utils-std

strings.c (4095B)

// 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 <ctype.h>  /* isprint() */
#include <errno.h>  /* errno */
#include <fcntl.h>  /* open(), O_RDONLY */
#include <limits.h> /* LONG_MAX */
#include <stdio.h>  /* fprintf(), BUFSIZ */
#include <stdlib.h> /* strtol() */
#include <string.h> /* strerror(), strncmp() */
#include <unistd.h> /* read(), write(), close(), getopt(), optarg, optind */
size_t opt_min_strlen = 4;
const char *opt_offset_format = NULL;
int delim = '\n';
static int
print_string(char *buffer, size_t offset)
{
	int ret = 0;
	if(opt_offset_format == NULL)
	{
		ret = printf("%s%c", buffer, delim);
	}
	else
	{
		ret = printf(opt_offset_format, offset, buffer, delim);
	}
	if(ret < 0)
	{
		fprintf(stderr, "strings: error: Failed writing: %s\n", strerror(errno));
		errno = 0;
		return 1;
	}
	else
	{
		return 0;
	}
}
static int
concat(int fd, const char *fdname)
{
	ssize_t c;
	static char read_buf[4096];
	static char write_buf[4096];
	size_t write_pos = 0;
	size_t offset = 0;
	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;
			}
		}
	}
	if(c < 0)
	{
		fprintf(stderr, "strings: error: Failed reading ‘%s’: %s\n", fdname, strerror(errno));
		errno = 0;
		return 1;
	}
	return 0;
}
static void
usage(void)
{
	fprintf(stderr, "strings: [-az] [-t format] [-n number] [file...]\n");
}
int
main(int argc, char *argv[])
{
	int c;
	while((c = getopt(argc, argv, ":an:t:z")) != -1)
	{
		char *endptr = NULL;
		switch(c)
		{
		case 'a':
			/* Structure is always ignored */
			break;
		case 'n':
			opt_min_strlen = strtol(optarg, &endptr, 10);
			if(endptr && *endptr != 0)
			{
				// extraneous characters is invalid
				errno = EINVAL;
			}
			if(errno != 0)
			{
				fprintf(stderr, "strings: error: Option `-n %s`: %s\n", optarg, strerror(errno));
				usage();
				return 1;
			}
			if(opt_min_strlen < 1)
			{
				fprintf(stderr, "strings: error: Option `-n %s` is too small\n", optarg);
				usage();
				return 1;
			}
			if(opt_min_strlen == LONG_MAX || opt_min_strlen > 4096)
			{
				fprintf(stderr, "strings: error: 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%c";
				break;
			case 'x':
				opt_offset_format = "%zx %s%c";
				break;
			case 'd':
				opt_offset_format = "%zd %s%c";
				break;
			default:
				fprintf(stderr, "strings: error: Unknown format: %s\n", optarg);
				usage();
				return 1;
			}
			break;
		case 'z':
			delim = '\0';
			break;
		case ':':
			fprintf(stderr, "strings: error: Missing operand for option: '-%c'\n", optopt);
			usage();
			return 1;
		case '?':
			fprintf(stderr, "strings: error: Unrecognised option: '-%c'\n", optopt);
			usage();
			return 1;
		default:
			abort();
		}
	}
	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: Failed 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: Failed closing ‘%s’: %s\n", argv[argi], strerror(errno));
				return 1;
			}
		}
	}
	return 0;
}