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 #include <linux/slab.h>
  28
  29 #include "udf_sb.h"
  30
  31 #define SURROGATE_MASK 0xfffff800
  32 #define SURROGATE_PAIR 0x0000d800
  33
  34 static int udf_uni2char_utf8(wchar_t uni,
  35                              unsigned char *out,
  36                              int boundlen)
  37 {
  38         int u_len = 0;
  39
  40         if (boundlen <= 0)
  41                 return -ENAMETOOLONG;
  42
  43         if ((uni & SURROGATE_MASK) == SURROGATE_PAIR)
  44                 return -EINVAL;
  45
  46         if (uni < 0x80) {
  47                 out[u_len++] = (unsigned char)uni;
  48         } else if (uni < 0x800) {
  49                 if (boundlen < 2)
  50                         return -ENAMETOOLONG;
  51                 out[u_len++] = (unsigned char)(0xc0 | (uni >> 6));
  52                 out[u_len++] = (unsigned char)(0x80 | (uni & 0x3f));
  53         } else {
  54                 if (boundlen < 3)
  55                         return -ENAMETOOLONG;
  56                 out[u_len++] = (unsigned char)(0xe0 | (uni >> 12));
  57                 out[u_len++] = (unsigned char)(0x80 | ((uni >> 6) & 0x3f));
  58                 out[u_len++] = (unsigned char)(0x80 | (uni & 0x3f));
  59         }
  60         return u_len;
  61 }
  62
  63 static int udf_char2uni_utf8(const unsigned char *in,
  64                              int boundlen,
  65                              wchar_t *uni)
  66 {
  67         unsigned int utf_char;
  68         unsigned char c;
  69         int utf_cnt, u_len;
  70
  71         utf_char = 0;
  72         utf_cnt = 0;
  73         for (u_len = 0; u_len < boundlen;) {
  74                 c = in[u_len++];
  75
  76                 /* Complete a multi-byte UTF-8 character */
  77                 if (utf_cnt) {
  78                         utf_char = (utf_char << 6) | (c & 0x3f);
  79                         if (--utf_cnt)
  80                                 continue;
  81                 } else {
  82                         /* Check for a multi-byte UTF-8 character */
  83                         if (c & 0x80) {
  84                                 /* Start a multi-byte UTF-8 character */
  85                                 if ((c & 0xe0) == 0xc0) {
  86                                         utf_char = c & 0x1f;
  87                                         utf_cnt = 1;
  88                                 } else if ((c & 0xf0) == 0xe0) {
  89                                         utf_char = c & 0x0f;
  90                                         utf_cnt = 2;
  91                                 } else if ((c & 0xf8) == 0xf0) {
  92                                         utf_char = c & 0x07;
  93                                         utf_cnt = 3;
  94                                 } else if ((c & 0xfc) == 0xf8) {
  95                                         utf_char = c & 0x03;
  96                                         utf_cnt = 4;
  97                                 } else if ((c & 0xfe) == 0xfc) {
  98                                         utf_char = c & 0x01;
  99                                         utf_cnt = 5;
 100                                 } else {
 101                                         utf_cnt = -1;
 102                                         break;
 103                                 }
 104                                 continue;
 105                         } else {
 106                                 /* Single byte UTF-8 character (most common) */
 107                                 utf_char = c;
 108                         }
 109                 }
 110                 *uni = utf_char;
 111                 break;
 112         }
 113         if (utf_cnt) {
 114                 *uni = '?';
 115                 return -EINVAL;
 116         }
 117         return u_len;
 118 }
 119
 120 #define ILLEGAL_CHAR_MARK       '_'
 121 #define EXT_MARK                '.'
 122 #define CRC_MARK                '#'
 123 #define EXT_SIZE                5
 124 /* Number of chars we need to store generated CRC to make filename unique */
 125 #define CRC_LEN                 5
 126
 127 static int udf_name_conv_char(uint8_t *str_o, int str_o_max_len,
 128                               int *str_o_idx,
 129                               const uint8_t *str_i, int str_i_max_len,
 130                               int *str_i_idx,
 131                               int u_ch, int *needsCRC,
 132                               int (*conv_f)(wchar_t, unsigned char *, int),
 133                               int translate)
 134 {
 135         uint32_t c;
 136         int illChar = 0;
 137         int len, gotch = 0;
 138
 139         for (; (!gotch) && (*str_i_idx < str_i_max_len); *str_i_idx += u_ch) {
 140                 if (*str_o_idx >= str_o_max_len) {
 141                         *needsCRC = 1;
 142                         return gotch;
 143                 }
 144
 145                 /* Expand OSTA compressed Unicode to Unicode */
 146                 c = str_i[*str_i_idx];
 147                 if (u_ch > 1)
 148                         c = (c << 8) | str_i[*str_i_idx + 1];
 149
 150                 if (translate && (c == '/' || c == 0))
 151                         illChar = 1;
 152                 else if (illChar)
 153                         break;
 154                 else
 155                         gotch = 1;
 156         }
 157         if (illChar) {
 158                 *needsCRC = 1;
 159                 c = ILLEGAL_CHAR_MARK;
 160                 gotch = 1;
 161         }
 162         if (gotch) {
 163                 len = conv_f(c, &str_o[*str_o_idx], str_o_max_len - *str_o_idx);
 164                 /* Valid character? */
 165                 if (len >= 0)
 166                         *str_o_idx += len;
 167                 else if (len == -ENAMETOOLONG) {
 168                         *needsCRC = 1;
 169                         gotch = 0;
 170                 } else {
 171                         str_o[(*str_o_idx)++] = '?';
 172                         *needsCRC = 1;
 173                 }
 174         }
 175         return gotch;
 176 }
 177
 178 static int udf_name_from_CS0(uint8_t *str_o, int str_max_len,
 179                              const uint8_t *ocu, int ocu_len,
 180                              int (*conv_f)(wchar_t, unsigned char *, int),
 181                              int translate)
 182 {
 183         uint32_t c;
 184         uint8_t cmp_id;
 185         int idx, len;
 186         int u_ch;
 187         int needsCRC = 0;
 188         int ext_i_len, ext_max_len;
 189         int str_o_len = 0;      /* Length of resulting output */
 190         int ext_o_len = 0;      /* Extension output length */
 191         int ext_crc_len = 0;    /* Extension output length if used with CRC */
 192         int i_ext = -1;         /* Extension position in input buffer */
 193         int o_crc = 0;          /* Rightmost possible output pos for CRC+ext */
 194         unsigned short valueCRC;
 195         uint8_t ext[EXT_SIZE * NLS_MAX_CHARSET_SIZE + 1];
 196         uint8_t crc[CRC_LEN];
 197
 198         if (str_max_len <= 0)
 199                 return 0;
 200
 201         if (ocu_len == 0) {
 202                 memset(str_o, 0, str_max_len);
 203                 return 0;
 204         }
 205
 206         cmp_id = ocu[0];
 207         if (cmp_id != 8 && cmp_id != 16) {
 208                 memset(str_o, 0, str_max_len);
 209                 pr_err("unknown compression code (%u)\n", cmp_id);
 210                 return -EINVAL;
 211         }
 212         u_ch = cmp_id >> 3;
 213
 214         ocu++;
 215         ocu_len--;
 216
 217         if (ocu_len % u_ch) {
 218                 pr_err("incorrect filename length (%d)\n", ocu_len + 1);
 219                 return -EINVAL;
 220         }
 221
 222         if (translate) {
 223                 /* Look for extension */
 224                 for (idx = ocu_len - u_ch, ext_i_len = 0;
 225                      (idx >= 0) && (ext_i_len < EXT_SIZE);
 226                      idx -= u_ch, ext_i_len++) {
 227                         c = ocu[idx];
 228                         if (u_ch > 1)
 229                                 c = (c << 8) | ocu[idx + 1];
 230
 231                         if (c == EXT_MARK) {
 232                                 if (ext_i_len)
 233                                         i_ext = idx;
 234                                 break;
 235                         }
 236                 }
 237                 if (i_ext >= 0) {
 238                         /* Convert extension */
 239                         ext_max_len = min_t(int, sizeof(ext), str_max_len);
 240                         ext[ext_o_len++] = EXT_MARK;
 241                         idx = i_ext + u_ch;
 242                         while (udf_name_conv_char(ext, ext_max_len, &ext_o_len,
 243                                                   ocu, ocu_len, &idx,
 244                                                   u_ch, &needsCRC,
 245                                                   conv_f, translate)) {
 246                                 if ((ext_o_len + CRC_LEN) < str_max_len)
 247                                         ext_crc_len = ext_o_len;
 248                         }
 249                 }
 250         }
 251
 252         idx = 0;
 253         while (1) {
 254                 if (translate && (idx == i_ext)) {
 255                         if (str_o_len > (str_max_len - ext_o_len))
 256                                 needsCRC = 1;
 257                         break;
 258                 }
 259
 260                 if (!udf_name_conv_char(str_o, str_max_len, &str_o_len,
 261                                         ocu, ocu_len, &idx,
 262                                         u_ch, &needsCRC, conv_f, translate))
 263                         break;
 264
 265                 if (translate &&
 266                     (str_o_len <= (str_max_len - ext_o_len - CRC_LEN)))
 267                         o_crc = str_o_len;
 268         }
 269
 270         if (translate) {
 271                 if (str_o_len <= 2 && str_o[0] == '.' &&
 272                     (str_o_len == 1 || str_o[1] == '.'))
 273                         needsCRC = 1;
 274                 if (needsCRC) {
 275                         str_o_len = o_crc;
 276                         valueCRC = crc_itu_t(0, ocu, ocu_len);
 277                         crc[0] = CRC_MARK;
 278                         crc[1] = hex_asc_upper_hi(valueCRC >> 8);
 279                         crc[2] = hex_asc_upper_lo(valueCRC >> 8);
 280                         crc[3] = hex_asc_upper_hi(valueCRC);
 281                         crc[4] = hex_asc_upper_lo(valueCRC);
 282                         len = min_t(int, CRC_LEN, str_max_len - str_o_len);
 283                         memcpy(&str_o[str_o_len], crc, len);
 284                         str_o_len += len;
 285                         ext_o_len = ext_crc_len;
 286                 }
 287                 if (ext_o_len > 0) {
 288                         memcpy(&str_o[str_o_len], ext, ext_o_len);
 289                         str_o_len += ext_o_len;
 290                 }
 291         }
 292
 293         return str_o_len;
 294 }
 295
 296 static int udf_name_to_CS0(uint8_t *ocu, int ocu_max_len,
 297                            const uint8_t *str_i, int str_len,
 298                            int (*conv_f)(const unsigned char *, int, wchar_t *))
 299 {
 300         int i, len;
 301         unsigned int max_val;
 302         wchar_t uni_char;
 303         int u_len, u_ch;
 304
 305         if (ocu_max_len <= 0)
 306                 return 0;
 307
 308         memset(ocu, 0, ocu_max_len);
 309         ocu[0] = 8;
 310         max_val = 0xff;
 311         u_ch = 1;
 312
 313 try_again:
 314         u_len = 1;
 315         for (i = 0; i < str_len; i++) {
 316                 /* Name didn't fit? */
 317                 if (u_len + u_ch > ocu_max_len)
 318                         return 0;
 319                 len = conv_f(&str_i[i], str_len - i, &uni_char);
 320                 if (!len)
 321                         continue;
 322                 /* Invalid character, deal with it */
 323                 if (len < 0) {
 324                         len = 1;
 325                         uni_char = '?';
 326                 }
 327
 328                 if (uni_char > max_val) {
 329                         max_val = 0xffff;
 330                         ocu[0] = 0x10;
 331                         u_ch = 2;
 332                         goto try_again;
 333                 }
 334
 335                 if (max_val == 0xffff)
 336                         ocu[u_len++] = (uint8_t)(uni_char >> 8);
 337                 ocu[u_len++] = (uint8_t)(uni_char & 0xff);
 338                 i += len - 1;
 339         }
 340
 341         return u_len;
 342 }
 343
 344 /*
 345  * Convert CS0 dstring to output charset. Warning: This function may truncate
 346  * input string if it is too long as it is used for informational strings only
 347  * and it is better to truncate the string than to refuse mounting a media.
 348  */
 349 int udf_dstrCS0toUTF8(uint8_t *utf_o, int o_len,
 350                       const uint8_t *ocu_i, int i_len)
 351 {
 352         int s_len = 0;
 353
 354         if (i_len > 0) {
 355                 s_len = ocu_i[i_len - 1];
 356                 if (s_len >= i_len) {
 357                         pr_warn("incorrect dstring lengths (%d/%d),"
 358                                 " truncating\n", s_len, i_len);
 359                         s_len = i_len - 1;
 360                         /* 2-byte encoding? Need to round properly... */
 361                         if (ocu_i[0] == 16)
 362                                 s_len -= (s_len - 1) & 2;
 363                 }
 364         }
 365
 366         return udf_name_from_CS0(utf_o, o_len, ocu_i, s_len,
 367                                  udf_uni2char_utf8, 0);
 368 }
 369
 370 int udf_get_filename(struct super_block *sb, const uint8_t *sname, int slen,
 371                      uint8_t *dname, int dlen)
 372 {
 373         int (*conv_f)(wchar_t, unsigned char *, int);
 374         int ret;
 375
 376         if (!slen)
 377                 return -EIO;
 378
 379         if (dlen <= 0)
 380                 return 0;
 381
 382         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 383                 conv_f = udf_uni2char_utf8;
 384         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 385                 conv_f = UDF_SB(sb)->s_nls_map->uni2char;
 386         } else
 387                 BUG();
 388
 389         ret = udf_name_from_CS0(dname, dlen, sname, slen, conv_f, 1);
 390         /* Zero length filename isn't valid... */
 391         if (ret == 0)
 392                 ret = -EINVAL;
 393         return ret;
 394 }
 395
 396 int udf_put_filename(struct super_block *sb, const uint8_t *sname, int slen,
 397                      uint8_t *dname, int dlen)
 398 {
 399         int (*conv_f)(const unsigned char *, int, wchar_t *);
 400
 401         if (UDF_QUERY_FLAG(sb, UDF_FLAG_UTF8)) {
 402                 conv_f = udf_char2uni_utf8;
 403         } else if (UDF_QUERY_FLAG(sb, UDF_FLAG_NLS_MAP)) {
 404                 conv_f = UDF_SB(sb)->s_nls_map->char2uni;
 405         } else
 406                 BUG();
 407
 408         return udf_name_to_CS0(dname, dlen, sname, slen, conv_f);
 409 }
 410