From: Walter Alejandro Iglesias <wai@roquesor.com>
Subject: base64 encoder in libc
To: tech@openbsd.org
Date: Mon, 12 Aug 2024 10:38:04 +0200

I found out that there is a base64 encoder in libc (used by OpenSMTP, by
the way), which seems to be undocumented (at least I couldn't find a man
page).

  https://www.lemoda.net/unix/base64/index.html

It allowed me to simplify my encoding function, hence to reduce a bit
the size of my diffs.  Updated version of my diffs:

  https://en.roquesor.com/Downloads/mail_patches.tar.gz

The new version of the test program:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <err.h>
#include <errno.h>
#include <stdint.h>
#include <unistd.h>
#include <ctype.h>
#include <locale.h>

#define b64enc __b64_ntop

static int newline_at_end = 0;

void		encode_word(char **p);
static void	filecopy(FILE *, FILE *);
int		b64enc(unsigned char const *src, size_t srclength,
    char *target, size_t targsize);
void		wrap_header(char **p, size_t name_len);
void		usage(void);

int
main(int argc, char *argv[])
{
	FILE *fp;
	int option;

	while ((option = getopt(argc, argv, "h")) != -1)
		switch (option) {
		case 'h':
			usage();
			break;
		default:
			usage();
		}

	argc -= optind;
	argv += optind;

	if (argc > 0)
		while (argc-- > 0)
			if ((fp = fopen(*argv++, "r")) == NULL)
				warn("%s", *(argv - 1));
			else {
				filecopy(fp, stdout);
				fclose(fp);
			}
	else
		filecopy(stdin, stdout);

	return errno;
}

void
filecopy(FILE * ifp, FILE * ofp)
{
	char *p = NULL;
	size_t size = 0;
	size_t i = 0;
	int c;

	while ((c = getc(ifp)) != EOF) {
		if (i == size) {
			p = realloc(p, size + 100);
			if (p == NULL)
				err(1, NULL);
			size += 100;
		}
		/* Strip control characters */
		if (!iscntrl(c) || c == '\t' || c == '\n')
			p[i++] = c;
	}

	if (i == size) {
		p = realloc(p, size + 1);
		if (p == NULL)
			err(1, NULL);
	}

	p[i] = '\0';

	encode_word(&p);
	wrap_header(&p, 0);
	printf("%s", p);

	free(p);
}

void
encode_word(char **p)
{
	char *s = NULL;
	char *r = NULL;
	char *word = NULL;
	char *encoded_word = NULL;
	size_t i = 0;
	size_t n = 0;
	size_t size = 0;
	size_t word_len = 0;
	size_t enc_word_len = 0;
	char otag[] = "=?UTF-8?B?";
	char ctag[] = "?=";
	size_t tags_len = strlen(otag) + strlen(ctag);
	int utf8word = 0;

	setlocale(LC_CTYPE, "en_US.UTF-8");

	size = strlen(*p);
	s = malloc(size);
	if (s == NULL)
		err(1, NULL);

	r = *p;

	while (r[n] != '\0') {
		/* We are at the first character of a word */
		if ((n == 0 || (isspace(r[n - 1]) || r[n - 1] == '\n')) &&
		    !isspace(r[n]) && r[n] != '\n') {

			/* Load the size of the word in a variable */
			word_len = strcspn(&r[n], "\t\n ");

			/*
			 * If we are not at the first character of the
			 * string count one more character[1].
			 */
			if (n > 0 && utf8word)
				word_len++;

			word = malloc(word_len + 1);
			if (word == NULL)
				err(1, NULL);

			/*
			 * (1) If the pervious word was encoded we need
			 * to include the previous space or tab in the
			 * encoded word.
			 */
			if (n > 0 && utf8word) {
				if (r[n - 1] == '\n')
					r[n - 1] = ' ';
				snprintf(word, word_len + 1, "%s", &r[n - 1]);
			} else
				snprintf(word, word_len + 1, "%s", &r[n]);

			/*
			 * Check if the word contains UTF-8 characters
			 */
			if (word_len > mbstowcs(NULL, word, 0) ) {
				enc_word_len = (word_len + 2) / 3 * 4;
				s = realloc(s, size + tags_len +
				    (enc_word_len - word_len) + 1);
				if (s == NULL)
					err(1, NULL);
				size += tags_len +
				    (enc_word_len - word_len) + 1;

				encoded_word = malloc(enc_word_len + 1);
				if (encoded_word == NULL)
					err(1, NULL);

				/* Encode the word */
				b64enc(word, word_len, encoded_word,
				    enc_word_len);
				encoded_word[enc_word_len] = '\0';

				/* Append the encoded word to the string */
				snprintf(&s[i], i + enc_word_len +
				    tags_len + 1, "%s%s%s", otag,
				    encoded_word, ctag);
				i += enc_word_len + tags_len;
			
				free(encoded_word);
				n += word_len;

				/*
				 * If we included the last space or tab in
				 * the encoded word shift the count to
				 * prevent the last caracter of the word
				 * be loaded twice in the destination
				 * string.
				 */
				if (n > 0 && utf8word)
					n--;

				utf8word = 1;
			} else
				utf8word = 0;
			free(word);
		}
		s[i++] = r[n++];
	}
	if (i == size) {
		s = realloc(s, size + 1);
		if (s == NULL)
			err(1, NULL);
	}
	s[i] = '\0';

	/* Remove this free() from mail patch */
	free(r);

	*p = s;
}

void
wrap_header(char **p, size_t name_len)
{
	char *s = NULL;
	char *r = NULL;
	size_t i = 0;
	size_t n = 0;
	size_t size = 0;
	int col = name_len;	/* let room for header name */
	int wrap = 72;

	size = strlen(*p);
	s = malloc(size);
	if (s == NULL)
		err(1, NULL);

	r = *p;

	while (r[n] != '\0') {
		/* Replace newlines by spaces */
		if (r[n] == '\n' && r[n + 1] != '\0')
			r[n] = ' ';
		if (n != 0 && isspace(r[n]) &&
		    col + strcspn(&r[n + 1], "\t\n ") >= wrap) {
			s = realloc(s, size + 1);
			if (s == NULL)
				err(1, NULL);
			size++;
			if (r[n + 1] != '\0')
				s[i++] = '\n';
			while (isspace(r[n + 1]))
				n++;
			col = 0;
			if (r[n] == '\t')
				col += 7;
		}
		s[i++] = r[n++];
		if (r[n] == '\t')
			col += 7;
		col++;
	}
	if (i == size) {
		s = realloc(s, size + 1);
		if (s == NULL)
			err(1, NULL);
	}
	s[i] = '\0';

	/* Remove this free from mail patch */
	free(r);

	*p = s;
}

void
usage(void)
{
	extern char *__progname;

	fprintf(stderr,
		"Usage: %s [-bhlnp] [-w width] [file ...]\n"
		"  -h   print this help\n",
		__progname);
	exit(1);
}

-- 
Walter