Blame crypto/asn1/a_utf8.c

Packit c4476c
/*
Packit c4476c
 * Copyright 1995-2016 The OpenSSL Project Authors. All Rights Reserved.
Packit c4476c
 *
Packit c4476c
 * Licensed under the OpenSSL license (the "License").  You may not use
Packit c4476c
 * this file except in compliance with the License.  You can obtain a copy
Packit c4476c
 * in the file LICENSE in the source distribution or at
Packit c4476c
 * https://www.openssl.org/source/license.html
Packit c4476c
 */
Packit c4476c
Packit c4476c
#include <stdio.h>
Packit c4476c
#include "internal/cryptlib.h"
Packit c4476c
#include <openssl/asn1.h>
Packit c4476c
Packit c4476c
/* UTF8 utilities */
Packit c4476c
Packit c4476c
/*-
Packit c4476c
 * This parses a UTF8 string one character at a time. It is passed a pointer
Packit c4476c
 * to the string and the length of the string. It sets 'value' to the value of
Packit c4476c
 * the current character. It returns the number of characters read or a
Packit c4476c
 * negative error code:
Packit c4476c
 * -1 = string too short
Packit c4476c
 * -2 = illegal character
Packit c4476c
 * -3 = subsequent characters not of the form 10xxxxxx
Packit c4476c
 * -4 = character encoded incorrectly (not minimal length).
Packit c4476c
 */
Packit c4476c
Packit c4476c
int UTF8_getc(const unsigned char *str, int len, unsigned long *val)
Packit c4476c
{
Packit c4476c
    const unsigned char *p;
Packit c4476c
    unsigned long value;
Packit c4476c
    int ret;
Packit c4476c
    if (len <= 0)
Packit c4476c
        return 0;
Packit c4476c
    p = str;
Packit c4476c
Packit c4476c
    /* Check syntax and work out the encoded value (if correct) */
Packit c4476c
    if ((*p & 0x80) == 0) {
Packit c4476c
        value = *p++ & 0x7f;
Packit c4476c
        ret = 1;
Packit c4476c
    } else if ((*p & 0xe0) == 0xc0) {
Packit c4476c
        if (len < 2)
Packit c4476c
            return -1;
Packit c4476c
        if ((p[1] & 0xc0) != 0x80)
Packit c4476c
            return -3;
Packit c4476c
        value = (*p++ & 0x1f) << 6;
Packit c4476c
        value |= *p++ & 0x3f;
Packit c4476c
        if (value < 0x80)
Packit c4476c
            return -4;
Packit c4476c
        ret = 2;
Packit c4476c
    } else if ((*p & 0xf0) == 0xe0) {
Packit c4476c
        if (len < 3)
Packit c4476c
            return -1;
Packit c4476c
        if (((p[1] & 0xc0) != 0x80)
Packit c4476c
            || ((p[2] & 0xc0) != 0x80))
Packit c4476c
            return -3;
Packit c4476c
        value = (*p++ & 0xf) << 12;
Packit c4476c
        value |= (*p++ & 0x3f) << 6;
Packit c4476c
        value |= *p++ & 0x3f;
Packit c4476c
        if (value < 0x800)
Packit c4476c
            return -4;
Packit c4476c
        ret = 3;
Packit c4476c
    } else if ((*p & 0xf8) == 0xf0) {
Packit c4476c
        if (len < 4)
Packit c4476c
            return -1;
Packit c4476c
        if (((p[1] & 0xc0) != 0x80)
Packit c4476c
            || ((p[2] & 0xc0) != 0x80)
Packit c4476c
            || ((p[3] & 0xc0) != 0x80))
Packit c4476c
            return -3;
Packit c4476c
        value = ((unsigned long)(*p++ & 0x7)) << 18;
Packit c4476c
        value |= (*p++ & 0x3f) << 12;
Packit c4476c
        value |= (*p++ & 0x3f) << 6;
Packit c4476c
        value |= *p++ & 0x3f;
Packit c4476c
        if (value < 0x10000)
Packit c4476c
            return -4;
Packit c4476c
        ret = 4;
Packit c4476c
    } else if ((*p & 0xfc) == 0xf8) {
Packit c4476c
        if (len < 5)
Packit c4476c
            return -1;
Packit c4476c
        if (((p[1] & 0xc0) != 0x80)
Packit c4476c
            || ((p[2] & 0xc0) != 0x80)
Packit c4476c
            || ((p[3] & 0xc0) != 0x80)
Packit c4476c
            || ((p[4] & 0xc0) != 0x80))
Packit c4476c
            return -3;
Packit c4476c
        value = ((unsigned long)(*p++ & 0x3)) << 24;
Packit c4476c
        value |= ((unsigned long)(*p++ & 0x3f)) << 18;
Packit c4476c
        value |= ((unsigned long)(*p++ & 0x3f)) << 12;
Packit c4476c
        value |= (*p++ & 0x3f) << 6;
Packit c4476c
        value |= *p++ & 0x3f;
Packit c4476c
        if (value < 0x200000)
Packit c4476c
            return -4;
Packit c4476c
        ret = 5;
Packit c4476c
    } else if ((*p & 0xfe) == 0xfc) {
Packit c4476c
        if (len < 6)
Packit c4476c
            return -1;
Packit c4476c
        if (((p[1] & 0xc0) != 0x80)
Packit c4476c
            || ((p[2] & 0xc0) != 0x80)
Packit c4476c
            || ((p[3] & 0xc0) != 0x80)
Packit c4476c
            || ((p[4] & 0xc0) != 0x80)
Packit c4476c
            || ((p[5] & 0xc0) != 0x80))
Packit c4476c
            return -3;
Packit c4476c
        value = ((unsigned long)(*p++ & 0x1)) << 30;
Packit c4476c
        value |= ((unsigned long)(*p++ & 0x3f)) << 24;
Packit c4476c
        value |= ((unsigned long)(*p++ & 0x3f)) << 18;
Packit c4476c
        value |= ((unsigned long)(*p++ & 0x3f)) << 12;
Packit c4476c
        value |= (*p++ & 0x3f) << 6;
Packit c4476c
        value |= *p++ & 0x3f;
Packit c4476c
        if (value < 0x4000000)
Packit c4476c
            return -4;
Packit c4476c
        ret = 6;
Packit c4476c
    } else
Packit c4476c
        return -2;
Packit c4476c
    *val = value;
Packit c4476c
    return ret;
Packit c4476c
}
Packit c4476c
Packit c4476c
/*
Packit c4476c
 * This takes a character 'value' and writes the UTF8 encoded value in 'str'
Packit c4476c
 * where 'str' is a buffer containing 'len' characters. Returns the number of
Packit c4476c
 * characters written or -1 if 'len' is too small. 'str' can be set to NULL
Packit c4476c
 * in which case it just returns the number of characters. It will need at
Packit c4476c
 * most 6 characters.
Packit c4476c
 */
Packit c4476c
Packit c4476c
int UTF8_putc(unsigned char *str, int len, unsigned long value)
Packit c4476c
{
Packit c4476c
    if (!str)
Packit c4476c
        len = 6;                /* Maximum we will need */
Packit c4476c
    else if (len <= 0)
Packit c4476c
        return -1;
Packit c4476c
    if (value < 0x80) {
Packit c4476c
        if (str)
Packit c4476c
            *str = (unsigned char)value;
Packit c4476c
        return 1;
Packit c4476c
    }
Packit c4476c
    if (value < 0x800) {
Packit c4476c
        if (len < 2)
Packit c4476c
            return -1;
Packit c4476c
        if (str) {
Packit c4476c
            *str++ = (unsigned char)(((value >> 6) & 0x1f) | 0xc0);
Packit c4476c
            *str = (unsigned char)((value & 0x3f) | 0x80);
Packit c4476c
        }
Packit c4476c
        return 2;
Packit c4476c
    }
Packit c4476c
    if (value < 0x10000) {
Packit c4476c
        if (len < 3)
Packit c4476c
            return -1;
Packit c4476c
        if (str) {
Packit c4476c
            *str++ = (unsigned char)(((value >> 12) & 0xf) | 0xe0);
Packit c4476c
            *str++ = (unsigned char)(((value >> 6) & 0x3f) | 0x80);
Packit c4476c
            *str = (unsigned char)((value & 0x3f) | 0x80);
Packit c4476c
        }
Packit c4476c
        return 3;
Packit c4476c
    }
Packit c4476c
    if (value < 0x200000) {
Packit c4476c
        if (len < 4)
Packit c4476c
            return -1;
Packit c4476c
        if (str) {
Packit c4476c
            *str++ = (unsigned char)(((value >> 18) & 0x7) | 0xf0);
Packit c4476c
            *str++ = (unsigned char)(((value >> 12) & 0x3f) | 0x80);
Packit c4476c
            *str++ = (unsigned char)(((value >> 6) & 0x3f) | 0x80);
Packit c4476c
            *str = (unsigned char)((value & 0x3f) | 0x80);
Packit c4476c
        }
Packit c4476c
        return 4;
Packit c4476c
    }
Packit c4476c
    if (value < 0x4000000) {
Packit c4476c
        if (len < 5)
Packit c4476c
            return -1;
Packit c4476c
        if (str) {
Packit c4476c
            *str++ = (unsigned char)(((value >> 24) & 0x3) | 0xf8);
Packit c4476c
            *str++ = (unsigned char)(((value >> 18) & 0x3f) | 0x80);
Packit c4476c
            *str++ = (unsigned char)(((value >> 12) & 0x3f) | 0x80);
Packit c4476c
            *str++ = (unsigned char)(((value >> 6) & 0x3f) | 0x80);
Packit c4476c
            *str = (unsigned char)((value & 0x3f) | 0x80);
Packit c4476c
        }
Packit c4476c
        return 5;
Packit c4476c
    }
Packit c4476c
    if (len < 6)
Packit c4476c
        return -1;
Packit c4476c
    if (str) {
Packit c4476c
        *str++ = (unsigned char)(((value >> 30) & 0x1) | 0xfc);
Packit c4476c
        *str++ = (unsigned char)(((value >> 24) & 0x3f) | 0x80);
Packit c4476c
        *str++ = (unsigned char)(((value >> 18) & 0x3f) | 0x80);
Packit c4476c
        *str++ = (unsigned char)(((value >> 12) & 0x3f) | 0x80);
Packit c4476c
        *str++ = (unsigned char)(((value >> 6) & 0x3f) | 0x80);
Packit c4476c
        *str = (unsigned char)((value & 0x3f) | 0x80);
Packit c4476c
    }
Packit c4476c
    return 6;
Packit c4476c
}