citrus_ues.c revision 260264
196643Sobrien/* $FreeBSD: stable/9/lib/libiconv_modules/UES/citrus_ues.c 260264 2014-01-04 17:27:43Z dim $ */
296643Sobrien/* $NetBSD: citrus_ues.c,v 1.1 2006/11/13 15:16:31 tnozaki Exp $ */
396643Sobrien
496643Sobrien/*-
596643Sobrien * Copyright (c)2006 Citrus Project,
6160819Ssimon * All rights reserved.
7160819Ssimon *
8162915Ssimon * Redistribution and use in source and binary forms, with or without
9162915Ssimon * modification, are permitted provided that the following conditions
10162915Ssimon * are met:
11162915Ssimon * 1. Redistributions of source code must retain the above copyright
12160819Ssimon *    notice, this list of conditions and the following disclaimer.
13160819Ssimon * 2. Redistributions in binary form must reproduce the above copyright
14160819Ssimon *    notice, this list of conditions and the following disclaimer in the
15160819Ssimon *    documentation and/or other materials provided with the distribution.
16160819Ssimon *
17160819Ssimon * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18160819Ssimon * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19160819Ssimon * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20160819Ssimon * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21160819Ssimon * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22160819Ssimon * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23160819Ssimon * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24160819Ssimon * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25160819Ssimon * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26160819Ssimon * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27160819Ssimon * SUCH DAMAGE.
28160819Ssimon */
29160819Ssimon
30160819Ssimon#include <sys/cdefs.h>
31160819Ssimon
32160819Ssimon#include <assert.h>
3396643Sobrien#include <errno.h>
34160819Ssimon#include <limits.h>
35160819Ssimon#include <stdio.h>
3696643Sobrien#include <stdint.h>
37160819Ssimon#include <stdlib.h>
38160819Ssimon#include <string.h>
39160819Ssimon#include <wchar.h>
40160819Ssimon
4196643Sobrien#include "citrus_namespace.h"
4296643Sobrien#include "citrus_types.h"
4396643Sobrien#include "citrus_bcs.h"
4496643Sobrien#include "citrus_module.h"
4596643Sobrien#include "citrus_stdenc.h"
4696643Sobrien#include "citrus_ues.h"
4796643Sobrien
4896643Sobrientypedef struct {
4996643Sobrien	size_t	 mb_cur_max;
5096643Sobrien	int	 mode;
5196643Sobrien#define MODE_C99	1
5296643Sobrien} _UESEncodingInfo;
5396643Sobrien
54160819Ssimontypedef struct {
5596643Sobrien	int	 chlen;
5696643Sobrien	char	 ch[12];
5796643Sobrien} _UESState;
5896643Sobrien
59160819Ssimon#define _CEI_TO_EI(_cei_)               (&(_cei_)->ei)
6096643Sobrien#define _CEI_TO_STATE(_cei_, _func_)    (_cei_)->states.s_##_func_
6196643Sobrien
62160819Ssimon#define _FUNCNAME(m)			_citrus_UES_##m
63160819Ssimon#define _ENCODING_INFO			_UESEncodingInfo
6496643Sobrien#define _ENCODING_STATE			_UESState
6596643Sobrien#define _ENCODING_MB_CUR_MAX(_ei_)	(_ei_)->mb_cur_max
6696643Sobrien#define _ENCODING_IS_STATE_DEPENDENT		0
6796643Sobrien#define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
6896643Sobrien
6996643Sobrienstatic __inline void
7096643Sobrien/*ARGSUSED*/
7196643Sobrien_citrus_UES_init_state(_UESEncodingInfo * __restrict ei __unused,
7296643Sobrien    _UESState * __restrict psenc)
7396643Sobrien{
7496643Sobrien
7596643Sobrien	psenc->chlen = 0;
7696643Sobrien}
7796643Sobrien
7896643Sobrien#if 0
7996643Sobrienstatic __inline void
8096643Sobrien/*ARGSUSED*/
8196643Sobrien_citrus_UES_pack_state(_UESEncodingInfo * __restrict ei __unused,
8296643Sobrien    void *__restrict pspriv, const _UESState * __restrict psenc)
8396643Sobrien{
8496643Sobrien
8596643Sobrien	memcpy(pspriv, (const void *)psenc, sizeof(*psenc));
8696643Sobrien}
8796643Sobrien
8896643Sobrienstatic __inline void
8996643Sobrien/*ARGSUSED*/
9096643Sobrien_citrus_UES_unpack_state(_UESEncodingInfo * __restrict ei __unused,
9196643Sobrien    _UESState * __restrict psenc, const void * __restrict pspriv)
9296643Sobrien{
9396643Sobrien
9496643Sobrien	memcpy((void *)psenc, pspriv, sizeof(*psenc));
9596643Sobrien}
9696643Sobrien#endif
97160819Ssimon
9896643Sobrienstatic __inline int
9996643Sobriento_int(int ch)
10096643Sobrien{
10196643Sobrien
10296643Sobrien	if (ch >= '0' && ch <= '9')
10396643Sobrien		return (ch - '0');
10496643Sobrien	else if (ch >= 'A' && ch <= 'F')
10596643Sobrien		return ((ch - 'A') + 10);
10696643Sobrien	else if (ch >= 'a' && ch <= 'f')
10796643Sobrien		return ((ch - 'a') + 10);
10896643Sobrien	return (-1);
10996643Sobrien}
11096643Sobrien
11196643Sobrien#define ESCAPE		'\\'
11296643Sobrien#define UCS2_ESC	'u'
11396643Sobrien#define UCS4_ESC	'U'
11496643Sobrien
11596643Sobrien#define UCS2_BIT	16
11696643Sobrien#define UCS4_BIT	32
11796643Sobrien#define BMP_MAX		UINT32_C(0xFFFF)
11896643Sobrien#define UCS2_MAX	UINT32_C(0x10FFFF)
11996643Sobrien#define UCS4_MAX	UINT32_C(0x7FFFFFFF)
12096643Sobrien
12196643Sobrienstatic const char *xdig = "0123456789abcdef";
12296643Sobrien
12396643Sobrienstatic __inline int
12496643Sobriento_str(char *s, wchar_t wc, int bit)
12596643Sobrien{
12696643Sobrien	char *p;
12796643Sobrien
12896643Sobrien	p = s;
12996643Sobrien	*p++ = ESCAPE;
13096643Sobrien	switch (bit) {
13196643Sobrien	case UCS2_BIT:
13296643Sobrien		*p++ = UCS2_ESC;
13396643Sobrien		break;
13496643Sobrien	case UCS4_BIT:
13596643Sobrien		*p++ = UCS4_ESC;
13696643Sobrien		break;
13796643Sobrien	default:
13896643Sobrien		abort();
13996643Sobrien	}
14096643Sobrien	do {
14196643Sobrien		*p++ = xdig[(wc >> (bit -= 4)) & 0xF];
14296643Sobrien	} while (bit > 0);
14396643Sobrien	return (p - s);
14496643Sobrien}
14596643Sobrien
14696643Sobrienstatic __inline bool
14796643Sobrienis_hi_surrogate(wchar_t wc)
14896643Sobrien{
14996643Sobrien
15096643Sobrien	return (wc >= 0xD800 && wc <= 0xDBFF);
15196643Sobrien}
15296643Sobrien
15396643Sobrienstatic __inline bool
15496643Sobrienis_lo_surrogate(wchar_t wc)
15596643Sobrien{
15696643Sobrien
15796643Sobrien	return (wc >= 0xDC00 && wc <= 0xDFFF);
15896643Sobrien}
15996643Sobrien
16096643Sobrienstatic __inline wchar_t
16196643Sobriensurrogate_to_ucs(wchar_t hi, wchar_t lo)
16296643Sobrien{
16396643Sobrien
16496643Sobrien	hi -= 0xD800;
16596643Sobrien	lo -= 0xDC00;
16696643Sobrien	return ((hi << 10 | lo) + 0x10000);
16796643Sobrien}
16896643Sobrien
16996643Sobrienstatic __inline void
17096643Sobrienucs_to_surrogate(wchar_t wc, wchar_t * __restrict hi, wchar_t * __restrict lo)
17196643Sobrien{
17296643Sobrien
17396643Sobrien	wc -= 0x10000;
17496643Sobrien	*hi = (wc >> 10) + 0xD800;
17596643Sobrien	*lo = (wc & 0x3FF) + 0xDC00;
17696643Sobrien}
17796643Sobrien
17896643Sobrienstatic __inline bool
17996643Sobrienis_basic(wchar_t wc)
18096643Sobrien{
18196643Sobrien
18296643Sobrien	return ((uint32_t)wc <= 0x9F && wc != 0x24 && wc != 0x40 &&
18396643Sobrien	    wc != 0x60);
18496643Sobrien}
18596643Sobrien
18696643Sobrienstatic int
18796643Sobrien_citrus_UES_mbrtowc_priv(_UESEncodingInfo * __restrict ei,
18896643Sobrien    wchar_t * __restrict pwc, char ** __restrict s, size_t n,
18996643Sobrien    _UESState * __restrict psenc, size_t * __restrict nresult)
19096643Sobrien{
19196643Sobrien	char *s0;
19296643Sobrien	int ch, head, num, tail;
19396643Sobrien	wchar_t hi, wc;
19496643Sobrien
19596643Sobrien	if (*s == NULL) {
19696643Sobrien		_citrus_UES_init_state(ei, psenc);
19796643Sobrien		*nresult = 0;
19896643Sobrien		return (0);
19996643Sobrien	}
20096643Sobrien	s0 = *s;
20196643Sobrien
20296643Sobrien	hi = (wchar_t)0;
20396643Sobrien	tail = 0;
20496643Sobrien
20596643Sobriensurrogate:
206	wc = (wchar_t)0;
207	head = tail;
208	if (psenc->chlen == head) {
209		if (n-- < 1)
210			goto restart;
211		psenc->ch[psenc->chlen++] = *s0++;
212	}
213	ch = (unsigned char)psenc->ch[head++];
214	if (ch == ESCAPE) {
215		if (psenc->chlen == head) {
216			if (n-- < 1)
217				goto restart;
218			psenc->ch[psenc->chlen++] = *s0++;
219		}
220		switch (psenc->ch[head]) {
221		case UCS2_ESC:
222			tail += 6;
223			break;
224		case UCS4_ESC:
225			if (ei->mode & MODE_C99) {
226				tail = 10;
227				break;
228			}
229		/*FALLTHROUGH*/
230		default:
231			tail = 0;
232		}
233		++head;
234	}
235	for (; head < tail; ++head) {
236		if (psenc->chlen == head) {
237			if (n-- < 1) {
238restart:
239				*s = s0;
240				*nresult = (size_t)-2;
241				return (0);
242			}
243			psenc->ch[psenc->chlen++] = *s0++;
244		}
245		num = to_int((int)(unsigned char)psenc->ch[head]);
246		if (num < 0) {
247			tail = 0;
248			break;
249		}
250		wc = (wc << 4) | num;
251	}
252	head = 0;
253	switch (tail) {
254	case 0:
255		break;
256	case 6:
257		if (hi != (wchar_t)0)
258			break;
259		if ((ei->mode & MODE_C99) == 0) {
260			if (is_hi_surrogate(wc) != 0) {
261				hi = wc;
262				goto surrogate;
263			}
264			if ((uint32_t)wc <= 0x7F /* XXX */ ||
265			    is_lo_surrogate(wc) != 0)
266				break;
267			goto done;
268		}
269	/*FALLTHROUGH*/
270	case 10:
271		if (is_basic(wc) == 0 && (uint32_t)wc <= UCS4_MAX &&
272		    is_hi_surrogate(wc) == 0 && is_lo_surrogate(wc) == 0)
273			goto done;
274		*nresult = (size_t)-1;
275		return (EILSEQ);
276	case 12:
277		if (is_lo_surrogate(wc) == 0)
278			break;
279		wc = surrogate_to_ucs(hi, wc);
280		goto done;
281	}
282	ch = (unsigned char)psenc->ch[0];
283	head = psenc->chlen;
284	if (--head > 0)
285		memmove(&psenc->ch[0], &psenc->ch[1], head);
286	wc = (wchar_t)ch;
287done:
288	psenc->chlen = head;
289	if (pwc != NULL)
290		*pwc = wc;
291	*nresult = (size_t)((wc == 0) ? 0 : (s0 - *s));
292	*s = s0;
293
294	return (0);
295}
296
297static int
298_citrus_UES_wcrtomb_priv(_UESEncodingInfo * __restrict ei,
299    char * __restrict s, size_t n, wchar_t wc,
300    _UESState * __restrict psenc, size_t * __restrict nresult)
301{
302	wchar_t hi, lo;
303
304	if (psenc->chlen != 0)
305		return (EINVAL);
306
307	if ((ei->mode & MODE_C99) ? is_basic(wc) : (uint32_t)wc <= 0x7F) {
308		if (n-- < 1)
309			goto e2big;
310		psenc->ch[psenc->chlen++] = (char)wc;
311	} else if ((uint32_t)wc <= BMP_MAX) {
312		if (n < 6)
313			goto e2big;
314		psenc->chlen = to_str(&psenc->ch[0], wc, UCS2_BIT);
315	} else if ((ei->mode & MODE_C99) == 0 && (uint32_t)wc <= UCS2_MAX) {
316		if (n < 12)
317			goto e2big;
318		ucs_to_surrogate(wc, &hi, &lo);
319		psenc->chlen += to_str(&psenc->ch[0], hi, UCS2_BIT);
320		psenc->chlen += to_str(&psenc->ch[6], lo, UCS2_BIT);
321	} else if ((ei->mode & MODE_C99) && (uint32_t)wc <= UCS4_MAX) {
322		if (n < 10)
323			goto e2big;
324		psenc->chlen = to_str(&psenc->ch[0], wc, UCS4_BIT);
325	} else {
326		*nresult = (size_t)-1;
327		return (EILSEQ);
328	}
329	memcpy(s, psenc->ch, psenc->chlen);
330	*nresult = psenc->chlen;
331	psenc->chlen = 0;
332
333	return (0);
334
335e2big:
336	*nresult = (size_t)-1;
337	return (E2BIG);
338}
339
340/*ARGSUSED*/
341static int
342_citrus_UES_stdenc_wctocs(_UESEncodingInfo * __restrict ei __unused,
343    _csid_t * __restrict csid, _index_t * __restrict idx, wchar_t wc)
344{
345
346	*csid = 0;
347	*idx = (_index_t)wc;
348
349	return (0);
350}
351
352static __inline int
353/*ARGSUSED*/
354_citrus_UES_stdenc_cstowc(_UESEncodingInfo * __restrict ei __unused,
355    wchar_t * __restrict wc, _csid_t csid, _index_t idx)
356{
357
358	if (csid != 0)
359		return (EILSEQ);
360	*wc = (wchar_t)idx;
361
362	return (0);
363}
364
365static __inline int
366/*ARGSUSED*/
367_citrus_UES_stdenc_get_state_desc_generic(_UESEncodingInfo * __restrict ei __unused,
368    _UESState * __restrict psenc, int * __restrict rstate)
369{
370
371	*rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
372	    _STDENC_SDGEN_INCOMPLETE_CHAR;
373	return (0);
374}
375
376static void
377/*ARGSUSED*/
378_citrus_UES_encoding_module_uninit(_UESEncodingInfo *ei __unused)
379{
380
381	/* ei seems to be unused */
382}
383
384static int
385/*ARGSUSED*/
386_citrus_UES_encoding_module_init(_UESEncodingInfo * __restrict ei,
387    const void * __restrict var, size_t lenvar)
388{
389	const char *p;
390
391	p = var;
392	memset((void *)ei, 0, sizeof(*ei));
393	while (lenvar > 0) {
394		switch (_bcs_toupper(*p)) {
395		case 'C':
396			MATCH(C99, ei->mode |= MODE_C99);
397			break;
398		}
399		++p;
400		--lenvar;
401	}
402	ei->mb_cur_max = (ei->mode & MODE_C99) ? 10 : 12;
403
404	return (0);
405}
406
407/* ----------------------------------------------------------------------
408 * public interface for stdenc
409 */
410
411_CITRUS_STDENC_DECLS(UES);
412_CITRUS_STDENC_DEF_OPS(UES);
413
414#include "citrus_stdenc_template.h"
415