xref: /netbsd-src/usr.bin/look/look.c (revision 9fbd88883c38d0c0fbfcbe66d76fe6b0fab3f9de)
1 /*	$NetBSD: look.c,v 1.10 1998/02/20 09:26:00 mycroft Exp $	*/
2 
3 /*-
4  * Copyright (c) 1991, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * David Hitz of Auspex Systems, Inc.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *	This product includes software developed by the University of
21  *	California, Berkeley and its contributors.
22  * 4. Neither the name of the University nor the names of its contributors
23  *    may be used to endorse or promote products derived from this software
24  *    without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36  * SUCH DAMAGE.
37  */
38 
39 #include <sys/cdefs.h>
40 #ifndef lint
41 __COPYRIGHT("@(#) Copyright (c) 1991, 1993\n\
42 	The Regents of the University of California.  All rights reserved.\n");
43 #endif /* not lint */
44 
45 #ifndef lint
46 #if 0
47 static char sccsid[] = "@(#)look.c	8.2 (Berkeley) 5/4/95";
48 #endif
49 __RCSID("$NetBSD: look.c,v 1.10 1998/02/20 09:26:00 mycroft Exp $");
50 #endif /* not lint */
51 
52 /*
53  * look -- find lines in a sorted list.
54  *
55  * The man page said that TABs and SPACEs participate in -d comparisons.
56  * In fact, they were ignored.  This implements historic practice, not
57  * the manual page.
58  */
59 
60 #include <sys/types.h>
61 #include <sys/mman.h>
62 #include <sys/stat.h>
63 
64 #include <ctype.h>
65 #include <errno.h>
66 #include <fcntl.h>
67 #include <limits.h>
68 #include <stdio.h>
69 #include <stdlib.h>
70 #include <string.h>
71 #include <unistd.h>
72 #include <err.h>
73 
74 #include "pathnames.h"
75 
76 /*
77  * FOLD and DICT convert characters to a normal form for comparison,
78  * according to the user specified flags.
79  *
80  * DICT expects integers because it uses a non-character value to
81  * indicate a character which should not participate in comparisons.
82  */
83 #define	EQUAL		0
84 #define	GREATER		1
85 #define	LESS		(-1)
86 #define NO_COMPARE	(-2)
87 
88 #define	FOLD(c)	(isascii(c) && isupper(c) ? tolower(c) : (c))
89 #define	DICT(c)	(isascii(c) && isalnum(c) ? (c) : NO_COMPARE)
90 
91 int dflag, fflag;
92 
93 char	*binary_search __P((char *, char *, char *));
94 int	 compare __P((char *, char *, char *));
95 char	*linear_search __P((char *, char *, char *));
96 int	 look __P((char *, char *, char *));
97 int	 main __P((int, char **));
98 void	 print_from __P((char *, char *, char *));
99 void	 usage __P((void));
100 
101 int
102 main(argc, argv)
103 	int argc;
104 	char *argv[];
105 {
106 	struct stat sb;
107 	int ch, fd, termchar;
108 	char *back, *file, *front, *string, *p;
109 
110 	string = NULL;
111 	file = _PATH_WORDS;
112 	termchar = '\0';
113 	while ((ch = getopt(argc, argv, "dft:")) != -1)
114 		switch(ch) {
115 		case 'd':
116 			dflag = 1;
117 			break;
118 		case 'f':
119 			fflag = 1;
120 			break;
121 		case 't':
122 			termchar = *optarg;
123 			break;
124 		case '?':
125 		default:
126 			usage();
127 		}
128 	argc -= optind;
129 	argv += optind;
130 
131 	switch (argc) {
132 	case 2:				/* Don't set -df for user. */
133 		string = *argv++;
134 		file = *argv;
135 		break;
136 	case 1:				/* But set -df by default. */
137 		dflag = fflag = 1;
138 		string = *argv;
139 		break;
140 	default:
141 		usage();
142 	}
143 
144 	if (termchar != '\0' && (p = strchr(string, termchar)) != NULL)
145 		*++p = '\0';
146 
147 	if ((fd = open(file, O_RDONLY, 0)) < 0 || fstat(fd, &sb))
148 		err(2, "%s", file);
149 	if (sb.st_size > SIZE_T_MAX)
150 		err(2, "%s: %s", file, strerror(EFBIG));
151 	if ((front = mmap(NULL, (size_t)sb.st_size,
152 	    PROT_READ, MAP_FILE|MAP_SHARED, fd, (off_t)0)) == NULL)
153 		err(2, "%s", file);
154 	back = front + sb.st_size;
155 	exit(look(string, front, back));
156 }
157 
158 int
159 look(string, front, back)
160 	char *string, *front, *back;
161 {
162 	int ch;
163 	char *readp, *writep;
164 
165 	/* Reformat string string to avoid doing it multiple times later. */
166 	for (readp = writep = string; (ch = *readp++) != 0; ) {
167 		if (fflag)
168 			ch = FOLD(ch);
169 		if (dflag)
170 			ch = DICT(ch);
171 		if (ch != NO_COMPARE)
172 			*(writep++) = ch;
173 	}
174 	*writep = '\0';
175 
176 	front = binary_search(string, front, back);
177 	front = linear_search(string, front, back);
178 
179 	if (front)
180 		print_from(string, front, back);
181 	return (front ? 0 : 1);
182 }
183 
184 
185 /*
186  * Binary search for "string" in memory between "front" and "back".
187  *
188  * This routine is expected to return a pointer to the start of a line at
189  * *or before* the first word matching "string".  Relaxing the constraint
190  * this way simplifies the algorithm.
191  *
192  * Invariants:
193  * 	front points to the beginning of a line at or before the first
194  *	matching string.
195  *
196  * 	back points to the beginning of a line at or after the first
197  *	matching line.
198  *
199  * Base of the Invariants.
200  * 	front = NULL;
201  *	back = EOF;
202  *
203  * Advancing the Invariants:
204  *
205  * 	p = first newline after halfway point from front to back.
206  *
207  * 	If the string at "p" is not greater than the string to match,
208  *	p is the new front.  Otherwise it is the new back.
209  *
210  * Termination:
211  *
212  * 	The definition of the routine allows it return at any point,
213  *	since front is always at or before the line to print.
214  *
215  * 	In fact, it returns when the chosen "p" equals "back".  This
216  *	implies that there exists a string is least half as long as
217  *	(back - front), which in turn implies that a linear search will
218  *	be no more expensive than the cost of simply printing a string or two.
219  *
220  * 	Trying to continue with binary search at this point would be
221  *	more trouble than it's worth.
222  */
223 #define	SKIP_PAST_NEWLINE(p, back) \
224 	while (p < back && *p++ != '\n');
225 
226 char *
227 binary_search(string, front, back)
228 	char *string, *front, *back;
229 {
230 	char *p;
231 
232 	p = front + (back - front) / 2;
233 	SKIP_PAST_NEWLINE(p, back);
234 
235 	/*
236 	 * If the file changes underneath us, make sure we don't
237 	 * infinitely loop.
238 	 */
239 	while (p < back && back > front) {
240 		if (compare(string, p, back) == GREATER)
241 			front = p;
242 		else
243 			back = p;
244 		p = front + (back - front) / 2;
245 		SKIP_PAST_NEWLINE(p, back);
246 	}
247 	return (front);
248 }
249 
250 /*
251  * Find the first line that starts with string, linearly searching from front
252  * to back.
253  *
254  * Return NULL for no such line.
255  *
256  * This routine assumes:
257  *
258  * 	o front points at the first character in a line.
259  *	o front is before or at the first line to be printed.
260  */
261 char *
262 linear_search(string, front, back)
263 	char *string, *front, *back;
264 {
265 	while (front < back) {
266 		switch (compare(string, front, back)) {
267 		case EQUAL:		/* Found it. */
268 			return (front);
269 			break;
270 		case LESS:		/* No such string. */
271 			return (NULL);
272 			break;
273 		case GREATER:		/* Keep going. */
274 			break;
275 		}
276 		SKIP_PAST_NEWLINE(front, back);
277 	}
278 	return (NULL);
279 }
280 
281 /*
282  * Print as many lines as match string, starting at front.
283  */
284 void
285 print_from(string, front, back)
286 	char *string, *front, *back;
287 {
288 	for (; front < back && compare(string, front, back) == EQUAL; ++front) {
289 		for (; front < back && *front != '\n'; ++front)
290 			if (putchar(*front) == EOF)
291 				err(2, "stdout");
292 		if (putchar('\n') == EOF)
293 			err(2, "stdout");
294 	}
295 }
296 
297 /*
298  * Return LESS, GREATER, or EQUAL depending on how the string1 compares with
299  * string2 (s1 ??? s2).
300  *
301  * 	o Matches up to len(s1) are EQUAL.
302  *	o Matches up to len(s2) are GREATER.
303  *
304  * Compare understands about the -f and -d flags, and treats comparisons
305  * appropriately.
306  *
307  * The string "s1" is null terminated.  The string s2 is '\n' terminated (or
308  * "back" terminated).
309  */
310 int
311 compare(s1, s2, back)
312 	char *s1, *s2, *back;
313 {
314 	int ch;
315 
316 	for (; *s1 && s2 < back && *s2 != '\n'; ++s1, ++s2) {
317 		ch = *s2;
318 		if (fflag)
319 			ch = FOLD(ch);
320 		if (dflag)
321 			ch = DICT(ch);
322 
323 		if (ch == NO_COMPARE) {
324 			++s2;		/* Ignore character in comparison. */
325 			continue;
326 		}
327 		if (*s1 != ch)
328 			return (*s1 < ch ? LESS : GREATER);
329 	}
330 	return (*s1 ? GREATER : EQUAL);
331 }
332 
333 void
334 usage()
335 {
336 	(void)fprintf(stderr, "usage: look [-df] [-t char] string [file]\n");
337 	exit(2);
338 }
339