1*0a6a1f1dSLionel Sambuc /* $NetBSD: citrus_hz.c,v 1.4 2014/06/24 22:24:18 spz Exp $ */
22fe8fb19SBen Gras
32fe8fb19SBen Gras /*-
42fe8fb19SBen Gras * Copyright (c)2004, 2006 Citrus Project,
52fe8fb19SBen Gras * All rights reserved.
62fe8fb19SBen Gras *
72fe8fb19SBen Gras * Redistribution and use in source and binary forms, with or without
82fe8fb19SBen Gras * modification, are permitted provided that the following conditions
92fe8fb19SBen Gras * are met:
102fe8fb19SBen Gras * 1. Redistributions of source code must retain the above copyright
112fe8fb19SBen Gras * notice, this list of conditions and the following disclaimer.
122fe8fb19SBen Gras * 2. Redistributions in binary form must reproduce the above copyright
132fe8fb19SBen Gras * notice, this list of conditions and the following disclaimer in the
142fe8fb19SBen Gras * documentation and/or other materials provided with the distribution.
152fe8fb19SBen Gras *
162fe8fb19SBen Gras * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
172fe8fb19SBen Gras * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
182fe8fb19SBen Gras * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
192fe8fb19SBen Gras * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
202fe8fb19SBen Gras * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
212fe8fb19SBen Gras * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
222fe8fb19SBen Gras * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
232fe8fb19SBen Gras * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
242fe8fb19SBen Gras * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
252fe8fb19SBen Gras * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
262fe8fb19SBen Gras * SUCH DAMAGE.
272fe8fb19SBen Gras *
282fe8fb19SBen Gras */
292fe8fb19SBen Gras
302fe8fb19SBen Gras #include <sys/cdefs.h>
312fe8fb19SBen Gras #if defined(LIBC_SCCS) && !defined(lint)
32*0a6a1f1dSLionel Sambuc __RCSID("$NetBSD: citrus_hz.c,v 1.4 2014/06/24 22:24:18 spz Exp $");
332fe8fb19SBen Gras #endif /* LIBC_SCCS and not lint */
342fe8fb19SBen Gras
352fe8fb19SBen Gras #include <sys/queue.h>
362fe8fb19SBen Gras #include <sys/types.h>
372fe8fb19SBen Gras #include <assert.h>
382fe8fb19SBen Gras #include <errno.h>
392fe8fb19SBen Gras #include <string.h>
402fe8fb19SBen Gras #include <stdint.h>
412fe8fb19SBen Gras #include <stdlib.h>
422fe8fb19SBen Gras #include <stddef.h>
432fe8fb19SBen Gras #include <limits.h>
442fe8fb19SBen Gras #include <wchar.h>
452fe8fb19SBen Gras
462fe8fb19SBen Gras #include "citrus_namespace.h"
472fe8fb19SBen Gras #include "citrus_types.h"
482fe8fb19SBen Gras #include "citrus_bcs.h"
492fe8fb19SBen Gras #include "citrus_module.h"
502fe8fb19SBen Gras #include "citrus_ctype.h"
512fe8fb19SBen Gras #include "citrus_stdenc.h"
522fe8fb19SBen Gras
532fe8fb19SBen Gras #include "citrus_hz.h"
542fe8fb19SBen Gras #include "citrus_prop.h"
552fe8fb19SBen Gras
562fe8fb19SBen Gras /*
572fe8fb19SBen Gras * wchar_t mapping:
582fe8fb19SBen Gras *
592fe8fb19SBen Gras * CTRL/ASCII 00000000 00000000 00000000 gxxxxxxx
602fe8fb19SBen Gras * GB2312 00000000 00000000 0xxxxxxx gxxxxxxx
612fe8fb19SBen Gras * 94/96*n (~M) 0mmmmmmm 0xxxxxxx 0xxxxxxx gxxxxxxx
622fe8fb19SBen Gras */
632fe8fb19SBen Gras
642fe8fb19SBen Gras #define ESCAPE_CHAR '~'
652fe8fb19SBen Gras
662fe8fb19SBen Gras typedef enum {
672fe8fb19SBen Gras CTRL = 0, ASCII = 1, GB2312 = 2, CS94 = 3, CS96 = 4
682fe8fb19SBen Gras } charset_t;
692fe8fb19SBen Gras
702fe8fb19SBen Gras typedef struct {
712fe8fb19SBen Gras int start, end, width;
722fe8fb19SBen Gras } range_t;
732fe8fb19SBen Gras
742fe8fb19SBen Gras static const range_t ranges[] = {
752fe8fb19SBen Gras #define RANGE(start, end) { start, end, (end - start) + 1 }
762fe8fb19SBen Gras /* CTRL */ RANGE(0x00, 0x1F),
772fe8fb19SBen Gras /* ASCII */ RANGE(0x20, 0x7F),
782fe8fb19SBen Gras /* GB2312 */ RANGE(0x21, 0x7E),
792fe8fb19SBen Gras /* CS94 */ RANGE(0x21, 0x7E),
802fe8fb19SBen Gras /* CS96 */ RANGE(0x20, 0x7F),
812fe8fb19SBen Gras #undef RANGE
822fe8fb19SBen Gras };
832fe8fb19SBen Gras
842fe8fb19SBen Gras typedef struct escape_t escape_t;
852fe8fb19SBen Gras typedef struct {
862fe8fb19SBen Gras charset_t charset;
872fe8fb19SBen Gras size_t length;
882fe8fb19SBen Gras #define ROWCOL_MAX 3
892fe8fb19SBen Gras escape_t *escape;
902fe8fb19SBen Gras } graphic_t;
912fe8fb19SBen Gras
922fe8fb19SBen Gras typedef TAILQ_HEAD(escape_list, escape_t) escape_list;
932fe8fb19SBen Gras struct escape_t {
942fe8fb19SBen Gras TAILQ_ENTRY(escape_t) entry;
952fe8fb19SBen Gras int ch;
962fe8fb19SBen Gras graphic_t *left, *right;
972fe8fb19SBen Gras escape_list *set;
982fe8fb19SBen Gras };
992fe8fb19SBen Gras
1002fe8fb19SBen Gras #define GL(escape) ((escape)->left)
1012fe8fb19SBen Gras #define GR(escape) ((escape)->right)
1022fe8fb19SBen Gras #define SET(escape) ((escape)->set)
1032fe8fb19SBen Gras #define ESC(escape) ((escape)->ch)
1042fe8fb19SBen Gras #define INIT(escape) (TAILQ_FIRST(SET(escape)))
1052fe8fb19SBen Gras
1062fe8fb19SBen Gras static __inline escape_t *
find_escape(escape_list * set,int ch)1072fe8fb19SBen Gras find_escape(escape_list *set, int ch)
1082fe8fb19SBen Gras {
1092fe8fb19SBen Gras escape_t *escape;
1102fe8fb19SBen Gras
1112fe8fb19SBen Gras _DIAGASSERT(set != NULL);
1122fe8fb19SBen Gras
1132fe8fb19SBen Gras TAILQ_FOREACH(escape, set, entry) {
1142fe8fb19SBen Gras if (ESC(escape) == ch)
1152fe8fb19SBen Gras break;
1162fe8fb19SBen Gras }
1172fe8fb19SBen Gras
1182fe8fb19SBen Gras return escape;
1192fe8fb19SBen Gras }
1202fe8fb19SBen Gras
1212fe8fb19SBen Gras typedef struct {
1222fe8fb19SBen Gras escape_list e0, e1;
1232fe8fb19SBen Gras graphic_t *ascii, *gb2312;
1242fe8fb19SBen Gras } _HZEncodingInfo;
1252fe8fb19SBen Gras
1262fe8fb19SBen Gras #define E0SET(ei) (&(ei)->e0)
1272fe8fb19SBen Gras #define E1SET(ei) (&(ei)->e1)
1282fe8fb19SBen Gras #define INIT0(ei) (TAILQ_FIRST(E0SET(ei)))
1292fe8fb19SBen Gras #define INIT1(ei) (TAILQ_FIRST(E1SET(ei)))
1302fe8fb19SBen Gras
1312fe8fb19SBen Gras typedef struct {
1322fe8fb19SBen Gras int chlen;
1332fe8fb19SBen Gras char ch[ROWCOL_MAX];
1342fe8fb19SBen Gras escape_t *inuse;
1352fe8fb19SBen Gras } _HZState;
1362fe8fb19SBen Gras
1372fe8fb19SBen Gras typedef struct {
1382fe8fb19SBen Gras _HZEncodingInfo ei;
1392fe8fb19SBen Gras struct {
1402fe8fb19SBen Gras /* for future multi-locale facility */
1412fe8fb19SBen Gras _HZState s_mblen;
1422fe8fb19SBen Gras _HZState s_mbrlen;
1432fe8fb19SBen Gras _HZState s_mbrtowc;
1442fe8fb19SBen Gras _HZState s_mbtowc;
1452fe8fb19SBen Gras _HZState s_mbsrtowcs;
14684d9c625SLionel Sambuc _HZState s_mbsnrtowcs;
1472fe8fb19SBen Gras _HZState s_wcrtomb;
1482fe8fb19SBen Gras _HZState s_wcsrtombs;
14984d9c625SLionel Sambuc _HZState s_wcsnrtombs;
1502fe8fb19SBen Gras _HZState s_wctomb;
1512fe8fb19SBen Gras } states;
1522fe8fb19SBen Gras } _HZCTypeInfo;
1532fe8fb19SBen Gras
1542fe8fb19SBen Gras #define _CEI_TO_EI(_cei_) (&(_cei_)->ei)
1552fe8fb19SBen Gras #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_
1562fe8fb19SBen Gras
1572fe8fb19SBen Gras #define _FUNCNAME(m) _citrus_HZ_##m
1582fe8fb19SBen Gras #define _ENCODING_INFO _HZEncodingInfo
1592fe8fb19SBen Gras #define _CTYPE_INFO _HZCTypeInfo
1602fe8fb19SBen Gras #define _ENCODING_STATE _HZState
1612fe8fb19SBen Gras #define _ENCODING_MB_CUR_MAX(_ei_) MB_LEN_MAX
1622fe8fb19SBen Gras #define _ENCODING_IS_STATE_DEPENDENT 1
1632fe8fb19SBen Gras #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) ((_ps_)->inuse == NULL)
1642fe8fb19SBen Gras
1652fe8fb19SBen Gras static __inline void
_citrus_HZ_init_state(_HZEncodingInfo * __restrict ei,_HZState * __restrict psenc)1662fe8fb19SBen Gras _citrus_HZ_init_state(_HZEncodingInfo * __restrict ei,
1672fe8fb19SBen Gras _HZState * __restrict psenc)
1682fe8fb19SBen Gras {
1692fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
1702fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
1712fe8fb19SBen Gras
1722fe8fb19SBen Gras psenc->chlen = 0;
1732fe8fb19SBen Gras psenc->inuse = INIT0(ei);
1742fe8fb19SBen Gras }
1752fe8fb19SBen Gras
1762fe8fb19SBen Gras static __inline void
1772fe8fb19SBen Gras /*ARGSUSED*/
_citrus_HZ_pack_state(_HZEncodingInfo * __restrict ei,void * __restrict pspriv,const _HZState * __restrict psenc)1782fe8fb19SBen Gras _citrus_HZ_pack_state(_HZEncodingInfo * __restrict ei,
1792fe8fb19SBen Gras void *__restrict pspriv, const _HZState * __restrict psenc)
1802fe8fb19SBen Gras {
1812fe8fb19SBen Gras /* ei may be unused */
1822fe8fb19SBen Gras _DIAGASSERT(pspriv != NULL);
1832fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
1842fe8fb19SBen Gras
1852fe8fb19SBen Gras memcpy(pspriv, (const void *)psenc, sizeof(*psenc));
1862fe8fb19SBen Gras }
1872fe8fb19SBen Gras
1882fe8fb19SBen Gras static __inline void
1892fe8fb19SBen Gras /*ARGSUSED*/
_citrus_HZ_unpack_state(_HZEncodingInfo * __restrict ei,_HZState * __restrict psenc,const void * __restrict pspriv)1902fe8fb19SBen Gras _citrus_HZ_unpack_state(_HZEncodingInfo * __restrict ei,
1912fe8fb19SBen Gras _HZState * __restrict psenc, const void * __restrict pspriv)
1922fe8fb19SBen Gras {
1932fe8fb19SBen Gras /* ei may be unused */
1942fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
1952fe8fb19SBen Gras _DIAGASSERT(pspriv != NULL);
1962fe8fb19SBen Gras
1972fe8fb19SBen Gras memcpy((void *)psenc, pspriv, sizeof(*psenc));
1982fe8fb19SBen Gras }
1992fe8fb19SBen Gras
2002fe8fb19SBen Gras static int
_citrus_HZ_mbrtowc_priv(_HZEncodingInfo * __restrict ei,wchar_t * __restrict pwc,const char ** __restrict s,size_t n,_HZState * __restrict psenc,size_t * __restrict nresult)2012fe8fb19SBen Gras _citrus_HZ_mbrtowc_priv(_HZEncodingInfo * __restrict ei,
2022fe8fb19SBen Gras wchar_t * __restrict pwc, const char ** __restrict s, size_t n,
2032fe8fb19SBen Gras _HZState * __restrict psenc, size_t * __restrict nresult)
2042fe8fb19SBen Gras {
2052fe8fb19SBen Gras const char *s0;
2062fe8fb19SBen Gras wchar_t wc;
2072fe8fb19SBen Gras int bit, head, tail, len, ch;
2082fe8fb19SBen Gras graphic_t *graphic;
2092fe8fb19SBen Gras escape_t *candidate, *init;
2102fe8fb19SBen Gras const range_t *range;
2112fe8fb19SBen Gras
2122fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
2132fe8fb19SBen Gras /* pwc may be null */
2142fe8fb19SBen Gras _DIAGASSERT(s != NULL);
2152fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
2162fe8fb19SBen Gras _DIAGASSERT(nresult != NULL);
2172fe8fb19SBen Gras
2182fe8fb19SBen Gras if (*s == NULL) {
2192fe8fb19SBen Gras _citrus_HZ_init_state(ei, psenc);
2202fe8fb19SBen Gras *nresult = 1;
2212fe8fb19SBen Gras return 0;
2222fe8fb19SBen Gras }
2232fe8fb19SBen Gras s0 = *s;
2242fe8fb19SBen Gras if (psenc->chlen < 0 || psenc->inuse == NULL)
2252fe8fb19SBen Gras return EINVAL;
2262fe8fb19SBen Gras
2272fe8fb19SBen Gras wc = (wchar_t)0;
2282fe8fb19SBen Gras bit = head = tail = 0;
2292fe8fb19SBen Gras graphic = NULL;
2302fe8fb19SBen Gras for (len = 0; len <= MB_LEN_MAX; /**/) {
2312fe8fb19SBen Gras if (psenc->chlen == tail) {
2322fe8fb19SBen Gras if (n-- < 1) {
2332fe8fb19SBen Gras *s = s0;
2342fe8fb19SBen Gras *nresult = (size_t)-2;
2352fe8fb19SBen Gras return 0;
2362fe8fb19SBen Gras }
2372fe8fb19SBen Gras psenc->ch[psenc->chlen++] = *s0++;
2382fe8fb19SBen Gras ++len;
2392fe8fb19SBen Gras }
2402fe8fb19SBen Gras ch = (unsigned char)psenc->ch[tail++];
2412fe8fb19SBen Gras if (tail == 1) {
2422fe8fb19SBen Gras if ((ch & ~0x80) <= 0x1F) {
2432fe8fb19SBen Gras if (psenc->inuse != INIT0(ei))
2442fe8fb19SBen Gras break;
2452fe8fb19SBen Gras wc = (wchar_t)ch;
2462fe8fb19SBen Gras goto done;
2472fe8fb19SBen Gras }
2482fe8fb19SBen Gras if (ch & 0x80) {
2492fe8fb19SBen Gras graphic = GR(psenc->inuse);
2502fe8fb19SBen Gras bit = 0x80;
2512fe8fb19SBen Gras ch &= ~0x80;
2522fe8fb19SBen Gras } else {
2532fe8fb19SBen Gras graphic = GL(psenc->inuse);
2542fe8fb19SBen Gras if (ch == ESCAPE_CHAR)
2552fe8fb19SBen Gras continue;
2562fe8fb19SBen Gras bit = 0x0;
2572fe8fb19SBen Gras }
2582fe8fb19SBen Gras if (graphic == NULL)
2592fe8fb19SBen Gras break;
2602fe8fb19SBen Gras } else if (tail == 2 && psenc->ch[0] == ESCAPE_CHAR) {
2612fe8fb19SBen Gras if (tail < psenc->chlen)
2622fe8fb19SBen Gras return EINVAL;
2632fe8fb19SBen Gras if (ch == ESCAPE_CHAR) {
2642fe8fb19SBen Gras ++head;
2652fe8fb19SBen Gras } else if (ch == '\n') {
2662fe8fb19SBen Gras if (psenc->inuse != INIT0(ei))
2672fe8fb19SBen Gras break;
2682fe8fb19SBen Gras tail = psenc->chlen = 0;
2692fe8fb19SBen Gras continue;
2702fe8fb19SBen Gras } else {
2712fe8fb19SBen Gras candidate = NULL;
2722fe8fb19SBen Gras init = INIT0(ei);
2732fe8fb19SBen Gras _DIAGASSERT(init != NULL);
2742fe8fb19SBen Gras if (psenc->inuse == init) {
2752fe8fb19SBen Gras init = INIT1(ei);
2762fe8fb19SBen Gras } else if (INIT(psenc->inuse) == init) {
2772fe8fb19SBen Gras if (ESC(init) != ch)
2782fe8fb19SBen Gras break;
2792fe8fb19SBen Gras candidate = init;
2802fe8fb19SBen Gras }
2812fe8fb19SBen Gras if (candidate == NULL) {
2822fe8fb19SBen Gras candidate = find_escape(
2832fe8fb19SBen Gras SET(psenc->inuse), ch);
2842fe8fb19SBen Gras if (candidate == NULL) {
2852fe8fb19SBen Gras if (init == NULL ||
2862fe8fb19SBen Gras ESC(init) != ch)
2872fe8fb19SBen Gras break;
2882fe8fb19SBen Gras candidate = init;
2892fe8fb19SBen Gras }
2902fe8fb19SBen Gras }
2912fe8fb19SBen Gras psenc->inuse = candidate;
2922fe8fb19SBen Gras tail = psenc->chlen = 0;
2932fe8fb19SBen Gras continue;
2942fe8fb19SBen Gras }
2952fe8fb19SBen Gras } else if (ch & 0x80) {
2962fe8fb19SBen Gras if (graphic != GR(psenc->inuse))
2972fe8fb19SBen Gras break;
2982fe8fb19SBen Gras ch &= ~0x80;
2992fe8fb19SBen Gras } else {
3002fe8fb19SBen Gras if (graphic != GL(psenc->inuse))
3012fe8fb19SBen Gras break;
3022fe8fb19SBen Gras }
3032fe8fb19SBen Gras _DIAGASSERT(graphic != NULL);
3042fe8fb19SBen Gras range = &ranges[(size_t)graphic->charset];
3052fe8fb19SBen Gras if (range->start > ch || range->end < ch)
3062fe8fb19SBen Gras break;
3072fe8fb19SBen Gras wc <<= 8;
3082fe8fb19SBen Gras wc |= ch;
3092fe8fb19SBen Gras if (graphic->length == (tail - head)) {
3102fe8fb19SBen Gras if (graphic->charset > GB2312)
3112fe8fb19SBen Gras bit |= ESC(psenc->inuse) << 24;
3122fe8fb19SBen Gras wc |= bit;
3132fe8fb19SBen Gras goto done;
3142fe8fb19SBen Gras }
3152fe8fb19SBen Gras }
3162fe8fb19SBen Gras *nresult = (size_t)-1;
3172fe8fb19SBen Gras return EILSEQ;
3182fe8fb19SBen Gras done:
3192fe8fb19SBen Gras if (tail < psenc->chlen)
3202fe8fb19SBen Gras return EINVAL;
3212fe8fb19SBen Gras *s = s0;
3222fe8fb19SBen Gras if (pwc != NULL)
3232fe8fb19SBen Gras *pwc = wc;
3242fe8fb19SBen Gras psenc->chlen = 0;
3252fe8fb19SBen Gras *nresult = (wc == 0) ? 0 : len;
3262fe8fb19SBen Gras
3272fe8fb19SBen Gras return 0;
3282fe8fb19SBen Gras }
3292fe8fb19SBen Gras
3302fe8fb19SBen Gras static int
_citrus_HZ_wcrtomb_priv(_HZEncodingInfo * __restrict ei,char * __restrict s,size_t n,wchar_t wc,_HZState * __restrict psenc,size_t * __restrict nresult)3312fe8fb19SBen Gras _citrus_HZ_wcrtomb_priv(_HZEncodingInfo * __restrict ei,
3322fe8fb19SBen Gras char * __restrict s, size_t n, wchar_t wc,
3332fe8fb19SBen Gras _HZState * __restrict psenc, size_t * __restrict nresult)
3342fe8fb19SBen Gras {
3352fe8fb19SBen Gras int bit, ch;
3362fe8fb19SBen Gras escape_t *candidate, *init;
3372fe8fb19SBen Gras graphic_t *graphic;
3382fe8fb19SBen Gras size_t len;
3392fe8fb19SBen Gras const range_t *range;
3402fe8fb19SBen Gras
3412fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
3422fe8fb19SBen Gras _DIAGASSERT(s != NULL);
3432fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
3442fe8fb19SBen Gras _DIAGASSERT(nresult != NULL);
3452fe8fb19SBen Gras
3462fe8fb19SBen Gras if (psenc->chlen != 0 || psenc->inuse == NULL)
3472fe8fb19SBen Gras return EINVAL;
3482fe8fb19SBen Gras if (wc & 0x80) {
3492fe8fb19SBen Gras bit = 0x80;
3502fe8fb19SBen Gras wc &= ~0x80;
3512fe8fb19SBen Gras } else {
3522fe8fb19SBen Gras bit = 0x0;
3532fe8fb19SBen Gras }
3542fe8fb19SBen Gras if ((uint32_t)wc <= 0x1F) {
3552fe8fb19SBen Gras candidate = INIT0(ei);
3562fe8fb19SBen Gras graphic = (bit == 0)
3572fe8fb19SBen Gras ? candidate->left : candidate->right;
3582fe8fb19SBen Gras if (graphic == NULL)
3592fe8fb19SBen Gras goto ilseq;
3602fe8fb19SBen Gras range = &ranges[(size_t)CTRL];
3612fe8fb19SBen Gras len = 1;
3622fe8fb19SBen Gras } else if ((uint32_t)wc <= 0x7F) {
3632fe8fb19SBen Gras graphic = ei->ascii;
3642fe8fb19SBen Gras if (graphic == NULL)
3652fe8fb19SBen Gras goto ilseq;
3662fe8fb19SBen Gras candidate = graphic->escape;
3672fe8fb19SBen Gras range = &ranges[(size_t)graphic->charset];
3682fe8fb19SBen Gras len = graphic->length;
3692fe8fb19SBen Gras } else if ((uint32_t)wc <= 0x7F7F) {
3702fe8fb19SBen Gras graphic = ei->gb2312;
3712fe8fb19SBen Gras if (graphic == NULL)
3722fe8fb19SBen Gras goto ilseq;
3732fe8fb19SBen Gras candidate = graphic->escape;
3742fe8fb19SBen Gras range = &ranges[(size_t)graphic->charset];
3752fe8fb19SBen Gras len = graphic->length;
3762fe8fb19SBen Gras } else {
3772fe8fb19SBen Gras ch = (wc >> 24) & 0xFF;
3782fe8fb19SBen Gras candidate = find_escape(E0SET(ei), ch);
3792fe8fb19SBen Gras if (candidate == NULL) {
3802fe8fb19SBen Gras candidate = find_escape(E1SET(ei), ch);
3812fe8fb19SBen Gras if (candidate == NULL)
3822fe8fb19SBen Gras goto ilseq;
3832fe8fb19SBen Gras }
3842fe8fb19SBen Gras wc &= ~0xFF000000;
3852fe8fb19SBen Gras graphic = (bit == 0)
3862fe8fb19SBen Gras ? candidate->left : candidate->right;
3872fe8fb19SBen Gras if (graphic == NULL)
3882fe8fb19SBen Gras goto ilseq;
3892fe8fb19SBen Gras range = &ranges[(size_t)graphic->charset];
3902fe8fb19SBen Gras len = graphic->length;
3912fe8fb19SBen Gras }
3922fe8fb19SBen Gras if (psenc->inuse != candidate) {
3932fe8fb19SBen Gras init = INIT0(ei);
3942fe8fb19SBen Gras if (SET(psenc->inuse) == SET(candidate)) {
3952fe8fb19SBen Gras if (INIT(psenc->inuse) != init ||
3962fe8fb19SBen Gras psenc->inuse == init || candidate == init)
3972fe8fb19SBen Gras init = NULL;
3982fe8fb19SBen Gras } else if (candidate == (init = INIT(candidate))) {
3992fe8fb19SBen Gras init = NULL;
4002fe8fb19SBen Gras }
4012fe8fb19SBen Gras if (init != NULL) {
4022fe8fb19SBen Gras if (n < 2)
4032fe8fb19SBen Gras return E2BIG;
4042fe8fb19SBen Gras n -= 2;
4052fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ESCAPE_CHAR;
4062fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ESC(init);
4072fe8fb19SBen Gras }
4082fe8fb19SBen Gras if (n < 2)
4092fe8fb19SBen Gras return E2BIG;
4102fe8fb19SBen Gras n -= 2;
4112fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ESCAPE_CHAR;
4122fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ESC(candidate);
4132fe8fb19SBen Gras psenc->inuse = candidate;
4142fe8fb19SBen Gras }
4152fe8fb19SBen Gras if (n < len)
4162fe8fb19SBen Gras return E2BIG;
4172fe8fb19SBen Gras while (len-- > 0) {
4182fe8fb19SBen Gras ch = (wc >> (len * 8)) & 0xFF;
4192fe8fb19SBen Gras if (range->start > ch || range->end < ch)
4202fe8fb19SBen Gras goto ilseq;
4212fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ch | bit;
4222fe8fb19SBen Gras }
4232fe8fb19SBen Gras memcpy(s, psenc->ch, psenc->chlen);
4242fe8fb19SBen Gras *nresult = psenc->chlen;
4252fe8fb19SBen Gras psenc->chlen = 0;
4262fe8fb19SBen Gras
4272fe8fb19SBen Gras return 0;
4282fe8fb19SBen Gras
4292fe8fb19SBen Gras ilseq:
4302fe8fb19SBen Gras *nresult = (size_t)-1;
4312fe8fb19SBen Gras return EILSEQ;
4322fe8fb19SBen Gras }
4332fe8fb19SBen Gras
4342fe8fb19SBen Gras static __inline int
_citrus_HZ_put_state_reset(_HZEncodingInfo * __restrict ei,char * __restrict s,size_t n,_HZState * __restrict psenc,size_t * __restrict nresult)4352fe8fb19SBen Gras _citrus_HZ_put_state_reset(_HZEncodingInfo * __restrict ei,
4362fe8fb19SBen Gras char * __restrict s, size_t n, _HZState * __restrict psenc,
4372fe8fb19SBen Gras size_t * __restrict nresult)
4382fe8fb19SBen Gras {
4392fe8fb19SBen Gras escape_t *candidate;
4402fe8fb19SBen Gras
4412fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
4422fe8fb19SBen Gras _DIAGASSERT(s != NULL);
4432fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
4442fe8fb19SBen Gras _DIAGASSERT(nresult != NULL);
4452fe8fb19SBen Gras
4462fe8fb19SBen Gras if (psenc->chlen != 0 || psenc->inuse == NULL)
4472fe8fb19SBen Gras return EINVAL;
4482fe8fb19SBen Gras candidate = INIT0(ei);
4492fe8fb19SBen Gras if (psenc->inuse != candidate) {
4502fe8fb19SBen Gras if (n < 2)
4512fe8fb19SBen Gras return E2BIG;
4522fe8fb19SBen Gras n -= 2;
4532fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ESCAPE_CHAR;
4542fe8fb19SBen Gras psenc->ch[psenc->chlen++] = ESC(candidate);
4552fe8fb19SBen Gras }
4562fe8fb19SBen Gras if (n < 1)
4572fe8fb19SBen Gras return E2BIG;
4582fe8fb19SBen Gras if (psenc->chlen > 0)
4592fe8fb19SBen Gras memcpy(s, psenc->ch, psenc->chlen);
4602fe8fb19SBen Gras *nresult = psenc->chlen;
4612fe8fb19SBen Gras _citrus_HZ_init_state(ei, psenc);
4622fe8fb19SBen Gras
4632fe8fb19SBen Gras return 0;
4642fe8fb19SBen Gras }
4652fe8fb19SBen Gras
4662fe8fb19SBen Gras static __inline int
_citrus_HZ_stdenc_get_state_desc_generic(_HZEncodingInfo * __restrict ei,_HZState * __restrict psenc,int * __restrict rstate)4672fe8fb19SBen Gras _citrus_HZ_stdenc_get_state_desc_generic(_HZEncodingInfo * __restrict ei,
4682fe8fb19SBen Gras _HZState * __restrict psenc, int * __restrict rstate)
4692fe8fb19SBen Gras {
4702fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
4712fe8fb19SBen Gras _DIAGASSERT(psenc != NULL);
4722fe8fb19SBen Gras _DIAGASSERT(rstate != NULL);
4732fe8fb19SBen Gras
4742fe8fb19SBen Gras if (psenc->chlen < 0 || psenc->inuse == NULL)
4752fe8fb19SBen Gras return EINVAL;
4762fe8fb19SBen Gras *rstate = (psenc->chlen == 0)
4772fe8fb19SBen Gras ? ((psenc->inuse == INIT0(ei))
4782fe8fb19SBen Gras ? _STDENC_SDGEN_INITIAL
4792fe8fb19SBen Gras : _STDENC_SDGEN_STABLE)
4802fe8fb19SBen Gras : ((psenc->ch[0] == ESCAPE_CHAR)
4812fe8fb19SBen Gras ? _STDENC_SDGEN_INCOMPLETE_SHIFT
4822fe8fb19SBen Gras : _STDENC_SDGEN_INCOMPLETE_CHAR);
4832fe8fb19SBen Gras
4842fe8fb19SBen Gras return 0;
4852fe8fb19SBen Gras }
4862fe8fb19SBen Gras
4872fe8fb19SBen Gras static __inline int
4882fe8fb19SBen Gras /*ARGSUSED*/
_citrus_HZ_stdenc_wctocs(_HZEncodingInfo * __restrict ei,_csid_t * __restrict csid,_index_t * __restrict idx,wchar_t wc)4892fe8fb19SBen Gras _citrus_HZ_stdenc_wctocs(_HZEncodingInfo * __restrict ei,
4902fe8fb19SBen Gras _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc)
4912fe8fb19SBen Gras {
4922fe8fb19SBen Gras int bit;
4932fe8fb19SBen Gras
4942fe8fb19SBen Gras _DIAGASSERT(csid != NULL);
4952fe8fb19SBen Gras _DIAGASSERT(idx != NULL);
4962fe8fb19SBen Gras
4972fe8fb19SBen Gras if (wc & 0x80) {
4982fe8fb19SBen Gras bit = 0x80;
4992fe8fb19SBen Gras wc &= ~0x80;
5002fe8fb19SBen Gras } else {
5012fe8fb19SBen Gras bit = 0x0;
5022fe8fb19SBen Gras }
5032fe8fb19SBen Gras if ((uint32_t)wc <= 0x7F) {
5042fe8fb19SBen Gras *csid = (_csid_t)bit;
5052fe8fb19SBen Gras *idx = (_index_t)wc;
5062fe8fb19SBen Gras } else if ((uint32_t)wc <= 0x7F7F) {
5072fe8fb19SBen Gras *csid = (_csid_t)(bit | 0x8000);
5082fe8fb19SBen Gras *idx = (_index_t)wc;
5092fe8fb19SBen Gras } else {
5102fe8fb19SBen Gras *csid = (_index_t)(wc & ~0x00FFFF7F);
5112fe8fb19SBen Gras *idx = (_csid_t)(wc & 0x00FFFF7F);
5122fe8fb19SBen Gras }
5132fe8fb19SBen Gras
5142fe8fb19SBen Gras return 0;
5152fe8fb19SBen Gras }
5162fe8fb19SBen Gras
5172fe8fb19SBen Gras static __inline int
5182fe8fb19SBen Gras /*ARGSUSED*/
_citrus_HZ_stdenc_cstowc(_HZEncodingInfo * __restrict ei,wchar_t * __restrict wc,_csid_t csid,_index_t idx)5192fe8fb19SBen Gras _citrus_HZ_stdenc_cstowc(_HZEncodingInfo * __restrict ei,
5202fe8fb19SBen Gras wchar_t * __restrict wc, _csid_t csid, _index_t idx)
5212fe8fb19SBen Gras {
5222fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
5232fe8fb19SBen Gras _DIAGASSERT(wc != NULL);
5242fe8fb19SBen Gras
5252fe8fb19SBen Gras *wc = (wchar_t)idx;
5262fe8fb19SBen Gras switch (csid) {
5272fe8fb19SBen Gras case 0x80:
5282fe8fb19SBen Gras case 0x8080:
5292fe8fb19SBen Gras *wc |= (wchar_t)0x80;
5302fe8fb19SBen Gras /*FALLTHROUGH*/
5312fe8fb19SBen Gras case 0x0:
5322fe8fb19SBen Gras case 0x8000:
5332fe8fb19SBen Gras break;
5342fe8fb19SBen Gras default:
5352fe8fb19SBen Gras *wc |= (wchar_t)csid;
5362fe8fb19SBen Gras }
5372fe8fb19SBen Gras
5382fe8fb19SBen Gras return 0;
5392fe8fb19SBen Gras }
5402fe8fb19SBen Gras
5412fe8fb19SBen Gras static void
_citrus_HZ_encoding_module_uninit(_HZEncodingInfo * ei)5422fe8fb19SBen Gras _citrus_HZ_encoding_module_uninit(_HZEncodingInfo *ei)
5432fe8fb19SBen Gras {
5442fe8fb19SBen Gras escape_t *escape;
5452fe8fb19SBen Gras
5462fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
5472fe8fb19SBen Gras while ((escape = TAILQ_FIRST(E0SET(ei))) != NULL) {
5482fe8fb19SBen Gras TAILQ_REMOVE(E0SET(ei), escape, entry);
5492fe8fb19SBen Gras free(GL(escape));
5502fe8fb19SBen Gras free(GR(escape));
5512fe8fb19SBen Gras free(escape);
5522fe8fb19SBen Gras }
5532fe8fb19SBen Gras while ((escape = TAILQ_FIRST(E1SET(ei))) != NULL) {
5542fe8fb19SBen Gras TAILQ_REMOVE(E1SET(ei), escape, entry);
5552fe8fb19SBen Gras free(GL(escape));
5562fe8fb19SBen Gras free(GR(escape));
5572fe8fb19SBen Gras free(escape);
5582fe8fb19SBen Gras }
5592fe8fb19SBen Gras }
5602fe8fb19SBen Gras
5612fe8fb19SBen Gras static int
_citrus_HZ_parse_char(void * context,const char * name,const char * s)562*0a6a1f1dSLionel Sambuc _citrus_HZ_parse_char(void *context, const char *name, const char *s)
5632fe8fb19SBen Gras {
5642fe8fb19SBen Gras void **p;
5652fe8fb19SBen Gras escape_t *escape;
5662fe8fb19SBen Gras
5672fe8fb19SBen Gras _DIAGASSERT(context != NULL && *context != NULL);
5682fe8fb19SBen Gras _DIAGASSERT(name != NULL);
5692fe8fb19SBen Gras _DIAGASSERT(s != NULL);
5702fe8fb19SBen Gras
571*0a6a1f1dSLionel Sambuc p = (void **)context;
5722fe8fb19SBen Gras escape = (escape_t *)p[0];
5732fe8fb19SBen Gras if (escape->ch != '\0')
5742fe8fb19SBen Gras return EINVAL;
5752fe8fb19SBen Gras escape->ch = *s++;
5762fe8fb19SBen Gras if (escape->ch == ESCAPE_CHAR || *s != '\0')
5772fe8fb19SBen Gras return EINVAL;
5782fe8fb19SBen Gras
5792fe8fb19SBen Gras return 0;
5802fe8fb19SBen Gras }
5812fe8fb19SBen Gras
5822fe8fb19SBen Gras static int
_citrus_HZ_parse_graphic(void * context,const char * name,const char * s)583*0a6a1f1dSLionel Sambuc _citrus_HZ_parse_graphic(void *context, const char *name, const char *s)
5842fe8fb19SBen Gras {
5852fe8fb19SBen Gras void **p;
5862fe8fb19SBen Gras _HZEncodingInfo *ei;
5872fe8fb19SBen Gras escape_t *escape;
5882fe8fb19SBen Gras graphic_t *graphic;
5892fe8fb19SBen Gras
5902fe8fb19SBen Gras _DIAGASSERT(context != NULL && *context != NULL);
5912fe8fb19SBen Gras _DIAGASSERT(name != NULL);
5922fe8fb19SBen Gras _DIAGASSERT(s != NULL);
5932fe8fb19SBen Gras
594*0a6a1f1dSLionel Sambuc p = (void **)context;
5952fe8fb19SBen Gras escape = (escape_t *)p[0];
5962fe8fb19SBen Gras ei = (_HZEncodingInfo *)p[1];
5972fe8fb19SBen Gras graphic = malloc(sizeof(*graphic));
5982fe8fb19SBen Gras if (graphic == NULL)
5992fe8fb19SBen Gras return ENOMEM;
6002fe8fb19SBen Gras memset(graphic, 0, sizeof(*graphic));
6012fe8fb19SBen Gras if (strcmp("GL", name) == 0) {
6022fe8fb19SBen Gras if (GL(escape) != NULL)
6032fe8fb19SBen Gras goto release;
6042fe8fb19SBen Gras GL(escape) = graphic;
6052fe8fb19SBen Gras } else if (strcmp("GR", name) == 0) {
6062fe8fb19SBen Gras if (GR(escape) != NULL)
6072fe8fb19SBen Gras goto release;
6082fe8fb19SBen Gras GR(escape) = graphic;
6092fe8fb19SBen Gras } else {
6102fe8fb19SBen Gras release:
6112fe8fb19SBen Gras free(graphic);
6122fe8fb19SBen Gras return EINVAL;
6132fe8fb19SBen Gras }
6142fe8fb19SBen Gras graphic->escape = escape;
6152fe8fb19SBen Gras if (_bcs_strncasecmp("ASCII", s, 5) == 0) {
6162fe8fb19SBen Gras if (s[5] != '\0')
6172fe8fb19SBen Gras return EINVAL;
6182fe8fb19SBen Gras graphic->charset = ASCII;
6192fe8fb19SBen Gras graphic->length = 1;
6202fe8fb19SBen Gras ei->ascii = graphic;
6212fe8fb19SBen Gras return 0;
6222fe8fb19SBen Gras } else if (_bcs_strncasecmp("GB2312", s, 6) == 0) {
6232fe8fb19SBen Gras if (s[6] != '\0')
6242fe8fb19SBen Gras return EINVAL;
6252fe8fb19SBen Gras graphic->charset = GB2312;
6262fe8fb19SBen Gras graphic->length = 2;
6272fe8fb19SBen Gras ei->gb2312 = graphic;
6282fe8fb19SBen Gras return 0;
6292fe8fb19SBen Gras } else if (strncmp("94*", s, 3) == 0) {
6302fe8fb19SBen Gras graphic->charset = CS94;
6312fe8fb19SBen Gras } else if (strncmp("96*", s, 3) == 0) {
6322fe8fb19SBen Gras graphic->charset = CS96;
6332fe8fb19SBen Gras } else {
6342fe8fb19SBen Gras return EINVAL;
6352fe8fb19SBen Gras }
6362fe8fb19SBen Gras s += 3;
6372fe8fb19SBen Gras switch(*s) {
6382fe8fb19SBen Gras case '1': case '2': case '3':
6392fe8fb19SBen Gras graphic->length = (size_t)(*s - '0');
6402fe8fb19SBen Gras if (*++s == '\0')
6412fe8fb19SBen Gras break;
6422fe8fb19SBen Gras /*FALLTHROUGH*/
6432fe8fb19SBen Gras default:
6442fe8fb19SBen Gras return EINVAL;
6452fe8fb19SBen Gras }
6462fe8fb19SBen Gras return 0;
6472fe8fb19SBen Gras }
6482fe8fb19SBen Gras
6492fe8fb19SBen Gras static const _citrus_prop_hint_t escape_hints[] = {
6502fe8fb19SBen Gras _CITRUS_PROP_HINT_STR("CH", &_citrus_HZ_parse_char),
6512fe8fb19SBen Gras _CITRUS_PROP_HINT_STR("GL", &_citrus_HZ_parse_graphic),
6522fe8fb19SBen Gras _CITRUS_PROP_HINT_STR("GR", &_citrus_HZ_parse_graphic),
6532fe8fb19SBen Gras _CITRUS_PROP_HINT_END
6542fe8fb19SBen Gras };
6552fe8fb19SBen Gras
6562fe8fb19SBen Gras static int
_citrus_HZ_parse_escape(void * context,const char * name,const char * s)657*0a6a1f1dSLionel Sambuc _citrus_HZ_parse_escape(void *context, const char *name, const char *s)
6582fe8fb19SBen Gras {
6592fe8fb19SBen Gras _HZEncodingInfo *ei;
6602fe8fb19SBen Gras escape_t *escape;
6612fe8fb19SBen Gras void *p[2];
6622fe8fb19SBen Gras
6632fe8fb19SBen Gras _DIAGASSERT(context != NULL);
6642fe8fb19SBen Gras _DIAGASSERT(name != NULL);
6652fe8fb19SBen Gras _DIAGASSERT(s != NULL);
6662fe8fb19SBen Gras
667*0a6a1f1dSLionel Sambuc ei = (_HZEncodingInfo *)context;
6682fe8fb19SBen Gras escape = malloc(sizeof(*escape));
6692fe8fb19SBen Gras if (escape == NULL)
6702fe8fb19SBen Gras return EINVAL;
6712fe8fb19SBen Gras memset(escape, 0, sizeof(*escape));
6722fe8fb19SBen Gras if (strcmp("0", name) == 0) {
6732fe8fb19SBen Gras escape->set = E0SET(ei);
6742fe8fb19SBen Gras TAILQ_INSERT_TAIL(E0SET(ei), escape, entry);
6752fe8fb19SBen Gras } else if (strcmp("1", name) == 0) {
6762fe8fb19SBen Gras escape->set = E1SET(ei);
6772fe8fb19SBen Gras TAILQ_INSERT_TAIL(E1SET(ei), escape, entry);
6782fe8fb19SBen Gras } else {
6792fe8fb19SBen Gras free(escape);
6802fe8fb19SBen Gras return EINVAL;
6812fe8fb19SBen Gras }
6822fe8fb19SBen Gras p[0] = (void *)escape;
6832fe8fb19SBen Gras p[1] = (void *)ei;
6842fe8fb19SBen Gras return _citrus_prop_parse_variable(
6852fe8fb19SBen Gras escape_hints, (void *)&p[0], s, strlen(s));
6862fe8fb19SBen Gras }
6872fe8fb19SBen Gras
6882fe8fb19SBen Gras static const _citrus_prop_hint_t root_hints[] = {
6892fe8fb19SBen Gras _CITRUS_PROP_HINT_STR("0", &_citrus_HZ_parse_escape),
6902fe8fb19SBen Gras _CITRUS_PROP_HINT_STR("1", &_citrus_HZ_parse_escape),
6912fe8fb19SBen Gras _CITRUS_PROP_HINT_END
6922fe8fb19SBen Gras };
6932fe8fb19SBen Gras
6942fe8fb19SBen Gras static int
_citrus_HZ_encoding_module_init(_HZEncodingInfo * __restrict ei,const void * __restrict var,size_t lenvar)6952fe8fb19SBen Gras _citrus_HZ_encoding_module_init(_HZEncodingInfo * __restrict ei,
6962fe8fb19SBen Gras const void * __restrict var, size_t lenvar)
6972fe8fb19SBen Gras {
6982fe8fb19SBen Gras int errnum;
6992fe8fb19SBen Gras
7002fe8fb19SBen Gras _DIAGASSERT(ei != NULL);
7012fe8fb19SBen Gras
7022fe8fb19SBen Gras memset(ei, 0, sizeof(*ei));
7032fe8fb19SBen Gras TAILQ_INIT(E0SET(ei));
7042fe8fb19SBen Gras TAILQ_INIT(E1SET(ei));
7052fe8fb19SBen Gras errnum = _citrus_prop_parse_variable(
7062fe8fb19SBen Gras root_hints, (void *)ei, var, lenvar);
7072fe8fb19SBen Gras if (errnum != 0)
7082fe8fb19SBen Gras _citrus_HZ_encoding_module_uninit(ei);
7092fe8fb19SBen Gras return errnum;
7102fe8fb19SBen Gras }
7112fe8fb19SBen Gras
7122fe8fb19SBen Gras /* ----------------------------------------------------------------------
7132fe8fb19SBen Gras * public interface for ctype
7142fe8fb19SBen Gras */
7152fe8fb19SBen Gras
7162fe8fb19SBen Gras _CITRUS_CTYPE_DECLS(HZ);
7172fe8fb19SBen Gras _CITRUS_CTYPE_DEF_OPS(HZ);
7182fe8fb19SBen Gras
7192fe8fb19SBen Gras #include "citrus_ctype_template.h"
7202fe8fb19SBen Gras
7212fe8fb19SBen Gras /* ----------------------------------------------------------------------
7222fe8fb19SBen Gras * public interface for stdenc
7232fe8fb19SBen Gras */
7242fe8fb19SBen Gras
7252fe8fb19SBen Gras _CITRUS_STDENC_DECLS(HZ);
7262fe8fb19SBen Gras _CITRUS_STDENC_DEF_OPS(HZ);
7272fe8fb19SBen Gras
7282fe8fb19SBen Gras #include "citrus_stdenc_template.h"
729