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";
16 #define CUT_OPT_BYTE_FLGS (1<<0)
17 #define CUT_OPT_CHAR_FLGS (1<<1)
18 #define CUT_OPT_FIELDS_FLGS (1<<2)
19 #define CUT_OPT_DELIM_FLGS (1<<3)
20 #define CUT_OPT_SUPPRESS_FLGS (1<<4)
22 static char delim = '\t'; /* delimiter, default is tab */
35 /* growable array holding a series of lists */
36 static struct cut_list *cut_lists;
37 static unsigned int nlists; /* number of elements in above list */
40 static int cmpfunc(const void *a, const void *b)
42 return (((struct cut_list *) a)->startpos -
43 ((struct cut_list *) b)->startpos);
47 static void cut_file(FILE * file)
50 unsigned int linenum = 0; /* keep these zero-based to be consistent */
52 /* go through every line in the file */
53 while ((line = xmalloc_getline(file)) != NULL) {
55 /* set up a list so we can keep track of what's been printed */
56 char * printed = xzalloc(strlen(line) * sizeof(char));
57 char * orig_line = line;
58 unsigned int cl_pos = 0;
61 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
62 if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
63 /* print the chars specified in each cut list */
64 for (; cl_pos < nlists; cl_pos++) {
65 spos = cut_lists[cl_pos].startpos;
66 while (spos < strlen(line)) {
72 if (spos > cut_lists[cl_pos].endpos
73 || cut_lists[cl_pos].endpos == NON_RANGE)
77 } else if (delim == '\n') { /* cut by lines */
78 spos = cut_lists[cl_pos].startpos;
80 /* get out if we have no more lists to process or if the lines
81 * are lower than what we're interested in */
82 if (linenum < spos || cl_pos >= nlists)
85 /* if the line we're looking for is lower than the one we were
86 * passed, it means we displayed it already, so move on */
87 while (spos < linenum) {
89 /* go to the next list if we're at the end of this one */
90 if (spos > cut_lists[cl_pos].endpos
91 || cut_lists[cl_pos].endpos == NON_RANGE) {
93 /* get out if there's no more lists to process */
96 spos = cut_lists[cl_pos].startpos;
97 /* get out if the current line is lower than the one
98 * we just became interested in */
104 /* If we made it here, it means we've found the line we're
105 * looking for, so print it */
108 } else { /* cut by fields */
109 int ndelim = -1; /* zero-based / one-based problem */
110 int nfields_printed = 0;
112 const char delimiter[2] = { delim, 0 };
114 /* does this line contain any delimiters? */
115 if (strchr(line, delim) == NULL) {
116 if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
121 /* process each list on this line, for as long as we've got
122 * a line to process */
123 for (; cl_pos < nlists && line; cl_pos++) {
124 spos = cut_lists[cl_pos].startpos;
126 /* find the field we're looking for */
127 while (line && ndelim < spos) {
128 field = strsep(&line, delimiter);
132 /* we found it, and it hasn't been printed yet */
133 if (field && ndelim == spos && !printed[ndelim]) {
134 /* if this isn't our first time through, we need to
135 * print the delimiter after the last field that was
137 if (nfields_printed > 0)
139 fputs(field, stdout);
140 printed[ndelim] = 'X';
141 nfields_printed++; /* shouldn't overflow.. */
146 /* keep going as long as we have a line to work with,
147 * this is a list, and we're not at the end of that
149 } while (spos <= cut_lists[cl_pos].endpos && line
150 && cut_lists[cl_pos].endpos != NON_RANGE);
153 /* if we printed anything at all, we need to finish it with a
154 * newline cuz we were handed a chomped line */
163 static const char _op_on_field[] = " only when operating on fields";
165 int cut_main(int argc, char **argv)
169 opt_complementary = "b--bcf:c--bcf:f--bcf";
170 getopt32(argc, argv, optstring, &sopt, &sopt, &sopt, <ok);
171 if (!(option_mask32 & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
172 bb_error_msg_and_die("expected a list of bytes, characters, or fields");
173 if (option_mask32 & BB_GETOPT_ERROR)
174 bb_error_msg_and_die("only one type of list may be specified");
176 if (option_mask32 & CUT_OPT_DELIM_FLGS) {
177 if (strlen(ltok) > 1) {
178 bb_error_msg_and_die("the delimiter must be a single character");
183 /* non-field (char or byte) cutting has some special handling */
184 if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
185 if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) {
187 ("suppressing non-delimited lines makes sense%s",
192 ("a delimiter may be specified%s", _op_on_field);
197 * parse list and put values into startpos and endpos.
198 * valid list formats: N, N-, N-M, -M
199 * more than one list can be separated by commas
205 /* take apart the lists, one by one (they are separated with commas */
206 while ((ltok = strsep(&sopt, ",")) != NULL) {
208 /* it's actually legal to pass an empty list */
209 if (strlen(ltok) == 0)
212 /* get the start pos */
213 ntok = strsep(<ok, "-");
216 ("internal error: ntok is null for start pos!?\n");
217 } else if (strlen(ntok) == 0) {
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 */
228 ntok = strsep(<ok, "-");
231 } else if (strlen(ntok) == 0) {
235 /* if the user specified and end position of 0, that means "til the
239 e--; /* again, arrays are zero based, lines are 1 based */
244 /* if there's something left to tokenize, the user passed
247 bb_error_msg_and_die("invalid byte or field list");
249 /* add the new list */
250 cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
251 cut_lists[nlists-1].startpos = s;
252 cut_lists[nlists-1].endpos = e;
255 /* make sure we got some cut positions out of all that */
257 bb_error_msg_and_die("missing list of positions");
259 /* now that the lists are parsed, we need to sort them to make life
260 * easier on us when it comes time to print the chars / fields / lines
262 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
265 /* argv[(optind)..(argc-1)] should be names of file to process. If no
266 * files were specified or '-' was specified, take input from stdin.
267 * Otherwise, we process all the files specified. */
268 if (argv[optind] == NULL
269 || (argv[optind][0] == '-' && argv[optind][1] == '\0')) {
274 for (; optind < argc; optind++) {
275 file = bb_wfopen(argv[optind], "r");
282 if (ENABLE_FEATURE_CLEAN_UP)