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