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 Fischer
9 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
14 /* This is a NOEXEC applet. Be very careful! */
18 static const char optstring[] ALIGN1 = "b:c:f:d:sn";
19 #define CUT_OPT_BYTE_FLGS (1 << 0)
20 #define CUT_OPT_CHAR_FLGS (1 << 1)
21 #define CUT_OPT_FIELDS_FLGS (1 << 2)
22 #define CUT_OPT_DELIM_FLGS (1 << 3)
23 #define CUT_OPT_SUPPRESS_FLGS (1 << 4)
36 /* growable array holding a series of lists */
37 static struct cut_list *cut_lists;
38 static unsigned int nlists; /* number of elements in above list */
41 static int cmpfunc(const void *a, const void *b)
43 return (((struct cut_list *) a)->startpos -
44 ((struct cut_list *) b)->startpos);
48 static void cut_file(FILE *file, char delim)
51 unsigned int linenum = 0; /* keep these zero-based to be consistent */
53 /* go through every line in the file */
54 while ((line = xmalloc_getline(file)) != NULL) {
56 /* set up a list so we can keep track of what's been printed */
57 char * printed = xzalloc(strlen(line) * sizeof(char));
58 char * orig_line = line;
59 unsigned int cl_pos = 0;
62 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
63 if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
64 /* print the chars specified in each cut list */
65 for (; cl_pos < nlists; cl_pos++) {
66 spos = cut_lists[cl_pos].startpos;
67 while (spos < strlen(line)) {
73 if (spos > cut_lists[cl_pos].endpos
74 || cut_lists[cl_pos].endpos == NON_RANGE)
78 } else if (delim == '\n') { /* cut by lines */
79 spos = cut_lists[cl_pos].startpos;
81 /* get out if we have no more lists to process or if the lines
82 * are lower than what we're interested in */
83 if (linenum < spos || cl_pos >= nlists)
86 /* if the line we're looking for is lower than the one we were
87 * passed, it means we displayed it already, so move on */
88 while (spos < linenum) {
90 /* go to the next list if we're at the end of this one */
91 if (spos > cut_lists[cl_pos].endpos
92 || cut_lists[cl_pos].endpos == NON_RANGE) {
94 /* get out if there's no more lists to process */
97 spos = cut_lists[cl_pos].startpos;
98 /* get out if the current line is lower than the one
99 * we just became interested in */
105 /* If we made it here, it means we've found the line we're
106 * looking for, so print it */
109 } else { /* cut by fields */
110 int ndelim = -1; /* zero-based / one-based problem */
111 int nfields_printed = 0;
113 const char delimiter[2] = { delim, 0 };
115 /* does this line contain any delimiters? */
116 if (strchr(line, delim) == NULL) {
117 if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
122 /* process each list on this line, for as long as we've got
123 * a line to process */
124 for (; cl_pos < nlists && line; cl_pos++) {
125 spos = cut_lists[cl_pos].startpos;
127 /* find the field we're looking for */
128 while (line && ndelim < spos) {
129 field = strsep(&line, delimiter);
133 /* we found it, and it hasn't been printed yet */
134 if (field && ndelim == spos && !printed[ndelim]) {
135 /* if this isn't our first time through, we need to
136 * print the delimiter after the last field that was
138 if (nfields_printed > 0)
140 fputs(field, stdout);
141 printed[ndelim] = 'X';
142 nfields_printed++; /* shouldn't overflow.. */
147 /* keep going as long as we have a line to work with,
148 * this is a list, and we're not at the end of that
150 } while (spos <= cut_lists[cl_pos].endpos && line
151 && cut_lists[cl_pos].endpos != NON_RANGE);
154 /* if we printed anything at all, we need to finish it with a
155 * newline cuz we were handed a chomped line */
164 int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
165 int cut_main(int argc ATTRIBUTE_UNUSED, char **argv)
167 char delim = '\t'; /* delimiter, default is tab */
170 opt_complementary = "b--bcf:c--bcf:f--bcf";
171 getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok);
174 if (!(option_mask32 & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
175 bb_error_msg_and_die("expected a list of bytes, characters, or fields");
177 if (option_mask32 & CUT_OPT_DELIM_FLGS) {
178 if (ltok[0] && ltok[1]) { /* more than 1 char? */
179 bb_error_msg_and_die("the delimiter must be a single character");
184 /* non-field (char or byte) cutting has some special handling */
185 if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
186 static const char _op_on_field[] ALIGN1 = " only when operating on fields";
188 if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) {
190 ("suppressing non-delimited lines makes sense%s",
195 ("a delimiter may be specified%s", _op_on_field);
200 * parse list and put values into startpos and endpos.
201 * valid list formats: N, N-, N-M, -M
202 * more than one list can be separated by commas
208 /* take apart the lists, one by one (they are separated with commas) */
209 while ((ltok = strsep(&sopt, ",")) != NULL) {
211 /* it's actually legal to pass an empty list */
215 /* get the start pos */
216 ntok = strsep(<ok, "-");
221 /* account for the fact that arrays are zero based, while
222 * the user expects the first char on the line to be char #1 */
227 /* get the end pos */
230 } else if (!ltok[0]) {
234 /* if the user specified and end position of 0, that means "til the
238 e--; /* again, arrays are zero based, lines are 1 based */
243 /* add the new list */
244 cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
245 cut_lists[nlists-1].startpos = s;
246 cut_lists[nlists-1].endpos = e;
249 /* make sure we got some cut positions out of all that */
251 bb_error_msg_and_die("missing list of positions");
253 /* now that the lists are parsed, we need to sort them to make life
254 * easier on us when it comes time to print the chars / fields / lines
256 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
260 int retval = EXIT_SUCCESS;
263 *--argv = (char *)"-";
266 FILE *file = fopen_or_warn_stdin(*argv);
268 retval = EXIT_FAILURE;
271 cut_file(file, delim);
272 fclose_if_not_stdin(file);
275 if (ENABLE_FEATURE_CLEAN_UP)
277 fflush_stdout_and_exit(retval);