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