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         memset(dest, 0, sizeof(struct ustr));
  37         memcpy(dest->u_name, src, strlen);
  38         dest->u_cmpID = 0x08;
  39         dest->u_len = strlen;
  40         return strlen;
  41 }
  42
  43 /*
  44  * udf_build_ustr
  45  */
  46 int udf_build_ustr(struct ustr *dest, dstring * ptr, int size)
  47 {
  48         int usesize;
  49
  50         if ((!dest) || (!ptr) || (!size))
  51                 return -1;
  52
  53         memset(dest, 0, sizeof(struct ustr));
  54         usesize = (size > UDF_NAME_LEN) ? UDF_NAME_LEN : size;
  55         dest->u_cmpID = ptr[0];
  56         dest->u_len = ptr[size - 1];
  57         memcpy(dest->u_name, ptr + 1, usesize - 1);
  58         return 0;
  59 }
  60
  61 /*
  62  * udf_build_ustr_exact
  63  */
  64 static int udf_build_ustr_exact(struct ustr *dest, dstring * ptr, int exactsize)
  65 {
  66         if ((!dest) || (!ptr) || (!exactsize))
  67                 return -1;
  68
  69         memset(dest, 0, sizeof(struct ustr));
  70         dest->u_cmpID = ptr[0];
  71         dest->u_len = exactsize - 1;
  72         memcpy(dest->u_name, ptr + 1, exactsize - 1);
  73         return 0;
  74 }
  75
  76 /*
  77  * udf_ocu_to_utf8
  78  *
  79  * PURPOSE
  80  *      Convert OSTA Compressed Unicode to the UTF-8 equivalent.
  81  *
  82  * DESCRIPTION
  83  *      This routine is only called by udf_filldir().
  84  *
  85  * PRE-CONDITIONS
  86  *      utf                     Pointer to UTF-8 output buffer.
  87  *      ocu                     Pointer to OSTA Compressed Unicode input buffer
  88  *                              of size UDF_NAME_LEN bytes.
  89  *                              both of type "struct ustr *"
  90  *
  91  * POST-CONDITIONS
  92  *      <return>                Zero on success.
  93  *
  94  * HISTORY
  95  *      November 12, 1997 - Andrew E. Mileski
  96  *      Written, tested, and released.
  97  */
  98 int udf_CS0toUTF8(struct ustr *utf_o, struct ustr *ocu_i)
  99 {
 100         uint8_t *ocu;
 101         uint32_t c;
 102         uint8_t cmp_id, ocu_len;
 103         int i;
 104
 105         ocu = ocu_i->u_name;
 106
 107         ocu_len = ocu_i->u_len;
 108         cmp_id = ocu_i->u_cmpID;
 109         utf_o->u_len = 0;
 110
 111         if (ocu_len == 0) {
 112                 memset(utf_o, 0, sizeof(struct ustr));
 113                 utf_o->u_cmpID = 0;
 114                 utf_o->u_len = 0;
 115                 return 0;
 116         }
 117
 118         if ((cmp_id != 8) && (cmp_id != 16)) {
 119                 printk(KERN_ERR "udf: unknown compression code (%d) stri=%s\n",
 120                        cmp_id, ocu_i->u_name);
 121                 return 0;
 122         }
 123
 124         for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
 125
 126                 /* Expand OSTA compressed Unicode to Unicode */
 127                 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 | ((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                                 continue;
 219                         } else
 220                                 /* Single byte UTF-8 character (most common) */
 221                                 utf_char = c;
 222                 }
 223
 224                 /* Choose no compression if necessary */
 225                 if (utf_char > max_val) {
 226                         if (0xffU == max_val) {
 227                                 max_val = 0xffffU;
 228                                 ocu[0] = (uint8_t) 0x10U;
 229                                 goto try_again;
 230                         }
 231                         goto error_out;
 232                 }
 233
 234                 if (max_val == 0xffffU) {
 235                         ocu[++u_len] = (uint8_t) (utf_char >> 8);
 236                 }
 237                 ocu[++u_len] = (uint8_t) (utf_char & 0xffU);
 238         }
 239
 240         if (utf_cnt) {
 241               error_out:
 242                 ocu[++u_len] = '?';
 243                 printk(KERN_DEBUG "udf: bad UTF-8 character\n");
 244         }
 245
 246         ocu[length - 1] = (uint8_t) u_len + 1;
 247         return u_len + 1;
 248 }
 249
 250 static int udf_CS0toNLS(struct nls_table *nls, struct ustr *utf_o,
 251                         struct ustr *ocu_i)
 252 {
 253         uint8_t *ocu;
 254         uint32_t c;
 255         uint8_t cmp_id, ocu_len;
 256         int i;
 257
 258         ocu = ocu_i->u_name;
 259
 260         ocu_len = ocu_i->u_len;
 261         cmp_id = ocu_i->u_cmpID;
 262         utf_o->u_len = 0;
 263
 264         if (ocu_len == 0) {
 265                 memset(utf_o, 0, sizeof(struct ustr));
 266                 utf_o->u_cmpID = 0;
 267                 utf_o->u_len = 0;
 268                 return 0;
 269         }
 270
 271         if ((cmp_id != 8) && (cmp_id != 16)) {
 272                 printk(KERN_ERR "udf: unknown compression code (%d) stri=%s\n",
 273                        cmp_id, ocu_i->u_name);
 274                 return 0;
 275         }
 276
 277         for (i = 0; (i < ocu_len) && (utf_o->u_len <= (UDF_NAME_LEN - 3));) {
 278                 /* Expand OSTA compressed Unicode to Unicode */
 279                 c = ocu[i++];
 280                 if (cmp_id == 16)
 281                         c = (c << 8) | ocu[i++];
 282
 283                 utf_o->u_len += nls->uni2char(c, &utf_o->u_name[utf_o->u_len],
 284                                               UDF_NAME_LEN - utf_o->u_len);
 285         }
 286         utf_o->u_cmpID = 8;
 287
 288         return utf_o->u_len;
 289 }
 290
 291 static int udf_NLStoCS0(struct nls_table *nls, dstring * ocu, struct ustr *uni,
 292                         int length)
 293 {
 294         unsigned len, i, max_val;
 295         uint16_t uni_char;
 296         int u_len;
 297
 298         memset(ocu, 0, sizeof(dstring) * length);
 299         ocu[0] = 8;
 300         max_val = 0xffU;
 301
 302       try_again:
 303         u_len = 0U;
 304         for (i = 0U; i < uni->u_len; i++) {
 305                 len = nls->char2uni(&uni->u_name[i], uni->u_len - i, &uni_char);
 306                 if (len <= 0)
 307                         continue;
 308
 309                 if (uni_char > max_val) {
 310                         max_val = 0xffffU;
 311                         ocu[0] = (uint8_t) 0x10U;
 312                         goto try_again;
 313                 }
 314
 315                 if (max_val == 0xffffU)
 316                         ocu[++u_len] = (uint8_t) (uni_char >> 8);
 317                 ocu[++u_len] = (uint8_t) (uni_char & 0xffU);
 318                 i += len - 1;
 319         }
 320
 321         ocu[length - 1] = (uint8_t) u_len + 1;
 322         return u_len + 1;
 323 }
 324
 325 int udf_get_filename(struct super_block *sb, uint8_t * sname, uint8_t * dname,
 326                      int flen)
 327 {
 328         struct ustr filename, unifilename;
 329         int len;
 330
 331         if (udf_build_ustr_exact(&unifilename, sname, flen)) {
 332                 return 0;
 333         }
 334
 335         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 336                 if (!udf_CS0toUTF8(&filename, &unifilename)) {
 337                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 338                                   sname);
 339                         return 0;
 340                 }
 341         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 342                 if (!udf_CS0toNLS
 343                     (UDF_SB(sb)->s_nls_map, &filename, &unifilename)) {
 344                         udf_debug("Failed in udf_get_filename: sname = %s\n",
 345                                   sname);
 346                         return 0;
 347                 }
 348         } else
 349                 return 0;
 350
 351         if ((len =
 352              udf_translate_to_linux(dname, filename.u_name, filename.u_len,
 353                                     unifilename.u_name, unifilename.u_len))) {
 354                 return len;
 355         }
 356         return 0;
 357 }
 358
 359 int udf_put_filename(struct super_block *sb, const uint8_t * sname,
 360                      uint8_t * dname, int flen)
 361 {
 362         struct ustr unifilename;
 363         int namelen;
 364
 365         if (!(udf_char_to_ustr(&unifilename, sname, flen))) {
 366                 return 0;
 367         }
 368
 369         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 370                 if (!
 371                     (namelen =
 372                      udf_UTF8toCS0(dname, &unifilename, UDF_NAME_LEN))) {
 373                         return 0;
 374                 }
 375         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 376                 if (!
 377                     (namelen =
 378                      udf_NLStoCS0(UDF_SB(sb)->s_nls_map, dname, &unifilename,
 379                                   UDF_NAME_LEN))) {
 380                         return 0;
 381                 }
 382         } else
 383                 return 0;
 384
 385         return namelen;
 386 }
 387
 388 #define ILLEGAL_CHAR_MARK       '_'
 389 #define EXT_MARK                        '.'
 390 #define CRC_MARK                        '#'
 391 #define EXT_SIZE                        5
 392
 393 static int udf_translate_to_linux(uint8_t * newName, uint8_t * udfName,
 394                                   int udfLen, uint8_t * fidName, int fidNameLen)
 395 {
 396         int index, newIndex = 0, needsCRC = 0;
 397         int extIndex = 0, newExtIndex = 0, hasExt = 0;
 398         unsigned short valueCRC;
 399         uint8_t curr;
 400         const uint8_t hexChar[] = "0123456789ABCDEF";
 401
 402         if (udfName[0] == '.' && (udfLen == 1 ||
 403                                   (udfLen == 2 && udfName[1] == '.'))) {
 404                 needsCRC = 1;
 405                 newIndex = udfLen;
 406                 memcpy(newName, udfName, udfLen);
 407         } else {
 408                 for (index = 0; index < udfLen; index++) {
 409                         curr = udfName[index];
 410                         if (curr == '/' || curr == 0) {
 411                                 needsCRC = 1;
 412                                 curr = ILLEGAL_CHAR_MARK;
 413                                 while (index + 1 < udfLen
 414                                        && (udfName[index + 1] == '/'
 415                                            || udfName[index + 1] == 0))
 416                                         index++;
 417                         }
 418                         if (curr == EXT_MARK
 419                             && (udfLen - index - 1) <= EXT_SIZE) {
 420                                 if (udfLen == index + 1)
 421                                         hasExt = 0;
 422                                 else {
 423                                         hasExt = 1;
 424                                         extIndex = index;
 425                                         newExtIndex = newIndex;
 426                                 }
 427                         }
 428                         if (newIndex < 256)
 429                                 newName[newIndex++] = curr;
 430                         else
 431                                 needsCRC = 1;
 432                 }
 433         }
 434         if (needsCRC) {
 435                 uint8_t ext[EXT_SIZE];
 436                 int localExtIndex = 0;
 437
 438                 if (hasExt) {
 439                         int maxFilenameLen;
 440                         for (index = 0;
 441                              index < EXT_SIZE && extIndex + index + 1 < udfLen;
 442                              index++) {
 443                                 curr = udfName[extIndex + index + 1];
 444
 445                                 if (curr == '/' || curr == 0) {
 446                                         needsCRC = 1;
 447                                         curr = ILLEGAL_CHAR_MARK;
 448                                         while (extIndex + index + 2 < udfLen
 449                                                && (index + 1 < EXT_SIZE
 450                                                    &&
 451                                                    (udfName
 452                                                     [extIndex + index + 2] ==
 453                                                     '/'
 454                                                     || udfName[extIndex +
 455                                                                index + 2] ==
 456                                                     0)))
 457                                                 index++;
 458                                 }
 459                                 ext[localExtIndex++] = curr;
 460                         }
 461                         maxFilenameLen = 250 - localExtIndex;
 462                         if (newIndex > maxFilenameLen)
 463                                 newIndex = maxFilenameLen;
 464                         else
 465                                 newIndex = newExtIndex;
 466                 } else if (newIndex > 250)
 467                         newIndex = 250;
 468                 newName[newIndex++] = CRC_MARK;
 469                 valueCRC = udf_crc(fidName, fidNameLen, 0);
 470                 newName[newIndex++] = hexChar[(valueCRC & 0xf000) >> 12];
 471                 newName[newIndex++] = hexChar[(valueCRC & 0x0f00) >> 8];
 472                 newName[newIndex++] = hexChar[(valueCRC & 0x00f0) >> 4];
 473                 newName[newIndex++] = hexChar[(valueCRC & 0x000f)];
 474
 475                 if (hasExt) {
 476                         newName[newIndex++] = EXT_MARK;
 477                         for (index = 0; index < localExtIndex; index++)
 478                                 newName[newIndex++] = ext[index];
 479                 }
 480         }
 481         return newIndex;
 482 }