fs/udf/unicode.c

   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;
 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                 utf_o->u_len += nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
 283                                               UDF_NAME_LEN - utf_o->u_len);
 284         }
 285         utf_o->u_cmpID = 8;
 286
 287         return utf_o->u_len;
 288 }
 289
 290 static int udf_NLStoCS0(struct nls_table *nls, dstring *ocu, struct ustr *uni,
 291                         int length)
 292 {
 293         unsigned len, i, max_val;
 294         uint16_t uni_char;
 295         int u_len;
 296
 297         memset(ocu, 0, sizeof(dstring) * length);
 298         ocu[0] = 8;
 299         max_val = 0xffU;
 300
 301 try_again:
 302         u_len = 0U;
 303         for (i = 0U; i < uni->u_len; i++) {
 304                 len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
 305                 if (len <= 0)
 306                         continue;
 307
 308                 if (uni_char > max_val) {
 309                         max_val = 0xffffU;
 310                         ocu[0] = (uint8_t)0x10U;
 311                         goto try_again;
 312                 }
 313
 314                 if (max_val == 0xffffU)
 315                         ocu[++u_len] = (uint8_t)(uni_char >> 8);
 316                 ocu[++u_len] = (uint8_t)(uni_char & 0xffU);
 317                 i += len - 1;
 318         }
 319
 320         ocu[length - 1] = (uint8_t)u_len + 1;
 321         return u_len + 1;
 322 }
 323
 324 int udf_get_filename(struct super_block *sb, uint8_t *sname, uint8_t *dname,
 325                      int flen)
 326 {
 327         struct ustr filename, unifilename;
 328         int len;
 329
 330         if (udf_build_ustr_exact(&unifilename, sname, flen))
 331                 return 0;
 332
 333         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 334                 if (!udf_CS0toUTF8(&filename, &unifilename)) {
 335                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 336                                   sname);
 337                         return 0;
 338                 }
 339         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 340                 if (!udf_CS0toNLS(UDF_SB(sb)->s_nls_map, &filename,
 341                                   &unifilename)) {
 342                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 343                                   sname);
 344                         return 0;
 345                 }
 346         } else
 347                 return 0;
 348
 349         len = udf_translate_to_linux(dname, filename.u_name, filename.u_len,
 350                                      unifilename.u_name, unifilename.u_len);
 351         if (len)
 352                 return len;
 353
 354         return 0;
 355 }
 356
 357 int udf_put_filename(struct super_block *sb, const uint8_t *sname,
 358                      uint8_t *dname, int flen)
 359 {
 360         struct ustr unifilename;
 361         int namelen;
 362
 363         if (!udf_char_to_ustr(&unifilename, sname, flen))
 364                 return 0;
 365
 366         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 367                 namelen = udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN);
 368                 if (!namelen)
 369                         return 0;
 370         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 371                 namelen = udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname,
 372                                         &unifilename, UDF_NAME_LEN);
 373                 if (!namelen)
 374                         return 0;
 375         } else
 376                 return 0;
 377
 378         return namelen;
 379 }
 380
 381 #define ILLEGAL_CHAR_MARK       '_'
 382 #define EXT_MARK                '.'
 383 #define CRC_MARK                '#'
 384 #define EXT_SIZE                5
 385
 386 static int udf_translate_to_linux(uint8_t *newName, uint8_t *udfName,
 387                                   int udfLen, uint8_t *fidName,
 388                                   int fidNameLen)
 389 {
 390         int index, newIndex = 0, needsCRC = 0;
 391         int extIndex = 0, newExtIndex = 0, hasExt = 0;
 392         unsigned short valueCRC;
 393         uint8_t curr;
 394         const uint8_t hexChar[] = "0123456789ABCDEF";
 395
 396         if (udfName[0] == '.' &&
 397             (udfLen == 1 || (udfLen == 2 && udfName[1] == '.'))) {
 398                 needsCRC = 1;
 399                 newIndex = udfLen;
 400                 memcpy(newName, udfName, udfLen);
 401         } else {
 402                 for (index = 0; index < udfLen; index++) {
 403                         curr = udfName[index];
 404                         if (curr == '/' || curr == 0) {
 405                                 needsCRC = 1;
 406                                 curr = ILLEGAL_CHAR_MARK;
 407                                 while (index + 1 < udfLen &&
 408                                                 (udfName[index + 1] == '/' ||
 409                                                  udfName[index + 1] == 0))
 410                                         index++;
 411                         }
 412                         if (curr == EXT_MARK &&
 413                                         (udfLen - index - 1) <= EXT_SIZE) {
 414                                 if (udfLen == index + 1)
 415                                         hasExt = 0;
 416                                 else {
 417                                         hasExt = 1;
 418                                         extIndex = index;
 419                                         newExtIndex = newIndex;
 420                                 }
 421                         }
 422                         if (newIndex < 256)
 423                                 newName[newIndex++] = curr;
 424                         else
 425                                 needsCRC = 1;
 426                 }
 427         }
 428         if (needsCRC) {
 429                 uint8_t ext[EXT_SIZE];
 430                 int localExtIndex = 0;
 431
 432                 if (hasExt) {
 433                         int maxFilenameLen;
 434                         for (index = 0;
 435                              index < EXT_SIZE && extIndex + index + 1 < udfLen;
 436                              index++) {
 437                                 curr = udfName[extIndex + index + 1];
 438
 439                                 if (curr == '/' || curr == 0) {
 440                                         needsCRC = 1;
 441                                         curr = ILLEGAL_CHAR_MARK;
 442                                         while (extIndex + index + 2 < udfLen &&
 443                                               (index + 1 < EXT_SIZE &&
 444                                                 (udfName[extIndex + index + 2] == '/' ||
 445                                                  udfName[extIndex + index + 2] == 0)))
 446                                                 index++;
 447                                 }
 448                                 ext[localExtIndex++] = curr;
 449                         }
 450                         maxFilenameLen = 250 - localExtIndex;
 451                         if (newIndex > maxFilenameLen)
 452                                 newIndex = maxFilenameLen;
 453                         else
 454                                 newIndex = newExtIndex;
 455                 } else if (newIndex > 250)
 456                         newIndex = 250;
 457                 newName[newIndex++] = CRC_MARK;
 458                 valueCRC = crc_itu_t(0, fidName, fidNameLen);
 459                 newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12];
 460                 newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8];
 461                 newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4];
 462                 newName[newIndex++] = hexChar[(valueCRC & 0x000f)];
 463
 464                 if (hasExt) {
 465                         newName[newIndex++] = EXT_MARK;
 466                         for (index = 0; index < localExtIndex; index++)
 467                                 newName[newIndex++] = ext[index];
 468                 }
 469         }
 470
 471         return newIndex;
 472 }