1 | /* cut - extract columns from a file or stdin. Author: Michael J. Holme
|
---|
2 | *
|
---|
3 | * Copyright 1989, Michael John Holme, All rights reserved.
|
---|
4 | * This code may be freely distributed, provided that this notice
|
---|
5 | * remains intact.
|
---|
6 | *
|
---|
7 | * V1.1: 6th September 1989
|
---|
8 | *
|
---|
9 | * Bugs, criticisms, etc,
|
---|
10 | * c/o Mark Powell
|
---|
11 | * JANET sq79@uk.ac.liv
|
---|
12 | * ARPA sq79%liv.ac.uk@nsfnet-relay.ac.uk
|
---|
13 | * UUCP ...!mcvax!ukc!liv.ac.uk!sq79
|
---|
14 | *-------------------------------------------------------------------------
|
---|
15 | * Changed for POSIX1003.2/Draft10 conformance
|
---|
16 | * Thomas Brupbacher (tobr@mw.lpc.ethz.ch), September 1990.
|
---|
17 | * Changes:
|
---|
18 | * - separation of error messages ( stderr) and output (stdout).
|
---|
19 | * - support for -b and -n (no effect, -b acts as -c)
|
---|
20 | * - support for -s
|
---|
21 | *-------------------------------------------------------------------------
|
---|
22 | */
|
---|
23 |
|
---|
24 | #include <sys/types.h>
|
---|
25 | #include <ctype.h>
|
---|
26 | #include <string.h>
|
---|
27 | #include <errno.h>
|
---|
28 | #include <stdlib.h>
|
---|
29 | #include <stdio.h>
|
---|
30 |
|
---|
31 | #define MAX_FIELD 80 /* Pointers to the beginning of each field
|
---|
32 | * are stored in columns[], if a line holds
|
---|
33 | * more than MAX_FIELD columns the array
|
---|
34 | * boundary is exceed. But unlikely at 80 */
|
---|
35 |
|
---|
36 | #define MAX_ARGS 32 /* Maximum number of fields following -f or
|
---|
37 | * -c switches */
|
---|
38 | int args[MAX_ARGS * 2];
|
---|
39 | int num_args;
|
---|
40 |
|
---|
41 | /* Lots of new defines, should easen maintainance... */
|
---|
42 | #define DUMP_STDIN 0 /* define for mode: no options */
|
---|
43 | #define OPTIONF 1 /* define for mode: option -f */
|
---|
44 | #define OPTIONC 2 /* define for mode: option -c */
|
---|
45 | #define OPTIONB 3 /* define for mode: option -b */
|
---|
46 | #define NOTSET 0 /* option not selected */
|
---|
47 | #define SET 1 /* option selected */
|
---|
48 |
|
---|
49 | /* Defines for the warnings */
|
---|
50 | #define DELIMITER_NOT_APPLICABLE 0
|
---|
51 | #define OVERRIDING_PREVIOUS_MODE 1
|
---|
52 | #define OPTION_NOT_APPLICABLE 2
|
---|
53 | #define UNKNOWN_OPTION 3
|
---|
54 | #define FILE_NOT_READABLE 4
|
---|
55 |
|
---|
56 | /* Defines for the fatal errors */
|
---|
57 | #define SYNTAX_ERROR 101
|
---|
58 | #define POSITION_ERROR 102
|
---|
59 | #define USAGE 103
|
---|
60 | #define LINE_TO_LONG_ERROR 104
|
---|
61 | #define RANGE_ERROR 105
|
---|
62 | #define MAX_FIELDS_EXEEDED_ERROR 106
|
---|
63 | #define MAX_ARGS_EXEEDED_ERROR 107
|
---|
64 |
|
---|
65 |
|
---|
66 | int mode; /* 0 = dump stdin to stdout, 1=-f, 2=-c */
|
---|
67 | int flag_i; /* SET = -i set on command line */
|
---|
68 | int flag_s; /* SET = -s set on command line */
|
---|
69 | char delim = '\t'; /* default delimiting character */
|
---|
70 | FILE *fd;
|
---|
71 | char *name;
|
---|
72 | char line[BUFSIZ];
|
---|
73 | int exit_status;
|
---|
74 |
|
---|
75 | _PROTOTYPE(int main, (int argc, char **argv));
|
---|
76 | _PROTOTYPE(void warn, (int warn_number, char *option));
|
---|
77 | _PROTOTYPE(void cuterror, (int err));
|
---|
78 | _PROTOTYPE(void get_args, (void));
|
---|
79 | _PROTOTYPE(void cut, (void));
|
---|
80 |
|
---|
81 | void warn(warn_number, option)
|
---|
82 | int warn_number;
|
---|
83 | char *option;
|
---|
84 | {
|
---|
85 | static char *warn_msg[] = {
|
---|
86 | "%s: Option -d allowed only with -f\n",
|
---|
87 | "%s: -%s overrides earlier option\n",
|
---|
88 | "%s: -%s not allowed in current mode\n",
|
---|
89 | "%s: Cannot open %s\n"
|
---|
90 | };
|
---|
91 |
|
---|
92 | fprintf(stderr, warn_msg[warn_number], name, option);
|
---|
93 | exit_status = warn_number + 1;
|
---|
94 |
|
---|
95 | }
|
---|
96 |
|
---|
97 | void cuterror(err)
|
---|
98 | int err;
|
---|
99 | {
|
---|
100 | static char *err_mes[] = {
|
---|
101 | "%s: syntax error\n",
|
---|
102 | "%s: position must be >0\n",
|
---|
103 | "%s: usage: cut [-f args [-i] [-d x]]|[-c args] [filename [...]]\n",
|
---|
104 | "%s: line longer than BUFSIZ\n",
|
---|
105 | "%s: range must not decrease from left to right\n",
|
---|
106 | "%s: MAX_FIELD exceeded\n",
|
---|
107 | "%s: MAX_ARGS exceeded\n"
|
---|
108 | };
|
---|
109 |
|
---|
110 | fprintf(stderr, err_mes[err - 101], name);
|
---|
111 | exit(err);
|
---|
112 | }
|
---|
113 |
|
---|
114 |
|
---|
115 | void get_args()
|
---|
116 | {
|
---|
117 | int i = 0;
|
---|
118 | int arg_ptr = 0;
|
---|
119 | int flag;
|
---|
120 |
|
---|
121 | num_args = 0;
|
---|
122 | do {
|
---|
123 | if (num_args == MAX_ARGS) cuterror(MAX_ARGS_EXEEDED_ERROR);
|
---|
124 | if (!isdigit(line[i]) && line[i] != '-') cuterror(SYNTAX_ERROR);
|
---|
125 |
|
---|
126 | args[arg_ptr] = 1;
|
---|
127 | args[arg_ptr + 1] = BUFSIZ;
|
---|
128 | flag = 1;
|
---|
129 |
|
---|
130 | while (line[i] != ',' && line[i] != 0) {
|
---|
131 | if (isdigit(line[i])) {
|
---|
132 | args[arg_ptr] = 0;
|
---|
133 | while (isdigit(line[i]))
|
---|
134 | args[arg_ptr] = 10 * args[arg_ptr] + line[i++] - '0';
|
---|
135 | if (!args[arg_ptr]) cuterror(POSITION_ERROR);
|
---|
136 | arg_ptr++;
|
---|
137 | }
|
---|
138 | if (line[i] == '-') {
|
---|
139 | arg_ptr |= 1;
|
---|
140 | i++;
|
---|
141 | flag = 0;
|
---|
142 | }
|
---|
143 | }
|
---|
144 | if (flag && arg_ptr & 1) args[arg_ptr] = args[arg_ptr - 1];
|
---|
145 | if (args[num_args * 2] > args[num_args * 2 + 1])
|
---|
146 | cuterror(RANGE_ERROR);
|
---|
147 | num_args++;
|
---|
148 | arg_ptr = num_args * 2;
|
---|
149 | }
|
---|
150 | while (line[i++]);
|
---|
151 | }
|
---|
152 |
|
---|
153 |
|
---|
154 | void cut()
|
---|
155 | {
|
---|
156 | int i, j, length, maxcol;
|
---|
157 | char *columns[MAX_FIELD];
|
---|
158 |
|
---|
159 | while (fgets(line, BUFSIZ, fd)) {
|
---|
160 | length = strlen(line) - 1;
|
---|
161 | *(line + length) = 0;
|
---|
162 | switch (mode) {
|
---|
163 | case DUMP_STDIN: printf("%s", line); break;
|
---|
164 | case OPTIONF:
|
---|
165 | maxcol = 0;
|
---|
166 | columns[maxcol++] = line;
|
---|
167 | for (i = 0; i < length; i++) {
|
---|
168 | if (*(line + i) == delim) {
|
---|
169 | *(line + i) = 0;
|
---|
170 | if (maxcol == MAX_FIELD)
|
---|
171 | cuterror(MAX_FIELDS_EXEEDED_ERROR);
|
---|
172 | columns[maxcol] = line + i + 1;
|
---|
173 | while (*(line + i + 1) == delim && flag_i) {
|
---|
174 | columns[maxcol]++;
|
---|
175 | i++;
|
---|
176 | }
|
---|
177 | maxcol++;
|
---|
178 | }
|
---|
179 | }
|
---|
180 | if (maxcol == 1) {
|
---|
181 | if (flag_s != SET) printf("%s", line);
|
---|
182 | } else {
|
---|
183 | for (i = 0; i < num_args; i++) {
|
---|
184 | for (j = args[i * 2]; j <= args[i * 2 + 1]; j++)
|
---|
185 | if (j <= maxcol) {
|
---|
186 | printf("%s", columns[j - 1]);
|
---|
187 | if (i != num_args - 1 || j != args[i * 2 + 1])
|
---|
188 | putchar(delim);
|
---|
189 | }
|
---|
190 | }
|
---|
191 | }
|
---|
192 | break;
|
---|
193 | case OPTIONC:
|
---|
194 | for (i = 0; i < num_args; i++) {
|
---|
195 | for (j = args[i * 2]; j <= (args[i * 2 + 1] > length ? length :
|
---|
196 | args[i * 2 + 1]); j++)
|
---|
197 | putchar(*(line + j - 1));
|
---|
198 | }
|
---|
199 | }
|
---|
200 | if (maxcol == 1 && flag_s == SET);
|
---|
201 | else
|
---|
202 | putchar('\n');
|
---|
203 | }
|
---|
204 | }
|
---|
205 |
|
---|
206 |
|
---|
207 | int main(argc, argv)
|
---|
208 | int argc;
|
---|
209 | char *argv[];
|
---|
210 | {
|
---|
211 | int i = 1;
|
---|
212 | int numberFilenames = 0;
|
---|
213 | name = argv[0];
|
---|
214 |
|
---|
215 | if (argc == 1) cuterror(USAGE);
|
---|
216 |
|
---|
217 | while (i < argc) {
|
---|
218 | if (argv[i][0] == '-') {
|
---|
219 | switch (argv[i++][1]) {
|
---|
220 | case 'd':
|
---|
221 | if (mode == OPTIONC || mode == OPTIONB)
|
---|
222 | warn(DELIMITER_NOT_APPLICABLE, "d");
|
---|
223 | delim = argv[i++][0];
|
---|
224 | break;
|
---|
225 | case 'f':
|
---|
226 | sprintf(line, "%s", argv[i++]);
|
---|
227 | if (mode == OPTIONC || mode == OPTIONB)
|
---|
228 | warn(OVERRIDING_PREVIOUS_MODE, "f");
|
---|
229 | mode = OPTIONF;
|
---|
230 | break;
|
---|
231 | case 'b':
|
---|
232 | sprintf(line, "%s", argv[i++]);
|
---|
233 | if (mode == OPTIONF || mode == OPTIONC)
|
---|
234 | warn(OVERRIDING_PREVIOUS_MODE, "b");
|
---|
235 | mode = OPTIONB;
|
---|
236 | break;
|
---|
237 | case 'c':
|
---|
238 | sprintf(line, "%s", argv[i++]);
|
---|
239 | if (mode == OPTIONF || mode == OPTIONB)
|
---|
240 | warn(OVERRIDING_PREVIOUS_MODE, "c");
|
---|
241 | mode = OPTIONC;
|
---|
242 | break;
|
---|
243 | case 'i': flag_i = SET; break;
|
---|
244 | case 's': flag_s = SET; break;
|
---|
245 | case '\0': /* - means: read from stdin */
|
---|
246 | numberFilenames++;
|
---|
247 | break;
|
---|
248 | case 'n': /* needed for Posix, but no effect here */
|
---|
249 | if (mode != OPTIONB)
|
---|
250 | warn(OPTION_NOT_APPLICABLE, "n");
|
---|
251 | break;
|
---|
252 | default:
|
---|
253 | warn(UNKNOWN_OPTION, &(argv[i - 1][1]));
|
---|
254 | }
|
---|
255 | } else {
|
---|
256 | i++;
|
---|
257 | numberFilenames++;
|
---|
258 | }
|
---|
259 | }
|
---|
260 |
|
---|
261 | /* Here follow the checks, if the selected options are reasonable. */
|
---|
262 | if (mode == OPTIONB) /* since in Minix char := byte */
|
---|
263 | mode = OPTIONC;
|
---|
264 | /* Flag -s is only allowed with -f, otherwise warn and reset flag_s */
|
---|
265 | if (flag_s == SET && (mode == OPTIONB || mode == OPTIONC)) {
|
---|
266 | warn(OPTION_NOT_APPLICABLE, "s");
|
---|
267 | flag_s = NOTSET;
|
---|
268 | }
|
---|
269 |
|
---|
270 | /* Flag -i is only allowed with -f, otherwise warn and reset flag_i */
|
---|
271 | if (flag_i == SET && mode == OPTIONF) {
|
---|
272 | warn(OPTION_NOT_APPLICABLE, "s");
|
---|
273 | flag_i = NOTSET;
|
---|
274 | }
|
---|
275 | get_args();
|
---|
276 | if (numberFilenames != 0) {
|
---|
277 | i = 1;
|
---|
278 | while (i < argc) {
|
---|
279 | if (argv[i][0] == '-') {
|
---|
280 | switch (argv[i][1]) {
|
---|
281 | case 'f':
|
---|
282 | case 'c':
|
---|
283 | case 'b':
|
---|
284 | case 'd': i += 2; break;
|
---|
285 | case 'n':
|
---|
286 | case 'i':
|
---|
287 | case 's': i++; break;
|
---|
288 | case '\0':
|
---|
289 | fd = stdin;
|
---|
290 | i++;
|
---|
291 | cut();
|
---|
292 | break;
|
---|
293 | default: i++;
|
---|
294 | }
|
---|
295 | } else {
|
---|
296 | if ((fd = fopen(argv[i++], "r")) == NULL) {
|
---|
297 | warn(FILE_NOT_READABLE, argv[i - 1]);
|
---|
298 | } else {
|
---|
299 | cut();
|
---|
300 | fclose(fd);
|
---|
301 | }
|
---|
302 | }
|
---|
303 | }
|
---|
304 | } else {
|
---|
305 | fd = stdin;
|
---|
306 | cut();
|
---|
307 | }
|
---|
308 |
|
---|
309 | return(exit_status);
|
---|
310 | }
|
---|