1 /* vi: set sw=4 ts=4: */
3 * cut.c - minimalist version of cut
5 * Copyright (C) 1999,2000,2001 by Lineo, inc.
6 * Written by Mark Whitley <markw@codepoet.org>
7 * debloated by Bernhard Reutner-Fischer
9 * Licensed under GPLv2 or later, see file LICENSE in this source tree.
12 //usage:#define cut_trivial_usage
13 //usage: "[OPTIONS] [FILE]..."
14 //usage:#define cut_full_usage "\n\n"
15 //usage: "Print selected fields from each input FILE to stdout\n"
17 //usage: "\n -b LIST Output only bytes from LIST"
18 //usage: "\n -c LIST Output only characters from LIST"
19 //usage: "\n -d CHAR Use CHAR instead of tab as the field delimiter"
20 //usage: "\n -s Output only the lines containing delimiter"
21 //usage: "\n -f N Print only these fields"
22 //usage: "\n -n Ignored"
24 //usage:#define cut_example_usage
25 //usage: "$ echo \"Hello world\" | cut -f 1 -d ' '\n"
27 //usage: "$ echo \"Hello world\" | cut -f 2 -d ' '\n"
32 /* This is a NOEXEC applet. Be very careful! */
36 static const char optstring[] ALIGN1 = "b:c:f:d:sn";
37 #define CUT_OPT_BYTE_FLGS (1 << 0)
38 #define CUT_OPT_CHAR_FLGS (1 << 1)
39 #define CUT_OPT_FIELDS_FLGS (1 << 2)
40 #define CUT_OPT_DELIM_FLGS (1 << 3)
41 #define CUT_OPT_SUPPRESS_FLGS (1 << 4)
54 static int cmpfunc(const void *a, const void *b)
56 return (((struct cut_list *) a)->startpos -
57 ((struct cut_list *) b)->startpos);
60 static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists)
63 unsigned linenum = 0; /* keep these zero-based to be consistent */
65 /* go through every line in the file */
66 while ((line = xmalloc_fgetline(file)) != NULL) {
68 /* set up a list so we can keep track of what's been printed */
69 int linelen = strlen(line);
70 char *printed = xzalloc(linelen + 1);
71 char *orig_line = line;
75 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
76 if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
77 /* print the chars specified in each cut list */
78 for (; cl_pos < nlists; cl_pos++) {
79 spos = cut_lists[cl_pos].startpos;
80 while (spos < linelen) {
86 if (spos > cut_lists[cl_pos].endpos
87 /* NON_RANGE is -1, so if below is true,
88 * the above was true too (spos is >= 0) */
89 /* || cut_lists[cl_pos].endpos == NON_RANGE */
95 } else if (delim == '\n') { /* cut by lines */
96 spos = cut_lists[cl_pos].startpos;
98 /* get out if we have no more lists to process or if the lines
99 * are lower than what we're interested in */
100 if (((int)linenum < spos) || (cl_pos >= nlists))
103 /* if the line we're looking for is lower than the one we were
104 * passed, it means we displayed it already, so move on */
105 while (spos < (int)linenum) {
107 /* go to the next list if we're at the end of this one */
108 if (spos > cut_lists[cl_pos].endpos
109 || cut_lists[cl_pos].endpos == NON_RANGE
112 /* get out if there's no more lists to process */
113 if (cl_pos >= nlists)
115 spos = cut_lists[cl_pos].startpos;
116 /* get out if the current line is lower than the one
117 * we just became interested in */
118 if ((int)linenum < spos)
123 /* If we made it here, it means we've found the line we're
124 * looking for, so print it */
127 } else { /* cut by fields */
128 int ndelim = -1; /* zero-based / one-based problem */
129 int nfields_printed = 0;
133 delimiter[0] = delim;
136 /* does this line contain any delimiters? */
137 if (strchr(line, delim) == NULL) {
138 if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
143 /* process each list on this line, for as long as we've got
144 * a line to process */
145 for (; cl_pos < nlists && line; cl_pos++) {
146 spos = cut_lists[cl_pos].startpos;
148 /* find the field we're looking for */
149 while (line && ndelim < spos) {
150 field = strsep(&line, delimiter);
154 /* we found it, and it hasn't been printed yet */
155 if (field && ndelim == spos && !printed[ndelim]) {
156 /* if this isn't our first time through, we need to
157 * print the delimiter after the last field that was
159 if (nfields_printed > 0)
161 fputs(field, stdout);
162 printed[ndelim] = 'X';
163 nfields_printed++; /* shouldn't overflow.. */
168 /* keep going as long as we have a line to work with,
169 * this is a list, and we're not at the end of that
171 } while (spos <= cut_lists[cl_pos].endpos && line
172 && cut_lists[cl_pos].endpos != NON_RANGE);
175 /* if we printed anything at all, we need to finish it with a
176 * newline cuz we were handed a chomped line */
185 int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
186 int cut_main(int argc UNUSED_PARAM, char **argv)
188 /* growable array holding a series of lists */
189 struct cut_list *cut_lists = NULL;
190 unsigned nlists = 0; /* number of elements in above list */
191 char delim = '\t'; /* delimiter, default is tab */
195 opt_complementary = "b--bcf:c--bcf:f--bcf";
196 opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok);
199 if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
200 bb_error_msg_and_die("expected a list of bytes, characters, or fields");
202 if (opt & CUT_OPT_DELIM_FLGS) {
203 if (ltok[0] && ltok[1]) { /* more than 1 char? */
204 bb_error_msg_and_die("the delimiter must be a single character");
209 /* non-field (char or byte) cutting has some special handling */
210 if (!(opt & CUT_OPT_FIELDS_FLGS)) {
211 static const char _op_on_field[] ALIGN1 = " only when operating on fields";
213 if (opt & CUT_OPT_SUPPRESS_FLGS) {
215 ("suppressing non-delimited lines makes sense%s",
220 ("a delimiter may be specified%s", _op_on_field);
225 * parse list and put values into startpos and endpos.
226 * valid list formats: N, N-, N-M, -M
227 * more than one list can be separated by commas
233 /* take apart the lists, one by one (they are separated with commas) */
234 while ((ltok = strsep(&sopt, ",")) != NULL) {
236 /* it's actually legal to pass an empty list */
240 /* get the start pos */
241 ntok = strsep(<ok, "-");
245 s = xatoi_positive(ntok);
246 /* account for the fact that arrays are zero based, while
247 * the user expects the first char on the line to be char #1 */
252 /* get the end pos */
255 } else if (!ltok[0]) {
258 e = xatoi_positive(ltok);
259 /* if the user specified and end position of 0,
260 * that means "til the end of the line" */
263 e--; /* again, arrays are zero based, lines are 1 based */
268 /* add the new list */
269 cut_lists = xrealloc_vector(cut_lists, 4, nlists);
270 /* NB: startpos is always >= 0,
271 * while endpos may be = NON_RANGE (-1) */
272 cut_lists[nlists].startpos = s;
273 cut_lists[nlists].endpos = e;
277 /* make sure we got some cut positions out of all that */
279 bb_error_msg_and_die("missing list of positions");
281 /* now that the lists are parsed, we need to sort them to make life
282 * easier on us when it comes time to print the chars / fields / lines
284 qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
288 int retval = EXIT_SUCCESS;
291 *--argv = (char *)"-";
294 FILE *file = fopen_or_warn_stdin(*argv);
296 retval = EXIT_FAILURE;
299 cut_file(file, delim, cut_lists, nlists);
300 fclose_if_not_stdin(file);
303 if (ENABLE_FEATURE_CLEAN_UP)
305 fflush_stdout_and_exit(retval);