box/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 tarball for details.
  10  *
  11  * See SuS3 sort standard at:
  12  * http://www.opengroup.org/onlinepubs/007904975/utilities/sort.html
  13  */
  14
  15 #include "libbb.h"
  16
  17 /* This is a NOEXEC applet. Be very careful! */
  18
  19
  20 /*
  21         sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
  22         sort -c [-bdfinru][-t char][-k keydef][file]
  23 */
  24
  25 /* These are sort types */
  26 static const char OPT_STR[] = "ngMucszbrdfimS:T:o:k:t:";
  27 enum {
  28         FLAG_n  = 1,            /* Numeric sort */
  29         FLAG_g  = 2,            /* Sort using strtod() */
  30         FLAG_M  = 4,            /* Sort date */
  31 /* ucsz apply to root level only, not keys.  b at root level implies bb */
  32         FLAG_u  = 8,            /* Unique */
  33         FLAG_c  = 0x10,         /* Check: no output, exit(!ordered) */
  34         FLAG_s  = 0x20,         /* Stable sort, no ascii fallback at end */
  35         FLAG_z  = 0x40,         /* Input is null terminated, not \n */
  36 /* These can be applied to search keys, the previous four can't */
  37         FLAG_b  = 0x80,         /* Ignore leading blanks */
  38         FLAG_r  = 0x100,        /* Reverse */
  39         FLAG_d  = 0x200,        /* Ignore !(isalnum()|isspace()) */
  40         FLAG_f  = 0x400,        /* Force uppercase */
  41         FLAG_i  = 0x800,        /* Ignore !isprint() */
  42         FLAG_m  = 0x1000,       /* ignored: merge already sorted files; do not sort */
  43         FLAG_S  = 0x2000,       /* ignored: -S, --buffer-size=SIZE */
  44         FLAG_T  = 0x4000,       /* ignored: -T, --temporary-directory=DIR */
  45         FLAG_o  = 0x8000,
  46         FLAG_k  = 0x10000,
  47         FLAG_t  = 0x20000,
  48         FLAG_bb = 0x80000000,   /* Ignore trailing blanks  */
  49 };
  50
  51 #if ENABLE_FEATURE_SORT_BIG
  52 static char key_separator;
  53
  54 static struct sort_key {
  55         struct sort_key *next_key;      /* linked list */
  56         unsigned range[4];      /* start word, start char, end word, end char */
  57         unsigned flags;
  58 } *key_list;
  59
  60 static char *get_key(char *str, struct sort_key *key, int flags)
  61 {
  62         int start = 0, end = 0, len, i, j;
  63
  64         /* Special case whole string, so we don't have to make a copy */
  65         if (key->range[0] == 1 && !key->range[1] && !key->range[2] && !key->range[3]
  66          && !(flags & (FLAG_b | FLAG_d | FLAG_f | FLAG_i | FLAG_bb))
  67         ) {
  68                 return str;
  69         }
  70
  71         /* Find start of key on first pass, end on second pass */
  72         len = strlen(str);
  73         for (j = 0; j < 2; j++) {
  74                 if (!key->range[2*j])
  75                         end = len;
  76                 /* Loop through fields */
  77                 else {
  78                         end = 0;
  79                         for (i = 1; i < key->range[2*j] + j; i++) {
  80                                 if (key_separator) {
  81                                         /* Skip body of key and separator */
  82                                         while (str[end]) {
  83                                                 if (str[end++] == key_separator)
  84                                                         break;
  85                                         }
  86                                 } else {
  87                                         /* Skip leading blanks */
  88                                         while (isspace(str[end]))
  89                                                 end++;
  90                                         /* Skip body of key */
  91                                         while (str[end]) {
  92                                                 if (isspace(str[end]))
  93                                                         break;
  94                                                 end++;
  95                                         }
  96                                 }
  97                         }
  98                 }
  99                 if (!j) start = end;
 100         }
 101         /* Strip leading whitespace if necessary */
 102 //XXX: skip_whitespace()
 103         if (flags & FLAG_b)
 104                 while (isspace(str[start])) start++;
 105         /* Strip trailing whitespace if necessary */
 106         if (flags & FLAG_bb)
 107                 while (end > start && isspace(str[end-1])) end--;
 108         /* Handle offsets on start and end */
 109         if (key->range[3]) {
 110                 end += key->range[3] - 1;
 111                 if (end > len) end = len;
 112         }
 113         if (key->range[1]) {
 114                 start += key->range[1] - 1;
 115                 if (start > len) start = len;
 116         }
 117         /* Make the copy */
 118         if (end < start) end = start;
 119         str = xstrndup(str+start, end-start);
 120         /* Handle -d */
 121         if (flags & FLAG_d) {
 122                 for (start = end = 0; str[end]; end++)
 123                         if (isspace(str[end]) || isalnum(str[end]))
 124                                 str[start++] = str[end];
 125                 str[start] = '\0';
 126         }
 127         /* Handle -i */
 128         if (flags & FLAG_i) {
 129                 for (start = end = 0; str[end]; end++)
 130                         if (isprint(str[end]))
 131                                 str[start++] = str[end];
 132                 str[start] = '\0';
 133         }
 134         /* Handle -f */
 135         if (flags & FLAG_f)
 136                 for (i = 0; str[i]; i++)
 137                         str[i] = toupper(str[i]);
 138
 139         return str;
 140 }
 141
 142 static struct sort_key *add_key(void)
 143 {
 144         struct sort_key **pkey = &key_list;
 145         while (*pkey)
 146                 pkey = &((*pkey)->next_key);
 147         return *pkey = xzalloc(sizeof(struct sort_key));
 148 }
 149
 150 #define GET_LINE(fp) \
 151         ((option_mask32 & FLAG_z) \
 152         ? bb_get_chunk_from_file(fp, NULL) \
 153         : xmalloc_getline(fp))
 154 #else
 155 #define GET_LINE(fp) xmalloc_getline(fp)
 156 #endif
 157
 158 /* Iterate through keys list and perform comparisons */
 159 static int compare_keys(const void *xarg, const void *yarg)
 160 {
 161         int flags = option_mask32, retval = 0;
 162         char *x, *y;
 163
 164 #if ENABLE_FEATURE_SORT_BIG
 165         struct sort_key *key;
 166
 167         for (key = key_list; !retval && key; key = key->next_key) {
 168                 flags = key->flags ? key->flags : option_mask32;
 169                 /* Chop out and modify key chunks, handling -dfib */
 170                 x = get_key(*(char **)xarg, key, flags);
 171                 y = get_key(*(char **)yarg, key, flags);
 172 #else
 173         /* This curly bracket serves no purpose but to match the nesting
 174            level of the for () loop we're not using */
 175         {
 176                 x = *(char **)xarg;
 177                 y = *(char **)yarg;
 178 #endif
 179                 /* Perform actual comparison */
 180                 switch (flags & 7) {
 181                 default:
 182                         bb_error_msg_and_die("unknown sort type");
 183                         break;
 184                 /* Ascii sort */
 185                 case 0:
 186 #if ENABLE_LOCALE_SUPPORT
 187                         retval = strcoll(x, y);
 188 #else
 189                         retval = strcmp(x, y);
 190 #endif
 191                         break;
 192 #if ENABLE_FEATURE_SORT_BIG
 193                 case FLAG_g: {
 194                         char *xx, *yy;
 195                         double dx = strtod(x, &xx);
 196                         double dy = strtod(y, &yy);
 197                         /* not numbers < NaN < -infinity < numbers < +infinity) */
 198                         if (x == xx)
 199                                 retval = (y == yy ? 0 : -1);
 200                         else if (y == yy)
 201                                 retval = 1;
 202                         /* Check for isnan */
 203                         else if (dx != dx)
 204                                 retval = (dy != dy) ? 0 : -1;
 205                         else if (dy != dy)
 206                                 retval = 1;
 207                         /* Check for infinity.  Could underflow, but it avoids libm. */
 208                         else if (1.0 / dx == 0.0) {
 209                                 if (dx < 0)
 210                                         retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1;
 211                                 else
 212                                         retval = (1.0 / dy == 0.0 && dy > 0) ? 0 : 1;
 213                         } else if (1.0 / dy == 0.0)
 214                                 retval = (dy < 0) ? 1 : -1;
 215                         else
 216                                 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 217                         break;
 218                 }
 219                 case FLAG_M: {
 220                         struct tm thyme;
 221                         int dx;
 222                         char *xx, *yy;
 223
 224                         xx = strptime(x, "%b", &thyme);
 225                         dx = thyme.tm_mon;
 226                         yy = strptime(y, "%b", &thyme);
 227                         if (!xx)
 228                                 retval = (!yy) ? 0 : -1;
 229                         else if (!yy)
 230                                 retval = 1;
 231                         else
 232                                 retval = (dx == thyme.tm_mon) ? 0 : dx - thyme.tm_mon;
 233                         break;
 234                 }
 235                 /* Full floating point version of -n */
 236                 case FLAG_n: {
 237                         double dx = atof(x);
 238                         double dy = atof(y);
 239                         retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
 240                         break;
 241                 }
 242                 } /* switch */
 243                 /* Free key copies. */
 244                 if (x != *(char **)xarg) free(x);
 245                 if (y != *(char **)yarg) free(y);
 246                 /* if (retval) break; - done by for () anyway */
 247 #else
 248                 /* Integer version of -n for tiny systems */
 249                 case FLAG_n:
 250                         retval = atoi(x) - atoi(y);
 251                         break;
 252                 } /* switch */
 253 #endif
 254         } /* for */
 255
 256         /* Perform fallback sort if necessary */
 257         if (!retval && !(option_mask32 & FLAG_s))
 258                 retval = strcmp(*(char **)xarg, *(char **)yarg);
 259
 260         if (flags & FLAG_r) return -retval;
 261         return retval;
 262 }
 263
 264 #if ENABLE_FEATURE_SORT_BIG
 265 static unsigned str2u(char **str)
 266 {
 267         unsigned long lu;
 268         if (!isdigit((*str)[0]))
 269                 bb_error_msg_and_die("bad field specification");
 270         lu = strtoul(*str, str, 10);
 271         if ((sizeof(long) > sizeof(int) && lu > INT_MAX) || !lu)
 272                 bb_error_msg_and_die("bad field specification");
 273         return lu;
 274 }
 275 #endif
 276
 277 int sort_main(int argc, char **argv);
 278 int sort_main(int argc, char **argv)
 279 {
 280         FILE *fp, *outfile = stdout;
 281         char *line, **lines = NULL;
 282         char *str_ignored, *str_o, *str_t;
 283         llist_t *lst_k = NULL;
 284         int i, flag;
 285         int linecount = 0;
 286
 287         xfunc_error_retval = 2;
 288
 289         /* Parse command line options */
 290         /* -o and -t can be given at most once */
 291         opt_complementary = "?:o--o:t--t:" /* -t, -o: maximum one of each */
 292                         "k::"; /* -k takes list */
 293         getopt32(argc, argv, OPT_STR, &str_ignored, &str_ignored, &str_o, &lst_k, &str_t);
 294 #if ENABLE_FEATURE_SORT_BIG
 295         if (option_mask32 & FLAG_o) outfile = xfopen(str_o, "w");
 296         if (option_mask32 & FLAG_t) {
 297                 if (!str_t[0] || str_t[1])
 298                         bb_error_msg_and_die("bad -t parameter");
 299                 key_separator = str_t[0];
 300         }
 301         /* parse sort key */
 302         while (lst_k) {
 303                 enum {
 304                         FLAG_allowed_for_k =
 305                                 FLAG_n | /* Numeric sort */
 306                                 FLAG_g | /* Sort using strtod() */
 307                                 FLAG_M | /* Sort date */
 308                                 FLAG_b | /* Ignore leading blanks */
 309                                 FLAG_r | /* Reverse */
 310                                 FLAG_d | /* Ignore !(isalnum()|isspace()) */
 311                                 FLAG_f | /* Force uppercase */
 312                                 FLAG_i | /* Ignore !isprint() */
 313                         0
 314                 };
 315                 struct sort_key *key = add_key();
 316                 char *str_k = lst_k->data;
 317                 const char *temp2;
 318
 319                 i = 0; /* i==0 before comma, 1 after (-k3,6) */
 320                 while (*str_k) {
 321                         /* Start of range */
 322                         /* Cannot use bb_strtou - suffix can be a letter */
 323                         key->range[2*i] = str2u(&str_k);
 324                         if (*str_k == '.') {
 325                                 str_k++;
 326                                 key->range[2*i+1] = str2u(&str_k);
 327                         }
 328                         while (*str_k) {
 329                                 if (*str_k == ',' && !i++) {
 330                                         str_k++;
 331                                         break;
 332                                 } /* no else needed: fall through to syntax error
 333                                          because comma isn't in OPT_STR */
 334                                 temp2 = strchr(OPT_STR, *str_k);
 335                                 if (!temp2)
 336                                         bb_error_msg_and_die("unknown key option");
 337                                 flag = 1 << (temp2 - OPT_STR);
 338                                 if (flag & ~FLAG_allowed_for_k)
 339                                         bb_error_msg_and_die("unknown sort type");
 340                                 /* b after ',' means strip _trailing_ space */
 341                                 if (i && flag == FLAG_b) flag = FLAG_bb;
 342                                 key->flags |= flag;
 343                                 str_k++;
 344                         }
 345                 }
 346                 /* leaking lst_k... */
 347                 lst_k = lst_k->link;
 348         }
 349 #endif
 350         /* global b strips leading and trailing spaces */
 351         if (option_mask32 & FLAG_b) option_mask32 |= FLAG_bb;
 352
 353         /* Open input files and read data */
 354         for (i = argv[optind] ? optind : optind-1; argv[i]; i++) {
 355                 fp = stdin;
 356                 if (i >= optind && NOT_LONE_DASH(argv[i]))
 357                         fp = xfopen(argv[i], "r");
 358                 for (;;) {
 359                         line = GET_LINE(fp);
 360                         if (!line) break;
 361                         if (!(linecount & 63))
 362                                 lines = xrealloc(lines, sizeof(char *) * (linecount + 64));
 363                         lines[linecount++] = line;
 364                 }
 365                 fclose(fp);
 366         }
 367 #if ENABLE_FEATURE_SORT_BIG
 368         /* if no key, perform alphabetic sort */
 369         if (!key_list)
 370                 add_key()->range[0] = 1;
 371         /* handle -c */
 372         if (option_mask32 & FLAG_c) {
 373                 int j = (option_mask32 & FLAG_u) ? -1 : 0;
 374                 for (i = 1; i < linecount; i++)
 375                         if (compare_keys(&lines[i-1], &lines[i]) > j) {
 376                                 fprintf(stderr, "Check line %d\n", i);
 377                                 return 1;
 378                         }
 379                 return 0;
 380         }
 381 #endif
 382         /* Perform the actual sort */
 383         qsort(lines, linecount, sizeof(char *), compare_keys);
 384         /* handle -u */
 385         if (option_mask32 & FLAG_u) {
 386                 flag = 0;
 387                 /* coreutils 6.3 drop lines for which only key is the same */
 388                 /* -- disabling last-resort compare... */
 389                 option_mask32 |= FLAG_s;
 390                 for (i = 1; i < linecount; i++) {
 391                         if (!compare_keys(&lines[flag], &lines[i]))
 392                                 free(lines[i]);
 393                         else
 394                                 lines[++flag] = lines[i];
 395                 }
 396                 if (linecount) linecount = flag+1;
 397         }
 398         /* Print it */
 399         for (i = 0; i < linecount; i++)
 400                 fprintf(outfile, "%s\n", lines[i]);
 401
 402         fflush_stdout_and_exit(EXIT_SUCCESS);
 403 }