xref: /freebsd/sys/libkern/iconv_ucs.c (revision 41059135ce931c0f1014a999ffabc6bc470ce856)
1 /*-
2  * Copyright (c) 2003, 2005 Ryuichiro Imura
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 
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
29 
30 #include <sys/param.h>
31 #include <sys/kernel.h>
32 #include <sys/systm.h>
33 #include <sys/malloc.h>
34 #include <sys/iconv.h>
35 
36 #include "iconv_converter_if.h"
37 
38 /*
39  * "UCS" converter
40  */
41 
42 #define	KICONV_UCS_COMBINE	0x1
43 #define	KICONV_UCS_FROM_UTF8	0x2
44 #define	KICONV_UCS_TO_UTF8	0x4
45 #define	KICONV_UCS_FROM_LE	0x8
46 #define	KICONV_UCS_TO_LE	0x10
47 #define	KICONV_UCS_FROM_UTF16	0x20
48 #define	KICONV_UCS_TO_UTF16	0x40
49 #define	KICONV_UCS_UCS4		0x80
50 
51 #define	ENCODING_UTF16	"UTF-16BE"
52 #define	ENCODING_UTF8	"UTF-8"
53 
54 static struct {
55 	const char *name;
56 	int from_flag, to_flag;
57 } unicode_family[] = {
58 	{ "UTF-8",	KICONV_UCS_FROM_UTF8,	KICONV_UCS_TO_UTF8 },
59 	{ "UCS-2LE",	KICONV_UCS_FROM_LE,	KICONV_UCS_TO_LE },
60 	{ "UTF-16BE",	KICONV_UCS_FROM_UTF16,	KICONV_UCS_TO_UTF16 },
61 	{ "UTF-16LE",	KICONV_UCS_FROM_UTF16|KICONV_UCS_FROM_LE,
62 	    KICONV_UCS_TO_UTF16|KICONV_UCS_TO_LE },
63 	{ NULL,		0,	0 }
64 };
65 
66 static uint32_t utf8_to_ucs4(const char *src, size_t *utf8width, size_t srclen);
67 static u_char *ucs4_to_utf8(uint32_t ucs4, char * dst, size_t *utf8width, size_t dstlen);
68 static uint32_t encode_surrogate(uint32_t code);
69 static uint32_t decode_surrogate(const u_char *ucs);
70 
71 #ifdef MODULE_DEPEND
72 MODULE_DEPEND(iconv_ucs, libiconv, 2, 2, 2);
73 #endif
74 
75 /*
76  * UCS converter instance
77  */
78 struct iconv_ucs {
79 	KOBJ_FIELDS;
80 	int			convtype;
81 	struct iconv_cspair *	d_csp;
82 	struct iconv_cspair *	d_cspf;
83 	void *			f_ctp;
84 	void *			t_ctp;
85 	void *			ctype;
86 };
87 
88 static int
89 iconv_ucs_open(struct iconv_converter_class *dcp,
90 	struct iconv_cspair *csp, struct iconv_cspair *cspf, void **dpp)
91 {
92 	struct iconv_ucs *dp;
93 	int i;
94 	const char *from, *to;
95 
96 	dp = (struct iconv_ucs *)kobj_create((struct kobj_class*)dcp, M_ICONV, M_WAITOK);
97 	to = csp->cp_to;
98 	from = cspf ? cspf->cp_from : csp->cp_from;
99 
100 	dp->convtype = 0;
101 
102 	if (cspf)
103 		dp->convtype |= KICONV_UCS_COMBINE;
104 	for (i = 0; unicode_family[i].name; i++) {
105 		if (strcasecmp(from, unicode_family[i].name) == 0)
106 			dp->convtype |= unicode_family[i].from_flag;
107 		if (strcasecmp(to, unicode_family[i].name) == 0)
108 			dp->convtype |= unicode_family[i].to_flag;
109 	}
110 	if (strcmp(ENCODING_UNICODE, ENCODING_UTF16) == 0)
111 		dp->convtype |= KICONV_UCS_UCS4;
112 	else
113 		dp->convtype &= ~KICONV_UCS_UCS4;
114 
115 	dp->f_ctp = dp->t_ctp = NULL;
116 	if (dp->convtype & KICONV_UCS_COMBINE) {
117 		if ((dp->convtype & KICONV_UCS_FROM_UTF8) == 0 &&
118 		    (dp->convtype & KICONV_UCS_FROM_LE) == 0) {
119 			iconv_open(ENCODING_UNICODE, from, &dp->f_ctp);
120 		}
121 		if ((dp->convtype & KICONV_UCS_TO_UTF8) == 0 &&
122 		    (dp->convtype & KICONV_UCS_TO_LE) == 0) {
123 			iconv_open(to, ENCODING_UNICODE, &dp->t_ctp);
124 		}
125 	}
126 
127 	dp->ctype = NULL;
128 	if (dp->convtype & (KICONV_UCS_FROM_UTF8 | KICONV_UCS_TO_UTF8))
129 		iconv_open(KICONV_WCTYPE_NAME, ENCODING_UTF8, &dp->ctype);
130 
131 	dp->d_csp = csp;
132 	if (dp->convtype & (KICONV_UCS_FROM_UTF8 | KICONV_UCS_FROM_LE)) {
133 		if (cspf) {
134 			dp->d_cspf = cspf;
135 			cspf->cp_refcount++;
136 		} else
137 			csp->cp_refcount++;
138 	}
139 	if (dp->convtype & (KICONV_UCS_TO_UTF8 | KICONV_UCS_TO_LE))
140 		csp->cp_refcount++;
141 	*dpp = (void*)dp;
142 	return 0;
143 }
144 
145 static int
146 iconv_ucs_close(void *data)
147 {
148 	struct iconv_ucs *dp = data;
149 
150 	if (dp->f_ctp)
151 		iconv_close(dp->f_ctp);
152 	if (dp->t_ctp)
153 		iconv_close(dp->t_ctp);
154 	if (dp->ctype)
155 		iconv_close(dp->ctype);
156 	if (dp->d_cspf)
157 		dp->d_cspf->cp_refcount--;
158 	else if (dp->convtype & (KICONV_UCS_FROM_UTF8 | KICONV_UCS_FROM_LE))
159 		dp->d_csp->cp_refcount--;
160 	if (dp->convtype & (KICONV_UCS_TO_UTF8 | KICONV_UCS_TO_LE))
161 		dp->d_csp->cp_refcount--;
162 	kobj_delete((struct kobj*)data, M_ICONV);
163 	return 0;
164 }
165 
166 static int
167 iconv_ucs_conv(void *d2p, const char **inbuf,
168 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft,
169 	int convchar, int casetype)
170 {
171 	struct iconv_ucs *dp = (struct iconv_ucs*)d2p;
172 	int ret = 0, i;
173 	size_t in, on, ir, or, inlen, outlen, ucslen;
174 	const char *src, *p;
175 	char *dst;
176 	u_char ucs[4], *q;
177 	uint32_t code;
178 
179 	if (inbuf == NULL || *inbuf == NULL || outbuf == NULL || *outbuf == NULL)
180 		return 0;
181 	ir = in = *inbytesleft;
182 	or = on = *outbytesleft;
183 	src = *inbuf;
184 	dst = *outbuf;
185 
186 	while (ir > 0 && or > 0) {
187 
188 		/*
189 		 * The first half of conversion.
190 		 * (convert any code into ENCODING_UNICODE)
191 		 */
192 		code = 0;
193 		p = src;
194 		if (dp->convtype & KICONV_UCS_FROM_UTF8) {
195 			/* convert UTF-8 to ENCODING_UNICODE */
196 			inlen = 0;
197 			code = utf8_to_ucs4(p, &inlen, ir);
198 			if (code == 0) {
199 				ret = -1;
200 				break;
201 			}
202 
203 			if (casetype == KICONV_FROM_LOWER && dp->ctype) {
204 				code = towlower(code, dp->ctype);
205 			} else if (casetype == KICONV_FROM_UPPER && dp->ctype) {
206 				code = towupper(code, dp->ctype);
207 			}
208 
209 			if ((code >= 0xd800 && code < 0xe000) || code >= 0x110000 ) {
210 				/* reserved for utf-16 surrogate pair */
211 				/* invalid unicode */
212 				ret = -1;
213 				break;
214 			}
215 
216 			if (inlen == 4) {
217 				if (dp->convtype & KICONV_UCS_UCS4) {
218 					ucslen = 4;
219 					code = encode_surrogate(code);
220 				} else {
221 					/* can't handle with ucs-2 */
222 					ret = -1;
223 					break;
224 				}
225 			} else {
226 				ucslen = 2;
227 			}
228 
229 			/* save UCS-4 into ucs[] */
230 			for (q = ucs, i = ucslen - 1 ; i >= 0 ; i--)
231 				*q++ = (code >> (i << 3)) & 0xff;
232 
233 		} else if (dp->convtype & KICONV_UCS_COMBINE && dp->f_ctp) {
234 			/* convert local code to ENCODING_UNICODE */
235 			ucslen = 4;
236 			inlen = ir;
237 			q = ucs;
238 			ret = iconv_convchr_case(dp->f_ctp, &p, &inlen, (char **)&q,
239 			    &ucslen, casetype & (KICONV_FROM_LOWER | KICONV_FROM_UPPER));
240 			if (ret)
241 				break;
242 			inlen = ir - inlen;
243 			ucslen = 4 - ucslen;
244 
245 		} else {
246 			/* src code is a proper subset of ENCODING_UNICODE */
247 			q = ucs;
248 			if (dp->convtype & KICONV_UCS_FROM_LE) {
249 				*q = *(p + 1);
250 				*(q + 1) = *p;
251 				p += 2;
252 			} else {
253 				*q = *p++;
254 				*(q + 1) = *p++;
255 			}
256 			if ((*q & 0xfc) == 0xd8) {
257 				if (dp->convtype & KICONV_UCS_UCS4 &&
258 				    dp->convtype & KICONV_UCS_FROM_UTF16) {
259 					inlen = ucslen = 4;
260 				} else {
261 					/* invalid unicode */
262 					ret = -1;
263 					break;
264 				}
265 			} else {
266 				inlen = ucslen = 2;
267 			}
268 			if (ir < inlen) {
269 				ret = -1;
270 				break;
271 			}
272 			if (ucslen == 4) {
273 				q += 2;
274 				if (dp->convtype & KICONV_UCS_FROM_LE) {
275 					*q = *(p + 1);
276 					*(q + 1) = *p;
277 				} else {
278 					*q = *p++;
279 					*(q + 1) = *p;
280 				}
281 				if ((*q & 0xfc) != 0xdc) {
282 					/* invalid unicode */
283 					ret = -1;
284 					break;
285 				}
286 			}
287 		}
288 
289 		/*
290 		 * The second half of conversion.
291 		 * (convert ENCODING_UNICODE into any code)
292 		 */
293 		p = ucs;
294 		if (dp->convtype & KICONV_UCS_TO_UTF8) {
295 			q = (u_char *)dst;
296 			if (ucslen == 4 && dp->convtype & KICONV_UCS_UCS4) {
297 				/* decode surrogate pair */
298 				code = decode_surrogate(p);
299 			} else {
300 				code = (ucs[0] << 8) | ucs[1];
301 			}
302 
303 			if (casetype == KICONV_LOWER && dp->ctype) {
304 				code = towlower(code, dp->ctype);
305 			} else if (casetype == KICONV_UPPER && dp->ctype) {
306 				code = towupper(code, dp->ctype);
307 			}
308 
309 			outlen = 0;
310 			if (ucs4_to_utf8(code, q, &outlen, or) == NULL) {
311 				ret = -1;
312 				break;
313 			}
314 
315 			src += inlen;
316 			ir -= inlen;
317 			dst += outlen;
318 			or -= outlen;
319 
320 		} else if (dp->convtype & KICONV_UCS_COMBINE && dp->t_ctp) {
321 			ret = iconv_convchr_case(dp->t_ctp, &p, &ucslen, &dst,
322 			    &or, casetype & (KICONV_LOWER | KICONV_UPPER));
323 			if (ret)
324 				break;
325 
326 			src += inlen;
327 			ir -= inlen;
328 
329 		} else {
330 			/* dst code is a proper subset of ENCODING_UNICODE */
331 			if (or < ucslen) {
332 				ret = -1;
333 				break;
334 			}
335 			src += inlen;
336 			ir -= inlen;
337 			or -= ucslen;
338 			if (dp->convtype & KICONV_UCS_TO_LE) {
339 				*dst++ = *(p + 1);
340 				*dst++ = *p;
341 				p += 2;
342 			} else {
343 				*dst++ = *p++;
344 				*dst++ = *p++;
345 			}
346 			if (ucslen == 4) {
347 				if ((dp->convtype & KICONV_UCS_UCS4) == 0 ||
348 				    (dp->convtype & KICONV_UCS_TO_UTF16) == 0) {
349 					ret = -1;
350 					break;
351 				}
352 				if (dp->convtype & KICONV_UCS_TO_LE) {
353 					*dst++ = *(p + 1);
354 					*dst++ = *p;
355 				} else {
356 					*dst++ = *p++;
357 					*dst++ = *p;
358 				}
359 			}
360 		}
361 
362 		if (convchar == 1)
363 			break;
364 	}
365 
366 	*inbuf += in - ir;
367 	*outbuf += on - or;
368 	*inbytesleft -= in - ir;
369 	*outbytesleft -= on - or;
370 	return (ret);
371 }
372 
373 static int
374 iconv_ucs_init(struct iconv_converter_class *dcp)
375 {
376 	int error;
377 
378 	error = iconv_add(ENCODING_UNICODE, ENCODING_UNICODE, ENCODING_UTF8);
379 	if (error)
380 		return (error);
381 	error = iconv_add(ENCODING_UNICODE, ENCODING_UTF8, ENCODING_UNICODE);
382 	if (error)
383 		return (error);
384 	return (0);
385 }
386 
387 static int
388 iconv_ucs_done(struct iconv_converter_class *dcp)
389 {
390 	return (0);
391 }
392 
393 static const char *
394 iconv_ucs_name(struct iconv_converter_class *dcp)
395 {
396 	return (ENCODING_UNICODE);
397 }
398 
399 static kobj_method_t iconv_ucs_methods[] = {
400 	KOBJMETHOD(iconv_converter_open,	iconv_ucs_open),
401 	KOBJMETHOD(iconv_converter_close,	iconv_ucs_close),
402 	KOBJMETHOD(iconv_converter_conv,	iconv_ucs_conv),
403 	KOBJMETHOD(iconv_converter_init,	iconv_ucs_init),
404 	KOBJMETHOD(iconv_converter_done,	iconv_ucs_done),
405 	KOBJMETHOD(iconv_converter_name,	iconv_ucs_name),
406 	{0, 0}
407 };
408 
409 KICONV_CONVERTER(ucs, sizeof(struct iconv_ucs));
410 
411 static uint32_t
412 utf8_to_ucs4(const char *src, size_t *utf8width, size_t srclen)
413 {
414 	size_t i, w = 0;
415 	uint32_t ucs4 = 0;
416 
417 	/*
418 	 * get leading 1 byte from utf-8
419 	 */
420 	if ((*src & 0x80) == 0) {
421 		/*
422 		 * leading 1 bit is "0"
423 		 *  utf-8: 0xxxxxxx
424 		 *  ucs-4: 00000000 00000000 00000000 0xxxxxxx
425 		 */
426 		w = 1;
427 		/* get trailing 7 bits */
428 		ucs4 = *src & 0x7f;
429 	} else if ((*src & 0xe0) == 0xc0) {
430 		/*
431 		 * leading 3 bits are "110"
432 		 *  utf-8: 110xxxxx 10yyyyyy
433 		 *  ucs-4: 00000000 00000000 00000xxx xxyyyyyy
434 		 */
435 		w = 2;
436 		/* get trailing 5 bits */
437 		ucs4 = *src & 0x1f;
438 	} else if ((*src & 0xf0) == 0xe0) {
439 		/*
440 		 * leading 4 bits are "1110"
441 		 *  utf-8: 1110xxxx 10yyyyyy 10zzzzzz
442 		 *  ucs-4: 00000000 00000000 xxxxyyyy yyzzzzzz
443 		 */
444 		w = 3;
445 		/* get trailing 4 bits */
446 		ucs4 = *src & 0x0f;
447 	} else if ((*src & 0xf8) == 0xf0) {
448 		/*
449 		 * leading 5 bits are "11110"
450 		 *  utf-8: 11110www 10xxxxxx 10yyyyyy 10zzzzzz
451 		 *  ucs-4: 00000000 000wwwxx xxxxyyyy yyzzzzzz
452 		 */
453 		w = 4;
454 		/* get trailing 3 bits */
455 		ucs4 = *src & 0x07;
456 	} else {
457 		/* out of utf-16 range or having illegal bits */
458 		return (0);
459 	}
460 
461 	if (srclen < w)
462 		return (0);
463 
464 	/*
465 	 * get left parts from utf-8
466 	 */
467 	for (i = 1 ; i < w ; i++) {
468 		if ((*(src + i) & 0xc0) != 0x80) {
469 			/* invalid: leading 2 bits are not "10" */
470 			return (0);
471 		}
472 		/* concatenate trailing 6 bits into ucs4 */
473 		ucs4 <<= 6;
474 		ucs4 |= *(src + i) & 0x3f;
475 	}
476 
477 	*utf8width = w;
478 	return (ucs4);
479 }
480 
481 static u_char *
482 ucs4_to_utf8(uint32_t ucs4, char *dst, size_t *utf8width, size_t dstlen)
483 {
484 	u_char lead, *p;
485 	size_t i, w;
486 
487 	/*
488 	 * determine utf-8 width and leading bits
489 	 */
490 	if (ucs4 < 0x80) {
491 		w = 1;
492 		lead = 0;	/* "0" */
493 	} else if (ucs4 < 0x800) {
494 		w = 2;
495 		lead = 0xc0;	/* "11" */
496 	} else if (ucs4 < 0x10000) {
497 		w = 3;
498 		lead = 0xe0;	/* "111" */
499 	} else if (ucs4 < 0x200000) {
500 		w = 4;
501 		lead = 0xf0;	/* "1111" */
502 	} else {
503 		return (NULL);
504 	}
505 
506 	if (dstlen < w)
507 		return (NULL);
508 
509 	/*
510 	 * construct utf-8
511 	 */
512 	p = dst;
513 	for (i = w - 1 ; i >= 1 ; i--) {
514 		/* get trailing 6 bits and put it with leading bit as "1" */
515 		*(p + i) = (ucs4 & 0x3f) | 0x80;
516 		ucs4 >>= 6;
517 	}
518 	*p = ucs4 | lead;
519 
520 	*utf8width = w;
521 
522 	return (p);
523 }
524 
525 static uint32_t
526 encode_surrogate(uint32_t code)
527 {
528 	return ((((code - 0x10000) << 6) & 0x3ff0000) |
529 	    ((code - 0x10000) & 0x3ff) | 0xd800dc00);
530 }
531 
532 static uint32_t
533 decode_surrogate(const u_char *ucs)
534 {
535 	return ((((ucs[0] & 0x3) << 18) | (ucs[1] << 10) |
536 	    ((ucs[2] & 0x3) << 8) | ucs[3]) + 0x10000);
537 }
538 
539