xref: /linux/fs/udf/unicode.c (revision 827634added7f38b7d724cab1dccdb2b004c13c3)
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 int udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
72 {
73 	if ((!dest) || (!ptr) || (!exactsize))
74 		return -1;
75 
76 	memset(dest, 0, sizeof(struct ustr));
77 	dest->u_cmpID = ptr[0];
78 	dest->u_len = exactsize - 1;
79 	memcpy(dest->u_name, ptr + 1, exactsize - 1);
80 
81 	return 0;
82 }
83 
84 /*
85  * udf_ocu_to_utf8
86  *
87  * PURPOSE
88  *	Convert OSTA Compressed Unicode to the UTF-8 equivalent.
89  *
90  * PRE-CONDITIONS
91  *	utf			Pointer to UTF-8 output buffer.
92  *	ocu			Pointer to OSTA Compressed Unicode input buffer
93  *				of size UDF_NAME_LEN bytes.
94  * 				both of type "struct ustr *"
95  *
96  * POST-CONDITIONS
97  *	<return>		Zero on success.
98  *
99  * HISTORY
100  *	November 12, 1997 - Andrew E. Mileski
101  *	Written, tested, and released.
102  */
103 int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
104 {
105 	const uint8_t *ocu;
106 	uint8_t cmp_id, ocu_len;
107 	int i;
108 
109 	ocu_len = ocu_i->u_len;
110 	if (ocu_len == 0) {
111 		memset(utf_o, 0, sizeof(struct ustr));
112 		return 0;
113 	}
114 
115 	cmp_id = ocu_i->u_cmpID;
116 	if (cmp_id != 8 && cmp_id != 16) {
117 		memset(utf_o, 0, sizeof(struct ustr));
118 		pr_err("unknown compression code (%d) stri=%s\n",
119 		       cmp_id, ocu_i->u_name);
120 		return 0;
121 	}
122 
123 	ocu = ocu_i->u_name;
124 	utf_o->u_len = 0;
125 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
126 
127 		/* Expand OSTA compressed Unicode to Unicode */
128 		uint32_t c = ocu[i++];
129 		if (cmp_id == 16)
130 			c = (c << 8) | ocu[i++];
131 
132 		/* Compress Unicode to UTF-8 */
133 		if (c < 0x80U)
134 			utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
135 		else if (c < 0x800U) {
136 			utf_o->u_name[utf_o->u_len++] =
137 						(uint8_t)(0xc0 | (c >> 6));
138 			utf_o->u_name[utf_o->u_len++] =
139 						(uint8_t)(0x80 | (c & 0x3f));
140 		} else {
141 			utf_o->u_name[utf_o->u_len++] =
142 						(uint8_t)(0xe0 | (c >> 12));
143 			utf_o->u_name[utf_o->u_len++] =
144 						(uint8_t)(0x80 |
145 							  ((c >> 6) & 0x3f));
146 			utf_o->u_name[utf_o->u_len++] =
147 						(uint8_t)(0x80 | (c & 0x3f));
148 		}
149 	}
150 	utf_o->u_cmpID = 8;
151 
152 	return utf_o->u_len;
153 }
154 
155 /*
156  *
157  * udf_utf8_to_ocu
158  *
159  * PURPOSE
160  *	Convert UTF-8 to the OSTA Compressed Unicode equivalent.
161  *
162  * DESCRIPTION
163  *	This routine is only called by udf_lookup().
164  *
165  * PRE-CONDITIONS
166  *	ocu			Pointer to OSTA Compressed Unicode output
167  *				buffer of size UDF_NAME_LEN bytes.
168  *	utf			Pointer to UTF-8 input buffer.
169  *	utf_len			Length of UTF-8 input buffer in bytes.
170  *
171  * POST-CONDITIONS
172  *	<return>		Zero on success.
173  *
174  * HISTORY
175  *	November 12, 1997 - Andrew E. Mileski
176  *	Written, tested, and released.
177  */
178 static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
179 {
180 	unsigned c, i, max_val, utf_char;
181 	int utf_cnt, u_len;
182 
183 	memset(ocu, 0, sizeof(dstring) * length);
184 	ocu[0] = 8;
185 	max_val = 0xffU;
186 
187 try_again:
188 	u_len = 0U;
189 	utf_char = 0U;
190 	utf_cnt = 0U;
191 	for (i = 0U; i < utf->u_len; i++) {
192 		c = (uint8_t)utf->u_name[i];
193 
194 		/* Complete a multi-byte UTF-8 character */
195 		if (utf_cnt) {
196 			utf_char = (utf_char << 6) | (c & 0x3fU);
197 			if (--utf_cnt)
198 				continue;
199 		} else {
200 			/* Check for a multi-byte UTF-8 character */
201 			if (c & 0x80U) {
202 				/* Start a multi-byte UTF-8 character */
203 				if ((c & 0xe0U) == 0xc0U) {
204 					utf_char = c & 0x1fU;
205 					utf_cnt = 1;
206 				} else if ((c & 0xf0U) == 0xe0U) {
207 					utf_char = c & 0x0fU;
208 					utf_cnt = 2;
209 				} else if ((c & 0xf8U) == 0xf0U) {
210 					utf_char = c & 0x07U;
211 					utf_cnt = 3;
212 				} else if ((c & 0xfcU) == 0xf8U) {
213 					utf_char = c & 0x03U;
214 					utf_cnt = 4;
215 				} else if ((c & 0xfeU) == 0xfcU) {
216 					utf_char = c & 0x01U;
217 					utf_cnt = 5;
218 				} else {
219 					goto error_out;
220 				}
221 				continue;
222 			} else {
223 				/* Single byte UTF-8 character (most common) */
224 				utf_char = c;
225 			}
226 		}
227 
228 		/* Choose no compression if necessary */
229 		if (utf_char > max_val) {
230 			if (max_val == 0xffU) {
231 				max_val = 0xffffU;
232 				ocu[0] = (uint8_t)0x10U;
233 				goto try_again;
234 			}
235 			goto error_out;
236 		}
237 
238 		if (max_val == 0xffffU)
239 			ocu[++u_len] = (uint8_t)(utf_char >> 8);
240 		ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
241 	}
242 
243 	if (utf_cnt) {
244 error_out:
245 		ocu[++u_len] = '?';
246 		printk(KERN_DEBUG pr_fmt("bad UTF-8 character\n"));
247 	}
248 
249 	ocu[length - 1] = (uint8_t)u_len + 1;
250 
251 	return u_len + 1;
252 }
253 
254 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
255 			const struct ustr *ocu_i)
256 {
257 	const uint8_t *ocu;
258 	uint8_t cmp_id, ocu_len;
259 	int i, len;
260 
261 
262 	ocu_len = ocu_i->u_len;
263 	if (ocu_len == 0) {
264 		memset(utf_o, 0, sizeof(struct ustr));
265 		return 0;
266 	}
267 
268 	cmp_id = ocu_i->u_cmpID;
269 	if (cmp_id != 8 && cmp_id != 16) {
270 		memset(utf_o, 0, sizeof(struct ustr));
271 		pr_err("unknown compression code (%d) stri=%s\n",
272 		       cmp_id, ocu_i->u_name);
273 		return 0;
274 	}
275 
276 	ocu = ocu_i->u_name;
277 	utf_o->u_len = 0;
278 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
279 		/* Expand OSTA compressed Unicode to Unicode */
280 		uint32_t c = ocu[i++];
281 		if (cmp_id == 16)
282 			c = (c << 8) | ocu[i++];
283 
284 		len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
285 				    UDF_NAME_LEN - utf_o->u_len);
286 		/* Valid character? */
287 		if (len >= 0)
288 			utf_o->u_len += len;
289 		else
290 			utf_o->u_name[utf_o->u_len++] = '?';
291 	}
292 	utf_o->u_cmpID = 8;
293 
294 	return utf_o->u_len;
295 }
296 
297 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
298 			int length)
299 {
300 	int len;
301 	unsigned i, max_val;
302 	uint16_t uni_char;
303 	int u_len;
304 
305 	memset(ocu, 0, sizeof(dstring) * length);
306 	ocu[0] = 8;
307 	max_val = 0xffU;
308 
309 try_again:
310 	u_len = 0U;
311 	for (i = 0U; i < uni->u_len; i++) {
312 		len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
313 		if (!len)
314 			continue;
315 		/* Invalid character, deal with it */
316 		if (len < 0) {
317 			len = 1;
318 			uni_char = '?';
319 		}
320 
321 		if (uni_char > max_val) {
322 			max_val = 0xffffU;
323 			ocu[0] = (uint8_t)0x10U;
324 			goto try_again;
325 		}
326 
327 		if (max_val == 0xffffU)
328 			ocu[++u_len] = (uint8_t)(uni_char >> 8);
329 		ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
330 		i += len - 1;
331 	}
332 
333 	ocu[length - 1] = (uint8_t)u_len + 1;
334 	return u_len + 1;
335 }
336 
337 int udf_get_filename(struct super_block *sb, uint8_t *sname, int slen,
338 		     uint8_t *dname, int dlen)
339 {
340 	struct ustr *filename, *unifilename;
341 	int len = 0;
342 
343 	filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
344 	if (!filename)
345 		return 0;
346 
347 	unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
348 	if (!unifilename)
349 		goto out1;
350 
351 	if (udf_build_ustr_exact(unifilename, sname, slen))
352 		goto out2;
353 
354 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
355 		if (!udf_CS0toUTF8(filename, unifilename)) {
356 			udf_debug("Failed in udf_get_filename: sname = %s\n",
357 				  sname);
358 			goto out2;
359 		}
360 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
361 		if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
362 				  unifilename)) {
363 			udf_debug("Failed in udf_get_filename: sname = %s\n",
364 				  sname);
365 			goto out2;
366 		}
367 	} else
368 		goto out2;
369 
370 	len = udf_translate_to_linux(dname, dlen,
371 				     filename->u_name, filename->u_len,
372 				     unifilename->u_name, unifilename->u_len);
373 out2:
374 	kfree(unifilename);
375 out1:
376 	kfree(filename);
377 	return len;
378 }
379 
380 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
381 		     uint8_t *dname, int flen)
382 {
383 	struct ustr unifilename;
384 	int namelen;
385 
386 	if (!udf_char_to_ustr(&unifilename, sname, flen))
387 		return 0;
388 
389 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
390 		namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
391 		if (!namelen)
392 			return 0;
393 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
394 		namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
395 					&unifilename, UDF_NAME_LEN);
396 		if (!namelen)
397 			return 0;
398 	} else
399 		return 0;
400 
401 	return namelen;
402 }
403 
404 #define ILLEGAL_CHAR_MARK	'_'
405 #define EXT_MARK		'.'
406 #define CRC_MARK		'#'
407 #define EXT_SIZE 		5
408 /* Number of chars we need to store generated CRC to make filename unique */
409 #define CRC_LEN			5
410 
411 static int udf_translate_to_linux(uint8_t *newName, int newLen,
412 				  uint8_t *udfName, int udfLen,
413 				  uint8_t *fidName, int fidNameLen)
414 {
415 	int index, newIndex = 0, needsCRC = 0;
416 	int extIndex = 0, newExtIndex = 0, hasExt = 0;
417 	unsigned short valueCRC;
418 	uint8_t curr;
419 
420 	if (udfName[0] == '.' &&
421 	    (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
422 		needsCRC = 1;
423 		newIndex = udfLen;
424 		memcpy(newName, udfName, udfLen);
425 	} else {
426 		for (index = 0; index < udfLen; index++) {
427 			curr = udfName[index];
428 			if (curr == '/' || curr == 0) {
429 				needsCRC = 1;
430 				curr = ILLEGAL_CHAR_MARK;
431 				while (index + 1 < udfLen &&
432 						(udfName[index + 1] == '/' ||
433 						 udfName[index + 1] == 0))
434 					index++;
435 			}
436 			if (curr == EXT_MARK &&
437 					(udfLen - index - 1) <= EXT_SIZE) {
438 				if (udfLen == index + 1)
439 					hasExt = 0;
440 				else {
441 					hasExt = 1;
442 					extIndex = index;
443 					newExtIndex = newIndex;
444 				}
445 			}
446 			if (newIndex < newLen)
447 				newName[newIndex++] = curr;
448 			else
449 				needsCRC = 1;
450 		}
451 	}
452 	if (needsCRC) {
453 		uint8_t ext[EXT_SIZE];
454 		int localExtIndex = 0;
455 
456 		if (hasExt) {
457 			int maxFilenameLen;
458 			for (index = 0;
459 			     index < EXT_SIZE && extIndex + index + 1 < udfLen;
460 			     index++) {
461 				curr = udfName[extIndex + index + 1];
462 
463 				if (curr == '/' || curr == 0) {
464 					needsCRC = 1;
465 					curr = ILLEGAL_CHAR_MARK;
466 					while (extIndex + index + 2 < udfLen &&
467 					      (index + 1 < EXT_SIZE &&
468 						(udfName[extIndex + index + 2] == '/' ||
469 						 udfName[extIndex + index + 2] == 0)))
470 						index++;
471 				}
472 				ext[localExtIndex++] = curr;
473 			}
474 			maxFilenameLen = newLen - CRC_LEN - localExtIndex;
475 			if (newIndex > maxFilenameLen)
476 				newIndex = maxFilenameLen;
477 			else
478 				newIndex = newExtIndex;
479 		} else if (newIndex > newLen - CRC_LEN)
480 			newIndex = newLen - CRC_LEN;
481 		newName[newIndex++] = CRC_MARK;
482 		valueCRC = crc_itu_t(0, fidName, fidNameLen);
483 		newName[newIndex++] = hex_asc_upper_hi(valueCRC >> 8);
484 		newName[newIndex++] = hex_asc_upper_lo(valueCRC >> 8);
485 		newName[newIndex++] = hex_asc_upper_hi(valueCRC);
486 		newName[newIndex++] = hex_asc_upper_lo(valueCRC);
487 
488 		if (hasExt) {
489 			newName[newIndex++] = EXT_MARK;
490 			for (index = 0; index < localExtIndex; index++)
491 				newName[newIndex++] = ext[index];
492 		}
493 	}
494 
495 	return newIndex;
496 }
497