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>
8 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
20 static const char optstring[] = "b:c:f:d:sn";
21 #define OPT_BYTE_FLGS 1
22 #define OPT_CHAR_FLGS 2
23 #define OPT_FIELDS_FLGS 4
24 #define OPT_DELIM_FLGS 8
25 #define OPT_SUPRESS_FLGS 16
26 static char part; /* (b)yte, (c)har, (f)ields */
27 static unsigned int supress_non_delimited_lines;
28 static char delim = '\t'; /* delimiter, default is tab */
41 static struct cut_list *cut_lists = NULL; /* growable array holding a series of lists */
42 static unsigned int nlists = 0; /* number of elements in above list */
45 static int cmpfunc(const void *a, const void *b)
47 struct cut_list *la = (struct cut_list *)a;
48 struct cut_list *lb = (struct cut_list *)b;
50 if (la->startpos > lb->startpos)
52 if (la->startpos < lb->startpos)
59 * parse_lists() - parses a list and puts values into startpos and endpos.
60 * valid list formats: N, N-, N-M, -M
61 * more than one list can be separated by commas
63 static void parse_lists(char *lists)
70 /* take apart the lists, one by one (they are separated with commas */
71 while ((ltok = strsep(&lists, ",")) != NULL) {
73 /* it's actually legal to pass an empty list */
74 if (strlen(ltok) == 0)
77 /* get the start pos */
78 ntok = strsep(<ok, "-");
80 fprintf(stderr, "Help ntok is null for starting position! What do I do?\n");
81 } else if (strlen(ntok) == 0) {
84 s = strtoul(ntok, &junk, 10);
85 if(*junk != '\0' || s < 0)
86 bb_error_msg_and_die("invalid byte or field list");
88 /* account for the fact that arrays are zero based, while the user
89 * expects the first char on the line to be char # 1 */
95 ntok = strsep(<ok, "-");
98 } else if (strlen(ntok) == 0) {
101 e = strtoul(ntok, &junk, 10);
102 if(*junk != '\0' || e < 0)
103 bb_error_msg_and_die("invalid byte or field list");
104 /* if the user specified and end position of 0, that means "til the
108 e--; /* again, arrays are zero based, lines are 1 based */
113 /* if there's something left to tokenize, the user past an invalid list */
115 bb_error_msg_and_die("invalid byte or field list");
117 /* add the new list */
118 cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
119 cut_lists[nlists-1].startpos = s;
120 cut_lists[nlists-1].endpos = e;
123 /* make sure we got some cut positions out of all that */
125 bb_error_msg_and_die("missing list of positions");
127 /* now that the lists are parsed, we need to sort them to make life easier
128 * on us when it comes time to print the chars / fields / lines */
129 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
134 static void cut_line_by_chars(const char *line)
137 /* set up a list so we can keep track of what's been printed */
138 char *printed = xzalloc(strlen(line));
140 /* print the chars specified in each cut list */
141 for (c = 0; c < nlists; c++) {
142 l = cut_lists[c].startpos;
143 while (l < strlen(line)) {
149 if (cut_lists[c].endpos == NON_RANGE || l > cut_lists[c].endpos)
153 putchar('\n'); /* cuz we were handed a chomped line */
158 static void cut_line_by_fields(char *line)
161 int ndelim = -1; /* zero-based / one-based problem */
162 int nfields_printed = 0;
164 char d[2] = { delim, 0 };
167 /* test the easy case first: does this line contain any delimiters? */
168 if (strchr(line, delim) == NULL) {
169 if (!supress_non_delimited_lines)
174 /* set up a list so we can keep track of what's been printed */
175 printed = xzalloc(strlen(line));
177 /* process each list on this line, for as long as we've got a line to process */
178 for (c = 0; c < nlists && line; c++) {
179 f = cut_lists[c].startpos;
182 /* find the field we're looking for */
183 while (line && ndelim < f) {
184 field = strsep(&line, d);
188 /* we found it, and it hasn't been printed yet */
189 if (field && ndelim == f && !printed[ndelim]) {
190 /* if this isn't our first time through, we need to print the
191 * delimiter after the last field that was printed */
192 if (nfields_printed > 0)
194 fputs(field, stdout);
195 printed[ndelim] = 'X';
201 /* keep going as long as we have a line to work with, this is a
202 * list, and we're not at the end of that list */
203 } while (line && cut_lists[c].endpos != NON_RANGE && f <= cut_lists[c].endpos);
206 /* if we printed anything at all, we need to finish it with a newline cuz
207 * we were handed a chomped line */
214 static void cut_file_by_lines(const char *line, unsigned int linenum)
219 /* I can't initialize this above cuz the "initializer isn't
220 * constant" *sigh* */
222 l = cut_lists[c].startpos;
224 /* get out if we have no more lists to process or if the lines are lower
225 * than what we're interested in */
226 if (c >= nlists || linenum < l)
229 /* if the line we're looking for is lower than the one we were passed, it
230 * means we displayed it already, so move on */
231 while (l < linenum) {
233 /* move on to the next list if we're at the end of this one */
234 if (cut_lists[c].endpos == NON_RANGE || l > cut_lists[c].endpos) {
236 /* get out if there's no more lists to process */
239 l = cut_lists[c].startpos;
240 /* get out if the current line is lower than the one we just became
247 /* If we made it here, it means we've found the line we're looking for, so print it */
255 static void cut_file(FILE *file)
258 unsigned int linenum = 0; /* keep these zero-based to be consistent */
260 /* go through every line in the file */
261 while ((line = bb_get_chomped_line_from_file(file)) != NULL) {
263 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
264 if ((part & (OPT_CHAR_FLGS | OPT_BYTE_FLGS)))
265 cut_line_by_chars(line);
267 /* cut based on fields */
270 cut_file_by_lines(line, linenum);
272 cut_line_by_fields(line);
281 int cut_main(int argc, char **argv)
286 bb_opt_complementally = "b--bcf:c--bcf:f--bcf";
287 opt = bb_getopt_ulflags(argc, argv, optstring, &sopt, &sopt, &sopt, &sdopt);
288 part = opt & (OPT_BYTE_FLGS|OPT_CHAR_FLGS|OPT_FIELDS_FLGS);
290 bb_error_msg_and_die("you must specify a list of bytes, characters, or fields");
291 if(opt & BB_GETOPT_ERROR)
292 bb_error_msg_and_die("only one type of list may be specified");
294 if((opt & (OPT_DELIM_FLGS))) {
295 if (strlen(sdopt) > 1) {
296 bb_error_msg_and_die("the delimiter must be a single character");
300 supress_non_delimited_lines = opt & OPT_SUPRESS_FLGS;
302 /* non-field (char or byte) cutting has some special handling */
303 if (part != OPT_FIELDS_FLGS) {
304 if (supress_non_delimited_lines) {
305 bb_error_msg_and_die("suppressing non-delimited lines makes sense"
306 " only when operating on fields");
309 bb_error_msg_and_die("a delimiter may be specified only when operating on fields");
313 /* argv[(optind)..(argc-1)] should be names of file to process. If no
314 * files were specified or '-' was specified, take input from stdin.
315 * Otherwise, we process all the files specified. */
316 if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
322 for (i = optind; i < argc; i++) {
323 file = bb_wfopen(argv[i], "r");