1145256Sjkoshy/*- 2177107Sjkoshy * Copyright (c) 2002-2004 Tim J. Robbins 3145256Sjkoshy * All rights reserved. 4145256Sjkoshy * 5145256Sjkoshy * Copyright (c) 2011 The FreeBSD Foundation 6145256Sjkoshy * All rights reserved. 7145256Sjkoshy * Portions of this software were developed by David Chisnall 8145256Sjkoshy * under sponsorship from the FreeBSD Foundation. 9145256Sjkoshy * 10145256Sjkoshy * Redistribution and use in source and binary forms, with or without 11145256Sjkoshy * modification, are permitted provided that the following conditions 12145256Sjkoshy * are met: 13145256Sjkoshy * 1. Redistributions of source code must retain the above copyright 14145256Sjkoshy * notice, this list of conditions and the following disclaimer. 15145256Sjkoshy * 2. Redistributions in binary form must reproduce the above copyright 16145256Sjkoshy * notice, this list of conditions and the following disclaimer in the 17145256Sjkoshy * documentation and/or other materials provided with the distribution. 18145256Sjkoshy * 19145256Sjkoshy * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 20145256Sjkoshy * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21145256Sjkoshy * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22145256Sjkoshy * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 23145256Sjkoshy * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24145256Sjkoshy * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25145256Sjkoshy * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26145256Sjkoshy * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27145256Sjkoshy * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28145256Sjkoshy * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29145256Sjkoshy * SUCH DAMAGE. 30145256Sjkoshy */ 31145256Sjkoshy/* 32145256Sjkoshy * PRC National Standard GB 18030-2000 encoding of Chinese text. 33145256Sjkoshy * 34145256Sjkoshy * See gb18030(5) for details. 35145256Sjkoshy */ 36145256Sjkoshy 37145256Sjkoshy#include <sys/param.h> 38145256Sjkoshy__FBSDID("$FreeBSD$"); 39145256Sjkoshy 40145256Sjkoshy#include <errno.h> 41145256Sjkoshy#include <runetype.h> 42145256Sjkoshy#include <stdlib.h> 43145256Sjkoshy#include <string.h> 44145256Sjkoshy#include <wchar.h> 45185363Sjkoshy#include "mblocal.h" 46185363Sjkoshy 47145256Sjkoshystatic size_t _GB18030_mbrtowc(wchar_t * __restrict, const char * __restrict, 48145340Smarcel size_t, mbstate_t * __restrict); 49145256Sjkoshystatic int _GB18030_mbsinit(const mbstate_t *); 50145256Sjkoshystatic size_t _GB18030_wcrtomb(char * __restrict, wchar_t, 51147191Sjkoshy mbstate_t * __restrict); 52147759Sjkoshy 53185363Sjkoshytypedef struct { 54185363Sjkoshy int count; 55185363Sjkoshy u_char bytes[4]; 56185363Sjkoshy} _GB18030State; 57206089Sfabient 58206089Sfabientint 59206089Sfabient_GB18030_init(struct xlocale_ctype *l, _RuneLocale *rl) 60206089Sfabient{ 61147191Sjkoshy 62145256Sjkoshy l->__mbrtowc = _GB18030_mbrtowc; 63147759Sjkoshy l->__wcrtomb = _GB18030_wcrtomb; 64147759Sjkoshy l->__mbsinit = _GB18030_mbsinit; 65147191Sjkoshy l->runes = rl; 66147191Sjkoshy l->__mb_cur_max = 4; 67145256Sjkoshy l->__mb_sb_limit = 128; 68145256Sjkoshy 69147191Sjkoshy return (0); 70145256Sjkoshy} 71145256Sjkoshy 72183725Sjkoshystatic int 73183725Sjkoshy_GB18030_mbsinit(const mbstate_t *ps) 74183725Sjkoshy{ 75183725Sjkoshy 76200928Srpaulo return (ps == NULL || ((const _GB18030State *)ps)->count == 0); 77200928Srpaulo} 78200928Srpaulo 79200928Srpaulostatic size_t 80204635Sgnn_GB18030_mbrtowc(wchar_t * __restrict pwc, const char * __restrict s, 81233320Sgonzo size_t n, mbstate_t * __restrict ps) 82204635Sgnn{ 83204635Sgnn _GB18030State *gs; 84233628Sfabient wchar_t wch; 85233628Sfabient int ch, len, ocount; 86204635Sgnn size_t ncopy; 87228869Sjhibbits 88228869Sjhibbits gs = (_GB18030State *)ps; 89228869Sjhibbits 90228869Sjhibbits if (gs->count < 0 || gs->count > sizeof(gs->bytes)) { 91204635Sgnn errno = EINVAL; 92145256Sjkoshy return ((size_t)-1); 93145256Sjkoshy } 94145256Sjkoshy 95145256Sjkoshy if (s == NULL) { 96145256Sjkoshy s = ""; 97145256Sjkoshy n = 1; 98145256Sjkoshy pwc = NULL; 99145256Sjkoshy } 100145256Sjkoshy 101145256Sjkoshy ncopy = MIN(MIN(n, MB_CUR_MAX), sizeof(gs->bytes) - gs->count); 102145256Sjkoshy memcpy(gs->bytes + gs->count, s, ncopy); 103145256Sjkoshy ocount = gs->count; 104145256Sjkoshy gs->count += ncopy; 105145256Sjkoshy s = (char *)gs->bytes; 106145256Sjkoshy n = gs->count; 107145256Sjkoshy 108145256Sjkoshy if (n == 0) 109183725Sjkoshy /* Incomplete multibyte sequence */ 110145256Sjkoshy return ((size_t)-2); 111145256Sjkoshy 112145256Sjkoshy /* 113145256Sjkoshy * Single byte: [00-7f] 114145256Sjkoshy * Two byte: [81-fe][40-7e,80-fe] 115145256Sjkoshy * Four byte: [81-fe][30-39][81-fe][30-39] 116145256Sjkoshy */ 117183725Sjkoshy ch = (unsigned char)*s++; 118183725Sjkoshy if (ch <= 0x7f) { 119183725Sjkoshy len = 1; 120183725Sjkoshy wch = ch; 121183725Sjkoshy } else if (ch >= 0x81 && ch <= 0xfe) { 122183725Sjkoshy wch = ch; 123183725Sjkoshy if (n < 2) 124183725Sjkoshy return ((size_t)-2); 125183725Sjkoshy ch = (unsigned char)*s++; 126183725Sjkoshy if ((ch >= 0x40 && ch <= 0x7e) || (ch >= 0x80 && ch <= 0xfe)) { 127183725Sjkoshy wch = (wch << 8) | ch; 128183725Sjkoshy len = 2; 129183725Sjkoshy } else if (ch >= 0x30 && ch <= 0x39) { 130183725Sjkoshy /* 131183725Sjkoshy * Strip high bit off the wide character we will 132183725Sjkoshy * eventually output so that it is positive when 133183725Sjkoshy * cast to wint_t on 32-bit twos-complement machines. 134183725Sjkoshy */ 135183725Sjkoshy wch = ((wch & 0x7f) << 8) | ch; 136183725Sjkoshy if (n < 3) 137183725Sjkoshy return ((size_t)-2); 138185363Sjkoshy ch = (unsigned char)*s++; 139183725Sjkoshy if (ch < 0x81 || ch > 0xfe) 140185363Sjkoshy goto ilseq; 141183725Sjkoshy wch = (wch << 8) | ch; 142185363Sjkoshy if (n < 4) 143183725Sjkoshy return ((size_t)-2); 144183725Sjkoshy ch = (unsigned char)*s++; 145183725Sjkoshy if (ch < 0x30 || ch > 0x39) 146185363Sjkoshy goto ilseq; 147185363Sjkoshy wch = (wch << 8) | ch; 148185363Sjkoshy len = 4; 149185363Sjkoshy } else 150185363Sjkoshy goto ilseq; 151185363Sjkoshy } else 152185363Sjkoshy goto ilseq; 153185363Sjkoshy 154200928Srpaulo if (pwc != NULL) 155204635Sgnn *pwc = wch; 156233335Sgonzo gs->count = 0; 157206089Sfabient return (wch == L'\0' ? 0 : len - ocount); 158228869Sjhibbitsilseq: 159185363Sjkoshy errno = EILSEQ; 160233628Sfabient return ((size_t)-1); 161233628Sfabient} 162185363Sjkoshy 163185363Sjkoshystatic size_t 164185363Sjkoshy_GB18030_wcrtomb(char * __restrict s, wchar_t wc, mbstate_t * __restrict ps) 165185363Sjkoshy{ 166185363Sjkoshy _GB18030State *gs; 167185363Sjkoshy size_t len; 168185363Sjkoshy int c; 169185363Sjkoshy 170185363Sjkoshy gs = (_GB18030State *)ps; 171185363Sjkoshy 172185363Sjkoshy if (gs->count != 0) { 173185363Sjkoshy errno = EINVAL; 174185363Sjkoshy return ((size_t)-1); 175185363Sjkoshy } 176185363Sjkoshy 177185363Sjkoshy if (s == NULL) 178185363Sjkoshy /* Reset to initial shift state (no-op) */ 179185363Sjkoshy return (1); 180185363Sjkoshy if ((wc & ~0x7fffffff) != 0) 181187761Sjeff goto ilseq; 182187761Sjeff if (wc & 0x7f000000) { 183187761Sjeff /* Replace high bit that mbrtowc() removed. */ 184187761Sjeff wc |= 0x80000000; 185187761Sjeff c = (wc >> 24) & 0xff; 186240164Sfabient if (c < 0x81 || c > 0xfe) 187240164Sfabient goto ilseq; 188240164Sfabient *s++ = c; 189240164Sfabient c = (wc >> 16) & 0xff; 190240164Sfabient if (c < 0x30 || c > 0x39) 191232366Sdavide goto ilseq; 192232366Sdavide *s++ = c; 193232366Sdavide c = (wc >> 8) & 0xff; 194232366Sdavide if (c < 0x81 || c > 0xfe) 195232366Sdavide goto ilseq; 196241738Ssbruno *s++ = c; 197241738Ssbruno c = wc & 0xff; 198241738Ssbruno if (c < 0x30 || c > 0x39) 199241738Ssbruno goto ilseq; 200241738Ssbruno *s++ = c; 201206089Sfabient len = 4; 202206089Sfabient } else if (wc & 0x00ff0000) 203206089Sfabient goto ilseq; 204206089Sfabient else if (wc & 0x0000ff00) { 205206089Sfabient c = (wc >> 8) & 0xff; 206206089Sfabient if (c < 0x81 || c > 0xfe) 207206089Sfabient goto ilseq; 208206089Sfabient *s++ = c; 209206089Sfabient c = wc & 0xff; 210206089Sfabient if (c < 0x40 || c == 0x7f || c == 0xff) 211232366Sdavide goto ilseq; 212232366Sdavide *s++ = c; 213232366Sdavide len = 2; 214232366Sdavide } else if (wc <= 0x7f) { 215232366Sdavide *s++ = wc; 216206089Sfabient len = 1; 217206089Sfabient } else 218206089Sfabient goto ilseq; 219206089Sfabient 220206089Sfabient return (len); 221185363Sjkoshyilseq: 222185363Sjkoshy errno = EILSEQ; 223185363Sjkoshy return ((size_t)-1); 224185363Sjkoshy} 225185363Sjkoshy