aboutsummaryrefslogtreecommitdiff
path: root/coreutils/cut.c
diff options
context:
space:
mode:
Diffstat (limited to 'coreutils/cut.c')
-rw-r--r--coreutils/cut.c285
1 files changed, 285 insertions, 0 deletions
diff --git a/coreutils/cut.c b/coreutils/cut.c
new file mode 100644
index 000000000..a538e3d20
--- /dev/null
+++ b/coreutils/cut.c
@@ -0,0 +1,285 @@
1/* vi: set sw=4 ts=4: */
2/*
3 * cut.c - minimalist version of cut
4 *
5 * Copyright (C) 1999,2000,2001 by Lineo, inc.
6 * Written by Mark Whitley <markw@codepoet.org>
7 * debloated by Bernhard Fischer
8 *
9 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
10 */
11
12#include "busybox.h"
13
14/* option vars */
15static 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)
21
22static char delim = '\t'; /* delimiter, default is tab */
23
24struct cut_list {
25 int startpos;
26 int endpos;
27};
28
29enum {
30 BOL = 0,
31 EOL = INT_MAX,
32 NON_RANGE = -1
33};
34
35/* growable array holding a series of lists */
36static struct cut_list *cut_lists;
37static unsigned int nlists; /* number of elements in above list */
38
39
40static int cmpfunc(const void *a, const void *b)
41{
42 return (((struct cut_list *) a)->startpos -
43 ((struct cut_list *) b)->startpos);
44
45}
46
47static void cut_file(FILE * file)
48{
49 char *line = NULL;
50 unsigned int linenum = 0; /* keep these zero-based to be consistent */
51
52 /* go through every line in the file */
53 while ((line = xmalloc_getline(file)) != NULL) {
54
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;
59 int spos;
60
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)) {
67 if (!printed[spos]) {
68 printed[spos] = 'X';
69 putchar(line[spos]);
70 }
71 spos++;
72 if (spos > cut_lists[cl_pos].endpos
73 || cut_lists[cl_pos].endpos == NON_RANGE)
74 break;
75 }
76 }
77 } else if (delim == '\n') { /* cut by lines */
78 spos = cut_lists[cl_pos].startpos;
79
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)
83 goto next_line;
84
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) {
88 spos++;
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) {
92 cl_pos++;
93 /* get out if there's no more lists to process */
94 if (cl_pos >= nlists)
95 goto next_line;
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 */
99 if (linenum < spos)
100 goto next_line;
101 }
102 }
103
104 /* If we made it here, it means we've found the line we're
105 * looking for, so print it */
106 puts(line);
107 goto next_line;
108 } else { /* cut by fields */
109 int ndelim = -1; /* zero-based / one-based problem */
110 int nfields_printed = 0;
111 char *field = NULL;
112 const char delimiter[2] = { delim, 0 };
113
114 /* does this line contain any delimiters? */
115 if (strchr(line, delim) == NULL) {
116 if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
117 puts(line);
118 goto next_line;
119 }
120
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;
125 do {
126 /* find the field we're looking for */
127 while (line && ndelim < spos) {
128 field = strsep(&line, delimiter);
129 ndelim++;
130 }
131
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
136 * printed */
137 if (nfields_printed > 0)
138 putchar(delim);
139 fputs(field, stdout);
140 printed[ndelim] = 'X';
141 nfields_printed++; /* shouldn't overflow.. */
142 }
143
144 spos++;
145
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
148 * list */
149 } while (spos <= cut_lists[cl_pos].endpos && line
150 && cut_lists[cl_pos].endpos != NON_RANGE);
151 }
152 }
153 /* if we printed anything at all, we need to finish it with a
154 * newline cuz we were handed a chomped line */
155 putchar('\n');
156 next_line:
157 linenum++;
158 free(printed);
159 free(orig_line);
160 }
161}
162
163static const char _op_on_field[] = " only when operating on fields";
164
165int cut_main(int argc, char **argv)
166{
167 char *sopt, *ltok;
168
169 opt_complementary = "b--bcf:c--bcf:f--bcf";
170 getopt32(argc, argv, optstring, &sopt, &sopt, &sopt, &ltok);
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");
175
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");
179 }
180 delim = ltok[0];
181 }
182
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) {
186 bb_error_msg_and_die
187 ("suppressing non-delimited lines makes sense%s",
188 _op_on_field);
189 }
190 if (delim != '\t') {
191 bb_error_msg_and_die
192 ("a delimiter may be specified%s", _op_on_field);
193 }
194 }
195
196 /*
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
200 */
201 {
202 char *ntok;
203 int s = 0, e = 0;
204
205 /* take apart the lists, one by one (they are separated with commas */
206 while ((ltok = strsep(&sopt, ",")) != NULL) {
207
208 /* it's actually legal to pass an empty list */
209 if (strlen(ltok) == 0)
210 continue;
211
212 /* get the start pos */
213 ntok = strsep(&ltok, "-");
214 if (ntok == NULL) {
215 bb_error_msg
216 ("internal error: ntok is null for start pos!?\n");
217 } else if (strlen(ntok) == 0) {
218 s = BOL;
219 } else {
220 s = xatoi_u(ntok);
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 */
223 if (s != 0)
224 s--;
225 }
226
227 /* get the end pos */
228 ntok = strsep(&ltok, "-");
229 if (ntok == NULL) {
230 e = NON_RANGE;
231 } else if (strlen(ntok) == 0) {
232 e = EOL;
233 } else {
234 e = xatoi_u(ntok);
235 /* if the user specified and end position of 0, that means "til the
236 * end of the line */
237 if (e == 0)
238 e = EOL;
239 e--; /* again, arrays are zero based, lines are 1 based */
240 if (e == s)
241 e = NON_RANGE;
242 }
243
244 /* if there's something left to tokenize, the user passed
245 * an invalid list */
246 if (ltok)
247 bb_error_msg_and_die("invalid byte or field list");
248
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;
253 }
254
255 /* make sure we got some cut positions out of all that */
256 if (nlists == 0)
257 bb_error_msg_and_die("missing list of positions");
258
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
261 */
262 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
263 }
264
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')) {
270 cut_file(stdin);
271 } else {
272 FILE *file;
273
274 for (; optind < argc; optind++) {
275 file = fopen_or_warn(argv[optind], "r");
276 if (file) {
277 cut_file(file);
278 fclose(file);
279 }
280 }
281 }
282 if (ENABLE_FEATURE_CLEAN_UP)
283 free(cut_lists);
284 return EXIT_SUCCESS;
285}
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-08-14 02:23:42 +0000