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:     "\nOptions:"
  22 //usage:        IF_FEATURE_SORT_BIG(
  23 //usage:     "\n        -b      Ignore leading blanks"
  24 //usage:     "\n        -c      Check whether input is sorted"
  25 //usage:     "\n        -d      Dictionary order (blank or alphanumeric only)"
  26 //usage:     "\n        -f      Ignore case"
  27 //usage:     "\n        -g      General numerical sort"
  28 //usage:     "\n        -i      Ignore unprintable characters"
  29 //usage:     "\n        -k      Sort key"
  30 //usage:     "\n        -M      Sort month"
  31 //usage:        )
  32 //usage:     "\n        -n      Sort numbers"
  33 //usage:        IF_FEATURE_SORT_BIG(
  34 //usage:     "\n        -o      Output to file"
  35 //usage:     "\n        -k      Sort by key"
  36 //usage:     "\n        -t CHAR Key separator"
  37 //usage:        )
  38 //usage:     "\n        -r      Reverse sort order"
  39 //usage:        IF_FEATURE_SORT_BIG(
  40 //usage:     "\n        -s      Stable (don't sort ties alphabetically)"
  41 //usage:        )
  42 //usage:     "\n        -u      Suppress duplicate lines"
  43 //usage:        IF_FEATURE_SORT_BIG(
  44 //usage:     "\n        -z      Lines are terminated by NUL, not newline"
  45 //usage:     "\n        -mST    Ignored for GNU compatibility")
  46 //usage:
  47 //usage:#define sort_example_usage
  48 //usage:       "$ echo -e \"e\\nf\\nb\\nd\\nc\\na\" | sort\n"
  49 //usage:       "a\n"
  50 //usage:       "b\n"
  51 //usage:       "c\n"
  52 //usage:       "d\n"
  53 //usage:       "e\n"
  54 //usage:       "f\n"
  55 //usage:        IF_FEATURE_SORT_BIG(
  56 //usage:                "$ echo -e \"c 3\\nb 2\\nd 2\" | $SORT -k 2,2n -k 1,1r\n"
  57 //usage:                "d 2\n"
  58 //usage:                "b 2\n"
  59 //usage:                "c 3\n"
  60 //usage:        )
  61 //usage:       ""
  62
  63 #include "libbb.h"
  64
  65 /* This is a NOEXEC applet. Be very careful! */
  66
  67
  68 /*
  69         sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
  70         sort -c [-bdfinru][-t char][-k keydef][file]
  71 */
  72
  73 /* These are sort types */
  74 static const char OPT_STR[] ALIGN1 = "ngMucszbrdfimS:T:o:k:t:";
  75 enum {
  76         FLAG_n  = 1,            /* Numeric sort */
  77         FLAG_g  = 2,            /* Sort using strtod() */
  78         FLAG_M  = 4,            /* Sort date */
  79 /* ucsz apply to root level only, not keys.  b at root level implies bb */
  80         FLAG_u  = 8,            /* Unique */
  81         FLAG_c  = 0x10,         /* Check: no output, exit(!ordered) */
  82         FLAG_s  = 0x20,         /* Stable sort, no ascii fallback at end */
  83         FLAG_z  = 0x40,         /* Input and output is NUL terminated, not \n */
  84 /* These can be applied to search keys, the previous four can't */
  85         FLAG_b  = 0x80,         /* Ignore leading blanks */
  86         FLAG_r  = 0x100,        /* Reverse */
  87         FLAG_d  = 0x200,        /* Ignore !(isalnum()|isspace()) */
  88         FLAG_f  = 0x400,        /* Force uppercase */
  89         FLAG_i  = 0x800,        /* Ignore !isprint() */
  90         FLAG_m  = 0x1000,       /* ignored: merge already sorted files; do not sort */
  91         FLAG_S  = 0x2000,       /* ignored: -S, --buffer-size=SIZE */
  92         FLAG_T  = 0x4000,       /* ignored: -T, --temporary-directory=DIR */
  93         FLAG_o  = 0x8000,
  94         FLAG_k  = 0x10000,
  95         FLAG_t  = 0x20000,
  96         FLAG_bb = 0x80000000,   /* Ignore trailing blanks  */
  97 };
  98
  99 #if ENABLE_FEATURE_SORT_BIG
 100 static char key_separator;
 101
 102 static struct sort_key {
 103         struct sort_key *next_key;  /* linked list */
 104         unsigned range[4];          /* start word, start char, end word, end char */
 105         unsigned flags;
 106 } *key_list;
 107
 108 static char *get_key(char *str, struct sort_key *key, int flags)
 109 {
 110         int start = 0, end = 0, len, j;
 111         unsigned i;
 112
 113         /* Special case whole string, so we don't have to make a copy */
 114         if (key->range[0] == 1 && !key->range[1] && !key->range[2] && !key->range[3]
 115          && !(flags & (FLAG_b | FLAG_d | FLAG_f | FLAG_i | FLAG_bb))
 116         ) {
 117                 return str;
 118         }
 119
 120         /* Find start of key on first pass, end on second pass */
 121         len = strlen(str);
 122         for (j = 0; j < 2; j++) {
 123                 if (!key->range[2*j])
 124                         end = len;
 125                 /* Loop through fields */
 126                 else {
 127                         end = 0;
 128                         for (i = 1; i < key->range[2*j] + j; i++) {
 129                                 if (key_separator) {
 130                                         /* Skip body of key and separator */
 131                                         while (str[end]) {
 132                                                 if (str[end++] == key_separator)
 133                                                         break;
 134                                         }
 135                                 } else {
 136                                         /* Skip leading blanks */
 137                                         while (isspace(str[end]))
 138                                                 end++;
 139                                         /* Skip body of key */
 140                                         while (str[end]) {
 141                                                 if (isspace(str[end]))
 142                                                         break;
 143                                                 end++;
 144                                         }
 145                                 }
 146                         }
 147                 }
 148                 if (!j) start = end;
 149         }
 150         /* Strip leading whitespace if necessary */
 151 //XXX: skip_whitespace()
 152         if (flags & FLAG_b)
 153                 while (isspace(str[start])) start++;
 154         /* Strip trailing whitespace if necessary */
 155         if (flags & FLAG_bb)
 156                 while (end > start && isspace(str[end-1])) end--;
 157         /* Handle offsets on start and end */
 158         if (key->range[3]) {
 159                 end += key->range[3] - 1;
 160                 if (end > len) end = len;
 161         }
 162         if (key->range[1]) {
 163                 start += key->range[1] - 1;
 164                 if (start > len) start = len;
 165         }
 166         /* Make the copy */
 167         if (end < start) end = start;
 168         str = xstrndup(str+start, end-start);
 169         /* Handle -d */
 170         if (flags & FLAG_d) {
 171                 for (start = end = 0; str[end]; end++)
 172                         if (isspace(str[end]) || isalnum(str[end]))
 173                                 str[start++] = str[end];
 174                 str[start] = '\0';
 175         }
 176         /* Handle -i */
 177         if (flags & FLAG_i) {
 178                 for (start = end = 0; str[end]; end++)
 179                         if (isprint_asciionly(str[end]))
 180                                 str[start++] = str[end];
 181                 str[start] = '\0';
 182         }
 183         /* Handle -f */
 184         if (flags & FLAG_f)
 185                 for (i = 0; str[i]; i++)
 186                         str[i] = toupper(str[i]);
 187
 188         return str;
 189 }
 190
 191 static struct sort_key *add_key(void)
 192 {
 193         struct sort_key **pkey = &key_list;
 194         while (*pkey)
 195                 pkey = &((*pkey)->next_key);
 196         return *pkey = xzalloc(sizeof(struct sort_key));
 197 }
 198
 199 #define GET_LINE(fp) \
 200         ((option_mask32 & FLAG_z) \
 201         ? bb_get_chunk_from_file(fp, NULL) \
 202         : xmalloc_fgetline(fp))
 203 #else
 204 #define GET_LINE(fp) xmalloc_fgetline(fp)
 205 #endif
 206
 207 /* Iterate through keys list and perform comparisons */
 208 static int compare_keys(const void *xarg, const void *yarg)
 209 {
 210         int flags = option_mask32, retval = 0;
 211         char *x, *y;
 212
 213 #if ENABLE_FEATURE_SORT_BIG
 214         struct sort_key *key;
 215
 216         for (key = key_list; !retval && key; key = key->next_key) {
 217                 flags = key->flags ? key->flags : option_mask32;
 218                 /* Chop out and modify key chunks, handling -dfib */
 219                 x = get_key(*(char **)xarg, key, flags);
 220                 y = get_key(*(char **)yarg, key, flags);
 221 #else
 222         /* This curly bracket serves no purpose but to match the nesting
 223            level of the for () loop we're not using */
 224         {
 225                 x = *(char **)xarg;
 226                 y = *(char **)yarg;
 227 #endif
 228                 /* Perform actual comparison */
 229                 switch (flags & 7) {
 230                 default:
 231                         bb_error_msg_and_die("unknown sort type");
 232                         break;
 233                 /* Ascii sort */
 234                 case 0:
 235 #if ENABLE_LOCALE_SUPPORT
 236                         retval = strcoll(x, y);
 237 #else
 238                         retval = strcmp(x, y);
 239 #endif
 240                         break;
 241 #if ENABLE_FEATURE_SORT_BIG
 242                 case FLAG_g: {
 243                         char *xx, *yy;
 244                         double dx = strtod(x, &xx);
 245                         double dy = strtod(y, &yy);
 246                         /* not numbers < NaN < -infinity < numbers < +infinity) */
 247                         if (x == xx)
 248                                 retval = (y == yy ? 0 : -1);
 249                         else if (y == yy)
 250                                 retval = 1;
 251                         /* Check for isnan */
 252                         else if (dx != dx)
 253                                 retval = (dy != dy) ? 0 : -1;
 254                         else if (dy != dy)
 255                                 retval = 1;
 256                         /* Check for infinity.  Could underflow, but it avoids libm. */
 257                         else if (1.0 / dx == 0.0) {
 258                                 if (dx < 0)
 259                                         retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1;
 260                                 else
 261                                         retval = (1.0 / dy == 0.0 && dy > 0) ? 0 : 1;
 262                         } else if (1.0 / dy == 0.0)
 263                                 retval = (dy < 0) ? 1 : -1;
 264                         else
 265                                 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 266                         break;
 267                 }
 268                 case FLAG_M: {
 269                         struct tm thyme;
 270                         int dx;
 271                         char *xx, *yy;
 272
 273                         xx = strptime(x, "%b", &thyme);
 274                         dx = thyme.tm_mon;
 275                         yy = strptime(y, "%b", &thyme);
 276                         if (!xx)
 277                                 retval = (!yy) ? 0 : -1;
 278                         else if (!yy)
 279                                 retval = 1;
 280                         else
 281                                 retval = (dx == thyme.tm_mon) ? 0 : dx - thyme.tm_mon;
 282                         break;
 283                 }
 284                 /* Full floating point version of -n */
 285                 case FLAG_n: {
 286                         double dx = atof(x);
 287                         double dy = atof(y);
 288                         retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 289                         break;
 290                 }
 291                 } /* switch */
 292                 /* Free key copies. */
 293                 if (x != *(char **)xarg) free(x);
 294                 if (y != *(char **)yarg) free(y);
 295                 /* if (retval) break; - done by for () anyway */
 296 #else
 297                 /* Integer version of -n for tiny systems */
 298                 case FLAG_n:
 299                         retval = atoi(x) - atoi(y);
 300                         break;
 301                 } /* switch */
 302 #endif
 303         } /* for */
 304
 305         /* Perform fallback sort if necessary */
 306         if (!retval && !(option_mask32 & FLAG_s))
 307                 retval = strcmp(*(char **)xarg, *(char **)yarg);
 308
 309         if (flags & FLAG_r) return -retval;
 310         return retval;
 311 }
 312
 313 #if ENABLE_FEATURE_SORT_BIG
 314 static unsigned str2u(char **str)
 315 {
 316         unsigned long lu;
 317         if (!isdigit((*str)[0]))
 318                 bb_error_msg_and_die("bad field specification");
 319         lu = strtoul(*str, str, 10);
 320         if ((sizeof(long) > sizeof(int) && lu > INT_MAX) || !lu)
 321                 bb_error_msg_and_die("bad field specification");
 322         return lu;
 323 }
 324 #endif
 325
 326 int sort_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 327 int sort_main(int argc UNUSED_PARAM, char **argv)
 328 {
 329         char *line, **lines;
 330         char *str_ignored, *str_o, *str_t;
 331         llist_t *lst_k = NULL;
 332         int i, flag;
 333         int linecount;
 334         unsigned opts;
 335
 336         xfunc_error_retval = 2;
 337
 338         /* Parse command line options */
 339         /* -o and -t can be given at most once */
 340         opt_complementary = "o--o:t--t:" /* -t, -o: at most one of each */
 341                         "k::"; /* -k takes list */
 342         opts = getopt32(argv, OPT_STR, &str_ignored, &str_ignored, &str_o, &lst_k, &str_t);
 343         /* global b strips leading and trailing spaces */
 344         if (opts & FLAG_b)
 345                 option_mask32 |= FLAG_bb;
 346 #if ENABLE_FEATURE_SORT_BIG
 347         if (opts & FLAG_t) {
 348                 if (!str_t[0] || str_t[1])
 349                         bb_error_msg_and_die("bad -t parameter");
 350                 key_separator = str_t[0];
 351         }
 352         /* note: below this point we use option_mask32, not opts,
 353          * since that reduces register pressure and makes code smaller */
 354
 355         /* parse sort key */
 356         while (lst_k) {
 357                 enum {
 358                         FLAG_allowed_for_k =
 359                                 FLAG_n | /* Numeric sort */
 360                                 FLAG_g | /* Sort using strtod() */
 361                                 FLAG_M | /* Sort date */
 362                                 FLAG_b | /* Ignore leading blanks */
 363                                 FLAG_r | /* Reverse */
 364                                 FLAG_d | /* Ignore !(isalnum()|isspace()) */
 365                                 FLAG_f | /* Force uppercase */
 366                                 FLAG_i | /* Ignore !isprint() */
 367                         0
 368                 };
 369                 struct sort_key *key = add_key();
 370                 char *str_k = llist_pop(&lst_k);
 371
 372                 i = 0; /* i==0 before comma, 1 after (-k3,6) */
 373                 while (*str_k) {
 374                         /* Start of range */
 375                         /* Cannot use bb_strtou - suffix can be a letter */
 376                         key->range[2*i] = str2u(&str_k);
 377                         if (*str_k == '.') {
 378                                 str_k++;
 379                                 key->range[2*i+1] = str2u(&str_k);
 380                         }
 381                         while (*str_k) {
 382                                 const char *temp2;
 383
 384                                 if (*str_k == ',' && !i++) {
 385                                         str_k++;
 386                                         break;
 387                                 } /* no else needed: fall through to syntax error
 388                                         because comma isn't in OPT_STR */
 389                                 temp2 = strchr(OPT_STR, *str_k);
 390                                 if (!temp2)
 391                                         bb_error_msg_and_die("unknown key option");
 392                                 flag = 1 << (temp2 - OPT_STR);
 393                                 if (flag & ~FLAG_allowed_for_k)
 394                                         bb_error_msg_and_die("unknown sort type");
 395                                 /* b after ',' means strip _trailing_ space */
 396                                 if (i && flag == FLAG_b)
 397                                         flag = FLAG_bb;
 398                                 key->flags |= flag;
 399                                 str_k++;
 400                         }
 401                 }
 402         }
 403 #endif
 404
 405         /* Open input files and read data */
 406         argv += optind;
 407         if (!*argv)
 408                 *--argv = (char*)"-";
 409         linecount = 0;
 410         lines = NULL;
 411         do {
 412                 /* coreutils 6.9 compat: abort on first open error,
 413                  * do not continue to next file: */
 414                 FILE *fp = xfopen_stdin(*argv);
 415                 for (;;) {
 416                         line = GET_LINE(fp);
 417                         if (!line)
 418                                 break;
 419                         lines = xrealloc_vector(lines, 6, linecount);
 420                         lines[linecount++] = line;
 421                 }
 422                 fclose_if_not_stdin(fp);
 423         } while (*++argv);
 424
 425 #if ENABLE_FEATURE_SORT_BIG
 426         /* if no key, perform alphabetic sort */
 427         if (!key_list)
 428                 add_key()->range[0] = 1;
 429         /* handle -c */
 430         if (option_mask32 & FLAG_c) {
 431                 int j = (option_mask32 & FLAG_u) ? -1 : 0;
 432                 for (i = 1; i < linecount; i++) {
 433                         if (compare_keys(&lines[i-1], &lines[i]) > j) {
 434                                 fprintf(stderr, "Check line %u\n", i);
 435                                 return EXIT_FAILURE;
 436                         }
 437                 }
 438                 return EXIT_SUCCESS;
 439         }
 440 #endif
 441         /* Perform the actual sort */
 442         qsort(lines, linecount, sizeof(lines[0]), compare_keys);
 443         /* handle -u */
 444         if (option_mask32 & FLAG_u) {
 445                 flag = 0;
 446                 /* coreutils 6.3 drop lines for which only key is the same */
 447                 /* -- disabling last-resort compare... */
 448                 option_mask32 |= FLAG_s;
 449                 for (i = 1; i < linecount; i++) {
 450                         if (compare_keys(&lines[flag], &lines[i]) == 0)
 451                                 free(lines[i]);
 452                         else
 453                                 lines[++flag] = lines[i];
 454                 }
 455                 if (linecount)
 456                         linecount = flag+1;
 457         }
 458
 459         /* Print it */
 460 #if ENABLE_FEATURE_SORT_BIG
 461         /* Open output file _after_ we read all input ones */
 462         if (option_mask32 & FLAG_o)
 463                 xmove_fd(xopen(str_o, O_WRONLY|O_CREAT|O_TRUNC), STDOUT_FILENO);
 464 #endif
 465         flag = (option_mask32 & FLAG_z) ? '\0' : '\n';
 466         for (i = 0; i < linecount; i++)
 467                 printf("%s%c", lines[i], flag);
 468
 469         fflush_stdout_and_exit(EXIT_SUCCESS);
 470 }