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.
15 static const char optstring[] = "b:c:f:d:sn";
17 #define CUT_OPT_BYTE_FLGS (1<<0)
18 #define CUT_OPT_CHAR_FLGS (1<<1)
19 #define CUT_OPT_FIELDS_FLGS (1<<2)
20 #define CUT_OPT_DELIM_FLGS (1<<3)
21 #define CUT_OPT_SUPPRESS_FLGS (1<<4)
24 static char delim = '\t'; /* delimiter, default is tab */
37 /* growable array holding a series of lists */
38 static struct cut_list *cut_lists;
39 static unsigned int nlists; /* number of elements in above list */
42 static int cmpfunc(const void *a, const void *b)
44 return (((struct cut_list *) a)->startpos -
45 ((struct cut_list *) b)->startpos);
49 static void cut_file(FILE * file)
52 unsigned int linenum = 0; /* keep these zero-based to be consistent */
54 /* go through every line in the file */
55 while ((line = bb_get_chomped_line_from_file(file)) != NULL) {
57 /* set up a list so we can keep track of what's been printed */
58 char * printed = xzalloc(strlen(line) * sizeof(char));
59 char * orig_line = line;
60 unsigned int cl_pos = 0;
63 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
64 if ((opt & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS))) {
65 /* print the chars specified in each cut list */
66 for (; cl_pos < nlists; cl_pos++) {
67 spos = cut_lists[cl_pos].startpos;
68 while (spos < strlen(line)) {
74 if (spos > cut_lists[cl_pos].endpos
75 || cut_lists[cl_pos].endpos == NON_RANGE)
79 } else if (delim == '\n') { /* cut by lines */
80 spos = cut_lists[cl_pos].startpos;
82 /* get out if we have no more lists to process or if the lines
83 * are lower than what we're interested in */
84 if (linenum < spos || cl_pos >= nlists)
87 /* if the line we're looking for is lower than the one we were
88 * passed, it means we displayed it already, so move on */
89 while (spos < linenum) {
91 /* go to the next list if we're at the end of this one */
92 if (spos > cut_lists[cl_pos].endpos
93 || cut_lists[cl_pos].endpos == NON_RANGE) {
95 /* get out if there's no more lists to process */
98 spos = cut_lists[cl_pos].startpos;
99 /* get out if the current line is lower than the one
100 * we just became interested in */
106 /* If we made it here, it means we've found the line we're
107 * looking for, so print it */
110 } else { /* cut by fields */
111 int ndelim = -1; /* zero-based / one-based problem */
112 int nfields_printed = 0;
114 const char delimiter[2] = { delim, 0 };
116 /* does this line contain any delimiters? */
117 if (strchr(line, delim) == NULL) {
118 if (!(opt & CUT_OPT_SUPPRESS_FLGS))
123 /* process each list on this line, for as long as we've got
124 * a line to process */
125 for (; cl_pos < nlists && line; cl_pos++) {
126 spos = cut_lists[cl_pos].startpos;
129 /* find the field we're looking for */
130 while (line && ndelim < spos) {
131 field = strsep(&line, delimiter);
135 /* we found it, and it hasn't been printed yet */
136 if (field && ndelim == spos && !printed[ndelim]) {
137 /* if this isn't our first time through, we need to
138 * print the delimiter after the last field that was
140 if (nfields_printed > 0)
142 fputs(field, stdout);
143 printed[ndelim] = 'X';
144 nfields_printed++; /* shouldn't overflow.. */
149 /* keep going as long as we have a line to work with,
150 * this is a list, and we're not at the end of that
152 } while (spos <= cut_lists[cl_pos].endpos && line
153 && cut_lists[cl_pos].endpos != NON_RANGE);
156 /* if we printed anything at all, we need to finish it with a
157 * newline cuz we were handed a chomped line */
166 static const char _op_on_field[] = " only when operating on fields";
168 int cut_main(int argc, char **argv)
172 opt_complementary = "b--bcf:c--bcf:f--bcf";
173 opt = getopt32(argc, argv, optstring, &sopt, &sopt, &sopt, <ok);
174 if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
176 ("expected a list of bytes, characters, or fields");
177 if (opt & BB_GETOPT_ERROR)
178 bb_error_msg_and_die("only one type of list may be specified");
180 if ((opt & (CUT_OPT_DELIM_FLGS))) {
181 if (strlen(ltok) > 1) {
182 bb_error_msg_and_die("the delimiter must be a single character");
187 /* non-field (char or byte) cutting has some special handling */
188 if (!(opt & CUT_OPT_FIELDS_FLGS)) {
189 if (opt & CUT_OPT_SUPPRESS_FLGS) {
191 ("suppressing non-delimited lines makes sense%s",
196 ("a delimiter may be specified%s", _op_on_field);
201 * parse list and put values into startpos and endpos.
202 * valid list formats: N, N-, N-M, -M
203 * more than one list can be separated by commas
209 /* take apart the lists, one by one (they are separated with commas */
210 while ((ltok = strsep(&sopt, ",")) != NULL) {
212 /* it's actually legal to pass an empty list */
213 if (strlen(ltok) == 0)
216 /* get the start pos */
217 ntok = strsep(<ok, "-");
220 ("internal error: ntok is null for start pos!?\n");
221 } else if (strlen(ntok) == 0) {
225 /* account for the fact that arrays are zero based, while
226 * the user expects the first char on the line to be char #1 */
231 /* get the end pos */
232 ntok = strsep(<ok, "-");
235 } else if (strlen(ntok) == 0) {
239 /* if the user specified and end position of 0, that means "til the
243 e--; /* again, arrays are zero based, lines are 1 based */
248 /* if there's something left to tokenize, the user passed
251 bb_error_msg_and_die("invalid byte or field list");
253 /* add the new list */
255 xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
256 cut_lists[nlists - 1].startpos = s;
257 cut_lists[nlists - 1].endpos = e;
260 /* make sure we got some cut positions out of all that */
262 bb_error_msg_and_die("missing list of positions");
264 /* now that the lists are parsed, we need to sort them to make life
265 * easier on us when it comes time to print the chars / fields / lines
267 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
270 /* argv[(optind)..(argc-1)] should be names of file to process. If no
271 * files were specified or '-' was specified, take input from stdin.
272 * Otherwise, we process all the files specified. */
273 if (argv[optind] == NULL
274 || (argv[optind][0] == '-' && argv[optind][1] == '\0')) {
279 for (; optind < argc; optind++) {
280 file = bb_wfopen(argv[optind], "r");
287 if (ENABLE_FEATURE_CLEAN_UP)