blob: a33a825f8c7ce919128adbf271d7f255233f6e2c
1 | /* vi: set sw=4 ts=4: */ |
2 | /* |
3 | * cut.c - minimalist version of cut |
4 | * |
5 | * Copyright (C) 1999,2000,2001 by Lineo, inc. |
6 | * Written by Mark Whitley <markw@codepoet.org> |
7 | * debloated by Bernhard Reutner-Fischer |
8 | * |
9 | * Licensed under GPLv2 or later, see file LICENSE in this source tree. |
10 | */ |
11 | //config:config CUT |
12 | //config: bool "cut" |
13 | //config: default y |
14 | //config: help |
15 | //config: cut is used to print selected parts of lines from |
16 | //config: each file to stdout. |
17 | |
18 | //applet:IF_CUT(APPLET_NOEXEC(cut, cut, BB_DIR_USR_BIN, BB_SUID_DROP, cut)) |
19 | |
20 | //kbuild:lib-$(CONFIG_CUT) += cut.o |
21 | |
22 | //usage:#define cut_trivial_usage |
23 | //usage: "[OPTIONS] [FILE]..." |
24 | //usage:#define cut_full_usage "\n\n" |
25 | //usage: "Print selected fields from each input FILE to stdout\n" |
26 | //usage: "\n -b LIST Output only bytes from LIST" |
27 | //usage: "\n -c LIST Output only characters from LIST" |
28 | //usage: "\n -d CHAR Use CHAR instead of tab as the field delimiter" |
29 | //usage: "\n -s Output only the lines containing delimiter" |
30 | //usage: "\n -f N Print only these fields" |
31 | //usage: "\n -n Ignored" |
32 | //usage: |
33 | //usage:#define cut_example_usage |
34 | //usage: "$ echo \"Hello world\" | cut -f 1 -d ' '\n" |
35 | //usage: "Hello\n" |
36 | //usage: "$ echo \"Hello world\" | cut -f 2 -d ' '\n" |
37 | //usage: "world\n" |
38 | |
39 | #include "libbb.h" |
40 | |
41 | /* This is a NOEXEC applet. Be very careful! */ |
42 | |
43 | |
44 | /* option vars */ |
45 | static const char optstring[] ALIGN1 = "b:c:f:d:sn"; |
46 | #define CUT_OPT_BYTE_FLGS (1 << 0) |
47 | #define CUT_OPT_CHAR_FLGS (1 << 1) |
48 | #define CUT_OPT_FIELDS_FLGS (1 << 2) |
49 | #define CUT_OPT_DELIM_FLGS (1 << 3) |
50 | #define CUT_OPT_SUPPRESS_FLGS (1 << 4) |
51 | |
52 | struct cut_list { |
53 | int startpos; |
54 | int endpos; |
55 | }; |
56 | |
57 | enum { |
58 | BOL = 0, |
59 | EOL = INT_MAX, |
60 | NON_RANGE = -1 |
61 | }; |
62 | |
63 | static int cmpfunc(const void *a, const void *b) |
64 | { |
65 | return (((struct cut_list *) a)->startpos - |
66 | ((struct cut_list *) b)->startpos); |
67 | } |
68 | |
69 | static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists) |
70 | { |
71 | char *line; |
72 | unsigned linenum = 0; /* keep these zero-based to be consistent */ |
73 | |
74 | /* go through every line in the file */ |
75 | while ((line = xmalloc_fgetline(file)) != NULL) { |
76 | |
77 | /* set up a list so we can keep track of what's been printed */ |
78 | int linelen = strlen(line); |
79 | char *printed = xzalloc(linelen + 1); |
80 | char *orig_line = line; |
81 | unsigned cl_pos = 0; |
82 | int spos; |
83 | |
84 | /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */ |
85 | if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) { |
86 | /* print the chars specified in each cut list */ |
87 | for (; cl_pos < nlists; cl_pos++) { |
88 | spos = cut_lists[cl_pos].startpos; |
89 | while (spos < linelen) { |
90 | if (!printed[spos]) { |
91 | printed[spos] = 'X'; |
92 | putchar(line[spos]); |
93 | } |
94 | spos++; |
95 | if (spos > cut_lists[cl_pos].endpos |
96 | /* NON_RANGE is -1, so if below is true, |
97 | * the above was true too (spos is >= 0) */ |
98 | /* || cut_lists[cl_pos].endpos == NON_RANGE */ |
99 | ) { |
100 | break; |
101 | } |
102 | } |
103 | } |
104 | } else if (delim == '\n') { /* cut by lines */ |
105 | spos = cut_lists[cl_pos].startpos; |
106 | |
107 | /* get out if we have no more lists to process or if the lines |
108 | * are lower than what we're interested in */ |
109 | if (((int)linenum < spos) || (cl_pos >= nlists)) |
110 | goto next_line; |
111 | |
112 | /* if the line we're looking for is lower than the one we were |
113 | * passed, it means we displayed it already, so move on */ |
114 | while (spos < (int)linenum) { |
115 | spos++; |
116 | /* go to the next list if we're at the end of this one */ |
117 | if (spos > cut_lists[cl_pos].endpos |
118 | || cut_lists[cl_pos].endpos == NON_RANGE |
119 | ) { |
120 | cl_pos++; |
121 | /* get out if there's no more lists to process */ |
122 | if (cl_pos >= nlists) |
123 | goto next_line; |
124 | spos = cut_lists[cl_pos].startpos; |
125 | /* get out if the current line is lower than the one |
126 | * we just became interested in */ |
127 | if ((int)linenum < spos) |
128 | goto next_line; |
129 | } |
130 | } |
131 | |
132 | /* If we made it here, it means we've found the line we're |
133 | * looking for, so print it */ |
134 | puts(line); |
135 | goto next_line; |
136 | } else { /* cut by fields */ |
137 | int ndelim = -1; /* zero-based / one-based problem */ |
138 | int nfields_printed = 0; |
139 | char *field = NULL; |
140 | char delimiter[2]; |
141 | |
142 | delimiter[0] = delim; |
143 | delimiter[1] = 0; |
144 | |
145 | /* does this line contain any delimiters? */ |
146 | if (strchr(line, delim) == NULL) { |
147 | if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS)) |
148 | puts(line); |
149 | goto next_line; |
150 | } |
151 | |
152 | /* process each list on this line, for as long as we've got |
153 | * a line to process */ |
154 | for (; cl_pos < nlists && line; cl_pos++) { |
155 | spos = cut_lists[cl_pos].startpos; |
156 | do { |
157 | /* find the field we're looking for */ |
158 | while (line && ndelim < spos) { |
159 | field = strsep(&line, delimiter); |
160 | ndelim++; |
161 | } |
162 | |
163 | /* we found it, and it hasn't been printed yet */ |
164 | if (field && ndelim == spos && !printed[ndelim]) { |
165 | /* if this isn't our first time through, we need to |
166 | * print the delimiter after the last field that was |
167 | * printed */ |
168 | if (nfields_printed > 0) |
169 | putchar(delim); |
170 | fputs(field, stdout); |
171 | printed[ndelim] = 'X'; |
172 | nfields_printed++; /* shouldn't overflow.. */ |
173 | } |
174 | |
175 | spos++; |
176 | |
177 | /* keep going as long as we have a line to work with, |
178 | * this is a list, and we're not at the end of that |
179 | * list */ |
180 | } while (spos <= cut_lists[cl_pos].endpos && line |
181 | && cut_lists[cl_pos].endpos != NON_RANGE); |
182 | } |
183 | } |
184 | /* if we printed anything at all, we need to finish it with a |
185 | * newline cuz we were handed a chomped line */ |
186 | putchar('\n'); |
187 | next_line: |
188 | linenum++; |
189 | free(printed); |
190 | free(orig_line); |
191 | } |
192 | } |
193 | |
194 | int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
195 | int cut_main(int argc UNUSED_PARAM, char **argv) |
196 | { |
197 | /* growable array holding a series of lists */ |
198 | struct cut_list *cut_lists = NULL; |
199 | unsigned nlists = 0; /* number of elements in above list */ |
200 | char delim = '\t'; /* delimiter, default is tab */ |
201 | char *sopt, *ltok; |
202 | unsigned opt; |
203 | |
204 | opt_complementary = "b--bcf:c--bcf:f--bcf"; |
205 | opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok); |
206 | // argc -= optind; |
207 | argv += optind; |
208 | if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS))) |
209 | bb_error_msg_and_die("expected a list of bytes, characters, or fields"); |
210 | |
211 | if (opt & CUT_OPT_DELIM_FLGS) { |
212 | if (ltok[0] && ltok[1]) { /* more than 1 char? */ |
213 | bb_error_msg_and_die("the delimiter must be a single character"); |
214 | } |
215 | delim = ltok[0]; |
216 | } |
217 | |
218 | /* non-field (char or byte) cutting has some special handling */ |
219 | if (!(opt & CUT_OPT_FIELDS_FLGS)) { |
220 | static const char _op_on_field[] ALIGN1 = " only when operating on fields"; |
221 | |
222 | if (opt & CUT_OPT_SUPPRESS_FLGS) { |
223 | bb_error_msg_and_die |
224 | ("suppressing non-delimited lines makes sense%s", |
225 | _op_on_field); |
226 | } |
227 | if (delim != '\t') { |
228 | bb_error_msg_and_die |
229 | ("a delimiter may be specified%s", _op_on_field); |
230 | } |
231 | } |
232 | |
233 | /* |
234 | * parse list and put values into startpos and endpos. |
235 | * valid list formats: N, N-, N-M, -M |
236 | * more than one list can be separated by commas |
237 | */ |
238 | { |
239 | char *ntok; |
240 | int s = 0, e = 0; |
241 | |
242 | /* take apart the lists, one by one (they are separated with commas) */ |
243 | while ((ltok = strsep(&sopt, ",")) != NULL) { |
244 | |
245 | /* it's actually legal to pass an empty list */ |
246 | if (!ltok[0]) |
247 | continue; |
248 | |
249 | /* get the start pos */ |
250 | ntok = strsep(<ok, "-"); |
251 | if (!ntok[0]) { |
252 | s = BOL; |
253 | } else { |
254 | s = xatoi_positive(ntok); |
255 | /* account for the fact that arrays are zero based, while |
256 | * the user expects the first char on the line to be char #1 */ |
257 | if (s != 0) |
258 | s--; |
259 | } |
260 | |
261 | /* get the end pos */ |
262 | if (ltok == NULL) { |
263 | e = NON_RANGE; |
264 | } else if (!ltok[0]) { |
265 | e = EOL; |
266 | } else { |
267 | e = xatoi_positive(ltok); |
268 | /* if the user specified and end position of 0, |
269 | * that means "til the end of the line" */ |
270 | if (e == 0) |
271 | e = EOL; |
272 | e--; /* again, arrays are zero based, lines are 1 based */ |
273 | if (e == s) |
274 | e = NON_RANGE; |
275 | } |
276 | |
277 | /* add the new list */ |
278 | cut_lists = xrealloc_vector(cut_lists, 4, nlists); |
279 | /* NB: startpos is always >= 0, |
280 | * while endpos may be = NON_RANGE (-1) */ |
281 | cut_lists[nlists].startpos = s; |
282 | cut_lists[nlists].endpos = e; |
283 | nlists++; |
284 | } |
285 | |
286 | /* make sure we got some cut positions out of all that */ |
287 | if (nlists == 0) |
288 | bb_error_msg_and_die("missing list of positions"); |
289 | |
290 | /* now that the lists are parsed, we need to sort them to make life |
291 | * easier on us when it comes time to print the chars / fields / lines |
292 | */ |
293 | qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc); |
294 | } |
295 | |
296 | { |
297 | int retval = EXIT_SUCCESS; |
298 | |
299 | if (!*argv) |
300 | *--argv = (char *)"-"; |
301 | |
302 | do { |
303 | FILE *file = fopen_or_warn_stdin(*argv); |
304 | if (!file) { |
305 | retval = EXIT_FAILURE; |
306 | continue; |
307 | } |
308 | cut_file(file, delim, cut_lists, nlists); |
309 | fclose_if_not_stdin(file); |
310 | } while (*++argv); |
311 | |
312 | if (ENABLE_FEATURE_CLEAN_UP) |
313 | free(cut_lists); |
314 | fflush_stdout_and_exit(retval); |
315 | } |
316 | } |
317 |