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
  15 //usage:#define sort_trivial_usage
  16 //usage:       "[-nru"
  17 //usage:        IF_FEATURE_SORT_BIG("gMcszbdfiokt] [-o FILE] [-k start[.offset][opts][,end[.offset][opts]] [-t CHAR")
  18 //usage:       "] [FILE]..."
  19 //usage:#define sort_full_usage "\n\n"
  20 //usage:       "Sort lines of text\n"
  21 //usage:        IF_FEATURE_SORT_BIG(
  22 //usage:     "\n        -o FILE Output to FILE"
  23 //usage:     "\n        -c      Check whether input is sorted"
  24 //usage:     "\n        -b      Ignore leading blanks"
  25 //usage:     "\n        -f      Ignore case"
  26 //usage:     "\n        -i      Ignore unprintable characters"
  27 //usage:     "\n        -d      Dictionary order (blank or alphanumeric only)"
  28 //usage:     "\n        -g      General numerical sort"
  29 //usage:     "\n        -M      Sort month"
  30 //usage:        )
  31 //-h, --human-numeric-sort: compare human readable numbers (e.g., 2K 1G)
  32 //usage:     "\n        -n      Sort numbers"
  33 //usage:        IF_FEATURE_SORT_BIG(
  34 //usage:     "\n        -t CHAR Field separator"
  35 //usage:     "\n        -k N[,M] Sort by Nth field"
  36 //usage:        )
  37 //usage:     "\n        -r      Reverse sort order"
  38 //usage:        IF_FEATURE_SORT_BIG(
  39 //usage:     "\n        -s      Stable (don't sort ties alphabetically)"
  40 //usage:        )
  41 //usage:     "\n        -u      Suppress duplicate lines"
  42 //usage:        IF_FEATURE_SORT_BIG(
  43 //usage:     "\n        -z      Lines are terminated by NUL, not newline"
  44 ////usage:     "\n      -m      Ignored for GNU compatibility"
  45 ////usage:     "\n      -S BUFSZ Ignored for GNU compatibility"
  46 ////usage:     "\n      -T TMPDIR Ignored for GNU compatibility"
  47 //usage:        )
  48 //usage:
  49 //usage:#define sort_example_usage
  50 //usage:       "$ echo -e \"e\\nf\\nb\\nd\\nc\\na\" | sort\n"
  51 //usage:       "a\n"
  52 //usage:       "b\n"
  53 //usage:       "c\n"
  54 //usage:       "d\n"
  55 //usage:       "e\n"
  56 //usage:       "f\n"
  57 //usage:        IF_FEATURE_SORT_BIG(
  58 //usage:                "$ echo -e \"c 3\\nb 2\\nd 2\" | $SORT -k 2,2n -k 1,1r\n"
  59 //usage:                "d 2\n"
  60 //usage:                "b 2\n"
  61 //usage:                "c 3\n"
  62 //usage:        )
  63 //usage:       ""
  64
  65 #include "libbb.h"
  66
  67 /* This is a NOEXEC applet. Be very careful! */
  68
  69
  70 /*
  71         sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
  72         sort -c [-bdfinru][-t char][-k keydef][file]
  73 */
  74
  75 /* These are sort types */
  76 static const char OPT_STR[] ALIGN1 = "ngMucszbrdfimS:T:o:k:*t:";
  77 enum {
  78         FLAG_n  = 1,            /* Numeric sort */
  79         FLAG_g  = 2,            /* Sort using strtod() */
  80         FLAG_M  = 4,            /* Sort date */
  81 /* ucsz apply to root level only, not keys.  b at root level implies bb */
  82         FLAG_u  = 8,            /* Unique */
  83         FLAG_c  = 0x10,         /* Check: no output, exit(!ordered) */
  84         FLAG_s  = 0x20,         /* Stable sort, no ascii fallback at end */
  85         FLAG_z  = 0x40,         /* Input and output is NUL terminated, not \n */
  86 /* These can be applied to search keys, the previous four can't */
  87         FLAG_b  = 0x80,         /* Ignore leading blanks */
  88         FLAG_r  = 0x100,        /* Reverse */
  89         FLAG_d  = 0x200,        /* Ignore !(isalnum()|isspace()) */
  90         FLAG_f  = 0x400,        /* Force uppercase */
  91         FLAG_i  = 0x800,        /* Ignore !isprint() */
  92         FLAG_m  = 0x1000,       /* ignored: merge already sorted files; do not sort */
  93         FLAG_S  = 0x2000,       /* ignored: -S, --buffer-size=SIZE */
  94         FLAG_T  = 0x4000,       /* ignored: -T, --temporary-directory=DIR */
  95         FLAG_o  = 0x8000,
  96         FLAG_k  = 0x10000,
  97         FLAG_t  = 0x20000,
  98         FLAG_bb = 0x80000000,   /* Ignore trailing blanks  */
  99 };
 100
 101 #if ENABLE_FEATURE_SORT_BIG
 102 static char key_separator;
 103
 104 static struct sort_key {
 105         struct sort_key *next_key;  /* linked list */
 106         unsigned range[4];          /* start word, start char, end word, end char */
 107         unsigned flags;
 108 } *key_list;
 109
 110 static char *get_key(char *str, struct sort_key *key, int flags)
 111 {
 112         int start = start; /* for compiler */
 113         int end;
 114         int len, j;
 115         unsigned i;
 116
 117         /* Special case whole string, so we don't have to make a copy */
 118         if (key->range[0] == 1 && !key->range[1] && !key->range[2] && !key->range[3]
 119          && !(flags & (FLAG_b | FLAG_d | FLAG_f | FLAG_i | FLAG_bb))
 120         ) {
 121                 return str;
 122         }
 123
 124         /* Find start of key on first pass, end on second pass */
 125         len = strlen(str);
 126         for (j = 0; j < 2; j++) {
 127                 if (!key->range[2*j])
 128                         end = len;
 129                 /* Loop through fields */
 130                 else {
 131                         unsigned char ch = 0;
 132
 133                         end = 0;
 134                         for (i = 1; i < key->range[2*j] + j; i++) {
 135                                 if (key_separator) {
 136                                         /* Skip body of key and separator */
 137                                         while ((ch = str[end]) != '\0') {
 138                                                         end++;
 139                                                 if (ch == key_separator)
 140                                                         break;
 141                                         }
 142                                 } else {
 143                                         /* Skip leading blanks */
 144                                         while (isspace(str[end]))
 145                                                 end++;
 146                                         /* Skip body of key */
 147                                         while (str[end] != '\0') {
 148                                                 if (isspace(str[end]))
 149                                                         break;
 150                                                 end++;
 151                                         }
 152                                 }
 153                         }
 154                         /* Remove last delim: "abc:def:" => "abc:def" */
 155                         if (j && ch) {
 156                                 //if (str[end-1] != key_separator)
 157                                 //  bb_error_msg(_and_die("BUG! "
 158                                 //  "str[start:%d,end:%d]:'%.*s'",
 159                                 //  start, end, (int)(end-start), &str[start]);
 160                                 end--;
 161                         }
 162                 }
 163                 if (!j) start = end;
 164         }
 165         /* Strip leading whitespace if necessary */
 166         if (flags & FLAG_b)
 167                 /* not using skip_whitespace() for speed */
 168                 while (isspace(str[start])) start++;
 169         /* Strip trailing whitespace if necessary */
 170         if (flags & FLAG_bb)
 171                 while (end > start && isspace(str[end-1])) end--;
 172         /* -kSTART,N.ENDCHAR: honor ENDCHAR (1-based) */
 173         if (key->range[3]) {
 174                 end = key->range[3];
 175                 if (end > len) end = len;
 176         }
 177         /* -kN.STARTCHAR[,...]: honor STARTCHAR (1-based) */
 178         if (key->range[1]) {
 179                 start += key->range[1] - 1;
 180                 if (start > len) start = len;
 181         }
 182         /* Make the copy */
 183         if (end < start)
 184                 end = start;
 185         str = xstrndup(str+start, end-start);
 186         /* Handle -d */
 187         if (flags & FLAG_d) {
 188                 for (start = end = 0; str[end]; end++)
 189                         if (isspace(str[end]) || isalnum(str[end]))
 190                                 str[start++] = str[end];
 191                 str[start] = '\0';
 192         }
 193         /* Handle -i */
 194         if (flags & FLAG_i) {
 195                 for (start = end = 0; str[end]; end++)
 196                         if (isprint_asciionly(str[end]))
 197                                 str[start++] = str[end];
 198                 str[start] = '\0';
 199         }
 200         /* Handle -f */
 201         if (flags & FLAG_f)
 202                 for (i = 0; str[i]; i++)
 203                         str[i] = toupper(str[i]);
 204
 205         return str;
 206 }
 207
 208 static struct sort_key *add_key(void)
 209 {
 210         struct sort_key **pkey = &key_list;
 211         while (*pkey)
 212                 pkey = &((*pkey)->next_key);
 213         return *pkey = xzalloc(sizeof(struct sort_key));
 214 }
 215
 216 #define GET_LINE(fp) \
 217         ((option_mask32 & FLAG_z) \
 218         ? bb_get_chunk_from_file(fp, NULL) \
 219         : xmalloc_fgetline(fp))
 220 #else
 221 #define GET_LINE(fp) xmalloc_fgetline(fp)
 222 #endif
 223
 224 /* Iterate through keys list and perform comparisons */
 225 static int compare_keys(const void *xarg, const void *yarg)
 226 {
 227         int flags = option_mask32, retval = 0;
 228         char *x, *y;
 229
 230 #if ENABLE_FEATURE_SORT_BIG
 231         struct sort_key *key;
 232
 233         for (key = key_list; !retval && key; key = key->next_key) {
 234                 flags = key->flags ? key->flags : option_mask32;
 235                 /* Chop out and modify key chunks, handling -dfib */
 236                 x = get_key(*(char **)xarg, key, flags);
 237                 y = get_key(*(char **)yarg, key, flags);
 238 #else
 239         /* This curly bracket serves no purpose but to match the nesting
 240          * level of the for () loop we're not using */
 241         {
 242                 x = *(char **)xarg;
 243                 y = *(char **)yarg;
 244 #endif
 245                 /* Perform actual comparison */
 246                 switch (flags & (FLAG_n | FLAG_M | FLAG_g)) {
 247                 default:
 248                         bb_error_msg_and_die("unknown sort type");
 249                         break;
 250                 /* Ascii sort */
 251                 case 0:
 252 #if ENABLE_LOCALE_SUPPORT
 253                         retval = strcoll(x, y);
 254 #else
 255                         retval = strcmp(x, y);
 256 #endif
 257                         break;
 258 #if ENABLE_FEATURE_SORT_BIG
 259                 case FLAG_g: {
 260                         char *xx, *yy;
 261                         double dx = strtod(x, &xx);
 262                         double dy = strtod(y, &yy);
 263                         /* not numbers < NaN < -infinity < numbers < +infinity) */
 264                         if (x == xx)
 265                                 retval = (y == yy ? 0 : -1);
 266                         else if (y == yy)
 267                                 retval = 1;
 268                         /* Check for isnan */
 269                         else if (dx != dx)
 270                                 retval = (dy != dy) ? 0 : -1;
 271                         else if (dy != dy)
 272                                 retval = 1;
 273                         /* Check for infinity.  Could underflow, but it avoids libm. */
 274                         else if (1.0 / dx == 0.0) {
 275                                 if (dx < 0)
 276                                         retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1;
 277                                 else
 278                                         retval = (1.0 / dy == 0.0 && dy > 0) ? 0 : 1;
 279                         } else if (1.0 / dy == 0.0)
 280                                 retval = (dy < 0) ? 1 : -1;
 281                         else
 282                                 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 283                         break;
 284                 }
 285                 case FLAG_M: {
 286                         struct tm thyme;
 287                         int dx;
 288                         char *xx, *yy;
 289
 290                         xx = strptime(x, "%b", &thyme);
 291                         dx = thyme.tm_mon;
 292                         yy = strptime(y, "%b", &thyme);
 293                         if (!xx)
 294                                 retval = (!yy) ? 0 : -1;
 295                         else if (!yy)
 296                                 retval = 1;
 297                         else
 298                                 retval = dx - thyme.tm_mon;
 299                         break;
 300                 }
 301                 /* Full floating point version of -n */
 302                 case FLAG_n: {
 303                         double dx = atof(x);
 304                         double dy = atof(y);
 305                         retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 306                         break;
 307                 }
 308                 } /* switch */
 309                 /* Free key copies. */
 310                 if (x != *(char **)xarg) free(x);
 311                 if (y != *(char **)yarg) free(y);
 312                 /* if (retval) break; - done by for () anyway */
 313 #else
 314                 /* Integer version of -n for tiny systems */
 315                 case FLAG_n:
 316                         retval = atoi(x) - atoi(y);
 317                         break;
 318                 } /* switch */
 319 #endif
 320         } /* for */
 321
 322         /* Perform fallback sort if necessary */
 323         if (!retval && !(option_mask32 & FLAG_s)) {
 324                 flags = option_mask32;
 325                 retval = strcmp(*(char **)xarg, *(char **)yarg);
 326         }
 327
 328         if (flags & FLAG_r)
 329                 return -retval;
 330
 331         return retval;
 332 }
 333
 334 #if ENABLE_FEATURE_SORT_BIG
 335 static unsigned str2u(char **str)
 336 {
 337         unsigned long lu;
 338         if (!isdigit((*str)[0]))
 339                 bb_error_msg_and_die("bad field specification");
 340         lu = strtoul(*str, str, 10);
 341         if ((sizeof(long) > sizeof(int) && lu > INT_MAX) || !lu)
 342                 bb_error_msg_and_die("bad field specification");
 343         return lu;
 344 }
 345 #endif
 346
 347 int sort_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 348 int sort_main(int argc UNUSED_PARAM, char **argv)
 349 {
 350         char *line, **lines;
 351         char *str_ignored, *str_o, *str_t;
 352         llist_t *lst_k = NULL;
 353         int i;
 354         int linecount;
 355         unsigned opts;
 356
 357         xfunc_error_retval = 2;
 358
 359         /* Parse command line options */
 360         /* -o and -t can be given at most once */
 361         opt_complementary = "o--o:t--t"; /* -t, -o: at most one of each */
 362         opts = getopt32(argv, OPT_STR, &str_ignored, &str_ignored, &str_o, &lst_k, &str_t);
 363         /* global b strips leading and trailing spaces */
 364         if (opts & FLAG_b)
 365                 option_mask32 |= FLAG_bb;
 366 #if ENABLE_FEATURE_SORT_BIG
 367         if (opts & FLAG_t) {
 368                 if (!str_t[0] || str_t[1])
 369                         bb_error_msg_and_die("bad -t parameter");
 370                 key_separator = str_t[0];
 371         }
 372         /* note: below this point we use option_mask32, not opts,
 373          * since that reduces register pressure and makes code smaller */
 374
 375         /* Parse sort key */
 376         while (lst_k) {
 377                 enum {
 378                         FLAG_allowed_for_k =
 379                                 FLAG_n | /* Numeric sort */
 380                                 FLAG_g | /* Sort using strtod() */
 381                                 FLAG_M | /* Sort date */
 382                                 FLAG_b | /* Ignore leading blanks */
 383                                 FLAG_r | /* Reverse */
 384                                 FLAG_d | /* Ignore !(isalnum()|isspace()) */
 385                                 FLAG_f | /* Force uppercase */
 386                                 FLAG_i | /* Ignore !isprint() */
 387                         0
 388                 };
 389                 struct sort_key *key = add_key();
 390                 char *str_k = llist_pop(&lst_k);
 391
 392                 i = 0; /* i==0 before comma, 1 after (-k3,6) */
 393                 while (*str_k) {
 394                         /* Start of range */
 395                         /* Cannot use bb_strtou - suffix can be a letter */
 396                         key->range[2*i] = str2u(&str_k);
 397                         if (*str_k == '.') {
 398                                 str_k++;
 399                                 key->range[2*i+1] = str2u(&str_k);
 400                         }
 401                         while (*str_k) {
 402                                 int flag;
 403                                 const char *idx;
 404
 405                                 if (*str_k == ',' && !i++) {
 406                                         str_k++;
 407                                         break;
 408                                 } /* no else needed: fall through to syntax error
 409                                         because comma isn't in OPT_STR */
 410                                 idx = strchr(OPT_STR, *str_k);
 411                                 if (!idx)
 412                                         bb_error_msg_and_die("unknown key option");
 413                                 flag = 1 << (idx - OPT_STR);
 414                                 if (flag & ~FLAG_allowed_for_k)
 415                                         bb_error_msg_and_die("unknown sort type");
 416                                 /* b after ',' means strip _trailing_ space */
 417                                 if (i && flag == FLAG_b)
 418                                         flag = FLAG_bb;
 419                                 key->flags |= flag;
 420                                 str_k++;
 421                         }
 422                 }
 423         }
 424 #endif
 425
 426         /* Open input files and read data */
 427         argv += optind;
 428         if (!*argv)
 429                 *--argv = (char*)"-";
 430         linecount = 0;
 431         lines = NULL;
 432         do {
 433                 /* coreutils 6.9 compat: abort on first open error,
 434                  * do not continue to next file: */
 435                 FILE *fp = xfopen_stdin(*argv);
 436                 for (;;) {
 437                         line = GET_LINE(fp);
 438                         if (!line)
 439                                 break;
 440                         lines = xrealloc_vector(lines, 6, linecount);
 441                         lines[linecount++] = line;
 442                 }
 443                 fclose_if_not_stdin(fp);
 444         } while (*++argv);
 445
 446 #if ENABLE_FEATURE_SORT_BIG
 447         /* If no key, perform alphabetic sort */
 448         if (!key_list)
 449                 add_key()->range[0] = 1;
 450         /* Handle -c */
 451         if (option_mask32 & FLAG_c) {
 452                 int j = (option_mask32 & FLAG_u) ? -1 : 0;
 453                 for (i = 1; i < linecount; i++) {
 454                         if (compare_keys(&lines[i-1], &lines[i]) > j) {
 455                                 fprintf(stderr, "Check line %u\n", i);
 456                                 return EXIT_FAILURE;
 457                         }
 458                 }
 459                 return EXIT_SUCCESS;
 460         }
 461 #endif
 462         /* Perform the actual sort */
 463         qsort(lines, linecount, sizeof(lines[0]), compare_keys);
 464
 465         /* Handle -u */
 466         if (option_mask32 & FLAG_u) {
 467                 int j = 0;
 468                 /* coreutils 6.3 drop lines for which only key is the same */
 469                 /* -- disabling last-resort compare... */
 470                 option_mask32 |= FLAG_s;
 471                 for (i = 1; i < linecount; i++) {
 472                         if (compare_keys(&lines[j], &lines[i]) == 0)
 473                                 free(lines[i]);
 474                         else
 475                                 lines[++j] = lines[i];
 476                 }
 477                 if (linecount)
 478                         linecount = j+1;
 479         }
 480
 481         /* Print it */
 482 #if ENABLE_FEATURE_SORT_BIG
 483         /* Open output file _after_ we read all input ones */
 484         if (option_mask32 & FLAG_o)
 485                 xmove_fd(xopen(str_o, O_WRONLY|O_CREAT|O_TRUNC), STDOUT_FILENO);
 486 #endif
 487         {
 488                 int ch = (option_mask32 & FLAG_z) ? '\0' : '\n';
 489                 for (i = 0; i < linecount; i++)
 490                         printf("%s%c", lines[i], ch);
 491         }
 492
 493         fflush_stdout_and_exit(EXIT_SUCCESS);
 494 }