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 83260264Sdim#if 0 84219019Sgaborstatic __inline void 85219019Sgabor/*ARGSUSED*/ 86219019Sgabor_citrus_GBK2K_pack_state(_GBK2KEncodingInfo * __restrict ei __unused, 87219019Sgabor void * __restrict pspriv, const _GBK2KState * __restrict s) 88219019Sgabor{ 89219019Sgabor 90219019Sgabor memcpy(pspriv, (const void *)s, sizeof(*s)); 91219019Sgabor} 92219019Sgabor 93219019Sgaborstatic __inline void 94219019Sgabor/*ARGSUSED*/ 95219019Sgabor_citrus_GBK2K_unpack_state(_GBK2KEncodingInfo * __restrict ei __unused, 96219019Sgabor _GBK2KState * __restrict s, const void * __restrict pspriv) 97219019Sgabor{ 98219019Sgabor 99219019Sgabor memcpy((void *)s, pspriv, sizeof(*s)); 100219019Sgabor} 101260264Sdim#endif 102219019Sgabor 103219019Sgaborstatic __inline bool 104219019Sgabor_mb_singlebyte(int c) 105219019Sgabor{ 106219019Sgabor 107219019Sgabor return ((c & 0xff) <= 0x7f); 108219019Sgabor} 109219019Sgabor 110219019Sgaborstatic __inline bool 111219019Sgabor_mb_leadbyte(int c) 112219019Sgabor{ 113219019Sgabor 114219019Sgabor c &= 0xff; 115219019Sgabor return (0x81 <= c && c <= 0xfe); 116219019Sgabor} 117219019Sgabor 118219019Sgaborstatic __inline bool 119219019Sgabor_mb_trailbyte(int c) 120219019Sgabor{ 121219019Sgabor 122219019Sgabor c &= 0xff; 123219019Sgabor return ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfe)); 124219019Sgabor} 125219019Sgabor 126219019Sgaborstatic __inline bool 127219019Sgabor_mb_surrogate(int c) 128219019Sgabor{ 129219019Sgabor 130219019Sgabor c &= 0xff; 131219019Sgabor return (0x30 <= c && c <= 0x39); 132219019Sgabor} 133219019Sgabor 134219019Sgaborstatic __inline int 135219019Sgabor_mb_count(wchar_t v) 136219019Sgabor{ 137219019Sgabor uint32_t c; 138219019Sgabor 139219019Sgabor c = (uint32_t)v; /* XXX */ 140219019Sgabor if (!(c & 0xffffff00)) 141219019Sgabor return (1); 142219019Sgabor if (!(c & 0xffff0000)) 143219019Sgabor return (2); 144219019Sgabor return (4); 145219019Sgabor} 146219019Sgabor 147219019Sgabor#define _PSENC (psenc->ch[psenc->chlen - 1]) 148219019Sgabor#define _PUSH_PSENC(c) (psenc->ch[psenc->chlen++] = (c)) 149219019Sgabor 150219019Sgaborstatic int 151219019Sgabor_citrus_GBK2K_mbrtowc_priv(_GBK2KEncodingInfo * __restrict ei, 152282275Stijl wchar_t * __restrict pwc, char ** __restrict s, size_t n, 153219019Sgabor _GBK2KState * __restrict psenc, size_t * __restrict nresult) 154219019Sgabor{ 155282275Stijl char *s0, *s1; 156219019Sgabor wchar_t wc; 157219019Sgabor int chlenbak, len; 158219019Sgabor 159219019Sgabor s0 = *s; 160219019Sgabor 161219019Sgabor if (s0 == NULL) { 162219019Sgabor /* _citrus_GBK2K_init_state(ei, psenc); */ 163219019Sgabor psenc->chlen = 0; 164219019Sgabor *nresult = 0; 165219019Sgabor return (0); 166219019Sgabor } 167219019Sgabor 168219019Sgabor chlenbak = psenc->chlen; 169219019Sgabor 170219019Sgabor switch (psenc->chlen) { 171219019Sgabor case 3: 172219019Sgabor if (!_mb_leadbyte (_PSENC)) 173219019Sgabor goto invalid; 174219019Sgabor /* FALLTHROUGH */ 175219019Sgabor case 2: 176219019Sgabor if (!_mb_surrogate(_PSENC) || _mb_trailbyte(_PSENC)) 177219019Sgabor goto invalid; 178219019Sgabor /* FALLTHROUGH */ 179219019Sgabor case 1: 180219019Sgabor if (!_mb_leadbyte (_PSENC)) 181219019Sgabor goto invalid; 182219019Sgabor /* FALLTHOROUGH */ 183219019Sgabor case 0: 184219019Sgabor break; 185219019Sgabor default: 186219019Sgabor goto invalid; 187219019Sgabor } 188219019Sgabor 189219019Sgabor for (;;) { 190219019Sgabor if (n-- < 1) 191219019Sgabor goto restart; 192219019Sgabor 193219019Sgabor _PUSH_PSENC(*s0++); 194219019Sgabor 195219019Sgabor switch (psenc->chlen) { 196219019Sgabor case 1: 197219019Sgabor if (_mb_singlebyte(_PSENC)) 198219019Sgabor goto convert; 199219019Sgabor if (_mb_leadbyte (_PSENC)) 200219019Sgabor continue; 201219019Sgabor goto ilseq; 202219019Sgabor case 2: 203219019Sgabor if (_mb_trailbyte (_PSENC)) 204219019Sgabor goto convert; 205219019Sgabor if (ei->mb_cur_max == 4 && 206219019Sgabor _mb_surrogate (_PSENC)) 207219019Sgabor continue; 208219019Sgabor goto ilseq; 209219019Sgabor case 3: 210219019Sgabor if (_mb_leadbyte (_PSENC)) 211219019Sgabor continue; 212219019Sgabor goto ilseq; 213219019Sgabor case 4: 214219019Sgabor if (_mb_surrogate (_PSENC)) 215219019Sgabor goto convert; 216219019Sgabor goto ilseq; 217219019Sgabor } 218219019Sgabor } 219219019Sgabor 220219019Sgaborconvert: 221219019Sgabor len = psenc->chlen; 222219019Sgabor s1 = &psenc->ch[0]; 223219019Sgabor wc = 0; 224219019Sgabor while (len-- > 0) 225219019Sgabor wc = (wc << 8) | (*s1++ & 0xff); 226219019Sgabor 227219019Sgabor if (pwc != NULL) 228219019Sgabor *pwc = wc; 229219019Sgabor *s = s0; 230219019Sgabor *nresult = (wc == 0) ? 0 : psenc->chlen - chlenbak; 231219019Sgabor /* _citrus_GBK2K_init_state(ei, psenc); */ 232219019Sgabor psenc->chlen = 0; 233219019Sgabor 234219019Sgabor return (0); 235219019Sgabor 236219019Sgaborrestart: 237219019Sgabor *s = s0; 238219019Sgabor *nresult = (size_t)-2; 239219019Sgabor 240219019Sgabor return (0); 241219019Sgabor 242219019Sgaborinvalid: 243219019Sgabor return (EINVAL); 244219019Sgabor 245219019Sgaborilseq: 246219019Sgabor *nresult = (size_t)-1; 247219019Sgabor return (EILSEQ); 248219019Sgabor} 249219019Sgabor 250219019Sgaborstatic int 251219019Sgabor_citrus_GBK2K_wcrtomb_priv(_GBK2KEncodingInfo * __restrict ei, 252219019Sgabor char * __restrict s, size_t n, wchar_t wc, _GBK2KState * __restrict psenc, 253219019Sgabor size_t * __restrict nresult) 254219019Sgabor{ 255219019Sgabor size_t len; 256219019Sgabor int ret; 257219019Sgabor 258219019Sgabor if (psenc->chlen != 0) { 259219019Sgabor ret = EINVAL; 260219019Sgabor goto err; 261219019Sgabor } 262219019Sgabor 263219019Sgabor len = _mb_count(wc); 264219019Sgabor if (n < len) { 265219019Sgabor ret = E2BIG; 266219019Sgabor goto err; 267219019Sgabor } 268219019Sgabor 269219019Sgabor switch (len) { 270219019Sgabor case 1: 271219019Sgabor if (!_mb_singlebyte(_PUSH_PSENC(wc ))) { 272219019Sgabor ret = EILSEQ; 273219019Sgabor goto err; 274219019Sgabor } 275219019Sgabor break; 276219019Sgabor case 2: 277219019Sgabor if (!_mb_leadbyte (_PUSH_PSENC(wc >> 8)) || 278219019Sgabor !_mb_trailbyte (_PUSH_PSENC(wc))) { 279219019Sgabor ret = EILSEQ; 280219019Sgabor goto err; 281219019Sgabor } 282219019Sgabor break; 283219019Sgabor case 4: 284219019Sgabor if (ei->mb_cur_max != 4 || 285219019Sgabor !_mb_leadbyte (_PUSH_PSENC(wc >> 24)) || 286219019Sgabor !_mb_surrogate (_PUSH_PSENC(wc >> 16)) || 287219019Sgabor !_mb_leadbyte (_PUSH_PSENC(wc >> 8)) || 288219019Sgabor !_mb_surrogate (_PUSH_PSENC(wc))) { 289219019Sgabor ret = EILSEQ; 290219019Sgabor goto err; 291219019Sgabor } 292219019Sgabor break; 293219019Sgabor } 294219019Sgabor 295219019Sgabor memcpy(s, psenc->ch, psenc->chlen); 296219019Sgabor *nresult = psenc->chlen; 297219019Sgabor /* _citrus_GBK2K_init_state(ei, psenc); */ 298219019Sgabor psenc->chlen = 0; 299219019Sgabor 300219019Sgabor return (0); 301219019Sgabor 302219019Sgaborerr: 303219019Sgabor *nresult = (size_t)-1; 304219019Sgabor return (ret); 305219019Sgabor} 306219019Sgabor 307219019Sgaborstatic __inline int 308219019Sgabor/*ARGSUSED*/ 309219019Sgabor_citrus_GBK2K_stdenc_wctocs(_GBK2KEncodingInfo * __restrict ei __unused, 310219019Sgabor _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc) 311219019Sgabor{ 312219019Sgabor uint8_t ch, cl; 313219019Sgabor 314219019Sgabor if ((uint32_t)wc < 0x80) { 315219019Sgabor /* ISO646 */ 316219019Sgabor *csid = 0; 317219019Sgabor *idx = (_index_t)wc; 318219019Sgabor } else if ((uint32_t)wc >= 0x10000) { 319219019Sgabor /* GBKUCS : XXX */ 320219019Sgabor *csid = 3; 321219019Sgabor *idx = (_index_t)wc; 322219019Sgabor } else { 323219019Sgabor ch = (uint8_t)(wc >> 8); 324219019Sgabor cl = (uint8_t)wc; 325219019Sgabor if (ch >= 0xA1 && cl >= 0xA1) { 326219019Sgabor /* EUC G1 */ 327219019Sgabor *csid = 1; 328219019Sgabor *idx = (_index_t)wc & 0x7F7FU; 329219019Sgabor } else { 330219019Sgabor /* extended area (0x8140-) */ 331219019Sgabor *csid = 2; 332219019Sgabor *idx = (_index_t)wc; 333219019Sgabor } 334219019Sgabor } 335219019Sgabor 336219019Sgabor return (0); 337219019Sgabor} 338219019Sgabor 339219019Sgaborstatic __inline int 340219019Sgabor/*ARGSUSED*/ 341219019Sgabor_citrus_GBK2K_stdenc_cstowc(_GBK2KEncodingInfo * __restrict ei, 342219019Sgabor wchar_t * __restrict wc, _csid_t csid, _index_t idx) 343219019Sgabor{ 344219019Sgabor 345219019Sgabor switch (csid) { 346219019Sgabor case 0: 347219019Sgabor /* ISO646 */ 348219019Sgabor *wc = (wchar_t)idx; 349219019Sgabor break; 350219019Sgabor case 1: 351219019Sgabor /* EUC G1 */ 352219019Sgabor *wc = (wchar_t)idx | 0x8080U; 353219019Sgabor break; 354219019Sgabor case 2: 355219019Sgabor /* extended area */ 356219019Sgabor *wc = (wchar_t)idx; 357219019Sgabor break; 358219019Sgabor case 3: 359219019Sgabor /* GBKUCS : XXX */ 360219019Sgabor if (ei->mb_cur_max != 4) 361219019Sgabor return (EINVAL); 362219019Sgabor *wc = (wchar_t)idx; 363219019Sgabor break; 364219019Sgabor default: 365219019Sgabor return (EILSEQ); 366219019Sgabor } 367219019Sgabor 368219019Sgabor return (0); 369219019Sgabor} 370219019Sgabor 371219019Sgaborstatic __inline int 372219019Sgabor/*ARGSUSED*/ 373219019Sgabor_citrus_GBK2K_stdenc_get_state_desc_generic(_GBK2KEncodingInfo * __restrict ei __unused, 374219019Sgabor _GBK2KState * __restrict psenc, int * __restrict rstate) 375219019Sgabor{ 376219019Sgabor 377219019Sgabor *rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL : 378219019Sgabor _STDENC_SDGEN_INCOMPLETE_CHAR; 379219019Sgabor return (0); 380219019Sgabor} 381219019Sgabor 382219019Sgaborstatic int 383219019Sgabor/*ARGSUSED*/ 384219019Sgabor_citrus_GBK2K_encoding_module_init(_GBK2KEncodingInfo * __restrict ei, 385219019Sgabor const void * __restrict var, size_t lenvar) 386219019Sgabor{ 387219019Sgabor const char *p; 388219019Sgabor 389219019Sgabor p = var; 390219019Sgabor memset((void *)ei, 0, sizeof(*ei)); 391219019Sgabor ei->mb_cur_max = 4; 392219019Sgabor while (lenvar > 0) { 393219019Sgabor switch (_bcs_tolower(*p)) { 394219019Sgabor case '2': 395219019Sgabor MATCH("2byte", ei->mb_cur_max = 2); 396219019Sgabor break; 397219019Sgabor } 398219019Sgabor p++; 399219019Sgabor lenvar--; 400219019Sgabor } 401219019Sgabor 402219019Sgabor return (0); 403219019Sgabor} 404219019Sgabor 405219019Sgaborstatic void 406219019Sgabor/*ARGSUSED*/ 407219019Sgabor_citrus_GBK2K_encoding_module_uninit(_GBK2KEncodingInfo *ei __unused) 408219019Sgabor{ 409219019Sgabor 410219019Sgabor} 411219019Sgabor 412219019Sgabor/* ---------------------------------------------------------------------- 413219019Sgabor * public interface for stdenc 414219019Sgabor */ 415219019Sgabor 416219019Sgabor_CITRUS_STDENC_DECLS(GBK2K); 417219019Sgabor_CITRUS_STDENC_DEF_OPS(GBK2K); 418219019Sgabor 419219019Sgabor#include "citrus_stdenc_template.h" 420