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 <sys/stat.h> 32 #include <sys/sysctl.h> 33 #include <sys/types.h> 34 35 #include <err.h> 36 #include <errno.h> 37 #include <getopt.h> 38 #include <limits.h> 39 #include <locale.h> 40 #include <md5.h> 41 #include <regex.h> 42 #include <signal.h> 43 #include <stdbool.h> 44 #include <stdio.h> 45 #include <stdlib.h> 46 #include <string.h> 47 #include <unistd.h> 48 #include <wchar.h> 49 #include <wctype.h> 50 51 #include "coll.h" 52 #include "file.h" 53 #include "sort.h" 54 55 #ifndef WITHOUT_NLS 56 #include <nl_types.h> 57 nl_catd catalog; 58 #endif 59 60 #define OPTIONS "bcCdfghik:Mmno:RrsS:t:T:uVz" 61 62 #define DEFAULT_RANDOM_SORT_SEED_FILE ("/dev/random") 63 #define MAX_DEFAULT_RANDOM_SEED_DATA_SIZE (1024) 64 65 static bool need_random = false; 66 static const char* random_source = DEFAULT_RANDOM_SORT_SEED_FILE; 67 static const void* random_seed = NULL; 68 static size_t random_seed_size = 0; 69 70 MD5_CTX md5_ctx; 71 72 /* 73 * Default messages to use when NLS is disabled or no catalogue 74 * is found. 75 */ 76 const char *nlsstr[] = { "", 77 /* 1*/"you cannot use -%c and -%c together", 78 /* 2*/"extra argument not allowed with -c", 79 /* 3*/"Unknown feature: %s", 80 /* 4*/"Wrong memory buffer specification", 81 /* 5*/"0 field in key specs", 82 /* 6*/"0 column in key specs", 83 /* 7*/"Wrong file mode", 84 /* 8*/"Cannot open file for reading", 85 /* 9*/"Radix sort cannot be used with these sort options", 86 /*10*/"The chosen sort method cannot be used with stable and/or unique sort", 87 /*11*/"Invalid key position", 88 /*12*/"Usage: %s [-bcCdfigMmnrsuz] [-kPOS1[,POS2] ... ] " 89 "[+POS1 [-POS2]] [-S memsize] [-T tmpdir] [-t separator] " 90 "[-o outfile] [--batch-size size] [--files0-from file] " 91 "[--heapsort] [--mergesort] [--radixsort] [--qsort] " 92 #if defined(SORT_THREADS) 93 "[--nthreads thread_no] " 94 #endif 95 "[--human-numeric-sort] " 96 "[--version-sort] [--random-sort [--random-source file]] " 97 "[--compress-program program] [file ...]\n" }; 98 99 struct sort_opts sort_opts_vals; 100 101 bool debug_sort = false; 102 bool need_hint = false; 103 104 #if defined(SORT_THREADS) 105 size_t ncpu = 1; 106 size_t nthreads = 1; 107 #endif 108 109 static bool gnusort_numeric_compatibility = false; 110 111 static struct sort_mods default_sort_mods_object; 112 struct sort_mods * const default_sort_mods = &default_sort_mods_object; 113 114 static bool print_symbols_on_debug = false; 115 116 /* 117 * Arguments from file (when file0-from option is used: 118 */ 119 static int argc_from_file0 = -1; 120 static char **argv_from_file0 = NULL; 121 122 /* 123 * Placeholder symbols for options which have no single-character equivalent 124 */ 125 enum 126 { 127 SORT_OPT = CHAR_MAX + 1, 128 HELP_OPT, 129 FF_OPT, 130 BS_OPT, 131 VERSION_OPT, 132 DEBUG_OPT, 133 #if defined(SORT_THREADS) 134 NTHREADS_OPT, 135 #endif 136 RANDOMSOURCE_OPT, 137 COMPRESSPROGRAM_OPT, 138 QSORT_OPT, 139 MERGESORT_OPT, 140 HEAPSORT_OPT, 141 RADIXSORT_OPT 142 }; 143 144 #define NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS 6 145 static const char mutually_exclusive_flags[NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS] = { 'M', 'n', 'g', 'R', 'h', 'V' }; 146 147 struct option long_options[] = { 148 { "batch-size", required_argument, NULL, BS_OPT }, 149 { "buffer-size", required_argument, NULL, 'S' }, 150 { "check", optional_argument, NULL, 'c' }, 151 { "check=silent|quiet", optional_argument, NULL, 'C' }, 152 { "compress-program", required_argument, NULL, COMPRESSPROGRAM_OPT }, 153 { "debug", no_argument, NULL, DEBUG_OPT }, 154 { "dictionary-order", no_argument, NULL, 'd' }, 155 { "field-separator", required_argument, NULL, 't' }, 156 { "files0-from", required_argument, NULL, FF_OPT }, 157 { "general-numeric-sort", no_argument, NULL, 'g' }, 158 { "heapsort", no_argument, NULL, HEAPSORT_OPT }, 159 { "help",no_argument, NULL, HELP_OPT }, 160 { "human-numeric-sort", no_argument, NULL, 'h' }, 161 { "ignore-leading-blanks", no_argument, NULL, 'b' }, 162 { "ignore-case", no_argument, NULL, 'f' }, 163 { "ignore-nonprinting", no_argument, NULL, 'i' }, 164 { "key", required_argument, NULL, 'k' }, 165 { "merge", no_argument, NULL, 'm' }, 166 { "mergesort", no_argument, NULL, MERGESORT_OPT }, 167 { "month-sort", no_argument, NULL, 'M' }, 168 { "numeric-sort", no_argument, NULL, 'n' }, 169 { "output", required_argument, NULL, 'o' }, 170 #if defined(SORT_THREADS) 171 { "nthreads", required_argument, NULL, NTHREADS_OPT }, 172 #endif 173 { "qsort", no_argument, NULL, QSORT_OPT }, 174 { "radixsort", no_argument, NULL, RADIXSORT_OPT }, 175 { "random-sort", no_argument, NULL, 'R' }, 176 { "random-source", required_argument, NULL, RANDOMSOURCE_OPT }, 177 { "reverse", no_argument, NULL, 'r' }, 178 { "sort", required_argument, NULL, SORT_OPT }, 179 { "stable", no_argument, NULL, 's' }, 180 { "temporary-directory",required_argument, NULL, 'T' }, 181 { "unique", no_argument, NULL, 'u' }, 182 { "version", no_argument, NULL, VERSION_OPT }, 183 { "version-sort",no_argument, NULL, 'V' }, 184 { "zero-terminated", no_argument, NULL, 'z' }, 185 { NULL, no_argument, NULL, 0 } 186 }; 187 188 void fix_obsolete_keys(int *argc, char **argv); 189 190 /* 191 * Check where sort modifier is present 192 */ 193 static bool 194 sort_modifier_empty(struct sort_mods *sm) 195 { 196 197 if (sm == NULL) 198 return (true); 199 return (!(sm->Mflag || sm->Vflag || sm->nflag || sm->gflag || 200 sm->rflag || sm->Rflag || sm->hflag || sm->dflag || sm->fflag)); 201 } 202 203 /* 204 * Print out usage text. 205 */ 206 static void 207 usage(bool opt_err) 208 { 209 struct option *o; 210 FILE *out; 211 212 out = stdout; 213 o = &(long_options[0]); 214 215 if (opt_err) 216 out = stderr; 217 fprintf(out, getstr(12), getprogname()); 218 if (opt_err) 219 exit(2); 220 exit(0); 221 } 222 223 /* 224 * Read input file names from a file (file0-from option). 225 */ 226 static void 227 read_fns_from_file0(const char *fn) 228 { 229 if (fn) { 230 struct file0_reader f0r; 231 FILE *f; 232 233 f = fopen(fn, "r"); 234 if (f == NULL) 235 err(2, NULL); 236 237 memset(&f0r, 0, sizeof(f0r)); 238 f0r.f = f; 239 240 while (!feof(f)) { 241 char *line = read_file0_line(&f0r); 242 243 if (line && *line) { 244 ++argc_from_file0; 245 if (argc_from_file0 < 1) 246 argc_from_file0 = 1; 247 argv_from_file0 = sort_realloc(argv_from_file0, 248 argc_from_file0 * sizeof(char *)); 249 if (argv_from_file0 == NULL) 250 err(2, NULL); 251 argv_from_file0[argc_from_file0 - 1] = 252 sort_strdup(line); 253 } 254 } 255 closefile(f, fn); 256 } 257 } 258 259 /* 260 * Check how much RAM is available for the sort. 261 */ 262 static void 263 set_hw_params(void) 264 { 265 #if defined(SORT_THREADS) 266 size_t ncpusz; 267 #endif 268 size_t pages, psize, psz, pszsz; 269 270 pages = psize = 0; 271 #if defined(SORT_THREADS) 272 ncpu = 1; 273 ncpusz = sizeof(size_t); 274 #endif 275 psz = pszsz = sizeof(size_t); 276 277 if (sysctlbyname("vm.stats.vm.v_free_count", &pages, &psz, 278 NULL, 0) < 0) { 279 perror("vm.stats.vm.v_free_count"); 280 return; 281 } 282 if (sysctlbyname("vm.stats.vm.v_page_size", &psize, &pszsz, 283 NULL, 0) < 0) { 284 perror("vm.stats.vm.v_page_size"); 285 return; 286 } 287 #if defined(SORT_THREADS) 288 if (sysctlbyname("hw.ncpu", &ncpu, &ncpusz, 289 NULL, 0) < 0) 290 ncpu = 1; 291 else if(ncpu > 32) 292 ncpu = 32; 293 294 nthreads = ncpu; 295 #endif 296 297 free_memory = (unsigned long long) pages * (unsigned long long) psize; 298 available_free_memory = (free_memory * 9) / 10; 299 } 300 301 /* 302 * Convert "plain" symbol to wide symbol, with default value. 303 */ 304 static void 305 conv_mbtowc(wchar_t *wc, const char *c, const wchar_t def) 306 { 307 308 if (wc && c) { 309 int res; 310 311 res = mbtowc(wc, c, MB_CUR_MAX); 312 if (res < 1) 313 *wc = def; 314 } 315 } 316 317 /* 318 * Set current locale symbols. 319 */ 320 static void 321 set_locale(void) 322 { 323 struct lconv *lc; 324 const char *locale; 325 326 setlocale(LC_ALL, ""); 327 328 lc = localeconv(); 329 330 if (lc) { 331 /* obtain LC_NUMERIC info */ 332 /* Convert to wide char form */ 333 conv_mbtowc(&symbol_decimal_point, lc->decimal_point, 334 symbol_decimal_point); 335 conv_mbtowc(&symbol_thousands_sep, lc->thousands_sep, 336 symbol_thousands_sep); 337 conv_mbtowc(&symbol_positive_sign, lc->positive_sign, 338 symbol_positive_sign); 339 conv_mbtowc(&symbol_negative_sign, lc->negative_sign, 340 symbol_negative_sign); 341 } 342 343 if (getenv("GNUSORT_NUMERIC_COMPATIBILITY")) 344 gnusort_numeric_compatibility = true; 345 346 locale = setlocale(LC_COLLATE, NULL); 347 348 if (locale) { 349 char *tmpl; 350 const char *cclocale; 351 352 tmpl = sort_strdup(locale); 353 cclocale = setlocale(LC_COLLATE, "C"); 354 if (cclocale && !strcmp(cclocale, tmpl)) 355 byte_sort = true; 356 else { 357 const char *pclocale; 358 359 pclocale = setlocale(LC_COLLATE, "POSIX"); 360 if (pclocale && !strcmp(pclocale, tmpl)) 361 byte_sort = true; 362 } 363 setlocale(LC_COLLATE, tmpl); 364 sort_free(tmpl); 365 } 366 } 367 368 /* 369 * Set directory temporary files. 370 */ 371 static void 372 set_tmpdir(void) 373 { 374 char *td; 375 376 td = getenv("TMPDIR"); 377 if (td != NULL) 378 tmpdir = sort_strdup(td); 379 } 380 381 /* 382 * Parse -S option. 383 */ 384 static unsigned long long 385 parse_memory_buffer_value(const char *value) 386 { 387 388 if (value == NULL) 389 return (available_free_memory); 390 else { 391 char *endptr; 392 unsigned long long membuf; 393 394 endptr = NULL; 395 errno = 0; 396 membuf = strtoll(value, &endptr, 10); 397 398 if (errno != 0) { 399 warn(getstr(4)); 400 membuf = available_free_memory; 401 } else { 402 switch (*endptr){ 403 case 'Y': 404 membuf *= 1024; 405 /* FALLTHROUGH */ 406 case 'Z': 407 membuf *= 1024; 408 /* FALLTHROUGH */ 409 case 'E': 410 membuf *= 1024; 411 /* FALLTHROUGH */ 412 case 'P': 413 membuf *= 1024; 414 /* FALLTHROUGH */ 415 case 'T': 416 membuf *= 1024; 417 /* FALLTHROUGH */ 418 case 'G': 419 membuf *= 1024; 420 /* FALLTHROUGH */ 421 case 'M': 422 membuf *= 1024; 423 /* FALLTHROUGH */ 424 case '\0': 425 case 'K': 426 membuf *= 1024; 427 /* FALLTHROUGH */ 428 case 'b': 429 break; 430 case '%': 431 membuf = (available_free_memory * membuf) / 432 100; 433 break; 434 default: 435 fprintf(stderr, "%s: %s\n", strerror(EINVAL), 436 optarg); 437 membuf = available_free_memory; 438 } 439 } 440 return (membuf); 441 } 442 } 443 444 /* 445 * Signal handler that clears the temporary files. 446 */ 447 static void 448 sig_handler(int sig, siginfo_t *siginfo, void *context) 449 { 450 451 sig = sig; 452 siginfo = siginfo; 453 context = context; 454 clear_tmp_files(); 455 exit(-1); 456 } 457 458 /* 459 * Set signal handler on panic signals. 460 */ 461 static void 462 set_signal_handler(void) 463 { 464 struct sigaction sa; 465 466 memset(&sa, 0, sizeof(sa)); 467 sa.sa_sigaction = &sig_handler; 468 sa.sa_flags = SA_SIGINFO; 469 470 if (sigaction(SIGTERM, &sa, NULL) < 0) { 471 perror("sigaction"); 472 return; 473 } 474 if (sigaction(SIGHUP, &sa, NULL) < 0) { 475 perror("sigaction"); 476 return; 477 } 478 if (sigaction(SIGINT, &sa, NULL) < 0) { 479 perror("sigaction"); 480 return; 481 } 482 if (sigaction(SIGQUIT, &sa, NULL) < 0) { 483 perror("sigaction"); 484 return; 485 } 486 if (sigaction(SIGABRT, &sa, NULL) < 0) { 487 perror("sigaction"); 488 return; 489 } 490 if (sigaction(SIGBUS, &sa, NULL) < 0) { 491 perror("sigaction"); 492 return; 493 } 494 if (sigaction(SIGSEGV, &sa, NULL) < 0) { 495 perror("sigaction"); 496 return; 497 } 498 if (sigaction(SIGUSR1, &sa, NULL) < 0) { 499 perror("sigaction"); 500 return; 501 } 502 if (sigaction(SIGUSR2, &sa, NULL) < 0) { 503 perror("sigaction"); 504 return; 505 } 506 } 507 508 /* 509 * Print "unknown" message and exit with status 2. 510 */ 511 static void 512 unknown(const char *what) 513 { 514 515 errx(2, getstr(3), what); 516 } 517 518 /* 519 * Check whether contradictory input options are used. 520 */ 521 static void 522 check_mutually_exclusive_flags(char c, bool *mef_flags) 523 { 524 int fo_index, mec; 525 bool found_others, found_this; 526 527 found_others = found_this =false; 528 fo_index = 0; 529 530 for (int i = 0; i < NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS; i++) { 531 mec = mutually_exclusive_flags[i]; 532 533 if (mec != c) { 534 if (mef_flags[i]) { 535 if (found_this) 536 errx(1, getstr(1), c, mec); 537 found_others = true; 538 fo_index = i; 539 } 540 } else { 541 if (found_others) 542 errx(1, getstr(1), c, 543 mutually_exclusive_flags[fo_index]); 544 mef_flags[i] = true; 545 found_this = true; 546 } 547 } 548 } 549 550 /* 551 * Initialise sort opts data. 552 */ 553 static void 554 set_sort_opts(void) 555 { 556 557 memset(&default_sort_mods_object, 0, 558 sizeof(default_sort_mods_object)); 559 memset(&sort_opts_vals, 0, sizeof(sort_opts_vals)); 560 default_sort_mods_object.func = 561 get_sort_func(&default_sort_mods_object); 562 } 563 564 /* 565 * Set a sort modifier on a sort modifiers object. 566 */ 567 static bool 568 set_sort_modifier(struct sort_mods *sm, int c) 569 { 570 571 if (sm) { 572 switch (c){ 573 case 'b': 574 sm->bflag = true; 575 break; 576 case 'd': 577 sm->dflag = true; 578 break; 579 case 'f': 580 sm->fflag = true; 581 break; 582 case 'g': 583 sm->gflag = true; 584 need_hint = true; 585 break; 586 case 'i': 587 sm->iflag = true; 588 break; 589 case 'R': 590 sm->Rflag = true; 591 need_random = true; 592 break; 593 case 'M': 594 initialise_months(); 595 sm->Mflag = true; 596 need_hint = true; 597 break; 598 case 'n': 599 sm->nflag = true; 600 need_hint = true; 601 print_symbols_on_debug = true; 602 break; 603 case 'r': 604 sm->rflag = true; 605 break; 606 case 'V': 607 sm->Vflag = true; 608 break; 609 case 'h': 610 sm->hflag = true; 611 need_hint = true; 612 print_symbols_on_debug = true; 613 break; 614 default: 615 return false; 616 } 617 sort_opts_vals.complex_sort = true; 618 sm->func = get_sort_func(sm); 619 } 620 return (true); 621 } 622 623 /* 624 * Parse POS in -k option. 625 */ 626 static int 627 parse_pos(const char *s, struct key_specs *ks, bool *mef_flags, bool second) 628 { 629 regmatch_t pmatch[4]; 630 regex_t re; 631 char *c, *f; 632 const char *sregexp = "^([0-9]+)(\\.[0-9]+)?([bdfirMngRhV]+)?$"; 633 size_t len, nmatch; 634 int ret; 635 636 ret = -1; 637 nmatch = 4; 638 c = f = NULL; 639 640 if (regcomp(&re, sregexp, REG_EXTENDED) != 0) 641 return (-1); 642 643 if (regexec(&re, s, nmatch, pmatch, 0) != 0) 644 goto end; 645 646 if (pmatch[0].rm_eo <= pmatch[0].rm_so) 647 goto end; 648 649 if (pmatch[1].rm_eo <= pmatch[1].rm_so) 650 goto end; 651 652 len = pmatch[1].rm_eo - pmatch[1].rm_so; 653 f = sort_malloc((len + 1) * sizeof(char)); 654 655 strncpy(f, s + pmatch[1].rm_so, len); 656 f[len] = '\0'; 657 658 if (second) { 659 errno = 0; 660 ks->f2 = (size_t) strtoul(f, NULL, 10); 661 if (errno != 0) 662 errx(2, "%s: -k", strerror(errno)); 663 if (ks->f2 == 0) { 664 warn(getstr(5)); 665 goto end; 666 } 667 } else { 668 errno = 0; 669 ks->f1 = (size_t) strtoul(f, NULL, 10); 670 if (errno != 0) 671 errx(2, "%s: -k", strerror(errno)); 672 if (ks->f1 == 0) { 673 warn(getstr(5)); 674 goto end; 675 } 676 } 677 678 if (pmatch[2].rm_eo > pmatch[2].rm_so) { 679 len = pmatch[2].rm_eo - pmatch[2].rm_so - 1; 680 c = sort_malloc((len + 1) * sizeof(char)); 681 682 strncpy(c, s + pmatch[2].rm_so + 1, len); 683 c[len] = '\0'; 684 685 if (second) { 686 errno = 0; 687 ks->c2 = (size_t) strtoul(c, NULL, 10); 688 if (errno != 0) 689 errx(2, "%s: -k", strerror(errno)); 690 } else { 691 errno = 0; 692 ks->c1 = (size_t) strtoul(c, NULL, 10); 693 if (errno != 0) 694 errx(2, "%s: -k", strerror(errno)); 695 if (ks->c1 == 0) { 696 warn(getstr(6)); 697 goto end; 698 } 699 } 700 } else { 701 if (second) 702 ks->c2 = 0; 703 else 704 ks->c1 = 1; 705 } 706 707 if (pmatch[3].rm_eo > pmatch[3].rm_so) { 708 regoff_t i = 0; 709 710 for (i = pmatch[3].rm_so; i < pmatch[3].rm_eo; i++) { 711 check_mutually_exclusive_flags(s[i], mef_flags); 712 if (s[i] == 'b') { 713 if (second) 714 ks->pos2b = true; 715 else 716 ks->pos1b = true; 717 } else if (!set_sort_modifier(&(ks->sm), s[i])) 718 goto end; 719 } 720 } 721 722 ret = 0; 723 724 end: 725 726 if (c) 727 sort_free(c); 728 if (f) 729 sort_free(f); 730 regfree(&re); 731 732 return (ret); 733 } 734 735 /* 736 * Parse -k option value. 737 */ 738 static int 739 parse_k(const char *s, struct key_specs *ks) 740 { 741 int ret = -1; 742 bool mef_flags[NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS] = 743 { false, false, false, false, false, false }; 744 745 if (s && *s) { 746 char *sptr; 747 748 sptr = strchr(s, ','); 749 if (sptr) { 750 size_t size1; 751 char *pos1, *pos2; 752 753 size1 = sptr - s; 754 755 if (size1 < 1) 756 return (-1); 757 pos1 = sort_malloc((size1 + 1) * sizeof(char)); 758 759 strncpy(pos1, s, size1); 760 pos1[size1] = '\0'; 761 762 ret = parse_pos(pos1, ks, mef_flags, false); 763 764 sort_free(pos1); 765 if (ret < 0) 766 return (ret); 767 768 pos2 = sort_strdup(sptr + 1); 769 ret = parse_pos(pos2, ks, mef_flags, true); 770 sort_free(pos2); 771 } else 772 ret = parse_pos(s, ks, mef_flags, false); 773 } 774 775 return (ret); 776 } 777 778 /* 779 * Parse POS in +POS -POS option. 780 */ 781 static int 782 parse_pos_obs(const char *s, int *nf, int *nc, char* sopts) 783 { 784 regex_t re; 785 regmatch_t pmatch[4]; 786 char *c, *f; 787 const char *sregexp = "^([0-9]+)(\\.[0-9]+)?([A-Za-z]+)?$"; 788 int ret; 789 size_t len, nmatch; 790 791 ret = -1; 792 nmatch = 4; 793 c = f = NULL; 794 *nc = *nf = 0; 795 796 if (regcomp(&re, sregexp, REG_EXTENDED) != 0) 797 return (-1); 798 799 if (regexec(&re, s, nmatch, pmatch, 0) != 0) 800 goto end; 801 802 if (pmatch[0].rm_eo <= pmatch[0].rm_so) 803 goto end; 804 805 if (pmatch[1].rm_eo <= pmatch[1].rm_so) 806 goto end; 807 808 len = pmatch[1].rm_eo - pmatch[1].rm_so; 809 f = sort_malloc((len + 1) * sizeof(char)); 810 811 strncpy(f, s + pmatch[1].rm_so, len); 812 f[len] = '\0'; 813 814 errno = 0; 815 *nf = (size_t) strtoul(f, NULL, 10); 816 if (errno != 0) 817 errx(2, getstr(11)); 818 819 if (pmatch[2].rm_eo > pmatch[2].rm_so) { 820 len = pmatch[2].rm_eo - pmatch[2].rm_so - 1; 821 c = sort_malloc((len + 1) * sizeof(char)); 822 823 strncpy(c, s + pmatch[2].rm_so + 1, len); 824 c[len] = '\0'; 825 826 errno = 0; 827 *nc = (size_t) strtoul(c, NULL, 10); 828 if (errno != 0) 829 errx(2, getstr(11)); 830 } 831 832 if (pmatch[3].rm_eo > pmatch[3].rm_so) { 833 834 len = pmatch[3].rm_eo - pmatch[3].rm_so; 835 836 strncpy(sopts, s + pmatch[3].rm_so, len); 837 sopts[len] = '\0'; 838 } 839 840 ret = 0; 841 842 end: 843 if (c) 844 sort_free(c); 845 if (f) 846 sort_free(f); 847 regfree(&re); 848 849 return (ret); 850 } 851 852 /* 853 * "Translate" obsolete +POS1 -POS2 syntax into new -kPOS1,POS2 syntax 854 */ 855 void 856 fix_obsolete_keys(int *argc, char **argv) 857 { 858 char sopt[129]; 859 860 for (int i = 1; i < *argc; i++) { 861 char *arg1; 862 863 arg1 = argv[i]; 864 865 if (strlen(arg1) > 1 && arg1[0] == '+') { 866 int c1, f1; 867 char sopts1[128]; 868 869 sopts1[0] = 0; 870 c1 = f1 = 0; 871 872 if (parse_pos_obs(arg1 + 1, &f1, &c1, sopts1) < 0) 873 continue; 874 else { 875 f1 += 1; 876 c1 += 1; 877 if (i + 1 < *argc) { 878 char *arg2 = argv[i + 1]; 879 880 if (strlen(arg2) > 1 && 881 arg2[0] == '-') { 882 int c2, f2; 883 char sopts2[128]; 884 885 sopts2[0] = 0; 886 c2 = f2 = 0; 887 888 if (parse_pos_obs(arg2 + 1, 889 &f2, &c2, sopts2) >= 0) { 890 if (c2 > 0) 891 f2 += 1; 892 sprintf(sopt, "-k%d.%d%s,%d.%d%s", 893 f1, c1, sopts1, f2, c2, sopts2); 894 argv[i] = sort_strdup(sopt); 895 for (int j = i + 1; j + 1 < *argc; j++) 896 argv[j] = argv[j + 1]; 897 *argc -= 1; 898 continue; 899 } 900 } 901 } 902 sprintf(sopt, "-k%d.%d", f1, c1); 903 argv[i] = sort_strdup(sopt); 904 } 905 } 906 } 907 } 908 909 /* 910 * Set random seed 911 */ 912 static void 913 set_random_seed(void) 914 { 915 if (need_random) { 916 917 if (strcmp(random_source, DEFAULT_RANDOM_SORT_SEED_FILE) == 0) { 918 FILE* fseed; 919 MD5_CTX ctx; 920 char rsd[MAX_DEFAULT_RANDOM_SEED_DATA_SIZE]; 921 size_t sz = 0; 922 923 fseed = openfile(random_source, "r"); 924 while (!feof(fseed)) { 925 int cr; 926 927 cr = fgetc(fseed); 928 if (cr == EOF) 929 break; 930 931 rsd[sz++] = (char) cr; 932 933 if (sz >= MAX_DEFAULT_RANDOM_SEED_DATA_SIZE) 934 break; 935 } 936 937 closefile(fseed, random_source); 938 939 MD5Init(&ctx); 940 MD5Update(&ctx, rsd, sz); 941 942 random_seed = MD5End(&ctx, NULL); 943 random_seed_size = strlen(random_seed); 944 945 } else { 946 MD5_CTX ctx; 947 char *b; 948 949 MD5Init(&ctx); 950 b = MD5File(random_source, NULL); 951 if (b == NULL) 952 err(2, NULL); 953 954 random_seed = b; 955 random_seed_size = strlen(b); 956 } 957 958 MD5Init(&md5_ctx); 959 if(random_seed_size>0) { 960 MD5Update(&md5_ctx, random_seed, random_seed_size); 961 } 962 } 963 } 964 965 /* 966 * Main function. 967 */ 968 int 969 main(int argc, char **argv) 970 { 971 char *outfile, *real_outfile; 972 int c, result; 973 bool mef_flags[NUMBER_OF_MUTUALLY_EXCLUSIVE_FLAGS] = 974 { false, false, false, false, false, false }; 975 976 result = 0; 977 outfile = sort_strdup("-"); 978 real_outfile = NULL; 979 980 struct sort_mods *sm = &default_sort_mods_object; 981 982 init_tmp_files(); 983 984 set_signal_handler(); 985 986 set_hw_params(); 987 set_locale(); 988 set_tmpdir(); 989 set_sort_opts(); 990 991 #if 0 992 { 993 static int counter = 0; 994 char fn[128]; 995 sprintf(fn, "/var/tmp/debug.sort.%d", counter++); 996 FILE* f = fopen(fn, "w"); 997 fprintf(f, ">>sort>>"); 998 for (int i = 0; i < argc; i++) { 999 fprintf(f, "<%s>", argv[i]); 1000 } 1001 fprintf(f, "<<sort<<\n"); 1002 fclose(f); 1003 } 1004 #endif 1005 1006 fix_obsolete_keys(&argc, argv); 1007 1008 while (((c = getopt_long(argc, argv, OPTIONS, long_options, NULL)) 1009 != -1)) { 1010 1011 check_mutually_exclusive_flags(c, mef_flags); 1012 1013 if (!set_sort_modifier(sm, c)) { 1014 1015 switch (c) { 1016 case 'c': 1017 sort_opts_vals.cflag = true; 1018 if (optarg) { 1019 if (!strcmp(optarg, "diagnose-first")) 1020 ; 1021 else if (!strcmp(optarg, "silent") || 1022 !strcmp(optarg, "quiet")) 1023 sort_opts_vals.csilentflag = true; 1024 else if (*optarg) 1025 unknown(optarg); 1026 } 1027 break; 1028 case 'C': 1029 sort_opts_vals.cflag = true; 1030 sort_opts_vals.csilentflag = true; 1031 break; 1032 case 'k': 1033 { 1034 sort_opts_vals.complex_sort = true; 1035 sort_opts_vals.kflag = true; 1036 1037 keys_num++; 1038 keys = sort_realloc(keys, keys_num * 1039 sizeof(struct key_specs)); 1040 memset(&(keys[keys_num - 1]), 0, 1041 sizeof(struct key_specs)); 1042 1043 if (parse_k(optarg, &(keys[keys_num - 1])) 1044 < 0) { 1045 errx(2, "%s: -k %s\n", 1046 strerror(EINVAL), optarg); 1047 } 1048 1049 break; 1050 } 1051 case 'm': 1052 sort_opts_vals.mflag = true; 1053 break; 1054 case 'o': 1055 outfile = sort_realloc(outfile, sizeof(char) * 1056 (strlen(optarg) + 1)); 1057 strlcpy(outfile, optarg, (strlen(optarg) + 1)); 1058 break; 1059 case 's': 1060 sort_opts_vals.sflag = true; 1061 break; 1062 case 'S': 1063 available_free_memory = 1064 parse_memory_buffer_value(optarg); 1065 break; 1066 case 'T': 1067 tmpdir = sort_strdup(optarg); 1068 break; 1069 case 't': 1070 if (strlen(optarg) > 1) { 1071 if (strcmp(optarg, "\\0")) { 1072 errx(2, "%s: %s\n", 1073 strerror(EINVAL), optarg); 1074 } 1075 *optarg = 0; 1076 } 1077 sort_opts_vals.tflag = true; 1078 sort_opts_vals.field_sep = btowc(optarg[0]); 1079 if (sort_opts_vals.field_sep == WEOF) { 1080 errno = EINVAL; 1081 err(2, NULL); 1082 } 1083 if (!gnusort_numeric_compatibility) { 1084 if (symbol_decimal_point == sort_opts_vals.field_sep) 1085 symbol_decimal_point = WEOF; 1086 if (symbol_thousands_sep == sort_opts_vals.field_sep) 1087 symbol_thousands_sep = WEOF; 1088 if (symbol_negative_sign == sort_opts_vals.field_sep) 1089 symbol_negative_sign = WEOF; 1090 if (symbol_positive_sign == sort_opts_vals.field_sep) 1091 symbol_positive_sign = WEOF; 1092 } 1093 break; 1094 case 'u': 1095 sort_opts_vals.uflag = true; 1096 /* stable sort for the correct unique val */ 1097 sort_opts_vals.sflag = true; 1098 break; 1099 case 'z': 1100 sort_opts_vals.zflag = true; 1101 break; 1102 case SORT_OPT: 1103 if (optarg) { 1104 if (!strcmp(optarg, "general-numeric")) 1105 set_sort_modifier(sm, 'g'); 1106 else if (!strcmp(optarg, "human-numeric")) 1107 set_sort_modifier(sm, 'h'); 1108 else if (!strcmp(optarg, "numeric")) 1109 set_sort_modifier(sm, 'n'); 1110 else if (!strcmp(optarg, "month")) 1111 set_sort_modifier(sm, 'M'); 1112 else if (!strcmp(optarg, "random")) 1113 set_sort_modifier(sm, 'R'); 1114 else 1115 unknown(optarg); 1116 } 1117 break; 1118 #if defined(SORT_THREADS) 1119 case NTHREADS_OPT: 1120 nthreads = (size_t)(atoi(optarg)); 1121 if (nthreads < 1) 1122 nthreads = 1; 1123 if (nthreads > 1024) 1124 nthreads = 1024; 1125 break; 1126 #endif 1127 case QSORT_OPT: 1128 sort_opts_vals.sort_method = SORT_QSORT; 1129 break; 1130 case MERGESORT_OPT: 1131 sort_opts_vals.sort_method = SORT_MERGESORT; 1132 break; 1133 case HEAPSORT_OPT: 1134 sort_opts_vals.sort_method = SORT_HEAPSORT; 1135 break; 1136 case RADIXSORT_OPT: 1137 sort_opts_vals.sort_method = SORT_RADIXSORT; 1138 break; 1139 case RANDOMSOURCE_OPT: 1140 random_source = strdup(optarg); 1141 break; 1142 case COMPRESSPROGRAM_OPT: 1143 compress_program = strdup(optarg); 1144 break; 1145 case FF_OPT: 1146 read_fns_from_file0(optarg); 1147 break; 1148 case BS_OPT: 1149 { 1150 errno = 0; 1151 long mof = strtol(optarg, NULL, 10); 1152 if (errno != 0) 1153 errx(2, "--batch-size: %s", 1154 strerror(errno)); 1155 if (mof >= 2) 1156 max_open_files = (size_t) mof + 1; 1157 } 1158 break; 1159 case VERSION_OPT: 1160 printf("%s\n", VERSION); 1161 exit(EXIT_SUCCESS); 1162 /* NOTREACHED */ 1163 break; 1164 case DEBUG_OPT: 1165 debug_sort = true; 1166 break; 1167 case HELP_OPT: 1168 usage(false); 1169 /* NOTREACHED */ 1170 break; 1171 default: 1172 usage(true); 1173 /* NOTREACHED */ 1174 } 1175 } 1176 } 1177 1178 argc -= optind; 1179 argv += optind; 1180 1181 #ifndef WITHOUT_NLS 1182 catalog = catopen("sort", NL_CAT_LOCALE); 1183 #endif 1184 1185 if (sort_opts_vals.cflag && sort_opts_vals.mflag) 1186 errx(1, getstr(1), 'm', 'c'); 1187 1188 #ifndef WITHOUT_NLS 1189 catclose(catalog); 1190 #endif 1191 1192 if (keys_num == 0) { 1193 keys_num = 1; 1194 keys = sort_realloc(keys, sizeof(struct key_specs)); 1195 memset(&(keys[0]), 0, sizeof(struct key_specs)); 1196 keys[0].c1 = 1; 1197 keys[0].pos1b = default_sort_mods->bflag; 1198 keys[0].pos2b = default_sort_mods->bflag; 1199 memcpy(&(keys[0].sm), default_sort_mods, 1200 sizeof(struct sort_mods)); 1201 } 1202 1203 for (size_t i = 0; i < keys_num; i++) { 1204 struct key_specs *ks; 1205 1206 ks = &(keys[i]); 1207 1208 if (sort_modifier_empty(&(ks->sm)) && !(ks->pos1b) && 1209 !(ks->pos2b)) { 1210 ks->pos1b = sm->bflag; 1211 ks->pos2b = sm->bflag; 1212 memcpy(&(ks->sm), sm, sizeof(struct sort_mods)); 1213 } 1214 1215 ks->sm.func = get_sort_func(&(ks->sm)); 1216 } 1217 1218 if (argc_from_file0 >= 0) { 1219 argc = argc_from_file0; 1220 argv = argv_from_file0; 1221 } 1222 1223 if (debug_sort) { 1224 #if defined(SORT_THREADS) 1225 nthreads = 1; 1226 #endif 1227 printf("Using collate rules of %s locale\n", 1228 setlocale(LC_COLLATE, NULL)); 1229 if (byte_sort) 1230 printf("Byte sort is used\n"); 1231 if (print_symbols_on_debug) { 1232 printf("Decimal Point: <%lc>\n", symbol_decimal_point); 1233 if (symbol_thousands_sep) 1234 printf("Thousands separator: <%lc>\n", 1235 symbol_thousands_sep); 1236 printf("Positive sign: <%lc>\n", symbol_positive_sign); 1237 printf("Negative sign: <%lc>\n", symbol_negative_sign); 1238 } 1239 } 1240 1241 set_random_seed(); 1242 1243 /* Case when the outfile equals one of the input files: */ 1244 if (strcmp(outfile, "-")) { 1245 1246 for(int i = 0; i < argc; ++i) { 1247 if (strcmp(argv[i], outfile) == 0) { 1248 real_outfile = sort_strdup(outfile); 1249 for(;;) { 1250 char* tmp = sort_malloc(strlen(outfile) + 1251 strlen(".tmp") + 1); 1252 1253 strcpy(tmp, outfile); 1254 strcpy(tmp + strlen(tmp), ".tmp"); 1255 sort_free(outfile); 1256 outfile = tmp; 1257 if (access(outfile, F_OK) < 0) 1258 break; 1259 } 1260 tmp_file_atexit(outfile); 1261 } 1262 } 1263 } 1264 1265 if (!sort_opts_vals.cflag && !sort_opts_vals.mflag) { 1266 struct file_list fl; 1267 struct sort_list list; 1268 1269 sort_list_init(&list); 1270 file_list_init(&fl, true); 1271 1272 if (argc < 1) 1273 procfile("-", &list, &fl); 1274 else { 1275 while (argc > 0) { 1276 procfile(*argv, &list, &fl); 1277 --argc; 1278 ++argv; 1279 } 1280 } 1281 1282 if (fl.count < 1) 1283 sort_list_to_file(&list, outfile); 1284 else { 1285 if (list.count > 0) { 1286 char *flast = new_tmp_file_name(); 1287 1288 sort_list_to_file(&list, flast); 1289 file_list_add(&fl, flast, false); 1290 } 1291 merge_files(&fl, outfile); 1292 } 1293 1294 file_list_clean(&fl); 1295 1296 /* 1297 * We are about to exit the program, so we can ignore 1298 * the clean-up for speed 1299 * 1300 * sort_list_clean(&list); 1301 */ 1302 1303 } else if (sort_opts_vals.cflag) { 1304 result = (argc == 0) ? (check("-")) : (check(*argv)); 1305 } else if (sort_opts_vals.mflag) { 1306 struct file_list fl; 1307 1308 file_list_init(&fl, false); 1309 file_list_populate(&fl, argc, argv, true); 1310 merge_files(&fl, outfile); 1311 file_list_clean(&fl); 1312 } 1313 1314 if (real_outfile) { 1315 unlink(real_outfile); 1316 if (rename(outfile, real_outfile) < 0) 1317 err(2, NULL); 1318 sort_free(real_outfile); 1319 } 1320 1321 sort_free(outfile); 1322 1323 return (result); 1324 } 1325