source3/lib/util_unistr.c

   1 /*
   2    Unix SMB/CIFS implementation.
   3    Samba utility functions
   4    Copyright (C) Andrew Tridgell 1992-2001
   5    Copyright (C) Simo Sorce 2001
   6    Copyright (C) Jeremy Allison 2005
   7
   8    This program is free software; you can redistribute it and/or modify
   9    it under the terms of the GNU General Public License as published by
  10    the Free Software Foundation; either version 3 of the License, or
  11    (at your option) any later version.
  12
  13    This program is distributed in the hope that it will be useful,
  14    but WITHOUT ANY WARRANTY; without even the implied warranty of
  15    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16    GNU General Public License for more details.
  17
  18    You should have received a copy of the GNU General Public License
  19    along with this program.  If not, see <http://www.gnu.org/licenses/>.
  20 */
  21
  22 #include "includes.h"
  23
  24 /* these 3 tables define the unicode case handling.  They are loaded
  25    at startup either via mmap() or read() from the lib directory */
  26 static uint8 *valid_table;
  27 static bool initialized;
  28
  29 /**
  30  * Destroy global objects allocated by load_case_tables()
  31  **/
  32 void gfree_case_tables(void)
  33 {
  34         if ( valid_table ) {
  35                 unmap_file(valid_table, 0x10000);
  36                 valid_table = NULL;
  37         }
  38         initialized = false;
  39 }
  40
  41 /**
  42  * Load the valid character map table from <tt>valid.dat</tt> or
  43  * create from the configured codepage.
  44  *
  45  * This function is called whenever the configuration is reloaded.
  46  * However, the valid character table is not changed if it's loaded
  47  * from a file, because we can't unmap files.
  48  **/
  49
  50 static void init_valid_table(void)
  51 {
  52         if (valid_table) {
  53                 return;
  54         }
  55
  56         valid_table = (uint8 *)map_file(data_path("valid.dat"), 0x10000);
  57         if (!valid_table) {
  58                 smb_panic("Could not load valid.dat file required for mangle method=hash");
  59                 return;
  60         }
  61 }
  62
  63 /*******************************************************************
  64  Write a string in (little-endian) unicode format. src is in
  65  the current DOS codepage. len is the length in bytes of the
  66  string pointed to by dst.
  67
  68  if null_terminate is True then null terminate the packet (adds 2 bytes)
  69
  70  the return value is the length in bytes consumed by the string, including the
  71  null termination if applied
  72 ********************************************************************/
  73
  74 size_t dos_PutUniCode(char *dst,const char *src, size_t len, bool null_terminate)
  75 {
  76         int flags = null_terminate ? STR_UNICODE|STR_NOALIGN|STR_TERMINATE
  77                                    : STR_UNICODE|STR_NOALIGN;
  78         return push_ucs2(NULL, dst, src, len, flags);
  79 }
  80
  81
  82 /* Converts a string from internal samba format to unicode
  83  */
  84
  85 int rpcstr_push(void *dest, const char *src, size_t dest_len, int flags)
  86 {
  87         return push_ucs2(NULL, dest, src, dest_len, flags|STR_UNICODE|STR_NOALIGN);
  88 }
  89
  90 /* Converts a string from internal samba format to unicode. Always terminates.
  91  * Actually just a wrapper round push_ucs2_talloc().
  92  */
  93
  94 int rpcstr_push_talloc(TALLOC_CTX *ctx, smb_ucs2_t **dest, const char *src)
  95 {
  96         size_t size;
  97         if (push_ucs2_talloc(ctx, dest, src, &size))
  98                 return size;
  99         else
 100                 return -1;
 101 }
 102
 103 /*******************************************************************
 104  Determine if a character is valid in a 8.3 name.
 105 ********************************************************************/
 106
 107 bool isvalid83_w(smb_ucs2_t c)
 108 {
 109         init_valid_table();
 110         return valid_table[SVAL(&c,0)] != 0;
 111 }
 112
 113 /*******************************************************************
 114  Count the number of characters in a smb_ucs2_t string.
 115 ********************************************************************/
 116
 117 size_t strlen_w(const smb_ucs2_t *src)
 118 {
 119         size_t len;
 120         smb_ucs2_t c;
 121
 122         for(len = 0; *(COPY_UCS2_CHAR(&c,src)); src++, len++) {
 123                 ;
 124         }
 125
 126         return len;
 127 }
 128
 129 /*******************************************************************
 130  Count up to max number of characters in a smb_ucs2_t string.
 131 ********************************************************************/
 132
 133 size_t strnlen_w(const smb_ucs2_t *src, size_t max)
 134 {
 135         size_t len;
 136         smb_ucs2_t c;
 137
 138         for(len = 0; (len < max) && *(COPY_UCS2_CHAR(&c,src)); src++, len++) {
 139                 ;
 140         }
 141
 142         return len;
 143 }
 144
 145 /*******************************************************************
 146  Wide strchr().
 147 ********************************************************************/
 148
 149 smb_ucs2_t *strchr_w(const smb_ucs2_t *s, smb_ucs2_t c)
 150 {
 151         smb_ucs2_t cp;
 152         while (*(COPY_UCS2_CHAR(&cp,s))) {
 153                 if (c == cp) {
 154                         return (smb_ucs2_t *)s;
 155                 }
 156                 s++;
 157         }
 158         if (c == cp) {
 159                 return (smb_ucs2_t *)s;
 160         }
 161
 162         return NULL;
 163 }
 164
 165 smb_ucs2_t *strchr_wa(const smb_ucs2_t *s, char c)
 166 {
 167         return strchr_w(s, UCS2_CHAR(c));
 168 }
 169
 170 /*******************************************************************
 171  Wide strrchr().
 172 ********************************************************************/
 173
 174 smb_ucs2_t *strrchr_w(const smb_ucs2_t *s, smb_ucs2_t c)
 175 {
 176         smb_ucs2_t cp;
 177         const smb_ucs2_t *p = s;
 178         int len = strlen_w(s);
 179
 180         if (len == 0) {
 181                 return NULL;
 182         }
 183         p += (len - 1);
 184         do {
 185                 if (c == *(COPY_UCS2_CHAR(&cp,p))) {
 186                         return (smb_ucs2_t *)p;
 187                 }
 188         } while (p-- != s);
 189         return NULL;
 190 }
 191
 192 /*******************************************************************
 193  Wide version of strrchr that returns after doing strrchr 'n' times.
 194 ********************************************************************/
 195
 196 smb_ucs2_t *strnrchr_w(const smb_ucs2_t *s, smb_ucs2_t c, unsigned int n)
 197 {
 198         smb_ucs2_t cp;
 199         const smb_ucs2_t *p = s;
 200         int len = strlen_w(s);
 201
 202         if (len == 0 || !n) {
 203                 return NULL;
 204         }
 205         p += (len - 1);
 206         do {
 207                 if (c == *(COPY_UCS2_CHAR(&cp,p))) {
 208                         n--;
 209                 }
 210
 211                 if (!n) {
 212                         return (smb_ucs2_t *)p;
 213                 }
 214         } while (p-- != s);
 215         return NULL;
 216 }
 217
 218 /*******************************************************************
 219  Wide strstr().
 220 ********************************************************************/
 221
 222 smb_ucs2_t *strstr_w(const smb_ucs2_t *s, const smb_ucs2_t *ins)
 223 {
 224         smb_ucs2_t *r;
 225         size_t inslen;
 226
 227         if (!s || !*s || !ins || !*ins) {
 228                 return NULL;
 229         }
 230
 231         inslen = strlen_w(ins);
 232         r = (smb_ucs2_t *)s;
 233
 234         while ((r = strchr_w(r, *ins))) {
 235                 if (strncmp_w(r, ins, inslen) == 0) {
 236                         return r;
 237                 }
 238                 r++;
 239         }
 240
 241         return NULL;
 242 }
 243
 244 /*******************************************************************
 245  Convert a string to lower case.
 246  return True if any char is converted
 247
 248  This is unsafe for any string involving a UTF16 character
 249 ********************************************************************/
 250
 251 bool strlower_w(smb_ucs2_t *s)
 252 {
 253         smb_ucs2_t cp;
 254         bool ret = False;
 255
 256         while (*(COPY_UCS2_CHAR(&cp,s))) {
 257                 smb_ucs2_t v = tolower_m(cp);
 258                 if (v != cp) {
 259                         COPY_UCS2_CHAR(s,&v);
 260                         ret = True;
 261                 }
 262                 s++;
 263         }
 264         return ret;
 265 }
 266
 267 /*******************************************************************
 268  Convert a string to upper case.
 269  return True if any char is converted
 270
 271  This is unsafe for any string involving a UTF16 character
 272 ********************************************************************/
 273
 274 bool strupper_w(smb_ucs2_t *s)
 275 {
 276         smb_ucs2_t cp;
 277         bool ret = False;
 278         while (*(COPY_UCS2_CHAR(&cp,s))) {
 279                 smb_ucs2_t v = toupper_m(cp);
 280                 if (v != cp) {
 281                         COPY_UCS2_CHAR(s,&v);
 282                         ret = True;
 283                 }
 284                 s++;
 285         }
 286         return ret;
 287 }
 288
 289 /*******************************************************************
 290  Convert a string to "normal" form.
 291 ********************************************************************/
 292
 293 void strnorm_w(smb_ucs2_t *s, int case_default)
 294 {
 295         if (case_default == CASE_UPPER) {
 296                 strupper_w(s);
 297         } else {
 298                 strlower_w(s);
 299         }
 300 }
 301
 302 int strcmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b)
 303 {
 304         smb_ucs2_t cpa, cpb;
 305
 306         while ((*(COPY_UCS2_CHAR(&cpb,b))) && (*(COPY_UCS2_CHAR(&cpa,a)) == cpb)) {
 307                 a++;
 308                 b++;
 309         }
 310         return (*(COPY_UCS2_CHAR(&cpa,a)) - *(COPY_UCS2_CHAR(&cpb,b)));
 311         /* warning: if *a != *b and both are not 0 we return a random
 312                 greater or lesser than 0 number not realted to which
 313                 string is longer */
 314 }
 315
 316 int strncmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b, size_t len)
 317 {
 318         smb_ucs2_t cpa, cpb;
 319         size_t n = 0;
 320
 321         while ((n < len) && (*(COPY_UCS2_CHAR(&cpb,b))) && (*(COPY_UCS2_CHAR(&cpa,a)) == cpb)) {
 322                 a++;
 323                 b++;
 324                 n++;
 325         }
 326         return (len - n)?(*(COPY_UCS2_CHAR(&cpa,a)) - *(COPY_UCS2_CHAR(&cpb,b))):0;
 327 }
 328
 329 /*******************************************************************
 330  Case insensitive string comparison.
 331 ********************************************************************/
 332
 333 int strcasecmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b)
 334 {
 335         smb_ucs2_t cpa, cpb;
 336
 337         while ((*COPY_UCS2_CHAR(&cpb,b)) && toupper_m(*(COPY_UCS2_CHAR(&cpa,a))) == toupper_m(cpb)) {
 338                 a++;
 339                 b++;
 340         }
 341         return (tolower_m(*(COPY_UCS2_CHAR(&cpa,a))) - tolower_m(*(COPY_UCS2_CHAR(&cpb,b))));
 342 }
 343
 344 /*******************************************************************
 345  Case insensitive string comparison, length limited.
 346 ********************************************************************/
 347
 348 int strncasecmp_w(const smb_ucs2_t *a, const smb_ucs2_t *b, size_t len)
 349 {
 350         smb_ucs2_t cpa, cpb;
 351         size_t n = 0;
 352
 353         while ((n < len) && *COPY_UCS2_CHAR(&cpb,b) && (toupper_m(*(COPY_UCS2_CHAR(&cpa,a))) == toupper_m(cpb))) {
 354                 a++;
 355                 b++;
 356                 n++;
 357         }
 358         return (len - n)?(tolower_m(*(COPY_UCS2_CHAR(&cpa,a))) - tolower_m(*(COPY_UCS2_CHAR(&cpb,b)))):0;
 359 }
 360
 361 /*******************************************************************
 362  Compare 2 strings.
 363 ********************************************************************/
 364
 365 bool strequal_w(const smb_ucs2_t *s1, const smb_ucs2_t *s2)
 366 {
 367         if (s1 == s2) {
 368                 return(True);
 369         }
 370         if (!s1 || !s2) {
 371                 return(False);
 372         }
 373
 374         return(strcasecmp_w(s1,s2)==0);
 375 }
 376
 377 /*******************************************************************
 378  Compare 2 strings up to and including the nth char.
 379 ******************************************************************/
 380
 381 bool strnequal_w(const smb_ucs2_t *s1,const smb_ucs2_t *s2,size_t n)
 382 {
 383         if (s1 == s2) {
 384                 return(True);
 385         }
 386         if (!s1 || !s2 || !n) {
 387                 return(False);
 388         }
 389
 390         return(strncasecmp_w(s1,s2,n)==0);
 391 }
 392
 393 /*******************************************************************
 394  Duplicate string.
 395 ********************************************************************/
 396
 397 smb_ucs2_t *strdup_w(const smb_ucs2_t *src)
 398 {
 399         return strndup_w(src, 0);
 400 }
 401
 402 /* if len == 0 then duplicate the whole string */
 403
 404 smb_ucs2_t *strndup_w(const smb_ucs2_t *src, size_t len)
 405 {
 406         smb_ucs2_t *dest;
 407
 408         if (!len) {
 409                 len = strlen_w(src);
 410         }
 411         dest = SMB_MALLOC_ARRAY(smb_ucs2_t, len + 1);
 412         if (!dest) {
 413                 DEBUG(0,("strdup_w: out of memory!\n"));
 414                 return NULL;
 415         }
 416
 417         memcpy(dest, src, len * sizeof(smb_ucs2_t));
 418         dest[len] = 0;
 419         return dest;
 420 }
 421
 422 /*******************************************************************
 423  Copy a string with max len.
 424 ********************************************************************/
 425
 426 smb_ucs2_t *strncpy_w(smb_ucs2_t *dest, const smb_ucs2_t *src, const size_t max)
 427 {
 428         smb_ucs2_t cp;
 429         size_t len;
 430
 431         if (!dest || !src) {
 432                 return NULL;
 433         }
 434
 435         for (len = 0; (*COPY_UCS2_CHAR(&cp,(src+len))) && (len < max); len++) {
 436                 cp = *COPY_UCS2_CHAR(dest+len,src+len);
 437         }
 438         cp = 0;
 439         for ( /*nothing*/ ; len < max; len++ ) {
 440                 cp = *COPY_UCS2_CHAR(dest+len,&cp);
 441         }
 442
 443         return dest;
 444 }
 445
 446 /*******************************************************************
 447  Append a string of len bytes and add a terminator.
 448 ********************************************************************/
 449
 450 smb_ucs2_t *strncat_w(smb_ucs2_t *dest, const smb_ucs2_t *src, const size_t max)
 451 {
 452         size_t start;
 453         size_t len;
 454         smb_ucs2_t z = 0;
 455
 456         if (!dest || !src) {
 457                 return NULL;
 458         }
 459
 460         start = strlen_w(dest);
 461         len = strnlen_w(src, max);
 462
 463         memcpy(&dest[start], src, len*sizeof(smb_ucs2_t));
 464         z = *COPY_UCS2_CHAR(dest+start+len,&z);
 465
 466         return dest;
 467 }
 468
 469 smb_ucs2_t *strcat_w(smb_ucs2_t *dest, const smb_ucs2_t *src)
 470 {
 471         size_t start;
 472         size_t len;
 473         smb_ucs2_t z = 0;
 474
 475         if (!dest || !src) {
 476                 return NULL;
 477         }
 478
 479         start = strlen_w(dest);
 480         len = strlen_w(src);
 481
 482         memcpy(&dest[start], src, len*sizeof(smb_ucs2_t));
 483         z = *COPY_UCS2_CHAR(dest+start+len,&z);
 484
 485         return dest;
 486 }
 487
 488
 489 /*******************************************************************
 490  Replace any occurence of oldc with newc in unicode string.
 491 ********************************************************************/
 492
 493 void string_replace_w(smb_ucs2_t *s, smb_ucs2_t oldc, smb_ucs2_t newc)
 494 {
 495         smb_ucs2_t cp;
 496
 497         for(;*(COPY_UCS2_CHAR(&cp,s));s++) {
 498                 if(cp==oldc) {
 499                         COPY_UCS2_CHAR(s,&newc);
 500                 }
 501         }
 502 }
 503
 504 /*******************************************************************
 505  Trim unicode string.
 506 ********************************************************************/
 507
 508 bool trim_string_w(smb_ucs2_t *s, const smb_ucs2_t *front,
 509                                   const smb_ucs2_t *back)
 510 {
 511         bool ret = False;
 512         size_t len, front_len, back_len;
 513
 514         if (!s) {
 515                 return False;
 516         }
 517
 518         len = strlen_w(s);
 519
 520         if (front && *front) {
 521                 front_len = strlen_w(front);
 522                 while (len && strncmp_w(s, front, front_len) == 0) {
 523                         memmove(s, (s + front_len), (len - front_len + 1) * sizeof(smb_ucs2_t));
 524                         len -= front_len;
 525                         ret = True;
 526                 }
 527         }
 528
 529         if (back && *back) {
 530                 back_len = strlen_w(back);
 531                 while (len && strncmp_w((s + (len - back_len)), back, back_len) == 0) {
 532                         s[len - back_len] = 0;
 533                         len -= back_len;
 534                         ret = True;
 535                 }
 536         }
 537
 538         return ret;
 539 }
 540
 541 /*
 542   The *_wa() functions take a combination of 7 bit ascii
 543   and wide characters They are used so that you can use string
 544   functions combining C string constants with ucs2 strings
 545
 546   The char* arguments must NOT be multibyte - to be completely sure
 547   of this only pass string constants */
 548
 549 int strcmp_wa(const smb_ucs2_t *a, const char *b)
 550 {
 551         smb_ucs2_t cp = 0;
 552
 553         while (*b && *(COPY_UCS2_CHAR(&cp,a)) == UCS2_CHAR(*b)) {
 554                 a++;
 555                 b++;
 556         }
 557         return (*(COPY_UCS2_CHAR(&cp,a)) - UCS2_CHAR(*b));
 558 }
 559
 560 int strncmp_wa(const smb_ucs2_t *a, const char *b, size_t len)
 561 {
 562         smb_ucs2_t cp = 0;
 563         size_t n = 0;
 564
 565         while ((n < len) && *b && *(COPY_UCS2_CHAR(&cp,a)) == UCS2_CHAR(*b)) {
 566                 a++;
 567                 b++;
 568                 n++;
 569         }
 570         return (len - n)?(*(COPY_UCS2_CHAR(&cp,a)) - UCS2_CHAR(*b)):0;
 571 }
 572
 573 smb_ucs2_t *strpbrk_wa(const smb_ucs2_t *s, const char *p)
 574 {
 575         smb_ucs2_t cp;
 576
 577         while (*(COPY_UCS2_CHAR(&cp,s))) {
 578                 int i;
 579                 for (i=0; p[i] && cp != UCS2_CHAR(p[i]); i++)
 580                         ;
 581                 if (p[i]) {
 582                         return (smb_ucs2_t *)s;
 583                 }
 584                 s++;
 585         }
 586         return NULL;
 587 }
 588
 589 smb_ucs2_t *strstr_wa(const smb_ucs2_t *s, const char *ins)
 590 {
 591         smb_ucs2_t *r;
 592         size_t inslen;
 593
 594         if (!s || !ins) {
 595                 return NULL;
 596         }
 597
 598         inslen = strlen(ins);
 599         r = (smb_ucs2_t *)s;
 600
 601         while ((r = strchr_w(r, UCS2_CHAR(*ins)))) {
 602                 if (strncmp_wa(r, ins, inslen) == 0)
 603                         return r;
 604                 r++;
 605         }
 606
 607         return NULL;
 608 }
 609
 610 /*************************************************************
 611  ascii only toupper - saves the need for smbd to be in C locale.
 612 *************************************************************/
 613
 614 int toupper_ascii(int c)
 615 {
 616         smb_ucs2_t uc = toupper_m(UCS2_CHAR(c));
 617         return UCS2_TO_CHAR(uc);
 618 }
 619
 620 /*************************************************************
 621  ascii only tolower - saves the need for smbd to be in C locale.
 622 *************************************************************/
 623
 624 int tolower_ascii(int c)
 625 {
 626         smb_ucs2_t uc = tolower_m(UCS2_CHAR(c));
 627         return UCS2_TO_CHAR(uc);
 628 }
 629
 630 /*************************************************************
 631  ascii only isupper - saves the need for smbd to be in C locale.
 632 *************************************************************/
 633
 634 int isupper_ascii(int c)
 635 {
 636         return isupper_m(UCS2_CHAR(c));
 637 }
 638
 639 /*************************************************************
 640  ascii only islower - saves the need for smbd to be in C locale.
 641 *************************************************************/
 642
 643 int islower_ascii(int c)
 644 {
 645         return islower_m(UCS2_CHAR(c));
 646 }