1235267Sgabor/*- 2235267Sgabor * Copyright (C) 2009 Gabor Kovesdan <gabor@FreeBSD.org> 3251245Sgabor * Copyright (C) 2012 Oleg Moskalenko <mom040267@gmail.com> 4235267Sgabor * All rights reserved. 5235267Sgabor * 6235267Sgabor * Redistribution and use in source and binary forms, with or without 7235267Sgabor * modification, are permitted provided that the following conditions 8235267Sgabor * are met: 9235267Sgabor * 1. Redistributions of source code must retain the above copyright 10235267Sgabor * notice, this list of conditions and the following disclaimer. 11235267Sgabor * 2. Redistributions in binary form must reproduce the above copyright 12235267Sgabor * notice, this list of conditions and the following disclaimer in the 13235267Sgabor * documentation and/or other materials provided with the distribution. 14235267Sgabor * 15235267Sgabor * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 16235267Sgabor * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 17235267Sgabor * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 18235267Sgabor * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 19235267Sgabor * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 20235267Sgabor * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 21235267Sgabor * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 22235267Sgabor * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 23235267Sgabor * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 24235267Sgabor * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 25235267Sgabor * SUCH DAMAGE. 26235267Sgabor */ 27235267Sgabor 28235267Sgabor#include <sys/cdefs.h> 29235267Sgabor__FBSDID("$FreeBSD: stable/10/usr.bin/sort/bwstring.c 309862 2016-12-12 00:47:12Z delphij $"); 30235267Sgabor 31235267Sgabor#include <ctype.h> 32235267Sgabor#include <errno.h> 33235267Sgabor#include <err.h> 34235267Sgabor#include <langinfo.h> 35235267Sgabor#include <math.h> 36235267Sgabor#include <stdlib.h> 37235267Sgabor#include <string.h> 38235267Sgabor#include <wchar.h> 39235267Sgabor#include <wctype.h> 40235267Sgabor 41235267Sgabor#include "bwstring.h" 42235267Sgabor#include "sort.h" 43235267Sgabor 44235435Sgaborbool byte_sort; 45235267Sgabor 46235435Sgaborstatic wchar_t **wmonths; 47235435Sgaborstatic unsigned char **cmonths; 48235267Sgabor 49235267Sgabor/* initialise months */ 50235267Sgabor 51235267Sgaborvoid 52235267Sgaborinitialise_months(void) 53235267Sgabor{ 54235267Sgabor const nl_item item[12] = { ABMON_1, ABMON_2, ABMON_3, ABMON_4, 55235267Sgabor ABMON_5, ABMON_6, ABMON_7, ABMON_8, ABMON_9, ABMON_10, 56235267Sgabor ABMON_11, ABMON_12 }; 57235267Sgabor unsigned char *tmp; 58235267Sgabor size_t len; 59235267Sgabor 60235267Sgabor if (MB_CUR_MAX == 1) { 61235267Sgabor if (cmonths == NULL) { 62235267Sgabor unsigned char *m; 63235267Sgabor 64235267Sgabor cmonths = sort_malloc(sizeof(unsigned char*) * 12); 65235267Sgabor for (int i = 0; i < 12; i++) { 66235267Sgabor cmonths[i] = NULL; 67235267Sgabor tmp = (unsigned char *) nl_langinfo(item[i]); 68235267Sgabor if (debug_sort) 69235267Sgabor printf("month[%d]=%s\n", i, tmp); 70281535Spfg if (*tmp == '\0') 71235267Sgabor continue; 72281535Spfg m = sort_strdup(tmp); 73281535Spfg len = strlen(tmp); 74235267Sgabor for (unsigned int j = 0; j < len; j++) 75235267Sgabor m[j] = toupper(m[j]); 76235267Sgabor cmonths[i] = m; 77235267Sgabor } 78235267Sgabor } 79235267Sgabor 80235267Sgabor } else { 81235267Sgabor if (wmonths == NULL) { 82235267Sgabor wchar_t *m; 83235267Sgabor 84235267Sgabor wmonths = sort_malloc(sizeof(wchar_t *) * 12); 85235267Sgabor for (int i = 0; i < 12; i++) { 86235267Sgabor wmonths[i] = NULL; 87235267Sgabor tmp = (unsigned char *) nl_langinfo(item[i]); 88235267Sgabor if (debug_sort) 89235267Sgabor printf("month[%d]=%s\n", i, tmp); 90281535Spfg if (*tmp == '\0') 91235267Sgabor continue; 92281535Spfg len = strlen(tmp); 93235267Sgabor m = sort_malloc(SIZEOF_WCHAR_STRING(len + 1)); 94281535Spfg if (mbstowcs(m, (char*)tmp, len) == 95281535Spfg ((size_t) - 1)) { 96281535Spfg sort_free(m); 97235267Sgabor continue; 98281535Spfg } 99235267Sgabor m[len] = L'\0'; 100235267Sgabor for (unsigned int j = 0; j < len; j++) 101235267Sgabor m[j] = towupper(m[j]); 102235267Sgabor wmonths[i] = m; 103235267Sgabor } 104235267Sgabor } 105235267Sgabor } 106235267Sgabor} 107235267Sgabor 108235267Sgabor/* 109235267Sgabor * Compare two wide-character strings 110235267Sgabor */ 111235267Sgaborstatic int 112235267Sgaborwide_str_coll(const wchar_t *s1, const wchar_t *s2) 113235267Sgabor{ 114235267Sgabor int ret = 0; 115235267Sgabor 116235267Sgabor errno = 0; 117235267Sgabor ret = wcscoll(s1, s2); 118235267Sgabor if (errno == EILSEQ) { 119235267Sgabor errno = 0; 120235267Sgabor ret = wcscmp(s1, s2); 121235267Sgabor if (errno != 0) { 122235267Sgabor for (size_t i = 0; ; ++i) { 123235267Sgabor wchar_t c1 = s1[i]; 124235267Sgabor wchar_t c2 = s2[i]; 125235267Sgabor if (c1 == L'\0') 126235267Sgabor return ((c2 == L'\0') ? 0 : -1); 127235267Sgabor if (c2 == L'\0') 128235267Sgabor return (+1); 129235267Sgabor if (c1 == c2) 130235267Sgabor continue; 131235267Sgabor return ((int)(c1 - c2)); 132235267Sgabor } 133235267Sgabor } 134235267Sgabor } 135235267Sgabor return (ret); 136235267Sgabor} 137235267Sgabor 138235267Sgabor/* counterparts of wcs functions */ 139235267Sgabor 140235267Sgaborvoid 141235267Sgaborbwsprintf(FILE *f, struct bwstring *bws, const char *prefix, const char *suffix) 142235267Sgabor{ 143235267Sgabor 144235267Sgabor if (MB_CUR_MAX == 1) 145235267Sgabor fprintf(f, "%s%s%s", prefix, bws->data.cstr, suffix); 146235267Sgabor else 147235267Sgabor fprintf(f, "%s%S%s", prefix, bws->data.wstr, suffix); 148235267Sgabor} 149235267Sgabor 150235267Sgaborconst void* bwsrawdata(const struct bwstring *bws) 151235267Sgabor{ 152235267Sgabor 153235267Sgabor return (&(bws->data)); 154235267Sgabor} 155235267Sgabor 156235267Sgaborsize_t bwsrawlen(const struct bwstring *bws) 157235267Sgabor{ 158235267Sgabor 159235267Sgabor return ((MB_CUR_MAX == 1) ? bws->len : SIZEOF_WCHAR_STRING(bws->len)); 160235267Sgabor} 161235267Sgabor 162235267Sgaborsize_t 163235267Sgaborbws_memsize(const struct bwstring *bws) 164235267Sgabor{ 165235267Sgabor 166235267Sgabor return ((MB_CUR_MAX == 1) ? (bws->len + 2 + sizeof(struct bwstring)) : 167235267Sgabor (SIZEOF_WCHAR_STRING(bws->len + 1) + sizeof(struct bwstring))); 168235267Sgabor} 169235267Sgabor 170235267Sgaborvoid 171235267Sgaborbws_setlen(struct bwstring *bws, size_t newlen) 172235267Sgabor{ 173235267Sgabor 174235267Sgabor if (bws && newlen != bws->len && newlen <= bws->len) { 175235267Sgabor bws->len = newlen; 176235267Sgabor if (MB_CUR_MAX == 1) 177235267Sgabor bws->data.cstr[newlen] = '\0'; 178235267Sgabor else 179235267Sgabor bws->data.wstr[newlen] = L'\0'; 180235267Sgabor } 181235267Sgabor} 182235267Sgabor 183235267Sgabor/* 184235267Sgabor * Allocate a new binary string of specified size 185235267Sgabor */ 186235267Sgaborstruct bwstring * 187235267Sgaborbwsalloc(size_t sz) 188235267Sgabor{ 189235267Sgabor struct bwstring *ret; 190235267Sgabor 191235267Sgabor if (MB_CUR_MAX == 1) 192235267Sgabor ret = sort_malloc(sizeof(struct bwstring) + 1 + sz); 193235267Sgabor else 194235267Sgabor ret = sort_malloc(sizeof(struct bwstring) + 195235267Sgabor SIZEOF_WCHAR_STRING(sz + 1)); 196235267Sgabor ret->len = sz; 197235267Sgabor 198235267Sgabor if (MB_CUR_MAX == 1) 199235267Sgabor ret->data.cstr[ret->len] = '\0'; 200235267Sgabor else 201235267Sgabor ret->data.wstr[ret->len] = L'\0'; 202235267Sgabor 203235267Sgabor return (ret); 204235267Sgabor} 205235267Sgabor 206235267Sgabor/* 207235267Sgabor * Create a copy of binary string. 208235267Sgabor * New string size equals the length of the old string. 209235267Sgabor */ 210235267Sgaborstruct bwstring * 211235267Sgaborbwsdup(const struct bwstring *s) 212235267Sgabor{ 213235267Sgabor 214235267Sgabor if (s == NULL) 215235267Sgabor return (NULL); 216235267Sgabor else { 217235267Sgabor struct bwstring *ret = bwsalloc(s->len); 218235267Sgabor 219235267Sgabor if (MB_CUR_MAX == 1) 220235267Sgabor memcpy(ret->data.cstr, s->data.cstr, (s->len)); 221235267Sgabor else 222235267Sgabor memcpy(ret->data.wstr, s->data.wstr, 223235267Sgabor SIZEOF_WCHAR_STRING(s->len)); 224235267Sgabor 225235267Sgabor return (ret); 226235267Sgabor } 227235267Sgabor} 228235267Sgabor 229235267Sgabor/* 230235267Sgabor * Create a new binary string from a raw binary buffer. 231235267Sgabor */ 232235267Sgaborstruct bwstring * 233235267Sgaborbwssbdup(const wchar_t *str, size_t len) 234235267Sgabor{ 235235267Sgabor 236235267Sgabor if (str == NULL) 237235267Sgabor return ((len == 0) ? bwsalloc(0) : NULL); 238235267Sgabor else { 239235267Sgabor struct bwstring *ret; 240235267Sgabor 241235267Sgabor ret = bwsalloc(len); 242235267Sgabor 243235267Sgabor if (MB_CUR_MAX == 1) 244235267Sgabor for (size_t i = 0; i < len; ++i) 245235267Sgabor ret->data.cstr[i] = (unsigned char) str[i]; 246235267Sgabor else 247235267Sgabor memcpy(ret->data.wstr, str, SIZEOF_WCHAR_STRING(len)); 248235267Sgabor 249235267Sgabor return (ret); 250235267Sgabor } 251235267Sgabor} 252235267Sgabor 253235267Sgabor/* 254235267Sgabor * Create a new binary string from a raw binary buffer. 255235267Sgabor */ 256235267Sgaborstruct bwstring * 257235267Sgaborbwscsbdup(const unsigned char *str, size_t len) 258235267Sgabor{ 259235267Sgabor struct bwstring *ret; 260235267Sgabor 261235267Sgabor ret = bwsalloc(len); 262235267Sgabor 263235267Sgabor if (str) { 264235267Sgabor if (MB_CUR_MAX == 1) 265235267Sgabor memcpy(ret->data.cstr, str, len); 266235267Sgabor else { 267235267Sgabor mbstate_t mbs; 268235267Sgabor const char *s; 269235267Sgabor size_t charlen, chars, cptr; 270235267Sgabor 271235267Sgabor charlen = chars = 0; 272235267Sgabor cptr = 0; 273235267Sgabor s = (const char *) str; 274235267Sgabor 275235267Sgabor memset(&mbs, 0, sizeof(mbs)); 276235267Sgabor 277235267Sgabor while (cptr < len) { 278235267Sgabor size_t n = MB_CUR_MAX; 279235267Sgabor 280235267Sgabor if (n > len - cptr) 281235267Sgabor n = len - cptr; 282235267Sgabor charlen = mbrlen(s + cptr, n, &mbs); 283235267Sgabor switch (charlen) { 284235267Sgabor case 0: 285235267Sgabor /* FALLTHROUGH */ 286235267Sgabor case (size_t) -1: 287235267Sgabor /* FALLTHROUGH */ 288235267Sgabor case (size_t) -2: 289235267Sgabor ret->data.wstr[chars++] = 290235267Sgabor (unsigned char) s[cptr]; 291235267Sgabor ++cptr; 292235267Sgabor break; 293235267Sgabor default: 294235267Sgabor n = mbrtowc(ret->data.wstr + (chars++), 295235267Sgabor s + cptr, charlen, &mbs); 296235267Sgabor if ((n == (size_t)-1) || (n == (size_t)-2)) 297235267Sgabor /* NOTREACHED */ 298235267Sgabor err(2, "mbrtowc error"); 299235267Sgabor cptr += charlen; 300235267Sgabor }; 301235267Sgabor } 302235267Sgabor 303235267Sgabor ret->len = chars; 304235267Sgabor ret->data.wstr[ret->len] = L'\0'; 305235267Sgabor } 306235267Sgabor } 307235267Sgabor return (ret); 308235267Sgabor} 309235267Sgabor 310235267Sgabor/* 311235267Sgabor * De-allocate object memory 312235267Sgabor */ 313235267Sgaborvoid 314235267Sgaborbwsfree(const struct bwstring *s) 315235267Sgabor{ 316235267Sgabor 317235267Sgabor if (s) 318235267Sgabor sort_free(s); 319235267Sgabor} 320235267Sgabor 321235267Sgabor/* 322235267Sgabor * Copy content of src binary string to dst. 323235267Sgabor * If the capacity of the dst string is not sufficient, 324235267Sgabor * then the data is truncated. 325235267Sgabor */ 326235267Sgaborsize_t 327235267Sgaborbwscpy(struct bwstring *dst, const struct bwstring *src) 328235267Sgabor{ 329235267Sgabor size_t nums = src->len; 330235267Sgabor 331235267Sgabor if (nums > dst->len) 332235267Sgabor nums = dst->len; 333235267Sgabor dst->len = nums; 334235267Sgabor 335235267Sgabor if (MB_CUR_MAX == 1) { 336235267Sgabor memcpy(dst->data.cstr, src->data.cstr, nums); 337235267Sgabor dst->data.cstr[dst->len] = '\0'; 338235267Sgabor } else { 339235267Sgabor memcpy(dst->data.wstr, src->data.wstr, 340235267Sgabor SIZEOF_WCHAR_STRING(nums + 1)); 341235267Sgabor dst->data.wstr[dst->len] = L'\0'; 342235267Sgabor } 343235267Sgabor 344235267Sgabor return (nums); 345235267Sgabor} 346235267Sgabor 347235267Sgabor/* 348235267Sgabor * Copy content of src binary string to dst, 349235267Sgabor * with specified number of symbols to be copied. 350235267Sgabor * If the capacity of the dst string is not sufficient, 351235267Sgabor * then the data is truncated. 352235267Sgabor */ 353235267Sgaborstruct bwstring * 354235267Sgaborbwsncpy(struct bwstring *dst, const struct bwstring *src, size_t size) 355235267Sgabor{ 356235267Sgabor size_t nums = src->len; 357235267Sgabor 358235267Sgabor if (nums > dst->len) 359235267Sgabor nums = dst->len; 360235267Sgabor if (nums > size) 361235267Sgabor nums = size; 362235267Sgabor dst->len = nums; 363235267Sgabor 364235267Sgabor if (MB_CUR_MAX == 1) { 365235267Sgabor memcpy(dst->data.cstr, src->data.cstr, nums); 366235267Sgabor dst->data.cstr[dst->len] = '\0'; 367235267Sgabor } else { 368235267Sgabor memcpy(dst->data.wstr, src->data.wstr, 369235267Sgabor SIZEOF_WCHAR_STRING(nums + 1)); 370235267Sgabor dst->data.wstr[dst->len] = L'\0'; 371235267Sgabor } 372235267Sgabor 373235267Sgabor return (dst); 374235267Sgabor} 375235267Sgabor 376235267Sgabor/* 377235267Sgabor * Copy content of src binary string to dst, 378235267Sgabor * with specified number of symbols to be copied. 379235267Sgabor * An offset value can be specified, from the start of src string. 380235267Sgabor * If the capacity of the dst string is not sufficient, 381235267Sgabor * then the data is truncated. 382235267Sgabor */ 383235267Sgaborstruct bwstring * 384235267Sgaborbwsnocpy(struct bwstring *dst, const struct bwstring *src, size_t offset, 385235267Sgabor size_t size) 386235267Sgabor{ 387235267Sgabor 388235267Sgabor if (offset >= src->len) { 389235267Sgabor dst->data.wstr[0] = 0; 390235267Sgabor dst->len = 0; 391235267Sgabor } else { 392235267Sgabor size_t nums = src->len - offset; 393235267Sgabor 394235267Sgabor if (nums > dst->len) 395235267Sgabor nums = dst->len; 396235267Sgabor if (nums > size) 397235267Sgabor nums = size; 398235267Sgabor dst->len = nums; 399235267Sgabor if (MB_CUR_MAX == 1) { 400235267Sgabor memcpy(dst->data.cstr, src->data.cstr + offset, 401235267Sgabor (nums)); 402235267Sgabor dst->data.cstr[dst->len] = '\0'; 403235267Sgabor } else { 404235267Sgabor memcpy(dst->data.wstr, src->data.wstr + offset, 405235267Sgabor SIZEOF_WCHAR_STRING(nums)); 406235267Sgabor dst->data.wstr[dst->len] = L'\0'; 407235267Sgabor } 408235267Sgabor } 409235267Sgabor return (dst); 410235267Sgabor} 411235267Sgabor 412235267Sgabor/* 413235267Sgabor * Write binary string to the file. 414235267Sgabor * The output is ended either with '\n' (nl == true) 415235267Sgabor * or '\0' (nl == false). 416235267Sgabor */ 417242430Sgaborsize_t 418235267Sgaborbwsfwrite(struct bwstring *bws, FILE *f, bool zero_ended) 419235267Sgabor{ 420235267Sgabor 421235267Sgabor if (MB_CUR_MAX == 1) { 422235267Sgabor size_t len = bws->len; 423235267Sgabor 424235267Sgabor if (!zero_ended) { 425235267Sgabor bws->data.cstr[len] = '\n'; 426235267Sgabor 427235267Sgabor if (fwrite(bws->data.cstr, len + 1, 1, f) < 1) 428235267Sgabor err(2, NULL); 429235267Sgabor 430235267Sgabor bws->data.cstr[len] = '\0'; 431235267Sgabor } else if (fwrite(bws->data.cstr, len + 1, 1, f) < 1) 432235267Sgabor err(2, NULL); 433235267Sgabor 434235267Sgabor return (len + 1); 435235267Sgabor 436235267Sgabor } else { 437235267Sgabor wchar_t eols; 438242430Sgabor size_t printed = 0; 439235267Sgabor 440235267Sgabor eols = zero_ended ? btowc('\0') : btowc('\n'); 441235267Sgabor 442242430Sgabor while (printed < BWSLEN(bws)) { 443235267Sgabor const wchar_t *s = bws->data.wstr + printed; 444235267Sgabor 445235267Sgabor if (*s == L'\0') { 446235267Sgabor int nums; 447235267Sgabor 448235267Sgabor nums = fwprintf(f, L"%lc", *s); 449235267Sgabor 450235267Sgabor if (nums != 1) 451235267Sgabor err(2, NULL); 452235267Sgabor ++printed; 453235267Sgabor } else { 454235267Sgabor int nums; 455235267Sgabor 456235267Sgabor nums = fwprintf(f, L"%ls", s); 457235267Sgabor 458235267Sgabor if (nums < 1) 459235267Sgabor err(2, NULL); 460235267Sgabor printed += nums; 461235267Sgabor } 462235267Sgabor } 463235267Sgabor fwprintf(f, L"%lc", eols); 464235267Sgabor return (printed + 1); 465235267Sgabor } 466235267Sgabor} 467235267Sgabor 468235267Sgabor/* 469235267Sgabor * Allocate and read a binary string from file. 470235267Sgabor * The strings are nl-ended or zero-ended, depending on the sort setting. 471235267Sgabor */ 472235267Sgaborstruct bwstring * 473235267Sgaborbwsfgetln(FILE *f, size_t *len, bool zero_ended, struct reader_buffer *rb) 474235267Sgabor{ 475242430Sgabor wint_t eols; 476235267Sgabor 477235267Sgabor eols = zero_ended ? btowc('\0') : btowc('\n'); 478235267Sgabor 479235267Sgabor if (!zero_ended && (MB_CUR_MAX > 1)) { 480235267Sgabor wchar_t *ret; 481235267Sgabor 482235267Sgabor ret = fgetwln(f, len); 483235267Sgabor 484235267Sgabor if (ret == NULL) { 485235267Sgabor if (!feof(f)) 486235267Sgabor err(2, NULL); 487235267Sgabor return (NULL); 488235267Sgabor } 489235267Sgabor if (*len > 0) { 490242430Sgabor if (ret[*len - 1] == (wchar_t)eols) 491235267Sgabor --(*len); 492235267Sgabor } 493235267Sgabor return (bwssbdup(ret, *len)); 494235267Sgabor 495235987Sgabor } else if (!zero_ended && (MB_CUR_MAX == 1)) { 496235987Sgabor char *ret; 497235987Sgabor 498235987Sgabor ret = fgetln(f, len); 499235987Sgabor 500235987Sgabor if (ret == NULL) { 501235987Sgabor if (!feof(f)) 502235987Sgabor err(2, NULL); 503235987Sgabor return (NULL); 504235987Sgabor } 505235987Sgabor if (*len > 0) { 506235987Sgabor if (ret[*len - 1] == '\n') 507235987Sgabor --(*len); 508235987Sgabor } 509242430Sgabor return (bwscsbdup((unsigned char*)ret, *len)); 510235987Sgabor 511235267Sgabor } else { 512235267Sgabor *len = 0; 513235267Sgabor 514235267Sgabor if (feof(f)) 515235267Sgabor return (NULL); 516235267Sgabor 517235267Sgabor if (2 >= rb->fgetwln_z_buffer_size) { 518235267Sgabor rb->fgetwln_z_buffer_size += 256; 519235267Sgabor rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer, 520235267Sgabor sizeof(wchar_t) * rb->fgetwln_z_buffer_size); 521235267Sgabor } 522235267Sgabor rb->fgetwln_z_buffer[*len] = 0; 523235267Sgabor 524235267Sgabor if (MB_CUR_MAX == 1) 525235267Sgabor while (!feof(f)) { 526242430Sgabor int c; 527242430Sgabor 528235267Sgabor c = fgetc(f); 529235267Sgabor 530235267Sgabor if (c == EOF) { 531235267Sgabor if (*len == 0) 532235267Sgabor return (NULL); 533235267Sgabor goto line_read_done; 534235267Sgabor } 535235267Sgabor if (c == eols) 536235267Sgabor goto line_read_done; 537235267Sgabor 538235267Sgabor if (*len + 1 >= rb->fgetwln_z_buffer_size) { 539235267Sgabor rb->fgetwln_z_buffer_size += 256; 540235267Sgabor rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer, 541235267Sgabor SIZEOF_WCHAR_STRING(rb->fgetwln_z_buffer_size)); 542235267Sgabor } 543235267Sgabor 544235267Sgabor rb->fgetwln_z_buffer[*len] = c; 545235267Sgabor rb->fgetwln_z_buffer[++(*len)] = 0; 546235267Sgabor } 547235267Sgabor else 548235267Sgabor while (!feof(f)) { 549242430Sgabor wint_t c = 0; 550242430Sgabor 551235267Sgabor c = fgetwc(f); 552235267Sgabor 553235267Sgabor if (c == WEOF) { 554235267Sgabor if (*len == 0) 555235267Sgabor return (NULL); 556235267Sgabor goto line_read_done; 557235267Sgabor } 558235267Sgabor if (c == eols) 559235267Sgabor goto line_read_done; 560235267Sgabor 561235267Sgabor if (*len + 1 >= rb->fgetwln_z_buffer_size) { 562235267Sgabor rb->fgetwln_z_buffer_size += 256; 563235267Sgabor rb->fgetwln_z_buffer = sort_realloc(rb->fgetwln_z_buffer, 564235267Sgabor SIZEOF_WCHAR_STRING(rb->fgetwln_z_buffer_size)); 565235267Sgabor } 566235267Sgabor 567235267Sgabor rb->fgetwln_z_buffer[*len] = c; 568235267Sgabor rb->fgetwln_z_buffer[++(*len)] = 0; 569235267Sgabor } 570235267Sgabor 571235267Sgaborline_read_done: 572235267Sgabor /* we do not count the last 0 */ 573235267Sgabor return (bwssbdup(rb->fgetwln_z_buffer, *len)); 574235267Sgabor } 575235267Sgabor} 576235267Sgabor 577235267Sgaborint 578235267Sgaborbwsncmp(const struct bwstring *bws1, const struct bwstring *bws2, 579235267Sgabor size_t offset, size_t len) 580235267Sgabor{ 581235267Sgabor size_t cmp_len, len1, len2; 582235267Sgabor int res = 0; 583235267Sgabor 584235267Sgabor cmp_len = 0; 585235267Sgabor len1 = bws1->len; 586235267Sgabor len2 = bws2->len; 587235267Sgabor 588235267Sgabor if (len1 <= offset) { 589235267Sgabor return ((len2 <= offset) ? 0 : -1); 590235267Sgabor } else { 591235267Sgabor if (len2 <= offset) 592235267Sgabor return (+1); 593235267Sgabor else { 594235267Sgabor len1 -= offset; 595235267Sgabor len2 -= offset; 596235267Sgabor 597235267Sgabor cmp_len = len1; 598235267Sgabor 599235267Sgabor if (len2 < cmp_len) 600235267Sgabor cmp_len = len2; 601235267Sgabor 602235267Sgabor if (len < cmp_len) 603235267Sgabor cmp_len = len; 604235267Sgabor 605235267Sgabor if (MB_CUR_MAX == 1) { 606235267Sgabor const unsigned char *s1, *s2; 607235267Sgabor 608235267Sgabor s1 = bws1->data.cstr + offset; 609235267Sgabor s2 = bws2->data.cstr + offset; 610235267Sgabor 611235267Sgabor res = memcmp(s1, s2, cmp_len); 612235267Sgabor 613235267Sgabor } else { 614235267Sgabor const wchar_t *s1, *s2; 615235267Sgabor 616235267Sgabor s1 = bws1->data.wstr + offset; 617235267Sgabor s2 = bws2->data.wstr + offset; 618235267Sgabor 619235267Sgabor res = memcmp(s1, s2, SIZEOF_WCHAR_STRING(cmp_len)); 620235267Sgabor } 621235267Sgabor } 622235267Sgabor } 623235267Sgabor 624235267Sgabor if (res == 0) { 625235267Sgabor if (len1 < cmp_len && len1 < len2) 626235267Sgabor res = -1; 627235267Sgabor else if (len2 < cmp_len && len2 < len1) 628235267Sgabor res = +1; 629235267Sgabor } 630235267Sgabor 631235267Sgabor return (res); 632235267Sgabor} 633235267Sgabor 634235267Sgaborint 635235267Sgaborbwscmp(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) 636235267Sgabor{ 637235267Sgabor size_t len1, len2, cmp_len; 638235267Sgabor int res; 639235267Sgabor 640235267Sgabor len1 = bws1->len; 641235267Sgabor len2 = bws2->len; 642235267Sgabor 643235267Sgabor len1 -= offset; 644235267Sgabor len2 -= offset; 645235267Sgabor 646235267Sgabor cmp_len = len1; 647235267Sgabor 648235267Sgabor if (len2 < cmp_len) 649235267Sgabor cmp_len = len2; 650235267Sgabor 651235267Sgabor res = bwsncmp(bws1, bws2, offset, cmp_len); 652235267Sgabor 653235267Sgabor if (res == 0) { 654235267Sgabor if( len1 < len2) 655235267Sgabor res = -1; 656235267Sgabor else if (len2 < len1) 657235267Sgabor res = +1; 658235267Sgabor } 659235267Sgabor 660235267Sgabor return (res); 661235267Sgabor} 662235267Sgabor 663235267Sgaborint 664235267Sgaborbws_iterator_cmp(bwstring_iterator iter1, bwstring_iterator iter2, size_t len) 665235267Sgabor{ 666235267Sgabor wchar_t c1, c2; 667235267Sgabor size_t i = 0; 668235267Sgabor 669235267Sgabor for (i = 0; i < len; ++i) { 670235267Sgabor c1 = bws_get_iter_value(iter1); 671235267Sgabor c2 = bws_get_iter_value(iter2); 672235267Sgabor if (c1 != c2) 673235267Sgabor return (c1 - c2); 674235267Sgabor iter1 = bws_iterator_inc(iter1, 1); 675235267Sgabor iter2 = bws_iterator_inc(iter2, 1); 676235267Sgabor } 677235267Sgabor 678235267Sgabor return (0); 679235267Sgabor} 680235267Sgabor 681235267Sgaborint 682235267Sgaborbwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) 683235267Sgabor{ 684235267Sgabor size_t len1, len2; 685235267Sgabor 686235267Sgabor len1 = bws1->len; 687235267Sgabor len2 = bws2->len; 688235267Sgabor 689235267Sgabor if (len1 <= offset) 690235267Sgabor return ((len2 <= offset) ? 0 : -1); 691235267Sgabor else { 692235267Sgabor if (len2 <= offset) 693235267Sgabor return (+1); 694235267Sgabor else { 695235267Sgabor len1 -= offset; 696235267Sgabor len2 -= offset; 697235267Sgabor 698235267Sgabor if (MB_CUR_MAX == 1) { 699235267Sgabor const unsigned char *s1, *s2; 700235267Sgabor 701235267Sgabor s1 = bws1->data.cstr + offset; 702235267Sgabor s2 = bws2->data.cstr + offset; 703235267Sgabor 704235267Sgabor if (byte_sort) { 705235267Sgabor int res = 0; 706235267Sgabor 707235267Sgabor if (len1 > len2) { 708235267Sgabor res = memcmp(s1, s2, len2); 709235267Sgabor if (!res) 710235267Sgabor res = +1; 711235267Sgabor } else if (len1 < len2) { 712235267Sgabor res = memcmp(s1, s2, len1); 713235267Sgabor if (!res) 714235267Sgabor res = -1; 715235267Sgabor } else 716235267Sgabor res = memcmp(s1, s2, len1); 717235267Sgabor 718235267Sgabor return (res); 719235267Sgabor 720235267Sgabor } else { 721235267Sgabor int res = 0; 722235267Sgabor size_t i, maxlen; 723235267Sgabor 724235267Sgabor i = 0; 725235267Sgabor maxlen = len1; 726235267Sgabor 727235267Sgabor if (maxlen > len2) 728235267Sgabor maxlen = len2; 729235267Sgabor 730235267Sgabor while (i < maxlen) { 731235267Sgabor /* goto next non-zero part: */ 732235267Sgabor while ((i < maxlen) && 733235267Sgabor !s1[i] && !s2[i]) 734235267Sgabor ++i; 735235267Sgabor 736235267Sgabor if (i >= maxlen) 737235267Sgabor break; 738235267Sgabor 739235267Sgabor if (s1[i] == 0) { 740235267Sgabor if (s2[i] == 0) 741235267Sgabor /* NOTREACHED */ 742235267Sgabor err(2, "bwscoll error 01"); 743235267Sgabor else 744235267Sgabor return (-1); 745235267Sgabor } else if (s2[i] == 0) 746235267Sgabor return (+1); 747235267Sgabor 748242430Sgabor res = strcoll((const char*)(s1 + i), (const char*)(s2 + i)); 749235267Sgabor if (res) 750235267Sgabor return (res); 751235267Sgabor 752235267Sgabor while ((i < maxlen) && 753235267Sgabor s1[i] && s2[i]) 754235267Sgabor ++i; 755235267Sgabor 756235267Sgabor if (i >= maxlen) 757235267Sgabor break; 758235267Sgabor 759235267Sgabor if (s1[i] == 0) { 760235267Sgabor if (s2[i] == 0) { 761235267Sgabor ++i; 762235267Sgabor continue; 763235267Sgabor } else 764235267Sgabor return (-1); 765235267Sgabor } else if (s2[i] == 0) 766235267Sgabor return (+1); 767235267Sgabor else 768235267Sgabor /* NOTREACHED */ 769235267Sgabor err(2, "bwscoll error 02"); 770235267Sgabor } 771235267Sgabor 772235267Sgabor if (len1 < len2) 773235267Sgabor return (-1); 774235267Sgabor else if (len1 > len2) 775235267Sgabor return (+1); 776235267Sgabor 777235267Sgabor return (0); 778235267Sgabor } 779235267Sgabor } else { 780235267Sgabor const wchar_t *s1, *s2; 781235267Sgabor size_t i, maxlen; 782235267Sgabor int res = 0; 783235267Sgabor 784235267Sgabor s1 = bws1->data.wstr + offset; 785235267Sgabor s2 = bws2->data.wstr + offset; 786235267Sgabor 787235267Sgabor i = 0; 788235267Sgabor maxlen = len1; 789235267Sgabor 790235267Sgabor if (maxlen > len2) 791235267Sgabor maxlen = len2; 792235267Sgabor 793235267Sgabor while (i < maxlen) { 794235267Sgabor 795235267Sgabor /* goto next non-zero part: */ 796235267Sgabor while ((i < maxlen) && 797235267Sgabor !s1[i] && !s2[i]) 798235267Sgabor ++i; 799235267Sgabor 800235267Sgabor if (i >= maxlen) 801235267Sgabor break; 802235267Sgabor 803235267Sgabor if (s1[i] == 0) { 804235267Sgabor if (s2[i] == 0) 805235267Sgabor /* NOTREACHED */ 806235267Sgabor err(2, "bwscoll error 1"); 807235267Sgabor else 808235267Sgabor return (-1); 809235267Sgabor } else if (s2[i] == 0) 810235267Sgabor return (+1); 811235267Sgabor 812235267Sgabor res = wide_str_coll(s1 + i, s2 + i); 813235267Sgabor if (res) 814235267Sgabor return (res); 815235267Sgabor 816235267Sgabor while ((i < maxlen) && s1[i] && s2[i]) 817235267Sgabor ++i; 818235267Sgabor 819235267Sgabor if (i >= maxlen) 820235267Sgabor break; 821235267Sgabor 822235267Sgabor if (s1[i] == 0) { 823235267Sgabor if (s2[i] == 0) { 824235267Sgabor ++i; 825235267Sgabor continue; 826235267Sgabor } else 827235267Sgabor return (-1); 828235267Sgabor } else if (s2[i] == 0) 829235267Sgabor return (+1); 830235267Sgabor else 831235267Sgabor /* NOTREACHED */ 832235267Sgabor err(2, "bwscoll error 2"); 833235267Sgabor } 834235267Sgabor 835235267Sgabor if (len1 < len2) 836235267Sgabor return (-1); 837235267Sgabor else if (len1 > len2) 838235267Sgabor return (+1); 839235267Sgabor 840235267Sgabor return (0); 841235267Sgabor } 842235267Sgabor } 843235267Sgabor } 844235267Sgabor} 845235267Sgabor 846235267Sgabor/* 847235267Sgabor * Correction of the system API 848235267Sgabor */ 849235267Sgabordouble 850235267Sgaborbwstod(struct bwstring *s0, bool *empty) 851235267Sgabor{ 852235267Sgabor double ret = 0; 853235267Sgabor 854235267Sgabor if (MB_CUR_MAX == 1) { 855235267Sgabor unsigned char *end, *s; 856235267Sgabor char *ep; 857235267Sgabor 858235267Sgabor s = s0->data.cstr; 859235267Sgabor end = s + s0->len; 860235267Sgabor ep = NULL; 861235267Sgabor 862235267Sgabor while (isblank(*s) && s < end) 863235267Sgabor ++s; 864235267Sgabor 865235267Sgabor if (!isprint(*s)) { 866235267Sgabor *empty = true; 867235267Sgabor return (0); 868235267Sgabor } 869235267Sgabor 870242430Sgabor ret = strtod((char*)s, &ep); 871235267Sgabor if ((unsigned char*) ep == s) { 872235267Sgabor *empty = true; 873235267Sgabor return (0); 874235267Sgabor } 875235267Sgabor } else { 876235267Sgabor wchar_t *end, *ep, *s; 877235267Sgabor 878235267Sgabor s = s0->data.wstr; 879235267Sgabor end = s + s0->len; 880235267Sgabor ep = NULL; 881235267Sgabor 882235267Sgabor while (iswblank(*s) && s < end) 883235267Sgabor ++s; 884235267Sgabor 885235267Sgabor if (!iswprint(*s)) { 886235267Sgabor *empty = true; 887235267Sgabor return (0); 888235267Sgabor } 889235267Sgabor 890235267Sgabor ret = wcstod(s, &ep); 891235267Sgabor if (ep == s) { 892235267Sgabor *empty = true; 893235267Sgabor return (0); 894235267Sgabor } 895235267Sgabor } 896235267Sgabor 897235267Sgabor *empty = false; 898235267Sgabor return (ret); 899235267Sgabor} 900235267Sgabor 901235267Sgabor/* 902235267Sgabor * A helper function for monthcoll. If a line matches 903235267Sgabor * a month name, it returns (number of the month - 1), 904235267Sgabor * while if there is no match, it just return -1. 905235267Sgabor */ 906235267Sgabor 907235267Sgaborint 908235267Sgaborbws_month_score(const struct bwstring *s0) 909235267Sgabor{ 910235267Sgabor 911235267Sgabor if (MB_CUR_MAX == 1) { 912235267Sgabor const unsigned char *end, *s; 913235267Sgabor 914235267Sgabor s = s0->data.cstr; 915235267Sgabor end = s + s0->len; 916235267Sgabor 917235267Sgabor while (isblank(*s) && s < end) 918235267Sgabor ++s; 919235267Sgabor 920235267Sgabor for (int i = 11; i >= 0; --i) { 921235267Sgabor if (cmonths[i] && 922242430Sgabor (s == (unsigned char*)strstr((const char*)s, (char*)(cmonths[i])))) 923235267Sgabor return (i); 924235267Sgabor } 925235267Sgabor 926235267Sgabor } else { 927235267Sgabor const wchar_t *end, *s; 928235267Sgabor 929235267Sgabor s = s0->data.wstr; 930235267Sgabor end = s + s0->len; 931235267Sgabor 932235267Sgabor while (iswblank(*s) && s < end) 933235267Sgabor ++s; 934235267Sgabor 935235267Sgabor for (int i = 11; i >= 0; --i) { 936235267Sgabor if (wmonths[i] && (s == wcsstr(s, wmonths[i]))) 937235267Sgabor return (i); 938235267Sgabor } 939235267Sgabor } 940235267Sgabor 941235267Sgabor return (-1); 942235267Sgabor} 943235267Sgabor 944235267Sgabor/* 945235267Sgabor * Rips out leading blanks (-b). 946235267Sgabor */ 947235267Sgaborstruct bwstring * 948235267Sgaborignore_leading_blanks(struct bwstring *str) 949235267Sgabor{ 950235267Sgabor 951235267Sgabor if (MB_CUR_MAX == 1) { 952235267Sgabor unsigned char *dst, *end, *src; 953235267Sgabor 954235267Sgabor src = str->data.cstr; 955235267Sgabor dst = src; 956235267Sgabor end = src + str->len; 957235267Sgabor 958235267Sgabor while (src < end && isblank(*src)) 959235267Sgabor ++src; 960235267Sgabor 961235267Sgabor if (src != dst) { 962235267Sgabor size_t newlen; 963235267Sgabor 964235267Sgabor newlen = BWSLEN(str) - (src - dst); 965235267Sgabor 966235267Sgabor while (src < end) { 967235267Sgabor *dst = *src; 968235267Sgabor ++dst; 969235267Sgabor ++src; 970235267Sgabor } 971235267Sgabor bws_setlen(str, newlen); 972235267Sgabor } 973235267Sgabor } else { 974235267Sgabor wchar_t *dst, *end, *src; 975235267Sgabor 976235267Sgabor src = str->data.wstr; 977235267Sgabor dst = src; 978235267Sgabor end = src + str->len; 979235267Sgabor 980235267Sgabor while (src < end && iswblank(*src)) 981235267Sgabor ++src; 982235267Sgabor 983235267Sgabor if (src != dst) { 984235267Sgabor 985235267Sgabor size_t newlen = BWSLEN(str) - (src - dst); 986235267Sgabor 987235267Sgabor while (src < end) { 988235267Sgabor *dst = *src; 989235267Sgabor ++dst; 990235267Sgabor ++src; 991235267Sgabor } 992235267Sgabor bws_setlen(str, newlen); 993235267Sgabor 994235267Sgabor } 995235267Sgabor } 996235267Sgabor return (str); 997235267Sgabor} 998235267Sgabor 999235267Sgabor/* 1000235267Sgabor * Rips out nonprinting characters (-i). 1001235267Sgabor */ 1002235267Sgaborstruct bwstring * 1003235267Sgaborignore_nonprinting(struct bwstring *str) 1004235267Sgabor{ 1005235267Sgabor size_t newlen = str->len; 1006235267Sgabor 1007235267Sgabor if (MB_CUR_MAX == 1) { 1008235267Sgabor unsigned char *dst, *end, *src; 1009235267Sgabor unsigned char c; 1010235267Sgabor 1011235267Sgabor src = str->data.cstr; 1012235267Sgabor dst = src; 1013235267Sgabor end = src + str->len; 1014235267Sgabor 1015235267Sgabor while (src < end) { 1016235267Sgabor c = *src; 1017235267Sgabor if (isprint(c)) { 1018235267Sgabor *dst = c; 1019235267Sgabor ++dst; 1020235267Sgabor ++src; 1021235267Sgabor } else { 1022235267Sgabor ++src; 1023235267Sgabor --newlen; 1024235267Sgabor } 1025235267Sgabor } 1026235267Sgabor } else { 1027235267Sgabor wchar_t *dst, *end, *src; 1028235267Sgabor wchar_t c; 1029235267Sgabor 1030235267Sgabor src = str->data.wstr; 1031235267Sgabor dst = src; 1032235267Sgabor end = src + str->len; 1033235267Sgabor 1034235267Sgabor while (src < end) { 1035235267Sgabor c = *src; 1036235267Sgabor if (iswprint(c)) { 1037235267Sgabor *dst = c; 1038235267Sgabor ++dst; 1039235267Sgabor ++src; 1040235267Sgabor } else { 1041235267Sgabor ++src; 1042235267Sgabor --newlen; 1043235267Sgabor } 1044235267Sgabor } 1045235267Sgabor } 1046235267Sgabor bws_setlen(str, newlen); 1047235267Sgabor 1048235267Sgabor return (str); 1049235267Sgabor} 1050235267Sgabor 1051235267Sgabor/* 1052235267Sgabor * Rips out any characters that are not alphanumeric characters 1053235267Sgabor * nor blanks (-d). 1054235267Sgabor */ 1055235267Sgaborstruct bwstring * 1056235267Sgabordictionary_order(struct bwstring *str) 1057235267Sgabor{ 1058235267Sgabor size_t newlen = str->len; 1059235267Sgabor 1060235267Sgabor if (MB_CUR_MAX == 1) { 1061235267Sgabor unsigned char *dst, *end, *src; 1062235267Sgabor unsigned char c; 1063235267Sgabor 1064235267Sgabor src = str->data.cstr; 1065235267Sgabor dst = src; 1066235267Sgabor end = src + str->len; 1067235267Sgabor 1068235267Sgabor while (src < end) { 1069235267Sgabor c = *src; 1070235267Sgabor if (isalnum(c) || isblank(c)) { 1071235267Sgabor *dst = c; 1072235267Sgabor ++dst; 1073235267Sgabor ++src; 1074235267Sgabor } else { 1075235267Sgabor ++src; 1076235267Sgabor --newlen; 1077235267Sgabor } 1078235267Sgabor } 1079235267Sgabor } else { 1080235267Sgabor wchar_t *dst, *end, *src; 1081235267Sgabor wchar_t c; 1082235267Sgabor 1083235267Sgabor src = str->data.wstr; 1084235267Sgabor dst = src; 1085235267Sgabor end = src + str->len; 1086235267Sgabor 1087235267Sgabor while (src < end) { 1088235267Sgabor c = *src; 1089235267Sgabor if (iswalnum(c) || iswblank(c)) { 1090235267Sgabor *dst = c; 1091235267Sgabor ++dst; 1092235267Sgabor ++src; 1093235267Sgabor } else { 1094235267Sgabor ++src; 1095235267Sgabor --newlen; 1096235267Sgabor } 1097235267Sgabor } 1098235267Sgabor } 1099235267Sgabor bws_setlen(str, newlen); 1100235267Sgabor 1101235267Sgabor return (str); 1102235267Sgabor} 1103235267Sgabor 1104235267Sgabor/* 1105235267Sgabor * Converts string to lower case(-f). 1106235267Sgabor */ 1107235267Sgaborstruct bwstring * 1108235267Sgaborignore_case(struct bwstring *str) 1109235267Sgabor{ 1110235267Sgabor 1111235267Sgabor if (MB_CUR_MAX == 1) { 1112235267Sgabor unsigned char *end, *s; 1113235267Sgabor 1114235267Sgabor s = str->data.cstr; 1115235267Sgabor end = s + str->len; 1116235267Sgabor 1117235267Sgabor while (s < end) { 1118235267Sgabor *s = toupper(*s); 1119235267Sgabor ++s; 1120235267Sgabor } 1121235267Sgabor } else { 1122235267Sgabor wchar_t *end, *s; 1123235267Sgabor 1124235267Sgabor s = str->data.wstr; 1125235267Sgabor end = s + str->len; 1126235267Sgabor 1127235267Sgabor while (s < end) { 1128235267Sgabor *s = towupper(*s); 1129235267Sgabor ++s; 1130235267Sgabor } 1131235267Sgabor } 1132235267Sgabor return (str); 1133235267Sgabor} 1134235267Sgabor 1135235267Sgaborvoid 1136235267Sgaborbws_disorder_warnx(struct bwstring *s, const char *fn, size_t pos) 1137235267Sgabor{ 1138235267Sgabor 1139235267Sgabor if (MB_CUR_MAX == 1) 1140235267Sgabor warnx("%s:%zu: disorder: %s", fn, pos + 1, s->data.cstr); 1141235267Sgabor else 1142235267Sgabor warnx("%s:%zu: disorder: %ls", fn, pos + 1, s->data.wstr); 1143235267Sgabor} 1144