 0x80; \ 544 inptr += 4; \ 545 continue; \ 546 } \ 547 } \ 548 else if (/*set == JISX0208_1978_set || */ set == JISX0208_1983_set) \ 549 { \ 550 size_t written = ucs4_to_jisx0208 (ch, outptr, outend - outptr); \ 551 \ 552 if (written != __UNKNOWN_10646_CHAR) \ 553 { \ 554 uint32_t jch = ucs4_to_jisx0213 (ch); \ 555 \ 556 if (jch & 0x0080) \ 557 { \ 558 /* A possible match in comp_table_data. Buffer it. */ \ 559 lasttwo = jch & 0x7f7f; \ 560 inptr += 4; \ 561 continue; \ 562 } \ 563 if (__glibc_unlikely (written == 0)) \ 564 { \ 565 result = __GCONV_FULL_OUTPUT; \ 566 break; \ 567 } \ 568 else \ 569 { \ 570 outptr += written; \ 571 inptr += 4; \ 572 continue; \ 573 } \ 574 } \ 575 } \ 576 else \ 577 { \ 578 /* (set == JISX0213_1_2000_set || set == JISX0213_1_2004_set \ 579 || set == JISX0213_2_set) */ \ 580 uint32_t jch = ucs4_to_jisx0213 (ch); \ 581 \ 582 if (jch != 0 \ 583 && (jch & 0x8000 \ 584 ? set == JISX0213_2_set \ 585 : (set == JISX0213_1_2004_set \ 586 || (set == JISX0213_1_2000_set \ 587 && !jisx0213_added_in_2004_p (jch))))) \ 588 { \ 589 if (jch & 0x0080) \ 590 { \ 591 /* A possible match in comp_table_data. Buffer it. */ \ 592 \ 593 /* We know it's a JISX 0213 plane 1 character. */ \ 594 assert ((jch & 0x8000) == 0); \ 595 \ 596 lasttwo = jch & 0x7f7f; \ 597 inptr += 4; \ 598 continue; \ 599 } \ 600 \ 601 if (__glibc_unlikely (outptr + 1 >= outend)) \ 602 { \ 603 result = __GCONV_FULL_OUTPUT; \ 604 break; \ 605 } \ 606 *outptr++ = (jch >> 8) & 0x7f; \ 607 *outptr++ = jch & 0x7f; \ 608 inptr += 4; \ 609 continue; \ 610 } \ 611 } \ 612 \ 613 /* The attempts to use the currently selected character set failed, \ 614 either because the character requires a different character set, \ 615 or because the character is unknown. */ \ 616 \ 617 if (ch <= 0x7f) \ 618 { \ 619 /* We must encode using ASCII. First write out the escape \ 620 sequence. */ \ 621 if (__glibc_unlikely (outptr + 3 > outend)) \ 622 { \ 623 result = __GCONV_FULL_OUTPUT; \ 624 break; \ 625 } \ 626 \ 627 *outptr++ = ESC; \ 628 *outptr++ = '('; \ 629 *outptr++ = 'B'; \ 630 set = ASCII_set; \ 631 \ 632 if (__glibc_unlikely (outptr >= outend)) \ 633 { \ 634 result = __GCONV_FULL_OUTPUT; \ 635 break; \ 636 } \ 637 *outptr++ = ch; \ 638 } \ 639 else \ 640 { \ 641 unsigned char buf[2]; \ 642 \ 643 /* Try JIS X 0201 Roman. */ \ 644 if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ 645 && buf[0] > 0x20 && buf[0] < 0x80) \ 646 { \ 647 if (set != JISX0201_Roman_set) \ 648 { \ 649 if (__glibc_unlikely (outptr + 3 > outend)) \ 650 { \ 651 result = __GCONV_FULL_OUTPUT; \ 652 break; \ 653 } \ 654 *outptr++ = ESC; \ 655 *outptr++ = '('; \ 656 *outptr++ = 'J'; \ 657 set = JISX0201_Roman_set; \ 658 } \ 659 \ 660 if (__glibc_unlikely (outptr >= outend)) \ 661 { \ 662 result = __GCONV_FULL_OUTPUT; \ 663 break; \ 664 } \ 665 *outptr++ = buf[0]; \ 666 } \ 667 else \ 668 { \ 669 uint32_t jch = ucs4_to_jisx0213 (ch); \ 670 \ 671 /* Try JIS X 0208. */ \ 672 size_t written = ucs4_to_jisx0208 (ch, buf, 2); \ 673 if (written != __UNKNOWN_10646_CHAR) \ 674 { \ 675 if (jch & 0x0080) \ 676 { \ 677 /* A possible match in comp_table_data. Buffer it. */ \ 678 lasttwo = ((set != JISX0208_1983_set ? 1 : 0) << 16) \ 679 | (jch & 0x7f7f); \ 680 set = JISX0208_1983_set; \ 681 inptr += 4; \ 682 continue; \ 683 } \ 684 \ 685 if (set != JISX0208_1983_set) \ 686 { \ 687 if (__glibc_unlikely (outptr + 3 > outend)) \ 688 { \ 689 result = __GCONV_FULL_OUTPUT; \ 690 break; \ 691 } \ 692 *outptr++ = ESC; \ 693 *outptr++ = '$'; \ 694 *outptr++ = 'B'; \ 695 set = JISX0208_1983_set; \ 696 } \ 697 \ 698 if (__glibc_unlikely (outptr + 2 > outend)) \ 699 { \ 700 result = __GCONV_FULL_OUTPUT; \ 701 break; \ 702 } \ 703 *outptr++ = buf[0]; \ 704 *outptr++ = buf[1]; \ 705 } \ 706 else \ 707 { \ 708 /* Try JIS X 0213. */ \ 709 if (jch != 0) \ 710 { \ 711 int new_set = \ 712 (jch & 0x8000 \ 713 ? JISX0213_2_set \ 714 : jisx0213_added_in_2004_p (jch) \ 715 ? JISX0213_1_2004_set \ 716 : JISX0213_1_2000_set); \ 717 \ 718 if (set != new_set) \ 719 { \ 720 if (__glibc_unlikely (outptr + 4 > outend)) \ 721 { \ 722 result = __GCONV_FULL_OUTPUT; \ 723 break; \ 724 } \ 725 *outptr++ = ESC; \ 726 *outptr++ = '$'; \ 727 *outptr++ = '('; \ 728 *outptr++ = \ 729 ((new_set - JISX0213_1_2000_set) >> 3) + 'O'; \ 730 set = new_set; \ 731 } \ 732 \ 733 if (jch & 0x0080) \ 734 { \ 735 /* A possible match in comp_table_data. \ 736 Buffer it. */ \ 737 \ 738 /* We know it's a JIS X 0213 plane 1 character. */ \ 739 assert ((jch & 0x8000) == 0); \ 740 \ 741 lasttwo = jch & 0x7f7f; \ 742 inptr += 4; \ 743 continue; \ 744 } \ 745 \ 746 if (__glibc_unlikely (outptr + 1 >= outend)) \ 747 { \ 748 result = __GCONV_FULL_OUTPUT; \ 749 break; \ 750 } \ 751 *outptr++ = (jch >> 8) & 0x7f; \ 752 *outptr++ = jch & 0x7f; \ 753 } \ 754 else \ 755 { \ 756 /* Try JIS X 0201 Katakana. This is officially not part \ 757 of ISO-2022-JP-3. Therefore we try it after all other \ 758 attempts. */ \ 759 if (ucs4_to_jisx0201 (ch, buf) != __UNKNOWN_10646_CHAR \ 760 && buf[0] >= 0x80) \ 761 { \ 762 if (set != JISX0201_Kana_set) \ 763 { \ 764 if (__builtin_expect (outptr + 3 > outend, 0)) \ 765 { \ 766 result = __GCONV_FULL_OUTPUT; \ 767 break; \ 768 } \ 769 *outptr++ = ESC; \ 770 *outptr++ = '('; \ 771 *outptr++ = 'I'; \ 772 set = JISX0201_Kana_set; \ 773 } \ 774 \ 775 if (__glibc_unlikely (outptr >= outend)) \ 776 { \ 777 result = __GCONV_FULL_OUTPUT; \ 778 break; \ 779 } \ 780 *outptr++ = buf[0] - 0x80; \ 781 } \ 782 else \ 783 { \ 784 UNICODE_TAG_HANDLER (ch, 4); \ 785 \ 786 /* Illegal character. */ \ 787 STANDARD_TO_LOOP_ERR_HANDLER (4); \ 788 } \ 789 } \ 790 } \ 791 } \ 792 } \ 793 \ 794 /* Now that we wrote the output increment the input pointer. */ \ 795 inptr += 4; \ 796 } 797 #define LOOP_NEED_FLAGS 798 #define EXTRA_LOOP_DECLS , int *statep 799 #define INIT_PARAMS int set = *statep & CURRENT_SEL_MASK; \ 800 uint32_t lasttwo = *statep >> 6 801 #define REINIT_PARAMS do \ 802 { \ 803 set = *statep & CURRENT_SEL_MASK; \ 804 lasttwo = *statep >> 6; \ 805 } \ 806 while (0) 807 #define UPDATE_PARAMS *statep = set | (lasttwo << 6) 808 #include <iconv/loop.c> 809 810 811 /* Now define the toplevel functions. */ 812 #include <iconv/skeleton.c> 813