utils-std

wc.c (5412B)

// 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 "../lib/bitmasks.h"
#include <assert.h>
#include <ctype.h> // isspace
#include <errno.h>
#include <locale.h> // setlocale
#include <stdbool.h>
#include <stdint.h> // uint8_t
#include <stdio.h>  // fprintf, fopen
#include <stdlib.h> // abort
#include <string.h> // strchr, strerror
#include <sys/stat.h>
#include <unistd.h> // getopt
#include <wchar.h>
#include <wctype.h> // iswspace
#ifdef HAS_GETOPT_LONG
#include <getopt.h>
#endif
static const char *argv0 = "wc";
static enum {
	WC_OPT_C = 1 << 0,
	WC_OPT_L = 1 << 1,
	WC_OPT_W = 1 << 2,
	WC_OPT_ALL = WC_OPT_C | WC_OPT_L | WC_OPT_W,
} wc_opts = 0;
off_t total_bytes = 0, total_lines = 0, total_words = 0;
static void
print_counts(off_t lines, off_t words, off_t bytes, const char *filename)
{
	const char *fmt = "%ld";
	if(FIELD_MATCH(wc_opts, WC_OPT_L))
	{
		printf(fmt, lines);
		fmt = " %ld";
	}
	if(FIELD_MATCH(wc_opts, WC_OPT_W))
	{
		printf(fmt, words);
		fmt = " %ld";
	}
	if(FIELD_MATCH(wc_opts, WC_OPT_C))
	{
		printf(fmt, bytes);
	}
	if(filename != NULL) printf(" %s", filename);
	printf("\n");
}
static int
wc_file_bytes(FILE *file, char *filename)
{
	off_t bytes = 0, lines = 0, words = 0, wordlen = 0;
	while(true)
	{
		assert(errno == 0);
		int c = getc(file);
		if(c == EOF)
		{
			if(errno != 0)
			{
				fprintf(stderr,
				        "%s: Failed reading from file '%s': %s\n",
				        argv0,
				        filename != NULL ? filename : "<stdin>",
				        strerror(errno));
				return -1;
			}
			break;
		}
		bytes++;
		if(c == '\n') lines++;
		if(isspace(c))
		{
			if(wordlen > 0)
			{
				words++;
				wordlen = 0;
			}
		}
		else
		{
			wordlen++;
		}
	}
	if(wordlen > 0) words++;
	total_bytes += bytes, total_lines += lines, total_words += words;
	print_counts(lines, words, bytes, filename);
	return 0;
}
static int
wc_file_chars(FILE *file, char *filename)
{
	off_t chars = 0, lines = 0, words = 0, wordlen = 0;
	while(true)
	{
		assert(errno == 0);
		wint_t c = getwc(file);
		if(c == WEOF)
		{
			if(errno != 0)
			{
				fprintf(stderr,
				        "%s: Failed reading from file '%s': %s\n",
				        argv0,
				        filename != NULL ? filename : "<stdin>",
				        strerror(errno));
				return -1;
			}
			break;
		}
		chars++;
		if(c == '\n') lines++;
		if(iswspace(c))
		{
			if(wordlen > 0)
			{
				words++;
				wordlen = 0;
			}
		}
		else
		{
			wordlen++;
		}
	}
	if(wordlen > 0) words++;
	print_counts(lines, words, chars, filename);
	return 0;
}
static void
usage(void)
{
	fprintf(stderr, "Usage: wc [-c|-m] [-lw] [file...]\n");
}
int
main(int argc, char *argv[])
{
	errno = 0;
	setlocale(LC_ALL, "");
	if(errno != 0)
	{
		fprintf(stderr, "%s: Warning: Failed to initialize locales: %s\n", argv0, strerror(errno));
		errno = 0;
	}
	int (*wc_file)(FILE *, char *) = &wc_file_bytes;
	int c = -1;
#ifdef HAS_GETOPT_LONG
	// Strictly for GNUisms compatibility so no long-only options
	// clang-format off
	static struct option opts[] = {
		{"bytes", no_argument, 0, 'c'},
		{"lines", no_argument, 0, 'l'},
		{"chars", no_argument, 0, 'm'},
		{"words", no_argument, 0, 'w'},
		{0, 0, 0, 0},
	};
	// clang-format on
	// Need + as first character to get POSIX-style option parsing
	while((c = getopt_long(argc, argv, "+:clmw", opts, NULL)) != -1)
#else
	while((c = getopt(argc, argv, ":clmw")) != -1)
#endif
	{
		switch(c)
		{
		case 'c':
			wc_opts |= WC_OPT_C;
			wc_file = wc_file_bytes;
			break;
		case 'l':
			wc_opts |= WC_OPT_L;
			break;
		case 'm':
			wc_opts |= WC_OPT_C;
			wc_file = wc_file_chars;
			break;
		case 'w':
			wc_opts |= WC_OPT_W;
			break;
		case ':':
			fprintf(stderr, "%s: Error: Missing operand for option: '-%c'\n", argv0, optopt);
			usage();
			return 1;
		case '?':
			fprintf(stderr, "%s: Error: Unrecognised option: '-%c'\n", argv0, optopt);
			usage();
			return 1;
		default:
			abort();
		}
	}
	if(wc_opts == 0) wc_opts = WC_OPT_ALL;
	assert(errno == 0);
	argc -= optind;
	argv += optind;
	if(argc < 1)
	{
		if(wc_file(stdin, NULL) < 0) return 1;
	}
	for(int i = 0; i < argc; i++)
	{
		char *path = argv[i];
		if(path[0] == '-' && path[1] == 0)
		{
			if(wc_file(stdin, NULL) < 0) return 1;
			continue;
		}
		// https://www.austingroupbugs.net/view.php?id=251
		if(strchr(path, '\n') != NULL)
			fprintf(
			    stderr,
			    "%s: Warning: Filename '%s' contains a newline while wc(1) uses newlines as separators\n",
			    argv0,
			    path);
		if(wc_opts == WC_OPT_C && wc_file == wc_file_bytes)
		{
			struct stat status;
			if(stat(path, &status) < 0)
			{
				fprintf(
				    stderr, "%s: Failed getting status for file '%s': %s\n", argv0, path, strerror(errno));
				return 1;
			}
			printf("%ld %s\n", status.st_size, path);
			total_bytes += status.st_size;
			continue;
		}
		FILE *file = fopen(path, "r");
		if(file == NULL)
		{
			fprintf(stderr, "%s: Failed opening file '%s': %s\n", argv0, path, strerror(errno));
			return 1;
		}
		if(wc_file(file, path) < 0) return 1;
		if(fclose(file) < 0)
		{
			fprintf(stderr, "%s: Failed closing file '%s': %s\n", argv0, path, strerror(errno));
			return 1;
		}
	}
	if(argc > 1) print_counts(total_lines, total_words, total_bytes, "total");
	return 0;
}