blob: 84449c775c812cfab64e160edfd85434d1b0ebd5
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 Reutner-Fischer |
8 | * |
9 | * Licensed under GPLv2 or later, see file LICENSE in this source tree. |
10 | */ |
11 | |
12 | //usage:#define cut_trivial_usage |
13 | //usage: "[OPTIONS] [FILE]..." |
14 | //usage:#define cut_full_usage "\n\n" |
15 | //usage: "Print selected fields from each input FILE to stdout\n" |
16 | //usage: "\n -b LIST Output only bytes from LIST" |
17 | //usage: "\n -c LIST Output only characters from LIST" |
18 | //usage: "\n -d CHAR Use CHAR instead of tab as the field delimiter" |
19 | //usage: "\n -s Output only the lines containing delimiter" |
20 | //usage: "\n -f N Print only these fields" |
21 | //usage: "\n -n Ignored" |
22 | //usage: |
23 | //usage:#define cut_example_usage |
24 | //usage: "$ echo \"Hello world\" | cut -f 1 -d ' '\n" |
25 | //usage: "Hello\n" |
26 | //usage: "$ echo \"Hello world\" | cut -f 2 -d ' '\n" |
27 | //usage: "world\n" |
28 | |
29 | #include "libbb.h" |
30 | |
31 | /* This is a NOEXEC applet. Be very careful! */ |
32 | |
33 | |
34 | /* option vars */ |
35 | static const char optstring[] ALIGN1 = "b:c:f:d:sn"; |
36 | #define CUT_OPT_BYTE_FLGS (1 << 0) |
37 | #define CUT_OPT_CHAR_FLGS (1 << 1) |
38 | #define CUT_OPT_FIELDS_FLGS (1 << 2) |
39 | #define CUT_OPT_DELIM_FLGS (1 << 3) |
40 | #define CUT_OPT_SUPPRESS_FLGS (1 << 4) |
41 | |
42 | struct cut_list { |
43 | int startpos; |
44 | int endpos; |
45 | }; |
46 | |
47 | enum { |
48 | BOL = 0, |
49 | EOL = INT_MAX, |
50 | NON_RANGE = -1 |
51 | }; |
52 | |
53 | static int cmpfunc(const void *a, const void *b) |
54 | { |
55 | return (((struct cut_list *) a)->startpos - |
56 | ((struct cut_list *) b)->startpos); |
57 | } |
58 | |
59 | static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists) |
60 | { |
61 | char *line; |
62 | unsigned linenum = 0; /* keep these zero-based to be consistent */ |
63 | |
64 | /* go through every line in the file */ |
65 | while ((line = xmalloc_fgetline(file)) != NULL) { |
66 | |
67 | /* set up a list so we can keep track of what's been printed */ |
68 | int linelen = strlen(line); |
69 | char *printed = xzalloc(linelen + 1); |
70 | char *orig_line = line; |
71 | unsigned cl_pos = 0; |
72 | int spos; |
73 | |
74 | /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */ |
75 | if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) { |
76 | /* print the chars specified in each cut list */ |
77 | for (; cl_pos < nlists; cl_pos++) { |
78 | spos = cut_lists[cl_pos].startpos; |
79 | while (spos < linelen) { |
80 | if (!printed[spos]) { |
81 | printed[spos] = 'X'; |
82 | putchar(line[spos]); |
83 | } |
84 | spos++; |
85 | if (spos > cut_lists[cl_pos].endpos |
86 | /* NON_RANGE is -1, so if below is true, |
87 | * the above was true too (spos is >= 0) */ |
88 | /* || cut_lists[cl_pos].endpos == NON_RANGE */ |
89 | ) { |
90 | break; |
91 | } |
92 | } |
93 | } |
94 | } else if (delim == '\n') { /* cut by lines */ |
95 | spos = cut_lists[cl_pos].startpos; |
96 | |
97 | /* get out if we have no more lists to process or if the lines |
98 | * are lower than what we're interested in */ |
99 | if (((int)linenum < spos) || (cl_pos >= nlists)) |
100 | goto next_line; |
101 | |
102 | /* if the line we're looking for is lower than the one we were |
103 | * passed, it means we displayed it already, so move on */ |
104 | while (spos < (int)linenum) { |
105 | spos++; |
106 | /* go to the next list if we're at the end of this one */ |
107 | if (spos > cut_lists[cl_pos].endpos |
108 | || cut_lists[cl_pos].endpos == NON_RANGE |
109 | ) { |
110 | cl_pos++; |
111 | /* get out if there's no more lists to process */ |
112 | if (cl_pos >= nlists) |
113 | goto next_line; |
114 | spos = cut_lists[cl_pos].startpos; |
115 | /* get out if the current line is lower than the one |
116 | * we just became interested in */ |
117 | if ((int)linenum < spos) |
118 | goto next_line; |
119 | } |
120 | } |
121 | |
122 | /* If we made it here, it means we've found the line we're |
123 | * looking for, so print it */ |
124 | puts(line); |
125 | goto next_line; |
126 | } else { /* cut by fields */ |
127 | int ndelim = -1; /* zero-based / one-based problem */ |
128 | int nfields_printed = 0; |
129 | char *field = NULL; |
130 | char delimiter[2]; |
131 | |
132 | delimiter[0] = delim; |
133 | delimiter[1] = 0; |
134 | |
135 | /* does this line contain any delimiters? */ |
136 | if (strchr(line, delim) == NULL) { |
137 | if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS)) |
138 | puts(line); |
139 | goto next_line; |
140 | } |
141 | |
142 | /* process each list on this line, for as long as we've got |
143 | * a line to process */ |
144 | for (; cl_pos < nlists && line; cl_pos++) { |
145 | spos = cut_lists[cl_pos].startpos; |
146 | do { |
147 | /* find the field we're looking for */ |
148 | while (line && ndelim < spos) { |
149 | field = strsep(&line, delimiter); |
150 | ndelim++; |
151 | } |
152 | |
153 | /* we found it, and it hasn't been printed yet */ |
154 | if (field && ndelim == spos && !printed[ndelim]) { |
155 | /* if this isn't our first time through, we need to |
156 | * print the delimiter after the last field that was |
157 | * printed */ |
158 | if (nfields_printed > 0) |
159 | putchar(delim); |
160 | fputs(field, stdout); |
161 | printed[ndelim] = 'X'; |
162 | nfields_printed++; /* shouldn't overflow.. */ |
163 | } |
164 | |
165 | spos++; |
166 | |
167 | /* keep going as long as we have a line to work with, |
168 | * this is a list, and we're not at the end of that |
169 | * list */ |
170 | } while (spos <= cut_lists[cl_pos].endpos && line |
171 | && cut_lists[cl_pos].endpos != NON_RANGE); |
172 | } |
173 | } |
174 | /* if we printed anything at all, we need to finish it with a |
175 | * newline cuz we were handed a chomped line */ |
176 | putchar('\n'); |
177 | next_line: |
178 | linenum++; |
179 | free(printed); |
180 | free(orig_line); |
181 | } |
182 | } |
183 | |
184 | int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
185 | int cut_main(int argc UNUSED_PARAM, char **argv) |
186 | { |
187 | /* growable array holding a series of lists */ |
188 | struct cut_list *cut_lists = NULL; |
189 | unsigned nlists = 0; /* number of elements in above list */ |
190 | char delim = '\t'; /* delimiter, default is tab */ |
191 | char *sopt, *ltok; |
192 | unsigned opt; |
193 | |
194 | opt_complementary = "b--bcf:c--bcf:f--bcf"; |
195 | opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok); |
196 | // argc -= optind; |
197 | argv += optind; |
198 | if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS))) |
199 | bb_error_msg_and_die("expected a list of bytes, characters, or fields"); |
200 | |
201 | if (opt & CUT_OPT_DELIM_FLGS) { |
202 | if (ltok[0] && ltok[1]) { /* more than 1 char? */ |
203 | bb_error_msg_and_die("the delimiter must be a single character"); |
204 | } |
205 | delim = ltok[0]; |
206 | } |
207 | |
208 | /* non-field (char or byte) cutting has some special handling */ |
209 | if (!(opt & CUT_OPT_FIELDS_FLGS)) { |
210 | static const char _op_on_field[] ALIGN1 = " only when operating on fields"; |
211 | |
212 | if (opt & CUT_OPT_SUPPRESS_FLGS) { |
213 | bb_error_msg_and_die |
214 | ("suppressing non-delimited lines makes sense%s", |
215 | _op_on_field); |
216 | } |
217 | if (delim != '\t') { |
218 | bb_error_msg_and_die |
219 | ("a delimiter may be specified%s", _op_on_field); |
220 | } |
221 | } |
222 | |
223 | /* |
224 | * parse list and put values into startpos and endpos. |
225 | * valid list formats: N, N-, N-M, -M |
226 | * more than one list can be separated by commas |
227 | */ |
228 | { |
229 | char *ntok; |
230 | int s = 0, e = 0; |
231 | |
232 | /* take apart the lists, one by one (they are separated with commas) */ |
233 | while ((ltok = strsep(&sopt, ",")) != NULL) { |
234 | |
235 | /* it's actually legal to pass an empty list */ |
236 | if (!ltok[0]) |
237 | continue; |
238 | |
239 | /* get the start pos */ |
240 | ntok = strsep(<ok, "-"); |
241 | if (!ntok[0]) { |
242 | s = BOL; |
243 | } else { |
244 | s = xatoi_positive(ntok); |
245 | /* account for the fact that arrays are zero based, while |
246 | * the user expects the first char on the line to be char #1 */ |
247 | if (s != 0) |
248 | s--; |
249 | } |
250 | |
251 | /* get the end pos */ |
252 | if (ltok == NULL) { |
253 | e = NON_RANGE; |
254 | } else if (!ltok[0]) { |
255 | e = EOL; |
256 | } else { |
257 | e = xatoi_positive(ltok); |
258 | /* if the user specified and end position of 0, |
259 | * that means "til the end of the line" */ |
260 | if (e == 0) |
261 | e = EOL; |
262 | e--; /* again, arrays are zero based, lines are 1 based */ |
263 | if (e == s) |
264 | e = NON_RANGE; |
265 | } |
266 | |
267 | /* add the new list */ |
268 | cut_lists = xrealloc_vector(cut_lists, 4, nlists); |
269 | /* NB: startpos is always >= 0, |
270 | * while endpos may be = NON_RANGE (-1) */ |
271 | cut_lists[nlists].startpos = s; |
272 | cut_lists[nlists].endpos = e; |
273 | nlists++; |
274 | } |
275 | |
276 | /* make sure we got some cut positions out of all that */ |
277 | if (nlists == 0) |
278 | bb_error_msg_and_die("missing list of positions"); |
279 | |
280 | /* now that the lists are parsed, we need to sort them to make life |
281 | * easier on us when it comes time to print the chars / fields / lines |
282 | */ |
283 | qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc); |
284 | } |
285 | |
286 | { |
287 | int retval = EXIT_SUCCESS; |
288 | |
289 | if (!*argv) |
290 | *--argv = (char *)"-"; |
291 | |
292 | do { |
293 | FILE *file = fopen_or_warn_stdin(*argv); |
294 | if (!file) { |
295 | retval = EXIT_FAILURE; |
296 | continue; |
297 | } |
298 | cut_file(file, delim, cut_lists, nlists); |
299 | fclose_if_not_stdin(file); |
300 | } while (*++argv); |
301 | |
302 | if (ENABLE_FEATURE_CLEAN_UP) |
303 | free(cut_lists); |
304 | fflush_stdout_and_exit(retval); |
305 | } |
306 | } |
307 |