xref: /openbmc/linux/fs/udf/unicode.c (revision 7dd65feb)
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 
28 #include "udf_sb.h"
29 
30 static int udf_translate_to_linux(uint8_t *, uint8_t *, int, uint8_t *, int);
31 
32 static int udf_char_to_ustr(struct ustr *dest, const uint8_t *src, int strlen)
33 {
34 	if ((!dest) || (!src) || (!strlen) || (strlen > UDF_NAME_LEN - 2))
35 		return 0;
36 
37 	memset(dest, 0, sizeof(struct ustr));
38 	memcpy(dest->u_name, src, strlen);
39 	dest->u_cmpID = 0x08;
40 	dest->u_len = strlen;
41 
42 	return strlen;
43 }
44 
45 /*
46  * udf_build_ustr
47  */
48 int udf_build_ustr(struct ustr *dest, dstring *ptr, int size)
49 {
50 	int usesize;
51 
52 	if (!dest || !ptr || !size)
53 		return -1;
54 	BUG_ON(size < 2);
55 
56 	usesize = min_t(size_t, ptr[size - 1], sizeof(dest->u_name));
57 	usesize = min(usesize, size - 2);
58 	dest->u_cmpID = ptr[0];
59 	dest->u_len = usesize;
60 	memcpy(dest->u_name, ptr + 1, usesize);
61 	memset(dest->u_name + usesize, 0, sizeof(dest->u_name) - usesize);
62 
63 	return 0;
64 }
65 
66 /*
67  * udf_build_ustr_exact
68  */
69 static int udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
70 {
71 	if ((!dest) || (!ptr) || (!exactsize))
72 		return -1;
73 
74 	memset(dest, 0, sizeof(struct ustr));
75 	dest->u_cmpID = ptr[0];
76 	dest->u_len = exactsize - 1;
77 	memcpy(dest->u_name, ptr + 1, exactsize - 1);
78 
79 	return 0;
80 }
81 
82 /*
83  * udf_ocu_to_utf8
84  *
85  * PURPOSE
86  *	Convert OSTA Compressed Unicode to the UTF-8 equivalent.
87  *
88  * PRE-CONDITIONS
89  *	utf			Pointer to UTF-8 output buffer.
90  *	ocu			Pointer to OSTA Compressed Unicode input buffer
91  *				of size UDF_NAME_LEN bytes.
92  * 				both of type "struct ustr *"
93  *
94  * POST-CONDITIONS
95  *	<return>		Zero on success.
96  *
97  * HISTORY
98  *	November 12, 1997 - Andrew E. Mileski
99  *	Written, tested, and released.
100  */
101 int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
102 {
103 	const uint8_t *ocu;
104 	uint8_t cmp_id, ocu_len;
105 	int i;
106 
107 	ocu_len = ocu_i->u_len;
108 	if (ocu_len == 0) {
109 		memset(utf_o, 0, sizeof(struct ustr));
110 		return 0;
111 	}
112 
113 	cmp_id = ocu_i->u_cmpID;
114 	if (cmp_id != 8 && cmp_id != 16) {
115 		memset(utf_o, 0, sizeof(struct ustr));
116 		printk(KERN_ERR "udf: unknown compression code (%d) stri=%s\n",
117 		       cmp_id, ocu_i->u_name);
118 		return 0;
119 	}
120 
121 	ocu = ocu_i->u_name;
122 	utf_o->u_len = 0;
123 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
124 
125 		/* Expand OSTA compressed Unicode to Unicode */
126 		uint32_t c = ocu[i++];
127 		if (cmp_id == 16)
128 			c = (c << 8) | ocu[i++];
129 
130 		/* Compress Unicode to UTF-8 */
131 		if (c < 0x80U)
132 			utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
133 		else if (c < 0x800U) {
134 			utf_o->u_name[utf_o->u_len++] =
135 						(uint8_t)(0xc0 | (c >> 6));
136 			utf_o->u_name[utf_o->u_len++] =
137 						(uint8_t)(0x80 | (c & 0x3f));
138 		} else {
139 			utf_o->u_name[utf_o->u_len++] =
140 						(uint8_t)(0xe0 | (c >> 12));
141 			utf_o->u_name[utf_o->u_len++] =
142 						(uint8_t)(0x80 |
143 							  ((c >> 6) & 0x3f));
144 			utf_o->u_name[utf_o->u_len++] =
145 						(uint8_t)(0x80 | (c & 0x3f));
146 		}
147 	}
148 	utf_o->u_cmpID = 8;
149 
150 	return utf_o->u_len;
151 }
152 
153 /*
154  *
155  * udf_utf8_to_ocu
156  *
157  * PURPOSE
158  *	Convert UTF-8 to the OSTA Compressed Unicode equivalent.
159  *
160  * DESCRIPTION
161  *	This routine is only called by udf_lookup().
162  *
163  * PRE-CONDITIONS
164  *	ocu			Pointer to OSTA Compressed Unicode output
165  *				buffer of size UDF_NAME_LEN bytes.
166  *	utf			Pointer to UTF-8 input buffer.
167  *	utf_len			Length of UTF-8 input buffer in bytes.
168  *
169  * POST-CONDITIONS
170  *	<return>		Zero on success.
171  *
172  * HISTORY
173  *	November 12, 1997 - Andrew E. Mileski
174  *	Written, tested, and released.
175  */
176 static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
177 {
178 	unsigned c, i, max_val, utf_char;
179 	int utf_cnt, u_len;
180 
181 	memset(ocu, 0, sizeof(dstring) * length);
182 	ocu[0] = 8;
183 	max_val = 0xffU;
184 
185 try_again:
186 	u_len = 0U;
187 	utf_char = 0U;
188 	utf_cnt = 0U;
189 	for (i = 0U; i < utf->u_len; i++) {
190 		c = (uint8_t)utf->u_name[i];
191 
192 		/* Complete a multi-byte UTF-8 character */
193 		if (utf_cnt) {
194 			utf_char = (utf_char << 6) | (c & 0x3fU);
195 			if (--utf_cnt)
196 				continue;
197 		} else {
198 			/* Check for a multi-byte UTF-8 character */
199 			if (c & 0x80U) {
200 				/* Start a multi-byte UTF-8 character */
201 				if ((c & 0xe0U) == 0xc0U) {
202 					utf_char = c & 0x1fU;
203 					utf_cnt = 1;
204 				} else if ((c & 0xf0U) == 0xe0U) {
205 					utf_char = c & 0x0fU;
206 					utf_cnt = 2;
207 				} else if ((c & 0xf8U) == 0xf0U) {
208 					utf_char = c & 0x07U;
209 					utf_cnt = 3;
210 				} else if ((c & 0xfcU) == 0xf8U) {
211 					utf_char = c & 0x03U;
212 					utf_cnt = 4;
213 				} else if ((c & 0xfeU) == 0xfcU) {
214 					utf_char = c & 0x01U;
215 					utf_cnt = 5;
216 				} else {
217 					goto error_out;
218 				}
219 				continue;
220 			} else {
221 				/* Single byte UTF-8 character (most common) */
222 				utf_char = c;
223 			}
224 		}
225 
226 		/* Choose no compression if necessary */
227 		if (utf_char > max_val) {
228 			if (max_val == 0xffU) {
229 				max_val = 0xffffU;
230 				ocu[0] = (uint8_t)0x10U;
231 				goto try_again;
232 			}
233 			goto error_out;
234 		}
235 
236 		if (max_val == 0xffffU)
237 			ocu[++u_len] = (uint8_t)(utf_char >> 8);
238 		ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
239 	}
240 
241 	if (utf_cnt) {
242 error_out:
243 		ocu[++u_len] = '?';
244 		printk(KERN_DEBUG "udf: bad UTF-8 character\n");
245 	}
246 
247 	ocu[length - 1] = (uint8_t)u_len + 1;
248 
249 	return u_len + 1;
250 }
251 
252 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
253 			const struct ustr *ocu_i)
254 {
255 	const uint8_t *ocu;
256 	uint8_t cmp_id, ocu_len;
257 	int i, len;
258 
259 
260 	ocu_len = ocu_i->u_len;
261 	if (ocu_len == 0) {
262 		memset(utf_o, 0, sizeof(struct ustr));
263 		return 0;
264 	}
265 
266 	cmp_id = ocu_i->u_cmpID;
267 	if (cmp_id != 8 && cmp_id != 16) {
268 		memset(utf_o, 0, sizeof(struct ustr));
269 		printk(KERN_ERR "udf: unknown compression code (%d) stri=%s\n",
270 		       cmp_id, ocu_i->u_name);
271 		return 0;
272 	}
273 
274 	ocu = ocu_i->u_name;
275 	utf_o->u_len = 0;
276 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
277 		/* Expand OSTA compressed Unicode to Unicode */
278 		uint32_t c = ocu[i++];
279 		if (cmp_id == 16)
280 			c = (c << 8) | ocu[i++];
281 
282 		len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
283 				    UDF_NAME_LEN - utf_o->u_len);
284 		/* Valid character? */
285 		if (len >= 0)
286 			utf_o->u_len += len;
287 		else
288 			utf_o->u_name[utf_o->u_len++] = '?';
289 	}
290 	utf_o->u_cmpID = 8;
291 
292 	return utf_o->u_len;
293 }
294 
295 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
296 			int length)
297 {
298 	int len;
299 	unsigned i, max_val;
300 	uint16_t uni_char;
301 	int u_len;
302 
303 	memset(ocu, 0, sizeof(dstring) * length);
304 	ocu[0] = 8;
305 	max_val = 0xffU;
306 
307 try_again:
308 	u_len = 0U;
309 	for (i = 0U; i < uni->u_len; i++) {
310 		len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
311 		if (!len)
312 			continue;
313 		/* Invalid character, deal with it */
314 		if (len < 0) {
315 			len = 1;
316 			uni_char = '?';
317 		}
318 
319 		if (uni_char > max_val) {
320 			max_val = 0xffffU;
321 			ocu[0] = (uint8_t)0x10U;
322 			goto try_again;
323 		}
324 
325 		if (max_val == 0xffffU)
326 			ocu[++u_len] = (uint8_t)(uni_char >> 8);
327 		ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
328 		i += len - 1;
329 	}
330 
331 	ocu[length - 1] = (uint8_t)u_len + 1;
332 	return u_len + 1;
333 }
334 
335 int udf_get_filename(struct super_block *sb, uint8_t *sname, uint8_t *dname,
336 		     int flen)
337 {
338 	struct ustr *filename, *unifilename;
339 	int len = 0;
340 
341 	filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
342 	if (!filename)
343 		return 0;
344 
345 	unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
346 	if (!unifilename)
347 		goto out1;
348 
349 	if (udf_build_ustr_exact(unifilename, sname, flen))
350 		goto out2;
351 
352 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
353 		if (!udf_CS0toUTF8(filename, unifilename)) {
354 			udf_debug("Failed in udf_get_filename: sname = %s\n",
355 				  sname);
356 			goto out2;
357 		}
358 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
359 		if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
360 				  unifilename)) {
361 			udf_debug("Failed in udf_get_filename: sname = %s\n",
362 				  sname);
363 			goto out2;
364 		}
365 	} else
366 		goto out2;
367 
368 	len = udf_translate_to_linux(dname, filename->u_name, filename->u_len,
369 				     unifilename->u_name, unifilename->u_len);
370 out2:
371 	kfree(unifilename);
372 out1:
373 	kfree(filename);
374 	return len;
375 }
376 
377 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
378 		     uint8_t *dname, int flen)
379 {
380 	struct ustr unifilename;
381 	int namelen;
382 
383 	if (!udf_char_to_ustr(&unifilename, sname, flen))
384 		return 0;
385 
386 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
387 		namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
388 		if (!namelen)
389 			return 0;
390 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
391 		namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
392 					&unifilename, UDF_NAME_LEN);
393 		if (!namelen)
394 			return 0;
395 	} else
396 		return 0;
397 
398 	return namelen;
399 }
400 
401 #define ILLEGAL_CHAR_MARK	'_'
402 #define EXT_MARK		'.'
403 #define CRC_MARK		'#'
404 #define EXT_SIZE 		5
405 
406 static int udf_translate_to_linux(uint8_t *newName, uint8_t *udfName,
407 				  int udfLen, uint8_t *fidName,
408 				  int fidNameLen)
409 {
410 	int index, newIndex = 0, needsCRC = 0;
411 	int extIndex = 0, newExtIndex = 0, hasExt = 0;
412 	unsigned short valueCRC;
413 	uint8_t curr;
414 	const uint8_t hexChar[] = "0123456789ABCDEF";
415 
416 	if (udfName[0] == '.' &&
417 	    (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
418 		needsCRC = 1;
419 		newIndex = udfLen;
420 		memcpy(newName, udfName, udfLen);
421 	} else {
422 		for (index = 0; index < udfLen; index++) {
423 			curr = udfName[index];
424 			if (curr == '/' || curr == 0) {
425 				needsCRC = 1;
426 				curr = ILLEGAL_CHAR_MARK;
427 				while (index + 1 < udfLen &&
428 						(udfName[index + 1] == '/' ||
429 						 udfName[index + 1] == 0))
430 					index++;
431 			}
432 			if (curr == EXT_MARK &&
433 					(udfLen - index - 1) <= EXT_SIZE) {
434 				if (udfLen == index + 1)
435 					hasExt = 0;
436 				else {
437 					hasExt = 1;
438 					extIndex = index;
439 					newExtIndex = newIndex;
440 				}
441 			}
442 			if (newIndex < 256)
443 				newName[newIndex++] = curr;
444 			else
445 				needsCRC = 1;
446 		}
447 	}
448 	if (needsCRC) {
449 		uint8_t ext[EXT_SIZE];
450 		int localExtIndex = 0;
451 
452 		if (hasExt) {
453 			int maxFilenameLen;
454 			for (index = 0;
455 			     index < EXT_SIZE && extIndex + index + 1 < udfLen;
456 			     index++) {
457 				curr = udfName[extIndex + index + 1];
458 
459 				if (curr == '/' || curr == 0) {
460 					needsCRC = 1;
461 					curr = ILLEGAL_CHAR_MARK;
462 					while (extIndex + index + 2 < udfLen &&
463 					      (index + 1 < EXT_SIZE &&
464 						(udfName[extIndex + index + 2] == '/' ||
465 						 udfName[extIndex + index + 2] == 0)))
466 						index++;
467 				}
468 				ext[localExtIndex++] = curr;
469 			}
470 			maxFilenameLen = 250 - localExtIndex;
471 			if (newIndex > maxFilenameLen)
472 				newIndex = maxFilenameLen;
473 			else
474 				newIndex = newExtIndex;
475 		} else if (newIndex > 250)
476 			newIndex = 250;
477 		newName[newIndex++] = CRC_MARK;
478 		valueCRC = crc_itu_t(0, fidName, fidNameLen);
479 		newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12];
480 		newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8];
481 		newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4];
482 		newName[newIndex++] = hexChar[(valueCRC & 0x000f)];
483 
484 		if (hasExt) {
485 			newName[newIndex++] = EXT_MARK;
486 			for (index = 0; index < localExtIndex; index++)
487 				newName[newIndex++] = ext[index];
488 		}
489 	}
490 
491 	return newIndex;
492 }
493