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 -k Sort key"
29 //usage: "\n -M Sort month"
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"
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 = 0, end = 0, len, j;
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))
119 /* Find start of key on first pass, end on second pass */
121 for (j = 0; j < 2; j++) {
122 if (!key->range[2*j])
124 /* Loop through fields */
127 for (i = 1; i < key->range[2*j] + j; i++) {
129 /* Skip body of key and separator */
131 if (str[end++] == key_separator)
135 /* Skip leading blanks */
136 while (isspace(str[end]))
138 /* Skip body of key */
140 if (isspace(str[end]))
149 /* Strip leading whitespace if necessary */
150 //XXX: skip_whitespace()
152 while (isspace(str[start])) start++;
153 /* Strip trailing whitespace if necessary */
155 while (end > start && isspace(str[end-1])) end--;
156 /* Handle offsets on start and end */
158 end += key->range[3] - 1;
159 if (end > len) end = len;
162 start += key->range[1] - 1;
163 if (start > len) start = len;
166 if (end < start) end = start;
167 str = xstrndup(str+start, end-start);
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];
176 if (flags & FLAG_i) {
177 for (start = end = 0; str[end]; end++)
178 if (isprint_asciionly(str[end]))
179 str[start++] = str[end];
184 for (i = 0; str[i]; i++)
185 str[i] = toupper(str[i]);
190 static struct sort_key *add_key(void)
192 struct sort_key **pkey = &key_list;
194 pkey = &((*pkey)->next_key);
195 return *pkey = xzalloc(sizeof(struct sort_key));
198 #define GET_LINE(fp) \
199 ((option_mask32 & FLAG_z) \
200 ? bb_get_chunk_from_file(fp, NULL) \
201 : xmalloc_fgetline(fp))
203 #define GET_LINE(fp) xmalloc_fgetline(fp)
206 /* Iterate through keys list and perform comparisons */
207 static int compare_keys(const void *xarg, const void *yarg)
209 int flags = option_mask32, retval = 0;
212 #if ENABLE_FEATURE_SORT_BIG
213 struct sort_key *key;
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);
221 /* This curly bracket serves no purpose but to match the nesting
222 level of the for () loop we're not using */
227 /* Perform actual comparison */
230 bb_error_msg_and_die("unknown sort type");
234 #if ENABLE_LOCALE_SUPPORT
235 retval = strcoll(x, y);
237 retval = strcmp(x, y);
240 #if ENABLE_FEATURE_SORT_BIG
243 double dx = strtod(x, &xx);
244 double dy = strtod(y, &yy);
245 /* not numbers < NaN < -infinity < numbers < +infinity) */
247 retval = (y == yy ? 0 : -1);
250 /* Check for isnan */
252 retval = (dy != dy) ? 0 : -1;
255 /* Check for infinity. Could underflow, but it avoids libm. */
256 else if (1.0 / dx == 0.0) {
258 retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1;
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;
264 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
272 xx = strptime(x, "%b", &thyme);
274 yy = strptime(y, "%b", &thyme);
276 retval = (!yy) ? 0 : -1;
280 retval = (dx == thyme.tm_mon) ? 0 : dx - thyme.tm_mon;
283 /* Full floating point version of -n */
287 retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0);
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 */
296 /* Integer version of -n for tiny systems */
298 retval = atoi(x) - atoi(y);
304 /* Perform fallback sort if necessary */
305 if (!retval && !(option_mask32 & FLAG_s))
306 retval = strcmp(*(char **)xarg, *(char **)yarg);
308 if (flags & FLAG_r) return -retval;
312 #if ENABLE_FEATURE_SORT_BIG
313 static unsigned str2u(char **str)
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");
325 int sort_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
326 int sort_main(int argc UNUSED_PARAM, char **argv)
329 char *str_ignored, *str_o, *str_t;
330 llist_t *lst_k = NULL;
335 xfunc_error_retval = 2;
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 */
344 option_mask32 |= FLAG_bb;
345 #if ENABLE_FEATURE_SORT_BIG
347 if (!str_t[0] || str_t[1])
348 bb_error_msg_and_die("bad -t parameter");
349 key_separator = str_t[0];
351 /* note: below this point we use option_mask32, not opts,
352 * since that reduces register pressure and makes code smaller */
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() */
368 struct sort_key *key = add_key();
369 char *str_k = llist_pop(&lst_k);
371 i = 0; /* i==0 before comma, 1 after (-k3,6) */
374 /* Cannot use bb_strtou - suffix can be a letter */
375 key->range[2*i] = str2u(&str_k);
378 key->range[2*i+1] = str2u(&str_k);
383 if (*str_k == ',' && !i++) {
386 } /* no else needed: fall through to syntax error
387 because comma isn't in OPT_STR */
388 temp2 = strchr(OPT_STR, *str_k);
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)
404 /* Open input files and read data */
407 *--argv = (char*)"-";
411 /* coreutils 6.9 compat: abort on first open error,
412 * do not continue to next file: */
413 FILE *fp = xfopen_stdin(*argv);
418 lines = xrealloc_vector(lines, 6, linecount);
419 lines[linecount++] = line;
421 fclose_if_not_stdin(fp);
424 #if ENABLE_FEATURE_SORT_BIG
425 /* if no key, perform alphabetic sort */
427 add_key()->range[0] = 1;
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);
440 /* Perform the actual sort */
441 qsort(lines, linecount, sizeof(lines[0]), compare_keys);
443 if (option_mask32 & FLAG_u) {
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)
452 lines[++flag] = lines[i];
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);
464 flag = (option_mask32 & FLAG_z) ? '\0' : '\n';
465 for (i = 0; i < linecount; i++)
466 printf("%s%c", lines[i], flag);
468 fflush_stdout_and_exit(EXIT_SUCCESS);