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