utils

strings.c (3973B)

// Collection of Unix tools, comparable to coreutils
// SPDX-FileCopyrightText: 2017-2023 Haelwenn (lanodan) Monnier <contact+utils@hacktivis.me>
// SPDX-License-Identifier: MPL-2.0
#define _POSIX_C_SOURCE 200809L
#include <ctype.h>  /* isprint() */
#include <errno.h>  /* errno */
#include <fcntl.h>  /* open(), O_RDONLY */
#include <limits.h> /* LONG_MIN, LONG_MAX */
#include <stdio.h>  /* fprintf(), BUFSIZ */
#include <stdlib.h> /* strtol() */
#include <string.h> /* strerror(), strncmp(), memset() */
#include <unistd.h> /* read(), write(), close(), getopt(), optarg, optind */
size_t opt_min_strlen   = 4;
char *opt_offset_format = NULL;
int
print_string(char *buffer, size_t offset)
{
	int ret = 0;
	if(opt_offset_format == NULL)
	{
		ret = printf("%s\n", buffer);
	}
	else
	{
		/* flawfinder: ignore. opt_offset_format isn't user-provided */
		ret = printf(opt_offset_format, offset, buffer);
	}
	if(ret < 0)
	{
		fprintf(stderr, "strings: Error writing: %s\n", strerror(errno));
		return 1;
	}
	else
	{
		return 0;
	}
}
int
concat(int fd, const char *fdname)
{
	ssize_t c;
	char read_buf[4096];
	char write_buf[4096];
	size_t write_pos = 0;
	size_t offset    = 0;
	memset(write_buf, 0, sizeof(write_buf));
	while((c = read(fd, read_buf, sizeof(read_buf))) > 0)
	{
		int read_pos = 0;
		char b       = 0;
		for(; read_pos < c; read_pos++)
		{
			b = read_buf[read_pos];
			if(isprint(b) && write_pos < 4096)
			{
				write_buf[write_pos++] = b;
			}
			else
			{
				if(write_pos >= opt_min_strlen)
				{
					write_buf[write_pos + 1] = 0;
					if(print_string(write_buf, offset) != 0)
					{
						return 1;
					}
				}
				offset += write_pos;
				offset++;
				write_pos = 0;
				memset(write_buf, 0, sizeof(write_buf));
			}
		}
	}
	if(c < 0)
	{
		fprintf(stderr, "strings: Error reading ‘%s’: %s\n", fdname, strerror(errno));
		return 1;
	}
	return 0;
}
void
usage()
{
	fprintf(stderr, "strings: [-a] [-t format] [-n number] [file...]\n");
}
int
main(int argc, char *argv[])
{
	int c;
	const char *errstr = NULL;
	/* flawfinder: ignore. Old implementations of getopt should fix themselves */
	while((c = getopt(argc, argv, ":an:t:")) != -1)
	{
		switch(c)
		{
		case 'a':
			/* Structure is always ignored */
			break;
		case 'n':
			errno          = 0;
			char *endptr   = "";
			opt_min_strlen = strtol(optarg, &endptr, 10);
			if(*endptr != 0)
			{
				// extraneous characters is invalid
				errno = EINVAL;
			}
			if(errno != 0)
			{
				fprintf(stderr, "strings: Option `-n %s`: %s\n", optarg, strerror(errno));
				usage();
				return 1;
			}
			if(opt_min_strlen == LLONG_MIN || opt_min_strlen < 1)
			{
				fprintf(stderr, "strings: Option `-n %s` is too small\n", optarg);
				usage();
				return 1;
			}
			if(opt_min_strlen == LLONG_MAX || opt_min_strlen > 4096)
			{
				fprintf(stderr, "strings: Option `-n %s` is too large\n", optarg);
				usage();
				return 1;
			}
			break;
		case 't':
			if(strnlen(optarg, 2) > 1)
			{
				usage();
				return 1;
			}
			switch(optarg[0])
			{
			case 'o':
				opt_offset_format = "%zo %s\n";
				break;
			case 'x':
				opt_offset_format = "%zx %s\n";
				break;
			case 'd':
				opt_offset_format = "%zd %s\n";
				break;
			default:
				fprintf(stderr, "strings: Unknown format: %s\n", optarg);
				usage();
				return 1;
			}
			break;
		}
	}
	argc -= optind;
	argv += optind;
	if(argc < 1)
	{
		return concat(0, "<stdin>");
	}
	for(int argi = 0; argi < argc; argi++)
	{
		if(strncmp(argv[argi], "-", 2) == 0)
		{
			if(concat(0, "<stdin>") != 0)
			{
				return 1;
			}
		}
		else
		{
			int fd = open(argv[argi], O_RDONLY);
			if(fd <= 0)
			{
				fprintf(stderr, "strings: Error opening ‘%s’: %s\n", argv[argi], strerror(errno));
				return 1;
			}
			if(concat(fd, argv[argi]) != 0)
			{
				return 1;
			}
			if(close(fd) < 0)
			{
				fprintf(stderr, "strings: Error closing ‘%s’: %s\n", argv[argi], strerror(errno));
				return 1;
			}
		}
	}
	return 0;
}