1ad30f8e7SGabor Kovesdan /* $FreeBSD$ */ 2ad30f8e7SGabor Kovesdan /* $NetBSD: citrus_mskanji.c,v 1.13 2008/06/14 16:01:08 tnozaki Exp $ */ 3ad30f8e7SGabor Kovesdan 4ad30f8e7SGabor Kovesdan /*- 5ad30f8e7SGabor Kovesdan * Copyright (c)2002 Citrus Project, 6ad30f8e7SGabor Kovesdan * All rights reserved. 7ad30f8e7SGabor Kovesdan * 8ad30f8e7SGabor Kovesdan * Redistribution and use in source and binary forms, with or without 9ad30f8e7SGabor Kovesdan * modification, are permitted provided that the following conditions 10ad30f8e7SGabor Kovesdan * are met: 11ad30f8e7SGabor Kovesdan * 1. Redistributions of source code must retain the above copyright 12ad30f8e7SGabor Kovesdan * notice, this list of conditions and the following disclaimer. 13ad30f8e7SGabor Kovesdan * 2. Redistributions in binary form must reproduce the above copyright 14ad30f8e7SGabor Kovesdan * notice, this list of conditions and the following disclaimer in the 15ad30f8e7SGabor Kovesdan * documentation and/or other materials provided with the distribution. 16ad30f8e7SGabor Kovesdan * 17ad30f8e7SGabor Kovesdan * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 18ad30f8e7SGabor Kovesdan * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19ad30f8e7SGabor Kovesdan * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20ad30f8e7SGabor Kovesdan * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 21ad30f8e7SGabor Kovesdan * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22ad30f8e7SGabor Kovesdan * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23ad30f8e7SGabor Kovesdan * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24ad30f8e7SGabor Kovesdan * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25ad30f8e7SGabor Kovesdan * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26ad30f8e7SGabor Kovesdan * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27ad30f8e7SGabor Kovesdan * SUCH DAMAGE. 28ad30f8e7SGabor Kovesdan */ 29ad30f8e7SGabor Kovesdan 30ad30f8e7SGabor Kovesdan /* 31ad30f8e7SGabor Kovesdan * ja_JP.SJIS locale table for BSD4.4/rune 32ad30f8e7SGabor Kovesdan * version 1.0 33ad30f8e7SGabor Kovesdan * (C) Sin'ichiro MIYATANI / Phase One, Inc 34ad30f8e7SGabor Kovesdan * May 12, 1995 35ad30f8e7SGabor Kovesdan * 36ad30f8e7SGabor Kovesdan * Redistribution and use in source and binary forms, with or without 37ad30f8e7SGabor Kovesdan * modification, are permitted provided that the following conditions 38ad30f8e7SGabor Kovesdan * are met: 39ad30f8e7SGabor Kovesdan * 1. Redistributions of source code must retain the above copyright 40ad30f8e7SGabor Kovesdan * notice, this list of conditions and the following disclaimer. 41ad30f8e7SGabor Kovesdan * 2. Redistributions in binary form must reproduce the above copyright 42ad30f8e7SGabor Kovesdan * notice, this list of conditions and the following disclaimer in the 43ad30f8e7SGabor Kovesdan * documentation and/or other materials provided with the distribution. 44ad30f8e7SGabor Kovesdan * 3. All advertising materials mentioning features or use of this software 45ad30f8e7SGabor Kovesdan * must display the following acknowledgement: 46ad30f8e7SGabor Kovesdan * This product includes software developed by Phase One, Inc. 47ad30f8e7SGabor Kovesdan * 4. The name of Phase One, Inc. may be used to endorse or promote products 48ad30f8e7SGabor Kovesdan * derived from this software without specific prior written permission. 49ad30f8e7SGabor Kovesdan * 50ad30f8e7SGabor Kovesdan * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 51ad30f8e7SGabor Kovesdan * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 52ad30f8e7SGabor Kovesdan * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 53ad30f8e7SGabor Kovesdan * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 54ad30f8e7SGabor Kovesdan * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 55ad30f8e7SGabor Kovesdan * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 56ad30f8e7SGabor Kovesdan * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 57ad30f8e7SGabor Kovesdan * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 58ad30f8e7SGabor Kovesdan * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 59ad30f8e7SGabor Kovesdan * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 60ad30f8e7SGabor Kovesdan * SUCH DAMAGE. 61ad30f8e7SGabor Kovesdan */ 62ad30f8e7SGabor Kovesdan 63ad30f8e7SGabor Kovesdan 64ad30f8e7SGabor Kovesdan #include <sys/cdefs.h> 65ad30f8e7SGabor Kovesdan #include <sys/types.h> 66ad30f8e7SGabor Kovesdan 67ad30f8e7SGabor Kovesdan #include <assert.h> 68ad30f8e7SGabor Kovesdan #include <errno.h> 69ad30f8e7SGabor Kovesdan #include <limits.h> 70ad30f8e7SGabor Kovesdan #include <stdbool.h> 71ad30f8e7SGabor Kovesdan #include <stddef.h> 72ad30f8e7SGabor Kovesdan #include <stdio.h> 73ad30f8e7SGabor Kovesdan #include <stdlib.h> 74ad30f8e7SGabor Kovesdan #include <string.h> 75ad30f8e7SGabor Kovesdan #include <wchar.h> 76ad30f8e7SGabor Kovesdan 77ad30f8e7SGabor Kovesdan #include "citrus_namespace.h" 78ad30f8e7SGabor Kovesdan #include "citrus_types.h" 79ad30f8e7SGabor Kovesdan #include "citrus_bcs.h" 80ad30f8e7SGabor Kovesdan #include "citrus_module.h" 81ad30f8e7SGabor Kovesdan #include "citrus_stdenc.h" 82ad30f8e7SGabor Kovesdan #include "citrus_mskanji.h" 83ad30f8e7SGabor Kovesdan 84ad30f8e7SGabor Kovesdan 85ad30f8e7SGabor Kovesdan /* ---------------------------------------------------------------------- 86ad30f8e7SGabor Kovesdan * private stuffs used by templates 87ad30f8e7SGabor Kovesdan */ 88ad30f8e7SGabor Kovesdan 89ad30f8e7SGabor Kovesdan typedef struct _MSKanjiState { 90ad30f8e7SGabor Kovesdan int chlen; 91ad30f8e7SGabor Kovesdan char ch[2]; 92ad30f8e7SGabor Kovesdan } _MSKanjiState; 93ad30f8e7SGabor Kovesdan 94ad30f8e7SGabor Kovesdan typedef struct { 95ad30f8e7SGabor Kovesdan int mode; 96ad30f8e7SGabor Kovesdan #define MODE_JIS2004 1 97ad30f8e7SGabor Kovesdan } _MSKanjiEncodingInfo; 98ad30f8e7SGabor Kovesdan 99ad30f8e7SGabor Kovesdan #define _CEI_TO_EI(_cei_) (&(_cei_)->ei) 100ad30f8e7SGabor Kovesdan #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_ 101ad30f8e7SGabor Kovesdan 102ad30f8e7SGabor Kovesdan #define _FUNCNAME(m) _citrus_MSKanji_##m 103ad30f8e7SGabor Kovesdan #define _ENCODING_INFO _MSKanjiEncodingInfo 104ad30f8e7SGabor Kovesdan #define _ENCODING_STATE _MSKanjiState 105ad30f8e7SGabor Kovesdan #define _ENCODING_MB_CUR_MAX(_ei_) 2 106ad30f8e7SGabor Kovesdan #define _ENCODING_IS_STATE_DEPENDENT 0 107ad30f8e7SGabor Kovesdan #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0 108ad30f8e7SGabor Kovesdan 109ad30f8e7SGabor Kovesdan 110ad30f8e7SGabor Kovesdan static bool 111ad30f8e7SGabor Kovesdan _mskanji1(int c) 112ad30f8e7SGabor Kovesdan { 113ad30f8e7SGabor Kovesdan 114ad30f8e7SGabor Kovesdan return ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc)); 115ad30f8e7SGabor Kovesdan } 116ad30f8e7SGabor Kovesdan 117ad30f8e7SGabor Kovesdan static bool 118ad30f8e7SGabor Kovesdan _mskanji2(int c) 119ad30f8e7SGabor Kovesdan { 120ad30f8e7SGabor Kovesdan 121ad30f8e7SGabor Kovesdan return ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc)); 122ad30f8e7SGabor Kovesdan } 123ad30f8e7SGabor Kovesdan 124ad30f8e7SGabor Kovesdan static __inline void 125ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 126ad30f8e7SGabor Kovesdan _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei __unused, 127ad30f8e7SGabor Kovesdan _MSKanjiState * __restrict s) 128ad30f8e7SGabor Kovesdan { 129ad30f8e7SGabor Kovesdan 130ad30f8e7SGabor Kovesdan s->chlen = 0; 131ad30f8e7SGabor Kovesdan } 132ad30f8e7SGabor Kovesdan 133b61949ddSDimitry Andric #if 0 134ad30f8e7SGabor Kovesdan static __inline void 135ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 136ad30f8e7SGabor Kovesdan _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei __unused, 137ad30f8e7SGabor Kovesdan void * __restrict pspriv, const _MSKanjiState * __restrict s) 138ad30f8e7SGabor Kovesdan { 139ad30f8e7SGabor Kovesdan 140ad30f8e7SGabor Kovesdan memcpy(pspriv, (const void *)s, sizeof(*s)); 141ad30f8e7SGabor Kovesdan } 142ad30f8e7SGabor Kovesdan 143ad30f8e7SGabor Kovesdan static __inline void 144ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 145ad30f8e7SGabor Kovesdan _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei __unused, 146ad30f8e7SGabor Kovesdan _MSKanjiState * __restrict s, const void * __restrict pspriv) 147ad30f8e7SGabor Kovesdan { 148ad30f8e7SGabor Kovesdan 149ad30f8e7SGabor Kovesdan memcpy((void *)s, pspriv, sizeof(*s)); 150ad30f8e7SGabor Kovesdan } 151b61949ddSDimitry Andric #endif 152ad30f8e7SGabor Kovesdan 153ad30f8e7SGabor Kovesdan static int 154ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 155ad30f8e7SGabor Kovesdan _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei, 156*1243a98eSTijl Coosemans wchar_t * __restrict pwc, char ** __restrict s, size_t n, 157ad30f8e7SGabor Kovesdan _MSKanjiState * __restrict psenc, size_t * __restrict nresult) 158ad30f8e7SGabor Kovesdan { 159*1243a98eSTijl Coosemans char *s0; 160ad30f8e7SGabor Kovesdan wchar_t wchar; 161ad30f8e7SGabor Kovesdan int chlenbak, len; 162ad30f8e7SGabor Kovesdan 163ad30f8e7SGabor Kovesdan s0 = *s; 164ad30f8e7SGabor Kovesdan 165ad30f8e7SGabor Kovesdan if (s0 == NULL) { 166ad30f8e7SGabor Kovesdan _citrus_MSKanji_init_state(ei, psenc); 167ad30f8e7SGabor Kovesdan *nresult = 0; /* state independent */ 168ad30f8e7SGabor Kovesdan return (0); 169ad30f8e7SGabor Kovesdan } 170ad30f8e7SGabor Kovesdan 171ad30f8e7SGabor Kovesdan chlenbak = psenc->chlen; 172ad30f8e7SGabor Kovesdan 173ad30f8e7SGabor Kovesdan /* make sure we have the first byte in the buffer */ 174ad30f8e7SGabor Kovesdan switch (psenc->chlen) { 175ad30f8e7SGabor Kovesdan case 0: 176ad30f8e7SGabor Kovesdan if (n < 1) 177ad30f8e7SGabor Kovesdan goto restart; 178ad30f8e7SGabor Kovesdan psenc->ch[0] = *s0++; 179ad30f8e7SGabor Kovesdan psenc->chlen = 1; 180ad30f8e7SGabor Kovesdan n--; 181ad30f8e7SGabor Kovesdan break; 182ad30f8e7SGabor Kovesdan case 1: 183ad30f8e7SGabor Kovesdan break; 184ad30f8e7SGabor Kovesdan default: 185ad30f8e7SGabor Kovesdan /* illegal state */ 186ad30f8e7SGabor Kovesdan goto encoding_error; 187ad30f8e7SGabor Kovesdan } 188ad30f8e7SGabor Kovesdan 189ad30f8e7SGabor Kovesdan len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1; 190ad30f8e7SGabor Kovesdan while (psenc->chlen < len) { 191ad30f8e7SGabor Kovesdan if (n < 1) 192ad30f8e7SGabor Kovesdan goto restart; 193ad30f8e7SGabor Kovesdan psenc->ch[psenc->chlen] = *s0++; 194ad30f8e7SGabor Kovesdan psenc->chlen++; 195ad30f8e7SGabor Kovesdan n--; 196ad30f8e7SGabor Kovesdan } 197ad30f8e7SGabor Kovesdan 198ad30f8e7SGabor Kovesdan *s = s0; 199ad30f8e7SGabor Kovesdan 200ad30f8e7SGabor Kovesdan switch (len) { 201ad30f8e7SGabor Kovesdan case 1: 202ad30f8e7SGabor Kovesdan wchar = psenc->ch[0] & 0xff; 203ad30f8e7SGabor Kovesdan break; 204ad30f8e7SGabor Kovesdan case 2: 205ad30f8e7SGabor Kovesdan if (!_mskanji2(psenc->ch[1] & 0xff)) 206ad30f8e7SGabor Kovesdan goto encoding_error; 207ad30f8e7SGabor Kovesdan wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff); 208ad30f8e7SGabor Kovesdan break; 209ad30f8e7SGabor Kovesdan default: 210ad30f8e7SGabor Kovesdan /* illegal state */ 211ad30f8e7SGabor Kovesdan goto encoding_error; 212ad30f8e7SGabor Kovesdan } 213ad30f8e7SGabor Kovesdan 214ad30f8e7SGabor Kovesdan psenc->chlen = 0; 215ad30f8e7SGabor Kovesdan 216ad30f8e7SGabor Kovesdan if (pwc) 217ad30f8e7SGabor Kovesdan *pwc = wchar; 218ad30f8e7SGabor Kovesdan *nresult = wchar ? len - chlenbak : 0; 219ad30f8e7SGabor Kovesdan return (0); 220ad30f8e7SGabor Kovesdan 221ad30f8e7SGabor Kovesdan encoding_error: 222ad30f8e7SGabor Kovesdan psenc->chlen = 0; 223ad30f8e7SGabor Kovesdan *nresult = (size_t)-1; 224ad30f8e7SGabor Kovesdan return (EILSEQ); 225ad30f8e7SGabor Kovesdan 226ad30f8e7SGabor Kovesdan restart: 227ad30f8e7SGabor Kovesdan *nresult = (size_t)-2; 228ad30f8e7SGabor Kovesdan *s = s0; 229ad30f8e7SGabor Kovesdan return (0); 230ad30f8e7SGabor Kovesdan } 231ad30f8e7SGabor Kovesdan 232ad30f8e7SGabor Kovesdan 233ad30f8e7SGabor Kovesdan static int 234ad30f8e7SGabor Kovesdan _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei __unused, 235ad30f8e7SGabor Kovesdan char * __restrict s, size_t n, wchar_t wc, 236ad30f8e7SGabor Kovesdan _MSKanjiState * __restrict psenc __unused, size_t * __restrict nresult) 237ad30f8e7SGabor Kovesdan { 238ad30f8e7SGabor Kovesdan int ret; 239ad30f8e7SGabor Kovesdan 240ad30f8e7SGabor Kovesdan /* check invalid sequence */ 241ad30f8e7SGabor Kovesdan if (wc & ~0xffff) { 242ad30f8e7SGabor Kovesdan ret = EILSEQ; 243ad30f8e7SGabor Kovesdan goto err; 244ad30f8e7SGabor Kovesdan } 245ad30f8e7SGabor Kovesdan 246ad30f8e7SGabor Kovesdan if (wc & 0xff00) { 247ad30f8e7SGabor Kovesdan if (n < 2) { 248ad30f8e7SGabor Kovesdan ret = E2BIG; 249ad30f8e7SGabor Kovesdan goto err; 250ad30f8e7SGabor Kovesdan } 251ad30f8e7SGabor Kovesdan 252ad30f8e7SGabor Kovesdan s[0] = (wc >> 8) & 0xff; 253ad30f8e7SGabor Kovesdan s[1] = wc & 0xff; 254ad30f8e7SGabor Kovesdan if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) { 255ad30f8e7SGabor Kovesdan ret = EILSEQ; 256ad30f8e7SGabor Kovesdan goto err; 257ad30f8e7SGabor Kovesdan } 258ad30f8e7SGabor Kovesdan 259ad30f8e7SGabor Kovesdan *nresult = 2; 260ad30f8e7SGabor Kovesdan return (0); 261ad30f8e7SGabor Kovesdan } else { 262ad30f8e7SGabor Kovesdan if (n < 1) { 263ad30f8e7SGabor Kovesdan ret = E2BIG; 264ad30f8e7SGabor Kovesdan goto err; 265ad30f8e7SGabor Kovesdan } 266ad30f8e7SGabor Kovesdan 267ad30f8e7SGabor Kovesdan s[0] = wc & 0xff; 268ad30f8e7SGabor Kovesdan if (_mskanji1(s[0] & 0xff)) { 269ad30f8e7SGabor Kovesdan ret = EILSEQ; 270ad30f8e7SGabor Kovesdan goto err; 271ad30f8e7SGabor Kovesdan } 272ad30f8e7SGabor Kovesdan 273ad30f8e7SGabor Kovesdan *nresult = 1; 274ad30f8e7SGabor Kovesdan return (0); 275ad30f8e7SGabor Kovesdan } 276ad30f8e7SGabor Kovesdan 277ad30f8e7SGabor Kovesdan err: 278ad30f8e7SGabor Kovesdan *nresult = (size_t)-1; 279ad30f8e7SGabor Kovesdan return (ret); 280ad30f8e7SGabor Kovesdan } 281ad30f8e7SGabor Kovesdan 282ad30f8e7SGabor Kovesdan 283ad30f8e7SGabor Kovesdan static __inline int 284ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 285ad30f8e7SGabor Kovesdan _citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei, 286ad30f8e7SGabor Kovesdan _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc) 287ad30f8e7SGabor Kovesdan { 288ad30f8e7SGabor Kovesdan _index_t col, row; 289ad30f8e7SGabor Kovesdan int offset; 290ad30f8e7SGabor Kovesdan 291ad30f8e7SGabor Kovesdan if ((_wc_t)wc < 0x80) { 292ad30f8e7SGabor Kovesdan /* ISO-646 */ 293ad30f8e7SGabor Kovesdan *csid = 0; 294ad30f8e7SGabor Kovesdan *idx = (_index_t)wc; 295ad30f8e7SGabor Kovesdan } else if ((_wc_t)wc < 0x100) { 296ad30f8e7SGabor Kovesdan /* KANA */ 297ad30f8e7SGabor Kovesdan *csid = 1; 298ad30f8e7SGabor Kovesdan *idx = (_index_t)wc & 0x7F; 299ad30f8e7SGabor Kovesdan } else { 300ad30f8e7SGabor Kovesdan /* Kanji (containing Gaiji zone) */ 301ad30f8e7SGabor Kovesdan /* 302ad30f8e7SGabor Kovesdan * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)): 303ad30f8e7SGabor Kovesdan * 0x8140 - 0x817E -> 0x2121 - 0x215F 304ad30f8e7SGabor Kovesdan * 0x8180 - 0x819E -> 0x2160 - 0x217E 305ad30f8e7SGabor Kovesdan * 0x819F - 0x81FC -> 0x2221 - 0x227E 306ad30f8e7SGabor Kovesdan * 307ad30f8e7SGabor Kovesdan * 0x8240 - 0x827E -> 0x2321 - 0x235F 308ad30f8e7SGabor Kovesdan * ... 309ad30f8e7SGabor Kovesdan * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E 310ad30f8e7SGabor Kovesdan * 311ad30f8e7SGabor Kovesdan * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F 312ad30f8e7SGabor Kovesdan * ... 313ad30f8e7SGabor Kovesdan * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E 314ad30f8e7SGabor Kovesdan * 315ad30f8e7SGabor Kovesdan * extended Gaiji zone: 316ad30f8e7SGabor Kovesdan * 0xF040 - 0xFCFC 317ad30f8e7SGabor Kovesdan * 318ad30f8e7SGabor Kovesdan * JIS X0213-plane2: 319ad30f8e7SGabor Kovesdan * 0xF040 - 0xF09E -> 0x2121 - 0x217E 320ad30f8e7SGabor Kovesdan * 0xF140 - 0xF19E -> 0x2321 - 0x237E 321ad30f8e7SGabor Kovesdan * ... 322ad30f8e7SGabor Kovesdan * 0xF240 - 0xF29E -> 0x2521 - 0x257E 323ad30f8e7SGabor Kovesdan * 324ad30f8e7SGabor Kovesdan * 0xF09F - 0xF0FC -> 0x2821 - 0x287E 325ad30f8e7SGabor Kovesdan * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E 326ad30f8e7SGabor Kovesdan * ... 327ad30f8e7SGabor Kovesdan * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E 328ad30f8e7SGabor Kovesdan * 329ad30f8e7SGabor Kovesdan * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E 330ad30f8e7SGabor Kovesdan * ... 331ad30f8e7SGabor Kovesdan * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E 332ad30f8e7SGabor Kovesdan */ 333ad30f8e7SGabor Kovesdan row = ((_wc_t)wc >> 8) & 0xFF; 334ad30f8e7SGabor Kovesdan col = (_wc_t)wc & 0xFF; 335ad30f8e7SGabor Kovesdan if (!_mskanji1(row) || !_mskanji2(col)) 336ad30f8e7SGabor Kovesdan return (EILSEQ); 337ad30f8e7SGabor Kovesdan if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) { 338ad30f8e7SGabor Kovesdan *csid = 2; 339ad30f8e7SGabor Kovesdan offset = 0x81; 340ad30f8e7SGabor Kovesdan } else { 341ad30f8e7SGabor Kovesdan *csid = 3; 342ad30f8e7SGabor Kovesdan if ((_wc_t)wc <= 0xF49E) { 343ad30f8e7SGabor Kovesdan offset = (_wc_t)wc >= 0xF29F || 344ad30f8e7SGabor Kovesdan ((_wc_t)wc >= 0xF09F && 345ad30f8e7SGabor Kovesdan (_wc_t)wc <= 0xF0FC) ? 0xED : 0xF0; 346ad30f8e7SGabor Kovesdan } else 347ad30f8e7SGabor Kovesdan offset = 0xCE; 348ad30f8e7SGabor Kovesdan } 349ad30f8e7SGabor Kovesdan row -= offset; 350ad30f8e7SGabor Kovesdan if (row >= 0x5F) 351ad30f8e7SGabor Kovesdan row -= 0x40; 352ad30f8e7SGabor Kovesdan row = row * 2 + 0x21; 353ad30f8e7SGabor Kovesdan col -= 0x1F; 354ad30f8e7SGabor Kovesdan if (col >= 0x61) 355ad30f8e7SGabor Kovesdan col -= 1; 356ad30f8e7SGabor Kovesdan if (col > 0x7E) { 357ad30f8e7SGabor Kovesdan row += 1; 358ad30f8e7SGabor Kovesdan col -= 0x5E; 359ad30f8e7SGabor Kovesdan } 360ad30f8e7SGabor Kovesdan *idx = ((_index_t)row << 8) | col; 361ad30f8e7SGabor Kovesdan } 362ad30f8e7SGabor Kovesdan 363ad30f8e7SGabor Kovesdan return (0); 364ad30f8e7SGabor Kovesdan } 365ad30f8e7SGabor Kovesdan 366ad30f8e7SGabor Kovesdan static __inline int 367ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 368ad30f8e7SGabor Kovesdan _citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei, 369ad30f8e7SGabor Kovesdan wchar_t * __restrict wc, _csid_t csid, _index_t idx) 370ad30f8e7SGabor Kovesdan { 371ad30f8e7SGabor Kovesdan uint32_t col, row; 372ad30f8e7SGabor Kovesdan int offset; 373ad30f8e7SGabor Kovesdan 374ad30f8e7SGabor Kovesdan switch (csid) { 375ad30f8e7SGabor Kovesdan case 0: 376ad30f8e7SGabor Kovesdan /* ISO-646 */ 377ad30f8e7SGabor Kovesdan if (idx >= 0x80) 378ad30f8e7SGabor Kovesdan return (EILSEQ); 379ad30f8e7SGabor Kovesdan *wc = (wchar_t)idx; 380ad30f8e7SGabor Kovesdan break; 381ad30f8e7SGabor Kovesdan case 1: 382ad30f8e7SGabor Kovesdan /* kana */ 383ad30f8e7SGabor Kovesdan if (idx >= 0x80) 384ad30f8e7SGabor Kovesdan return (EILSEQ); 385ad30f8e7SGabor Kovesdan *wc = (wchar_t)idx + 0x80; 386ad30f8e7SGabor Kovesdan break; 387ad30f8e7SGabor Kovesdan case 3: 388ad30f8e7SGabor Kovesdan if ((ei->mode & MODE_JIS2004) == 0) 389ad30f8e7SGabor Kovesdan return (EILSEQ); 390ad30f8e7SGabor Kovesdan /*FALLTHROUGH*/ 391ad30f8e7SGabor Kovesdan case 2: 392ad30f8e7SGabor Kovesdan /* kanji */ 393ad30f8e7SGabor Kovesdan row = (idx >> 8); 394ad30f8e7SGabor Kovesdan if (row < 0x21) 395ad30f8e7SGabor Kovesdan return (EILSEQ); 396ad30f8e7SGabor Kovesdan if (csid == 3) { 397ad30f8e7SGabor Kovesdan if (row <= 0x2F) 398ad30f8e7SGabor Kovesdan offset = (row == 0x22 || row >= 0x26) ? 399ad30f8e7SGabor Kovesdan 0xED : 0xF0; 400ad30f8e7SGabor Kovesdan else if (row >= 0x4D && row <= 0x7E) 401ad30f8e7SGabor Kovesdan offset = 0xCE; 402ad30f8e7SGabor Kovesdan else 403ad30f8e7SGabor Kovesdan return (EILSEQ); 404ad30f8e7SGabor Kovesdan } else { 405ad30f8e7SGabor Kovesdan if (row > 0x97) 406ad30f8e7SGabor Kovesdan return (EILSEQ); 407ad30f8e7SGabor Kovesdan offset = (row < 0x5F) ? 0x81 : 0xC1; 408ad30f8e7SGabor Kovesdan } 409ad30f8e7SGabor Kovesdan col = idx & 0xFF; 410ad30f8e7SGabor Kovesdan if (col < 0x21 || col > 0x7E) 411ad30f8e7SGabor Kovesdan return (EILSEQ); 412ad30f8e7SGabor Kovesdan row -= 0x21; col -= 0x21; 413ad30f8e7SGabor Kovesdan if ((row & 1) == 0) { 414ad30f8e7SGabor Kovesdan col += 0x40; 415ad30f8e7SGabor Kovesdan if (col >= 0x7F) 416ad30f8e7SGabor Kovesdan col += 1; 417ad30f8e7SGabor Kovesdan } else 418ad30f8e7SGabor Kovesdan col += 0x9F; 419ad30f8e7SGabor Kovesdan row = row / 2 + offset; 420ad30f8e7SGabor Kovesdan *wc = ((wchar_t)row << 8) | col; 421ad30f8e7SGabor Kovesdan break; 422ad30f8e7SGabor Kovesdan default: 423ad30f8e7SGabor Kovesdan return (EILSEQ); 424ad30f8e7SGabor Kovesdan } 425ad30f8e7SGabor Kovesdan 426ad30f8e7SGabor Kovesdan return (0); 427ad30f8e7SGabor Kovesdan } 428ad30f8e7SGabor Kovesdan 429ad30f8e7SGabor Kovesdan static __inline int 430ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 431ad30f8e7SGabor Kovesdan _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei __unused, 432ad30f8e7SGabor Kovesdan _MSKanjiState * __restrict psenc, int * __restrict rstate) 433ad30f8e7SGabor Kovesdan { 434ad30f8e7SGabor Kovesdan 435ad30f8e7SGabor Kovesdan *rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL : 436ad30f8e7SGabor Kovesdan _STDENC_SDGEN_INCOMPLETE_CHAR; 437ad30f8e7SGabor Kovesdan return (0); 438ad30f8e7SGabor Kovesdan } 439ad30f8e7SGabor Kovesdan 440ad30f8e7SGabor Kovesdan static int 441ad30f8e7SGabor Kovesdan /*ARGSUSED*/ 442ad30f8e7SGabor Kovesdan _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo * __restrict ei, 443ad30f8e7SGabor Kovesdan const void * __restrict var, size_t lenvar) 444ad30f8e7SGabor Kovesdan { 445ad30f8e7SGabor Kovesdan const char *p; 446ad30f8e7SGabor Kovesdan 447ad30f8e7SGabor Kovesdan p = var; 448ad30f8e7SGabor Kovesdan memset((void *)ei, 0, sizeof(*ei)); 449ad30f8e7SGabor Kovesdan while (lenvar > 0) { 450ad30f8e7SGabor Kovesdan switch (_bcs_toupper(*p)) { 451ad30f8e7SGabor Kovesdan case 'J': 452ad30f8e7SGabor Kovesdan MATCH(JIS2004, ei->mode |= MODE_JIS2004); 453ad30f8e7SGabor Kovesdan break; 454ad30f8e7SGabor Kovesdan } 455ad30f8e7SGabor Kovesdan ++p; 456ad30f8e7SGabor Kovesdan --lenvar; 457ad30f8e7SGabor Kovesdan } 458ad30f8e7SGabor Kovesdan 459ad30f8e7SGabor Kovesdan return (0); 460ad30f8e7SGabor Kovesdan } 461ad30f8e7SGabor Kovesdan 462ad30f8e7SGabor Kovesdan static void 463ad30f8e7SGabor Kovesdan _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei __unused) 464ad30f8e7SGabor Kovesdan { 465ad30f8e7SGabor Kovesdan 466ad30f8e7SGabor Kovesdan } 467ad30f8e7SGabor Kovesdan 468ad30f8e7SGabor Kovesdan /* ---------------------------------------------------------------------- 469ad30f8e7SGabor Kovesdan * public interface for stdenc 470ad30f8e7SGabor Kovesdan */ 471ad30f8e7SGabor Kovesdan 472ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DECLS(MSKanji); 473ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DEF_OPS(MSKanji); 474ad30f8e7SGabor Kovesdan 475ad30f8e7SGabor Kovesdan #include "citrus_stdenc_template.h" 476