Home | History | Annotate | Line # | Download | only in grep
file.c revision 1.2.12.1
      1  1.2.12.1    matt /*	$NetBSD: file.c,v 1.2.12.1 2008/01/09 02:00:42 matt Exp $	*/
      2       1.2  rillig 
      3       1.1    cjep /*-
      4       1.1    cjep  * Copyright (c) 1999 James Howard and Dag-Erling Codan Smrgrav
      5       1.1    cjep  * All rights reserved.
      6       1.1    cjep  *
      7       1.1    cjep  * Redistribution and use in source and binary forms, with or without
      8       1.1    cjep  * modification, are permitted provided that the following conditions
      9       1.1    cjep  * are met:
     10       1.1    cjep  * 1. Redistributions of source code must retain the above copyright
     11       1.1    cjep  *    notice, this list of conditions and the following disclaimer.
     12       1.1    cjep  * 2. Redistributions in binary form must reproduce the above copyright
     13       1.1    cjep  *    notice, this list of conditions and the following disclaimer in the
     14       1.1    cjep  *    documentation and/or other materials provided with the distribution.
     15       1.1    cjep  *
     16       1.1    cjep  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     17       1.1    cjep  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     18       1.1    cjep  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     19       1.1    cjep  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     20       1.1    cjep  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     21       1.1    cjep  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     22       1.1    cjep  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     23       1.1    cjep  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     24       1.1    cjep  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     25       1.1    cjep  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     26       1.1    cjep  * SUCH DAMAGE.
     27       1.1    cjep  *
     28       1.1    cjep  */
     29       1.1    cjep 
     30       1.2  rillig #include <sys/cdefs.h>
     31       1.2  rillig #ifndef lint
     32  1.2.12.1    matt __RCSID("$NetBSD: file.c,v 1.2.12.1 2008/01/09 02:00:42 matt Exp $");
     33       1.2  rillig #endif /* not lint */
     34       1.2  rillig 
     35       1.1    cjep #include <sys/param.h>
     36       1.1    cjep 
     37  1.2.12.1    matt #include <zlib.h>
     38       1.1    cjep #include <err.h>
     39       1.1    cjep #include <stdio.h>
     40       1.1    cjep #include <stdlib.h>
     41       1.1    cjep 
     42       1.1    cjep #include "grep.h"
     43       1.1    cjep 
     44       1.2  rillig static char fname[MAXPATHLEN];
     45       1.2  rillig static char *lnbuf;
     46       1.2  rillig static size_t lnbuflen;
     47       1.1    cjep 
     48       1.1    cjep #define FILE_STDIO	0
     49       1.1    cjep #define FILE_MMAP	1
     50       1.1    cjep #define FILE_GZIP	2
     51       1.1    cjep 
     52       1.1    cjep struct file {
     53       1.2  rillig 	int type;
     54       1.2  rillig 	FILE *f;
     55       1.2  rillig 	mmf_t *mmf;
     56       1.2  rillig 	gzFile *gzf;
     57       1.1    cjep };
     58       1.1    cjep 
     59       1.1    cjep static char *
     60       1.2  rillig grepfgetln(FILE *f, size_t *len)
     61       1.2  rillig {
     62       1.2  rillig 	size_t n;
     63       1.2  rillig 	int c;
     64       1.2  rillig 
     65       1.2  rillig 	for (n = 0; ; ++n) {
     66       1.2  rillig 		c = getc(f);
     67       1.2  rillig 		if (c == EOF) {
     68       1.2  rillig 			if (feof(f))
     69       1.2  rillig 				break;
     70       1.2  rillig 			err(2, "%s", fname);
     71       1.2  rillig 			/* ERROR */
     72       1.2  rillig 
     73       1.2  rillig 		}
     74       1.2  rillig 		if (c == line_endchar)
     75       1.2  rillig 			break;
     76       1.2  rillig 		if (n >= lnbuflen) {
     77       1.2  rillig 			lnbuflen *= 2;
     78       1.2  rillig 			lnbuf = grep_realloc(lnbuf, ++lnbuflen);
     79       1.2  rillig 		}
     80       1.2  rillig 		lnbuf[n] = c;
     81       1.2  rillig 	}
     82       1.2  rillig 	if (feof(f) && n == 0)
     83       1.2  rillig 		return NULL;
     84       1.2  rillig 	*len = n;
     85       1.2  rillig 	return lnbuf;
     86       1.2  rillig }
     87       1.2  rillig 
     88       1.2  rillig static char *
     89       1.1    cjep gzfgetln(gzFile *f, size_t *len)
     90       1.1    cjep {
     91       1.2  rillig 	size_t n;
     92       1.2  rillig 	int c;
     93       1.1    cjep 
     94       1.1    cjep 	for (n = 0; ; ++n) {
     95       1.1    cjep 		c = gzgetc(f);
     96       1.1    cjep 		if (c == -1) {
     97       1.1    cjep 			const char *gzerrstr;
     98       1.1    cjep 			int gzerr;
     99       1.1    cjep 
    100       1.1    cjep 			if (gzeof(f))
    101       1.1    cjep 				break;
    102       1.2  rillig 
    103       1.1    cjep 			gzerrstr = gzerror(f, &gzerr);
    104       1.1    cjep 			if (gzerr == Z_ERRNO)
    105       1.2  rillig 				err(2, "%s", fname);
    106       1.1    cjep 			else
    107       1.2  rillig 				errx(2, "%s: %s", fname, gzerrstr);
    108       1.1    cjep 		}
    109       1.2  rillig 		if (c == line_endchar)
    110       1.1    cjep 			break;
    111       1.1    cjep 		if (n >= lnbuflen) {
    112       1.1    cjep 			lnbuflen *= 2;
    113       1.1    cjep 			lnbuf = grep_realloc(lnbuf, ++lnbuflen);
    114       1.1    cjep 		}
    115       1.1    cjep 		lnbuf[n] = c;
    116       1.1    cjep 	}
    117       1.1    cjep 
    118       1.1    cjep 	if (gzeof(f) && n == 0)
    119       1.1    cjep 		return NULL;
    120       1.1    cjep 	*len = n;
    121       1.1    cjep 	return lnbuf;
    122       1.1    cjep }
    123       1.1    cjep 
    124       1.1    cjep file_t *
    125       1.2  rillig grep_fdopen(int fd, const char *mode)
    126       1.1    cjep {
    127       1.1    cjep 	file_t *f;
    128       1.1    cjep 
    129       1.1    cjep 	if (fd == 0)
    130       1.1    cjep 		sprintf(fname, "(standard input)");
    131       1.1    cjep 	else
    132       1.1    cjep 		sprintf(fname, "(fd %d)", fd);
    133       1.2  rillig 
    134       1.1    cjep 	f = grep_malloc(sizeof *f);
    135       1.2  rillig 
    136       1.2  rillig 	if (zgrep) {
    137       1.1    cjep 		f->type = FILE_GZIP;
    138       1.1    cjep 		if ((f->gzf = gzdopen(fd, mode)) != NULL)
    139       1.1    cjep 			return f;
    140       1.1    cjep 	} else {
    141       1.1    cjep 		f->type = FILE_STDIO;
    142       1.1    cjep 		if ((f->f = fdopen(fd, mode)) != NULL)
    143       1.1    cjep 			return f;
    144       1.1    cjep 	}
    145       1.2  rillig 
    146       1.1    cjep 	free(f);
    147       1.1    cjep 	return NULL;
    148       1.1    cjep }
    149       1.1    cjep 
    150       1.1    cjep file_t *
    151       1.2  rillig grep_open(const char *path, const char *mode)
    152       1.1    cjep {
    153       1.1    cjep 	file_t *f;
    154       1.1    cjep 
    155       1.1    cjep 	snprintf(fname, MAXPATHLEN, "%s", path);
    156       1.2  rillig 
    157       1.1    cjep 	f = grep_malloc(sizeof *f);
    158       1.2  rillig 
    159       1.2  rillig 	if (zgrep) {
    160       1.1    cjep 		f->type = FILE_GZIP;
    161       1.1    cjep 		if ((f->gzf = gzopen(fname, mode)) != NULL)
    162       1.1    cjep 			return f;
    163       1.1    cjep 	} else {
    164       1.1    cjep 		/* try mmap first; if it fails, try stdio */
    165       1.1    cjep 		if ((f->mmf = mmopen(fname, mode)) != NULL) {
    166       1.1    cjep 			f->type = FILE_MMAP;
    167       1.1    cjep 			return f;
    168       1.1    cjep 		}
    169       1.1    cjep 		f->type = FILE_STDIO;
    170       1.1    cjep 		if ((f->f = fopen(path, mode)) != NULL)
    171       1.1    cjep 			return f;
    172       1.1    cjep 	}
    173       1.2  rillig 
    174       1.1    cjep 	free(f);
    175       1.1    cjep 	return NULL;
    176       1.1    cjep }
    177       1.1    cjep 
    178       1.1    cjep int
    179       1.1    cjep grep_bin_file(file_t *f)
    180       1.1    cjep {
    181       1.1    cjep 	switch (f->type) {
    182       1.1    cjep 	case FILE_STDIO:
    183       1.1    cjep 		return bin_file(f->f);
    184       1.1    cjep 	case FILE_MMAP:
    185       1.2  rillig 		return mmbin_file(f->mmf);
    186       1.1    cjep 	case FILE_GZIP:
    187       1.1    cjep 		return gzbin_file(f->gzf);
    188       1.1    cjep 	default:
    189       1.1    cjep 		/* can't happen */
    190       1.2  rillig 		errx(2, "invalid file type");
    191       1.1    cjep 	}
    192       1.1    cjep }
    193       1.1    cjep 
    194       1.1    cjep char *
    195       1.1    cjep grep_fgetln(file_t *f, size_t *l)
    196       1.1    cjep {
    197       1.1    cjep 	switch (f->type) {
    198       1.1    cjep 	case FILE_STDIO:
    199       1.2  rillig 		if (line_endchar == '\n')
    200       1.2  rillig 			return fgetln(f->f, l);
    201       1.2  rillig 		else
    202       1.2  rillig 			return grepfgetln(f->f, l);
    203       1.1    cjep 	case FILE_MMAP:
    204       1.1    cjep 		return mmfgetln(f->mmf, l);
    205       1.1    cjep 	case FILE_GZIP:
    206       1.1    cjep 		return gzfgetln(f->gzf, l);
    207       1.1    cjep 	default:
    208       1.1    cjep 		/* can't happen */
    209       1.2  rillig 		errx(2, "invalid file type");
    210       1.1    cjep 	}
    211       1.1    cjep }
    212       1.1    cjep 
    213       1.1    cjep void
    214       1.1    cjep grep_close(file_t *f)
    215       1.1    cjep {
    216       1.1    cjep 	switch (f->type) {
    217       1.1    cjep 	case FILE_STDIO:
    218       1.1    cjep 		fclose(f->f);
    219       1.1    cjep 		break;
    220       1.1    cjep 	case FILE_MMAP:
    221       1.1    cjep 		mmclose(f->mmf);
    222       1.1    cjep 		break;
    223       1.1    cjep 	case FILE_GZIP:
    224       1.1    cjep 		gzclose(f->gzf);
    225       1.1    cjep 		break;
    226       1.1    cjep 	default:
    227       1.1    cjep 		/* can't happen */
    228       1.2  rillig 		errx(2, "invalid file type");
    229       1.1    cjep 	}
    230       1.1    cjep }
    231