xf.li | bdd93d5 | 2023-05-12 07:10:14 -0700 | [diff] [blame^] | 1 | /* Conversion module for ISO-2022-CN. |
| 2 | Copyright (C) 1999-2016 Free Software Foundation, Inc. |
| 3 | This file is part of the GNU C Library. |
| 4 | Contributed by Ulrich Drepper <drepper@cygnus.com>, 1999. |
| 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 | <http://www.gnu.org/licenses/>. */ |
| 19 | |
| 20 | #include <dlfcn.h> |
| 21 | #include <gconv.h> |
| 22 | #include <stdint.h> |
| 23 | #include <string.h> |
| 24 | #include "gb2312.h" |
| 25 | #include "cns11643l1.h" |
| 26 | #include "cns11643l2.h" |
| 27 | |
| 28 | #include <assert.h> |
| 29 | |
| 30 | /* This makes obvious what everybody knows: 0x1b is the Esc character. */ |
| 31 | #define ESC 0x1b |
| 32 | |
| 33 | /* We have single-byte shift-in and shift-out sequences, and the single |
| 34 | shift sequence SS2 which replaces the SS2 designation for the next |
| 35 | two bytes. */ |
| 36 | #define SI 0x0f |
| 37 | #define SO 0x0e |
| 38 | #define SS2_0 ESC |
| 39 | #define SS2_1 0x4e |
| 40 | |
| 41 | /* Definitions used in the body of the `gconv' function. */ |
| 42 | #define CHARSET_NAME "ISO-2022-CN//" |
| 43 | #define DEFINE_INIT 1 |
| 44 | #define DEFINE_FINI 1 |
| 45 | #define FROM_LOOP from_iso2022cn_loop |
| 46 | #define TO_LOOP to_iso2022cn_loop |
| 47 | #define ONE_DIRECTION 0 |
| 48 | #define FROM_LOOP_MIN_NEEDED_FROM 1 |
| 49 | #define FROM_LOOP_MAX_NEEDED_FROM 4 |
| 50 | #define FROM_LOOP_MIN_NEEDED_TO 4 |
| 51 | #define FROM_LOOP_MAX_NEEDED_TO 4 |
| 52 | #define TO_LOOP_MIN_NEEDED_FROM 4 |
| 53 | #define TO_LOOP_MAX_NEEDED_FROM 4 |
| 54 | #define TO_LOOP_MIN_NEEDED_TO 1 |
| 55 | #define TO_LOOP_MAX_NEEDED_TO 6 |
| 56 | #define PREPARE_LOOP \ |
| 57 | int save_set; \ |
| 58 | int *setp = &data->__statep->__count; |
| 59 | #define EXTRA_LOOP_ARGS , setp |
| 60 | |
| 61 | |
| 62 | /* The COUNT element of the state keeps track of the currently selected |
| 63 | character set. The possible values are: */ |
| 64 | enum |
| 65 | { |
| 66 | ASCII_set = 0, |
| 67 | GB2312_set = 8, |
| 68 | CNS11643_1_set = 16, |
| 69 | CNS11643_2_set = 24, |
| 70 | CURRENT_SEL_MASK = 24, |
| 71 | GB2312_ann = 32, |
| 72 | CNS11643_1_ann = 64, |
| 73 | CNS11643_2_ann = 128, |
| 74 | CURRENT_ANN_MASK = 224 |
| 75 | }; |
| 76 | |
| 77 | |
| 78 | /* Since this is a stateful encoding we have to provide code which resets |
| 79 | the output state to the initial state. This has to be done during the |
| 80 | flushing. */ |
| 81 | #define EMIT_SHIFT_TO_INIT \ |
| 82 | if (data->__statep->__count != ASCII_set) \ |
| 83 | { \ |
| 84 | if (FROM_DIRECTION) \ |
| 85 | /* It's easy, we don't have to emit anything, we just reset the \ |
| 86 | state for the input. */ \ |
| 87 | data->__statep->__count = ASCII_set; \ |
| 88 | else \ |
| 89 | { \ |
| 90 | /* We are not in the initial state. To switch back we have \ |
| 91 | to emit `SI'. */ \ |
| 92 | if (__glibc_unlikely (outbuf == outend)) \ |
| 93 | /* We don't have enough room in the output buffer. */ \ |
| 94 | status = __GCONV_FULL_OUTPUT; \ |
| 95 | else \ |
| 96 | { \ |
| 97 | /* Write out the shift sequence. */ \ |
| 98 | *outbuf++ = SI; \ |
| 99 | data->__statep->__count = ASCII_set; \ |
| 100 | } \ |
| 101 | } \ |
| 102 | } |
| 103 | |
| 104 | |
| 105 | /* Since we might have to reset input pointer we must be able to save |
| 106 | and retore the state. */ |
| 107 | #define SAVE_RESET_STATE(Save) \ |
| 108 | if (Save) \ |
| 109 | save_set = *setp; \ |
| 110 | else \ |
| 111 | *setp = save_set |
| 112 | |
| 113 | |
| 114 | /* First define the conversion function from ISO-2022-CN to UCS4. */ |
| 115 | #define MIN_NEEDED_INPUT FROM_LOOP_MIN_NEEDED_FROM |
| 116 | #define MAX_NEEDED_INPUT FROM_LOOP_MAX_NEEDED_FROM |
| 117 | #define MIN_NEEDED_OUTPUT FROM_LOOP_MIN_NEEDED_TO |
| 118 | #define MAX_NEEDED_OUTPUT FROM_LOOP_MAX_NEEDED_TO |
| 119 | #define LOOPFCT FROM_LOOP |
| 120 | #define BODY \ |
| 121 | { \ |
| 122 | uint32_t ch = *inptr; \ |
| 123 | \ |
| 124 | /* This is a 7bit character set, disallow all 8bit characters. */ \ |
| 125 | if (__glibc_unlikely (ch >= 0x7f)) \ |
| 126 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
| 127 | \ |
| 128 | /* Recognize escape sequences. */ \ |
| 129 | if (__builtin_expect (ch, 0) == ESC) \ |
| 130 | { \ |
| 131 | /* There are two kinds of escape sequences we have to handle: \ |
| 132 | - those announcing the use of GB and CNS characters on the \ |
| 133 | line; we can simply ignore them \ |
| 134 | - the initial byte of the SS2 sequence. \ |
| 135 | */ \ |
| 136 | if (__builtin_expect (inptr + 2 > inend, 0) \ |
| 137 | || (inptr[1] == '$' \ |
| 138 | && (__builtin_expect (inptr + 3 > inend, 0) \ |
| 139 | || (inptr[2] == ')' \ |
| 140 | && __builtin_expect (inptr + 4 > inend, 0)) \ |
| 141 | || (inptr[2] == '*' \ |
| 142 | && __builtin_expect (inptr + 4 > inend, 0)))) \ |
| 143 | || (inptr[1] == SS2_1 \ |
| 144 | && __builtin_expect (inptr + 4 > inend, 0))) \ |
| 145 | { \ |
| 146 | result = __GCONV_INCOMPLETE_INPUT; \ |
| 147 | break; \ |
| 148 | } \ |
| 149 | if (inptr[1] == '$' \ |
| 150 | && ((inptr[2] == ')' && (inptr[3] == 'A' || inptr[3] == 'G')) \ |
| 151 | || (inptr[2] == '*' && inptr[3] == 'H'))) \ |
| 152 | { \ |
| 153 | /* OK, we accept those character sets. */ \ |
| 154 | if (inptr[3] == 'A') \ |
| 155 | ann = GB2312_ann; \ |
| 156 | else if (inptr[3] == 'G') \ |
| 157 | ann = CNS11643_1_ann; \ |
| 158 | inptr += 4; \ |
| 159 | continue; \ |
| 160 | } \ |
| 161 | } \ |
| 162 | else if (__builtin_expect (ch, 0) == SO) \ |
| 163 | { \ |
| 164 | /* Switch to use GB2312 or CNS 11643 plane 1, depending on which \ |
| 165 | S0 designation came last. The only problem is what to do with \ |
| 166 | faulty input files where no designator came. \ |
| 167 | XXX For now I'll default to use GB2312. If this is not the \ |
| 168 | best behaviour (e.g., we should flag an error) let me know. */ \ |
| 169 | ++inptr; \ |
| 170 | set = ann == CNS11643_1_ann ? CNS11643_1_set : GB2312_set; \ |
| 171 | continue; \ |
| 172 | } \ |
| 173 | else if (__builtin_expect (ch, 0) == SI) \ |
| 174 | { \ |
| 175 | /* Switch to use ASCII. */ \ |
| 176 | ++inptr; \ |
| 177 | set = ASCII_set; \ |
| 178 | continue; \ |
| 179 | } \ |
| 180 | \ |
| 181 | if (__builtin_expect (ch, 0) == ESC && inptr[1] == SS2_1) \ |
| 182 | { \ |
| 183 | /* This is a character from CNS 11643 plane 2. \ |
| 184 | XXX We could test here whether the use of this character \ |
| 185 | set was announced. */ \ |
| 186 | inptr += 2; \ |
| 187 | ch = cns11643l2_to_ucs4 (&inptr, 2, 0); \ |
| 188 | if (__builtin_expect (ch, 0) == __UNKNOWN_10646_CHAR) \ |
| 189 | { \ |
| 190 | inptr -= 2; \ |
| 191 | STANDARD_FROM_LOOP_ERR_HANDLER (2); \ |
| 192 | } \ |
| 193 | } \ |
| 194 | else if (set == ASCII_set) \ |
| 195 | { \ |
| 196 | /* Almost done, just advance the input pointer. */ \ |
| 197 | ++inptr; \ |
| 198 | } \ |
| 199 | else \ |
| 200 | { \ |
| 201 | /* That's pretty easy, we have a dedicated functions for this. */ \ |
| 202 | if (set == GB2312_set) \ |
| 203 | ch = gb2312_to_ucs4 (&inptr, inend - inptr, 0); \ |
| 204 | else \ |
| 205 | { \ |
| 206 | assert (set == CNS11643_1_set); \ |
| 207 | ch = cns11643l1_to_ucs4 (&inptr, inend - inptr, 0); \ |
| 208 | } \ |
| 209 | \ |
| 210 | if (__builtin_expect (ch, 1) == 0) \ |
| 211 | { \ |
| 212 | result = __GCONV_INCOMPLETE_INPUT; \ |
| 213 | break; \ |
| 214 | } \ |
| 215 | else if (__builtin_expect (ch, 1) == __UNKNOWN_10646_CHAR) \ |
| 216 | { \ |
| 217 | STANDARD_FROM_LOOP_ERR_HANDLER (1); \ |
| 218 | } \ |
| 219 | } \ |
| 220 | \ |
| 221 | put32 (outptr, ch); \ |
| 222 | outptr += 4; \ |
| 223 | } |
| 224 | #define LOOP_NEED_FLAGS |
| 225 | #define EXTRA_LOOP_DECLS , int *setp |
| 226 | #define INIT_PARAMS int set = *setp & CURRENT_SEL_MASK; \ |
| 227 | int ann = *setp & CURRENT_ANN_MASK |
| 228 | #define UPDATE_PARAMS *setp = set | ann |
| 229 | #include <iconv/loop.c> |
| 230 | |
| 231 | |
| 232 | /* Next, define the other direction. */ |
| 233 | #define MIN_NEEDED_INPUT TO_LOOP_MIN_NEEDED_FROM |
| 234 | #define MAX_NEEDED_INPUT TO_LOOP_MAX_NEEDED_FROM |
| 235 | #define MIN_NEEDED_OUTPUT TO_LOOP_MIN_NEEDED_TO |
| 236 | #define MAX_NEEDED_OUTPUT TO_LOOP_MAX_NEEDED_TO |
| 237 | #define LOOPFCT TO_LOOP |
| 238 | #define BODY \ |
| 239 | { \ |
| 240 | uint32_t ch = get32 (inptr); \ |
| 241 | \ |
| 242 | /* First see whether we can write the character using the currently \ |
| 243 | selected character set. */ \ |
| 244 | if (ch < 0x80) \ |
| 245 | { \ |
| 246 | if (set != ASCII_set) \ |
| 247 | { \ |
| 248 | *outptr++ = SI; \ |
| 249 | set = ASCII_set; \ |
| 250 | if (__glibc_unlikely (outptr == outend)) \ |
| 251 | { \ |
| 252 | result = __GCONV_FULL_OUTPUT; \ |
| 253 | break; \ |
| 254 | } \ |
| 255 | } \ |
| 256 | \ |
| 257 | *outptr++ = ch; \ |
| 258 | \ |
| 259 | /* At the end of the line we have to clear the `ann' flags since \ |
| 260 | every line must contain this information again. */ \ |
| 261 | if (ch == L'\n') \ |
| 262 | ann = 0; \ |
| 263 | } \ |
| 264 | else \ |
| 265 | { \ |
| 266 | unsigned char buf[2]; \ |
| 267 | /* Fake initialization to keep gcc quiet. */ \ |
| 268 | asm ("" : "=m" (buf)); \ |
| 269 | \ |
| 270 | int used; \ |
| 271 | size_t written = 0; \ |
| 272 | \ |
| 273 | if (set == GB2312_set || (ann & CNS11643_1_ann) == 0) \ |
| 274 | { \ |
| 275 | written = ucs4_to_gb2312 (ch, buf, 2); \ |
| 276 | used = GB2312_set; \ |
| 277 | } \ |
| 278 | else \ |
| 279 | { \ |
| 280 | written = ucs4_to_cns11643l1 (ch, buf, 2); \ |
| 281 | used = CNS11643_1_set; \ |
| 282 | } \ |
| 283 | \ |
| 284 | if (written == __UNKNOWN_10646_CHAR) \ |
| 285 | { \ |
| 286 | /* Cannot convert it using the currently selected SO set. \ |
| 287 | Next try the SS2 set. */ \ |
| 288 | written = ucs4_to_cns11643l2 (ch, buf, 2); \ |
| 289 | if (written != __UNKNOWN_10646_CHAR) \ |
| 290 | /* Yep, that worked. */ \ |
| 291 | used = CNS11643_2_set; \ |
| 292 | else \ |
| 293 | { \ |
| 294 | /* Well, see whether we have to change the SO set. */ \ |
| 295 | if (used == GB2312_set) \ |
| 296 | written = ucs4_to_cns11643l1 (ch, buf, 2); \ |
| 297 | else \ |
| 298 | written = ucs4_to_gb2312 (ch, buf, 2); \ |
| 299 | \ |
| 300 | if (__builtin_expect (written, 0) != __UNKNOWN_10646_CHAR) \ |
| 301 | /* Oh well, then switch SO. */ \ |
| 302 | used = GB2312_set + CNS11643_1_set - used; \ |
| 303 | else \ |
| 304 | { \ |
| 305 | UNICODE_TAG_HANDLER (ch, 4); \ |
| 306 | \ |
| 307 | /* Even this does not work. Error. */ \ |
| 308 | STANDARD_TO_LOOP_ERR_HANDLER (4); \ |
| 309 | } \ |
| 310 | } \ |
| 311 | } \ |
| 312 | assert (written == 2); \ |
| 313 | \ |
| 314 | /* See whether we have to emit an escape sequence. */ \ |
| 315 | if (set != used) \ |
| 316 | { \ |
| 317 | /* First see whether we announced that we use this \ |
| 318 | character set. */ \ |
| 319 | if ((ann & (16 << (used >> 3))) == 0) \ |
| 320 | { \ |
| 321 | const char *escseq; \ |
| 322 | \ |
| 323 | if (__glibc_unlikely (outptr + 4 > outend)) \ |
| 324 | { \ |
| 325 | result = __GCONV_FULL_OUTPUT; \ |
| 326 | break; \ |
| 327 | } \ |
| 328 | \ |
| 329 | assert ((used >> 3) >= 1 && (used >> 3) <= 3); \ |
| 330 | escseq = ")A)G*H" + ((used >> 3) - 1) * 2; \ |
| 331 | *outptr++ = ESC; \ |
| 332 | *outptr++ = '$'; \ |
| 333 | *outptr++ = *escseq++; \ |
| 334 | *outptr++ = *escseq++; \ |
| 335 | \ |
| 336 | if (used == GB2312_set) \ |
| 337 | ann = (ann & CNS11643_2_ann) | GB2312_ann; \ |
| 338 | else if (used == CNS11643_1_set) \ |
| 339 | ann = (ann & CNS11643_2_ann) | CNS11643_1_ann; \ |
| 340 | else \ |
| 341 | ann |= CNS11643_2_ann; \ |
| 342 | } \ |
| 343 | \ |
| 344 | if (used == CNS11643_2_set) \ |
| 345 | { \ |
| 346 | if (__glibc_unlikely (outptr + 2 > outend)) \ |
| 347 | { \ |
| 348 | result = __GCONV_FULL_OUTPUT; \ |
| 349 | break; \ |
| 350 | } \ |
| 351 | *outptr++ = SS2_0; \ |
| 352 | *outptr++ = SS2_1; \ |
| 353 | } \ |
| 354 | else \ |
| 355 | { \ |
| 356 | /* We only have to emit something is currently ASCII is \ |
| 357 | selected. Otherwise we are switching within the \ |
| 358 | SO charset. */ \ |
| 359 | if (set == ASCII_set) \ |
| 360 | { \ |
| 361 | if (__glibc_unlikely (outptr + 1 > outend)) \ |
| 362 | { \ |
| 363 | result = __GCONV_FULL_OUTPUT; \ |
| 364 | break; \ |
| 365 | } \ |
| 366 | *outptr++ = SO; \ |
| 367 | } \ |
| 368 | } \ |
| 369 | \ |
| 370 | /* Always test the length here since we have used up all the \ |
| 371 | guaranteed output buffer slots. */ \ |
| 372 | if (__glibc_unlikely (outptr + 2 > outend)) \ |
| 373 | { \ |
| 374 | result = __GCONV_FULL_OUTPUT; \ |
| 375 | break; \ |
| 376 | } \ |
| 377 | } \ |
| 378 | else if (__glibc_unlikely (outptr + 2 > outend)) \ |
| 379 | { \ |
| 380 | result = __GCONV_FULL_OUTPUT; \ |
| 381 | break; \ |
| 382 | } \ |
| 383 | \ |
| 384 | *outptr++ = buf[0]; \ |
| 385 | *outptr++ = buf[1]; \ |
| 386 | set = used; \ |
| 387 | } \ |
| 388 | \ |
| 389 | /* Now that we wrote the output increment the input pointer. */ \ |
| 390 | inptr += 4; \ |
| 391 | } |
| 392 | #define LOOP_NEED_FLAGS |
| 393 | #define EXTRA_LOOP_DECLS , int *setp |
| 394 | #define INIT_PARAMS int set = *setp & CURRENT_SEL_MASK; \ |
| 395 | int ann = *setp & CURRENT_ANN_MASK |
| 396 | #define REINIT_PARAMS do \ |
| 397 | { \ |
| 398 | set = *setp & CURRENT_SEL_MASK; \ |
| 399 | ann = *setp & CURRENT_ANN_MASK; \ |
| 400 | } \ |
| 401 | while (0) |
| 402 | #define UPDATE_PARAMS *setp = set | ann |
| 403 | #include <iconv/loop.c> |
| 404 | |
| 405 | |
| 406 | /* Now define the toplevel functions. */ |
| 407 | #include <iconv/skeleton.c> |