usr.bin/sort/bwstring.c

   1 /*-
   2  * Copyright (C) 2009 Gabor Kovesdan <gabor@FreeBSD.org>
   3  * Copyright (C) 2012 Oleg Moskalenko <oleg.moskalenko@citrix.com>
   4  * All rights reserved.
   5  *
   6  * Redistribution and use in source and binary forms, with or without
   7  * modification, are permitted provided that the following conditions
   8  * are met:
   9  * 1. Redistributions of source code must retain the above copyright
  10  *    notice, this list of conditions and the following disclaimer.
  11  * 2. Redistributions in binary form must reproduce the above copyright
  12  *    notice, this list of conditions and the following disclaimer in the
  13  *    documentation and/or other materials provided with the distribution.
  14  *
  15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  16  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  17  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  18  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  19  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  20  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  21  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  22  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  23  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  24  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  25  * SUCH DAMAGE.
  26  */
  27
  28 #include <sys/cdefs.h>
  29 __FBSDID("$FreeBSD$");
  30
  31 #include <ctype.h>
  32 #include <errno.h>
  33 #include <err.h>
  34 #include <langinfo.h>
  35 #include <math.h>
  36 #include <stdlib.h>
  37 #include <string.h>
  38 #include <wchar.h>
  39 #include <wctype.h>
  40
  41 #include "bwstring.h"
  42 #include "sort.h"
  43
  44 bool byte_sort;
  45
  46 static wchar_t **wmonths;
  47 static unsigned char **cmonths;
  48
  49 /* initialise months */
  50
  51 void
  52 initialise_months(void)
  53 {
  54         const nl_item item[12] = { ABMON_1, ABMON_2, ABMON_3, ABMON_4,
  55             ABMON_5, ABMON_6, ABMON_7, ABMON_8, ABMON_9, ABMON_10,
  56             ABMON_11, ABMON_12 };
  57         unsigned char *tmp;
  58         size_t len;
  59
  60         if (MB_CUR_MAX == 1) {
  61                 if (cmonths == NULL) {
  62                         unsigned char *m;
  63
  64                         cmonths = sort_malloc(sizeof(unsigned char*) * 12);
  65                         for (int i = 0; i < 12; i++) {
  66                                 cmonths[i] = NULL;
  67                                 tmp = (unsigned char *) nl_langinfo(item[i]);
  68                                 if (tmp == NULL)
  69                                         continue;
  70                                 if (debug_sort)
  71                                         printf("month[%d]=%s\n", i, tmp);
  72                                 len = strlen(tmp);
  73                                 if (len < 1)
  74                                         continue;
  75                                 while (isblank(*tmp))
  76                                         ++tmp;
  77                                 m = sort_malloc(len + 1);
  78                                 memcpy(m, tmp, len + 1);
  79                                 m[len] = '\0';
  80                                 for (unsigned int j = 0; j < len; j++)
  81                                         m[j] = toupper(m[j]);
  82                                 cmonths[i] = m;
  83                         }
  84                 }
  85
  86         } else {
  87                 if (wmonths == NULL) {
  88                         wchar_t *m;
  89
  90                         wmonths = sort_malloc(sizeof(wchar_t *) * 12);
  91                         for (int i = 0; i < 12; i++) {
  92                                 wmonths[i] = NULL;
  93                                 tmp = (unsigned char *) nl_langinfo(item[i]);
  94                                 if (tmp == NULL)
  95                                         continue;
  96                                 if (debug_sort)
  97                                         printf("month[%d]=%s\n", i, tmp);
  98                                 len = strlen(tmp);
  99                                 if (len < 1)
 100                                         continue;
 101                                 while (isblank(*tmp))
 102                                         ++tmp;
 103                                 m = sort_malloc(SIZEOF_WCHAR_STRING(len + 1));
 104                                 if (mbstowcs(m, tmp, len) == ((size_t) -1))
 105                                         continue;
 106                                 m[len] = L'\0';
 107                                 for (unsigned int j = 0; j < len; j++)
 108                                         m[j] = towupper(m[j]);
 109                                 wmonths[i] = m;
 110                         }
 111                 }
 112         }
 113 }
 114
 115 /*
 116  * Compare two wide-character strings
 117  */
 118 static int
 119 wide_str_coll(const wchar_t *s1, const wchar_t *s2)
 120 {
 121         int ret = 0;
 122
 123         errno = 0;
 124         ret = wcscoll(s1, s2);
 125         if (errno == EILSEQ) {
 126                 errno = 0;
 127                 ret = wcscmp(s1, s2);
 128                 if (errno != 0) {
 129                         for (size_t i = 0; ; ++i) {
 130                                 wchar_t c1 = s1[i];
 131                                 wchar_t c2 = s2[i];
 132                                 if (c1 == L'\0')
 133                                         return ((c2 == L'\0') ? 0 : -1);
 134                                 if (c2 == L'\0')
 135                                         return (+1);
 136                                 if (c1 == c2)
 137                                         continue;
 138                                 return ((int)(c1 - c2));
 139                         }
 140                 }
 141         }
 142         return (ret);
 143 }
 144
 145 /* counterparts of wcs functions */
 146
 147 void
 148 bwsprintf(FILE *f, struct bwstring *bws, const char *prefix, const char *suffix)
 149 {
 150
 151         if (MB_CUR_MAX == 1)
 152                 fprintf(f, "%s%s%s", prefix, bws->data.cstr, suffix);
 153         else
 154                 fprintf(f, "%s%S%s", prefix, bws->data.wstr, suffix);
 155 }
 156
 157 const void* bwsrawdata(const struct bwstring *bws)
 158 {
 159
 160         return (&(bws->data));
 161 }
 162
 163 size_t bwsrawlen(const struct bwstring *bws)
 164 {
 165
 166         return ((MB_CUR_MAX == 1) ? bws->len : SIZEOF_WCHAR_STRING(bws->len));
 167 }
 168
 169 size_t
 170 bws_memsize(const struct bwstring *bws)
 171 {
 172
 173         return ((MB_CUR_MAX == 1) ? (bws->len + 2 + sizeof(struct bwstring)) :
 174             (SIZEOF_WCHAR_STRING(bws->len + 1) + sizeof(struct bwstring)));
 175 }
 176
 177 void
 178 bws_setlen(struct bwstring *bws, size_t newlen)
 179 {
 180
 181         if (bws && newlen != bws->len && newlen <= bws->len) {
 182                 bws->len = newlen;
 183                 if (MB_CUR_MAX == 1)
 184                         bws->data.cstr[newlen] = '\0';
 185                 else
 186                         bws->data.wstr[newlen] = L'\0';
 187         }
 188 }
 189
 190 /*
 191  * Allocate a new binary string of specified size
 192  */
 193 struct bwstring *
 194 bwsalloc(size_t sz)
 195 {
 196         struct bwstring *ret;
 197
 198         if (MB_CUR_MAX == 1)
 199                 ret = sort_malloc(sizeof(struct bwstring) + 1 + sz);
 200         else
 201                 ret = sort_malloc(sizeof(struct bwstring) +
 202                     SIZEOF_WCHAR_STRING(sz + 1));
 203         ret->len = sz;
 204
 205         if (MB_CUR_MAX == 1)
 206                 ret->data.cstr[ret->len] = '\0';
 207         else
 208                 ret->data.wstr[ret->len] = L'\0';
 209
 210         return (ret);
 211 }
 212
 213 /*
 214  * Create a copy of binary string.
 215  * New string size equals the length of the old string.
 216  */
 217 struct bwstring *
 218 bwsdup(const struct bwstring *s)
 219 {
 220
 221         if (s == NULL)
 222                 return (NULL);
 223         else {
 224                 struct bwstring *ret = bwsalloc(s->len);
 225
 226                 if (MB_CUR_MAX == 1)
 227                         memcpy(ret->data.cstr, s->data.cstr, (s->len));
 228                 else
 229                         memcpy(ret->data.wstr, s->data.wstr,
 230                             SIZEOF_WCHAR_STRING(s->len));
 231
 232                 return (ret);
 233         }
 234 }
 235
 236 /*
 237  * Create a new binary string from a raw binary buffer.
 238  */
 239 struct bwstring *
 240 bwssbdup(const wchar_t *str, size_t len)
 241 {
 242
 243         if (str == NULL)
 244                 return ((len == 0) ? bwsalloc(0) : NULL);
 245         else {
 246                 struct bwstring *ret;
 247
 248                 ret = bwsalloc(len);
 249
 250                 if (MB_CUR_MAX == 1)
 251                         for (size_t i = 0; i < len; ++i)
 252                                 ret->data.cstr[i] = (unsigned char) str[i];
 253                 else
 254                         memcpy(ret->data.wstr, str, SIZEOF_WCHAR_STRING(len));
 255
 256                 return (ret);
 257         }
 258 }
 259
 260 /*
 261  * Create a new binary string from a raw binary buffer.
 262  */
 263 struct bwstring *
 264 bwscsbdup(const unsigned char *str, size_t len)
 265 {
 266         struct bwstring *ret;
 267
 268         ret = bwsalloc(len);
 269
 270         if (str) {
 271                 if (MB_CUR_MAX == 1)
 272                         memcpy(ret->data.cstr, str, len);
 273                 else {
 274                         mbstate_t mbs;
 275                         const char *s;
 276                         size_t charlen, chars, cptr;
 277
 278                         charlen = chars = 0;
 279                         cptr = 0;
 280                         s = (const char *) str;
 281
 282                         memset(&mbs, 0, sizeof(mbs));
 283
 284                         while (cptr < len) {
 285                                 size_t n = MB_CUR_MAX;
 286
 287                                 if (n > len - cptr)
 288                                         n = len - cptr;
 289                                 charlen = mbrlen(s + cptr, n, &mbs);
 290                                 switch (charlen) {
 291                                 case 0:
 292                                         /* FALLTHROUGH */
 293                                 case (size_t) -1:
 294                                         /* FALLTHROUGH */
 295                                 case (size_t) -2:
 296                                         ret->data.wstr[chars++] =
 297                                             (unsigned char) s[cptr];
 298                                         ++cptr;
 299                                         break;
 300                                 default:
 301                                         n = mbrtowc(ret->data.wstr + (chars++),
 302                                             s + cptr, charlen, &mbs);
 303                                         if ((n == (size_t)-1) || (n == (size_t)-2))
 304                                                 /* NOTREACHED */
 305                                                 err(2, "mbrtowc error");
 306                                         cptr += charlen;
 307                                 };
 308                         }
 309
 310                         ret->len = chars;
 311                         ret->data.wstr[ret->len] = L'\0';
 312                 }
 313         }
 314         return (ret);
 315 }
 316
 317 /*
 318  * De-allocate object memory
 319  */
 320 void
 321 bwsfree(const struct bwstring *s)
 322 {
 323
 324         if (s)
 325                 sort_free(s);
 326 }
 327
 328 /*
 329  * Copy content of src binary string to dst.
 330  * If the capacity of the dst string is not sufficient,
 331  * then the data is truncated.
 332  */
 333 size_t
 334 bwscpy(struct bwstring *dst, const struct bwstring *src)
 335 {
 336         size_t nums = src->len;
 337
 338         if (nums > dst->len)
 339                 nums = dst->len;
 340         dst->len = nums;
 341
 342         if (MB_CUR_MAX == 1) {
 343                 memcpy(dst->data.cstr, src->data.cstr, nums);
 344                 dst->data.cstr[dst->len] = '\0';
 345         } else {
 346                 memcpy(dst->data.wstr, src->data.wstr,
 347                     SIZEOF_WCHAR_STRING(nums + 1));
 348                 dst->data.wstr[dst->len] = L'\0';
 349         }
 350
 351         return (nums);
 352 }
 353
 354 /*
 355  * Copy content of src binary string to dst,
 356  * with specified number of symbols to be copied.
 357  * If the capacity of the dst string is not sufficient,
 358  * then the data is truncated.
 359  */
 360 struct bwstring *
 361 bwsncpy(struct bwstring *dst, const struct bwstring *src, size_t size)
 362 {
 363         size_t nums = src->len;
 364
 365         if (nums > dst->len)
 366                 nums = dst->len;
 367         if (nums > size)
 368                 nums = size;
 369         dst->len = nums;
 370
 371         if (MB_CUR_MAX == 1) {
 372                 memcpy(dst->data.cstr, src->data.cstr, nums);
 373                 dst->data.cstr[dst->len] = '\0';
 374         } else {
 375                 memcpy(dst->data.wstr, src->data.wstr,
 376                     SIZEOF_WCHAR_STRING(nums + 1));
 377                 dst->data.wstr[dst->len] = L'\0';
 378         }
 379
 380         return (dst);
 381 }
 382
 383 /*
 384  * Copy content of src binary string to dst,
 385  * with specified number of symbols to be copied.
 386  * An offset value can be specified, from the start of src string.
 387  * If the capacity of the dst string is not sufficient,
 388  * then the data is truncated.
 389  */
 390 struct bwstring *
 391 bwsnocpy(struct bwstring *dst, const struct bwstring *src, size_t offset,
 392     size_t size)
 393 {
 394
 395         if (offset >= src->len) {
 396                 dst->data.wstr[0] = 0;
 397                 dst->len = 0;
 398         } else {
 399                 size_t nums = src->len - offset;
 400
 401                 if (nums > dst->len)
 402                         nums = dst->len;
 403                 if (nums > size)
 404                         nums = size;
 405                 dst->len = nums;
 406                 if (MB_CUR_MAX == 1) {
 407                         memcpy(dst->data.cstr, src->data.cstr + offset,
 408                             (nums));
 409                         dst->data.cstr[dst->len] = '\0';
 410                 } else {
 411                         memcpy(dst->data.wstr, src->data.wstr + offset,
 412                             SIZEOF_WCHAR_STRING(nums));
 413                         dst->data.wstr[dst->len] = L'\0';
 414                 }
 415         }
 416         return (dst);
 417 }
 418
 419 /*
 420  * Write binary string to the file.
 421  * The output is ended either with '\n' (nl == true)
 422  * or '\0' (nl == false).
 423  */
 424 int
 425 bwsfwrite(struct bwstring *bws, FILE *f, bool zero_ended)
 426 {
 427
 428         if (MB_CUR_MAX == 1) {
 429                 size_t len = bws->len;
 430
 431                 if (!zero_ended) {
 432                         bws->data.cstr[len] = '\n';
 433
 434                         if (fwrite(bws->data.cstr, len + 1, 1, f) < 1)
 435                                 err(2, NULL);
 436
 437                         bws->data.cstr[len] = '\0';
 438                 } else if (fwrite(bws->data.cstr, len + 1, 1, f) < 1)
 439                         err(2, NULL);
 440
 441                 return (len + 1);
 442
 443         } else {
 444                 wchar_t eols;
 445                 int printed = 0;
 446
 447                 eols = zero_ended ? btowc('\0') : btowc('\n');
 448
 449                 while (printed < (int) BWSLEN(bws)) {
 450                         const wchar_t *s = bws->data.wstr + printed;
 451
 452                         if (*s == L'\0') {
 453                                 int nums;
 454
 455                                 nums = fwprintf(f, L"%lc", *s);
 456
 457                                 if (nums != 1)
 458                                         err(2, NULL);
 459                                 ++printed;
 460                         } else {
 461                                 int nums;
 462
 463                                 nums = fwprintf(f, L"%ls", s);
 464
 465                                 if (nums < 1)
 466                                         err(2, NULL);
 467                                 printed += nums;
 468                         }
 469                 }
 470                 fwprintf(f, L"%lc", eols);
 471                 return (printed + 1);
 472         }
 473 }
 474
 475 /*
 476  * Allocate and read a binary string from file.
 477  * The strings are nl-ended or zero-ended, depending on the sort setting.
 478  */
 479 struct bwstring *
 480 bwsfgetln(FILE *f, size_t *len, bool zero_ended, struct reader_buffer *rb)
 481 {
 482         wchar_t eols;
 483
 484         eols = zero_ended ? btowc('\0') : btowc('\n');
 485
 486         if (!zero_ended && (MB_CUR_MAX > 1)) {
 487                 wchar_t *ret;
 488
 489                 ret = fgetwln(f, len);
 490
 491                 if (ret == NULL) {
 492                         if (!feof(f))
 493                                 err(2, NULL);
 494                         return (NULL);
 495                 }
 496                 if (*len > 0) {
 497                         if (ret[*len - 1] == eols)
 498                                 --(*len);
 499                 }
 500                 return (bwssbdup(ret, *len));
 501
 502         } else {
 503                 wchar_t c = 0;
 504
 505                 *len = 0;
 506
 507                 if (feof(f))
 508                         return (NULL);
 509
 510                 if (2 >= rb->fgetwln_z_buffer_size) {
 511                         rb->fgetwln_z_buffer_size += 256;
 512                         rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer,
 513                             sizeof(wchar_t) * rb->fgetwln_z_buffer_size);
 514                 }
 515                 rb->fgetwln_z_buffer[*len] = 0;
 516
 517                 if (MB_CUR_MAX == 1)
 518                         while (!feof(f)) {
 519                                 c = fgetc(f);
 520
 521                                 if (c == EOF) {
 522                                         if (*len == 0)
 523                                                 return (NULL);
 524                                         goto line_read_done;
 525                                 }
 526                                 if (c == eols)
 527                                         goto line_read_done;
 528
 529                                 if (*len + 1 >= rb->fgetwln_z_buffer_size) {
 530                                         rb->fgetwln_z_buffer_size += 256;
 531                                         rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer,
 532                                             SIZEOF_WCHAR_STRING(rb->fgetwln_z_buffer_size));
 533                                 }
 534
 535                                 rb->fgetwln_z_buffer[*len] = c;
 536                                 rb->fgetwln_z_buffer[++(*len)] = 0;
 537                         }
 538                 else
 539                         while (!feof(f)) {
 540                                 c = fgetwc(f);
 541
 542                                 if (c == WEOF) {
 543                                         if (*len == 0)
 544                                                 return (NULL);
 545                                         goto line_read_done;
 546                                 }
 547                                 if (c == eols)
 548                                         goto line_read_done;
 549
 550                                 if (*len + 1 >= rb->fgetwln_z_buffer_size) {
 551                                         rb->fgetwln_z_buffer_size += 256;
 552                                         rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer,
 553                                             SIZEOF_WCHAR_STRING(rb->fgetwln_z_buffer_size));
 554                                 }
 555
 556                                 rb->fgetwln_z_buffer[*len] = c;
 557                                 rb->fgetwln_z_buffer[++(*len)] = 0;
 558                         }
 559
 560 line_read_done:
 561                 /* we do not count the last 0 */
 562                 return (bwssbdup(rb->fgetwln_z_buffer, *len));
 563         }
 564 }
 565
 566 int
 567 bwsncmp(const struct bwstring *bws1, const struct bwstring *bws2,
 568     size_t offset, size_t len)
 569 {
 570         size_t cmp_len, len1, len2;
 571         int res = 0;
 572
 573         cmp_len = 0;
 574         len1 = bws1->len;
 575         len2 = bws2->len;
 576
 577         if (len1 <= offset) {
 578                 return ((len2 <= offset) ? 0 : -1);
 579         } else {
 580                 if (len2 <= offset)
 581                         return (+1);
 582                 else {
 583                         len1 -= offset;
 584                         len2 -= offset;
 585
 586                         cmp_len = len1;
 587
 588                         if (len2 < cmp_len)
 589                                 cmp_len = len2;
 590
 591                         if (len < cmp_len)
 592                                 cmp_len = len;
 593
 594                         if (MB_CUR_MAX == 1) {
 595                                 const unsigned char *s1, *s2;
 596
 597                                 s1 = bws1->data.cstr + offset;
 598                                 s2 = bws2->data.cstr + offset;
 599
 600                                 res = memcmp(s1, s2, cmp_len);
 601
 602                         } else {
 603                                 const wchar_t *s1, *s2;
 604
 605                                 s1 = bws1->data.wstr + offset;
 606                                 s2 = bws2->data.wstr + offset;
 607
 608                                 res = memcmp(s1, s2, SIZEOF_WCHAR_STRING(cmp_len));
 609                         }
 610                 }
 611         }
 612
 613         if (res == 0) {
 614                 if (len1 < cmp_len && len1 < len2)
 615                         res = -1;
 616                 else if (len2 < cmp_len && len2 < len1)
 617                         res = +1;
 618         }
 619
 620         return (res);
 621 }
 622
 623 int
 624 bwscmp(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset)
 625 {
 626         size_t len1, len2, cmp_len;
 627         int res;
 628
 629         len1 = bws1->len;
 630         len2 = bws2->len;
 631
 632         len1 -= offset;
 633         len2 -= offset;
 634
 635         cmp_len = len1;
 636
 637         if (len2 < cmp_len)
 638                 cmp_len = len2;
 639
 640         res = bwsncmp(bws1, bws2, offset, cmp_len);
 641
 642         if (res == 0) {
 643                 if( len1 < len2)
 644                         res = -1;
 645                 else if (len2 < len1)
 646                         res = +1;
 647         }
 648
 649         return (res);
 650 }
 651
 652 int
 653 bws_iterator_cmp(bwstring_iterator iter1, bwstring_iterator iter2, size_t len)
 654 {
 655         wchar_t c1, c2;
 656         size_t i = 0;
 657
 658         for (i = 0; i < len; ++i) {
 659                 c1 = bws_get_iter_value(iter1);
 660                 c2 = bws_get_iter_value(iter2);
 661                 if (c1 != c2)
 662                         return (c1 - c2);
 663                 iter1 = bws_iterator_inc(iter1, 1);
 664                 iter2 = bws_iterator_inc(iter2, 1);
 665         }
 666
 667         return (0);
 668 }
 669
 670 int
 671 bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset)
 672 {
 673         size_t len1, len2;
 674
 675         len1 = bws1->len;
 676         len2 = bws2->len;
 677
 678         if (len1 <= offset)
 679                 return ((len2 <= offset) ? 0 : -1);
 680         else {
 681                 if (len2 <= offset)
 682                         return (+1);
 683                 else {
 684                         len1 -= offset;
 685                         len2 -= offset;
 686
 687                         if (MB_CUR_MAX == 1) {
 688                                 const unsigned char *s1, *s2;
 689
 690                                 s1 = bws1->data.cstr + offset;
 691                                 s2 = bws2->data.cstr + offset;
 692
 693                                 if (byte_sort) {
 694                                         int res = 0;
 695
 696                                         if (len1 > len2) {
 697                                                 res = memcmp(s1, s2, len2);
 698                                                 if (!res)
 699                                                         res = +1;
 700                                         } else if (len1 < len2) {
 701                                                 res = memcmp(s1, s2, len1);
 702                                                 if (!res)
 703                                                         res = -1;
 704                                         } else
 705                                                 res = memcmp(s1, s2, len1);
 706
 707                                         return (res);
 708
 709                                 } else {
 710                                         int res = 0;
 711                                         size_t i, maxlen;
 712
 713                                         i = 0;
 714                                         maxlen = len1;
 715
 716                                         if (maxlen > len2)
 717                                                 maxlen = len2;
 718
 719                                         while (i < maxlen) {
 720                                                 /* goto next non-zero part: */
 721                                                 while ((i < maxlen) &&
 722                                                     !s1[i] && !s2[i])
 723                                                         ++i;
 724
 725                                                 if (i >= maxlen)
 726                                                         break;
 727
 728                                                 if (s1[i] == 0) {
 729                                                         if (s2[i] == 0)
 730                                                                 /* NOTREACHED */
 731                                                                 err(2, "bwscoll error 01");
 732                                                         else
 733                                                                 return (-1);
 734                                                 } else if (s2[i] == 0)
 735                                                         return (+1);
 736
 737                                                 res = strcoll(s1 + i, s2 + i);
 738                                                 if (res)
 739                                                         return (res);
 740
 741                                                 while ((i < maxlen) &&
 742                                                     s1[i] && s2[i])
 743                                                         ++i;
 744
 745                                                 if (i >= maxlen)
 746                                                         break;
 747
 748                                                 if (s1[i] == 0) {
 749                                                         if (s2[i] == 0) {
 750                                                                 ++i;
 751                                                                 continue;
 752                                                         } else
 753                                                                 return (-1);
 754                                                 } else if (s2[i] == 0)
 755                                                         return (+1);
 756                                                 else
 757                                                         /* NOTREACHED */
 758                                                         err(2, "bwscoll error 02");
 759                                         }
 760
 761                                         if (len1 < len2)
 762                                                 return (-1);
 763                                         else if (len1 > len2)
 764                                                 return (+1);
 765
 766                                         return (0);
 767                                 }
 768                         } else {
 769                                 const wchar_t *s1, *s2;
 770                                 size_t i, maxlen;
 771                                 int res = 0;
 772
 773                                 s1 = bws1->data.wstr + offset;
 774                                 s2 = bws2->data.wstr + offset;
 775
 776                                 i = 0;
 777                                 maxlen = len1;
 778
 779                                 if (maxlen > len2)
 780                                         maxlen = len2;
 781
 782                                 while (i < maxlen) {
 783
 784                                         /* goto next non-zero part: */
 785                                         while ((i < maxlen) &&
 786                                             !s1[i] && !s2[i])
 787                                                 ++i;
 788
 789                                         if (i >= maxlen)
 790                                                 break;
 791
 792                                         if (s1[i] == 0) {
 793                                                 if (s2[i] == 0)
 794                                                         /* NOTREACHED */
 795                                                         err(2, "bwscoll error 1");
 796                                                 else
 797                                                         return (-1);
 798                                         } else if (s2[i] == 0)
 799                                                 return (+1);
 800
 801                                         res = wide_str_coll(s1 + i, s2 + i);
 802                                         if (res)
 803                                                 return (res);
 804
 805                                         while ((i < maxlen) && s1[i] && s2[i])
 806                                                 ++i;
 807
 808                                         if (i >= maxlen)
 809                                                 break;
 810
 811                                         if (s1[i] == 0) {
 812                                                 if (s2[i] == 0) {
 813                                                         ++i;
 814                                                         continue;
 815                                                 } else
 816                                                         return (-1);
 817                                         } else if (s2[i] == 0)
 818                                                 return (+1);
 819                                         else
 820                                                 /* NOTREACHED */
 821                                                 err(2, "bwscoll error 2");
 822                                 }
 823
 824                                 if (len1 < len2)
 825                                         return (-1);
 826                                 else if (len1 > len2)
 827                                         return (+1);
 828
 829                                 return (0);
 830                         }
 831                 }
 832         }
 833 }
 834
 835 /*
 836  * Correction of the system API
 837  */
 838 double
 839 bwstod(struct bwstring *s0, bool *empty)
 840 {
 841         double ret = 0;
 842
 843         if (MB_CUR_MAX == 1) {
 844                 unsigned char *end, *s;
 845                 char *ep;
 846
 847                 s = s0->data.cstr;
 848                 end = s + s0->len;
 849                 ep = NULL;
 850
 851                 while (isblank(*s) && s < end)
 852                         ++s;
 853
 854                 if (!isprint(*s)) {
 855                         *empty = true;
 856                         return (0);
 857                 }
 858
 859                 ret = strtod(s, &ep);
 860                 if ((unsigned char*) ep == s) {
 861                         *empty = true;
 862                         return (0);
 863                 }
 864         } else {
 865                 wchar_t *end, *ep, *s;
 866
 867                 s = s0->data.wstr;
 868                 end = s + s0->len;
 869                 ep = NULL;
 870
 871                 while (iswblank(*s) && s < end)
 872                         ++s;
 873
 874                 if (!iswprint(*s)) {
 875                         *empty = true;
 876                         return (0);
 877                 }
 878
 879                 ret = wcstod(s, &ep);
 880                 if (ep == s) {
 881                         *empty = true;
 882                         return (0);
 883                 }
 884         }
 885
 886         *empty = false;
 887         return (ret);
 888 }
 889
 890 /*
 891  * A helper function for monthcoll.  If a line matches
 892  * a month name, it returns (number of the month - 1),
 893  * while if there is no match, it just return -1.
 894  */
 895
 896 int
 897 bws_month_score(const struct bwstring *s0)
 898 {
 899
 900         if (MB_CUR_MAX == 1) {
 901                 const unsigned char *end, *s;
 902                 size_t len;
 903
 904                 s = s0->data.cstr;
 905                 end = s + s0->len;
 906
 907                 while (isblank(*s) && s < end)
 908                         ++s;
 909
 910                 len = strlen(s);
 911
 912                 for (int i = 11; i >= 0; --i) {
 913                         if (cmonths[i] &&
 914                             (s == (unsigned char*)strstr(s, cmonths[i])))
 915                                 return (i);
 916                 }
 917
 918         } else {
 919                 const wchar_t *end, *s;
 920                 size_t len;
 921
 922                 s = s0->data.wstr;
 923                 end = s + s0->len;
 924
 925                 while (iswblank(*s) && s < end)
 926                         ++s;
 927
 928                 len = wcslen(s);
 929
 930                 for (int i = 11; i >= 0; --i) {
 931                         if (wmonths[i] && (s == wcsstr(s, wmonths[i])))
 932                                 return (i);
 933                 }
 934         }
 935
 936         return (-1);
 937 }
 938
 939 /*
 940  * Rips out leading blanks (-b).
 941  */
 942 struct bwstring *
 943 ignore_leading_blanks(struct bwstring *str)
 944 {
 945
 946         if (MB_CUR_MAX == 1) {
 947                 unsigned char *dst, *end, *src;
 948
 949                 src = str->data.cstr;
 950                 dst = src;
 951                 end = src + str->len;
 952
 953                 while (src < end && isblank(*src))
 954                         ++src;
 955
 956                 if (src != dst) {
 957                         size_t newlen;
 958
 959                         newlen = BWSLEN(str) - (src - dst);
 960
 961                         while (src < end) {
 962                                 *dst = *src;
 963                                 ++dst;
 964                                 ++src;
 965                         }
 966                         bws_setlen(str, newlen);
 967                 }
 968         } else {
 969                 wchar_t *dst, *end, *src;
 970
 971                 src = str->data.wstr;
 972                 dst = src;
 973                 end = src + str->len;
 974
 975                 while (src < end && iswblank(*src))
 976                         ++src;
 977
 978                 if (src != dst) {
 979
 980                         size_t newlen = BWSLEN(str) - (src - dst);
 981
 982                         while (src < end) {
 983                                 *dst = *src;
 984                                 ++dst;
 985                                 ++src;
 986                         }
 987                         bws_setlen(str, newlen);
 988
 989                 }
 990         }
 991         return (str);
 992 }
 993
 994 /*
 995  * Rips out nonprinting characters (-i).
 996  */
 997 struct bwstring *
 998 ignore_nonprinting(struct bwstring *str)
 999 {
1000         size_t newlen = str->len;
1001
1002         if (MB_CUR_MAX == 1) {
1003                 unsigned char *dst, *end, *src;
1004                 unsigned char c;
1005
1006                 src = str->data.cstr;
1007                 dst = src;
1008                 end = src + str->len;
1009
1010                 while (src < end) {
1011                         c = *src;
1012                         if (isprint(c)) {
1013                                 *dst = c;
1014                                 ++dst;
1015                                 ++src;
1016                         } else {
1017                                 ++src;
1018                                 --newlen;
1019                         }
1020                 }
1021         } else {
1022                 wchar_t *dst, *end, *src;
1023                 wchar_t c;
1024
1025                 src = str->data.wstr;
1026                 dst = src;
1027                 end = src + str->len;
1028
1029                 while (src < end) {
1030                         c = *src;
1031                         if (iswprint(c)) {
1032                                 *dst = c;
1033                                 ++dst;
1034                                 ++src;
1035                         } else {
1036                                 ++src;
1037                                 --newlen;
1038                         }
1039                 }
1040         }
1041         bws_setlen(str, newlen);
1042
1043         return (str);
1044 }
1045
1046 /*
1047  * Rips out any characters that are not alphanumeric characters
1048  * nor blanks (-d).
1049  */
1050 struct bwstring *
1051 dictionary_order(struct bwstring *str)
1052 {
1053         size_t newlen = str->len;
1054
1055         if (MB_CUR_MAX == 1) {
1056                 unsigned char *dst, *end, *src;
1057                 unsigned char c;
1058
1059                 src = str->data.cstr;
1060                 dst = src;
1061                 end = src + str->len;
1062
1063                 while (src < end) {
1064                         c = *src;
1065                         if (isalnum(c) || isblank(c)) {
1066                                 *dst = c;
1067                                 ++dst;
1068                                 ++src;
1069                         } else {
1070                                 ++src;
1071                                 --newlen;
1072                         }
1073                 }
1074         } else {
1075                 wchar_t *dst, *end, *src;
1076                 wchar_t c;
1077
1078                 src = str->data.wstr;
1079                 dst = src;
1080                 end = src + str->len;
1081
1082                 while (src < end) {
1083                         c = *src;
1084                         if (iswalnum(c) || iswblank(c)) {
1085                                 *dst = c;
1086                                 ++dst;
1087                                 ++src;
1088                         } else {
1089                                 ++src;
1090                                 --newlen;
1091                         }
1092                 }
1093         }
1094         bws_setlen(str, newlen);
1095
1096         return (str);
1097 }
1098
1099 /*
1100  * Converts string to lower case(-f).
1101  */
1102 struct bwstring *
1103 ignore_case(struct bwstring *str)
1104 {
1105
1106         if (MB_CUR_MAX == 1) {
1107                 unsigned char *end, *s;
1108
1109                 s = str->data.cstr;
1110                 end = s + str->len;
1111
1112                 while (s < end) {
1113                         *s = toupper(*s);
1114                         ++s;
1115                 }
1116         } else {
1117                 wchar_t *end, *s;
1118
1119                 s = str->data.wstr;
1120                 end = s + str->len;
1121
1122                 while (s < end) {
1123                         *s = towupper(*s);
1124                         ++s;
1125                 }
1126         }
1127         return (str);
1128 }
1129
1130 void
1131 bws_disorder_warnx(struct bwstring *s, const char *fn, size_t pos)
1132 {
1133
1134         if (MB_CUR_MAX == 1)
1135                 warnx("%s:%zu: disorder: %s", fn, pos + 1, s->data.cstr);
1136         else
1137                 warnx("%s:%zu: disorder: %ls", fn, pos + 1, s->data.wstr);
1138 }