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 //config: bool "cut (5.8 kb)"
15 //config: cut is used to print selected parts of lines from
16 //config: each file to stdout.
18 //applet:IF_CUT(APPLET_NOEXEC(cut, cut, BB_DIR_USR_BIN, BB_SUID_DROP, cut))
20 //kbuild:lib-$(CONFIG_CUT) += cut.o
22 //usage:#define cut_trivial_usage
23 //usage: "[OPTIONS] [FILE]..."
24 //usage:#define cut_full_usage "\n\n"
25 //usage: "Print selected fields from each input FILE to stdout\n"
26 //usage: "\n -b LIST Output only bytes from LIST"
27 //usage: "\n -c LIST Output only characters from LIST"
28 //usage: "\n -d CHAR Use CHAR instead of tab as the field delimiter"
29 //usage: "\n -s Output only the lines containing delimiter"
30 //usage: "\n -f N Print only these fields"
31 //usage: "\n -n Ignored"
33 //usage:#define cut_example_usage
34 //usage: "$ echo \"Hello world\" | cut -f 1 -d ' '\n"
36 //usage: "$ echo \"Hello world\" | cut -f 2 -d ' '\n"
41 /* This is a NOEXEC applet. Be very careful! */
45 #define OPT_STR "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)
63 static int cmpfunc(const void *a, const void *b)
65 return (((struct cut_list *) a)->startpos -
66 ((struct cut_list *) b)->startpos);
69 static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists)
72 unsigned linenum = 0; /* keep these zero-based to be consistent */
74 /* go through every line in the file */
75 while ((line = xmalloc_fgetline(file)) != NULL) {
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;
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) {
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 */
104 } else if (delim == '\n') { /* cut by lines */
105 spos = cut_lists[cl_pos].startpos;
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))
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) {
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
121 /* get out if there's no more lists to process */
122 if (cl_pos >= nlists)
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)
132 /* If we made it here, it means we've found the line we're
133 * looking for, so print it */
136 } else { /* cut by fields */
137 int ndelim = -1; /* zero-based / one-based problem */
138 int nfields_printed = 0;
142 delimiter[0] = delim;
145 /* does this line contain any delimiters? */
146 if (strchr(line, delim) == NULL) {
147 if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
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;
157 /* find the field we're looking for */
158 while (line && ndelim < spos) {
159 field = strsep(&line, delimiter);
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
168 if (nfields_printed > 0)
170 fputs(field, stdout);
171 printed[ndelim] = 'X';
172 nfields_printed++; /* shouldn't overflow.. */
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
180 } while (spos <= cut_lists[cl_pos].endpos && line
181 && cut_lists[cl_pos].endpos != NON_RANGE);
184 /* if we printed anything at all, we need to finish it with a
185 * newline cuz we were handed a chomped line */
194 int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
195 int cut_main(int argc UNUSED_PARAM, char **argv)
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 */
204 opt = getopt32(argv, "^"
206 "\0" "b--bcf:c--bcf:f--bcf",
207 &sopt, &sopt, &sopt, <ok
211 if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
212 bb_simple_error_msg_and_die("expected a list of bytes, characters, or fields");
214 if (opt & CUT_OPT_DELIM_FLGS) {
215 if (ltok[0] && ltok[1]) { /* more than 1 char? */
216 bb_simple_error_msg_and_die("the delimiter must be a single character");
221 /* non-field (char or byte) cutting has some special handling */
222 if (!(opt & CUT_OPT_FIELDS_FLGS)) {
223 static const char _op_on_field[] ALIGN1 = " only when operating on fields";
225 if (opt & CUT_OPT_SUPPRESS_FLGS) {
227 ("suppressing non-delimited lines makes sense%s",
232 ("a delimiter may be specified%s", _op_on_field);
237 * parse list and put values into startpos and endpos.
238 * valid list formats: N, N-, N-M, -M
239 * more than one list can be separated by commas
245 /* take apart the lists, one by one (they are separated with commas) */
246 while ((ltok = strsep(&sopt, ",")) != NULL) {
248 /* it's actually legal to pass an empty list */
252 /* get the start pos */
253 ntok = strsep(<ok, "-");
257 s = xatoi_positive(ntok);
258 /* account for the fact that arrays are zero based, while
259 * the user expects the first char on the line to be char #1 */
264 /* get the end pos */
267 } else if (!ltok[0]) {
270 e = xatoi_positive(ltok);
271 /* if the user specified and end position of 0,
272 * that means "til the end of the line" */
275 e--; /* again, arrays are zero based, lines are 1 based */
280 /* add the new list */
281 cut_lists = xrealloc_vector(cut_lists, 4, nlists);
282 /* NB: startpos is always >= 0,
283 * while endpos may be = NON_RANGE (-1) */
284 cut_lists[nlists].startpos = s;
285 cut_lists[nlists].endpos = e;
289 /* make sure we got some cut positions out of all that */
291 bb_simple_error_msg_and_die("missing list of positions");
293 /* now that the lists are parsed, we need to sort them to make life
294 * easier on us when it comes time to print the chars / fields / lines
296 qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
300 int retval = EXIT_SUCCESS;
303 *--argv = (char *)"-";
306 FILE *file = fopen_or_warn_stdin(*argv);
308 retval = EXIT_FAILURE;
311 cut_file(file, delim, cut_lists, nlists);
312 fclose_if_not_stdin(file);
315 if (ENABLE_FEATURE_CLEAN_UP)
317 fflush_stdout_and_exit(retval);