xref: /linux/fs/udf/unicode.c (revision 3b27d13940c3710a1128527c43719cb0bb05d73b)
1 /*
2  * unicode.c
3  *
4  * PURPOSE
5  *	Routines for converting between UTF-8 and OSTA Compressed Unicode.
6  *      Also handles filename mangling
7  *
8  * DESCRIPTION
9  *	OSTA Compressed Unicode is explained in the OSTA UDF specification.
10  *		http://www.osta.org/
11  *	UTF-8 is explained in the IETF RFC XXXX.
12  *		ftp://ftp.internic.net/rfc/rfcxxxx.txt
13  *
14  * COPYRIGHT
15  *	This file is distributed under the terms of the GNU General Public
16  *	License (GPL). Copies of the GPL can be obtained from:
17  *		ftp://prep.ai.mit.edu/pub/gnu/GPL
18  *	Each contributing author retains all rights to their own work.
19  */
20 
21 #include "udfdecl.h"
22 
23 #include <linux/kernel.h>
24 #include <linux/string.h>	/* for memset */
25 #include <linux/nls.h>
26 #include <linux/crc-itu-t.h>
27 #include <linux/slab.h>
28 
29 #include "udf_sb.h"
30 
31 static int udf_translate_to_linux(uint8_t *, int, uint8_t *, int, uint8_t *,
32 				  int);
33 
34 static int udf_char_to_ustr(struct ustr *dest, const uint8_t *src, int strlen)
35 {
36 	if ((!dest) || (!src) || (!strlen) || (strlen > UDF_NAME_LEN - 2))
37 		return 0;
38 
39 	memset(dest, 0, sizeof(struct ustr));
40 	memcpy(dest->u_name, src, strlen);
41 	dest->u_cmpID = 0x08;
42 	dest->u_len = strlen;
43 
44 	return strlen;
45 }
46 
47 /*
48  * udf_build_ustr
49  */
50 int udf_build_ustr(struct ustr *dest, dstring *ptr, int size)
51 {
52 	int usesize;
53 
54 	if (!dest || !ptr || !size)
55 		return -1;
56 	BUG_ON(size < 2);
57 
58 	usesize = min_t(size_t, ptr[size - 1], sizeof(dest->u_name));
59 	usesize = min(usesize, size - 2);
60 	dest->u_cmpID = ptr[0];
61 	dest->u_len = usesize;
62 	memcpy(dest->u_name, ptr + 1, usesize);
63 	memset(dest->u_name + usesize, 0, sizeof(dest->u_name) - usesize);
64 
65 	return 0;
66 }
67 
68 /*
69  * udf_build_ustr_exact
70  */
71 static void udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
72 {
73 	memset(dest, 0, sizeof(struct ustr));
74 	dest->u_cmpID = ptr[0];
75 	dest->u_len = exactsize - 1;
76 	memcpy(dest->u_name, ptr + 1, exactsize - 1);
77 }
78 
79 /*
80  * udf_CS0toUTF8
81  *
82  * PURPOSE
83  *	Convert OSTA Compressed Unicode to the UTF-8 equivalent.
84  *
85  * PRE-CONDITIONS
86  *	utf			Pointer to UTF-8 output buffer.
87  *	ocu			Pointer to OSTA Compressed Unicode input buffer
88  *				of size UDF_NAME_LEN bytes.
89  * 				both of type "struct ustr *"
90  *
91  * POST-CONDITIONS
92  *	<return>		>= 0 on success.
93  *
94  * HISTORY
95  *	November 12, 1997 - Andrew E. Mileski
96  *	Written, tested, and released.
97  */
98 int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
99 {
100 	const uint8_t *ocu;
101 	uint8_t cmp_id, ocu_len;
102 	int i;
103 
104 	ocu_len = ocu_i->u_len;
105 	if (ocu_len == 0) {
106 		memset(utf_o, 0, sizeof(struct ustr));
107 		return 0;
108 	}
109 
110 	cmp_id = ocu_i->u_cmpID;
111 	if (cmp_id != 8 && cmp_id != 16) {
112 		memset(utf_o, 0, sizeof(struct ustr));
113 		pr_err("unknown compression code (%d) stri=%s\n",
114 		       cmp_id, ocu_i->u_name);
115 		return -EINVAL;
116 	}
117 
118 	ocu = ocu_i->u_name;
119 	utf_o->u_len = 0;
120 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
121 
122 		/* Expand OSTA compressed Unicode to Unicode */
123 		uint32_t c = ocu[i++];
124 		if (cmp_id == 16)
125 			c = (c << 8) | ocu[i++];
126 
127 		/* Compress Unicode to UTF-8 */
128 		if (c < 0x80U)
129 			utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
130 		else if (c < 0x800U) {
131 			utf_o->u_name[utf_o->u_len++] =
132 						(uint8_t)(0xc0 | (c >> 6));
133 			utf_o->u_name[utf_o->u_len++] =
134 						(uint8_t)(0x80 | (c & 0x3f));
135 		} else {
136 			utf_o->u_name[utf_o->u_len++] =
137 						(uint8_t)(0xe0 | (c >> 12));
138 			utf_o->u_name[utf_o->u_len++] =
139 						(uint8_t)(0x80 |
140 							  ((c >> 6) & 0x3f));
141 			utf_o->u_name[utf_o->u_len++] =
142 						(uint8_t)(0x80 | (c & 0x3f));
143 		}
144 	}
145 	utf_o->u_cmpID = 8;
146 
147 	return utf_o->u_len;
148 }
149 
150 /*
151  *
152  * udf_UTF8toCS0
153  *
154  * PURPOSE
155  *	Convert UTF-8 to the OSTA Compressed Unicode equivalent.
156  *
157  * DESCRIPTION
158  *	This routine is only called by udf_lookup().
159  *
160  * PRE-CONDITIONS
161  *	ocu			Pointer to OSTA Compressed Unicode output
162  *				buffer of size UDF_NAME_LEN bytes.
163  *	utf			Pointer to UTF-8 input buffer.
164  *	utf_len			Length of UTF-8 input buffer in bytes.
165  *
166  * POST-CONDITIONS
167  *	<return>		Zero on success.
168  *
169  * HISTORY
170  *	November 12, 1997 - Andrew E. Mileski
171  *	Written, tested, and released.
172  */
173 static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
174 {
175 	unsigned c, i, max_val, utf_char;
176 	int utf_cnt, u_len;
177 
178 	memset(ocu, 0, sizeof(dstring) * length);
179 	ocu[0] = 8;
180 	max_val = 0xffU;
181 
182 try_again:
183 	u_len = 0U;
184 	utf_char = 0U;
185 	utf_cnt = 0U;
186 	for (i = 0U; i < utf->u_len; i++) {
187 		c = (uint8_t)utf->u_name[i];
188 
189 		/* Complete a multi-byte UTF-8 character */
190 		if (utf_cnt) {
191 			utf_char = (utf_char << 6) | (c & 0x3fU);
192 			if (--utf_cnt)
193 				continue;
194 		} else {
195 			/* Check for a multi-byte UTF-8 character */
196 			if (c & 0x80U) {
197 				/* Start a multi-byte UTF-8 character */
198 				if ((c & 0xe0U) == 0xc0U) {
199 					utf_char = c & 0x1fU;
200 					utf_cnt = 1;
201 				} else if ((c & 0xf0U) == 0xe0U) {
202 					utf_char = c & 0x0fU;
203 					utf_cnt = 2;
204 				} else if ((c & 0xf8U) == 0xf0U) {
205 					utf_char = c & 0x07U;
206 					utf_cnt = 3;
207 				} else if ((c & 0xfcU) == 0xf8U) {
208 					utf_char = c & 0x03U;
209 					utf_cnt = 4;
210 				} else if ((c & 0xfeU) == 0xfcU) {
211 					utf_char = c & 0x01U;
212 					utf_cnt = 5;
213 				} else {
214 					goto error_out;
215 				}
216 				continue;
217 			} else {
218 				/* Single byte UTF-8 character (most common) */
219 				utf_char = c;
220 			}
221 		}
222 
223 		/* Choose no compression if necessary */
224 		if (utf_char > max_val) {
225 			if (max_val == 0xffU) {
226 				max_val = 0xffffU;
227 				ocu[0] = (uint8_t)0x10U;
228 				goto try_again;
229 			}
230 			goto error_out;
231 		}
232 
233 		if (max_val == 0xffffU)
234 			ocu[++u_len] = (uint8_t)(utf_char >> 8);
235 		ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
236 	}
237 
238 	if (utf_cnt) {
239 error_out:
240 		ocu[++u_len] = '?';
241 		printk(KERN_DEBUG pr_fmt("bad UTF-8 character\n"));
242 	}
243 
244 	ocu[length - 1] = (uint8_t)u_len + 1;
245 
246 	return u_len + 1;
247 }
248 
249 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
250 			const struct ustr *ocu_i)
251 {
252 	const uint8_t *ocu;
253 	uint8_t cmp_id, ocu_len;
254 	int i, len;
255 
256 
257 	ocu_len = ocu_i->u_len;
258 	if (ocu_len == 0) {
259 		memset(utf_o, 0, sizeof(struct ustr));
260 		return 0;
261 	}
262 
263 	cmp_id = ocu_i->u_cmpID;
264 	if (cmp_id != 8 && cmp_id != 16) {
265 		memset(utf_o, 0, sizeof(struct ustr));
266 		pr_err("unknown compression code (%d) stri=%s\n",
267 		       cmp_id, ocu_i->u_name);
268 		return -EINVAL;
269 	}
270 
271 	ocu = ocu_i->u_name;
272 	utf_o->u_len = 0;
273 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
274 		/* Expand OSTA compressed Unicode to Unicode */
275 		uint32_t c = ocu[i++];
276 		if (cmp_id == 16)
277 			c = (c << 8) | ocu[i++];
278 
279 		len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
280 				    UDF_NAME_LEN - utf_o->u_len);
281 		/* Valid character? */
282 		if (len >= 0)
283 			utf_o->u_len += len;
284 		else
285 			utf_o->u_name[utf_o->u_len++] = '?';
286 	}
287 	utf_o->u_cmpID = 8;
288 
289 	return utf_o->u_len;
290 }
291 
292 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
293 			int length)
294 {
295 	int len;
296 	unsigned i, max_val;
297 	uint16_t uni_char;
298 	int u_len;
299 
300 	memset(ocu, 0, sizeof(dstring) * length);
301 	ocu[0] = 8;
302 	max_val = 0xffU;
303 
304 try_again:
305 	u_len = 0U;
306 	for (i = 0U; i < uni->u_len; i++) {
307 		len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
308 		if (!len)
309 			continue;
310 		/* Invalid character, deal with it */
311 		if (len < 0) {
312 			len = 1;
313 			uni_char = '?';
314 		}
315 
316 		if (uni_char > max_val) {
317 			max_val = 0xffffU;
318 			ocu[0] = (uint8_t)0x10U;
319 			goto try_again;
320 		}
321 
322 		if (max_val == 0xffffU)
323 			ocu[++u_len] = (uint8_t)(uni_char >> 8);
324 		ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
325 		i += len - 1;
326 	}
327 
328 	ocu[length - 1] = (uint8_t)u_len + 1;
329 	return u_len + 1;
330 }
331 
332 int udf_get_filename(struct super_block *sb, uint8_t *sname, int slen,
333 		     uint8_t *dname, int dlen)
334 {
335 	struct ustr *filename, *unifilename;
336 	int ret;
337 
338 	if (!slen)
339 		return -EIO;
340 
341 	filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
342 	if (!filename)
343 		return -ENOMEM;
344 
345 	unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
346 	if (!unifilename) {
347 		ret = -ENOMEM;
348 		goto out1;
349 	}
350 
351 	udf_build_ustr_exact(unifilename, sname, slen);
352 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
353 		ret = udf_CS0toUTF8(filename, unifilename);
354 		if (ret < 0) {
355 			udf_debug("Failed in udf_get_filename: sname = %s\n",
356 				  sname);
357 			goto out2;
358 		}
359 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
360 		ret = udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
361 				   unifilename);
362 		if (ret < 0) {
363 			udf_debug("Failed in udf_get_filename: sname = %s\n",
364 				  sname);
365 			goto out2;
366 		}
367 	} else
368 		BUG();
369 
370 	ret = udf_translate_to_linux(dname, dlen,
371 				     filename->u_name, filename->u_len,
372 				     unifilename->u_name, unifilename->u_len);
373 	/* Zero length filename isn't valid... */
374 	if (ret == 0)
375 		ret = -EINVAL;
376 out2:
377 	kfree(unifilename);
378 out1:
379 	kfree(filename);
380 	return ret;
381 }
382 
383 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
384 		     uint8_t *dname, int flen)
385 {
386 	struct ustr unifilename;
387 	int namelen;
388 
389 	if (!udf_char_to_ustr(&unifilename, sname, flen))
390 		return 0;
391 
392 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
393 		namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
394 		if (!namelen)
395 			return 0;
396 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
397 		namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
398 					&unifilename, UDF_NAME_LEN);
399 		if (!namelen)
400 			return 0;
401 	} else
402 		return 0;
403 
404 	return namelen;
405 }
406 
407 #define ILLEGAL_CHAR_MARK	'_'
408 #define EXT_MARK		'.'
409 #define CRC_MARK		'#'
410 #define EXT_SIZE 		5
411 /* Number of chars we need to store generated CRC to make filename unique */
412 #define CRC_LEN			5
413 
414 static int udf_translate_to_linux(uint8_t *newName, int newLen,
415 				  uint8_t *udfName, int udfLen,
416 				  uint8_t *fidName, int fidNameLen)
417 {
418 	int index, newIndex = 0, needsCRC = 0;
419 	int extIndex = 0, newExtIndex = 0, hasExt = 0;
420 	unsigned short valueCRC;
421 	uint8_t curr;
422 
423 	if (udfName[0] == '.' &&
424 	    (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
425 		needsCRC = 1;
426 		newIndex = udfLen;
427 		memcpy(newName, udfName, udfLen);
428 	} else {
429 		for (index = 0; index < udfLen; index++) {
430 			curr = udfName[index];
431 			if (curr == '/' || curr == 0) {
432 				needsCRC = 1;
433 				curr = ILLEGAL_CHAR_MARK;
434 				while (index + 1 < udfLen &&
435 						(udfName[index + 1] == '/' ||
436 						 udfName[index + 1] == 0))
437 					index++;
438 			}
439 			if (curr == EXT_MARK &&
440 					(udfLen - index - 1) <= EXT_SIZE) {
441 				if (udfLen == index + 1)
442 					hasExt = 0;
443 				else {
444 					hasExt = 1;
445 					extIndex = index;
446 					newExtIndex = newIndex;
447 				}
448 			}
449 			if (newIndex < newLen)
450 				newName[newIndex++] = curr;
451 			else
452 				needsCRC = 1;
453 		}
454 	}
455 	if (needsCRC) {
456 		uint8_t ext[EXT_SIZE];
457 		int localExtIndex = 0;
458 
459 		if (hasExt) {
460 			int maxFilenameLen;
461 			for (index = 0;
462 			     index < EXT_SIZE && extIndex + index + 1 < udfLen;
463 			     index++) {
464 				curr = udfName[extIndex + index + 1];
465 
466 				if (curr == '/' || curr == 0) {
467 					needsCRC = 1;
468 					curr = ILLEGAL_CHAR_MARK;
469 					while (extIndex + index + 2 < udfLen &&
470 					      (index + 1 < EXT_SIZE &&
471 						(udfName[extIndex + index + 2] == '/' ||
472 						 udfName[extIndex + index + 2] == 0)))
473 						index++;
474 				}
475 				ext[localExtIndex++] = curr;
476 			}
477 			maxFilenameLen = newLen - CRC_LEN - localExtIndex;
478 			if (newIndex > maxFilenameLen)
479 				newIndex = maxFilenameLen;
480 			else
481 				newIndex = newExtIndex;
482 		} else if (newIndex > newLen - CRC_LEN)
483 			newIndex = newLen - CRC_LEN;
484 		newName[newIndex++] = CRC_MARK;
485 		valueCRC = crc_itu_t(0, fidName, fidNameLen);
486 		newName[newIndex++] = hex_asc_upper_hi(valueCRC >> 8);
487 		newName[newIndex++] = hex_asc_upper_lo(valueCRC >> 8);
488 		newName[newIndex++] = hex_asc_upper_hi(valueCRC);
489 		newName[newIndex++] = hex_asc_upper_lo(valueCRC);
490 
491 		if (hasExt) {
492 			newName[newIndex++] = EXT_MARK;
493 			for (index = 0; index < localExtIndex; index++)
494 				newName[newIndex++] = ext[index];
495 		}
496 	}
497 
498 	return newIndex;
499 }
500