usr/src/lib/libc/port/locale/collate.c

   1 /*
   2  * Copyright 2014 Garrett D'Amore <garrett@damore.org>
   3  * Copyright 2010 Nexenta Systems, Inc.  All rights reserved.
   4  * Copyright (c) 1995 Alex Tatmanjants <alex@elvisti.kiev.ua>
   5  *              at Electronni Visti IA, Kiev, Ukraine.
   6  *                      All rights reserved.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in the
  15  *    documentation and/or other materials provided with the distribution.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND
  18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE
  21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  */
  29
  30 #include "lint.h"
  31 #include "file64.h"
  32 #include <stdio.h>
  33 #include <stdlib.h>
  34 #include <stddef.h>
  35 #include <string.h>
  36 #include <wchar.h>
  37 #include <errno.h>
  38 #include <unistd.h>
  39 #include <ctype.h>
  40 #include <unistd.h>
  41 #include <fcntl.h>
  42 #include <assert.h>
  43 #include <sys/stat.h>
  44 #include <sys/mman.h>
  45
  46 #include "collate.h"
  47 #include "localeimpl.h"
  48
  49 /* Check file format vs libc runtime. (See collatefile.h) */
  50 #if COLL_WEIGHTS_MAX != COLLATE_WEIGHTS_MAX
  51 #error "COLL_WEIGHTS_MAX != COLLATE_WEIGHTS_MAX"
  52 #endif
  53
  54 /*
  55  * See the comments in usr/src/cmd/localedef/collate.c for further
  56  * information.  It would also be very helpful to have a copy of the
  57  * POSIX standard for collation (in the locale format manual page)
  58  * handy (www.opengroup.org).
  59  */
  60
  61 /*
  62  * POSIX uses empty tables and falls down to strcmp.
  63  */
  64 struct lc_collate lc_collate_posix = {
  65         .lc_is_posix = 1,
  66 };
  67
  68 struct locdata __posix_collate_locdata = {
  69         .l_lname = "C",
  70         .l_data = { &lc_collate_posix }
  71 };
  72
  73
  74 struct locdata *
  75 __lc_collate_load(const char *locname)
  76 {
  77         int i, chains, z;
  78         char buf[PATH_MAX];
  79         char *TMP;
  80         char *map;
  81         collate_info_t *info;
  82         struct stat sbuf;
  83         int fd;
  84         struct locdata *ldata;
  85         struct lc_collate *lcc;
  86
  87         /*
  88          * Slurp the locale file into the cache.
  89          */
  90
  91         (void) snprintf(buf, sizeof (buf), "%s/%s/LC_COLLATE/LCL_DATA",
  92             _PathLocale, locname);
  93
  94         if ((fd = open(buf, O_RDONLY)) < 0) {
  95                 errno = EINVAL;
  96                 return (NULL);
  97         }
  98         if (fstat(fd, &sbuf) < 0) {
  99                 (void) close(fd);
 100                 errno = EINVAL;
 101                 return (NULL);
 102         }
 103         if (sbuf.st_size < (COLLATE_STR_LEN + sizeof (info))) {
 104                 (void) close(fd);
 105                 errno = EINVAL;
 106                 return (NULL);
 107         }
 108         map = mmap(NULL, sbuf.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 109         (void) close(fd);
 110         if ((TMP = map) == NULL) {
 111                 errno = EINVAL;
 112                 return (NULL);
 113         }
 114
 115         if (strncmp(TMP, COLLATE_VERSION, COLLATE_STR_LEN) != 0) {
 116                 (void) munmap(map, sbuf.st_size);
 117                 errno = EINVAL;
 118                 return (NULL);
 119         }
 120         TMP += COLLATE_STR_LEN;
 121
 122         info = (void *)TMP;
 123         TMP += sizeof (*info);
 124
 125         if ((info->directive_count < 1) ||
 126             (info->directive_count >= COLL_WEIGHTS_MAX) ||
 127             ((chains = info->chain_count) < 0)) {
 128                 (void) munmap(map, sbuf.st_size);
 129                 errno = EINVAL;
 130                 return (NULL);
 131         }
 132
 133         i = (sizeof (collate_char_t) * (UCHAR_MAX + 1)) +
 134             (sizeof (collate_chain_t) * chains) +
 135             (sizeof (collate_large_t) * info->large_count);
 136         for (z = 0; z < info->directive_count; z++) {
 137                 i += sizeof (collate_subst_t) * info->subst_count[z];
 138         }
 139         if (i != (sbuf.st_size - (TMP - map))) {
 140                 (void) munmap(map, sbuf.st_size);
 141                 errno = EINVAL;
 142                 return (NULL);
 143         }
 144
 145
 146         if ((ldata = __locdata_alloc(locname, sizeof (*lcc))) == NULL) {
 147                 (void) munmap(map, sbuf.st_size);
 148                 return (NULL);
 149         }
 150         lcc = ldata->l_data[0];
 151         ldata->l_map = map;
 152         ldata->l_map_len = sbuf.st_size;
 153
 154         lcc->lc_info = info;
 155         lcc->lc_directive_count = info->directive_count;
 156         lcc->lc_large_count = info->large_count;
 157
 158         for (z = 0; z < COLL_WEIGHTS_MAX; z++) {
 159                 lcc->lc_directive[z] = info->directive[z];
 160                 lcc->lc_subst_count[z] = info->subst_count[z];
 161                 lcc->lc_pri_count[z] = info->pri_count[z];
 162                 lcc->lc_undef_pri[z] = info->undef_pri[z];
 163         }
 164
 165         lcc->lc_char_table = (void *)TMP;
 166         TMP += sizeof (collate_char_t) * (UCHAR_MAX + 1);
 167
 168         for (z = 0; z < lcc->lc_directive_count; z++) {
 169                 int count;
 170                 if ((count = lcc->lc_subst_count[z]) > 0) {
 171                         lcc->lc_subst_table[z] = (void *)TMP;
 172                         TMP += count * sizeof (collate_subst_t);
 173                 } else {
 174                         lcc->lc_subst_table[z] = NULL;
 175                 }
 176         }
 177
 178         if (chains > 0) {
 179                 lcc->lc_chain_table = (void *)TMP;
 180                 TMP += chains * sizeof (collate_chain_t);
 181         } else
 182                 lcc->lc_chain_table = NULL;
 183         lcc->lc_chain_count = chains;
 184         if (lcc->lc_large_count > 0)
 185                 lcc->lc_large_table = (void *)TMP;
 186         else
 187                 lcc->lc_large_table = NULL;
 188
 189         return (ldata);
 190 }
 191
 192 static const int32_t *
 193 substsearch(const struct lc_collate *lcc, const wchar_t key, int pass)
 194 {
 195         const collate_subst_t *p;
 196         int n = lcc->lc_subst_count[pass];
 197
 198         if (n == 0)
 199                 return (NULL);
 200
 201         if (pass >= lcc->lc_directive_count)
 202                 return (NULL);
 203
 204         if (!(key & COLLATE_SUBST_PRIORITY))
 205                 return (NULL);
 206
 207         p = lcc->lc_subst_table[pass] + (key & ~COLLATE_SUBST_PRIORITY);
 208         assert(p->key == key);
 209         return (p->pri);
 210 }
 211
 212 /*
 213  * Note: for performance reasons, we have expanded bsearch here.  This avoids
 214  * function call overhead with each comparison.
 215  */
 216
 217 static collate_chain_t *
 218 chainsearch(const struct lc_collate *lcc, const wchar_t *key, int *len)
 219 {
 220         int low = 0;
 221         int high = lcc->lc_chain_count - 1;
 222         int next, compar, l;
 223         collate_chain_t *p;
 224         collate_chain_t *tab = lcc->lc_chain_table;
 225
 226         if (high < 0)
 227                 return (NULL);
 228
 229         while (low <= high) {
 230                 next = (low + high) / 2;
 231                 p = tab + next;
 232                 compar = *key - *p->str;
 233                 if (compar == 0) {
 234                         l = wcsnlen(p->str, COLLATE_STR_LEN);
 235                         compar = wcsncmp(key, p->str, l);
 236                         if (compar == 0) {
 237                                 *len = l;
 238                                 return (p);
 239                         }
 240                 }
 241                 if (compar > 0)
 242                         low = next + 1;
 243                 else
 244                         high = next - 1;
 245         }
 246         return (NULL);
 247 }
 248
 249 static collate_large_t *
 250 largesearch(const struct lc_collate *lcc, const wchar_t key)
 251 {
 252         int low = 0;
 253         int high = lcc->lc_info->large_count - 1;
 254         int next, compar;
 255         collate_large_t *p;
 256         collate_large_t *tab = lcc->lc_large_table;
 257
 258         if (high < 0)
 259                 return (NULL);
 260
 261         while (low <= high) {
 262                 next = (low + high) / 2;
 263                 p = tab + next;
 264                 compar = key - p->val;
 265                 if (compar == 0)
 266                         return (p);
 267                 if (compar > 0)
 268                         low = next + 1;
 269                 else
 270                         high = next - 1;
 271         }
 272         return (NULL);
 273 }
 274
 275 void
 276 _collate_lookup(const struct lc_collate *lcc, const wchar_t *t,
 277     int *len, int *pri, int which, const int **state)
 278 {
 279         collate_chain_t *p2;
 280         collate_large_t *match;
 281         int p, l;
 282         const int *sptr;
 283
 284         /*
 285          * If this is the "last" pass for the UNDEFINED, then
 286          * we just return the priority itself.
 287          */
 288         if (which >= lcc->lc_directive_count) {
 289                 *pri = *t;
 290                 *len = 1;
 291                 *state = NULL;
 292                 return;
 293         }
 294
 295         /*
 296          * If we have remaining substitution data from a previous
 297          * call, consume it first.
 298          */
 299         if ((sptr = *state) != NULL) {
 300                 *pri = *sptr;
 301                 sptr++;
 302                 if ((sptr == *state) || (sptr == NULL))
 303                         *state = NULL;
 304                 else
 305                         *state = sptr;
 306                 *len = 0;
 307                 return;
 308         }
 309
 310         /* No active substitutions */
 311         *len = 1;
 312
 313         /*
 314          * Check for composites such as dipthongs that collate as a
 315          * single element (aka chains or collating-elements).
 316          */
 317         if (((p2 = chainsearch(lcc, t, &l)) != NULL) &&
 318             ((p = p2->pri[which]) >= 0)) {
 319
 320                 *len = l;
 321                 *pri = p;
 322
 323         } else if (*t <= UCHAR_MAX) {
 324
 325                 /*
 326                  * Character is a small (8-bit) character.
 327                  * We just look these up directly for speed.
 328                  */
 329                 *pri = lcc->lc_char_table[*t].pri[which];
 330
 331         } else if ((lcc->lc_info->large_count > 0) &&
 332             ((match = largesearch(lcc, *t)) != NULL)) {
 333
 334                 /*
 335                  * Character was found in the extended table.
 336                  */
 337                 *pri = match->pri.pri[which];
 338
 339         } else {
 340                 /*
 341                  * Character lacks a specific definition.
 342                  */
 343                 if (lcc->lc_directive[which] & DIRECTIVE_UNDEFINED) {
 344                         /* Mask off sign bit to prevent ordering confusion. */
 345                         *pri = (*t & COLLATE_MAX_PRIORITY);
 346                 } else {
 347                         *pri = lcc->lc_undef_pri[which];
 348                 }
 349                 /* No substitutions for undefined characters! */
 350                 return;
 351         }
 352
 353         /*
 354          * Try substituting (expanding) the character.  We are
 355          * currently doing this *after* the chain compression.  I
 356          * think it should not matter, but this way might be slightly
 357          * faster.
 358          *
 359          * We do this after the priority search, as this will help us
 360          * to identify a single key value.  In order for this to work,
 361          * its important that the priority assigned to a given element
 362          * to be substituted be unique for that level.  The localedef
 363          * code ensures this for us.
 364          */
 365         if ((sptr = substsearch(lcc, *pri, which)) != NULL) {
 366                 if ((*pri = *sptr) > 0) {
 367                         sptr++;
 368                         *state = *sptr ? sptr : NULL;
 369                 }
 370         }
 371
 372 }
 373
 374 /*
 375  * This is the meaty part of wcsxfrm & strxfrm.  Note that it does
 376  * NOT NULL terminate.  That is left to the caller.
 377  */
 378 size_t
 379 _collate_wxfrm(const struct lc_collate *lcc, const wchar_t *src, wchar_t *xf,
 380     size_t room)
 381 {
 382         int             pri;
 383         int             len;
 384         const wchar_t   *t;
 385         wchar_t         *tr = NULL;
 386         int             direc;
 387         int             pass;
 388         const int32_t   *state;
 389         size_t          want = 0;
 390         size_t          need = 0;
 391         int             ndir = lcc->lc_directive_count;
 392
 393         assert(src);
 394
 395         for (pass = 0; pass <= ndir; pass++) {
 396
 397                 state = NULL;
 398
 399                 if (pass != 0) {
 400                         /* insert level separator from the previous pass */
 401                         if (room) {
 402                                 *xf++ = 1;
 403                                 room--;
 404                         }
 405                         want++;
 406                 }
 407
 408                 /* special pass for undefined */
 409                 if (pass == ndir) {
 410                         direc = DIRECTIVE_FORWARD | DIRECTIVE_UNDEFINED;
 411                 } else {
 412                         direc = lcc->lc_directive[pass];
 413                 }
 414
 415                 t = src;
 416
 417                 if (direc & DIRECTIVE_BACKWARD) {
 418                         wchar_t *bp, *fp, c;
 419                         free(tr);
 420                         if ((tr = wcsdup(t)) == NULL) {
 421                                 errno = ENOMEM;
 422                                 goto fail;
 423                         }
 424                         bp = tr;
 425                         fp = tr + wcslen(tr) - 1;
 426                         while (bp < fp) {
 427                                 c = *bp;
 428                                 *bp++ = *fp;
 429                                 *fp-- = c;
 430                         }
 431                         t = (const wchar_t *)tr;
 432                 }
 433
 434                 if (direc & DIRECTIVE_POSITION) {
 435                         while (*t || state) {
 436                                 _collate_lookup(lcc, t, &len, &pri, pass,
 437                                     &state);
 438                                 t += len;
 439                                 if (pri <= 0) {
 440                                         if (pri < 0) {
 441                                                 errno = EINVAL;
 442                                                 goto fail;
 443                                         }
 444                                         state = NULL;
 445                                         pri = COLLATE_MAX_PRIORITY;
 446                                 }
 447                                 if (room) {
 448                                         *xf++ = pri;
 449                                         room--;
 450                                 }
 451                                 want++;
 452                                 need = want;
 453                         }
 454                 } else {
 455                         while (*t || state) {
 456                                 _collate_lookup(lcc, t, &len, &pri, pass,
 457                                     &state);
 458                                 t += len;
 459                                 if (pri <= 0) {
 460                                         if (pri < 0) {
 461                                                 errno = EINVAL;
 462                                                 goto fail;
 463                                         }
 464                                         state = NULL;
 465                                         continue;
 466                                 }
 467                                 if (room) {
 468                                         *xf++ = pri;
 469                                         room--;
 470                                 }
 471                                 want++;
 472                                 need = want;
 473                         }
 474                 }
 475         }
 476
 477 end:
 478         free(tr);
 479         return (need);
 480
 481 fail:
 482         free(tr);
 483         return ((size_t)(-1));
 484 }
 485
 486 /*
 487  * In the non-POSIX case, we transform each character into a string of
 488  * characters representing the character's priority.  Since char is usually
 489  * signed, we are limited by 7 bits per byte.  To avoid zero, we need to add
 490  * XFRM_OFFSET, so we can't use a full 7 bits.  For simplicity, we choose 6
 491  * bits per byte.
 492  *
 493  * It turns out that we sometimes have real priorities that are
 494  * 31-bits wide.  (But: be careful using priorities where the high
 495  * order bit is set -- i.e. the priority is negative.  The sort order
 496  * may be surprising!)
 497  *
 498  * TODO: This would be a good area to optimize somewhat.  It turns out
 499  * that real prioririties *except for the last UNDEFINED pass* are generally
 500  * very small.  We need the localedef code to precalculate the max
 501  * priority for us, and ideally also give us a mask, and then we could
 502  * severely limit what we expand to.
 503  */
 504 #define XFRM_BYTES      6
 505 #define XFRM_OFFSET     ('0')   /* make all printable characters */
 506 #define XFRM_SHIFT      6
 507 #define XFRM_MASK       ((1 << XFRM_SHIFT) - 1)
 508 #define XFRM_SEP        ('.')   /* chosen to be less than XFRM_OFFSET */
 509
 510 static int
 511 xfrm(locale_t loc, unsigned char *p, int pri, int pass)
 512 {
 513         /* we use unsigned to ensure zero fill on right shift */
 514         uint32_t val = (uint32_t)loc->collate->lc_pri_count[pass];
 515         int nc = 0;
 516
 517         while (val) {
 518                 *p = (pri & XFRM_MASK) + XFRM_OFFSET;
 519                 pri >>= XFRM_SHIFT;
 520                 val >>= XFRM_SHIFT;
 521                 p++;
 522                 nc++;
 523         }
 524         return (nc);
 525 }
 526
 527 size_t
 528 _collate_sxfrm(const wchar_t *src, char *xf, size_t room, locale_t loc)
 529 {
 530         int             pri;
 531         int             len;
 532         const wchar_t   *t;
 533         wchar_t         *tr = NULL;
 534         int             direc;
 535         int             pass;
 536         const int32_t   *state;
 537         size_t          want = 0;
 538         size_t          need = 0;
 539         int             b;
 540         uint8_t         buf[XFRM_BYTES];
 541         const struct lc_collate *lcc = loc->collate;
 542         int             ndir = lcc->lc_directive_count;
 543
 544         assert(src);
 545
 546         for (pass = 0; pass <= ndir; pass++) {
 547
 548                 state = NULL;
 549
 550                 if (pass != 0) {
 551                         /* insert level separator from the previous pass */
 552                         if (room) {
 553                                 *xf++ = XFRM_SEP;
 554                                 room--;
 555                         }
 556                         want++;
 557                 }
 558
 559                 /* special pass for undefined */
 560                 if (pass == ndir) {
 561                         direc = DIRECTIVE_FORWARD | DIRECTIVE_UNDEFINED;
 562                 } else {
 563                         direc = lcc->lc_directive[pass];
 564                 }
 565
 566                 t = src;
 567
 568                 if (direc & DIRECTIVE_BACKWARD) {
 569                         wchar_t *bp, *fp, c;
 570                         free(tr);
 571                         if ((tr = wcsdup(t)) == NULL) {
 572                                 errno = ENOMEM;
 573                                 goto fail;
 574                         }
 575                         bp = tr;
 576                         fp = tr + wcslen(tr) - 1;
 577                         while (bp < fp) {
 578                                 c = *bp;
 579                                 *bp++ = *fp;
 580                                 *fp-- = c;
 581                         }
 582                         t = (const wchar_t *)tr;
 583                 }
 584
 585                 if (direc & DIRECTIVE_POSITION) {
 586                         while (*t || state) {
 587
 588                                 _collate_lookup(lcc, t, &len, &pri, pass,
 589                                     &state);
 590                                 t += len;
 591                                 if (pri <= 0) {
 592                                         if (pri < 0) {
 593                                                 errno = EINVAL;
 594                                                 goto fail;
 595                                         }
 596                                         state = NULL;
 597                                         pri = COLLATE_MAX_PRIORITY;
 598                                 }
 599
 600                                 b = xfrm(loc, buf, pri, pass);
 601                                 want += b;
 602                                 if (room) {
 603                                         while (b) {
 604                                                 b--;
 605                                                 if (room) {
 606                                                         *xf++ = buf[b];
 607                                                         room--;
 608                                                 }
 609                                         }
 610                                 }
 611                                 need = want;
 612                         }
 613                 } else {
 614                         while (*t || state) {
 615                                 _collate_lookup(lcc, t, &len, &pri, pass,
 616                                     &state);
 617                                 t += len;
 618                                 if (pri <= 0) {
 619                                         if (pri < 0) {
 620                                                 errno = EINVAL;
 621                                                 goto fail;
 622                                         }
 623                                         state = NULL;
 624                                         continue;
 625                                 }
 626
 627                                 b = xfrm(loc, buf, pri, pass);
 628                                 want += b;
 629                                 if (room) {
 630
 631                                         while (b) {
 632                                                 b--;
 633                                                 if (room) {
 634                                                         *xf++ = buf[b];
 635                                                         room--;
 636                                                 }
 637                                         }
 638                                 }
 639                                 need = want;
 640                         }
 641                 }
 642         }
 643
 644 end:
 645         free(tr);
 646         return (need);
 647
 648 fail:
 649         free(tr);
 650         return ((size_t)(-1));
 651 }