1219019Sgabor/* $FreeBSD$ */ 2219019Sgabor/* $NetBSD: citrus_gbk2k.c,v 1.7 2008/06/14 16:01:07 tnozaki Exp $ */ 3219019Sgabor 4219019Sgabor/*- 5219019Sgabor * Copyright (c)2003 Citrus Project, 6219019Sgabor * All rights reserved. 7219019Sgabor * 8219019Sgabor * Redistribution and use in source and binary forms, with or without 9219019Sgabor * modification, are permitted provided that the following conditions 10219019Sgabor * are met: 11219019Sgabor * 1. Redistributions of source code must retain the above copyright 12219019Sgabor * notice, this list of conditions and the following disclaimer. 13219019Sgabor * 2. Redistributions in binary form must reproduce the above copyright 14219019Sgabor * notice, this list of conditions and the following disclaimer in the 15219019Sgabor * documentation and/or other materials provided with the distribution. 16219019Sgabor * 17219019Sgabor * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18219019Sgabor * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19219019Sgabor * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20219019Sgabor * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 21219019Sgabor * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22219019Sgabor * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23219019Sgabor * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24219019Sgabor * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25219019Sgabor * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26219019Sgabor * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27219019Sgabor * SUCH DAMAGE. 28219019Sgabor */ 29219019Sgabor 30219019Sgabor#include <sys/cdefs.h> 31219019Sgabor#include <sys/types.h> 32219019Sgabor 33219019Sgabor#include <assert.h> 34219019Sgabor#include <errno.h> 35219019Sgabor#include <limits.h> 36219019Sgabor#include <stdbool.h> 37219019Sgabor#include <stddef.h> 38219019Sgabor#include <stdio.h> 39219019Sgabor#include <stdlib.h> 40219019Sgabor#include <string.h> 41219019Sgabor#include <wchar.h> 42219019Sgabor 43219019Sgabor#include "citrus_namespace.h" 44219019Sgabor#include "citrus_types.h" 45219019Sgabor#include "citrus_bcs.h" 46219019Sgabor#include "citrus_module.h" 47219019Sgabor#include "citrus_stdenc.h" 48219019Sgabor#include "citrus_gbk2k.h" 49219019Sgabor 50219019Sgabor 51219019Sgabor/* ---------------------------------------------------------------------- 52219019Sgabor * private stuffs used by templates 53219019Sgabor */ 54219019Sgabor 55219019Sgabortypedef struct _GBK2KState { 56219019Sgabor int chlen; 57219019Sgabor char ch[4]; 58219019Sgabor} _GBK2KState; 59219019Sgabor 60219019Sgabortypedef struct { 61219019Sgabor int mb_cur_max; 62219019Sgabor} _GBK2KEncodingInfo; 63219019Sgabor 64219019Sgabor#define _CEI_TO_EI(_cei_) (&(_cei_)->ei) 65219019Sgabor#define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_ 66219019Sgabor 67219019Sgabor#define _FUNCNAME(m) _citrus_GBK2K_##m 68219019Sgabor#define _ENCODING_INFO _GBK2KEncodingInfo 69219019Sgabor#define _ENCODING_STATE _GBK2KState 70219019Sgabor#define _ENCODING_MB_CUR_MAX(_ei_) (_ei_)->mb_cur_max 71219019Sgabor#define _ENCODING_IS_STATE_DEPENDENT 0 72219019Sgabor#define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0 73219019Sgabor 74219019Sgaborstatic __inline void 75219019Sgabor/*ARGSUSED*/ 76219019Sgabor_citrus_GBK2K_init_state(_GBK2KEncodingInfo * __restrict ei __unused, 77219019Sgabor _GBK2KState * __restrict s) 78219019Sgabor{ 79219019Sgabor 80219019Sgabor memset(s, 0, sizeof(*s)); 81219019Sgabor} 82219019Sgabor 83219019Sgaborstatic __inline void 84219019Sgabor/*ARGSUSED*/ 85219019Sgabor_citrus_GBK2K_pack_state(_GBK2KEncodingInfo * __restrict ei __unused, 86219019Sgabor void * __restrict pspriv, const _GBK2KState * __restrict s) 87219019Sgabor{ 88219019Sgabor 89219019Sgabor memcpy(pspriv, (const void *)s, sizeof(*s)); 90219019Sgabor} 91219019Sgabor 92219019Sgaborstatic __inline void 93219019Sgabor/*ARGSUSED*/ 94219019Sgabor_citrus_GBK2K_unpack_state(_GBK2KEncodingInfo * __restrict ei __unused, 95219019Sgabor _GBK2KState * __restrict s, const void * __restrict pspriv) 96219019Sgabor{ 97219019Sgabor 98219019Sgabor memcpy((void *)s, pspriv, sizeof(*s)); 99219019Sgabor} 100219019Sgabor 101219019Sgaborstatic __inline bool 102219019Sgabor_mb_singlebyte(int c) 103219019Sgabor{ 104219019Sgabor 105219019Sgabor return ((c & 0xff) <= 0x7f); 106219019Sgabor} 107219019Sgabor 108219019Sgaborstatic __inline bool 109219019Sgabor_mb_leadbyte(int c) 110219019Sgabor{ 111219019Sgabor 112219019Sgabor c &= 0xff; 113219019Sgabor return (0x81 <= c && c <= 0xfe); 114219019Sgabor} 115219019Sgabor 116219019Sgaborstatic __inline bool 117219019Sgabor_mb_trailbyte(int c) 118219019Sgabor{ 119219019Sgabor 120219019Sgabor c &= 0xff; 121219019Sgabor return ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfe)); 122219019Sgabor} 123219019Sgabor 124219019Sgaborstatic __inline bool 125219019Sgabor_mb_surrogate(int c) 126219019Sgabor{ 127219019Sgabor 128219019Sgabor c &= 0xff; 129219019Sgabor return (0x30 <= c && c <= 0x39); 130219019Sgabor} 131219019Sgabor 132219019Sgaborstatic __inline int 133219019Sgabor_mb_count(wchar_t v) 134219019Sgabor{ 135219019Sgabor uint32_t c; 136219019Sgabor 137219019Sgabor c = (uint32_t)v; /* XXX */ 138219019Sgabor if (!(c & 0xffffff00)) 139219019Sgabor return (1); 140219019Sgabor if (!(c & 0xffff0000)) 141219019Sgabor return (2); 142219019Sgabor return (4); 143219019Sgabor} 144219019Sgabor 145219019Sgabor#define _PSENC (psenc->ch[psenc->chlen - 1]) 146219019Sgabor#define _PUSH_PSENC(c) (psenc->ch[psenc->chlen++] = (c)) 147219019Sgabor 148219019Sgaborstatic int 149219019Sgabor_citrus_GBK2K_mbrtowc_priv(_GBK2KEncodingInfo * __restrict ei, 150252583Speter wchar_t * __restrict pwc, const char ** __restrict s, size_t n, 151219019Sgabor _GBK2KState * __restrict psenc, size_t * __restrict nresult) 152219019Sgabor{ 153252583Speter const char *s0, *s1; 154219019Sgabor wchar_t wc; 155219019Sgabor int chlenbak, len; 156219019Sgabor 157219019Sgabor s0 = *s; 158219019Sgabor 159219019Sgabor if (s0 == NULL) { 160219019Sgabor /* _citrus_GBK2K_init_state(ei, psenc); */ 161219019Sgabor psenc->chlen = 0; 162219019Sgabor *nresult = 0; 163219019Sgabor return (0); 164219019Sgabor } 165219019Sgabor 166219019Sgabor chlenbak = psenc->chlen; 167219019Sgabor 168219019Sgabor switch (psenc->chlen) { 169219019Sgabor case 3: 170219019Sgabor if (!_mb_leadbyte (_PSENC)) 171219019Sgabor goto invalid; 172219019Sgabor /* FALLTHROUGH */ 173219019Sgabor case 2: 174219019Sgabor if (!_mb_surrogate(_PSENC) || _mb_trailbyte(_PSENC)) 175219019Sgabor goto invalid; 176219019Sgabor /* FALLTHROUGH */ 177219019Sgabor case 1: 178219019Sgabor if (!_mb_leadbyte (_PSENC)) 179219019Sgabor goto invalid; 180219019Sgabor /* FALLTHOROUGH */ 181219019Sgabor case 0: 182219019Sgabor break; 183219019Sgabor default: 184219019Sgabor goto invalid; 185219019Sgabor } 186219019Sgabor 187219019Sgabor for (;;) { 188219019Sgabor if (n-- < 1) 189219019Sgabor goto restart; 190219019Sgabor 191219019Sgabor _PUSH_PSENC(*s0++); 192219019Sgabor 193219019Sgabor switch (psenc->chlen) { 194219019Sgabor case 1: 195219019Sgabor if (_mb_singlebyte(_PSENC)) 196219019Sgabor goto convert; 197219019Sgabor if (_mb_leadbyte (_PSENC)) 198219019Sgabor continue; 199219019Sgabor goto ilseq; 200219019Sgabor case 2: 201219019Sgabor if (_mb_trailbyte (_PSENC)) 202219019Sgabor goto convert; 203219019Sgabor if (ei->mb_cur_max == 4 && 204219019Sgabor _mb_surrogate (_PSENC)) 205219019Sgabor continue; 206219019Sgabor goto ilseq; 207219019Sgabor case 3: 208219019Sgabor if (_mb_leadbyte (_PSENC)) 209219019Sgabor continue; 210219019Sgabor goto ilseq; 211219019Sgabor case 4: 212219019Sgabor if (_mb_surrogate (_PSENC)) 213219019Sgabor goto convert; 214219019Sgabor goto ilseq; 215219019Sgabor } 216219019Sgabor } 217219019Sgabor 218219019Sgaborconvert: 219219019Sgabor len = psenc->chlen; 220219019Sgabor s1 = &psenc->ch[0]; 221219019Sgabor wc = 0; 222219019Sgabor while (len-- > 0) 223219019Sgabor wc = (wc << 8) | (*s1++ & 0xff); 224219019Sgabor 225219019Sgabor if (pwc != NULL) 226219019Sgabor *pwc = wc; 227219019Sgabor *s = s0; 228219019Sgabor *nresult = (wc == 0) ? 0 : psenc->chlen - chlenbak; 229219019Sgabor /* _citrus_GBK2K_init_state(ei, psenc); */ 230219019Sgabor psenc->chlen = 0; 231219019Sgabor 232219019Sgabor return (0); 233219019Sgabor 234219019Sgaborrestart: 235219019Sgabor *s = s0; 236219019Sgabor *nresult = (size_t)-2; 237219019Sgabor 238219019Sgabor return (0); 239219019Sgabor 240219019Sgaborinvalid: 241219019Sgabor return (EINVAL); 242219019Sgabor 243219019Sgaborilseq: 244219019Sgabor *nresult = (size_t)-1; 245219019Sgabor return (EILSEQ); 246219019Sgabor} 247219019Sgabor 248219019Sgaborstatic int 249219019Sgabor_citrus_GBK2K_wcrtomb_priv(_GBK2KEncodingInfo * __restrict ei, 250219019Sgabor char * __restrict s, size_t n, wchar_t wc, _GBK2KState * __restrict psenc, 251219019Sgabor size_t * __restrict nresult) 252219019Sgabor{ 253219019Sgabor size_t len; 254219019Sgabor int ret; 255219019Sgabor 256219019Sgabor if (psenc->chlen != 0) { 257219019Sgabor ret = EINVAL; 258219019Sgabor goto err; 259219019Sgabor } 260219019Sgabor 261219019Sgabor len = _mb_count(wc); 262219019Sgabor if (n < len) { 263219019Sgabor ret = E2BIG; 264219019Sgabor goto err; 265219019Sgabor } 266219019Sgabor 267219019Sgabor switch (len) { 268219019Sgabor case 1: 269219019Sgabor if (!_mb_singlebyte(_PUSH_PSENC(wc ))) { 270219019Sgabor ret = EILSEQ; 271219019Sgabor goto err; 272219019Sgabor } 273219019Sgabor break; 274219019Sgabor case 2: 275219019Sgabor if (!_mb_leadbyte (_PUSH_PSENC(wc >> 8)) || 276219019Sgabor !_mb_trailbyte (_PUSH_PSENC(wc))) { 277219019Sgabor ret = EILSEQ; 278219019Sgabor goto err; 279219019Sgabor } 280219019Sgabor break; 281219019Sgabor case 4: 282219019Sgabor if (ei->mb_cur_max != 4 || 283219019Sgabor !_mb_leadbyte (_PUSH_PSENC(wc >> 24)) || 284219019Sgabor !_mb_surrogate (_PUSH_PSENC(wc >> 16)) || 285219019Sgabor !_mb_leadbyte (_PUSH_PSENC(wc >> 8)) || 286219019Sgabor !_mb_surrogate (_PUSH_PSENC(wc))) { 287219019Sgabor ret = EILSEQ; 288219019Sgabor goto err; 289219019Sgabor } 290219019Sgabor break; 291219019Sgabor } 292219019Sgabor 293219019Sgabor memcpy(s, psenc->ch, psenc->chlen); 294219019Sgabor *nresult = psenc->chlen; 295219019Sgabor /* _citrus_GBK2K_init_state(ei, psenc); */ 296219019Sgabor psenc->chlen = 0; 297219019Sgabor 298219019Sgabor return (0); 299219019Sgabor 300219019Sgaborerr: 301219019Sgabor *nresult = (size_t)-1; 302219019Sgabor return (ret); 303219019Sgabor} 304219019Sgabor 305219019Sgaborstatic __inline int 306219019Sgabor/*ARGSUSED*/ 307219019Sgabor_citrus_GBK2K_stdenc_wctocs(_GBK2KEncodingInfo * __restrict ei __unused, 308219019Sgabor _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc) 309219019Sgabor{ 310219019Sgabor uint8_t ch, cl; 311219019Sgabor 312219019Sgabor if ((uint32_t)wc < 0x80) { 313219019Sgabor /* ISO646 */ 314219019Sgabor *csid = 0; 315219019Sgabor *idx = (_index_t)wc; 316219019Sgabor } else if ((uint32_t)wc >= 0x10000) { 317219019Sgabor /* GBKUCS : XXX */ 318219019Sgabor *csid = 3; 319219019Sgabor *idx = (_index_t)wc; 320219019Sgabor } else { 321219019Sgabor ch = (uint8_t)(wc >> 8); 322219019Sgabor cl = (uint8_t)wc; 323219019Sgabor if (ch >= 0xA1 && cl >= 0xA1) { 324219019Sgabor /* EUC G1 */ 325219019Sgabor *csid = 1; 326219019Sgabor *idx = (_index_t)wc & 0x7F7FU; 327219019Sgabor } else { 328219019Sgabor /* extended area (0x8140-) */ 329219019Sgabor *csid = 2; 330219019Sgabor *idx = (_index_t)wc; 331219019Sgabor } 332219019Sgabor } 333219019Sgabor 334219019Sgabor return (0); 335219019Sgabor} 336219019Sgabor 337219019Sgaborstatic __inline int 338219019Sgabor/*ARGSUSED*/ 339219019Sgabor_citrus_GBK2K_stdenc_cstowc(_GBK2KEncodingInfo * __restrict ei, 340219019Sgabor wchar_t * __restrict wc, _csid_t csid, _index_t idx) 341219019Sgabor{ 342219019Sgabor 343219019Sgabor switch (csid) { 344219019Sgabor case 0: 345219019Sgabor /* ISO646 */ 346219019Sgabor *wc = (wchar_t)idx; 347219019Sgabor break; 348219019Sgabor case 1: 349219019Sgabor /* EUC G1 */ 350219019Sgabor *wc = (wchar_t)idx | 0x8080U; 351219019Sgabor break; 352219019Sgabor case 2: 353219019Sgabor /* extended area */ 354219019Sgabor *wc = (wchar_t)idx; 355219019Sgabor break; 356219019Sgabor case 3: 357219019Sgabor /* GBKUCS : XXX */ 358219019Sgabor if (ei->mb_cur_max != 4) 359219019Sgabor return (EINVAL); 360219019Sgabor *wc = (wchar_t)idx; 361219019Sgabor break; 362219019Sgabor default: 363219019Sgabor return (EILSEQ); 364219019Sgabor } 365219019Sgabor 366219019Sgabor return (0); 367219019Sgabor} 368219019Sgabor 369219019Sgaborstatic __inline int 370219019Sgabor/*ARGSUSED*/ 371219019Sgabor_citrus_GBK2K_stdenc_get_state_desc_generic(_GBK2KEncodingInfo * __restrict ei __unused, 372219019Sgabor _GBK2KState * __restrict psenc, int * __restrict rstate) 373219019Sgabor{ 374219019Sgabor 375219019Sgabor *rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL : 376219019Sgabor _STDENC_SDGEN_INCOMPLETE_CHAR; 377219019Sgabor return (0); 378219019Sgabor} 379219019Sgabor 380219019Sgaborstatic int 381219019Sgabor/*ARGSUSED*/ 382219019Sgabor_citrus_GBK2K_encoding_module_init(_GBK2KEncodingInfo * __restrict ei, 383219019Sgabor const void * __restrict var, size_t lenvar) 384219019Sgabor{ 385219019Sgabor const char *p; 386219019Sgabor 387219019Sgabor p = var; 388219019Sgabor memset((void *)ei, 0, sizeof(*ei)); 389219019Sgabor ei->mb_cur_max = 4; 390219019Sgabor while (lenvar > 0) { 391219019Sgabor switch (_bcs_tolower(*p)) { 392219019Sgabor case '2': 393219019Sgabor MATCH("2byte", ei->mb_cur_max = 2); 394219019Sgabor break; 395219019Sgabor } 396219019Sgabor p++; 397219019Sgabor lenvar--; 398219019Sgabor } 399219019Sgabor 400219019Sgabor return (0); 401219019Sgabor} 402219019Sgabor 403219019Sgaborstatic void 404219019Sgabor/*ARGSUSED*/ 405219019Sgabor_citrus_GBK2K_encoding_module_uninit(_GBK2KEncodingInfo *ei __unused) 406219019Sgabor{ 407219019Sgabor 408219019Sgabor} 409219019Sgabor 410219019Sgabor/* ---------------------------------------------------------------------- 411219019Sgabor * public interface for stdenc 412219019Sgabor */ 413219019Sgabor 414219019Sgabor_CITRUS_STDENC_DECLS(GBK2K); 415219019Sgabor_CITRUS_STDENC_DEF_OPS(GBK2K); 416219019Sgabor 417219019Sgabor#include "citrus_stdenc_template.h" 418