Home | History | Annotate | Line # | Download | only in grep
grep.c revision 1.1.1.1
      1 /*-
      2  * Copyright (c) 1999 James Howard and Dag-Erling Codan Smrgrav
      3  * All rights reserved.
      4  *
      5  * Redistribution and use in source and binary forms, with or without
      6  * modification, are permitted provided that the following conditions
      7  * are met:
      8  * 1. Redistributions of source code must retain the above copyright
      9  *    notice, this list of conditions and the following disclaimer.
     10  * 2. Redistributions in binary form must reproduce the above copyright
     11  *    notice, this list of conditions and the following disclaimer in the
     12  *    documentation and/or other materials provided with the distribution.
     13  *
     14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     24  * SUCH DAMAGE.
     25  *
     26  *	$Id: grep.c,v 1.1.1.1 2004/01/02 14:58:44 cjep Exp $
     27  */
     28 
     29 #include <sys/types.h>
     30 #include <sys/stat.h>
     31 
     32 #include <err.h>
     33 #include <errno.h>
     34 #include <getopt.h>
     35 #include <regex.h>
     36 #include <stdio.h>
     37 #include <stdlib.h>
     38 #include <string.h>
     39 #include <unistd.h>
     40 
     41 #include "grep.h"
     42 
     43 /* Flags passed to regcomp() and regexec() */
     44 int	 cflags;
     45 int	 eflags = REG_STARTEND;
     46 
     47 int	 matchall;	/* shortcut */
     48 int	 patterns, pattern_sz;
     49 char   **pattern;
     50 regex_t	*r_pattern;
     51 
     52 /* For regex errors  */
     53 char	 re_error[RE_ERROR_BUF + 1];
     54 
     55 /* Command-line flags */
     56 int	 Aflag;		/* -A x: print x lines trailing each match */
     57 int	 Bflag;		/* -B x: print x lines leading each match */
     58 int	 Eflag;		/* -E: interpret pattern as extended regexp */
     59 int	 Fflag;		/* -F: interpret pattern as list of fixed strings */
     60 int	 Gflag;		/* -G: interpret pattern as basic regexp */
     61 int	 Hflag;		/* -H: if -R, follow explicitly listed symlinks */
     62 int	 Lflag;		/* -L: only show names of files with no matches */
     63 int	 Pflag;		/* -P: if -R, no symlinks are followed */
     64 int	 Rflag;		/* -R: recursively search directory trees */
     65 int	 Sflag;		/* -S: if -R, follow all symlinks */
     66 int	 Vflag;		/* -V: display version information */
     67 int	 Zflag;		/* -Z: decompress input before processing */
     68 int	 aflag;		/* -a: only search ascii files */
     69 int	 bflag;		/* -b: show block numbers for each match */
     70 int	 cflag;		/* -c: only show a count of matching lines */
     71 int	 hflag;		/* -h: don't print filename headers */
     72 int	 iflag;		/* -i: ignore case */
     73 int	 lflag;		/* -l: only show names of files with matches */
     74 int	 nflag;		/* -n: show line numbers in front of matching lines */
     75 int	 oflag;		/* -o: always print file name */
     76 int	 qflag;		/* -q: quiet mode (don't output anything) */
     77 int	 sflag;		/* -s: silent mode (ignore errors) */
     78 int	 vflag;		/* -v: only show non-matching lines */
     79 int	 wflag;		/* -w: pattern must start and end on word boundaries */
     80 int	 xflag;		/* -x: pattern must match entire line */
     81 
     82 /* Housekeeping */
     83 int	 first;		/* flag whether or not this is our fist match */
     84 int	 tail;		/* lines left to print */
     85 int	 lead;		/* number of lines in leading context queue */
     86 
     87 char	*progname;
     88 
     89 static void
     90 usage(void)
     91 {
     92 	fprintf(stderr, "usage: %s %s %s\n",
     93 		progname,
     94 		"[-[AB] num] [-CEFGHLPRSVZabchilnoqsvwx]",
     95 		"[-e patttern] [-f file]");
     96 	exit(2);
     97 }
     98 
     99 static char *optstr = "0123456789A:B:CEFGHLPSRUVZabce:f:hilnoqrsuvwxy";
    100 
    101 struct option long_options[] =
    102 {
    103 	{"basic-regexp",        no_argument,       NULL, 'G'},
    104 	{"extended-regexp",     no_argument,       NULL, 'E'},
    105 	{"fixed-strings",       no_argument,       NULL, 'F'},
    106 	{"after-context",       required_argument, NULL, 'A'},
    107 	{"before-context",      required_argument, NULL, 'B'},
    108 	{"context",             optional_argument, NULL, 'C'},
    109 	{"version",             no_argument,       NULL, 'V'},
    110 	{"byte-offset",         no_argument,       NULL, 'b'},
    111 	{"count",               no_argument,       NULL, 'c'},
    112 	{"regexp",              required_argument, NULL, 'e'},
    113 	{"file",                required_argument, NULL, 'f'},
    114 	{"no-filename",         no_argument,       NULL, 'h'},
    115 	{"ignore-case",         no_argument,       NULL, 'i'},
    116 	{"files-without-match", no_argument,       NULL, 'L'},
    117 	{"files-with-matches",  no_argument,       NULL, 'l'},
    118 	{"line-number",         no_argument,       NULL, 'n'},
    119 	{"quiet",               no_argument,       NULL, 'q'},
    120 	{"silent",              no_argument,       NULL, 'q'},
    121 	{"recursive",           no_argument,       NULL, 'r'},
    122 	{"no-messages",         no_argument,       NULL, 's'},
    123 	{"text",                no_argument,       NULL, 'a'},
    124 	{"revert-match",        no_argument,       NULL, 'v'},
    125 	{"word-regexp",         no_argument,       NULL, 'w'},
    126 	{"line-regexp",         no_argument,       NULL, 'x'},
    127 	{"binary",              no_argument,       NULL, 'U'},
    128 	{"unix-byte-offsets",   no_argument,       NULL, 'u'},
    129 	{"decompress",          no_argument,       NULL, 'Z'},
    130 
    131 	{NULL,                  no_argument,       NULL, 0}
    132 };
    133 
    134 
    135 static void
    136 add_pattern(char *pat, size_t len)
    137 {
    138 	if (len == 0 || matchall) {
    139 		matchall = 1;
    140 		return;
    141 	}
    142 	if (patterns == pattern_sz) {
    143 		pattern_sz *= 2;
    144 		pattern = grep_realloc(pattern, ++pattern_sz);
    145 	}
    146 	if (pat[len-1] == '\n')
    147 		--len;
    148 	pattern[patterns] = grep_malloc(len+1);
    149 	strncpy(pattern[patterns], pat, len);
    150 	pattern[patterns][len] = '\0';
    151 	++patterns;
    152 }
    153 
    154 static void
    155 read_patterns(char *fn)
    156 {
    157 	FILE *f;
    158 	char *line;
    159 	size_t len;
    160 	int nl;
    161 
    162 	if ((f = fopen(fn, "r")) == NULL)
    163 		err(1, "%s", fn);
    164 	nl = 0;
    165 	while ((line = fgetln(f, &len)) != NULL) {
    166 		if (*line == '\n') {
    167 			++nl;
    168 			continue;
    169 		}
    170 		if (nl) {
    171 			matchall = 1;
    172 			break;
    173 		}
    174 		nl = 0;
    175 		add_pattern(line, len);
    176 	}
    177 	if (ferror(f))
    178 		err(1, "%s", fn);
    179 	fclose(f);
    180 }
    181 
    182 int
    183 main(int argc, char *argv[])
    184 {
    185 	char *tmp;
    186 	int c, i;
    187 
    188 	if ((progname = strrchr(*argv, '/')) != NULL)
    189 		++progname;
    190 	else
    191 		progname = *argv;
    192 
    193 	while ((c = getopt_long(argc, argv, optstr,
    194 				long_options, (int *)NULL)) != -1) {
    195 		switch (c) {
    196 		case '0': case '1': case '2': case '3': case '4':
    197 		case '5': case '6': case '7': case '8': case '9':
    198 			tmp = argv[optind - 1];
    199 			if (tmp[0] == '-' && tmp[1] == c && !tmp[2])
    200 				Aflag = Bflag = strtol(++tmp, (char **)NULL, 10);
    201 			else
    202 				Aflag = Bflag = strtol(argv[optind] + 1, (char **)NULL, 10);
    203 			break;
    204 		case 'A':
    205 			Aflag = strtol(optarg, (char **)NULL, 10);
    206 			break;
    207 		case 'B':
    208 			Bflag = strtol(optarg, (char **)NULL, 10);
    209 			break;
    210 		case 'C':
    211 			if (optarg == NULL)
    212 				Aflag = Bflag = 2;
    213 			else
    214 				Aflag = Bflag = strtol(optarg, (char **)NULL, 10);
    215 			break;
    216 		case 'E':
    217 			Eflag++;
    218 			break;
    219 		case 'F':
    220 			Fflag++;
    221 			break;
    222 		case 'G':
    223 			Gflag++;
    224 			break;
    225 		case 'H':
    226 			Hflag++;
    227 			break;
    228 		case 'L':
    229 			lflag = 0;
    230 			Lflag = qflag = 1;
    231 			break;
    232 		case 'P':
    233 			Pflag++;
    234 			break;
    235 		case 'S':
    236 			Sflag++;
    237 			break;
    238 		case 'R':
    239 		case 'r':
    240 			Rflag++;
    241 			oflag++;
    242 			break;
    243 		case 'U':
    244 		case 'u':
    245 			/* these are here for compatability */
    246 			break;
    247 		case 'V':
    248 			fprintf(stderr, "grep version %u.%u\n", VER_MAJ, VER_MIN);
    249 			fprintf(stderr, argv[0]);
    250 			usage();
    251 			break;
    252 		case 'Z':
    253 			Zflag++;
    254 			break;
    255 		case 'a':
    256 			aflag = 1;
    257 			break;
    258 		case 'b':
    259 			bflag = 1;
    260 			break;
    261 		case 'c':
    262 			cflag = 1;
    263 			break;
    264 		case 'e':
    265 			add_pattern(optarg, strlen(optarg));
    266 			break;
    267 		case 'f':
    268 			read_patterns(optarg);
    269 			break;
    270 		case 'h':
    271 			oflag = 0;
    272 			hflag = 1;
    273 			break;
    274 		case 'i':
    275 		case 'y':
    276 			cflags |= REG_ICASE;
    277 			break;
    278 		case 'l':
    279 			Lflag = 0;
    280 			lflag = qflag = 1;
    281 			break;
    282 		case 'n':
    283 			nflag = 1;
    284 			break;
    285 		case 'o':
    286 			hflag = 0;
    287 			oflag = 1;
    288 			break;
    289 		case 'q':
    290 			qflag = 1;
    291 			break;
    292 		case 's':
    293 			sflag = 1;
    294 			break;
    295 		case 'v':
    296 			vflag = 1;
    297 			break;
    298 		case 'w':
    299 			wflag = 1;
    300 			break;
    301 		case 'x':
    302 			xflag = 1;
    303 			break;
    304 		default:
    305 			usage();
    306 		}
    307 	}
    308 
    309 	argc -= optind;
    310 	argv += optind;
    311 
    312 	if (argc == 0 && patterns == 0)
    313 		usage();
    314 
    315 	if (patterns == 0) {
    316 		add_pattern(*argv, strlen(*argv));
    317 		--argc;
    318 		++argv;
    319 	}
    320 
    321 	switch (*progname) {
    322 	case 'e':
    323 		Eflag++;
    324 		break;
    325 	case 'f':
    326 		Fflag++;
    327 		break;
    328 	case 'g':
    329 		Gflag++;
    330 		break;
    331 	case 'z':
    332 		Zflag++;
    333 		break;
    334 	}
    335 
    336 	cflags |= Eflag ? REG_EXTENDED : REG_BASIC;
    337 	r_pattern = grep_malloc(patterns * sizeof(regex_t));
    338 	for (i = 0; i < patterns; ++i) {
    339 		if ((c = regcomp(&r_pattern[i], pattern[i], cflags))) {
    340 			regerror(c, &r_pattern[i], re_error, RE_ERROR_BUF);
    341 			errx(1, "%s", re_error);
    342 		}
    343 	}
    344 
    345 	if ((argc == 0 || argc == 1) && !oflag)
    346 		hflag = 1;
    347 
    348 	if (argc == 0)
    349 		exit(!procfile(NULL));
    350 
    351 	if (Rflag)
    352 		c = grep_tree(argv);
    353 	else
    354 		for (c = 0; argc--; ++argv)
    355 			c += procfile(*argv);
    356 
    357 	exit(!c);
    358 }
    359