xref: /original-bsd/usr.bin/uniq/uniq.c (revision 13deef58)
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  * Case Larsen.
7  *
8  * %sccs.include.redist.c%
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
18 static char sccsid[] = "@(#)uniq.c	5.5 (Berkeley) 07/15/92";
19 #endif /* not lint */
20 
21 #include <errno.h>
22 #include <stdio.h>
23 #include <ctype.h>
24 #include <stdlib.h>
25 #include <string.h>
26 
27 #define	MAXLINELEN	(8 * 1024)
28 
29 int cflag, dflag, uflag;
30 int numchars, numfields, repeats;
31 
32 void	 err __P((const char *, ...));
33 FILE	*file __P((char *, char *));
34 void	 show __P((FILE *, char *));
35 char	*skip __P((char *));
36 void	 obsolete __P((char *[]));
37 void	 usage __P((void));
38 
39 int
40 main (argc, argv)
41 	int argc;
42 	char *argv[];
43 {
44 	register char *t1, *t2;
45 	FILE *ifp, *ofp;
46 	int ch;
47 	char *prevline, *thisline, *p;
48 
49 	obsolete(argv);
50 	while ((ch = getopt(argc, argv, "-cdf:s:u")) != EOF)
51 		switch (ch) {
52 		case '-':
53 			--optind;
54 			goto done;
55 		case 'c':
56 			cflag = 1;
57 			break;
58 		case 'd':
59 			dflag = 1;
60 			break;
61 		case 'f':
62 			numfields = strtol(optarg, &p, 10);
63 			if (numfields < 0 || *p)
64 				err("illegal field skip value: %s", optarg);
65 			break;
66 		case 's':
67 			numchars = strtol(optarg, &p, 10);
68 			if (numchars < 0 || *p)
69 				err("illegal character skip value: %s", optarg);
70 			break;
71 		case 'u':
72 			uflag = 1;
73 			break;
74 		case '?':
75 		default:
76 			usage();
77 	}
78 
79 done:	argc -= optind;
80 	argv +=optind;
81 
82 	/* If no flags are set, default is -d -u. */
83 	if (cflag) {
84 		if (dflag || uflag)
85 			usage();
86 	} else if (!dflag && !uflag)
87 		dflag = uflag = 1;
88 
89 	switch(argc) {
90 	case 0:
91 		ifp = stdin;
92 		ofp = stdout;
93 		break;
94 	case 1:
95 		ifp = file(argv[0], "r");
96 		ofp = stdout;
97 		break;
98 	case 2:
99 		ifp = file(argv[0], "r");
100 		ofp = file(argv[1], "w");
101 		break;
102 	default:
103 		usage();
104 	}
105 
106 	prevline = malloc(MAXLINELEN);
107 	thisline = malloc(MAXLINELEN);
108 	if (prevline == NULL || thisline == NULL)
109 		err("%s", strerror(errno));
110 
111 	if (fgets(prevline, MAXLINELEN, ifp) == NULL)
112 		exit(0);
113 
114 	while (fgets(thisline, MAXLINELEN, ifp)) {
115 		/* If requested get the chosen fields + character offsets. */
116 		if (numfields || numchars) {
117 			t1 = skip(thisline);
118 			t2 = skip(prevline);
119 		} else {
120 			t1 = thisline;
121 			t2 = prevline;
122 		}
123 
124 		/* If different, print; set previous to new value. */
125 		if (strcmp(t1, t2)) {
126 			show(ofp, prevline);
127 			t1 = prevline;
128 			prevline = thisline;
129 			thisline = t1;
130 			repeats = 0;
131 		} else
132 			++repeats;
133 	}
134 	show(ofp, prevline);
135 	exit(0);
136 }
137 
138 /*
139  * show --
140  *	Output a line depending on the flags and number of repetitions
141  *	of the line.
142  */
143 void
144 show(ofp, str)
145 	FILE *ofp;
146 	char *str;
147 {
148 	if (cflag)
149 		(void)fprintf(ofp, "%4d %s", repeats + 1, str);
150 	if (dflag && repeats || uflag && !repeats)
151 		(void)fprintf(ofp, "%s", str);
152 }
153 
154 char *
155 skip(str)
156 	register char *str;
157 {
158 	register int infield, nchars, nfields;
159 
160 	for (nfields = numfields, infield = 0; nfields && *str; ++str)
161 		if (isspace(*str)) {
162 			if (infield) {
163 				infield = 0;
164 				--nfields;
165 			}
166 		} else if (!infield)
167 			infield = 1;
168 	for (nchars = numchars; nchars-- && *str; ++str);
169 	return(str);
170 }
171 
172 FILE *
173 file(name, mode)
174 	char *name, *mode;
175 {
176 	FILE *fp;
177 
178 	if ((fp = fopen(name, mode)) == NULL)
179 		err("%s: %s", name, strerror(errno));
180 	return(fp);
181 }
182 
183 void
184 obsolete(argv)
185 	char *argv[];
186 {
187 	int len;
188 	char *ap, *p, *start;
189 
190 	while (ap = *++argv) {
191 		/* Return if "--" or not an option of any form. */
192 		if (ap[0] != '-') {
193 			if (ap[0] != '+')
194 				return;
195 		} else if (ap[1] == '-')
196 			return;
197 		if (!isdigit(ap[1]))
198 			continue;
199 		/*
200 		 * Digit signifies an old-style option.  Malloc space for dash,
201 		 * new option and argument.
202 		 */
203 		len = strlen(ap);
204 		if ((start = p = malloc(len + 3)) == NULL)
205 			err("%s", strerror(errno));
206 		*p++ = '-';
207 		*p++ = ap[0] == '+' ? 's' : 'f';
208 		(void)strcpy(p, ap + 1);
209 		*argv = start;
210 	}
211 }
212 
213 void
214 usage()
215 {
216 	(void)fprintf(stderr,
217 	    "usage: uniq [-c | -du] [-f fields] [-s chars] [input [output]]\n");
218 	exit(1);
219 }
220 
221 #if __STDC__
222 #include <stdarg.h>
223 #else
224 #include <varargs.h>
225 #endif
226 
227 void
228 #if __STDC__
229 err(const char *fmt, ...)
230 #else
231 err(fmt, va_alist)
232 	char *fmt;
233         va_dcl
234 #endif
235 {
236 	va_list ap;
237 #if __STDC__
238 	va_start(ap, fmt);
239 #else
240 	va_start(ap);
241 #endif
242 	(void)fprintf(stderr, "uniq: ");
243 	(void)vfprintf(stderr, fmt, ap);
244 	va_end(ap);
245 	(void)fprintf(stderr, "\n");
246 	exit(1);
247 	/* NOTREACHED */
248 }
249