Blame lib/str-unicode.c

Packit Service 4684c1
/*
Packit Service 4684c1
 * Copyright (C) 2016, 2017 Red Hat, Inc.
Packit Service 4684c1
 *
Packit Service 4684c1
 * Author: Nikos Mavrogiannopoulos
Packit Service 4684c1
 *
Packit Service 4684c1
 * This file is part of GnuTLS.
Packit Service 4684c1
 *
Packit Service 4684c1
 * The GnuTLS is free software; you can redistribute it and/or
Packit Service 4684c1
 * modify it under the terms of the GNU Lesser General Public License
Packit Service 4684c1
 * as published by the Free Software Foundation; either version 2.1 of
Packit Service 4684c1
 * the License, or (at your option) any later version.
Packit Service 4684c1
 *
Packit Service 4684c1
 * This library is distributed in the hope that it will be useful, but
Packit Service 4684c1
 * WITHOUT ANY WARRANTY; without even the implied warranty of
Packit Service 4684c1
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
Packit Service 4684c1
 * Lesser General Public License for more details.
Packit Service 4684c1
 *
Packit Service 4684c1
 * You should have received a copy of the GNU Lesser General Public License
Packit Service 4684c1
 * along with this program.  If not, see <https://www.gnu.org/licenses/>
Packit Service 4684c1
 *
Packit Service 4684c1
 */
Packit Service 4684c1
Packit Service 4684c1
#include "gnutls_int.h"
Packit Service 4684c1
#include "errors.h"
Packit Service 4684c1
#include "str.h"
Packit Service 4684c1
#include <uninorm.h>
Packit Service 4684c1
#include <unistr.h>
Packit Service 4684c1
#include <unictype.h>
Packit Service 4684c1
Packit Service 4684c1
/* rfc5892#section-2.6 exceptions
Packit Service 4684c1
 */
Packit Service 4684c1
inline static int is_allowed_exception(uint32_t ch)
Packit Service 4684c1
{
Packit Service 4684c1
	switch (ch) {
Packit Service 4684c1
		case 0xB7:
Packit Service 4684c1
		case 0x0375:
Packit Service 4684c1
		case 0x05F3:
Packit Service 4684c1
		case 0x05F4:
Packit Service 4684c1
		case 0x30FB:
Packit Service 4684c1
		case 0x0660:
Packit Service 4684c1
		case 0x0661:
Packit Service 4684c1
		case 0x0662:
Packit Service 4684c1
		case 0x0663:
Packit Service 4684c1
		case 0x0664:
Packit Service 4684c1
		case 0x0665:
Packit Service 4684c1
		case 0x0666:
Packit Service 4684c1
		case 0x0667:
Packit Service 4684c1
		case 0x0668:
Packit Service 4684c1
		case 0x0669:
Packit Service 4684c1
		case 0x06F0:
Packit Service 4684c1
		case 0x06F1:
Packit Service 4684c1
		case 0x06F2:
Packit Service 4684c1
		case 0x06F3:
Packit Service 4684c1
		case 0x06F4:
Packit Service 4684c1
		case 0x06F5:
Packit Service 4684c1
		case 0x06F6:
Packit Service 4684c1
		case 0x06F7:
Packit Service 4684c1
		case 0x06F8:
Packit Service 4684c1
		case 0x06F9:
Packit Service 4684c1
		case 0x0640:
Packit Service 4684c1
		case 0x07FA:
Packit Service 4684c1
		case 0x302E:
Packit Service 4684c1
		case 0x302F:
Packit Service 4684c1
		case 0x3031:
Packit Service 4684c1
		case 0x3032:
Packit Service 4684c1
		case 0x3033:
Packit Service 4684c1
		case 0x3034:
Packit Service 4684c1
		case 0x3035:
Packit Service 4684c1
		case 0x303B:
Packit Service 4684c1
			return 0; /* disallowed */
Packit Service 4684c1
		case 0xDF:
Packit Service 4684c1
		case 0x03C2:
Packit Service 4684c1
		case 0x06FD:
Packit Service 4684c1
		case 0x06FE:
Packit Service 4684c1
		case 0x0F0B:
Packit Service 4684c1
		case 0x3007:
Packit Service 4684c1
			return 1; /* allowed */
Packit Service 4684c1
		default:
Packit Service 4684c1
			return -1; /* not exception */
Packit Service 4684c1
	}
Packit Service 4684c1
}
Packit Service 4684c1
Packit Service 4684c1
/* Checks whether the provided string is in the valid set of FreeFormClass (RFC7564
Packit Service 4684c1
 * as an RFC7613 requirement), and converts all spaces to the ASCII-space. */
Packit Service 4684c1
static int check_for_valid_freeformclass(uint32_t *ucs4, unsigned ucs4_size)
Packit Service 4684c1
{
Packit Service 4684c1
	unsigned i;
Packit Service 4684c1
	int rc;
Packit Service 4684c1
	uint32_t tmp[4];
Packit Service 4684c1
	size_t tmp_size;
Packit Service 4684c1
	uint32_t *nrm;
Packit Service 4684c1
	uc_general_category_t cat;
Packit Service 4684c1
	unsigned is_invalid;
Packit Service 4684c1
Packit Service 4684c1
	/* make the union of Valid categories, excluding any invalid (i.e., control) */
Packit Service 4684c1
	cat = uc_general_category_or(UC_CATEGORY_Ll, UC_CATEGORY_Lu); /* LetterDigits */
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Lo);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Nd);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Lm);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Mn);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Mc);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Lt); /* OtherLetterDigits */
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Nl);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_No);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Me);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Sm); /* Symbols */
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Sc);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_So);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Sk);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Pc); /* Punctuation */
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Pd);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Ps);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Pe);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Pi);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Pf);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Po);
Packit Service 4684c1
	cat = uc_general_category_or(cat, UC_CATEGORY_Zs); /* Spaces */
Packit Service 4684c1
	cat = uc_general_category_and_not(cat, UC_CATEGORY_Cc); /* Not in Control */
Packit Service 4684c1
Packit Service 4684c1
	/* check for being in the allowed sets in rfc7564#section-4.3 */
Packit Service 4684c1
	for (i=0;i
Packit Service 4684c1
		is_invalid = 0;
Packit Service 4684c1
Packit Service 4684c1
		/* Disallowed 
Packit Service 4684c1
		   o  Old Hangul Jamo characters, i.e., the OldHangulJamo ("I") category
Packit Service 4684c1
		      (not handled in this code)
Packit Service 4684c1
Packit Service 4684c1
		   o  Control characters, i.e., the Controls ("L") category
Packit Service 4684c1
Packit Service 4684c1
		   o  Ignorable characters, i.e., the PrecisIgnorableProperties ("M")
Packit Service 4684c1
		 */
Packit Service 4684c1
		if (uc_is_property_default_ignorable_code_point(ucs4[i]) ||
Packit Service 4684c1
		    uc_is_property_not_a_character(ucs4[i])) {
Packit Service 4684c1
			return gnutls_assert_val(GNUTLS_E_INVALID_UTF8_STRING);
Packit Service 4684c1
		}
Packit Service 4684c1
Packit Service 4684c1
Packit Service 4684c1
		/* Contextual rules - we do not implement them / we reject chars from these sets
Packit Service 4684c1
		   o  A number of characters from the Exceptions ("F") category defined
Packit Service 4684c1
Packit Service 4684c1
		   o  Joining characters, i.e., the JoinControl ("H") category defined
Packit Service 4684c1
		 */
Packit Service 4684c1
		rc = is_allowed_exception(ucs4[i]);
Packit Service 4684c1
		if (rc == 0 || uc_is_property_join_control(ucs4[i]))
Packit Service 4684c1
			return gnutls_assert_val(GNUTLS_E_INVALID_UTF8_STRING);
Packit Service 4684c1
Packit Service 4684c1
		if (rc == 1) /* exceptionally allowed, continue */
Packit Service 4684c1
			continue;
Packit Service 4684c1
Packit Service 4684c1
Packit Service 4684c1
		/* Replace all spaces; an RFC7613 requirement
Packit Service 4684c1
		 */
Packit Service 4684c1
		if (uc_is_general_category(ucs4[i], UC_CATEGORY_Zs)) /* replace */
Packit Service 4684c1
			ucs4[i] = 0x20;
Packit Service 4684c1
Packit Service 4684c1
		/* Valid */
Packit Service 4684c1
		if ((ucs4[i] < 0x21 || ucs4[i] > 0x7E) && !uc_is_general_category(ucs4[i], cat))
Packit Service 4684c1
			is_invalid = 1;
Packit Service 4684c1
Packit Service 4684c1
		/* HasCompat */
Packit Service 4684c1
		if (is_invalid) {
Packit Service 4684c1
			tmp_size = sizeof(tmp)/sizeof(tmp[0]);
Packit Service 4684c1
			nrm = u32_normalize(UNINORM_NFKC, &ucs4[i], 1, tmp, &tmp_size);
Packit Service 4684c1
			if (nrm == NULL || (tmp_size == 1 && nrm[0] == ucs4[i]))
Packit Service 4684c1
				return gnutls_assert_val(GNUTLS_E_INVALID_UTF8_STRING);
Packit Service 4684c1
		}
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	return 0;
Packit Service 4684c1
}
Packit Service 4684c1
Packit Service 4684c1
Packit Service 4684c1
/**
Packit Service 4684c1
 * gnutls_utf8_password_normalize:
Packit Service 4684c1
 * @password: contain the UTF-8 formatted password
Packit Service 4684c1
 * @plen: the length of the provided password
Packit Service 4684c1
 * @out: the result in an null-terminated allocated string
Packit Service 4684c1
 * @flags: should be zero
Packit Service 4684c1
 *
Packit Service 4684c1
 * This function will convert the provided UTF-8 password according
Packit Service 4684c1
 * to the normalization rules in RFC7613.
Packit Service 4684c1
 *
Packit Service 4684c1
 * If the flag %GNUTLS_UTF8_IGNORE_ERRS is specified, any UTF-8 encoding
Packit Service 4684c1
 * errors will be ignored, and in that case the output will be a copy of the input.
Packit Service 4684c1
 *
Packit Service 4684c1
 * Returns: %GNUTLS_E_INVALID_UTF8_STRING on invalid UTF-8 data, or 0 on success.
Packit Service 4684c1
 *
Packit Service 4684c1
 * Since: 3.5.7
Packit Service 4684c1
 **/
Packit Service 4684c1
int gnutls_utf8_password_normalize(const unsigned char *password, unsigned plen,
Packit Service 4684c1
				   gnutls_datum_t *out, unsigned flags)
Packit Service 4684c1
{
Packit Service 4684c1
	size_t ucs4_size = 0, nrm_size = 0;
Packit Service 4684c1
	size_t final_size = 0;
Packit Service 4684c1
	uint8_t *final = NULL;
Packit Service 4684c1
	uint32_t *ucs4 = NULL;
Packit Service 4684c1
	uint32_t *nrm = NULL;
Packit Service 4684c1
	uint8_t *nrmu8 = NULL;
Packit Service 4684c1
	int ret;
Packit Service 4684c1
Packit Service 4684c1
	if (plen == 0) {
Packit Service 4684c1
		out->data = (uint8_t*)gnutls_strdup("");
Packit Service 4684c1
		out->size = 0;
Packit Service 4684c1
		if (out->data == NULL)
Packit Service 4684c1
			return gnutls_assert_val(GNUTLS_E_MEMORY_ERROR);
Packit Service 4684c1
		return 0;
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	/* check for invalid UTF-8 */
Packit Service 4684c1
	if (u8_check((uint8_t*)password, plen) != NULL) {
Packit Service 4684c1
		gnutls_assert();
Packit Service 4684c1
		if (flags & GNUTLS_UTF8_IGNORE_ERRS) {
Packit Service 4684c1
 raw_copy:
Packit Service 4684c1
			out->data = gnutls_malloc(plen+1);
Packit Service 4684c1
			if (out->data == NULL)
Packit Service 4684c1
				return gnutls_assert_val(GNUTLS_E_MEMORY_ERROR);
Packit Service 4684c1
			out->size = plen;
Packit Service 4684c1
			memcpy(out->data, password, plen);
Packit Service 4684c1
			out->data[plen] = 0;
Packit Service 4684c1
			return 0;
Packit Service 4684c1
		} else {
Packit Service 4684c1
			return GNUTLS_E_INVALID_UTF8_STRING;
Packit Service 4684c1
		}
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	/* convert to UTF-32 */
Packit Service 4684c1
	ucs4 = u8_to_u32((uint8_t*)password, plen, NULL, &ucs4_size);
Packit Service 4684c1
	if (ucs4 == NULL) {
Packit Service 4684c1
		gnutls_assert();
Packit Service 4684c1
		ret = GNUTLS_E_PARSING_ERROR;
Packit Service 4684c1
		goto fail;
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	ret = check_for_valid_freeformclass(ucs4, ucs4_size);
Packit Service 4684c1
	if (ret < 0) {
Packit Service 4684c1
		gnutls_assert();
Packit Service 4684c1
		if (flags & GNUTLS_UTF8_IGNORE_ERRS) {
Packit Service 4684c1
			free(ucs4);
Packit Service 4684c1
			goto raw_copy;
Packit Service 4684c1
		}
Packit Service 4684c1
		if (ret == GNUTLS_E_INVALID_UTF8_STRING)
Packit Service 4684c1
			ret = GNUTLS_E_INVALID_PASSWORD_STRING;
Packit Service 4684c1
		goto fail;
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	/* normalize to NFC */
Packit Service 4684c1
	nrm = u32_normalize(UNINORM_NFC, ucs4, ucs4_size, NULL, &nrm_size);
Packit Service 4684c1
	if (nrm == NULL) {
Packit Service 4684c1
		gnutls_assert();
Packit Service 4684c1
		ret = GNUTLS_E_INVALID_PASSWORD_STRING;
Packit Service 4684c1
		goto fail;
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	/* convert back to UTF-8 */
Packit Service 4684c1
	final_size = 0;
Packit Service 4684c1
	nrmu8 = u32_to_u8(nrm, nrm_size, NULL, &final_size);
Packit Service 4684c1
	if (nrmu8 == NULL) {
Packit Service 4684c1
		gnutls_assert();
Packit Service 4684c1
		ret = GNUTLS_E_INVALID_PASSWORD_STRING;
Packit Service 4684c1
		goto fail;
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	/* copy to output with null terminator */
Packit Service 4684c1
	final = gnutls_malloc(final_size+1);
Packit Service 4684c1
	if (final == NULL) {
Packit Service 4684c1
		gnutls_assert();
Packit Service 4684c1
		ret = GNUTLS_E_MEMORY_ERROR;
Packit Service 4684c1
		goto fail;
Packit Service 4684c1
	}
Packit Service 4684c1
Packit Service 4684c1
	memcpy(final, nrmu8, final_size);
Packit Service 4684c1
	final[final_size] = 0;
Packit Service 4684c1
Packit Service 4684c1
	free(ucs4);
Packit Service 4684c1
	free(nrm);
Packit Service 4684c1
	free(nrmu8);
Packit Service 4684c1
Packit Service 4684c1
	out->data = final;
Packit Service 4684c1
	out->size = final_size;
Packit Service 4684c1
Packit Service 4684c1
	return 0;
Packit Service 4684c1
Packit Service 4684c1
 fail:
Packit Service 4684c1
	gnutls_free(final);
Packit Service 4684c1
	free(ucs4);
Packit Service 4684c1
	free(nrm);
Packit Service 4684c1
	free(nrmu8);
Packit Service 4684c1
	return ret;
Packit Service 4684c1
}
Packit Service 4684c1