xref: /freebsd/sys/libkern/iconv_ucs.c (revision 1603881667360c015f6685131f2f25474fa67a72)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2003, 2005 Ryuichiro Imura
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26  * SUCH DAMAGE.
27  */
28 
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
31 
32 #include <sys/param.h>
33 #include <sys/kernel.h>
34 #include <sys/systm.h>
35 #include <sys/malloc.h>
36 #include <sys/iconv.h>
37 
38 #include "iconv_converter_if.h"
39 
40 /*
41  * "UCS" converter
42  */
43 
44 #define	KICONV_UCS_COMBINE	0x1
45 #define	KICONV_UCS_FROM_UTF8	0x2
46 #define	KICONV_UCS_TO_UTF8	0x4
47 #define	KICONV_UCS_FROM_LE	0x8
48 #define	KICONV_UCS_TO_LE	0x10
49 #define	KICONV_UCS_FROM_UTF16	0x20
50 #define	KICONV_UCS_TO_UTF16	0x40
51 #define	KICONV_UCS_UCS4		0x80
52 
53 #define	ENCODING_UTF16	"UTF-16BE"
54 #define	ENCODING_UTF8	"UTF-8"
55 
56 static struct {
57 	const char *name;
58 	int from_flag, to_flag;
59 } unicode_family[] = {
60 	{ "UTF-8",	KICONV_UCS_FROM_UTF8,	KICONV_UCS_TO_UTF8 },
61 	{ "UCS-2LE",	KICONV_UCS_FROM_LE,	KICONV_UCS_TO_LE },
62 	{ "UTF-16BE",	KICONV_UCS_FROM_UTF16,	KICONV_UCS_TO_UTF16 },
63 	{ "UTF-16LE",	KICONV_UCS_FROM_UTF16|KICONV_UCS_FROM_LE,
64 	    KICONV_UCS_TO_UTF16|KICONV_UCS_TO_LE },
65 	{ NULL,		0,	0 }
66 };
67 
68 static uint32_t utf8_to_ucs4(const char *src, size_t *utf8width, size_t srclen);
69 static u_char *ucs4_to_utf8(uint32_t ucs4, char * dst, size_t *utf8width, size_t dstlen);
70 static uint32_t encode_surrogate(uint32_t code);
71 static uint32_t decode_surrogate(const u_char *ucs);
72 
73 #ifdef MODULE_DEPEND
74 MODULE_DEPEND(iconv_ucs, libiconv, 2, 2, 2);
75 #endif
76 
77 /*
78  * UCS converter instance
79  */
80 struct iconv_ucs {
81 	KOBJ_FIELDS;
82 	int			convtype;
83 	struct iconv_cspair *	d_csp;
84 	struct iconv_cspair *	d_cspf;
85 	void *			f_ctp;
86 	void *			t_ctp;
87 	void *			ctype;
88 };
89 
90 static int
91 iconv_ucs_open(struct iconv_converter_class *dcp,
92 	struct iconv_cspair *csp, struct iconv_cspair *cspf, void **dpp)
93 {
94 	struct iconv_ucs *dp;
95 	int i;
96 	const char *from, *to;
97 
98 	dp = (struct iconv_ucs *)kobj_create((struct kobj_class*)dcp, M_ICONV, M_WAITOK);
99 	to = csp->cp_to;
100 	from = cspf ? cspf->cp_from : csp->cp_from;
101 
102 	dp->convtype = 0;
103 
104 	if (cspf)
105 		dp->convtype |= KICONV_UCS_COMBINE;
106 	for (i = 0; unicode_family[i].name; i++) {
107 		if (strcasecmp(from, unicode_family[i].name) == 0)
108 			dp->convtype |= unicode_family[i].from_flag;
109 		if (strcasecmp(to, unicode_family[i].name) == 0)
110 			dp->convtype |= unicode_family[i].to_flag;
111 	}
112 	if (strcmp(ENCODING_UNICODE, ENCODING_UTF16) == 0)
113 		dp->convtype |= KICONV_UCS_UCS4;
114 	else
115 		dp->convtype &= ~KICONV_UCS_UCS4;
116 
117 	dp->f_ctp = dp->t_ctp = NULL;
118 	if (dp->convtype & KICONV_UCS_COMBINE) {
119 		if ((dp->convtype & KICONV_UCS_FROM_UTF8) == 0 &&
120 		    (dp->convtype & KICONV_UCS_FROM_LE) == 0) {
121 			iconv_open(ENCODING_UNICODE, from, &dp->f_ctp);
122 		}
123 		if ((dp->convtype & KICONV_UCS_TO_UTF8) == 0 &&
124 		    (dp->convtype & KICONV_UCS_TO_LE) == 0) {
125 			iconv_open(to, ENCODING_UNICODE, &dp->t_ctp);
126 		}
127 	}
128 
129 	dp->ctype = NULL;
130 	if (dp->convtype & (KICONV_UCS_FROM_UTF8 | KICONV_UCS_TO_UTF8))
131 		iconv_open(KICONV_WCTYPE_NAME, ENCODING_UTF8, &dp->ctype);
132 
133 	dp->d_csp = csp;
134 	if (dp->convtype & (KICONV_UCS_FROM_UTF8 | KICONV_UCS_FROM_LE)) {
135 		if (cspf) {
136 			dp->d_cspf = cspf;
137 			cspf->cp_refcount++;
138 		} else
139 			csp->cp_refcount++;
140 	}
141 	if (dp->convtype & (KICONV_UCS_TO_UTF8 | KICONV_UCS_TO_LE))
142 		csp->cp_refcount++;
143 	*dpp = (void*)dp;
144 	return 0;
145 }
146 
147 static int
148 iconv_ucs_close(void *data)
149 {
150 	struct iconv_ucs *dp = data;
151 
152 	if (dp->f_ctp)
153 		iconv_close(dp->f_ctp);
154 	if (dp->t_ctp)
155 		iconv_close(dp->t_ctp);
156 	if (dp->ctype)
157 		iconv_close(dp->ctype);
158 	if (dp->d_cspf)
159 		dp->d_cspf->cp_refcount--;
160 	else if (dp->convtype & (KICONV_UCS_FROM_UTF8 | KICONV_UCS_FROM_LE))
161 		dp->d_csp->cp_refcount--;
162 	if (dp->convtype & (KICONV_UCS_TO_UTF8 | KICONV_UCS_TO_LE))
163 		dp->d_csp->cp_refcount--;
164 	kobj_delete((struct kobj*)data, M_ICONV);
165 	return 0;
166 }
167 
168 static int
169 iconv_ucs_conv(void *d2p, const char **inbuf,
170 	size_t *inbytesleft, char **outbuf, size_t *outbytesleft,
171 	int convchar, int casetype)
172 {
173 	struct iconv_ucs *dp = (struct iconv_ucs*)d2p;
174 	int ret = 0, i;
175 	size_t in, on, ir, or, inlen, outlen, ucslen;
176 	const char *src, *p;
177 	char *dst;
178 	u_char ucs[4], *q;
179 	uint32_t code;
180 
181 	if (inbuf == NULL || *inbuf == NULL || outbuf == NULL || *outbuf == NULL)
182 		return 0;
183 	ir = in = *inbytesleft;
184 	or = on = *outbytesleft;
185 	src = *inbuf;
186 	dst = *outbuf;
187 
188 	while (ir > 0 && or > 0) {
189 		/*
190 		 * The first half of conversion.
191 		 * (convert any code into ENCODING_UNICODE)
192 		 */
193 		code = 0;
194 		p = src;
195 		if (dp->convtype & KICONV_UCS_FROM_UTF8) {
196 			/* convert UTF-8 to ENCODING_UNICODE */
197 			inlen = 0;
198 			code = utf8_to_ucs4(p, &inlen, ir);
199 			if (code == 0) {
200 				ret = -1;
201 				break;
202 			}
203 
204 			if (casetype == KICONV_FROM_LOWER && dp->ctype) {
205 				code = towlower(code, dp->ctype);
206 			} else if (casetype == KICONV_FROM_UPPER && dp->ctype) {
207 				code = towupper(code, dp->ctype);
208 			}
209 
210 			if ((code >= 0xd800 && code < 0xe000) || code >= 0x110000 ) {
211 				/* reserved for utf-16 surrogate pair */
212 				/* invalid unicode */
213 				ret = -1;
214 				break;
215 			}
216 
217 			if (inlen == 4) {
218 				if (dp->convtype & KICONV_UCS_UCS4) {
219 					ucslen = 4;
220 					code = encode_surrogate(code);
221 				} else {
222 					/* can't handle with ucs-2 */
223 					ret = -1;
224 					break;
225 				}
226 			} else {
227 				ucslen = 2;
228 			}
229 
230 			/* save UCS-4 into ucs[] */
231 			for (q = ucs, i = ucslen - 1 ; i >= 0 ; i--)
232 				*q++ = (code >> (i << 3)) & 0xff;
233 
234 		} else if (dp->convtype & KICONV_UCS_COMBINE && dp->f_ctp) {
235 			/* convert local code to ENCODING_UNICODE */
236 			ucslen = 4;
237 			inlen = ir;
238 			q = ucs;
239 			ret = iconv_convchr_case(dp->f_ctp, &p, &inlen, (char **)&q,
240 			    &ucslen, casetype & (KICONV_FROM_LOWER | KICONV_FROM_UPPER));
241 			if (ret)
242 				break;
243 			inlen = ir - inlen;
244 			ucslen = 4 - ucslen;
245 
246 		} else {
247 			/* src code is a proper subset of ENCODING_UNICODE */
248 			q = ucs;
249 			if (dp->convtype & KICONV_UCS_FROM_LE) {
250 				*q = *(p + 1);
251 				*(q + 1) = *p;
252 				p += 2;
253 			} else {
254 				*q = *p++;
255 				*(q + 1) = *p++;
256 			}
257 			if ((*q & 0xfc) == 0xd8) {
258 				if (dp->convtype & KICONV_UCS_UCS4 &&
259 				    dp->convtype & KICONV_UCS_FROM_UTF16) {
260 					inlen = ucslen = 4;
261 				} else {
262 					/* invalid unicode */
263 					ret = -1;
264 					break;
265 				}
266 			} else {
267 				inlen = ucslen = 2;
268 			}
269 			if (ir < inlen) {
270 				ret = -1;
271 				break;
272 			}
273 			if (ucslen == 4) {
274 				q += 2;
275 				if (dp->convtype & KICONV_UCS_FROM_LE) {
276 					*q = *(p + 1);
277 					*(q + 1) = *p;
278 				} else {
279 					*q = *p++;
280 					*(q + 1) = *p;
281 				}
282 				if ((*q & 0xfc) != 0xdc) {
283 					/* invalid unicode */
284 					ret = -1;
285 					break;
286 				}
287 			}
288 		}
289 
290 		/*
291 		 * The second half of conversion.
292 		 * (convert ENCODING_UNICODE into any code)
293 		 */
294 		p = ucs;
295 		if (dp->convtype & KICONV_UCS_TO_UTF8) {
296 			q = (u_char *)dst;
297 			if (ucslen == 4 && dp->convtype & KICONV_UCS_UCS4) {
298 				/* decode surrogate pair */
299 				code = decode_surrogate(p);
300 			} else {
301 				code = (ucs[0] << 8) | ucs[1];
302 			}
303 
304 			if (casetype == KICONV_LOWER && dp->ctype) {
305 				code = towlower(code, dp->ctype);
306 			} else if (casetype == KICONV_UPPER && dp->ctype) {
307 				code = towupper(code, dp->ctype);
308 			}
309 
310 			outlen = 0;
311 			if (ucs4_to_utf8(code, q, &outlen, or) == NULL) {
312 				ret = -1;
313 				break;
314 			}
315 
316 			src += inlen;
317 			ir -= inlen;
318 			dst += outlen;
319 			or -= outlen;
320 
321 		} else if (dp->convtype & KICONV_UCS_COMBINE && dp->t_ctp) {
322 			ret = iconv_convchr_case(dp->t_ctp, &p, &ucslen, &dst,
323 			    &or, casetype & (KICONV_LOWER | KICONV_UPPER));
324 			if (ret)
325 				break;
326 
327 			src += inlen;
328 			ir -= inlen;
329 
330 		} else {
331 			/* dst code is a proper subset of ENCODING_UNICODE */
332 			if (or < ucslen) {
333 				ret = -1;
334 				break;
335 			}
336 			src += inlen;
337 			ir -= inlen;
338 			or -= ucslen;
339 			if (dp->convtype & KICONV_UCS_TO_LE) {
340 				*dst++ = *(p + 1);
341 				*dst++ = *p;
342 				p += 2;
343 			} else {
344 				*dst++ = *p++;
345 				*dst++ = *p++;
346 			}
347 			if (ucslen == 4) {
348 				if ((dp->convtype & KICONV_UCS_UCS4) == 0 ||
349 				    (dp->convtype & KICONV_UCS_TO_UTF16) == 0) {
350 					ret = -1;
351 					break;
352 				}
353 				if (dp->convtype & KICONV_UCS_TO_LE) {
354 					*dst++ = *(p + 1);
355 					*dst++ = *p;
356 				} else {
357 					*dst++ = *p++;
358 					*dst++ = *p;
359 				}
360 			}
361 		}
362 
363 		if (convchar == 1)
364 			break;
365 	}
366 
367 	*inbuf += in - ir;
368 	*outbuf += on - or;
369 	*inbytesleft -= in - ir;
370 	*outbytesleft -= on - or;
371 	return (ret);
372 }
373 
374 static int
375 iconv_ucs_init(struct iconv_converter_class *dcp)
376 {
377 	int error;
378 
379 	error = iconv_add(ENCODING_UNICODE, ENCODING_UNICODE, ENCODING_UTF8);
380 	if (error)
381 		return (error);
382 	error = iconv_add(ENCODING_UNICODE, ENCODING_UTF8, ENCODING_UNICODE);
383 	if (error)
384 		return (error);
385 	return (0);
386 }
387 
388 static int
389 iconv_ucs_done(struct iconv_converter_class *dcp)
390 {
391 	return (0);
392 }
393 
394 static const char *
395 iconv_ucs_name(struct iconv_converter_class *dcp)
396 {
397 	return (ENCODING_UNICODE);
398 }
399 
400 static kobj_method_t iconv_ucs_methods[] = {
401 	KOBJMETHOD(iconv_converter_open,	iconv_ucs_open),
402 	KOBJMETHOD(iconv_converter_close,	iconv_ucs_close),
403 	KOBJMETHOD(iconv_converter_conv,	iconv_ucs_conv),
404 	KOBJMETHOD(iconv_converter_init,	iconv_ucs_init),
405 	KOBJMETHOD(iconv_converter_done,	iconv_ucs_done),
406 	KOBJMETHOD(iconv_converter_name,	iconv_ucs_name),
407 	{0, 0}
408 };
409 
410 KICONV_CONVERTER(ucs, sizeof(struct iconv_ucs));
411 
412 static uint32_t
413 utf8_to_ucs4(const char *src, size_t *utf8width, size_t srclen)
414 {
415 	size_t i, w = 0;
416 	uint32_t ucs4 = 0;
417 
418 	/*
419 	 * get leading 1 byte from utf-8
420 	 */
421 	if ((*src & 0x80) == 0) {
422 		/*
423 		 * leading 1 bit is "0"
424 		 *  utf-8: 0xxxxxxx
425 		 *  ucs-4: 00000000 00000000 00000000 0xxxxxxx
426 		 */
427 		w = 1;
428 		/* get trailing 7 bits */
429 		ucs4 = *src & 0x7f;
430 	} else if ((*src & 0xe0) == 0xc0) {
431 		/*
432 		 * leading 3 bits are "110"
433 		 *  utf-8: 110xxxxx 10yyyyyy
434 		 *  ucs-4: 00000000 00000000 00000xxx xxyyyyyy
435 		 */
436 		w = 2;
437 		/* get trailing 5 bits */
438 		ucs4 = *src & 0x1f;
439 	} else if ((*src & 0xf0) == 0xe0) {
440 		/*
441 		 * leading 4 bits are "1110"
442 		 *  utf-8: 1110xxxx 10yyyyyy 10zzzzzz
443 		 *  ucs-4: 00000000 00000000 xxxxyyyy yyzzzzzz
444 		 */
445 		w = 3;
446 		/* get trailing 4 bits */
447 		ucs4 = *src & 0x0f;
448 	} else if ((*src & 0xf8) == 0xf0) {
449 		/*
450 		 * leading 5 bits are "11110"
451 		 *  utf-8: 11110www 10xxxxxx 10yyyyyy 10zzzzzz
452 		 *  ucs-4: 00000000 000wwwxx xxxxyyyy yyzzzzzz
453 		 */
454 		w = 4;
455 		/* get trailing 3 bits */
456 		ucs4 = *src & 0x07;
457 	} else {
458 		/* out of utf-16 range or having illegal bits */
459 		return (0);
460 	}
461 
462 	if (srclen < w)
463 		return (0);
464 
465 	/*
466 	 * get left parts from utf-8
467 	 */
468 	for (i = 1 ; i < w ; i++) {
469 		if ((*(src + i) & 0xc0) != 0x80) {
470 			/* invalid: leading 2 bits are not "10" */
471 			return (0);
472 		}
473 		/* concatenate trailing 6 bits into ucs4 */
474 		ucs4 <<= 6;
475 		ucs4 |= *(src + i) & 0x3f;
476 	}
477 
478 	*utf8width = w;
479 	return (ucs4);
480 }
481 
482 static u_char *
483 ucs4_to_utf8(uint32_t ucs4, char *dst, size_t *utf8width, size_t dstlen)
484 {
485 	u_char lead, *p;
486 	size_t i, w;
487 
488 	/*
489 	 * determine utf-8 width and leading bits
490 	 */
491 	if (ucs4 < 0x80) {
492 		w = 1;
493 		lead = 0;	/* "0" */
494 	} else if (ucs4 < 0x800) {
495 		w = 2;
496 		lead = 0xc0;	/* "11" */
497 	} else if (ucs4 < 0x10000) {
498 		w = 3;
499 		lead = 0xe0;	/* "111" */
500 	} else if (ucs4 < 0x200000) {
501 		w = 4;
502 		lead = 0xf0;	/* "1111" */
503 	} else {
504 		return (NULL);
505 	}
506 
507 	if (dstlen < w)
508 		return (NULL);
509 
510 	/*
511 	 * construct utf-8
512 	 */
513 	p = dst;
514 	for (i = w - 1 ; i >= 1 ; i--) {
515 		/* get trailing 6 bits and put it with leading bit as "1" */
516 		*(p + i) = (ucs4 & 0x3f) | 0x80;
517 		ucs4 >>= 6;
518 	}
519 	*p = ucs4 | lead;
520 
521 	*utf8width = w;
522 
523 	return (p);
524 }
525 
526 static uint32_t
527 encode_surrogate(uint32_t code)
528 {
529 	return ((((code - 0x10000) << 6) & 0x3ff0000) |
530 	    ((code - 0x10000) & 0x3ff) | 0xd800dc00);
531 }
532 
533 static uint32_t
534 decode_surrogate(const u_char *ucs)
535 {
536 	return ((((ucs[0] & 0x3) << 18) | (ucs[1] << 10) |
537 	    ((ucs[2] & 0x3) << 8) | ucs[3]) + 0x10000);
538 }
539