1219019Sgabor/* $FreeBSD$ */ 2219019Sgabor/* $NetBSD: citrus_dechanyu.c,v 1.3 2008/06/14 16:01:07 tnozaki Exp $ */ 3219019Sgabor 4219019Sgabor/*- 5219019Sgabor * Copyright (c)2007 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#include <sys/cdefs.h> 30219019Sgabor#include <sys/types.h> 31219019Sgabor 32219019Sgabor#include <assert.h> 33219019Sgabor#include <errno.h> 34219019Sgabor#include <limits.h> 35219019Sgabor#include <stddef.h> 36219019Sgabor#include <stdint.h> 37219019Sgabor#include <stdio.h> 38219019Sgabor#include <stdlib.h> 39219019Sgabor#include <string.h> 40219019Sgabor#include <wchar.h> 41219019Sgabor 42219019Sgabor#include "citrus_namespace.h" 43219019Sgabor#include "citrus_types.h" 44219019Sgabor#include "citrus_bcs.h" 45219019Sgabor#include "citrus_module.h" 46219019Sgabor#include "citrus_stdenc.h" 47219019Sgabor#include "citrus_dechanyu.h" 48219019Sgabor 49219019Sgabor/* ---------------------------------------------------------------------- 50219019Sgabor * private stuffs used by templates 51219019Sgabor */ 52219019Sgabor 53219019Sgabortypedef struct { 54219019Sgabor size_t chlen; 55219019Sgabor char ch[4]; 56219019Sgabor} _DECHanyuState; 57219019Sgabor 58219019Sgabortypedef struct { 59219019Sgabor int dummy; 60219019Sgabor} _DECHanyuEncodingInfo; 61219019Sgabor 62219019Sgabor#define _CEI_TO_EI(_cei_) (&(_cei_)->ei) 63219019Sgabor#define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.__CONCAT(s_,_func_) 64219019Sgabor 65219019Sgabor#define _FUNCNAME(m) __CONCAT(_citrus_DECHanyu_,m) 66219019Sgabor#define _ENCODING_INFO _DECHanyuEncodingInfo 67219019Sgabor#define _ENCODING_STATE _DECHanyuState 68219019Sgabor#define _ENCODING_MB_CUR_MAX(_ei_) 4 69219019Sgabor#define _ENCODING_IS_STATE_DEPENDENT 0 70219019Sgabor#define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0 71219019Sgabor 72219019Sgaborstatic __inline void 73219019Sgabor/*ARGSUSED*/ 74219019Sgabor_citrus_DECHanyu_init_state(_DECHanyuEncodingInfo * __restrict ei __unused, 75219019Sgabor _DECHanyuState * __restrict psenc) 76219019Sgabor{ 77219019Sgabor 78219019Sgabor psenc->chlen = 0; 79219019Sgabor} 80219019Sgabor 81219019Sgaborstatic __inline void 82219019Sgabor/*ARGSUSED*/ 83219019Sgabor_citrus_DECHanyu_pack_state(_DECHanyuEncodingInfo * __restrict ei __unused, 84219019Sgabor void * __restrict pspriv, const _DECHanyuState * __restrict psenc) 85219019Sgabor{ 86219019Sgabor 87219019Sgabor memcpy(pspriv, (const void *)psenc, sizeof(*psenc)); 88219019Sgabor} 89219019Sgabor 90219019Sgaborstatic __inline void 91219019Sgabor/*ARGSUSED*/ 92219019Sgabor_citrus_DECHanyu_unpack_state(_DECHanyuEncodingInfo * __restrict ei __unused, 93219019Sgabor _DECHanyuState * __restrict psenc, 94219019Sgabor const void * __restrict pspriv) 95219019Sgabor{ 96219019Sgabor 97219019Sgabor memcpy((void *)psenc, pspriv, sizeof(*psenc)); 98219019Sgabor} 99219019Sgabor 100219019Sgaborstatic void 101219019Sgabor/*ARGSUSED*/ 102219019Sgabor_citrus_DECHanyu_encoding_module_uninit(_DECHanyuEncodingInfo *ei __unused) 103219019Sgabor{ 104219019Sgabor 105219019Sgabor /* ei may be null */ 106219019Sgabor} 107219019Sgabor 108219019Sgaborstatic int 109219019Sgabor/*ARGSUSED*/ 110219019Sgabor_citrus_DECHanyu_encoding_module_init(_DECHanyuEncodingInfo * __restrict ei __unused, 111219019Sgabor const void * __restrict var __unused, size_t lenvar __unused) 112219019Sgabor{ 113219019Sgabor 114219019Sgabor /* ei may be null */ 115219019Sgabor return (0); 116219019Sgabor} 117219019Sgabor 118219019Sgaborstatic __inline bool 119219019Sgaboris_singlebyte(int c) 120219019Sgabor{ 121219019Sgabor 122219019Sgabor return (c <= 0x7F); 123219019Sgabor} 124219019Sgabor 125219019Sgaborstatic __inline bool 126219019Sgaboris_leadbyte(int c) 127219019Sgabor{ 128219019Sgabor 129219019Sgabor return (c >= 0xA1 && c <= 0xFE); 130219019Sgabor} 131219019Sgabor 132219019Sgaborstatic __inline bool 133219019Sgaboris_trailbyte(int c) 134219019Sgabor{ 135219019Sgabor 136219019Sgabor c &= ~0x80; 137219019Sgabor return (c >= 0x21 && c <= 0x7E); 138219019Sgabor} 139219019Sgabor 140219019Sgaborstatic __inline bool 141219019Sgaboris_hanyu1(int c) 142219019Sgabor{ 143219019Sgabor 144219019Sgabor return (c == 0xC2); 145219019Sgabor} 146219019Sgabor 147219019Sgaborstatic __inline bool 148219019Sgaboris_hanyu2(int c) 149219019Sgabor{ 150219019Sgabor 151219019Sgabor return (c == 0xCB); 152219019Sgabor} 153219019Sgabor 154219019Sgabor#define HANYUBIT 0xC2CB0000 155219019Sgabor 156219019Sgaborstatic __inline bool 157219019Sgaboris_94charset(int c) 158219019Sgabor{ 159219019Sgabor 160219019Sgabor return (c >= 0x21 && c <= 0x7E); 161219019Sgabor} 162219019Sgabor 163219019Sgaborstatic int 164219019Sgabor/*ARGSUSED*/ 165219019Sgabor_citrus_DECHanyu_mbrtowc_priv(_DECHanyuEncodingInfo * __restrict ei, 166252583Speter wchar_t * __restrict pwc, const char ** __restrict s, size_t n, 167219019Sgabor _DECHanyuState * __restrict psenc, size_t * __restrict nresult) 168219019Sgabor{ 169252583Speter const char *s0; 170219019Sgabor wchar_t wc; 171219019Sgabor int ch; 172219019Sgabor 173219019Sgabor if (*s == NULL) { 174219019Sgabor _citrus_DECHanyu_init_state(ei, psenc); 175219019Sgabor *nresult = _ENCODING_IS_STATE_DEPENDENT; 176219019Sgabor return (0); 177219019Sgabor } 178219019Sgabor s0 = *s; 179219019Sgabor 180219019Sgabor wc = (wchar_t)0; 181219019Sgabor switch (psenc->chlen) { 182219019Sgabor case 0: 183219019Sgabor if (n-- < 1) 184219019Sgabor goto restart; 185219019Sgabor ch = *s0++ & 0xFF; 186219019Sgabor if (is_singlebyte(ch)) { 187219019Sgabor if (pwc != NULL) 188219019Sgabor *pwc = (wchar_t)ch; 189219019Sgabor *nresult = (size_t)((ch == 0) ? 0 : 1); 190219019Sgabor *s = s0; 191219019Sgabor return (0); 192219019Sgabor } 193219019Sgabor if (!is_leadbyte(ch)) 194219019Sgabor goto ilseq; 195219019Sgabor psenc->ch[psenc->chlen++] = ch; 196219019Sgabor break; 197219019Sgabor case 1: 198219019Sgabor ch = psenc->ch[0] & 0xFF; 199219019Sgabor if (!is_leadbyte(ch)) 200219019Sgabor return (EINVAL); 201219019Sgabor break; 202219019Sgabor case 2: case 3: 203219019Sgabor ch = psenc->ch[0] & 0xFF; 204219019Sgabor if (is_hanyu1(ch)) { 205219019Sgabor ch = psenc->ch[1] & 0xFF; 206219019Sgabor if (is_hanyu2(ch)) { 207219019Sgabor wc |= (wchar_t)HANYUBIT; 208219019Sgabor break; 209219019Sgabor } 210219019Sgabor } 211219019Sgabor /*FALLTHROUGH*/ 212219019Sgabor default: 213219019Sgabor return (EINVAL); 214219019Sgabor } 215219019Sgabor 216219019Sgabor switch (psenc->chlen) { 217219019Sgabor case 1: 218219019Sgabor if (is_hanyu1(ch)) { 219219019Sgabor if (n-- < 1) 220219019Sgabor goto restart; 221219019Sgabor ch = *s0++ & 0xFF; 222219019Sgabor if (!is_hanyu2(ch)) 223219019Sgabor goto ilseq; 224219019Sgabor psenc->ch[psenc->chlen++] = ch; 225219019Sgabor wc |= (wchar_t)HANYUBIT; 226219019Sgabor if (n-- < 1) 227219019Sgabor goto restart; 228219019Sgabor ch = *s0++ & 0xFF; 229219019Sgabor if (!is_leadbyte(ch)) 230219019Sgabor goto ilseq; 231219019Sgabor psenc->ch[psenc->chlen++] = ch; 232219019Sgabor } 233219019Sgabor break; 234219019Sgabor case 2: 235219019Sgabor if (n-- < 1) 236219019Sgabor goto restart; 237219019Sgabor ch = *s0++ & 0xFF; 238219019Sgabor if (!is_leadbyte(ch)) 239219019Sgabor goto ilseq; 240219019Sgabor psenc->ch[psenc->chlen++] = ch; 241219019Sgabor break; 242219019Sgabor case 3: 243219019Sgabor ch = psenc->ch[2] & 0xFF; 244219019Sgabor if (!is_leadbyte(ch)) 245219019Sgabor return (EINVAL); 246219019Sgabor } 247219019Sgabor if (n-- < 1) 248219019Sgabor goto restart; 249219019Sgabor wc |= (wchar_t)(ch << 8); 250219019Sgabor ch = *s0++ & 0xFF; 251219019Sgabor if (!is_trailbyte(ch)) 252219019Sgabor goto ilseq; 253219019Sgabor wc |= (wchar_t)ch; 254219019Sgabor if (pwc != NULL) 255219019Sgabor *pwc = wc; 256219019Sgabor *nresult = (size_t)(s0 - *s); 257219019Sgabor *s = s0; 258219019Sgabor psenc->chlen = 0; 259219019Sgabor 260219019Sgabor return (0); 261219019Sgabor 262219019Sgaborrestart: 263219019Sgabor *nresult = (size_t)-2; 264219019Sgabor *s = s0; 265219019Sgabor return (0); 266219019Sgabor 267219019Sgaborilseq: 268219019Sgabor *nresult = (size_t)-1; 269219019Sgabor return (EILSEQ); 270219019Sgabor} 271219019Sgabor 272219019Sgaborstatic int 273219019Sgabor/*ARGSUSED*/ 274219019Sgabor_citrus_DECHanyu_wcrtomb_priv(_DECHanyuEncodingInfo * __restrict ei __unused, 275219019Sgabor char * __restrict s, size_t n, wchar_t wc, 276219019Sgabor _DECHanyuState * __restrict psenc, size_t * __restrict nresult) 277219019Sgabor{ 278219019Sgabor int ch; 279219019Sgabor 280219019Sgabor if (psenc->chlen != 0) 281219019Sgabor return (EINVAL); 282219019Sgabor 283219019Sgabor /* XXX: assume wchar_t as int */ 284219019Sgabor if ((uint32_t)wc <= 0x7F) { 285219019Sgabor ch = wc & 0xFF; 286219019Sgabor } else { 287219019Sgabor if ((uint32_t)wc > 0xFFFF) { 288219019Sgabor if ((wc & ~0xFFFF) != (wchar_t)HANYUBIT) 289219019Sgabor goto ilseq; 290219019Sgabor psenc->ch[psenc->chlen++] = (wc >> 24) & 0xFF; 291219019Sgabor psenc->ch[psenc->chlen++] = (wc >> 16) & 0xFF; 292219019Sgabor wc &= 0xFFFF; 293219019Sgabor } 294219019Sgabor ch = (wc >> 8) & 0xFF; 295219019Sgabor if (!is_leadbyte(ch)) 296219019Sgabor goto ilseq; 297219019Sgabor psenc->ch[psenc->chlen++] = ch; 298219019Sgabor ch = wc & 0xFF; 299219019Sgabor if (!is_trailbyte(ch)) 300219019Sgabor goto ilseq; 301219019Sgabor } 302219019Sgabor psenc->ch[psenc->chlen++] = ch; 303219019Sgabor if (n < psenc->chlen) { 304219019Sgabor *nresult = (size_t)-1; 305219019Sgabor return (E2BIG); 306219019Sgabor } 307219019Sgabor memcpy(s, psenc->ch, psenc->chlen); 308219019Sgabor *nresult = psenc->chlen; 309219019Sgabor psenc->chlen = 0; 310219019Sgabor 311219019Sgabor return (0); 312219019Sgabor 313219019Sgaborilseq: 314219019Sgabor *nresult = (size_t)-1; 315219019Sgabor return (EILSEQ); 316219019Sgabor} 317219019Sgabor 318219019Sgaborstatic __inline int 319219019Sgabor/*ARGSUSED*/ 320219019Sgabor_citrus_DECHanyu_stdenc_wctocs(_DECHanyuEncodingInfo * __restrict ei __unused, 321219019Sgabor _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc) 322219019Sgabor{ 323219019Sgabor wchar_t mask; 324219019Sgabor int plane; 325219019Sgabor 326219019Sgabor plane = 0; 327219019Sgabor mask = 0x7F; 328219019Sgabor /* XXX: assume wchar_t as int */ 329219019Sgabor if ((uint32_t)wc > 0x7F) { 330219019Sgabor if ((uint32_t)wc > 0xFFFF) { 331219019Sgabor if ((wc & ~0xFFFF) != (wchar_t)HANYUBIT) 332219019Sgabor return (EILSEQ); 333219019Sgabor plane += 2; 334219019Sgabor } 335219019Sgabor if (!is_leadbyte((wc >> 8) & 0xFF) || 336219019Sgabor !is_trailbyte(wc & 0xFF)) 337219019Sgabor return (EILSEQ); 338219019Sgabor plane += (wc & 0x80) ? 1 : 2; 339219019Sgabor mask |= 0x7F00; 340219019Sgabor } 341219019Sgabor *csid = plane; 342219019Sgabor *idx = (_index_t)(wc & mask); 343219019Sgabor 344219019Sgabor return (0); 345219019Sgabor} 346219019Sgabor 347219019Sgaborstatic __inline int 348219019Sgabor/*ARGSUSED*/ 349219019Sgabor_citrus_DECHanyu_stdenc_cstowc(_DECHanyuEncodingInfo * __restrict ei __unused, 350219019Sgabor wchar_t * __restrict wc, _csid_t csid, _index_t idx) 351219019Sgabor{ 352219019Sgabor 353219019Sgabor if (csid == 0) { 354219019Sgabor if (idx > 0x7F) 355219019Sgabor return (EILSEQ); 356219019Sgabor } else if (csid <= 4) { 357219019Sgabor if (!is_94charset(idx >> 8)) 358219019Sgabor return (EILSEQ); 359219019Sgabor if (!is_94charset(idx & 0xFF)) 360219019Sgabor return (EILSEQ); 361219019Sgabor if (csid % 2) 362219019Sgabor idx |= 0x80; 363219019Sgabor idx |= 0x8000; 364219019Sgabor if (csid > 2) 365219019Sgabor idx |= HANYUBIT; 366219019Sgabor } else 367219019Sgabor return (EILSEQ); 368219019Sgabor *wc = (wchar_t)idx; 369219019Sgabor return (0); 370219019Sgabor} 371219019Sgabor 372219019Sgaborstatic __inline int 373219019Sgabor/*ARGSUSED*/ 374219019Sgabor_citrus_DECHanyu_stdenc_get_state_desc_generic( 375219019Sgabor _DECHanyuEncodingInfo * __restrict ei __unused, 376219019Sgabor _DECHanyuState * __restrict psenc, int * __restrict rstate) 377219019Sgabor{ 378219019Sgabor 379219019Sgabor *rstate = (psenc->chlen == 0) 380219019Sgabor ? _STDENC_SDGEN_INITIAL 381219019Sgabor : _STDENC_SDGEN_INCOMPLETE_CHAR; 382219019Sgabor return (0); 383219019Sgabor} 384219019Sgabor 385219019Sgabor/* ---------------------------------------------------------------------- 386219019Sgabor * public interface for stdenc 387219019Sgabor */ 388219019Sgabor 389219019Sgabor_CITRUS_STDENC_DECLS(DECHanyu); 390219019Sgabor_CITRUS_STDENC_DEF_OPS(DECHanyu); 391219019Sgabor 392219019Sgabor#include "citrus_stdenc_template.h" 393