Commit | Line | Data |
---|---|---|
f2287932 KB |
1 | /* |
2 | * Copyright (c) 1989 The Regents of the University of California. | |
3 | * All rights reserved. | |
4 | * | |
5 | * This code is derived from software contributed to Berkeley by | |
6 | * Adam S. Moskowitz of Menlo Consulting and Marciano Pitargue. | |
7 | * | |
f15db449 | 8 | * %sccs.include.redist.c% |
f2287932 KB |
9 | */ |
10 | ||
11 | #ifndef lint | |
12 | char copyright[] = | |
13 | "@(#) Copyright (c) 1989 The Regents of the University of California.\n\ | |
14 | All rights reserved.\n"; | |
15 | #endif /* not lint */ | |
16 | ||
17 | #ifndef lint | |
c1cea706 | 18 | static char sccsid[] = "@(#)cut.c 5.4 (Berkeley) %G%"; |
f2287932 KB |
19 | #endif /* not lint */ |
20 | ||
21 | #include <limits.h> | |
22 | #include <stdio.h> | |
23 | #include <ctype.h> | |
24 | ||
25 | int cflag; | |
26 | char dchar; | |
27 | int dflag; | |
28 | int fflag; | |
29 | int sflag; | |
30 | ||
31 | main(argc, argv) | |
32 | int argc; | |
33 | char **argv; | |
34 | { | |
35 | extern char *optarg; | |
36 | extern int errno, optind; | |
37 | FILE *fp; | |
38 | int ch, (*fcn)(), c_cut(), f_cut(); | |
39 | char *strerror(); | |
40 | ||
41 | dchar = '\t'; /* default delimiter is \t */ | |
42 | ||
43 | while ((ch = getopt(argc, argv, "c:d:f:s")) != EOF) | |
44 | switch(ch) { | |
45 | case 'c': | |
46 | fcn = c_cut; | |
47 | get_list(optarg); | |
48 | cflag = 1; | |
49 | break; | |
50 | case 'd': | |
51 | dchar = *optarg; | |
52 | dflag = 1; | |
53 | break; | |
54 | case 'f': | |
55 | get_list(optarg); | |
56 | fcn = f_cut; | |
57 | fflag = 1; | |
58 | break; | |
59 | case 's': | |
60 | sflag = 1; | |
61 | break; | |
62 | case '?': | |
63 | default: | |
64 | usage(); | |
65 | } | |
66 | argc -= optind; | |
67 | argv += optind; | |
68 | ||
69 | if (fflag) { | |
70 | if (cflag) | |
71 | usage(); | |
72 | } else if (!cflag || dflag || sflag) | |
73 | usage(); | |
74 | ||
75 | if (*argv) | |
76 | for (; *argv; ++argv) { | |
77 | if (!(fp = fopen(*argv, "r"))) { | |
78 | (void)fprintf(stderr, | |
79 | "cut: %s: %s\n", *argv, strerror(errno)); | |
80 | exit(1); | |
81 | } | |
82 | fcn(fp, *argv); | |
83 | } | |
84 | else | |
85 | fcn(stdin, "stdin"); | |
86 | exit(0); | |
87 | } | |
88 | ||
89 | int autostart, autostop, maxval; | |
90 | ||
c1cea706 | 91 | char positions[_POSIX2_LINE_MAX + 1]; |
f2287932 KB |
92 | |
93 | get_list(list) | |
94 | char *list; | |
95 | { | |
96 | register char *pos; | |
97 | register int setautostart, start, stop; | |
98 | char *p, *strtok(); | |
99 | ||
100 | /* | |
101 | * set a byte in the positions array to indicate if a field or | |
102 | * column is to be selected; use +1, it's 1-based, not 0-based. | |
103 | * This parser is less restrictive than the Draft 9 POSIX spec. | |
104 | * POSIX doesn't allow lists that aren't in increasing order or | |
105 | * overlapping lists. We also handle "-3-5" although there's no | |
106 | * real reason too. | |
107 | */ | |
108 | for (; p = strtok(list, ", \t"); list = NULL) { | |
109 | setautostart = start = stop = 0; | |
110 | if (*p == '-') { | |
111 | ++p; | |
112 | setautostart = 1; | |
113 | } | |
114 | if (isdigit(*p)) { | |
115 | start = stop = strtol(p, &p, 10); | |
116 | if (setautostart && start > autostart) | |
117 | autostart = start; | |
118 | } | |
119 | if (*p == '-') { | |
120 | if (isdigit(p[1])) | |
121 | stop = strtol(p + 1, &p, 10); | |
122 | if (*p == '-') { | |
123 | ++p; | |
124 | if (!autostop || autostop > stop) | |
125 | autostop = stop; | |
126 | } | |
127 | } | |
128 | if (*p) | |
129 | badlist("illegal list value"); | |
130 | if (!stop || !start) | |
131 | badlist("values may not include zero"); | |
c1cea706 | 132 | if (stop > _POSIX2_LINE_MAX) { |
f2287932 KB |
133 | /* positions used rather than allocate a new buffer */ |
134 | (void)sprintf(positions, "%d too large (max %d)", | |
c1cea706 | 135 | stop, _POSIX2_LINE_MAX); |
f2287932 KB |
136 | badlist(positions); |
137 | } | |
138 | if (maxval < stop) | |
139 | maxval = stop; | |
140 | for (pos = positions + start; start++ <= stop; *pos++ = 1); | |
141 | } | |
142 | ||
143 | /* overlapping ranges */ | |
144 | if (autostop && maxval > autostop) | |
145 | maxval = autostop; | |
146 | ||
147 | /* set autostart */ | |
148 | if (autostart) | |
149 | memset(positions + 1, '1', autostart); | |
150 | } | |
151 | ||
152 | /* ARGSUSED */ | |
153 | c_cut(fp, fname) | |
154 | FILE *fp; | |
155 | char *fname; | |
156 | { | |
157 | register int ch, col; | |
158 | register char *pos; | |
159 | ||
160 | for (;;) { | |
161 | pos = positions + 1; | |
162 | for (col = maxval; col; --col) { | |
163 | if ((ch = getc(fp)) == EOF) | |
164 | return; | |
165 | if (ch == '\n') | |
166 | break; | |
167 | if (*pos++) | |
168 | putchar(ch); | |
169 | } | |
170 | if (ch != '\n') | |
171 | if (autostop) | |
172 | while ((ch = getc(fp)) != EOF && ch != '\n') | |
173 | putchar(ch); | |
174 | else | |
175 | while ((ch = getc(fp)) != EOF && ch != '\n'); | |
176 | putchar('\n'); | |
177 | } | |
178 | } | |
179 | ||
180 | f_cut(fp, fname) | |
181 | FILE *fp; | |
182 | char *fname; | |
183 | { | |
184 | register int ch, field, isdelim; | |
185 | register char *pos, *p, sep; | |
186 | int output; | |
c1cea706 | 187 | char lbuf[_POSIX2_LINE_MAX + 1]; |
f2287932 KB |
188 | |
189 | for (sep = dchar, output = 0; fgets(lbuf, sizeof(lbuf), fp);) { | |
190 | for (isdelim = 0, p = lbuf;; ++p) { | |
191 | if (!(ch = *p)) { | |
192 | (void)fprintf(stderr, | |
193 | "cut: %s: line too long.\n", fname); | |
194 | exit(1); | |
195 | } | |
196 | /* this should work if newline is delimiter */ | |
197 | if (ch == sep) | |
198 | isdelim = 1; | |
199 | if (ch == '\n') { | |
200 | if (!isdelim && !sflag) | |
201 | (void)printf("%s", lbuf); | |
202 | break; | |
203 | } | |
204 | } | |
205 | if (!isdelim) | |
206 | continue; | |
207 | ||
208 | pos = positions + 1; | |
209 | for (field = maxval, p = lbuf; field; --field, ++pos) { | |
210 | if (*pos) { | |
211 | if (output++) | |
212 | putchar(sep); | |
213 | while ((ch = *p++) != '\n' && ch != sep) | |
214 | putchar(ch); | |
215 | } else | |
216 | while ((ch = *p++) != '\n' && ch != sep); | |
217 | if (ch == '\n') | |
218 | break; | |
219 | } | |
220 | if (ch != '\n') | |
221 | if (autostop) { | |
222 | if (output) | |
223 | putchar(sep); | |
224 | for (; (ch = *p) != '\n'; ++p) | |
225 | putchar(ch); | |
226 | } else | |
227 | for (; (ch = *p) != '\n'; ++p); | |
228 | putchar('\n'); | |
229 | } | |
230 | } | |
231 | ||
232 | badlist(msg) | |
233 | char *msg; | |
234 | { | |
235 | (void)fprintf(stderr, "cut: [-cf] list: %s.\n", msg); | |
236 | exit(1); | |
237 | } | |
238 | ||
239 | usage() | |
240 | { | |
241 | (void)fprintf(stderr, | |
242 | "usage:\tcut -c list [file1 ...]\n\tcut -f list [-s] [-d delim] [file ...]\n"); | |
243 | exit(1); | |
244 | } |