Contents of /trunk/mkinitrd-magellan/busybox/coreutils/sort.c
Parent Directory | Revision Log
Revision 532 -
(show annotations)
(download)
Sat Sep 1 22:45:15 2007 UTC (17 years ago) by niro
File MIME type: text/plain
File size: 10551 byte(s)
Sat Sep 1 22:45:15 2007 UTC (17 years ago) by niro
File MIME type: text/plain
File size: 10551 byte(s)
-import if magellan mkinitrd; it is a fork of redhats mkinitrd-5.0.8 with all magellan patches and features; deprecates magellan-src/mkinitrd
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 "busybox.h" |
16 | |
17 | /* |
18 | sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...] |
19 | sort -c [-bdfinru][-t char][-k keydef][file] |
20 | */ |
21 | |
22 | /* These are sort types */ |
23 | static const char OPT_STR[] = "ngMucszbrdfimS:T:o:k:t:"; |
24 | enum { |
25 | FLAG_n = 1, /* Numeric sort */ |
26 | FLAG_g = 2, /* Sort using strtod() */ |
27 | FLAG_M = 4, /* Sort date */ |
28 | /* ucsz apply to root level only, not keys. b at root level implies bb */ |
29 | FLAG_u = 8, /* Unique */ |
30 | FLAG_c = 0x10, /* Check: no output, exit(!ordered) */ |
31 | FLAG_s = 0x20, /* Stable sort, no ascii fallback at end */ |
32 | FLAG_z = 0x40, /* Input is null terminated, not \n */ |
33 | /* These can be applied to search keys, the previous four can't */ |
34 | FLAG_b = 0x80, /* Ignore leading blanks */ |
35 | FLAG_r = 0x100, /* Reverse */ |
36 | FLAG_d = 0x200, /* Ignore !(isalnum()|isspace()) */ |
37 | FLAG_f = 0x400, /* Force uppercase */ |
38 | FLAG_i = 0x800, /* Ignore !isprint() */ |
39 | FLAG_m = 0x1000, /* ignored: merge already sorted files; do not sort */ |
40 | FLAG_S = 0x2000, /* ignored: -S, --buffer-size=SIZE */ |
41 | FLAG_T = 0x4000, /* ignored: -T, --temporary-directory=DIR */ |
42 | FLAG_o = 0x8000, |
43 | FLAG_k = 0x10000, |
44 | FLAG_t = 0x20000, |
45 | FLAG_bb = 0x80000000, /* Ignore trailing blanks */ |
46 | }; |
47 | |
48 | #if ENABLE_FEATURE_SORT_BIG |
49 | static char key_separator; |
50 | |
51 | static struct sort_key { |
52 | struct sort_key *next_key; /* linked list */ |
53 | unsigned range[4]; /* start word, start char, end word, end char */ |
54 | unsigned flags; |
55 | } *key_list; |
56 | |
57 | static char *get_key(char *str, struct sort_key *key, int flags) |
58 | { |
59 | int start = 0, end = 0, len, i, j; |
60 | |
61 | /* Special case whole string, so we don't have to make a copy */ |
62 | if (key->range[0] == 1 && !key->range[1] && !key->range[2] && !key->range[3] |
63 | && !(flags & (FLAG_b | FLAG_d | FLAG_f | FLAG_i | FLAG_bb)) |
64 | ) { |
65 | return str; |
66 | } |
67 | |
68 | /* Find start of key on first pass, end on second pass */ |
69 | len = strlen(str); |
70 | for (j = 0; j < 2; j++) { |
71 | if (!key->range[2*j]) |
72 | end = len; |
73 | /* Loop through fields */ |
74 | else { |
75 | end = 0; |
76 | for (i = 1; i < key->range[2*j] + j; i++) { |
77 | if (key_separator) { |
78 | /* Skip body of key and separator */ |
79 | while (str[end]) { |
80 | if (str[end++] == key_separator) |
81 | break; |
82 | } |
83 | } else { |
84 | /* Skip leading blanks */ |
85 | while (isspace(str[end])) |
86 | end++; |
87 | /* Skip body of key */ |
88 | while (str[end]) { |
89 | if (isspace(str[end])) |
90 | break; |
91 | end++; |
92 | } |
93 | } |
94 | } |
95 | } |
96 | if (!j) start = end; |
97 | } |
98 | /* Strip leading whitespace if necessary */ |
99 | //XXX: skip_whitespace() |
100 | if (flags & FLAG_b) |
101 | while (isspace(str[start])) start++; |
102 | /* Strip trailing whitespace if necessary */ |
103 | if (flags & FLAG_bb) |
104 | while (end > start && isspace(str[end-1])) end--; |
105 | /* Handle offsets on start and end */ |
106 | if (key->range[3]) { |
107 | end += key->range[3] - 1; |
108 | if (end > len) end = len; |
109 | } |
110 | if (key->range[1]) { |
111 | start += key->range[1] - 1; |
112 | if (start > len) start = len; |
113 | } |
114 | /* Make the copy */ |
115 | if (end < start) end = start; |
116 | str = xstrndup(str+start, end-start); |
117 | /* Handle -d */ |
118 | if (flags & FLAG_d) { |
119 | for (start = end = 0; str[end]; end++) |
120 | if (isspace(str[end]) || isalnum(str[end])) |
121 | str[start++] = str[end]; |
122 | str[start] = '\0'; |
123 | } |
124 | /* Handle -i */ |
125 | if (flags & FLAG_i) { |
126 | for (start = end = 0; str[end]; end++) |
127 | if (isprint(str[end])) |
128 | str[start++] = str[end]; |
129 | str[start] = '\0'; |
130 | } |
131 | /* Handle -f */ |
132 | if (flags & FLAG_f) |
133 | for (i = 0; str[i]; i++) |
134 | str[i] = toupper(str[i]); |
135 | |
136 | return str; |
137 | } |
138 | |
139 | static struct sort_key *add_key(void) |
140 | { |
141 | struct sort_key **pkey = &key_list; |
142 | while (*pkey) |
143 | pkey = &((*pkey)->next_key); |
144 | return *pkey = xzalloc(sizeof(struct sort_key)); |
145 | } |
146 | |
147 | #define GET_LINE(fp) \ |
148 | ((option_mask32 & FLAG_z) \ |
149 | ? bb_get_chunk_from_file(fp, NULL) \ |
150 | : xmalloc_getline(fp)) |
151 | #else |
152 | #define GET_LINE(fp) xmalloc_getline(fp) |
153 | #endif |
154 | |
155 | /* Iterate through keys list and perform comparisons */ |
156 | static int compare_keys(const void *xarg, const void *yarg) |
157 | { |
158 | int flags = option_mask32, retval = 0; |
159 | char *x, *y; |
160 | |
161 | #if ENABLE_FEATURE_SORT_BIG |
162 | struct sort_key *key; |
163 | |
164 | for (key = key_list; !retval && key; key = key->next_key) { |
165 | flags = key->flags ? key->flags : option_mask32; |
166 | /* Chop out and modify key chunks, handling -dfib */ |
167 | x = get_key(*(char **)xarg, key, flags); |
168 | y = get_key(*(char **)yarg, key, flags); |
169 | #else |
170 | /* This curly bracket serves no purpose but to match the nesting |
171 | level of the for () loop we're not using */ |
172 | { |
173 | x = *(char **)xarg; |
174 | y = *(char **)yarg; |
175 | #endif |
176 | /* Perform actual comparison */ |
177 | switch (flags & 7) { |
178 | default: |
179 | bb_error_msg_and_die("unknown sort type"); |
180 | break; |
181 | /* Ascii sort */ |
182 | case 0: |
183 | #if ENABLE_LOCALE_SUPPORT |
184 | retval = strcoll(x, y); |
185 | #else |
186 | retval = strcmp(x, y); |
187 | #endif |
188 | break; |
189 | #if ENABLE_FEATURE_SORT_BIG |
190 | case FLAG_g: { |
191 | char *xx, *yy; |
192 | double dx = strtod(x, &xx); |
193 | double dy = strtod(y, &yy); |
194 | /* not numbers < NaN < -infinity < numbers < +infinity) */ |
195 | if (x == xx) |
196 | retval = (y == yy ? 0 : -1); |
197 | else if (y == yy) |
198 | retval = 1; |
199 | /* Check for isnan */ |
200 | else if (dx != dx) |
201 | retval = (dy != dy) ? 0 : -1; |
202 | else if (dy != dy) |
203 | retval = 1; |
204 | /* Check for infinity. Could underflow, but it avoids libm. */ |
205 | else if (1.0 / dx == 0.0) { |
206 | if (dx < 0) |
207 | retval = (1.0 / dy == 0.0 && dy < 0) ? 0 : -1; |
208 | else |
209 | retval = (1.0 / dy == 0.0 && dy > 0) ? 0 : 1; |
210 | } else if (1.0 / dy == 0.0) |
211 | retval = (dy < 0) ? 1 : -1; |
212 | else |
213 | retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0); |
214 | break; |
215 | } |
216 | case FLAG_M: { |
217 | struct tm thyme; |
218 | int dx; |
219 | char *xx, *yy; |
220 | |
221 | xx = strptime(x, "%b", &thyme); |
222 | dx = thyme.tm_mon; |
223 | yy = strptime(y, "%b", &thyme); |
224 | if (!xx) |
225 | retval = (!yy) ? 0 : -1; |
226 | else if (!yy) |
227 | retval = 1; |
228 | else |
229 | retval = (dx == thyme.tm_mon) ? 0 : dx - thyme.tm_mon; |
230 | break; |
231 | } |
232 | /* Full floating point version of -n */ |
233 | case FLAG_n: { |
234 | double dx = atof(x); |
235 | double dy = atof(y); |
236 | retval = (dx > dy) ? 1 : ((dx < dy) ? -1 : 0); |
237 | break; |
238 | } |
239 | } /* switch */ |
240 | /* Free key copies. */ |
241 | if (x != *(char **)xarg) free(x); |
242 | if (y != *(char **)yarg) free(y); |
243 | /* if (retval) break; - done by for () anyway */ |
244 | #else |
245 | /* Integer version of -n for tiny systems */ |
246 | case FLAG_n: |
247 | retval = atoi(x) - atoi(y); |
248 | break; |
249 | } /* switch */ |
250 | #endif |
251 | } /* for */ |
252 | |
253 | /* Perform fallback sort if necessary */ |
254 | if (!retval && !(option_mask32 & FLAG_s)) |
255 | retval = strcmp(*(char **)xarg, *(char **)yarg); |
256 | |
257 | if (flags & FLAG_r) return -retval; |
258 | return retval; |
259 | } |
260 | |
261 | #if ENABLE_FEATURE_SORT_BIG |
262 | static unsigned str2u(char **str) |
263 | { |
264 | unsigned long lu; |
265 | if (!isdigit((*str)[0])) |
266 | bb_error_msg_and_die("bad field specification"); |
267 | lu = strtoul(*str, str, 10); |
268 | if ((sizeof(long) > sizeof(int) && lu > INT_MAX) || !lu) |
269 | bb_error_msg_and_die("bad field specification"); |
270 | return lu; |
271 | } |
272 | #endif |
273 | |
274 | int sort_main(int argc, char **argv) |
275 | { |
276 | FILE *fp, *outfile = stdout; |
277 | char *line, **lines = NULL; |
278 | char *str_ignored, *str_o, *str_k, *str_t; |
279 | int i, flag; |
280 | int linecount = 0; |
281 | |
282 | xfunc_error_retval = 2; |
283 | |
284 | /* Parse command line options */ |
285 | /* -o and -t can be given at most once */ |
286 | opt_complementary = "?:o--o:t--t"; |
287 | getopt32(argc, argv, OPT_STR, &str_ignored, &str_ignored, &str_o, &str_k, &str_t); |
288 | #if ENABLE_FEATURE_SORT_BIG |
289 | if (option_mask32 & FLAG_o) outfile = xfopen(str_o, "w"); |
290 | if (option_mask32 & FLAG_t) { |
291 | if (!str_t[0] || str_t[1]) |
292 | bb_error_msg_and_die("bad -t parameter"); |
293 | key_separator = str_t[0]; |
294 | } |
295 | /* parse sort key */ |
296 | if (option_mask32 & FLAG_k) { |
297 | enum { |
298 | FLAG_allowed_for_k = |
299 | FLAG_n | /* Numeric sort */ |
300 | FLAG_g | /* Sort using strtod() */ |
301 | FLAG_M | /* Sort date */ |
302 | FLAG_b | /* Ignore leading blanks */ |
303 | FLAG_r | /* Reverse */ |
304 | FLAG_d | /* Ignore !(isalnum()|isspace()) */ |
305 | FLAG_f | /* Force uppercase */ |
306 | FLAG_i | /* Ignore !isprint() */ |
307 | 0 |
308 | }; |
309 | struct sort_key *key = add_key(); |
310 | const char *temp2; |
311 | |
312 | i = 0; /* i==0 before comma, 1 after (-k3,6) */ |
313 | while (*str_k) { |
314 | /* Start of range */ |
315 | /* Cannot use bb_strtou - suffix can be a letter */ |
316 | key->range[2*i] = str2u(&str_k); |
317 | if (*str_k == '.') { |
318 | str_k++; |
319 | key->range[2*i+1] = str2u(&str_k); |
320 | } |
321 | while (*str_k) { |
322 | if (*str_k == ',' && !i++) { |
323 | str_k++; |
324 | break; |
325 | } /* no else needed: fall through to syntax error |
326 | because comma isn't in OPT_STR */ |
327 | temp2 = strchr(OPT_STR, *str_k); |
328 | if (!temp2) |
329 | bb_error_msg_and_die("unknown key option"); |
330 | flag = 1 << (temp2 - OPT_STR); |
331 | if (flag & ~FLAG_allowed_for_k) |
332 | bb_error_msg_and_die("unknown sort type"); |
333 | /* b after ',' means strip _trailing_ space */ |
334 | if (i && flag == FLAG_b) flag = FLAG_bb; |
335 | key->flags |= flag; |
336 | str_k++; |
337 | } |
338 | } |
339 | } |
340 | #endif |
341 | /* global b strips leading and trailing spaces */ |
342 | if (option_mask32 & FLAG_b) option_mask32 |= FLAG_bb; |
343 | |
344 | /* Open input files and read data */ |
345 | for (i = argv[optind] ? optind : optind-1; argv[i]; i++) { |
346 | fp = stdin; |
347 | if (i >= optind && NOT_LONE_DASH(argv[i])) |
348 | fp = xfopen(argv[i], "r"); |
349 | for (;;) { |
350 | line = GET_LINE(fp); |
351 | if (!line) break; |
352 | if (!(linecount & 63)) |
353 | lines = xrealloc(lines, sizeof(char *) * (linecount + 64)); |
354 | lines[linecount++] = line; |
355 | } |
356 | fclose(fp); |
357 | } |
358 | #if ENABLE_FEATURE_SORT_BIG |
359 | /* if no key, perform alphabetic sort */ |
360 | if (!key_list) |
361 | add_key()->range[0] = 1; |
362 | /* handle -c */ |
363 | if (option_mask32 & FLAG_c) { |
364 | int j = (option_mask32 & FLAG_u) ? -1 : 0; |
365 | for (i = 1; i < linecount; i++) |
366 | if (compare_keys(&lines[i-1], &lines[i]) > j) { |
367 | fprintf(stderr, "Check line %d\n", i); |
368 | return 1; |
369 | } |
370 | return 0; |
371 | } |
372 | #endif |
373 | /* Perform the actual sort */ |
374 | qsort(lines, linecount, sizeof(char *), compare_keys); |
375 | /* handle -u */ |
376 | if (option_mask32 & FLAG_u) { |
377 | flag = 0; |
378 | /* coreutils 6.3 drop lines for which only key is the same */ |
379 | /* -- disabling last-resort compare... */ |
380 | option_mask32 |= FLAG_s; |
381 | for (i = 1; i < linecount; i++) { |
382 | if (!compare_keys(&lines[flag], &lines[i])) |
383 | free(lines[i]); |
384 | else |
385 | lines[++flag] = lines[i]; |
386 | } |
387 | if (linecount) linecount = flag+1; |
388 | } |
389 | /* Print it */ |
390 | for (i = 0; i < linecount; i++) |
391 | fprintf(outfile, "%s\n", lines[i]); |
392 | |
393 | fflush_stdout_and_exit(EXIT_SUCCESS); |
394 | } |