4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore/*
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore * Copyright 2013 Garrett D'Amore <garrett@damore.org>
6b5e5868e7ebf1aff3a5abd7d0c4ef0e5fbf3648Garrett D'Amore * Copyright 2010 Nexenta Systems, Inc. All rights reserved.
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * Copyright (c) 2004 Tim J. Robbins. All rights reserved.
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * Copyright (c) 2003 David Xu <davidxu@freebsd.org>
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * All rights reserved.
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore *
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * Redistribution and use in source and binary forms, with or without
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * modification, are permitted provided that the following conditions
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * are met:
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * 1. Redistributions of source code must retain the above copyright
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * notice, this list of conditions and the following disclaimer.
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * 2. Redistributions in binary form must reproduce the above copyright
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * notice, this list of conditions and the following disclaimer in the
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * documentation and/or other materials provided with the distribution.
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore *
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore * SUCH DAMAGE.
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore */
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include "lint.h"
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include <sys/types.h>
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include <errno.h>
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include "runetype.h"
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include <stdlib.h>
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include <string.h>
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include <wchar.h>
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore#include "mblocal.h"
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore#include "lctype.h"
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic size_t _GB2312_mbrtowc(wchar_t *_RESTRICT_KYWD,
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore const char *_RESTRICT_KYWD,
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore size_t, mbstate_t *_RESTRICT_KYWD);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic int _GB2312_mbsinit(const mbstate_t *);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic size_t _GB2312_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore mbstate_t *_RESTRICT_KYWD);
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amorestatic size_t _GB2312_mbsnrtowcs(wchar_t *_RESTRICT_KYWD,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore const char **_RESTRICT_KYWD, size_t, size_t,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore mbstate_t *_RESTRICT_KYWD);
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amorestatic size_t _GB2312_wcsnrtombs(char *_RESTRICT_KYWD,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore const wchar_t **_RESTRICT_KYWD, size_t, size_t,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore mbstate_t *_RESTRICT_KYWD);
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amoretypedef struct {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore int count;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore uchar_t bytes[2];
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore} _GB2312State;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amorevoid
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore_GB2312_init(struct lc_ctype *lct)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore{
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_mbrtowc = _GB2312_mbrtowc;
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_wcrtomb = _GB2312_wcrtomb;
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_mbsinit = _GB2312_mbsinit;
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_mbsnrtowcs = _GB2312_mbsnrtowcs;
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_wcsnrtombs = _GB2312_wcsnrtombs;
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_max_mblen = 2;
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore lct->lc_is_ascii = 0;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore}
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic int
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore_GB2312_mbsinit(const mbstate_t *ps)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore{
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (ps == NULL || ((const _GB2312State *)ps)->count == 0);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore}
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic int
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore_GB2312_check(const char *str, size_t n)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore{
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore const uchar_t *s = (const uchar_t *)str;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (n == 0)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore /* Incomplete multibyte sequence */
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (-2);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (s[0] >= 0xa1 && s[0] <= 0xfe) {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (n < 2)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore /* Incomplete multibyte sequence */
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (-2);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (s[1] < 0xa1 || s[1] > 0xfe)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore /* Invalid multibyte sequence */
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (-1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (2);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore } else if (s[0] & 0x80) {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore /* Invalid multibyte sequence */
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (-1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore }
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore}
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic size_t
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore_GB2312_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore size_t n, mbstate_t *_RESTRICT_KYWD ps)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore{
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore _GB2312State *gs;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore wchar_t wc;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore int i, len, ocount;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore size_t ncopy;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore gs = (_GB2312State *)ps;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (gs->count < 0 || gs->count > sizeof (gs->bytes)) {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore errno = EINVAL;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return ((size_t)-1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore }
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (s == NULL) {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore s = "";
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore n = 1;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore pwc = NULL;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore }
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore ncopy = MIN(MIN(n, MB_CUR_MAX), sizeof (gs->bytes) - gs->count);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore (void) memcpy(gs->bytes + gs->count, s, ncopy);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore ocount = gs->count;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore gs->count += ncopy;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore s = (char *)gs->bytes;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore n = gs->count;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if ((len = _GB2312_check(s, n)) < 0)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return ((size_t)len);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore wc = 0;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore i = len;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore while (i-- > 0)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore wc = (wc << 8) | (unsigned char)*s++;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (pwc != NULL)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore *pwc = wc;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore gs->count = 0;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (wc == L'\0' ? 0 : len - ocount);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore}
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amorestatic size_t
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore_GB2312_wcrtomb(char *_RESTRICT_KYWD s, wchar_t wc,
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore mbstate_t *_RESTRICT_KYWD ps)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore{
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore _GB2312State *gs;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore gs = (_GB2312State *)ps;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (gs->count != 0) {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore errno = EINVAL;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return ((size_t)-1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore }
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (s == NULL)
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore /* Reset to initial shift state (no-op) */
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore if (wc & 0x8000) {
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore *s++ = (wc >> 8) & 0xff;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore *s = wc & 0xff;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (2);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore }
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore *s = wc & 0xff;
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore return (1);
4297a3b0d0a35d80f86fff155e288e885a100e6dGarrett D'Amore}
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amorestatic size_t
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore_GB2312_mbsnrtowcs(wchar_t *_RESTRICT_KYWD dst,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore const char **_RESTRICT_KYWD src, size_t nms, size_t len,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore mbstate_t *_RESTRICT_KYWD ps)
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore{
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore return (__mbsnrtowcs_std(dst, src, nms, len, ps, _GB2312_mbrtowc));
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore}
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amorestatic size_t
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore_GB2312_wcsnrtombs(char *_RESTRICT_KYWD dst,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore const wchar_t **_RESTRICT_KYWD src, size_t nwc, size_t len,
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore mbstate_t *_RESTRICT_KYWD ps)
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore{
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore return (__wcsnrtombs_std(dst, src, nwc, len, ps, _GB2312_wcrtomb));
2d08521bd15501c8370ba2153b9cca4f094979d0Garrett D'Amore}