1 /* vi: set sw=4 ts=4: */
3 * SuS3 compliant sort implementation for busybox
5 * Copyright (C) 2004 by Rob Landley <rob@landley.net>
7 * MAINTAINER: Rob Landley <rob@landley.net>
9 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
11 * See SuS3 sort standard at:
12 * http://www.opengroup.org/onlinepubs/007904975/utilities/sort.html
15 //usage:#define sort_trivial_usage
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"
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"
37 //usage: "\n -r Reverse sort order"
38 //usage: IF_FEATURE_SORT_BIG(
39 //usage: "\n -s Stable (don't sort ties alphabetically)"
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")
46 //usage:#define sort_example_usage
47 //usage: "$ echo -e \"e\\nf\\nb\\nd\\nc\\na\" | sort\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"
64 /* This is a NOEXEC applet. Be very careful! */
68 sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
69 sort -c [-bdfinru][-t char][-k keydef][file]
72 /* These are sort types */
73 static const char OPT_STR[] ALIGN1 = "ngMucszbrdfimS:T:o:k:t:";
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 */
95 FLAG_bb = 0x80000000, /* Ignore trailing blanks */
98 #if ENABLE_FEATURE_SORT_BIG
99 static char key_separator;
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 */
107 static char *get_key(char *str, struct sort_key *key, int flags)
109 int start = start; /* for compiler */
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))
121 /* Find start of key on first pass, end on second pass */
123 for (j = 0; j < 2; j++) {
124 if (!key->range[2*j])
126 /* Loop through fields */
128 unsigned char ch = 0;
131 for (i = 1; i < key->range[2*j] + j; i++) {
133 /* Skip body of key and separator */
134 while ((ch = str[end]) != '\0') {
136 if (ch == key_separator)
140 /* Skip leading blanks */
141 while (isspace(str[end]))
143 /* Skip body of key */
144 while (str[end] != '\0') {
145 if (isspace(str[end]))
151 /* Remove last delim: "abc:def:" => "abc:def" */
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]);
162 /* Strip leading whitespace if necessary */
164 /* not using skip_whitespace() for speed */
165 while (isspace(str[start])) start++;
166 /* Strip trailing whitespace if necessary */
168 while (end > start && isspace(str[end-1])) end--;
169 /* -kSTART,N.ENDCHAR: honor ENDCHAR (1-based) */
172 if (end > len) end = len;
174 /* -kN.STARTCHAR[,...]: honor STARTCHAR (1-based) */
176 start += key->range[1] - 1;
177 if (start > len) start = len;
182 str = xstrndup(str+start, end-start);
184 if (flags & FLAG_d) {
185 for (start = end = 0; str[end]; end++)
186 if (isspace(str[end]) || isalnum(str[end]))
187 str[start++] = str[end];
191 if (flags & FLAG_i) {
192 for (start = end = 0; str[end]; end++)
193 if (isprint_asciionly(str[end]))
194 str[start++] = str[end];
199 for (i = 0; str[i]; i++)
200 str[i] = toupper(str[i]);
205 static struct sort_key *add_key(void)
207 struct sort_key **pkey = &key_list;
209 pkey = &((*pkey)->next_key);
210 return *pkey = xzalloc(sizeof(struct sort_key));
213 #define GET_LINE(fp) \
214 ((option_mask32 & FLAG_z) \
215 ? bb_get_chunk_from_file(fp, NULL) \
216 : xmalloc_fgetline(fp))
218 #define GET_LINE(fp) xmalloc_fgetline(fp)
221 /* Iterate through keys list and perform comparisons */
222 static int compare_keys(const void *xarg, const void *yarg)
224 int flags = option_mask32, retval = 0;
227 #if ENABLE_FEATURE_SORT_BIG
228 struct sort_key *key;
230 for (key = key_list; !retval && key; key = key->next_key) {
231 flags = key->flags ? key->flags : option_mask32;
232 /* Chop out and modify key chunks, handling -dfib */
233 x = get_key(*(char **)xarg, key, flags);
234 y = get_key(*(char **)yarg, key, flags);
236 /* This curly bracket serves no purpose but to match the nesting
237 * level of the for () loop we're not using */
242 /* Perform actual comparison */
243 switch (flags & (FLAG_n | FLAG_M | FLAG_g)) {
245 bb_error_msg_and_die("unknown sort type");
249 #if ENABLE_LOCALE_SUPPORT
250 retval = strcoll(x, y);
252 retval = strcmp(x, y);
255 #if ENABLE_FEATURE_SORT_BIG
258 double dx = strtod(x, &xx);
259 double dy = strtod(y, &yy);
260 /* not numbers < NaN < -infinity < numbers < +infinity) */
262 retval = (y == yy ? 0 : -1);
265 /* Check for isnan */
267 retval = (dy != dy) ? 0 : -1;
270 /* Check for infinity. Could underflow, but it avoids libm. */
271 else if (1.0 / dx == 0.0) {
273 retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1;
275 retval = (1.0 / dy == 0.0 && dy > 0) ? 0 : 1;
276 } else if (1.0 / dy == 0.0)
277 retval = (dy < 0) ? 1 : -1;
279 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
287 xx = strptime(x, "%b", &thyme);
289 yy = strptime(y, "%b", &thyme);
291 retval = (!yy) ? 0 : -1;
295 retval = dx - thyme.tm_mon;
298 /* Full floating point version of -n */
302 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
306 /* Free key copies. */
307 if (x != *(char **)xarg) free(x);
308 if (y != *(char **)yarg) free(y);
309 /* if (retval) break; - done by for () anyway */
311 /* Integer version of -n for tiny systems */
313 retval = atoi(x) - atoi(y);
319 /* Perform fallback sort if necessary */
320 if (!retval && !(option_mask32 & FLAG_s)) {
321 flags = option_mask32;
322 retval = strcmp(*(char **)xarg, *(char **)yarg);
331 #if ENABLE_FEATURE_SORT_BIG
332 static unsigned str2u(char **str)
335 if (!isdigit((*str)[0]))
336 bb_error_msg_and_die("bad field specification");
337 lu = strtoul(*str, str, 10);
338 if ((sizeof(long) > sizeof(int) && lu > INT_MAX) || !lu)
339 bb_error_msg_and_die("bad field specification");
344 int sort_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
345 int sort_main(int argc UNUSED_PARAM, char **argv)
348 char *str_ignored, *str_o, *str_t;
349 llist_t *lst_k = NULL;
354 xfunc_error_retval = 2;
356 /* Parse command line options */
357 /* -o and -t can be given at most once */
358 opt_complementary = "o--o:t--t:" /* -t, -o: at most one of each */
359 "k::"; /* -k takes list */
360 opts = getopt32(argv, OPT_STR, &str_ignored, &str_ignored, &str_o, &lst_k, &str_t);
361 /* global b strips leading and trailing spaces */
363 option_mask32 |= FLAG_bb;
364 #if ENABLE_FEATURE_SORT_BIG
366 if (!str_t[0] || str_t[1])
367 bb_error_msg_and_die("bad -t parameter");
368 key_separator = str_t[0];
370 /* note: below this point we use option_mask32, not opts,
371 * since that reduces register pressure and makes code smaller */
377 FLAG_n | /* Numeric sort */
378 FLAG_g | /* Sort using strtod() */
379 FLAG_M | /* Sort date */
380 FLAG_b | /* Ignore leading blanks */
381 FLAG_r | /* Reverse */
382 FLAG_d | /* Ignore !(isalnum()|isspace()) */
383 FLAG_f | /* Force uppercase */
384 FLAG_i | /* Ignore !isprint() */
387 struct sort_key *key = add_key();
388 char *str_k = llist_pop(&lst_k);
390 i = 0; /* i==0 before comma, 1 after (-k3,6) */
393 /* Cannot use bb_strtou - suffix can be a letter */
394 key->range[2*i] = str2u(&str_k);
397 key->range[2*i+1] = str2u(&str_k);
403 if (*str_k == ',' && !i++) {
406 } /* no else needed: fall through to syntax error
407 because comma isn't in OPT_STR */
408 idx = strchr(OPT_STR, *str_k);
410 bb_error_msg_and_die("unknown key option");
411 flag = 1 << (idx - OPT_STR);
412 if (flag & ~FLAG_allowed_for_k)
413 bb_error_msg_and_die("unknown sort type");
414 /* b after ',' means strip _trailing_ space */
415 if (i && flag == FLAG_b)
424 /* Open input files and read data */
427 *--argv = (char*)"-";
431 /* coreutils 6.9 compat: abort on first open error,
432 * do not continue to next file: */
433 FILE *fp = xfopen_stdin(*argv);
438 lines = xrealloc_vector(lines, 6, linecount);
439 lines[linecount++] = line;
441 fclose_if_not_stdin(fp);
444 #if ENABLE_FEATURE_SORT_BIG
445 /* If no key, perform alphabetic sort */
447 add_key()->range[0] = 1;
449 if (option_mask32 & FLAG_c) {
450 int j = (option_mask32 & FLAG_u) ? -1 : 0;
451 for (i = 1; i < linecount; i++) {
452 if (compare_keys(&lines[i-1], &lines[i]) > j) {
453 fprintf(stderr, "Check line %u\n", i);
460 /* Perform the actual sort */
461 qsort(lines, linecount, sizeof(lines[0]), compare_keys);
464 if (option_mask32 & FLAG_u) {
466 /* coreutils 6.3 drop lines for which only key is the same */
467 /* -- disabling last-resort compare... */
468 option_mask32 |= FLAG_s;
469 for (i = 1; i < linecount; i++) {
470 if (compare_keys(&lines[j], &lines[i]) == 0)
473 lines[++j] = lines[i];
480 #if ENABLE_FEATURE_SORT_BIG
481 /* Open output file _after_ we read all input ones */
482 if (option_mask32 & FLAG_o)
483 xmove_fd(xopen(str_o, O_WRONLY|O_CREAT|O_TRUNC), STDOUT_FILENO);
486 int ch = (option_mask32 & FLAG_z) ? '\0' : '\n';
487 for (i = 0; i < linecount; i++)
488 printf("%s%c", lines[i], ch);
491 fflush_stdout_and_exit(EXIT_SUCCESS);