[apple/libc.git] / locale / FreeBSD / utf8.c

/*-
 * Copyright (c) 2002-2004 Tim J. Robbins
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#include <sys/param.h>
__FBSDID("$FreeBSD: src/lib/libc/locale/utf8.c,v 1.16 2007/10/15 09:51:30 ache Exp $");

#include "xlocale_private.h"

#include <errno.h>
#include <limits.h>
#include <runetype.h>
#include <stdlib.h>
#include <string.h>
#include <wchar.h>
#include "mblocal.h"

#define UTF8_MB_CUR_MAX		6

static size_t	_UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict,
		    size_t, mbstate_t * __restrict, locale_t);
static int	_UTF8_mbsinit(const mbstate_t *, locale_t);
static size_t	_UTF8_mbsnrtowcs(wchar_t * __restrict,
		    const char ** __restrict, size_t, size_t,
		    mbstate_t * __restrict, locale_t);
static size_t	_UTF8_wcrtomb(char * __restrict, wchar_t,
		    mbstate_t * __restrict, locale_t);
static size_t	_UTF8_wcsnrtombs(char * __restrict, const wchar_t ** __restrict,
		    size_t, size_t, mbstate_t * __restrict, locale_t);

typedef struct {
	wchar_t	ch;
	int	want;
	wchar_t	lbound;
} _UTF8State;

__private_extern__ int
_UTF8_init(struct __xlocale_st_runelocale *xrl)
{

	xrl->__mbrtowc = _UTF8_mbrtowc;
	xrl->__wcrtomb = _UTF8_wcrtomb;
	xrl->__mbsinit = _UTF8_mbsinit;
	xrl->__mbsnrtowcs = _UTF8_mbsnrtowcs;
	xrl->__wcsnrtombs = _UTF8_wcsnrtombs;
	xrl->__mb_cur_max = UTF8_MB_CUR_MAX;
	/*
	 * UCS-4 encoding used as the internal representation, so
	 * slots 0x0080-0x00FF are occuped and must be excluded
	 * from the single byte ctype by setting the limit.
	 */
	xrl->__mb_sb_limit = 128;

	return (0);
}

static int
_UTF8_mbsinit(const mbstate_t *ps, locale_t loc)
{

	return (ps == NULL || ((const _UTF8State *)ps)->want == 0);
}

static size_t
_UTF8_mbrtowc(wchar_t * __restrict pwc, const char * __restrict s, size_t n,
    mbstate_t * __restrict ps, locale_t loc)
{
	_UTF8State *us;
	int ch, i, mask, want;
	wchar_t lbound, wch;

	us = (_UTF8State *)ps;

	if (us->want < 0 || us->want > 6) {
		errno = EINVAL;
		return ((size_t)-1);
	}

	if (s == NULL) {
		s = "";
		n = 1;
		pwc = NULL;
	}

	if (n == 0)
		/* Incomplete multibyte sequence */
		return ((size_t)-2);

	if (us->want == 0 && ((ch = (unsigned char)*s) & ~0x7f) == 0) {
		/* Fast path for plain ASCII characters. */
		if (pwc != NULL)
			*pwc = ch;
		return (ch != '\0' ? 1 : 0);
	}

	if (us->want == 0) {
		/*
		 * Determine the number of octets that make up this character
		 * from the first octet, and a mask that extracts the
		 * interesting bits of the first octet. We already know
		 * the character is at least two bytes long.
		 *
		 * We also specify a lower bound for the character code to
		 * detect redundant, non-"shortest form" encodings. For
		 * example, the sequence C0 80 is _not_ a legal representation
		 * of the null character. This enforces a 1-to-1 mapping
		 * between character codes and their multibyte representations.
		 */
		ch = (unsigned char)*s;
		if ((ch & 0x80) == 0) {
			mask = 0x7f;
			want = 1;
			lbound = 0;
		} else if ((ch & 0xe0) == 0xc0) {
			mask = 0x1f;
			want = 2;
			lbound = 0x80;
		} else if ((ch & 0xf0) == 0xe0) {
			mask = 0x0f;
			want = 3;
			lbound = 0x800;
		} else if ((ch & 0xf8) == 0xf0) {
			mask = 0x07;
			want = 4;
			lbound = 0x10000;
		} else if ((ch & 0xfc) == 0xf8) {
			mask = 0x03;
			want = 5;
			lbound = 0x200000;
		} else if ((ch & 0xfe) == 0xfc) {
			mask = 0x01;
			want = 6;
			lbound = 0x4000000;
		} else {
			/*
			 * Malformed input; input is not UTF-8.
			 */
			errno = EILSEQ;
			return ((size_t)-1);
		}
	} else {
		want = us->want;
		lbound = us->lbound;
	}

	/*
	 * Decode the octet sequence representing the character in chunks
	 * of 6 bits, most significant first.
	 */
	if (us->want == 0)
		wch = (unsigned char)*s++ & mask;
	else
		wch = us->ch;
	for (i = (us->want == 0) ? 1 : 0; i < MIN(want, n); i++) {
		if ((*s & 0xc0) != 0x80) {
			/*
			 * Malformed input; bad characters in the middle
			 * of a character.
			 */
			errno = EILSEQ;
			return ((size_t)-1);
		}
		wch <<= 6;
		wch |= *s++ & 0x3f;
	}
	if (i < want) {
		/* Incomplete multibyte sequence. */
		us->want = want - i;
		us->lbound = lbound;
		us->ch = wch;
		return ((size_t)-2);
	}
	if (wch < lbound) {
		/*
		 * Malformed input; redundant encoding.
		 */
		errno = EILSEQ;
		return ((size_t)-1);
	}
	if (pwc != NULL)
		*pwc = wch;
	us->want = 0;
	return (wch == L'\0' ? 0 : want);
}

static size_t
_UTF8_mbsnrtowcs(wchar_t * __restrict dst, const char ** __restrict src,
    size_t nms, size_t len, mbstate_t * __restrict ps, locale_t loc)
{
	_UTF8State *us;
	const char *s;
	size_t nchr;
	wchar_t wc;
	size_t nb;

	us = (_UTF8State *)ps;

	s = *src;
	nchr = 0;

	if (dst == NULL) {
		/*
		 * The fast path in the loop below is not safe if an ASCII
		 * character appears as anything but the first byte of a
		 * multibyte sequence. Check now to avoid doing it in the loop.
		 */
		if (nms > 0 && us->want > 0 && (signed char)*s > 0) {
			errno = EILSEQ;
			return ((size_t)-1);
		}
		for (;;) {
			if (nms > 0 && (signed char)*s > 0)
				/*
				 * Fast path for plain ASCII characters
				 * excluding NUL.
				 */
				nb = 1;
			else if ((nb = _UTF8_mbrtowc(&wc, s, nms, ps, loc)) ==
			    (size_t)-1)
				/* Invalid sequence - mbrtowc() sets errno. */
				return ((size_t)-1);
			else if (nb == 0 || nb == (size_t)-2)
				return (nchr);
			s += nb;
			nms -= nb;
			nchr++;
		}
		/*NOTREACHED*/
	}

	/*
	 * The fast path in the loop below is not safe if an ASCII
	 * character appears as anything but the first byte of a
	 * multibyte sequence. Check now to avoid doing it in the loop.
	 */
	if (nms > 0 && len > 0 && us->want > 0 && (signed char)*s > 0) {
		errno = EILSEQ;
		return ((size_t)-1);
	}
	while (len-- > 0) {
		if (nms > 0 && (signed char)*s > 0) {
			/*
			 * Fast path for plain ASCII characters
			 * excluding NUL.
			 */
			*dst = (wchar_t)*s;
			nb = 1;
		} else if ((nb = _UTF8_mbrtowc(dst, s, nms, ps, loc)) ==
		    (size_t)-1) {
			*src = s;
			return ((size_t)-1);
		} else if (nb == (size_t)-2) {
			*src = s + nms;
			return (nchr);
		} else if (nb == 0) {
			*src = NULL;
			return (nchr);
		}
		s += nb;
		nms -= nb;
		nchr++;
		dst++;
	}
	*src = s;
	return (nchr);
}

static size_t
_UTF8_wcrtomb(char * __restrict s, wchar_t wc, mbstate_t * __restrict ps, locale_t loc)
{
	_UTF8State *us;
	unsigned char lead;
	int i, len;

	us = (_UTF8State *)ps;

	if (us->want != 0) {
		errno = EINVAL;
		return ((size_t)-1);
	}

	if (s == NULL)
		/* Reset to initial shift state (no-op) */
		return (1);

	if ((wc & ~0x7f) == 0) {
		/* Fast path for plain ASCII characters. */
		*s = (char)wc;
		return (1);
	}

	/*
	 * Determine the number of octets needed to represent this character.
	 * We always output the shortest sequence possible. Also specify the
	 * first few bits of the first octet, which contains the information
	 * about the sequence length.
	 */
	if ((wc & ~0x7f) == 0) {
		lead = 0;
		len = 1;
	} else if ((wc & ~0x7ff) == 0) {
		lead = 0xc0;
		len = 2;
	} else if ((wc & ~0xffff) == 0) {
		lead = 0xe0;
		len = 3;
	} else if ((wc & ~0x1fffff) == 0) {
		lead = 0xf0;
		len = 4;
	} else if ((wc & ~0x3ffffff) == 0) {
		lead = 0xf8;
		len = 5;
	} else if ((wc & ~0x7fffffff) == 0) {
		lead = 0xfc;
		len = 6;
	} else {
		errno = EILSEQ;
		return ((size_t)-1);
	}

	/*
	 * Output the octets representing the character in chunks
	 * of 6 bits, least significant last. The first octet is
	 * a special case because it contains the sequence length
	 * information.
	 */
	for (i = len - 1; i > 0; i--) {
		s[i] = (wc & 0x3f) | 0x80;
		wc >>= 6;
	}
	*s = (wc & 0xff) | lead;

	return (len);
}

static size_t
_UTF8_wcsnrtombs(char * __restrict dst, const wchar_t ** __restrict src,
    size_t nwc, size_t len, mbstate_t * __restrict ps, locale_t loc)
{
	_UTF8State *us;
	char buf[MB_LEN_MAX];
	const wchar_t *s;
	size_t nbytes;
	size_t nb;

	us = (_UTF8State *)ps;

	if (us->want != 0) {
		errno = EINVAL;
		return ((size_t)-1);
	}

	s = *src;
	nbytes = 0;

	if (dst == NULL) {
		while (nwc-- > 0) {
			if (0 <= *s && *s < 0x80)
				/* Fast path for plain ASCII characters. */
				nb = 1;
			else if ((nb = _UTF8_wcrtomb(buf, *s, ps, loc)) ==
			    (size_t)-1)
				/* Invalid character - wcrtomb() sets errno. */
				return ((size_t)-1);
			if (*s == L'\0')
				return (nbytes + nb - 1);
			s++;
			nbytes += nb;
		}
		return (nbytes);
	}

	while (len > 0 && nwc-- > 0) {
		if (0 <= *s && *s < 0x80) {
			/* Fast path for plain ASCII characters. */
			nb = 1;
			*dst = *s;
		} else if (len > (size_t)UTF8_MB_CUR_MAX) {
			/* Enough space to translate in-place. */
			if ((nb = _UTF8_wcrtomb(dst, *s, ps, loc)) == (size_t)-1) {
				*src = s;
				return ((size_t)-1);
			}
		} else {
			/*
			 * May not be enough space; use temp. buffer.
			 */
			if ((nb = _UTF8_wcrtomb(buf, *s, ps, loc)) == (size_t)-1) {
				*src = s;
				return ((size_t)-1);
			}
			if (nb > (int)len)
				/* MB sequence for character won't fit. */
				break;
			memcpy(dst, buf, nb);
		}
		if (*s == L'\0') {
			*src = NULL;
			return (nbytes + nb - 1);
		}
		s++;
		dst += nb;
		len -= nb;
		nbytes += nb;
	}
	*src = s;
	return (nbytes);
}
Commit	Line	Data
9385eb3d	1	/*-
3d9156a7	2	* Copyright (c) 2002-2004 Tim J. Robbins
9385eb3d A	3	* All rights reserved.
	4	*
	5	* Redistribution and use in source and binary forms, with or without
	6	* modification, are permitted provided that the following conditions
	7	* are met:
	8	* 1. Redistributions of source code must retain the above copyright
	9	* notice, this list of conditions and the following disclaimer.
	10	* 2. Redistributions in binary form must reproduce the above copyright
	11	* notice, this list of conditions and the following disclaimer in the
	12	* documentation and/or other materials provided with the distribution.
	13	*
	14	* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
	15	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	16	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	17	* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
	18	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	19	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	20	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	21	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	22	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	23	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	24	* SUCH DAMAGE.
	25	*/
	26
3d9156a7	27	#include <sys/param.h>
1f2f436a	28	__FBSDID("$FreeBSD: src/lib/libc/locale/utf8.c,v 1.16 2007/10/15 09:51:30 ache Exp $");
9385eb3d	29
ad3c9f2a A	30	#include "xlocale_private.h"
ad3c9f2a A	31
3d9156a7 A	32	#include <errno.h>
	33	#include <limits.h>
	34	#include <runetype.h>
9385eb3d	35	#include <stdlib.h>
3d9156a7 A	36	#include <string.h>
	37	#include <wchar.h>
	38	#include "mblocal.h"
9385eb3d	39
ad3c9f2a	40	#define UTF8_MB_CUR_MAX 6
1f2f436a A	41
1f2f436a A	42	static size_t _UTF8_mbrtowc(wchar_t * __restrict, const char * __restrict,
ad3c9f2a A	43	size_t, mbstate_t * __restrict, locale_t);
ad3c9f2a A	44	static int _UTF8_mbsinit(const mbstate_t *, locale_t);
1f2f436a A	45	static size_t _UTF8_mbsnrtowcs(wchar_t * __restrict,
1f2f436a A	46	const char ** __restrict, size_t, size_t,
ad3c9f2a	47	mbstate_t * __restrict, locale_t);
1f2f436a	48	static size_t _UTF8_wcrtomb(char * __restrict, wchar_t,
ad3c9f2a	49	mbstate_t * __restrict, locale_t);
1f2f436a	50	static size_t _UTF8_wcsnrtombs(char * __restrict, const wchar_t ** __restrict,
ad3c9f2a	51	size_t, size_t, mbstate_t * __restrict, locale_t);
3d9156a7 A	52
	53	typedef struct {
	54	wchar_t ch;
	55	int want;
	56	wchar_t lbound;
	57	} _UTF8State;
9385eb3d	58
ad3c9f2a A	59	__private_extern__ int
ad3c9f2a A	60	_UTF8_init(struct __xlocale_st_runelocale *xrl)
9385eb3d A	61	{
9385eb3d A	62
ad3c9f2a A	63	xrl->__mbrtowc = _UTF8_mbrtowc;
	64	xrl->__wcrtomb = _UTF8_wcrtomb;
	65	xrl->__mbsinit = _UTF8_mbsinit;
	66	xrl->__mbsnrtowcs = _UTF8_mbsnrtowcs;
	67	xrl->__wcsnrtombs = _UTF8_wcsnrtombs;
	68	xrl->__mb_cur_max = UTF8_MB_CUR_MAX;
1f2f436a A	69	/*
	70	* UCS-4 encoding used as the internal representation, so
	71	* slots 0x0080-0x00FF are occuped and must be excluded
	72	* from the single byte ctype by setting the limit.
	73	*/
ad3c9f2a	74	xrl->__mb_sb_limit = 128;
9385eb3d A	75
	76	return (0);
	77	}
	78
1f2f436a	79	static int
ad3c9f2a	80	_UTF8_mbsinit(const mbstate_t *ps, locale_t loc)
3d9156a7 A	81	{
	82
	83	return (ps == NULL \|\| ((const _UTF8State *)ps)->want == 0);
	84	}
	85
1f2f436a	86	static size_t
3d9156a7	87	_UTF8_mbrtowc(wchar_t * __restrict pwc, const char * __restrict s, size_t n,
ad3c9f2a	88	mbstate_t * __restrict ps, locale_t loc)
9385eb3d	89	{
3d9156a7 A	90	_UTF8State *us;
	91	int ch, i, mask, want;
	92	wchar_t lbound, wch;
	93
	94	us = (_UTF8State *)ps;
9385eb3d	95
3d9156a7 A	96	if (us->want < 0 \|\| us->want > 6) {
	97	errno = EINVAL;
	98	return ((size_t)-1);
9385eb3d A	99	}
9385eb3d A	100
3d9156a7 A	101	if (s == NULL) {
	102	s = "";
	103	n = 1;
	104	pwc = NULL;
9385eb3d A	105	}
9385eb3d A	106
3d9156a7 A	107	if (n == 0)
	108	/* Incomplete multibyte sequence */
	109	return ((size_t)-2);
	110
	111	if (us->want == 0 && ((ch = (unsigned char)*s) & ~0x7f) == 0) {
	112	/* Fast path for plain ASCII characters. */
	113	if (pwc != NULL)
	114	*pwc = ch;
	115	return (ch != '\0' ? 1 : 0);
	116	}
	117
	118	if (us->want == 0) {
9385eb3d	119	/*
3d9156a7 A	120	* Determine the number of octets that make up this character
	121	* from the first octet, and a mask that extracts the
	122	* interesting bits of the first octet. We already know
	123	* the character is at least two bytes long.
	124	*
	125	* We also specify a lower bound for the character code to
	126	* detect redundant, non-"shortest form" encodings. For
	127	* example, the sequence C0 80 is _not_ a legal representation
	128	* of the null character. This enforces a 1-to-1 mapping
	129	* between character codes and their multibyte representations.
9385eb3d	130	*/
3d9156a7 A	131	ch = (unsigned char)*s;
	132	if ((ch & 0x80) == 0) {
	133	mask = 0x7f;
	134	want = 1;
	135	lbound = 0;
	136	} else if ((ch & 0xe0) == 0xc0) {
	137	mask = 0x1f;
	138	want = 2;
	139	lbound = 0x80;
	140	} else if ((ch & 0xf0) == 0xe0) {
	141	mask = 0x0f;
	142	want = 3;
	143	lbound = 0x800;
	144	} else if ((ch & 0xf8) == 0xf0) {
	145	mask = 0x07;
	146	want = 4;
	147	lbound = 0x10000;
	148	} else if ((ch & 0xfc) == 0xf8) {
	149	mask = 0x03;
	150	want = 5;
	151	lbound = 0x200000;
1f2f436a	152	} else if ((ch & 0xfe) == 0xfc) {
3d9156a7 A	153	mask = 0x01;
	154	want = 6;
	155	lbound = 0x4000000;
	156	} else {
	157	/*
	158	* Malformed input; input is not UTF-8.
	159	*/
	160	errno = EILSEQ;
	161	return ((size_t)-1);
	162	}
	163	} else {
	164	want = us->want;
	165	lbound = us->lbound;
9385eb3d A	166	}
	167
	168	/*
	169	* Decode the octet sequence representing the character in chunks
	170	* of 6 bits, most significant first.
	171	*/
3d9156a7 A	172	if (us->want == 0)
	173	wch = (unsigned char)*s++ & mask;
	174	else
	175	wch = us->ch;
	176	for (i = (us->want == 0) ? 1 : 0; i < MIN(want, n); i++) {
	177	if ((*s & 0xc0) != 0x80) {
9385eb3d A	178	/*
	179	* Malformed input; bad characters in the middle
	180	* of a character.
	181	*/
3d9156a7 A	182	errno = EILSEQ;
3d9156a7 A	183	return ((size_t)-1);
9385eb3d A	184	}
9385eb3d A	185	wch <<= 6;
3d9156a7	186	wch \|= *s++ & 0x3f;
9385eb3d	187	}
3d9156a7 A	188	if (i < want) {
	189	/* Incomplete multibyte sequence. */
	190	us->want = want - i;
	191	us->lbound = lbound;
	192	us->ch = wch;
	193	return ((size_t)-2);
	194	}
	195	if (wch < lbound) {
9385eb3d A	196	/*
	197	* Malformed input; redundant encoding.
	198	*/
3d9156a7 A	199	errno = EILSEQ;
	200	return ((size_t)-1);
	201	}
	202	if (pwc != NULL)
	203	*pwc = wch;
	204	us->want = 0;
	205	return (wch == L'\0' ? 0 : want);
9385eb3d A	206	}
9385eb3d A	207
1f2f436a	208	static size_t
3d9156a7	209	_UTF8_mbsnrtowcs(wchar_t * __restrict dst, const char ** __restrict src,
ad3c9f2a	210	size_t nms, size_t len, mbstate_t * __restrict ps, locale_t loc)
9385eb3d	211	{
3d9156a7 A	212	_UTF8State *us;
	213	const char *s;
	214	size_t nchr;
	215	wchar_t wc;
	216	size_t nb;
	217
	218	us = (_UTF8State *)ps;
	219
	220	s = *src;
	221	nchr = 0;
	222
	223	if (dst == NULL) {
	224	/*
	225	* The fast path in the loop below is not safe if an ASCII
	226	* character appears as anything but the first byte of a
	227	* multibyte sequence. Check now to avoid doing it in the loop.
	228	*/
	229	if (nms > 0 && us->want > 0 && (signed char)*s > 0) {
	230	errno = EILSEQ;
	231	return ((size_t)-1);
	232	}
	233	for (;;) {
	234	if (nms > 0 && (signed char)*s > 0)
	235	/*
	236	* Fast path for plain ASCII characters
	237	* excluding NUL.
	238	*/
	239	nb = 1;
ad3c9f2a	240	else if ((nb = _UTF8_mbrtowc(&wc, s, nms, ps, loc)) ==
3d9156a7 A	241	(size_t)-1)
	242	/* Invalid sequence - mbrtowc() sets errno. */
	243	return ((size_t)-1);
	244	else if (nb == 0 \|\| nb == (size_t)-2)
	245	return (nchr);
	246	s += nb;
	247	nms -= nb;
	248	nchr++;
	249	}
	250	/NOTREACHED/
	251	}
	252
	253	/*
	254	* The fast path in the loop below is not safe if an ASCII
	255	* character appears as anything but the first byte of a
	256	* multibyte sequence. Check now to avoid doing it in the loop.
	257	*/
	258	if (nms > 0 && len > 0 && us->want > 0 && (signed char)*s > 0) {
	259	errno = EILSEQ;
	260	return ((size_t)-1);
	261	}
	262	while (len-- > 0) {
	263	if (nms > 0 && (signed char)*s > 0) {
	264	/*
	265	* Fast path for plain ASCII characters
	266	* excluding NUL.
	267	*/
	268	dst = (wchar_t)s;
	269	nb = 1;
ad3c9f2a	270	} else if ((nb = _UTF8_mbrtowc(dst, s, nms, ps, loc)) ==
3d9156a7 A	271	(size_t)-1) {
	272	*src = s;
	273	return ((size_t)-1);
	274	} else if (nb == (size_t)-2) {
	275	*src = s + nms;
	276	return (nchr);
	277	} else if (nb == 0) {
	278	*src = NULL;
	279	return (nchr);
	280	}
	281	s += nb;
	282	nms -= nb;
	283	nchr++;
	284	dst++;
	285	}
	286	*src = s;
	287	return (nchr);
	288	}
	289
1f2f436a	290	static size_t
ad3c9f2a	291	_UTF8_wcrtomb(char * __restrict s, wchar_t wc, mbstate_t * __restrict ps, locale_t loc)
3d9156a7 A	292	{
3d9156a7 A	293	_UTF8State *us;
9385eb3d A	294	unsigned char lead;
	295	int i, len;
	296
3d9156a7 A	297	us = (_UTF8State *)ps;
	298
	299	if (us->want != 0) {
	300	errno = EINVAL;
	301	return ((size_t)-1);
	302	}
	303
	304	if (s == NULL)
	305	/* Reset to initial shift state (no-op) */
	306	return (1);
	307
	308	if ((wc & ~0x7f) == 0) {
	309	/* Fast path for plain ASCII characters. */
	310	*s = (char)wc;
	311	return (1);
	312	}
	313
9385eb3d A	314	/*
	315	* Determine the number of octets needed to represent this character.
	316	* We always output the shortest sequence possible. Also specify the
	317	* first few bits of the first octet, which contains the information
	318	* about the sequence length.
	319	*/
3d9156a7	320	if ((wc & ~0x7f) == 0) {
9385eb3d A	321	lead = 0;
9385eb3d A	322	len = 1;
3d9156a7	323	} else if ((wc & ~0x7ff) == 0) {
9385eb3d A	324	lead = 0xc0;
9385eb3d A	325	len = 2;
3d9156a7	326	} else if ((wc & ~0xffff) == 0) {
9385eb3d A	327	lead = 0xe0;
9385eb3d A	328	len = 3;
3d9156a7	329	} else if ((wc & ~0x1fffff) == 0) {
9385eb3d A	330	lead = 0xf0;
9385eb3d A	331	len = 4;
3d9156a7	332	} else if ((wc & ~0x3ffffff) == 0) {
9385eb3d A	333	lead = 0xf8;
9385eb3d A	334	len = 5;
3d9156a7	335	} else if ((wc & ~0x7fffffff) == 0) {
9385eb3d A	336	lead = 0xfc;
	337	len = 6;
	338	} else {
3d9156a7 A	339	errno = EILSEQ;
3d9156a7 A	340	return ((size_t)-1);
9385eb3d A	341	}
9385eb3d A	342
3d9156a7 A	343	/*
	344	* Output the octets representing the character in chunks
	345	* of 6 bits, least significant last. The first octet is
	346	* a special case because it contains the sequence length
	347	* information.
	348	*/
	349	for (i = len - 1; i > 0; i--) {
	350	s[i] = (wc & 0x3f) \| 0x80;
	351	wc >>= 6;
9385eb3d	352	}
3d9156a7	353	*s = (wc & 0xff) \| lead;
9385eb3d A	354
	355	return (len);
	356	}
3d9156a7	357
1f2f436a	358	static size_t
3d9156a7	359	_UTF8_wcsnrtombs(char * __restrict dst, const wchar_t ** __restrict src,
ad3c9f2a	360	size_t nwc, size_t len, mbstate_t * __restrict ps, locale_t loc)
3d9156a7 A	361	{
	362	_UTF8State *us;
	363	char buf[MB_LEN_MAX];
	364	const wchar_t *s;
	365	size_t nbytes;
	366	size_t nb;
	367
	368	us = (_UTF8State *)ps;
	369
	370	if (us->want != 0) {
	371	errno = EINVAL;
	372	return ((size_t)-1);
	373	}
	374
	375	s = *src;
	376	nbytes = 0;
	377
	378	if (dst == NULL) {
	379	while (nwc-- > 0) {
	380	if (0 <= s && s < 0x80)
	381	/* Fast path for plain ASCII characters. */
	382	nb = 1;
ad3c9f2a	383	else if ((nb = _UTF8_wcrtomb(buf, *s, ps, loc)) ==
3d9156a7 A	384	(size_t)-1)
	385	/* Invalid character - wcrtomb() sets errno. */
	386	return ((size_t)-1);
	387	if (*s == L'\0')
	388	return (nbytes + nb - 1);
	389	s++;
	390	nbytes += nb;
	391	}
	392	return (nbytes);
	393	}
	394
	395	while (len > 0 && nwc-- > 0) {
	396	if (0 <= s && s < 0x80) {
	397	/* Fast path for plain ASCII characters. */
	398	nb = 1;
	399	dst = s;
ad3c9f2a	400	} else if (len > (size_t)UTF8_MB_CUR_MAX) {
3d9156a7	401	/* Enough space to translate in-place. */
ad3c9f2a	402	if ((nb = _UTF8_wcrtomb(dst, *s, ps, loc)) == (size_t)-1) {
3d9156a7 A	403	*src = s;
	404	return ((size_t)-1);
	405	}
	406	} else {
	407	/*
	408	* May not be enough space; use temp. buffer.
	409	*/
ad3c9f2a	410	if ((nb = _UTF8_wcrtomb(buf, *s, ps, loc)) == (size_t)-1) {
3d9156a7 A	411	*src = s;
	412	return ((size_t)-1);
	413	}
	414	if (nb > (int)len)
	415	/* MB sequence for character won't fit. */
	416	break;
	417	memcpy(dst, buf, nb);
	418	}
	419	if (*s == L'\0') {
	420	*src = NULL;
	421	return (nbytes + nb - 1);
	422	}
	423	s++;
	424	dst += nb;
	425	len -= nb;
	426	nbytes += nb;
	427	}
	428	*src = s;
	429	return (nbytes);
	430	}