utils-std

cmp.c (5841B)

// utils-std: Collection of commonly available Unix tools
// SPDX-FileCopyrightText: 2017 Haelwenn (lanodan) Monnier <contact+utils@hacktivis.me>
// SPDX-License-Identifier: MPL-2.0
#define _POSIX_C_SOURCE 200809L
#include "../config.h"
#include "../libutils/getopt_nolong.h"
#include <assert.h>
#include <errno.h>
#include <fcntl.h> // open, posix_fadvise
#include <stdbool.h>
#include <stdio.h>    // fopen, fprintf, getline
#include <stdlib.h>   // abort, strtoul
#include <string.h>   // strerror
#include <sys/stat.h> // fstat
#include <unistd.h>   // getopt
#ifdef HAS_GETOPT_LONG
#include <getopt.h>
#endif
static bool opt_s = false, opt_l = false;
static unsigned long max_bytes = 0;
const char *argv0 = "cmp";
#undef MIN
#define MIN(a, b) (((a) < (b)) ? (a) : (b))
static int
do_cmp(FILE *file1, const char *name1, FILE *file2, const char *name2)
{
	char *line1 = NULL, *line2 = NULL;
	size_t len1 = 0, len2 = 0;
	unsigned long pos = 1, ln = 1;
	errno = 0;
	while(true)
	{
		ssize_t nread1 = getline(&line1, &len1, file1);
		if(nread1 < 0)
		{
			if(!ferror(file1)) return 0;
			fprintf(stderr,
			        "%s: error: Failed to read line %ld from file '%s': %s\n",
			        argv0,
			        ln,
			        name1,
			        strerror(errno));
			return 2;
		}
		ssize_t nread2 = getline(&line2, &len2, file2);
		if(nread2 < 0)
		{
			if(!ferror(file2))
			{
				if(!opt_s) fprintf(stderr, "%s: error: EOF on %s line %ld\n", argv0, name2, ln);
				return 2;
			}
			fprintf(stderr,
			        "%s: error: Failed to read line %ld from file '%s': %s\n",
			        argv0,
			        ln,
			        name1,
			        strerror(errno));
			return 2;
		}
		for(ssize_t i = 0; i < MIN(nread1, nread2); i++)
		{
			if(max_bytes != 0 && pos + i >= max_bytes) return 0;
			if(line1[i] != line2[i])
			{
				if(opt_s) return 1;
				if(opt_l)
					printf("%ld %o %o\n", pos + i, line1[i], line2[i]);
				else
					printf("%s %s differ: char %zd, line %ld\n", name1, name2, i + 1, ln);
				return 1;
			}
		}
		assert(nread1 == nread2);
		pos += nread1;
		ln++;
	}
	return 0;
}
static bool
fstat_cmp(int fd1, char *name1, int fd2, char *name2)
{
	struct stat buf1, buf2;
	errno = 0;
	if(fstat(fd1, &buf1) != 0)
	{
		fprintf(stderr,
		        "%s: warning: Failed getting status for file '%s': %s\n",
		        argv0,
		        name1,
		        strerror(errno));
		return false;
	}
	errno = 0;
	if(fstat(fd2, &buf2) != 0)
	{
		fprintf(stderr,
		        "%s: warning: Failed getting status for file '%s': %s\n",
		        argv0,
		        name2,
		        strerror(errno));
		return false;
	}
	return buf1.st_ino == buf2.st_ino && buf1.st_dev == buf2.st_dev;
}
static void
usage(void)
{
	fprintf(stderr, "Usage: cmp [-l|-s] [-n max_bytes] file1 file2\n");
}
int
main(int argc, char *argv[])
{
	char *endptr = NULL;
#ifdef HAS_GETOPT_LONG
	// Strictly for GNUisms compatibility so no long-only options
	// clang-format off
	static struct option opts[] = {
		{"verbose", no_argument, NULL, 'l'},
		{"bytes", required_argument, NULL, 'n'},
		{"quiet", no_argument, NULL, 's'},
		{"silent", no_argument, NULL, 's'},
		{0, 0, 0, 0},
	};
	// clang-format on
	// Need + as first character to get POSIX-style option parsing
	for(int c = -1; (c = getopt_long(argc, argv, "+:ln:s", opts, NULL)) != -1;)
#else
	for(int c = -1; (c = getopt_nolong(argc, argv, ":ln:s")) != -1;)
#endif
	{
		switch(c)
		{
		case 'l':
			opt_l = true;
			break;
		case 's':
			opt_s = true;
			break;
		case 'n':
			errno = 0;
			max_bytes = strtoul(optarg, &endptr, 0);
			if(errno != 0)
			{
				fprintf(stderr, "%s: error: Failed parsing '-n %s': %s\n", argv0, optarg, strerror(errno));
				return 2;
			}
			if(endptr != NULL && endptr[0] != 0)
			{
				fprintf(stderr,
				        "%s: error: Non-numeric characters passed to '-n %s': %s\n",
				        argv0,
				        optarg,
				        endptr);
				return 2;
			}
			break;
		case ':':
			fprintf(stderr, "%s: error: Missing operand for option: '-%c'\n", argv0, optopt);
			usage();
			return 2;
		case '?':
			GETOPT_UNKNOWN_OPT
			usage();
			return 2;
		default:
			abort();
		}
	}
	argc -= optind;
	argv += optind;
	if(argc != 2)
	{
		fprintf(stderr, "%s: error: Expected 2 arguments, got %d arguments\n", argv0, argc);
		return 2;
	}
	if(strcmp(argv[0], argv[1]) == 0) return 0;
	int fd1 = STDIN_FILENO;
	if(!(argv[0][0] == '-' && argv[0][1] == 0))
	{
		errno = 0;
		fd1 = open(argv[0], O_RDONLY);
		if(errno != 0)
		{
			fprintf(stderr, "%s: error: Failed opening file '%s': %s\n", argv0, argv[0], strerror(errno));
			return 2;
		}
	}
	int fd2 = STDIN_FILENO;
	if(!(argv[1][0] == '-' && argv[1][1] == 0))
	{
		errno = 0;
		fd2 = open(argv[1], O_RDONLY);
		if(errno != 0)
		{
			fprintf(stderr, "%s: error: Failed opening file '%s': %s\n", argv0, argv[1], strerror(errno));
			return 2;
		}
	}
	if(fstat_cmp(fd1, argv[0], fd2, argv[1]))
	{
		if(fd1 != STDIN_FILENO) close(fd1);
		if(fd2 != STDIN_FILENO) close(fd2);
		return 0;
	}
	FILE *file1 = stdin;
	FILE *file2 = stdin;
	if(fd1 != STDIN_FILENO)
	{
		errno = 0;
		file1 = fdopen(fd1, "r");
		if(file1 == (FILE *)NULL)
		{
			fprintf(stderr,
			        "%s: error: Failed associating stream for file '%s': %s\n",
			        argv0,
			        argv[0],
			        strerror(errno));
			return 2;
		}
		posix_fadvise(fd1, 0, 0, POSIX_FADV_SEQUENTIAL);
	}
	if(fd2 != STDIN_FILENO)
	{
		errno = 0;
		file2 = fdopen(fd2, "r");
		if(file2 == (FILE *)NULL)
		{
			fprintf(stderr,
			        "%s: error: Failed associating stream for file '%s': %s\n",
			        argv0,
			        argv[1],
			        strerror(errno));
			return 2;
		}
		posix_fadvise(fd2, 0, 0, POSIX_FADV_SEQUENTIAL);
	}
	int ret = do_cmp(file1, argv[0], file2, argv[1]);
	if(file1 != stdin) fclose(file1);
	if(file2 != stdin) fclose(file2);
	return ret;
}