coreutils/sort.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * SuS3 compliant sort implementation for busybox
   4  *
   5  * Copyright (C) 2004 by Rob Landley <rob@landley.net>
   6  *
   7  * MAINTAINER: Rob Landley <rob@landley.net>
   8  *
   9  * Licensed under GPLv2 or later, see file LICENSE in this source tree.
  10  *
  11  * See SuS3 sort standard at:
  12  * http://www.opengroup.org/onlinepubs/007904975/utilities/sort.html
  13  */
  14 //config:config SORT
  15 //config:       bool "sort (8.1 kb)"
  16 //config:       default y
  17 //config:       help
  18 //config:       sort is used to sort lines of text in specified files.
  19 //config:
  20 //config:config FEATURE_SORT_BIG
  21 //config:       bool "Full SuSv3 compliant sort (support -ktcbdfioghM)"
  22 //config:       default y
  23 //config:       depends on SORT
  24 //config:       help
  25 //config:       Without this, sort only supports -rusz, and an integer version
  26 //config:       of -n. Selecting this adds sort keys, floating point support, and
  27 //config:       more. This adds a little over 3k to a nonstatic build on x86.
  28 //config:
  29 //config:       The SuSv3 sort standard is available at:
  30 //config:       http://www.opengroup.org/onlinepubs/007904975/utilities/sort.html
  31 //config:
  32 //config:config FEATURE_SORT_OPTIMIZE_MEMORY
  33 //config:       bool "Use less memory (but might be slower)"
  34 //config:       default n   # defaults to N since we are size-paranoid tribe
  35 //config:       depends on SORT
  36 //config:       help
  37 //config:       Attempt to use less memory (by storing only one copy
  38 //config:       of duplicated lines, and such). Useful if you work on huge files.
  39
  40 //applet:IF_SORT(APPLET_NOEXEC(sort, sort, BB_DIR_USR_BIN, BB_SUID_DROP, sort))
  41
  42 //kbuild:lib-$(CONFIG_SORT) += sort.o
  43
  44 //usage:#define sort_trivial_usage
  45 //usage:       "[-nru"
  46 //usage:        IF_FEATURE_SORT_BIG("ghMcszbdfiokt] [-o FILE] [-k START[.OFS][OPTS][,END[.OFS][OPTS]] [-t CHAR")
  47 //usage:       "] [FILE]..."
  48 //usage:#define sort_full_usage "\n\n"
  49 //usage:       "Sort lines of text\n"
  50 //usage:        IF_FEATURE_SORT_BIG(
  51 //usage:     "\n        -o FILE Output to FILE"
  52 //usage:     "\n        -c      Check whether input is sorted"
  53 //usage:     "\n        -b      Ignore leading blanks"
  54 //usage:     "\n        -f      Ignore case"
  55 //usage:     "\n        -i      Ignore unprintable characters"
  56 //usage:     "\n        -d      Dictionary order (blank or alphanumeric only)"
  57 //usage:        )
  58 //-h, --human-numeric-sort: compare human readable numbers (e.g., 2K 1G)
  59 //usage:     "\n        -n      Sort numbers"
  60 //usage:        IF_FEATURE_SORT_BIG(
  61 //usage:     "\n        -g      General numerical sort"
  62 //usage:     "\n        -h      Sort human readable numbers (2K 1G)"
  63 //usage:     "\n        -M      Sort month"
  64 //usage:     "\n        -V      Sort version"
  65 //usage:     "\n        -t CHAR Field separator"
  66 //usage:     "\n        -k N[,M] Sort by Nth field"
  67 //usage:        )
  68 //usage:     "\n        -r      Reverse sort order"
  69 //usage:     "\n        -s      Stable (don't sort ties alphabetically)"
  70 //usage:     "\n        -u      Suppress duplicate lines"
  71 //usage:     "\n        -z      NUL terminated input and output"
  72 ///////:     "\n        -m      Ignored for GNU compatibility"
  73 ///////:     "\n        -S BUFSZ Ignored for GNU compatibility"
  74 ///////:     "\n        -T TMPDIR Ignored for GNU compatibility"
  75 //usage:
  76 //usage:#define sort_example_usage
  77 //usage:       "$ echo -e \"e\\nf\\nb\\nd\\nc\\na\" | sort\n"
  78 //usage:       "a\n"
  79 //usage:       "b\n"
  80 //usage:       "c\n"
  81 //usage:       "d\n"
  82 //usage:       "e\n"
  83 //usage:       "f\n"
  84 //usage:        IF_FEATURE_SORT_BIG(
  85 //usage:                "$ echo -e \"c 3\\nb 2\\nd 2\" | $SORT -k 2,2n -k 1,1r\n"
  86 //usage:                "d 2\n"
  87 //usage:                "b 2\n"
  88 //usage:                "c 3\n"
  89 //usage:        )
  90 //usage:       ""
  91
  92 #include "libbb.h"
  93
  94 /* These are sort types */
  95 enum {
  96         FLAG_n  = 1 << 0,       /* Numeric sort */
  97         FLAG_g  = 1 << 1,       /* Sort using strtod() */
  98         FLAG_h  = 1 << 2,       /* Sort using strtod(), plus KMGT suffixes */
  99         FLAG_M  = 1 << 3,       /* Sort date */
 100         FLAG_V  = 1 << 4,       /* Sort version */
 101 /* ucsz apply to root level only, not keys.  b at root level implies bb */
 102         FLAG_u  = 1 << 5,       /* Unique */
 103         FLAG_c  = 1 << 6,       /* Check: no output, exit(!ordered) */
 104         FLAG_s  = 1 << 7,       /* Stable sort, no ascii fallback at end */
 105         FLAG_z  = 1 << 8,       /* Input and output is NUL terminated, not \n */
 106 /* These can be applied to search keys, the previous four can't */
 107         FLAG_b  = 1 << 9,       /* Ignore leading blanks */
 108         FLAG_r  = 1 << 10,      /* Reverse */
 109         FLAG_d  = 1 << 11,      /* Ignore !(isalnum()|isspace()) */
 110         FLAG_f  = 1 << 12,      /* Force uppercase */
 111         FLAG_i  = 1 << 13,      /* Ignore !isprint() */
 112         FLAG_m  = 1 << 14,      /* ignored: merge already sorted files; do not sort */
 113         FLAG_S  = 1 << 15,      /* ignored: -S, --buffer-size=SIZE */
 114         FLAG_T  = 1 << 16,      /* ignored: -T, --temporary-directory=DIR */
 115         FLAG_o  = 1 << 17,
 116         FLAG_k  = 1 << 18,
 117         FLAG_t  = 1 << 19,
 118         FLAG_bb = 0x80000000,   /* Ignore trailing blanks  */
 119         FLAG_no_tie_break = 0x40000000,
 120 };
 121
 122 static const char sort_opt_str[] ALIGN1 = "^"
 123                         "nghMVucszbrdfimS:T:o:k:*t:"
 124                         "\0" "o--o:t--t"/*-t, -o: at most one of each*/;
 125 /*
 126  * OPT_STR must not be string literal, needs to have stable address:
 127  * code uses "strchr(OPT_STR,c) - OPT_STR" idiom.
 128  */
 129 #define OPT_STR (sort_opt_str + 1)
 130
 131 #if ENABLE_FEATURE_SORT_BIG
 132 static char key_separator;
 133
 134 static struct sort_key {
 135         struct sort_key *next_key;  /* linked list */
 136         unsigned range[4];          /* start word, start char, end word, end char */
 137         unsigned flags;
 138 } *key_list;
 139
 140
 141 /* This is a NOEXEC applet. Be very careful! */
 142
 143
 144 static char *get_key(char *str, struct sort_key *key, int flags)
 145 {
 146         int start = start; /* for compiler */
 147         int end;
 148         int len, j;
 149         unsigned i;
 150
 151         /* Special case whole string, so we don't have to make a copy */
 152         if (key->range[0] == 1 && !key->range[1] && !key->range[2] && !key->range[3]
 153          && !(flags & (FLAG_b | FLAG_d | FLAG_f | FLAG_i | FLAG_bb))
 154         ) {
 155                 return str;
 156         }
 157
 158         /* Find start of key on first pass, end on second pass */
 159         len = strlen(str);
 160         for (j = 0; j < 2; j++) {
 161                 if (!key->range[2*j])
 162                         end = len;
 163                 /* Loop through fields */
 164                 else {
 165                         unsigned char ch = 0;
 166
 167                         end = 0;
 168                         for (i = 1; i < key->range[2*j] + j; i++) {
 169                                 if (key_separator) {
 170                                         /* Skip body of key and separator */
 171                                         while ((ch = str[end]) != '\0') {
 172                                                         end++;
 173                                                 if (ch == key_separator)
 174                                                         break;
 175                                         }
 176                                 } else {
 177                                         /* Skip leading blanks */
 178                                         while (isspace(str[end]))
 179                                                 end++;
 180                                         /* Skip body of key */
 181                                         while (str[end] != '\0') {
 182                                                 if (isspace(str[end]))
 183                                                         break;
 184                                                 end++;
 185                                         }
 186                                 }
 187                         }
 188                         /* Remove last delim: "abc:def:" => "abc:def" */
 189                         if (j && ch) {
 190                                 //if (str[end-1] != key_separator)
 191                                 //  bb_error_msg(_and_die("BUG! "
 192                                 //  "str[start:%d,end:%d]:'%.*s'",
 193                                 //  start, end, (int)(end-start), &str[start]);
 194                                 end--;
 195                         }
 196                 }
 197                 if (!j) start = end;
 198         }
 199         /* Strip leading whitespace if necessary */
 200         if (flags & FLAG_b)
 201                 /* not using skip_whitespace() for speed */
 202                 while (isspace(str[start])) start++;
 203         /* Strip trailing whitespace if necessary */
 204         if (flags & FLAG_bb)
 205                 while (end > start && isspace(str[end-1])) end--;
 206         /* -kSTART,N.ENDCHAR: honor ENDCHAR (1-based) */
 207         if (key->range[3]) {
 208                 end = key->range[3];
 209                 if (end > len) end = len;
 210         }
 211         /* -kN.STARTCHAR[,...]: honor STARTCHAR (1-based) */
 212         if (key->range[1]) {
 213                 start += key->range[1] - 1;
 214                 if (start > len) start = len;
 215         }
 216         /* Make the copy */
 217         if (end < start)
 218                 end = start;
 219         str = xstrndup(str+start, end-start);
 220         /* Handle -d */
 221         if (flags & FLAG_d) {
 222                 for (start = end = 0; str[end]; end++)
 223                         if (isspace(str[end]) || isalnum(str[end]))
 224                                 str[start++] = str[end];
 225                 str[start] = '\0';
 226         }
 227         /* Handle -i */
 228         if (flags & FLAG_i) {
 229                 for (start = end = 0; str[end]; end++)
 230                         if (isprint_asciionly(str[end]))
 231                                 str[start++] = str[end];
 232                 str[start] = '\0';
 233         }
 234         /* Handle -f */
 235         if (flags & FLAG_f)
 236                 for (i = 0; str[i]; i++)
 237                         str[i] = toupper(str[i]);
 238
 239         return str;
 240 }
 241
 242 static struct sort_key *add_key(void)
 243 {
 244         struct sort_key **pkey = &key_list;
 245         while (*pkey)
 246                 pkey = &((*pkey)->next_key);
 247         return *pkey = xzalloc(sizeof(struct sort_key));
 248 }
 249
 250 #define GET_LINE(fp) \
 251         ((option_mask32 & FLAG_z) \
 252         ? bb_get_chunk_from_file(fp, NULL) \
 253         : xmalloc_fgetline(fp))
 254 #else
 255 #define GET_LINE(fp) xmalloc_fgetline(fp)
 256 #endif
 257
 258 #if ENABLE_FEATURE_SORT_BIG
 259 static int scale_suffix(const char *tail)
 260 {
 261         static const char suffix[] ALIGN1 = "kmgtpezy";
 262         const char *s;
 263         int n;
 264
 265         if (!tail[0])
 266                 return -1;
 267         s = strchr(suffix, tail[0] | 0x20);
 268         if (!s)
 269                 return -1;
 270         n = s - suffix;
 271         if (n != 0 && tail[0] >= 'a')
 272                 return -1; /* mg... not accepted, only MG... */
 273         return n;
 274 }
 275 #endif
 276
 277 /* Iterate through keys list and perform comparisons */
 278 static int compare_keys(const void *xarg, const void *yarg)
 279 {
 280         int flags = option_mask32, retval = 0;
 281         char *x, *y;
 282
 283 #if ENABLE_FEATURE_SORT_BIG
 284         struct sort_key *key;
 285
 286         for (key = key_list; !retval && key; key = key->next_key) {
 287                 flags = key->flags ? key->flags : option_mask32;
 288                 /* Chop out and modify key chunks, handling -dfib */
 289                 x = get_key(*(char **)xarg, key, flags);
 290                 y = get_key(*(char **)yarg, key, flags);
 291 #else
 292         /* This curly bracket serves no purpose but to match the nesting
 293          * level of the for () loop we're not using */
 294         {
 295                 x = *(char **)xarg;
 296                 y = *(char **)yarg;
 297 #endif
 298                 /* Perform actual comparison */
 299                 switch (flags & (FLAG_n | FLAG_g | FLAG_h | FLAG_M | FLAG_V)) {
 300                 default:
 301                         bb_simple_error_msg_and_die("unknown sort type");
 302                         break;
 303 #if defined(HAVE_STRVERSCMP) && HAVE_STRVERSCMP == 1
 304                 case FLAG_V:
 305                         retval = strverscmp(x, y);
 306                         break;
 307 #endif
 308                 /* Ascii sort */
 309                 case 0:
 310 #if ENABLE_LOCALE_SUPPORT
 311                         retval = strcoll(x, y);
 312 #else
 313                         retval = strcmp(x, y);
 314 #endif
 315                         break;
 316 #if ENABLE_FEATURE_SORT_BIG
 317                 case FLAG_g:
 318                 case FLAG_h: {
 319                         char *xx, *yy;
 320 //TODO: needs setlocale(LC_NUMERIC, "C")?
 321                         double dx = strtod(x, &xx);
 322                         double dy = strtod(y, &yy);
 323                         /* not numbers < NaN < -infinity < numbers < +infinity) */
 324                         if (x == xx)
 325                                 retval = (y == yy ? 0 : -1);
 326                         else if (y == yy)
 327                                 retval = 1;
 328                         /* Check for isnan */
 329                         else if (dx != dx)
 330                                 retval = (dy != dy) ? 0 : -1;
 331                         else if (dy != dy)
 332                                 retval = 1;
 333                         else {
 334                                 if (flags & FLAG_h) {
 335                                         int xs = scale_suffix(xx);
 336                                         int ys = scale_suffix(yy);
 337                                         if (xs != ys) {
 338                                                 retval = xs - ys;
 339                                                 break;
 340                                         }
 341                                 }
 342                                 /* Check for infinity.  Could underflow, but it avoids libm. */
 343                                 if (1.0 / dx == 0.0) {
 344                                         if (dx < 0)
 345                                                 retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1;
 346                                         else
 347                                                 retval = (1.0 / dy == 0.0 && dy > 0) ? 0 : 1;
 348                                 } else if (1.0 / dy == 0.0)
 349                                         retval = (dy < 0) ? 1 : -1;
 350                                 else
 351                                         retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 352                         }
 353                         break;
 354                 }
 355                 case FLAG_M: {
 356                         struct tm thyme;
 357                         int dx;
 358                         char *xx, *yy;
 359
 360                         xx = strptime(skip_whitespace(x), "%b", &thyme);
 361                         dx = thyme.tm_mon;
 362                         yy = strptime(skip_whitespace(y), "%b", &thyme);
 363                         if (!xx)
 364                                 retval = (!yy) ? 0 : -1;
 365                         else if (!yy)
 366                                 retval = 1;
 367                         else
 368                                 retval = dx - thyme.tm_mon;
 369                         break;
 370                 }
 371                 /* Full floating point version of -n */
 372                 case FLAG_n: {
 373                         double dx = atof(x);
 374                         double dy = atof(y);
 375                         retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 376                         break;
 377                 }
 378                 } /* switch */
 379                 /* Free key copies. */
 380                 if (x != *(char **)xarg) free(x);
 381                 if (y != *(char **)yarg) free(y);
 382                 /* if (retval) break; - done by for () anyway */
 383 #else
 384                 /* Integer version of -n for tiny systems */
 385                 case FLAG_n:
 386                         retval = atoi(x) - atoi(y);
 387                         break;
 388                 } /* switch */
 389 #endif
 390         } /* for */
 391
 392         if (retval == 0) {
 393                 /* So far lines are "the same" */
 394
 395                 if (option_mask32 & FLAG_s) {
 396                         /* "Stable sort": later line is "greater than",
 397                          * IOW: do not allow qsort() to swap equal lines.
 398                          */
 399                         uint32_t *p32;
 400                         uint32_t x32, y32;
 401                         char *line;
 402                         unsigned len;
 403
 404                         line = *(char**)xarg;
 405                         len = (strlen(line) + 4) & (~3u);
 406                         p32 = (void*)(line + len);
 407                         x32 = *p32;
 408                         line = *(char**)yarg;
 409                         len = (strlen(line) + 4) & (~3u);
 410                         p32 = (void*)(line + len);
 411                         y32 = *p32;
 412
 413                         /* If x > y, 1, else -1 */
 414                         retval = (x32 > y32) * 2 - 1;
 415                         /* Here, -r has no effect! */
 416                         return retval;
 417                 }
 418                 if (!(option_mask32 & FLAG_no_tie_break)) {
 419                         /* fallback sort */
 420                         flags = option_mask32;
 421                         retval = strcmp(*(char **)xarg, *(char **)yarg);
 422                 }
 423         }
 424
 425         if (flags & FLAG_r)
 426                 return -retval;
 427
 428         return retval;
 429 }
 430
 431 #if ENABLE_FEATURE_SORT_BIG
 432 static unsigned str2u(char **str)
 433 {
 434         unsigned long lu;
 435         if (!isdigit((*str)[0]))
 436                 bb_simple_error_msg_and_die("bad field specification");
 437         lu = strtoul(*str, str, 10);
 438         if ((sizeof(long) > sizeof(int) && lu > INT_MAX) || !lu)
 439                 bb_simple_error_msg_and_die("bad field specification");
 440         return lu;
 441 }
 442 #endif
 443
 444 int sort_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 445 int sort_main(int argc UNUSED_PARAM, char **argv)
 446 {
 447         char **lines;
 448         char *str_ignored, *str_o, *str_t;
 449         llist_t *lst_k = NULL;
 450         int i;
 451         int linecount;
 452         unsigned opts;
 453 #if ENABLE_FEATURE_SORT_OPTIMIZE_MEMORY
 454         bool can_drop_dups;
 455         size_t prev_len = 0;
 456         char *prev_line = (char*) "";
 457         /* Postpone optimizing if the input is small, < 16k lines:
 458          * even just free()ing duplicate lines takes time.
 459          */
 460         size_t count_to_optimize_dups = 0x3fff;
 461 #endif
 462
 463         xfunc_error_retval = 2;
 464
 465         /* Parse command line options */
 466         opts = getopt32(argv,
 467                         sort_opt_str,
 468                         &str_ignored, &str_ignored, &str_o, &lst_k, &str_t
 469         );
 470 #if ENABLE_FEATURE_SORT_OPTIMIZE_MEMORY
 471         /* Can drop dups only if -u but no "complicating" options,
 472          * IOW: if we do a full line compares. Safe options:
 473          * -o FILE Output to FILE
 474          * -z      Lines are terminated by NUL, not newline
 475          * -r      Reverse sort order
 476          * -s      Stable (don't sort ties alphabetically)
 477          * Not sure about these:
 478          * -b      Ignore leading blanks
 479          * -f      Ignore case
 480          * -i      Ignore unprintable characters
 481          * -d      Dictionary order (blank or alphanumeric only)
 482          * -n      Sort numbers
 483          * -g      General numerical sort
 484          * -M      Sort month
 485          */
 486         can_drop_dups = ((opts & ~(FLAG_o|FLAG_z|FLAG_r|FLAG_s)) == FLAG_u);
 487         /* Stable sort needs every line to be uniquely allocated,
 488          * disable optimization to reuse strings:
 489          */
 490         if (opts & FLAG_s)
 491                 count_to_optimize_dups = (size_t)-1L;
 492 #endif
 493         /* global b strips leading and trailing spaces */
 494         if (opts & FLAG_b)
 495                 option_mask32 |= FLAG_bb;
 496 #if ENABLE_FEATURE_SORT_BIG
 497         if (opts & FLAG_t) {
 498                 if (!str_t[0] || str_t[1])
 499                         bb_simple_error_msg_and_die("bad -t parameter");
 500                 key_separator = str_t[0];
 501         }
 502         /* note: below this point we use option_mask32, not opts,
 503          * since that reduces register pressure and makes code smaller */
 504
 505         /* Parse sort key */
 506         while (lst_k) {
 507                 enum {
 508                         FLAG_allowed_for_k =
 509                                 FLAG_n | /* Numeric sort */
 510                                 FLAG_g | /* Sort using strtod() */
 511                                 FLAG_h | /* Sort using strtod(), plus KMGT suffixes */
 512                                 FLAG_M | /* Sort date */
 513                                 FLAG_b | /* Ignore leading blanks */
 514                                 FLAG_r | /* Reverse */
 515                                 FLAG_d | /* Ignore !(isalnum()|isspace()) */
 516                                 FLAG_f | /* Force uppercase */
 517                                 FLAG_i | /* Ignore !isprint() */
 518                         0
 519                 };
 520                 struct sort_key *key = add_key();
 521                 char *str_k = llist_pop(&lst_k);
 522
 523                 i = 0; /* i==0 before comma, 1 after (-k3,6) */
 524                 while (*str_k) {
 525                         /* Start of range */
 526                         /* Cannot use bb_strtou - suffix can be a letter */
 527                         key->range[2*i] = str2u(&str_k);
 528                         if (*str_k == '.') {
 529                                 str_k++;
 530                                 key->range[2*i+1] = str2u(&str_k);
 531                         }
 532                         while (*str_k) {
 533                                 int flag;
 534                                 const char *idx;
 535
 536                                 if (*str_k == ',' && !i++) {
 537                                         str_k++;
 538                                         break;
 539                                 } /* no else needed: fall through to syntax error
 540                                         because comma isn't in OPT_STR */
 541                                 idx = strchr(OPT_STR, *str_k);
 542                                 if (!idx)
 543                                         bb_simple_error_msg_and_die("unknown key option");
 544                                 flag = 1 << (idx - OPT_STR);
 545                                 if (flag & ~FLAG_allowed_for_k)
 546                                         bb_simple_error_msg_and_die("unknown sort type");
 547                                 /* b after ',' means strip _trailing_ space */
 548                                 if (i && flag == FLAG_b)
 549                                         flag = FLAG_bb;
 550                                 key->flags |= flag;
 551                                 str_k++;
 552                         }
 553                 }
 554         }
 555 #endif
 556
 557         /* Open input files and read data */
 558         argv += optind;
 559         if (!*argv)
 560                 *--argv = (char*)"-";
 561         linecount = 0;
 562         lines = NULL;
 563         do {
 564                 /* coreutils 6.9 compat: abort on first open error,
 565                  * do not continue to next file: */
 566                 FILE *fp = xfopen_stdin(*argv);
 567                 for (;;) {
 568                         char *line = GET_LINE(fp);
 569                         if (!line)
 570                                 break;
 571
 572 #if ENABLE_FEATURE_SORT_OPTIMIZE_MEMORY
 573                         if (count_to_optimize_dups != 0)
 574                                 count_to_optimize_dups--;
 575                         if (count_to_optimize_dups == 0) {
 576                                 size_t len;
 577                                 char *new_line;
 578
 579                                 /* On kernel/linux/arch/ *.[ch] files,
 580                                  * this reduces memory usage by 6%.
 581                                  *  yes | head -99999999 | sort
 582                                  * goes down from 1900Mb to 380 Mb.
 583                                  */
 584                                 len = strlen(line);
 585                                 if (len <= prev_len) {
 586                                         new_line = prev_line + (prev_len - len);
 587                                         if (strcmp(line, new_line) == 0) {
 588                                                 /* it's a tail of the prev line */
 589                                                 if (can_drop_dups && prev_len == len) {
 590                                                         /* it's identical to prev line */
 591                                                         free(line);
 592                                                         continue;
 593                                                 }
 594                                                 free(line);
 595                                                 line = new_line;
 596                                                 /* continue using longer prev_line
 597                                                  * for future tail tests.
 598                                                  */
 599                                                 goto skip;
 600                                         }
 601                                 }
 602                                 prev_len = len;
 603                                 prev_line = line;
 604  skip: ;
 605                         }
 606 #else
 607 //TODO: lighter version which only drops total dups if can_drop_dups == true
 608 #endif
 609                         lines = xrealloc_vector(lines, 6, linecount);
 610                         lines[linecount++] = line;
 611                 }
 612                 fclose_if_not_stdin(fp);
 613         } while (*++argv);
 614
 615 #if ENABLE_FEATURE_SORT_BIG
 616         /* If no key, perform alphabetic sort */
 617         if (!key_list)
 618                 add_key()->range[0] = 1;
 619         /* Handle -c */
 620         if (option_mask32 & FLAG_c) {
 621                 int j = (option_mask32 & FLAG_u) ? -1 : 0;
 622                 for (i = 1; i < linecount; i++) {
 623                         if (compare_keys(&lines[i-1], &lines[i]) > j) {
 624                                 fprintf(stderr, "Check line %u\n", i);
 625                                 return EXIT_FAILURE;
 626                         }
 627                 }
 628                 return EXIT_SUCCESS;
 629         }
 630 #endif
 631
 632         /* For stable sort, store original line position beyond terminating NUL */
 633         if (option_mask32 & FLAG_s) {
 634                 for (i = 0; i < linecount; i++) {
 635                         uint32_t *p32;
 636                         char *line;
 637                         unsigned len;
 638
 639                         line = lines[i];
 640                         len = (strlen(line) + 4) & (~3u);
 641                         lines[i] = line = xrealloc(line, len + 4);
 642                         p32 = (void*)(line + len);
 643                         *p32 = i;
 644                 }
 645                 /*option_mask32 |= FLAG_no_tie_break;*/
 646                 /* ^^^redundant: if FLAG_s, compare_keys() does no tie break */
 647         }
 648
 649         /* Perform the actual sort */
 650         qsort(lines, linecount, sizeof(lines[0]), compare_keys);
 651
 652         /* Handle -u */
 653         if (option_mask32 & FLAG_u) {
 654                 int j = 0;
 655                 /* coreutils 6.3 drop lines for which only key is the same:
 656                  * - disabling last-resort compare, or else compare_keys()
 657                  * will be the same only for completely identical lines
 658                  * - disabling -s (same reasons)
 659                  */
 660                 option_mask32 = (option_mask32 | FLAG_no_tie_break) & (~FLAG_s);
 661                 for (i = 1; i < linecount; i++) {
 662                         if (compare_keys(&lines[j], &lines[i]) == 0)
 663                                 free(lines[i]);
 664                         else
 665                                 lines[++j] = lines[i];
 666                 }
 667                 if (linecount)
 668                         linecount = j+1;
 669         }
 670
 671         /* Print it */
 672 #if ENABLE_FEATURE_SORT_BIG
 673         /* Open output file _after_ we read all input ones */
 674         if (option_mask32 & FLAG_o)
 675                 xmove_fd(xopen(str_o, O_WRONLY|O_CREAT|O_TRUNC), STDOUT_FILENO);
 676 #endif
 677         {
 678                 int ch = (option_mask32 & FLAG_z) ? '\0' : '\n';
 679                 for (i = 0; i < linecount; i++)
 680                         printf("%s%c", lines[i], ch);
 681         }
 682
 683         fflush_stdout_and_exit_SUCCESS();
 684 }