xref: /openbmc/linux/fs/udf/unicode.c (revision 8e9356c6)
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 *, uint8_t *, int, uint8_t *, int);
32 
33 static int udf_char_to_ustr(struct ustr *dest, const uint8_t *src, int strlen)
34 {
35 	if ((!dest) || (!src) || (!strlen) || (strlen > UDF_NAME_LEN - 2))
36 		return 0;
37 
38 	memset(dest, 0, sizeof(struct ustr));
39 	memcpy(dest->u_name, src, strlen);
40 	dest->u_cmpID = 0x08;
41 	dest->u_len = strlen;
42 
43 	return strlen;
44 }
45 
46 /*
47  * udf_build_ustr
48  */
49 int udf_build_ustr(struct ustr *dest, dstring *ptr, int size)
50 {
51 	int usesize;
52 
53 	if (!dest || !ptr || !size)
54 		return -1;
55 	BUG_ON(size < 2);
56 
57 	usesize = min_t(size_t, ptr[size - 1], sizeof(dest->u_name));
58 	usesize = min(usesize, size - 2);
59 	dest->u_cmpID = ptr[0];
60 	dest->u_len = usesize;
61 	memcpy(dest->u_name, ptr + 1, usesize);
62 	memset(dest->u_name + usesize, 0, sizeof(dest->u_name) - usesize);
63 
64 	return 0;
65 }
66 
67 /*
68  * udf_build_ustr_exact
69  */
70 static int udf_build_ustr_exact(struct ustr *dest, dstring *ptr, int exactsize)
71 {
72 	if ((!dest) || (!ptr) || (!exactsize))
73 		return -1;
74 
75 	memset(dest, 0, sizeof(struct ustr));
76 	dest->u_cmpID = ptr[0];
77 	dest->u_len = exactsize - 1;
78 	memcpy(dest->u_name, ptr + 1, exactsize - 1);
79 
80 	return 0;
81 }
82 
83 /*
84  * udf_ocu_to_utf8
85  *
86  * PURPOSE
87  *	Convert OSTA Compressed Unicode to the UTF-8 equivalent.
88  *
89  * PRE-CONDITIONS
90  *	utf			Pointer to UTF-8 output buffer.
91  *	ocu			Pointer to OSTA Compressed Unicode input buffer
92  *				of size UDF_NAME_LEN bytes.
93  * 				both of type "struct ustr *"
94  *
95  * POST-CONDITIONS
96  *	<return>		Zero on success.
97  *
98  * HISTORY
99  *	November 12, 1997 - Andrew E. Mileski
100  *	Written, tested, and released.
101  */
102 int udf_CS0toUTF8(struct ustr *utf_o, const struct ustr *ocu_i)
103 {
104 	const uint8_t *ocu;
105 	uint8_t cmp_id, ocu_len;
106 	int i;
107 
108 	ocu_len = ocu_i->u_len;
109 	if (ocu_len == 0) {
110 		memset(utf_o, 0, sizeof(struct ustr));
111 		return 0;
112 	}
113 
114 	cmp_id = ocu_i->u_cmpID;
115 	if (cmp_id != 8 && cmp_id != 16) {
116 		memset(utf_o, 0, sizeof(struct ustr));
117 		pr_err("unknown compression code (%d) stri=%s\n",
118 		       cmp_id, ocu_i->u_name);
119 		return 0;
120 	}
121 
122 	ocu = ocu_i->u_name;
123 	utf_o->u_len = 0;
124 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
125 
126 		/* Expand OSTA compressed Unicode to Unicode */
127 		uint32_t c = ocu[i++];
128 		if (cmp_id == 16)
129 			c = (c << 8) | ocu[i++];
130 
131 		/* Compress Unicode to UTF-8 */
132 		if (c < 0x80U)
133 			utf_o->u_name[utf_o->u_len++] = (uint8_t)c;
134 		else if (c < 0x800U) {
135 			utf_o->u_name[utf_o->u_len++] =
136 						(uint8_t)(0xc0 | (c >> 6));
137 			utf_o->u_name[utf_o->u_len++] =
138 						(uint8_t)(0x80 | (c & 0x3f));
139 		} else {
140 			utf_o->u_name[utf_o->u_len++] =
141 						(uint8_t)(0xe0 | (c >> 12));
142 			utf_o->u_name[utf_o->u_len++] =
143 						(uint8_t)(0x80 |
144 							  ((c >> 6) & 0x3f));
145 			utf_o->u_name[utf_o->u_len++] =
146 						(uint8_t)(0x80 | (c & 0x3f));
147 		}
148 	}
149 	utf_o->u_cmpID = 8;
150 
151 	return utf_o->u_len;
152 }
153 
154 /*
155  *
156  * udf_utf8_to_ocu
157  *
158  * PURPOSE
159  *	Convert UTF-8 to the OSTA Compressed Unicode equivalent.
160  *
161  * DESCRIPTION
162  *	This routine is only called by udf_lookup().
163  *
164  * PRE-CONDITIONS
165  *	ocu			Pointer to OSTA Compressed Unicode output
166  *				buffer of size UDF_NAME_LEN bytes.
167  *	utf			Pointer to UTF-8 input buffer.
168  *	utf_len			Length of UTF-8 input buffer in bytes.
169  *
170  * POST-CONDITIONS
171  *	<return>		Zero on success.
172  *
173  * HISTORY
174  *	November 12, 1997 - Andrew E. Mileski
175  *	Written, tested, and released.
176  */
177 static int udf_UTF8toCS0(dstring *ocu, struct ustr *utf, int length)
178 {
179 	unsigned c, i, max_val, utf_char;
180 	int utf_cnt, u_len;
181 
182 	memset(ocu, 0, sizeof(dstring) * length);
183 	ocu[0] = 8;
184 	max_val = 0xffU;
185 
186 try_again:
187 	u_len = 0U;
188 	utf_char = 0U;
189 	utf_cnt = 0U;
190 	for (i = 0U; i < utf->u_len; i++) {
191 		c = (uint8_t)utf->u_name[i];
192 
193 		/* Complete a multi-byte UTF-8 character */
194 		if (utf_cnt) {
195 			utf_char = (utf_char << 6) | (c & 0x3fU);
196 			if (--utf_cnt)
197 				continue;
198 		} else {
199 			/* Check for a multi-byte UTF-8 character */
200 			if (c & 0x80U) {
201 				/* Start a multi-byte UTF-8 character */
202 				if ((c & 0xe0U) == 0xc0U) {
203 					utf_char = c & 0x1fU;
204 					utf_cnt = 1;
205 				} else if ((c & 0xf0U) == 0xe0U) {
206 					utf_char = c & 0x0fU;
207 					utf_cnt = 2;
208 				} else if ((c & 0xf8U) == 0xf0U) {
209 					utf_char = c & 0x07U;
210 					utf_cnt = 3;
211 				} else if ((c & 0xfcU) == 0xf8U) {
212 					utf_char = c & 0x03U;
213 					utf_cnt = 4;
214 				} else if ((c & 0xfeU) == 0xfcU) {
215 					utf_char = c & 0x01U;
216 					utf_cnt = 5;
217 				} else {
218 					goto error_out;
219 				}
220 				continue;
221 			} else {
222 				/* Single byte UTF-8 character (most common) */
223 				utf_char = c;
224 			}
225 		}
226 
227 		/* Choose no compression if necessary */
228 		if (utf_char > max_val) {
229 			if (max_val == 0xffU) {
230 				max_val = 0xffffU;
231 				ocu[0] = (uint8_t)0x10U;
232 				goto try_again;
233 			}
234 			goto error_out;
235 		}
236 
237 		if (max_val == 0xffffU)
238 			ocu[++u_len] = (uint8_t)(utf_char >> 8);
239 		ocu[++u_len] = (uint8_t)(utf_char & 0xffU);
240 	}
241 
242 	if (utf_cnt) {
243 error_out:
244 		ocu[++u_len] = '?';
245 		printk(KERN_DEBUG pr_fmt("bad UTF-8 character\n"));
246 	}
247 
248 	ocu[length - 1] = (uint8_t)u_len + 1;
249 
250 	return u_len + 1;
251 }
252 
253 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
254 			const struct ustr *ocu_i)
255 {
256 	const uint8_t *ocu;
257 	uint8_t cmp_id, ocu_len;
258 	int i, len;
259 
260 
261 	ocu_len = ocu_i->u_len;
262 	if (ocu_len == 0) {
263 		memset(utf_o, 0, sizeof(struct ustr));
264 		return 0;
265 	}
266 
267 	cmp_id = ocu_i->u_cmpID;
268 	if (cmp_id != 8 && cmp_id != 16) {
269 		memset(utf_o, 0, sizeof(struct ustr));
270 		pr_err("unknown compression code (%d) stri=%s\n",
271 		       cmp_id, ocu_i->u_name);
272 		return 0;
273 	}
274 
275 	ocu = ocu_i->u_name;
276 	utf_o->u_len = 0;
277 	for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
278 		/* Expand OSTA compressed Unicode to Unicode */
279 		uint32_t c = ocu[i++];
280 		if (cmp_id == 16)
281 			c = (c << 8) | ocu[i++];
282 
283 		len = nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
284 				    UDF_NAME_LEN - utf_o->u_len);
285 		/* Valid character? */
286 		if (len >= 0)
287 			utf_o->u_len += len;
288 		else
289 			utf_o->u_name[utf_o->u_len++] = '?';
290 	}
291 	utf_o->u_cmpID = 8;
292 
293 	return utf_o->u_len;
294 }
295 
296 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
297 			int length)
298 {
299 	int len;
300 	unsigned i, max_val;
301 	uint16_t uni_char;
302 	int u_len;
303 
304 	memset(ocu, 0, sizeof(dstring) * length);
305 	ocu[0] = 8;
306 	max_val = 0xffU;
307 
308 try_again:
309 	u_len = 0U;
310 	for (i = 0U; i < uni->u_len; i++) {
311 		len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
312 		if (!len)
313 			continue;
314 		/* Invalid character, deal with it */
315 		if (len < 0) {
316 			len = 1;
317 			uni_char = '?';
318 		}
319 
320 		if (uni_char > max_val) {
321 			max_val = 0xffffU;
322 			ocu[0] = (uint8_t)0x10U;
323 			goto try_again;
324 		}
325 
326 		if (max_val == 0xffffU)
327 			ocu[++u_len] = (uint8_t)(uni_char >> 8);
328 		ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
329 		i += len - 1;
330 	}
331 
332 	ocu[length - 1] = (uint8_t)u_len + 1;
333 	return u_len + 1;
334 }
335 
336 int udf_get_filename(struct super_block *sb, uint8_t *sname, uint8_t *dname,
337 		     int flen)
338 {
339 	struct ustr *filename, *unifilename;
340 	int len = 0;
341 
342 	filename = kmalloc(sizeof(struct ustr), GFP_NOFS);
343 	if (!filename)
344 		return 0;
345 
346 	unifilename = kmalloc(sizeof(struct ustr), GFP_NOFS);
347 	if (!unifilename)
348 		goto out1;
349 
350 	if (udf_build_ustr_exact(unifilename, sname, flen))
351 		goto out2;
352 
353 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
354 		if (!udf_CS0toUTF8(filename, unifilename)) {
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 		if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, filename,
361 				  unifilename)) {
362 			udf_debug("Failed in udf_get_filename: sname = %s\n",
363 				  sname);
364 			goto out2;
365 		}
366 	} else
367 		goto out2;
368 
369 	len = udf_translate_to_linux(dname, filename->u_name, filename->u_len,
370 				     unifilename->u_name, unifilename->u_len);
371 out2:
372 	kfree(unifilename);
373 out1:
374 	kfree(filename);
375 	return len;
376 }
377 
378 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
379 		     uint8_t *dname, int flen)
380 {
381 	struct ustr unifilename;
382 	int namelen;
383 
384 	if (!udf_char_to_ustr(&unifilename, sname, flen))
385 		return 0;
386 
387 	if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
388 		namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
389 		if (!namelen)
390 			return 0;
391 	} else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
392 		namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
393 					&unifilename, UDF_NAME_LEN);
394 		if (!namelen)
395 			return 0;
396 	} else
397 		return 0;
398 
399 	return namelen;
400 }
401 
402 #define ILLEGAL_CHAR_MARK	'_'
403 #define EXT_MARK		'.'
404 #define CRC_MARK		'#'
405 #define EXT_SIZE 		5
406 
407 static int udf_translate_to_linux(uint8_t *newName, uint8_t *udfName,
408 				  int udfLen, uint8_t *fidName,
409 				  int fidNameLen)
410 {
411 	int index, newIndex = 0, needsCRC = 0;
412 	int extIndex = 0, newExtIndex = 0, hasExt = 0;
413 	unsigned short valueCRC;
414 	uint8_t curr;
415 	const uint8_t hexChar[] = "0123456789ABCDEF";
416 
417 	if (udfName[0] == '.' &&
418 	    (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
419 		needsCRC = 1;
420 		newIndex = udfLen;
421 		memcpy(newName, udfName, udfLen);
422 	} else {
423 		for (index = 0; index < udfLen; index++) {
424 			curr = udfName[index];
425 			if (curr == '/' || curr == 0) {
426 				needsCRC = 1;
427 				curr = ILLEGAL_CHAR_MARK;
428 				while (index + 1 < udfLen &&
429 						(udfName[index + 1] == '/' ||
430 						 udfName[index + 1] == 0))
431 					index++;
432 			}
433 			if (curr == EXT_MARK &&
434 					(udfLen - index - 1) <= EXT_SIZE) {
435 				if (udfLen == index + 1)
436 					hasExt = 0;
437 				else {
438 					hasExt = 1;
439 					extIndex = index;
440 					newExtIndex = newIndex;
441 				}
442 			}
443 			if (newIndex < 256)
444 				newName[newIndex++] = curr;
445 			else
446 				needsCRC = 1;
447 		}
448 	}
449 	if (needsCRC) {
450 		uint8_t ext[EXT_SIZE];
451 		int localExtIndex = 0;
452 
453 		if (hasExt) {
454 			int maxFilenameLen;
455 			for (index = 0;
456 			     index < EXT_SIZE && extIndex + index + 1 < udfLen;
457 			     index++) {
458 				curr = udfName[extIndex + index + 1];
459 
460 				if (curr == '/' || curr == 0) {
461 					needsCRC = 1;
462 					curr = ILLEGAL_CHAR_MARK;
463 					while (extIndex + index + 2 < udfLen &&
464 					      (index + 1 < EXT_SIZE &&
465 						(udfName[extIndex + index + 2] == '/' ||
466 						 udfName[extIndex + index + 2] == 0)))
467 						index++;
468 				}
469 				ext[localExtIndex++] = curr;
470 			}
471 			maxFilenameLen = 250 - localExtIndex;
472 			if (newIndex > maxFilenameLen)
473 				newIndex = maxFilenameLen;
474 			else
475 				newIndex = newExtIndex;
476 		} else if (newIndex > 250)
477 			newIndex = 250;
478 		newName[newIndex++] = CRC_MARK;
479 		valueCRC = crc_itu_t(0, fidName, fidNameLen);
480 		newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12];
481 		newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8];
482 		newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4];
483 		newName[newIndex++] = hexChar[(valueCRC & 0x000f)];
484 
485 		if (hasExt) {
486 			newName[newIndex++] = EXT_MARK;
487 			for (index = 0; index < localExtIndex; index++)
488 				newName[newIndex++] = ext[index];
489 		}
490 	}
491 
492 	return newIndex;
493 }
494