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