usr.bin/sort/bwstring.c

   1 /*-
   2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
   3  *
   4  * Copyright (C) 2009 Gabor Kovesdan <gabor@FreeBSD.org>
   5  * Copyright (C) 2012 Oleg Moskalenko <mom040267@gmail.com>
   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 AND CONTRIBUTORS ``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 OR CONTRIBUTORS 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 <sys/cdefs.h>
  31 __FBSDID("$FreeBSD$");
  32
  33 #include <ctype.h>
  34 #include <errno.h>
  35 #include <err.h>
  36 #include <langinfo.h>
  37 #include <math.h>
  38 #include <stdlib.h>
  39 #include <string.h>
  40 #include <wchar.h>
  41 #include <wctype.h>
  42
  43 #include "bwstring.h"
  44 #include "sort.h"
  45
  46 bool byte_sort;
  47
  48 static wchar_t **wmonths;
  49 static unsigned char **cmonths;
  50
  51 /* initialise months */
  52
  53 void
  54 initialise_months(void)
  55 {
  56         const nl_item item[12] = { ABMON_1, ABMON_2, ABMON_3, ABMON_4,
  57             ABMON_5, ABMON_6, ABMON_7, ABMON_8, ABMON_9, ABMON_10,
  58             ABMON_11, ABMON_12 };
  59         unsigned char *tmp;
  60         size_t len;
  61
  62         if (MB_CUR_MAX == 1) {
  63                 if (cmonths == NULL) {
  64                         unsigned char *m;
  65
  66                         cmonths = sort_malloc(sizeof(unsigned char*) * 12);
  67                         for (int i = 0; i < 12; i++) {
  68                                 cmonths[i] = NULL;
  69                                 tmp = (unsigned char *) nl_langinfo(item[i]);
  70                                 if (debug_sort)
  71                                         printf("month[%d]=%s\n", i, tmp);
  72                                 if (*tmp == '\0')
  73                                         continue;
  74                                 m = sort_strdup(tmp);
  75                                 len = strlen(tmp);
  76                                 for (unsigned int j = 0; j < len; j++)
  77                                         m[j] = toupper(m[j]);
  78                                 cmonths[i] = m;
  79                         }
  80                 }
  81
  82         } else {
  83                 if (wmonths == NULL) {
  84                         wchar_t *m;
  85
  86                         wmonths = sort_malloc(sizeof(wchar_t *) * 12);
  87                         for (int i = 0; i < 12; i++) {
  88                                 wmonths[i] = NULL;
  89                                 tmp = (unsigned char *) nl_langinfo(item[i]);
  90                                 if (debug_sort)
  91                                         printf("month[%d]=%s\n", i, tmp);
  92                                 if (*tmp == '\0')
  93                                         continue;
  94                                 len = strlen(tmp);
  95                                 m = sort_malloc(SIZEOF_WCHAR_STRING(len + 1));
  96                                 if (mbstowcs(m, (char*)tmp, len) ==
  97                                     ((size_t) - 1)) {
  98                                         sort_free(m);
  99                                         continue;
 100                                 }
 101                                 m[len] = L'\0';
 102                                 for (unsigned int j = 0; j < len; j++)
 103                                         m[j] = towupper(m[j]);
 104                                 wmonths[i] = m;
 105                         }
 106                 }
 107         }
 108 }
 109
 110 /*
 111  * Compare two wide-character strings
 112  */
 113 static int
 114 wide_str_coll(const wchar_t *s1, const wchar_t *s2)
 115 {
 116         int ret = 0;
 117
 118         errno = 0;
 119         ret = wcscoll(s1, s2);
 120         if (errno == EILSEQ) {
 121                 errno = 0;
 122                 ret = wcscmp(s1, s2);
 123                 if (errno != 0) {
 124                         for (size_t i = 0; ; ++i) {
 125                                 wchar_t c1 = s1[i];
 126                                 wchar_t c2 = s2[i];
 127                                 if (c1 == L'\0')
 128                                         return ((c2 == L'\0') ? 0 : -1);
 129                                 if (c2 == L'\0')
 130                                         return (+1);
 131                                 if (c1 == c2)
 132                                         continue;
 133                                 return ((int)(c1 - c2));
 134                         }
 135                 }
 136         }
 137         return (ret);
 138 }
 139
 140 /* counterparts of wcs functions */
 141
 142 void
 143 bwsprintf(FILE *f, struct bwstring *bws, const char *prefix, const char *suffix)
 144 {
 145
 146         if (MB_CUR_MAX == 1)
 147                 fprintf(f, "%s%s%s", prefix, bws->data.cstr, suffix);
 148         else
 149                 fprintf(f, "%s%S%s", prefix, bws->data.wstr, suffix);
 150 }
 151
 152 const void* bwsrawdata(const struct bwstring *bws)
 153 {
 154
 155         return (&(bws->data));
 156 }
 157
 158 size_t bwsrawlen(const struct bwstring *bws)
 159 {
 160
 161         return ((MB_CUR_MAX == 1) ? bws->len : SIZEOF_WCHAR_STRING(bws->len));
 162 }
 163
 164 size_t
 165 bws_memsize(const struct bwstring *bws)
 166 {
 167
 168         return ((MB_CUR_MAX == 1) ? (bws->len + 2 + sizeof(struct bwstring)) :
 169             (SIZEOF_WCHAR_STRING(bws->len + 1) + sizeof(struct bwstring)));
 170 }
 171
 172 void
 173 bws_setlen(struct bwstring *bws, size_t newlen)
 174 {
 175
 176         if (bws && newlen != bws->len && newlen <= bws->len) {
 177                 bws->len = newlen;
 178                 if (MB_CUR_MAX == 1)
 179                         bws->data.cstr[newlen] = '\0';
 180                 else
 181                         bws->data.wstr[newlen] = L'\0';
 182         }
 183 }
 184
 185 /*
 186  * Allocate a new binary string of specified size
 187  */
 188 struct bwstring *
 189 bwsalloc(size_t sz)
 190 {
 191         struct bwstring *ret;
 192
 193         if (MB_CUR_MAX == 1)
 194                 ret = sort_malloc(sizeof(struct bwstring) + 1 + sz);
 195         else
 196                 ret = sort_malloc(sizeof(struct bwstring) +
 197                     SIZEOF_WCHAR_STRING(sz + 1));
 198         ret->len = sz;
 199
 200         if (MB_CUR_MAX == 1)
 201                 ret->data.cstr[ret->len] = '\0';
 202         else
 203                 ret->data.wstr[ret->len] = L'\0';
 204
 205         return (ret);
 206 }
 207
 208 /*
 209  * Create a copy of binary string.
 210  * New string size equals the length of the old string.
 211  */
 212 struct bwstring *
 213 bwsdup(const struct bwstring *s)
 214 {
 215
 216         if (s == NULL)
 217                 return (NULL);
 218         else {
 219                 struct bwstring *ret = bwsalloc(s->len);
 220
 221                 if (MB_CUR_MAX == 1)
 222                         memcpy(ret->data.cstr, s->data.cstr, (s->len));
 223                 else
 224                         memcpy(ret->data.wstr, s->data.wstr,
 225                             SIZEOF_WCHAR_STRING(s->len));
 226
 227                 return (ret);
 228         }
 229 }
 230
 231 /*
 232  * Create a new binary string from a wide character buffer.
 233  */
 234 struct bwstring *
 235 bwssbdup(const wchar_t *str, size_t len)
 236 {
 237
 238         if (str == NULL)
 239                 return ((len == 0) ? bwsalloc(0) : NULL);
 240         else {
 241                 struct bwstring *ret;
 242
 243                 ret = bwsalloc(len);
 244
 245                 if (MB_CUR_MAX == 1)
 246                         for (size_t i = 0; i < len; ++i)
 247                                 ret->data.cstr[i] = (unsigned char) str[i];
 248                 else
 249                         memcpy(ret->data.wstr, str, SIZEOF_WCHAR_STRING(len));
 250
 251                 return (ret);
 252         }
 253 }
 254
 255 /*
 256  * Create a new binary string from a raw binary buffer.
 257  */
 258 struct bwstring *
 259 bwscsbdup(const unsigned char *str, size_t len)
 260 {
 261         struct bwstring *ret;
 262
 263         ret = bwsalloc(len);
 264
 265         if (str) {
 266                 if (MB_CUR_MAX == 1)
 267                         memcpy(ret->data.cstr, str, len);
 268                 else {
 269                         mbstate_t mbs;
 270                         const char *s;
 271                         size_t charlen, chars, cptr;
 272
 273                         charlen = chars = 0;
 274                         cptr = 0;
 275                         s = (const char *) str;
 276
 277                         memset(&mbs, 0, sizeof(mbs));
 278
 279                         while (cptr < len) {
 280                                 size_t n = MB_CUR_MAX;
 281
 282                                 if (n > len - cptr)
 283                                         n = len - cptr;
 284                                 charlen = mbrlen(s + cptr, n, &mbs);
 285                                 switch (charlen) {
 286                                 case 0:
 287                                         /* FALLTHROUGH */
 288                                 case (size_t) -1:
 289                                         /* FALLTHROUGH */
 290                                 case (size_t) -2:
 291                                         ret->data.wstr[chars++] =
 292                                             (unsigned char) s[cptr];
 293                                         ++cptr;
 294                                         break;
 295                                 default:
 296                                         n = mbrtowc(ret->data.wstr + (chars++),
 297                                             s + cptr, charlen, &mbs);
 298                                         if ((n == (size_t)-1) || (n == (size_t)-2))
 299                                                 /* NOTREACHED */
 300                                                 err(2, "mbrtowc error");
 301                                         cptr += charlen;
 302                                 }
 303                         }
 304
 305                         ret->len = chars;
 306                         ret->data.wstr[ret->len] = L'\0';
 307                 }
 308         }
 309         return (ret);
 310 }
 311
 312 /*
 313  * De-allocate object memory
 314  */
 315 void
 316 bwsfree(const struct bwstring *s)
 317 {
 318
 319         if (s)
 320                 sort_free(s);
 321 }
 322
 323 /*
 324  * Copy content of src binary string to dst.
 325  * If the capacity of the dst string is not sufficient,
 326  * then the data is truncated.
 327  */
 328 size_t
 329 bwscpy(struct bwstring *dst, const struct bwstring *src)
 330 {
 331         size_t nums = src->len;
 332
 333         if (nums > dst->len)
 334                 nums = dst->len;
 335         dst->len = nums;
 336
 337         if (MB_CUR_MAX == 1) {
 338                 memcpy(dst->data.cstr, src->data.cstr, nums);
 339                 dst->data.cstr[dst->len] = '\0';
 340         } else {
 341                 memcpy(dst->data.wstr, src->data.wstr,
 342                     SIZEOF_WCHAR_STRING(nums + 1));
 343                 dst->data.wstr[dst->len] = L'\0';
 344         }
 345
 346         return (nums);
 347 }
 348
 349 /*
 350  * Copy content of src binary string to dst,
 351  * with specified number of symbols to be copied.
 352  * If the capacity of the dst string is not sufficient,
 353  * then the data is truncated.
 354  */
 355 struct bwstring *
 356 bwsncpy(struct bwstring *dst, const struct bwstring *src, size_t size)
 357 {
 358         size_t nums = src->len;
 359
 360         if (nums > dst->len)
 361                 nums = dst->len;
 362         if (nums > size)
 363                 nums = size;
 364         dst->len = nums;
 365
 366         if (MB_CUR_MAX == 1) {
 367                 memcpy(dst->data.cstr, src->data.cstr, nums);
 368                 dst->data.cstr[dst->len] = '\0';
 369         } else {
 370                 memcpy(dst->data.wstr, src->data.wstr,
 371                     SIZEOF_WCHAR_STRING(nums + 1));
 372                 dst->data.wstr[dst->len] = L'\0';
 373         }
 374
 375         return (dst);
 376 }
 377
 378 /*
 379  * Copy content of src binary string to dst,
 380  * with specified number of symbols to be copied.
 381  * An offset value can be specified, from the start of src string.
 382  * If the capacity of the dst string is not sufficient,
 383  * then the data is truncated.
 384  */
 385 struct bwstring *
 386 bwsnocpy(struct bwstring *dst, const struct bwstring *src, size_t offset,
 387     size_t size)
 388 {
 389
 390         if (offset >= src->len) {
 391                 dst->data.wstr[0] = 0;
 392                 dst->len = 0;
 393         } else {
 394                 size_t nums = src->len - offset;
 395
 396                 if (nums > dst->len)
 397                         nums = dst->len;
 398                 if (nums > size)
 399                         nums = size;
 400                 dst->len = nums;
 401                 if (MB_CUR_MAX == 1) {
 402                         memcpy(dst->data.cstr, src->data.cstr + offset,
 403                             (nums));
 404                         dst->data.cstr[dst->len] = '\0';
 405                 } else {
 406                         memcpy(dst->data.wstr, src->data.wstr + offset,
 407                             SIZEOF_WCHAR_STRING(nums));
 408                         dst->data.wstr[dst->len] = L'\0';
 409                 }
 410         }
 411         return (dst);
 412 }
 413
 414 /*
 415  * Write binary string to the file.
 416  * The output is ended either with '\n' (nl == true)
 417  * or '\0' (nl == false).
 418  */
 419 size_t
 420 bwsfwrite(struct bwstring *bws, FILE *f, bool zero_ended)
 421 {
 422
 423         if (MB_CUR_MAX == 1) {
 424                 size_t len = bws->len;
 425
 426                 if (!zero_ended) {
 427                         bws->data.cstr[len] = '\n';
 428
 429                         if (fwrite(bws->data.cstr, len + 1, 1, f) < 1)
 430                                 err(2, NULL);
 431
 432                         bws->data.cstr[len] = '\0';
 433                 } else if (fwrite(bws->data.cstr, len + 1, 1, f) < 1)
 434                         err(2, NULL);
 435
 436                 return (len + 1);
 437
 438         } else {
 439                 wchar_t eols;
 440                 size_t printed = 0;
 441
 442                 eols = zero_ended ? btowc('\0') : btowc('\n');
 443
 444                 while (printed < BWSLEN(bws)) {
 445                         const wchar_t *s = bws->data.wstr + printed;
 446
 447                         if (*s == L'\0') {
 448                                 int nums;
 449
 450                                 nums = fwprintf(f, L"%lc", *s);
 451
 452                                 if (nums != 1)
 453                                         err(2, NULL);
 454                                 ++printed;
 455                         } else {
 456                                 int nums;
 457
 458                                 nums = fwprintf(f, L"%ls", s);
 459
 460                                 if (nums < 1)
 461                                         err(2, NULL);
 462                                 printed += nums;
 463                         }
 464                 }
 465                 fwprintf(f, L"%lc", eols);
 466                 return (printed + 1);
 467         }
 468 }
 469
 470 /*
 471  * Allocate and read a binary string from file.
 472  * The strings are nl-ended or zero-ended, depending on the sort setting.
 473  */
 474 struct bwstring *
 475 bwsfgetln(FILE *f, size_t *len, bool zero_ended, struct reader_buffer *rb)
 476 {
 477         wint_t eols;
 478
 479         eols = zero_ended ? btowc('\0') : btowc('\n');
 480
 481         if (!zero_ended && (MB_CUR_MAX > 1)) {
 482                 wchar_t *ret;
 483
 484                 ret = fgetwln(f, len);
 485
 486                 if (ret == NULL) {
 487                         if (!feof(f))
 488                                 err(2, NULL);
 489                         return (NULL);
 490                 }
 491                 if (*len > 0) {
 492                         if (ret[*len - 1] == (wchar_t)eols)
 493                                 --(*len);
 494                 }
 495                 return (bwssbdup(ret, *len));
 496
 497         } else if (!zero_ended && (MB_CUR_MAX == 1)) {
 498                 char *ret;
 499
 500                 ret = fgetln(f, len);
 501
 502                 if (ret == NULL) {
 503                         if (!feof(f))
 504                                 err(2, NULL);
 505                         return (NULL);
 506                 }
 507                 if (*len > 0) {
 508                         if (ret[*len - 1] == '\n')
 509                                 --(*len);
 510                 }
 511                 return (bwscsbdup((unsigned char*)ret, *len));
 512
 513         } else {
 514                 *len = 0;
 515
 516                 if (feof(f))
 517                         return (NULL);
 518
 519                 if (2 >= rb->fgetwln_z_buffer_size) {
 520                         rb->fgetwln_z_buffer_size += 256;
 521                         rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer,
 522                             sizeof(wchar_t) * rb->fgetwln_z_buffer_size);
 523                 }
 524                 rb->fgetwln_z_buffer[*len] = 0;
 525
 526                 if (MB_CUR_MAX == 1)
 527                         while (!feof(f)) {
 528                                 int c;
 529
 530                                 c = fgetc(f);
 531
 532                                 if (c == EOF) {
 533                                         if (*len == 0)
 534                                                 return (NULL);
 535                                         goto line_read_done;
 536                                 }
 537                                 if (c == eols)
 538                                         goto line_read_done;
 539
 540                                 if (*len + 1 >= rb->fgetwln_z_buffer_size) {
 541                                         rb->fgetwln_z_buffer_size += 256;
 542                                         rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer,
 543                                             SIZEOF_WCHAR_STRING(rb->fgetwln_z_buffer_size));
 544                                 }
 545
 546                                 rb->fgetwln_z_buffer[*len] = c;
 547                                 rb->fgetwln_z_buffer[++(*len)] = 0;
 548                         }
 549                 else
 550                         while (!feof(f)) {
 551                                 wint_t c = 0;
 552
 553                                 c = fgetwc(f);
 554
 555                                 if (c == WEOF) {
 556                                         if (*len == 0)
 557                                                 return (NULL);
 558                                         goto line_read_done;
 559                                 }
 560                                 if (c == eols)
 561                                         goto line_read_done;
 562
 563                                 if (*len + 1 >= rb->fgetwln_z_buffer_size) {
 564                                         rb->fgetwln_z_buffer_size += 256;
 565                                         rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer,
 566                                             SIZEOF_WCHAR_STRING(rb->fgetwln_z_buffer_size));
 567                                 }
 568
 569                                 rb->fgetwln_z_buffer[*len] = c;
 570                                 rb->fgetwln_z_buffer[++(*len)] = 0;
 571                         }
 572
 573 line_read_done:
 574                 /* we do not count the last 0 */
 575                 return (bwssbdup(rb->fgetwln_z_buffer, *len));
 576         }
 577 }
 578
 579 int
 580 bwsncmp(const struct bwstring *bws1, const struct bwstring *bws2,
 581     size_t offset, size_t len)
 582 {
 583         size_t cmp_len, len1, len2;
 584         int res = 0;
 585
 586         cmp_len = 0;
 587         len1 = bws1->len;
 588         len2 = bws2->len;
 589
 590         if (len1 <= offset) {
 591                 return ((len2 <= offset) ? 0 : -1);
 592         } else {
 593                 if (len2 <= offset)
 594                         return (+1);
 595                 else {
 596                         len1 -= offset;
 597                         len2 -= offset;
 598
 599                         cmp_len = len1;
 600
 601                         if (len2 < cmp_len)
 602                                 cmp_len = len2;
 603
 604                         if (len < cmp_len)
 605                                 cmp_len = len;
 606
 607                         if (MB_CUR_MAX == 1) {
 608                                 const unsigned char *s1, *s2;
 609
 610                                 s1 = bws1->data.cstr + offset;
 611                                 s2 = bws2->data.cstr + offset;
 612
 613                                 res = memcmp(s1, s2, cmp_len);
 614
 615                         } else {
 616                                 const wchar_t *s1, *s2;
 617
 618                                 s1 = bws1->data.wstr + offset;
 619                                 s2 = bws2->data.wstr + offset;
 620
 621                                 res = memcmp(s1, s2, SIZEOF_WCHAR_STRING(cmp_len));
 622                         }
 623                 }
 624         }
 625
 626         if (res == 0) {
 627                 if (len1 < cmp_len && len1 < len2)
 628                         res = -1;
 629                 else if (len2 < cmp_len && len2 < len1)
 630                         res = +1;
 631         }
 632
 633         return (res);
 634 }
 635
 636 int
 637 bwscmp(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset)
 638 {
 639         size_t len1, len2, cmp_len;
 640         int res;
 641
 642         len1 = bws1->len;
 643         len2 = bws2->len;
 644
 645         len1 -= offset;
 646         len2 -= offset;
 647
 648         cmp_len = len1;
 649
 650         if (len2 < cmp_len)
 651                 cmp_len = len2;
 652
 653         res = bwsncmp(bws1, bws2, offset, cmp_len);
 654
 655         if (res == 0) {
 656                 if( len1 < len2)
 657                         res = -1;
 658                 else if (len2 < len1)
 659                         res = +1;
 660         }
 661
 662         return (res);
 663 }
 664
 665 int
 666 bws_iterator_cmp(bwstring_iterator iter1, bwstring_iterator iter2, size_t len)
 667 {
 668         wchar_t c1, c2;
 669         size_t i = 0;
 670
 671         for (i = 0; i < len; ++i) {
 672                 c1 = bws_get_iter_value(iter1);
 673                 c2 = bws_get_iter_value(iter2);
 674                 if (c1 != c2)
 675                         return (c1 - c2);
 676                 iter1 = bws_iterator_inc(iter1, 1);
 677                 iter2 = bws_iterator_inc(iter2, 1);
 678         }
 679
 680         return (0);
 681 }
 682
 683 int
 684 bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset)
 685 {
 686         size_t len1, len2;
 687
 688         len1 = bws1->len;
 689         len2 = bws2->len;
 690
 691         if (len1 <= offset)
 692                 return ((len2 <= offset) ? 0 : -1);
 693         else {
 694                 if (len2 <= offset)
 695                         return (+1);
 696                 else {
 697                         len1 -= offset;
 698                         len2 -= offset;
 699
 700                         if (MB_CUR_MAX == 1) {
 701                                 const unsigned char *s1, *s2;
 702
 703                                 s1 = bws1->data.cstr + offset;
 704                                 s2 = bws2->data.cstr + offset;
 705
 706                                 if (byte_sort) {
 707                                         int res = 0;
 708
 709                                         if (len1 > len2) {
 710                                                 res = memcmp(s1, s2, len2);
 711                                                 if (!res)
 712                                                         res = +1;
 713                                         } else if (len1 < len2) {
 714                                                 res = memcmp(s1, s2, len1);
 715                                                 if (!res)
 716                                                         res = -1;
 717                                         } else
 718                                                 res = memcmp(s1, s2, len1);
 719
 720                                         return (res);
 721
 722                                 } else {
 723                                         int res = 0;
 724                                         size_t i, maxlen;
 725
 726                                         i = 0;
 727                                         maxlen = len1;
 728
 729                                         if (maxlen > len2)
 730                                                 maxlen = len2;
 731
 732                                         while (i < maxlen) {
 733                                                 /* goto next non-zero part: */
 734                                                 while ((i < maxlen) &&
 735                                                     !s1[i] && !s2[i])
 736                                                         ++i;
 737
 738                                                 if (i >= maxlen)
 739                                                         break;
 740
 741                                                 if (s1[i] == 0) {
 742                                                         if (s2[i] == 0)
 743                                                                 /* NOTREACHED */
 744                                                                 err(2, "bwscoll error 01");
 745                                                         else
 746                                                                 return (-1);
 747                                                 } else if (s2[i] == 0)
 748                                                         return (+1);
 749
 750                                                 res = strcoll((const char*)(s1 + i), (const char*)(s2 + i));
 751                                                 if (res)
 752                                                         return (res);
 753
 754                                                 while ((i < maxlen) &&
 755                                                     s1[i] && s2[i])
 756                                                         ++i;
 757
 758                                                 if (i >= maxlen)
 759                                                         break;
 760
 761                                                 if (s1[i] == 0) {
 762                                                         if (s2[i] == 0) {
 763                                                                 ++i;
 764                                                                 continue;
 765                                                         } else
 766                                                                 return (-1);
 767                                                 } else if (s2[i] == 0)
 768                                                         return (+1);
 769                                                 else
 770                                                         /* NOTREACHED */
 771                                                         err(2, "bwscoll error 02");
 772                                         }
 773
 774                                         if (len1 < len2)
 775                                                 return (-1);
 776                                         else if (len1 > len2)
 777                                                 return (+1);
 778
 779                                         return (0);
 780                                 }
 781                         } else {
 782                                 const wchar_t *s1, *s2;
 783                                 size_t i, maxlen;
 784                                 int res = 0;
 785
 786                                 s1 = bws1->data.wstr + offset;
 787                                 s2 = bws2->data.wstr + offset;
 788
 789                                 i = 0;
 790                                 maxlen = len1;
 791
 792                                 if (maxlen > len2)
 793                                         maxlen = len2;
 794
 795                                 while (i < maxlen) {
 796
 797                                         /* goto next non-zero part: */
 798                                         while ((i < maxlen) &&
 799                                             !s1[i] && !s2[i])
 800                                                 ++i;
 801
 802                                         if (i >= maxlen)
 803                                                 break;
 804
 805                                         if (s1[i] == 0) {
 806                                                 if (s2[i] == 0)
 807                                                         /* NOTREACHED */
 808                                                         err(2, "bwscoll error 1");
 809                                                 else
 810                                                         return (-1);
 811                                         } else if (s2[i] == 0)
 812                                                 return (+1);
 813
 814                                         res = wide_str_coll(s1 + i, s2 + i);
 815                                         if (res)
 816                                                 return (res);
 817
 818                                         while ((i < maxlen) && s1[i] && s2[i])
 819                                                 ++i;
 820
 821                                         if (i >= maxlen)
 822                                                 break;
 823
 824                                         if (s1[i] == 0) {
 825                                                 if (s2[i] == 0) {
 826                                                         ++i;
 827                                                         continue;
 828                                                 } else
 829                                                         return (-1);
 830                                         } else if (s2[i] == 0)
 831                                                 return (+1);
 832                                         else
 833                                                 /* NOTREACHED */
 834                                                 err(2, "bwscoll error 2");
 835                                 }
 836
 837                                 if (len1 < len2)
 838                                         return (-1);
 839                                 else if (len1 > len2)
 840                                         return (+1);
 841
 842                                 return (0);
 843                         }
 844                 }
 845         }
 846 }
 847
 848 /*
 849  * Correction of the system API
 850  */
 851 double
 852 bwstod(struct bwstring *s0, bool *empty)
 853 {
 854         double ret = 0;
 855
 856         if (MB_CUR_MAX == 1) {
 857                 unsigned char *end, *s;
 858                 char *ep;
 859
 860                 s = s0->data.cstr;
 861                 end = s + s0->len;
 862                 ep = NULL;
 863
 864                 while (isblank(*s) && s < end)
 865                         ++s;
 866
 867                 if (!isprint(*s)) {
 868                         *empty = true;
 869                         return (0);
 870                 }
 871
 872                 ret = strtod((char*)s, &ep);
 873                 if ((unsigned char*) ep == s) {
 874                         *empty = true;
 875                         return (0);
 876                 }
 877         } else {
 878                 wchar_t *end, *ep, *s;
 879
 880                 s = s0->data.wstr;
 881                 end = s + s0->len;
 882                 ep = NULL;
 883
 884                 while (iswblank(*s) && s < end)
 885                         ++s;
 886
 887                 if (!iswprint(*s)) {
 888                         *empty = true;
 889                         return (0);
 890                 }
 891
 892                 ret = wcstod(s, &ep);
 893                 if (ep == s) {
 894                         *empty = true;
 895                         return (0);
 896                 }
 897         }
 898
 899         *empty = false;
 900         return (ret);
 901 }
 902
 903 /*
 904  * A helper function for monthcoll.  If a line matches
 905  * a month name, it returns (number of the month - 1),
 906  * while if there is no match, it just return -1.
 907  */
 908
 909 int
 910 bws_month_score(const struct bwstring *s0)
 911 {
 912
 913         if (MB_CUR_MAX == 1) {
 914                 const unsigned char *end, *s;
 915
 916                 s = s0->data.cstr;
 917                 end = s + s0->len;
 918
 919                 while (isblank(*s) && s < end)
 920                         ++s;
 921
 922                 for (int i = 11; i >= 0; --i) {
 923                         if (cmonths[i] &&
 924                             (s == (unsigned char*)strstr((const char*)s, (char*)(cmonths[i]))))
 925                                 return (i);
 926                 }
 927
 928         } else {
 929                 const wchar_t *end, *s;
 930
 931                 s = s0->data.wstr;
 932                 end = s + s0->len;
 933
 934                 while (iswblank(*s) && s < end)
 935                         ++s;
 936
 937                 for (int i = 11; i >= 0; --i) {
 938                         if (wmonths[i] && (s == wcsstr(s, wmonths[i])))
 939                                 return (i);
 940                 }
 941         }
 942
 943         return (-1);
 944 }
 945
 946 /*
 947  * Rips out leading blanks (-b).
 948  */
 949 struct bwstring *
 950 ignore_leading_blanks(struct bwstring *str)
 951 {
 952
 953         if (MB_CUR_MAX == 1) {
 954                 unsigned char *dst, *end, *src;
 955
 956                 src = str->data.cstr;
 957                 dst = src;
 958                 end = src + str->len;
 959
 960                 while (src < end && isblank(*src))
 961                         ++src;
 962
 963                 if (src != dst) {
 964                         size_t newlen;
 965
 966                         newlen = BWSLEN(str) - (src - dst);
 967
 968                         while (src < end) {
 969                                 *dst = *src;
 970                                 ++dst;
 971                                 ++src;
 972                         }
 973                         bws_setlen(str, newlen);
 974                 }
 975         } else {
 976                 wchar_t *dst, *end, *src;
 977
 978                 src = str->data.wstr;
 979                 dst = src;
 980                 end = src + str->len;
 981
 982                 while (src < end && iswblank(*src))
 983                         ++src;
 984
 985                 if (src != dst) {
 986
 987                         size_t newlen = BWSLEN(str) - (src - dst);
 988
 989                         while (src < end) {
 990                                 *dst = *src;
 991                                 ++dst;
 992                                 ++src;
 993                         }
 994                         bws_setlen(str, newlen);
 995
 996                 }
 997         }
 998         return (str);
 999 }
1000
1001 /*
1002  * Rips out nonprinting characters (-i).
1003  */
1004 struct bwstring *
1005 ignore_nonprinting(struct bwstring *str)
1006 {
1007         size_t newlen = str->len;
1008
1009         if (MB_CUR_MAX == 1) {
1010                 unsigned char *dst, *end, *src;
1011                 unsigned char c;
1012
1013                 src = str->data.cstr;
1014                 dst = src;
1015                 end = src + str->len;
1016
1017                 while (src < end) {
1018                         c = *src;
1019                         if (isprint(c)) {
1020                                 *dst = c;
1021                                 ++dst;
1022                                 ++src;
1023                         } else {
1024                                 ++src;
1025                                 --newlen;
1026                         }
1027                 }
1028         } else {
1029                 wchar_t *dst, *end, *src;
1030                 wchar_t c;
1031
1032                 src = str->data.wstr;
1033                 dst = src;
1034                 end = src + str->len;
1035
1036                 while (src < end) {
1037                         c = *src;
1038                         if (iswprint(c)) {
1039                                 *dst = c;
1040                                 ++dst;
1041                                 ++src;
1042                         } else {
1043                                 ++src;
1044                                 --newlen;
1045                         }
1046                 }
1047         }
1048         bws_setlen(str, newlen);
1049
1050         return (str);
1051 }
1052
1053 /*
1054  * Rips out any characters that are not alphanumeric characters
1055  * nor blanks (-d).
1056  */
1057 struct bwstring *
1058 dictionary_order(struct bwstring *str)
1059 {
1060         size_t newlen = str->len;
1061
1062         if (MB_CUR_MAX == 1) {
1063                 unsigned char *dst, *end, *src;
1064                 unsigned char c;
1065
1066                 src = str->data.cstr;
1067                 dst = src;
1068                 end = src + str->len;
1069
1070                 while (src < end) {
1071                         c = *src;
1072                         if (isalnum(c) || isblank(c)) {
1073                                 *dst = c;
1074                                 ++dst;
1075                                 ++src;
1076                         } else {
1077                                 ++src;
1078                                 --newlen;
1079                         }
1080                 }
1081         } else {
1082                 wchar_t *dst, *end, *src;
1083                 wchar_t c;
1084
1085                 src = str->data.wstr;
1086                 dst = src;
1087                 end = src + str->len;
1088
1089                 while (src < end) {
1090                         c = *src;
1091                         if (iswalnum(c) || iswblank(c)) {
1092                                 *dst = c;
1093                                 ++dst;
1094                                 ++src;
1095                         } else {
1096                                 ++src;
1097                                 --newlen;
1098                         }
1099                 }
1100         }
1101         bws_setlen(str, newlen);
1102
1103         return (str);
1104 }
1105
1106 /*
1107  * Converts string to lower case(-f).
1108  */
1109 struct bwstring *
1110 ignore_case(struct bwstring *str)
1111 {
1112
1113         if (MB_CUR_MAX == 1) {
1114                 unsigned char *end, *s;
1115
1116                 s = str->data.cstr;
1117                 end = s + str->len;
1118
1119                 while (s < end) {
1120                         *s = toupper(*s);
1121                         ++s;
1122                 }
1123         } else {
1124                 wchar_t *end, *s;
1125
1126                 s = str->data.wstr;
1127                 end = s + str->len;
1128
1129                 while (s < end) {
1130                         *s = towupper(*s);
1131                         ++s;
1132                 }
1133         }
1134         return (str);
1135 }
1136
1137 void
1138 bws_disorder_warnx(struct bwstring *s, const char *fn, size_t pos)
1139 {
1140
1141         if (MB_CUR_MAX == 1)
1142                 warnx("%s:%zu: disorder: %s", fn, pos + 1, s->data.cstr);
1143         else
1144                 warnx("%s:%zu: disorder: %ls", fn, pos + 1, s->data.wstr);
1145 }