xref: /illumos-gate/usr/src/lib/libc/port/locale/gb18030.c (revision bbf21555)
14297a3b0SGarrett D'Amore /*
22d08521bSGarrett D'Amore  * Copyright 2013 Garrett D'Amore <garrett@damore.org>
36b5e5868SGarrett D'Amore  * Copyright 2010 Nexenta Systems, Inc.  All rights reserved.
44297a3b0SGarrett D'Amore  * Copyright (c) 2002-2004 Tim J. Robbins
54297a3b0SGarrett D'Amore  * All rights reserved.
64297a3b0SGarrett D'Amore  *
74297a3b0SGarrett D'Amore  * Redistribution and use in source and binary forms, with or without
84297a3b0SGarrett D'Amore  * modification, are permitted provided that the following conditions
94297a3b0SGarrett D'Amore  * are met:
104297a3b0SGarrett D'Amore  * 1. Redistributions of source code must retain the above copyright
114297a3b0SGarrett D'Amore  *    notice, this list of conditions and the following disclaimer.
124297a3b0SGarrett D'Amore  * 2. Redistributions in binary form must reproduce the above copyright
134297a3b0SGarrett D'Amore  *    notice, this list of conditions and the following disclaimer in the
144297a3b0SGarrett D'Amore  *    documentation and/or other materials provided with the distribution.
154297a3b0SGarrett D'Amore  *
164297a3b0SGarrett D'Amore  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
174297a3b0SGarrett D'Amore  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
184297a3b0SGarrett D'Amore  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
194297a3b0SGarrett D'Amore  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
204297a3b0SGarrett D'Amore  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
214297a3b0SGarrett D'Amore  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
224297a3b0SGarrett D'Amore  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
234297a3b0SGarrett D'Amore  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
244297a3b0SGarrett D'Amore  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
254297a3b0SGarrett D'Amore  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
264297a3b0SGarrett D'Amore  * SUCH DAMAGE.
274297a3b0SGarrett D'Amore  */
284297a3b0SGarrett D'Amore 
294297a3b0SGarrett D'Amore /*
304297a3b0SGarrett D'Amore  * PRC National Standard GB 18030-2000 encoding of Chinese text.
314297a3b0SGarrett D'Amore  *
32*bbf21555SRichard Lowe  * See gb18030(7) for details.
334297a3b0SGarrett D'Amore  */
344297a3b0SGarrett D'Amore 
354297a3b0SGarrett D'Amore #include "lint.h"
364297a3b0SGarrett D'Amore #include <sys/types.h>
374297a3b0SGarrett D'Amore #include <errno.h>
384297a3b0SGarrett D'Amore #include <stdlib.h>
394297a3b0SGarrett D'Amore #include <string.h>
404297a3b0SGarrett D'Amore #include <wchar.h>
414297a3b0SGarrett D'Amore #include "mblocal.h"
422d08521bSGarrett D'Amore #include "lctype.h"
434297a3b0SGarrett D'Amore 
444297a3b0SGarrett D'Amore 
454297a3b0SGarrett D'Amore static size_t	_GB18030_mbrtowc(wchar_t *_RESTRICT_KYWD,
464297a3b0SGarrett D'Amore 		    const char *_RESTRICT_KYWD,
470ac311baSRobert Mustacchi 		    size_t, mbstate_t *_RESTRICT_KYWD, boolean_t);
484297a3b0SGarrett D'Amore static int	_GB18030_mbsinit(const mbstate_t *);
494297a3b0SGarrett D'Amore static size_t	_GB18030_wcrtomb(char *_RESTRICT_KYWD, wchar_t,
504297a3b0SGarrett D'Amore 		    mbstate_t *_RESTRICT_KYWD);
512d08521bSGarrett D'Amore static size_t	_GB18030_mbsnrtowcs(wchar_t *_RESTRICT_KYWD,
522d08521bSGarrett D'Amore 		    const char **_RESTRICT_KYWD, size_t, size_t,
532d08521bSGarrett D'Amore 		    mbstate_t *_RESTRICT_KYWD);
542d08521bSGarrett D'Amore static size_t	_GB18030_wcsnrtombs(char *_RESTRICT_KYWD,
552d08521bSGarrett D'Amore 		    const wchar_t **_RESTRICT_KYWD, size_t, size_t,
562d08521bSGarrett D'Amore 		    mbstate_t *_RESTRICT_KYWD);
572d08521bSGarrett D'Amore 
582d08521bSGarrett D'Amore void
_GB18030_init(struct lc_ctype * lct)592d08521bSGarrett D'Amore _GB18030_init(struct lc_ctype *lct)
604297a3b0SGarrett D'Amore {
614297a3b0SGarrett D'Amore 
622d08521bSGarrett D'Amore 	lct->lc_mbrtowc = _GB18030_mbrtowc;
632d08521bSGarrett D'Amore 	lct->lc_wcrtomb = _GB18030_wcrtomb;
642d08521bSGarrett D'Amore 	lct->lc_mbsinit = _GB18030_mbsinit;
652d08521bSGarrett D'Amore 	lct->lc_mbsnrtowcs = _GB18030_mbsnrtowcs;
662d08521bSGarrett D'Amore 	lct->lc_wcsnrtombs = _GB18030_wcsnrtombs;
672d08521bSGarrett D'Amore 	lct->lc_max_mblen = 4;
682d08521bSGarrett D'Amore 	lct->lc_is_ascii = 0;
694297a3b0SGarrett D'Amore }
704297a3b0SGarrett D'Amore 
714297a3b0SGarrett D'Amore static int
_GB18030_mbsinit(const mbstate_t * ps)724297a3b0SGarrett D'Amore _GB18030_mbsinit(const mbstate_t *ps)
734297a3b0SGarrett D'Amore {
744297a3b0SGarrett D'Amore 
754297a3b0SGarrett D'Amore 	return (ps == NULL || ((const _GB18030State *)ps)->count == 0);
764297a3b0SGarrett D'Amore }
774297a3b0SGarrett D'Amore 
784297a3b0SGarrett D'Amore static size_t
_GB18030_mbrtowc(wchar_t * _RESTRICT_KYWD pwc,const char * _RESTRICT_KYWD s,size_t n,mbstate_t * _RESTRICT_KYWD ps,boolean_t zero)794297a3b0SGarrett D'Amore _GB18030_mbrtowc(wchar_t *_RESTRICT_KYWD pwc, const char *_RESTRICT_KYWD s,
800ac311baSRobert Mustacchi     size_t n, mbstate_t *_RESTRICT_KYWD ps, boolean_t zero)
814297a3b0SGarrett D'Amore {
824297a3b0SGarrett D'Amore 	_GB18030State *gs;
834297a3b0SGarrett D'Amore 	wchar_t wch;
844297a3b0SGarrett D'Amore 	int ch, len, ocount;
854297a3b0SGarrett D'Amore 	size_t ncopy;
864297a3b0SGarrett D'Amore 
874297a3b0SGarrett D'Amore 	gs = (_GB18030State *)ps;
884297a3b0SGarrett D'Amore 
894297a3b0SGarrett D'Amore 	if (gs->count < 0 || gs->count > sizeof (gs->bytes)) {
904297a3b0SGarrett D'Amore 		errno = EINVAL;
914297a3b0SGarrett D'Amore 		return ((size_t)-1);
924297a3b0SGarrett D'Amore 	}
934297a3b0SGarrett D'Amore 
944297a3b0SGarrett D'Amore 	if (s == NULL) {
954297a3b0SGarrett D'Amore 		s = "";
964297a3b0SGarrett D'Amore 		n = 1;
974297a3b0SGarrett D'Amore 		pwc = NULL;
984297a3b0SGarrett D'Amore 	}
994297a3b0SGarrett D'Amore 
1004297a3b0SGarrett D'Amore 	ncopy = MIN(MIN(n, MB_CUR_MAX), sizeof (gs->bytes) - gs->count);
1014297a3b0SGarrett D'Amore 	(void) memcpy(gs->bytes + gs->count, s, ncopy);
1024297a3b0SGarrett D'Amore 	ocount = gs->count;
1034297a3b0SGarrett D'Amore 	gs->count += ncopy;
1044297a3b0SGarrett D'Amore 	s = (char *)gs->bytes;
1054297a3b0SGarrett D'Amore 	n = gs->count;
1064297a3b0SGarrett D'Amore 
1074297a3b0SGarrett D'Amore 	if (n == 0)
1084297a3b0SGarrett D'Amore 		/* Incomplete multibyte sequence */
1094297a3b0SGarrett D'Amore 		return ((size_t)-2);
1104297a3b0SGarrett D'Amore 
1114297a3b0SGarrett D'Amore 	/*
1124297a3b0SGarrett D'Amore 	 * Single byte:		[00-7f]
1134297a3b0SGarrett D'Amore 	 * Two byte:		[81-fe][40-7e,80-fe]
1144297a3b0SGarrett D'Amore 	 * Four byte:		[81-fe][30-39][81-fe][30-39]
1154297a3b0SGarrett D'Amore 	 */
1164297a3b0SGarrett D'Amore 	ch = (unsigned char)*s++;
1174297a3b0SGarrett D'Amore 	if (ch <= 0x7f) {
1184297a3b0SGarrett D'Amore 		len = 1;
1194297a3b0SGarrett D'Amore 		wch = ch;
1204297a3b0SGarrett D'Amore 	} else if (ch >= 0x81 && ch <= 0xfe) {
1214297a3b0SGarrett D'Amore 		wch = ch;
1224297a3b0SGarrett D'Amore 		if (n < 2)
1234297a3b0SGarrett D'Amore 			return ((size_t)-2);
1244297a3b0SGarrett D'Amore 		ch = (unsigned char)*s++;
1254297a3b0SGarrett D'Amore 		if ((ch >= 0x40 && ch <= 0x7e) || (ch >= 0x80 && ch <= 0xfe)) {
1264297a3b0SGarrett D'Amore 			wch = (wch << 8) | ch;
1274297a3b0SGarrett D'Amore 			len = 2;
1284297a3b0SGarrett D'Amore 		} else if (ch >= 0x30 && ch <= 0x39) {
1294297a3b0SGarrett D'Amore 			/*
1304297a3b0SGarrett D'Amore 			 * Strip high bit off the wide character we will
1314297a3b0SGarrett D'Amore 			 * eventually output so that it is positive when
1324297a3b0SGarrett D'Amore 			 * cast to wint_t on 32-bit twos-complement machines.
1334297a3b0SGarrett D'Amore 			 */
1344297a3b0SGarrett D'Amore 			wch = ((wch & 0x7f) << 8) | ch;
1354297a3b0SGarrett D'Amore 			if (n < 3)
1364297a3b0SGarrett D'Amore 				return ((size_t)-2);
1374297a3b0SGarrett D'Amore 			ch = (unsigned char)*s++;
1384297a3b0SGarrett D'Amore 			if (ch < 0x81 || ch > 0xfe)
1394297a3b0SGarrett D'Amore 				goto ilseq;
1404297a3b0SGarrett D'Amore 			wch = (wch << 8) | ch;
1414297a3b0SGarrett D'Amore 			if (n < 4)
1424297a3b0SGarrett D'Amore 				return ((size_t)-2);
1434297a3b0SGarrett D'Amore 			ch = (unsigned char)*s++;
1444297a3b0SGarrett D'Amore 			if (ch < 0x30 || ch > 0x39)
1454297a3b0SGarrett D'Amore 				goto ilseq;
1464297a3b0SGarrett D'Amore 			wch = (wch << 8) | ch;
1474297a3b0SGarrett D'Amore 			len = 4;
1484297a3b0SGarrett D'Amore 		} else
1494297a3b0SGarrett D'Amore 			goto ilseq;
1504297a3b0SGarrett D'Amore 	} else
1514297a3b0SGarrett D'Amore 		goto ilseq;
1524297a3b0SGarrett D'Amore 
1534297a3b0SGarrett D'Amore 	if (pwc != NULL)
1544297a3b0SGarrett D'Amore 		*pwc = wch;
1554297a3b0SGarrett D'Amore 	gs->count = 0;
1560ac311baSRobert Mustacchi 	if (zero || wch != L'\0') {
1570ac311baSRobert Mustacchi 		return (len - ocount);
1580ac311baSRobert Mustacchi 	} else {
1590ac311baSRobert Mustacchi 		return (0);
1600ac311baSRobert Mustacchi 	}
1614297a3b0SGarrett D'Amore ilseq:
1624297a3b0SGarrett D'Amore 	errno = EILSEQ;
1634297a3b0SGarrett D'Amore 	return ((size_t)-1);
1644297a3b0SGarrett D'Amore }
1654297a3b0SGarrett D'Amore 
1664297a3b0SGarrett D'Amore static size_t
_GB18030_wcrtomb(char * _RESTRICT_KYWD s,wchar_t wc,mbstate_t * _RESTRICT_KYWD ps)1674297a3b0SGarrett D'Amore _GB18030_wcrtomb(char *_RESTRICT_KYWD s, wchar_t wc,
1684297a3b0SGarrett D'Amore     mbstate_t *_RESTRICT_KYWD ps)
1694297a3b0SGarrett D'Amore {
1704297a3b0SGarrett D'Amore 	_GB18030State *gs;
1714297a3b0SGarrett D'Amore 	size_t len;
1724297a3b0SGarrett D'Amore 	int c;
1734297a3b0SGarrett D'Amore 
1744297a3b0SGarrett D'Amore 	gs = (_GB18030State *)ps;
1754297a3b0SGarrett D'Amore 
1764297a3b0SGarrett D'Amore 	if (gs->count != 0) {
1774297a3b0SGarrett D'Amore 		errno = EINVAL;
1784297a3b0SGarrett D'Amore 		return ((size_t)-1);
1794297a3b0SGarrett D'Amore 	}
1804297a3b0SGarrett D'Amore 
1814297a3b0SGarrett D'Amore 	if (s == NULL)
1824297a3b0SGarrett D'Amore 		/* Reset to initial shift state (no-op) */
1834297a3b0SGarrett D'Amore 		return (1);
1844297a3b0SGarrett D'Amore 	if ((wc & ~0x7fffffff) != 0)
1854297a3b0SGarrett D'Amore 		goto ilseq;
1864297a3b0SGarrett D'Amore 	if (wc & 0x7f000000) {
1874297a3b0SGarrett D'Amore 		/* Replace high bit that mbrtowc() removed. */
1884297a3b0SGarrett D'Amore 		wc |= 0x80000000;
1894297a3b0SGarrett D'Amore 		c = (wc >> 24) & 0xff;
1904297a3b0SGarrett D'Amore 		if (c < 0x81 || c > 0xfe)
1914297a3b0SGarrett D'Amore 			goto ilseq;
1924297a3b0SGarrett D'Amore 		*s++ = c;
1934297a3b0SGarrett D'Amore 		c = (wc >> 16) & 0xff;
1944297a3b0SGarrett D'Amore 		if (c < 0x30 || c > 0x39)
1954297a3b0SGarrett D'Amore 			goto ilseq;
1964297a3b0SGarrett D'Amore 		*s++ = c;
1974297a3b0SGarrett D'Amore 		c = (wc >> 8) & 0xff;
1984297a3b0SGarrett D'Amore 		if (c < 0x81 || c > 0xfe)
1994297a3b0SGarrett D'Amore 			goto ilseq;
2004297a3b0SGarrett D'Amore 		*s++ = c;
2014297a3b0SGarrett D'Amore 		c = wc & 0xff;
2024297a3b0SGarrett D'Amore 		if (c < 0x30 || c > 0x39)
2034297a3b0SGarrett D'Amore 			goto ilseq;
2044297a3b0SGarrett D'Amore 		*s++ = c;
2054297a3b0SGarrett D'Amore 		len = 4;
2064297a3b0SGarrett D'Amore 	} else if (wc & 0x00ff0000)
2074297a3b0SGarrett D'Amore 		goto ilseq;
2084297a3b0SGarrett D'Amore 	else if (wc & 0x0000ff00) {
2094297a3b0SGarrett D'Amore 		c = (wc >> 8) & 0xff;
2104297a3b0SGarrett D'Amore 		if (c < 0x81 || c > 0xfe)
2114297a3b0SGarrett D'Amore 			goto ilseq;
2124297a3b0SGarrett D'Amore 		*s++ = c;
2134297a3b0SGarrett D'Amore 		c = wc & 0xff;
2144297a3b0SGarrett D'Amore 		if (c < 0x40 || c == 0x7f || c == 0xff)
2154297a3b0SGarrett D'Amore 			goto ilseq;
2164297a3b0SGarrett D'Amore 		*s++ = c;
2174297a3b0SGarrett D'Amore 		len = 2;
2184297a3b0SGarrett D'Amore 	} else if (wc <= 0x7f) {
2194297a3b0SGarrett D'Amore 		*s++ = wc;
2204297a3b0SGarrett D'Amore 		len = 1;
2214297a3b0SGarrett D'Amore 	} else
2224297a3b0SGarrett D'Amore 		goto ilseq;
2234297a3b0SGarrett D'Amore 
2244297a3b0SGarrett D'Amore 	return (len);
2254297a3b0SGarrett D'Amore ilseq:
2264297a3b0SGarrett D'Amore 	errno = EILSEQ;
2274297a3b0SGarrett D'Amore 	return ((size_t)-1);
2284297a3b0SGarrett D'Amore }
2292d08521bSGarrett D'Amore 
2302d08521bSGarrett D'Amore static size_t
_GB18030_mbsnrtowcs(wchar_t * _RESTRICT_KYWD dst,const char ** _RESTRICT_KYWD src,size_t nms,size_t len,mbstate_t * _RESTRICT_KYWD ps)2312d08521bSGarrett D'Amore _GB18030_mbsnrtowcs(wchar_t *_RESTRICT_KYWD dst,
2322d08521bSGarrett D'Amore     const char **_RESTRICT_KYWD src, size_t nms, size_t len,
2332d08521bSGarrett D'Amore     mbstate_t *_RESTRICT_KYWD ps)
2342d08521bSGarrett D'Amore {
2352d08521bSGarrett D'Amore 	return (__mbsnrtowcs_std(dst, src, nms, len, ps, _GB18030_mbrtowc));
2362d08521bSGarrett D'Amore }
2372d08521bSGarrett D'Amore 
2382d08521bSGarrett D'Amore static size_t
_GB18030_wcsnrtombs(char * _RESTRICT_KYWD dst,const wchar_t ** _RESTRICT_KYWD src,size_t nwc,size_t len,mbstate_t * _RESTRICT_KYWD ps)2392d08521bSGarrett D'Amore _GB18030_wcsnrtombs(char *_RESTRICT_KYWD dst,
2402d08521bSGarrett D'Amore     const wchar_t **_RESTRICT_KYWD src, size_t nwc, size_t len,
2412d08521bSGarrett D'Amore     mbstate_t *_RESTRICT_KYWD ps)
2422d08521bSGarrett D'Amore {
2432d08521bSGarrett D'Amore 	return (__wcsnrtombs_std(dst, src, nwc, len, ps, _GB18030_wcrtomb));
2442d08521bSGarrett D'Amore }
245