/* coding.c

 *

 * Copyright (C) 2006-2026 wolfSSL Inc.

 *

 * This file is part of wolfSSL.

 *

 * wolfSSL is free software; you can redistribute it and/or modify

 * it under the terms of the GNU General Public License as published by

 * the Free Software Foundation; either version 3 of the License, or

 * (at your option) any later version.

 *

 * wolfSSL is distributed in the hope that it will be useful,

 * but WITHOUT ANY WARRANTY; without even the implied warranty of

 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

 * GNU General Public License for more details.

 *

 * You should have received a copy of the GNU General Public License

 * along with this program; if not, write to the Free Software

 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1335, USA

 */

#include <wolfssl/wolfcrypt/libwolfssl_sources.h>

#include <wolfssl/wolfcrypt/coding.h>

#ifndef NO_CODING

#ifndef NO_ASN

    #include <wolfssl/wolfcrypt/asn.h> /* For PEM_LINE_SZ */

#endif

#ifdef NO_INLINE

    #include <wolfssl/wolfcrypt/misc.h>

#else

    #define WOLFSSL_MISC_INCLUDED

    #include <wolfcrypt/src/misc.c>

#endif

enum {

    BAD         = 0xFF,  /* invalid encoding */

    PAD         = '=',

    BASE64_MIN  = 0x2B,

    BASE16_MIN  = 0x30

};

#ifndef BASE64_LINE_SZ

    #ifdef NO_ASN

        #define BASE64_LINE_SZ 64

    #else

        #define BASE64_LINE_SZ PEM_LINE_SZ

    #endif

#endif

#ifdef WOLFSSL_BASE64_DECODE

static WC_INLINE byte Base64_Char2Val_CT(byte c)

{

    int v;

    int smallEnd   = (int)c - 0x7b;

    int smallStart = (int)c - 0x61;

    int bigEnd     = (int)c - 0x5b;

    int bigStart   = (int)c - 0x41;

    int numEnd     = (int)c - 0x3a;

    int numStart   = (int)c - 0x30;

    int slashEnd   = (int)c - 0x30;

    int slashStart = (int)c - 0x2f;

    int plusEnd    = (int)c - 0x2c;

    int plusStart  = (int)c - 0x2b;

    v  = ((smallStart >> 8) ^ (smallEnd >> 8)) & (smallStart + 26 + 1);

    v |= ((bigStart   >> 8) ^ (bigEnd   >> 8)) & (bigStart   +  0 + 1);

    v |= ((numStart   >> 8) ^ (numEnd   >> 8)) & (numStart   + 52 + 1);

    v |= ((slashStart >> 8) ^ (slashEnd >> 8)) & (slashStart + 63 + 1);

    v |= ((plusStart  >> 8) ^ (plusEnd  >> 8)) & (plusStart  + 62 + 1);

    return WC_OCTET(v - 1);

}

#ifndef BASE64_NO_TABLE

static

ALIGN64 const byte base64Decode_table[] = {    /* + starts at 0x2B */

/* 0x28:       + , - . / */                   62, BAD, BAD, BAD,  63,

/* 0x30: 0 1 2 3 4 5 6 7 */    52,  53,  54,  55,  56,  57,  58,  59,

/* 0x38: 8 9 : ; < = > ? */    60,  61, BAD, BAD, BAD, BAD, BAD, BAD,

/* 0x40: @ A B C D E F G */   BAD,   0,   1,   2,   3,   4,   5,   6,

/* 0x48: H I J K L M N O */     7,   8,   9,  10,  11,  12,  13,  14,

/* 0x50: P Q R S T U V W */    15,  16,  17,  18,  19,  20,  21,  22,

/* 0x58: X Y Z [ \ ] ^ _ */    23,  24,  25, BAD, BAD, BAD, BAD, BAD,

/* 0x60: ` a b c d e f g */   BAD,  26,  27,  28,  29,  30,  31,  32,

/* 0x68: h i j k l m n o */    33,  34,  35,  36,  37,  38,  39,  40,

/* 0x70: p q r s t u v w */    41,  42,  43,  44,  45,  46,  47,  48,

/* 0x78: x y z           */    49,  50,  51

                            };

#define BASE64DECODE_TABLE_SZ    (byte)(sizeof(base64Decode_table))

static WC_INLINE byte Base64_Char2Val_by_table(byte c)

{

#ifdef WC_CACHE_RESISTANT_BASE64_TABLE

    /* 80 characters in table.

     * 64 bytes in a cache line - first line has 64, second has 16

     */

    byte v;

    byte mask;

    c = (byte)(c - BASE64_MIN);

    mask = (byte)((((byte)(0x3f - c)) >> 7) - 1);

    /* Load a value from the first cache line and use when mask set. */

    v  = (byte)(base64Decode_table[ c & 0x3f        ] &   mask);

    /* Load a value from the second cache line and use when mask not set. */

    v |= (byte)(base64Decode_table[(c & 0x0f) | 0x40] & (~mask));

    return v;

#else

    return base64Decode_table[c - BASE64_MIN];

#endif

}

#endif /* !BASE64_NO_TABLE */

int Base64_SkipNewline(const byte* in, word32 *inLen,

  word32 *outJ)

{

    word32 len = *inLen;

    word32 j = *outJ;

    byte curChar;

    if (len == 0) {

        return BUFFER_E;

    }

    curChar = in[j];

    while (len > 1 && curChar == ' ') {

        /* skip whitespace in the middle or end of line */

        curChar = in[++j];

        len--;

    }

    if (curChar == '\r' || curChar == '\n') {

        j++;

        len--;

        if (curChar == '\r') {

            if (len) {

                curChar = in[j++];

                len--;

            }

        }

        if (curChar != '\n') {

            WOLFSSL_MSG("Bad end of line in Base64 Decode");

            return ASN_INPUT_E;

        }

        if (len) {

            curChar = in[j];

        }

    }

    while (len && curChar == ' ') {

        if (--len > 0) {

            curChar = in[++j];

        }

    }

    if (!len) {

        return BUFFER_E;

    }

    *inLen = len;

    *outJ = j;

    return 0;

}

#ifndef BASE64_NO_TABLE

int Base64_Decode_nonCT(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    word32 i = 0;

    word32 j = 0;

    int ret;

    const byte maxIdx = BASE64DECODE_TABLE_SZ + BASE64_MIN - 1;

    if ((in == NULL && inLen > 0) || out == NULL || outLen == NULL)

        return BAD_FUNC_ARG;

    while (inLen > 3) {

        int pad3 = 0;

        int pad4 = 0;

        byte b1, b2, b3;

        byte e1, e2, e3, e4;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            if (ret == WC_NO_ERR_TRACE(BUFFER_E)) {

                /* Running out of buffer here is not an error */

                break;

            }

            return ret;

        }

        e1 = in[j++];

        if (e1 == '\0') {

            inLen = 0;

            break;

        }

        inLen--;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            return ret;

        }

        e2 = in[j++];

        inLen--;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            return ret;

        }

        e3 = in[j++];

        inLen--;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            return ret;

        }

        e4 = in[j++];

        inLen--;

        if (e3 == PAD)

            pad3 = 1;

        if (e4 == PAD)

            pad4 = 1;

        if (pad3 && !pad4)

            return ASN_INPUT_E;

        if (e1 < BASE64_MIN || e2 < BASE64_MIN || e3 < BASE64_MIN ||

                                                              e4 < BASE64_MIN) {

            WOLFSSL_MSG("Bad Base64 Decode data, too small");

            return ASN_INPUT_E;

        }

        if (e1 > maxIdx || e2 > maxIdx || e3 > maxIdx || e4 > maxIdx) {

            WOLFSSL_MSG("Bad Base64 Decode data, too big");

            return ASN_INPUT_E;

        }

        e1 = Base64_Char2Val_by_table(e1);

        e2 = Base64_Char2Val_by_table(e2);

        e3 = (byte)((e3 == PAD) ? 0 : Base64_Char2Val_by_table(e3));

        e4 = (byte)((e4 == PAD) ? 0 : Base64_Char2Val_by_table(e4));

        if (e1 == BAD || e2 == BAD || e3 == BAD || e4 == BAD) {

            WOLFSSL_MSG("Bad Base64 Decode bad character");

            return ASN_INPUT_E;

        }

        if (i + 1 + !pad3 + !pad4 > *outLen) {

            WOLFSSL_MSG("Bad Base64 Decode out buffer, too small");

            return BUFFER_E;

        }

        b1 = (byte)((e1 << 2) | (e2 >> 4));

        b2 = (byte)(((e2 & 0xF) << 4) | (e3 >> 2));

        b3 = (byte)(((e3 & 0x3) << 6) | e4);

        out[i++] = b1;

        if (!pad3)

            out[i++] = b2;

        if (!pad4)

            out[i++] = b3;

        else

            break;

    }

    /* If there is still input available, and it's not whitespace or nulls, then

     * the input is invalid.

     */

    while (inLen > 0) {

        word32 cur_j = j;

        if (in[j] == 0)

            break;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            if (ret == WC_NO_ERR_TRACE(BUFFER_E)) {

                /* Running out of buffer here is not an error */

                break;

            }

            return ret;

        }

        if (j == cur_j)

            return ASN_INPUT_E;

    }

    /* If the output buffer has a room for an extra byte, add a null terminator */

    if (*outLen > i)

        out[i]= '\0';

    /* Note, *outLen won't reflect the optional terminating null. */

    *outLen = i;

    return 0;

}

#endif /* !BASE64_NO_TABLE */

int Base64_Decode(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    word32 i = 0;

    word32 j = 0;

    int ret;

    if ((in == NULL && inLen > 0) || out == NULL || outLen == NULL)

        return BAD_FUNC_ARG;

    while (inLen > 3) {

        int pad3 = 0;

        int pad4 = 0;

        byte b1, b2, b3;

        byte e1, e2, e3, e4;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            if (ret == WC_NO_ERR_TRACE(BUFFER_E)) {

                /* Running out of buffer here is not an error */

                break;

            }

            return ret;

        }

        e1 = in[j++];

        if (e1 == '\0') {

            inLen = 0;

            break;

        }

        inLen--;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            return ret;

        }

        e2 = in[j++];

        inLen--;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            return ret;

        }

        e3 = in[j++];

        inLen--;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            return ret;

        }

        e4 = in[j++];

        inLen--;

        if (e3 == PAD)

            pad3 = 1;

        if (e4 == PAD)

            pad4 = 1;

        if (pad3 && !pad4)

            return ASN_INPUT_E;

        e1 = Base64_Char2Val_CT(e1);

        e2 = Base64_Char2Val_CT(e2);

        e3 = (byte)((e3 == PAD) ? 0 : Base64_Char2Val_CT(e3));

        e4 = (byte)((e4 == PAD) ? 0 : Base64_Char2Val_CT(e4));

        if (e1 == BAD || e2 == BAD || e3 == BAD || e4 == BAD) {

            WOLFSSL_MSG("Bad Base64 Decode bad character");

            return ASN_INPUT_E;

        }

        /* Output space check needs to follow input character validation to

         * assure ASN_INPUT_E is returned on truncated input with the

         * terminating null included in the input buffer.

         */

        if (i + 1 + !pad3 + !pad4 > *outLen) {

            WOLFSSL_MSG("Bad Base64 Decode out buffer, too small");

            return BUFFER_E;

        }

        b1 = (byte)((e1 << 2) | (e2 >> 4));

        b2 = (byte)(((e2 & 0xF) << 4) | (e3 >> 2));

        b3 = (byte)(((e3 & 0x3) << 6) | e4);

        out[i++] = b1;

        if (!pad3)

            out[i++] = b2;

        if (!pad4)

            out[i++] = b3;

        else

            break;

    }

    /* If there is still input available, and it's not whitespace or nulls, then

     * the input is invalid.

     */

    while (inLen > 0) {

        word32 cur_j = j;

        if (in[j] == 0)

            break;

        if ((ret = Base64_SkipNewline(in, &inLen, &j)) != 0) {

            if (ret == WC_NO_ERR_TRACE(BUFFER_E)) {

                /* Running out of buffer here is not an error */

                break;

            }

            return ret;

        }

        if (j == cur_j)

            return ASN_INPUT_E;

    }

    /* If the output buffer has a room for an extra byte, add a null terminator */

    if (*outLen > i)

        out[i]= '\0';

    /* Note, *outLen won't reflect the optional terminating null. */

    *outLen = i;

    return 0;

}

#ifdef BASE64_NO_TABLE

int Base64_Decode_nonCT(const byte* in, word32 inLen, byte* out, word32* outLen) {

    return Base64_Decode(in, inLen, out, outLen);

}

#endif /* BASE64_NO_TABLE */

#endif /* WOLFSSL_BASE64_DECODE */

#if defined(WOLFSSL_BASE64_ENCODE)

static

const byte base64Encode[] = { 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J',

                              'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T',

                              'U', 'V', 'W', 'X', 'Y', 'Z',

                              'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j',

                              'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't',

                              'u', 'v', 'w', 'x', 'y', 'z',

                              '0', '1', '2', '3', '4', '5', '6', '7', '8', '9',

                              '+', '/'

                            };

/* make sure *i (idx) won't exceed max, store and possibly escape to out,

 * raw means use e w/o decode,  0 on success */

static int CEscape(int escaped, byte e, byte* out, word32* i, word32 maxSz,

                  int raw, int getSzOnly)

{

    int    doEscape = 0;

    word32 needed = 1;

    word32 idx = *i;

    byte basic;

    byte plus    = 0;

    byte equals  = 0;

    byte newline = 0;

    if (raw)

        basic = e;

    else if (e < sizeof(base64Encode))

        basic = base64Encode[e];

    else

        return BAD_FUNC_ARG;

    /* check whether to escape. Only escape for EncodeEsc */

    if (escaped == WC_ESC_NL_ENC) {

        switch ((char)basic) {

            case '+' :

                plus     = 1;

                doEscape = 1;

                needed  += 2;

                break;

            case '=' :

                equals   = 1;

                doEscape = 1;

                needed  += 2;

                break;

            case '\n' :

                newline  = 1;

                doEscape = 1;

                needed  += 2;

                break;

            default:

                /* do nothing */

                break;

        }

    }

    /* check size */

    if ( (idx+needed) > maxSz && !getSzOnly) {

        WOLFSSL_MSG("Escape buffer max too small");

        return BUFFER_E;

    }

    /* store it */

    if (doEscape == 0) {

        if(getSzOnly)

            idx++;

        else

            out[idx++] = basic;

    }

    else {

        if(getSzOnly)

            idx+=3;

        else {

            out[idx++] = '%';  /* start escape */

            if (plus) {

                out[idx++] = '2';

                out[idx++] = 'B';

            }

            else if (equals) {

                out[idx++] = '3';

                out[idx++] = 'D';

            }

            else if (newline) {

                out[idx++] = '0';

                out[idx++] = 'A';

            }

        }

    }

    *i = idx;

    return 0;

}

/* internal worker, handles both escaped and normal line endings.

   If out buffer is NULL, will return sz needed in outLen */

static int DoBase64_Encode(const byte* in, word32 inLen, byte* out,

                           word32* outLen, int escaped)

{

    int    ret = 0;

    word32 i = 0,

           j = 0,

           n = 0;   /* new line counter */

    int    getSzOnly = (out == NULL);

    word32 outSz;

    word32 addSz;

    if (in == NULL && inLen > 0)

        return BAD_FUNC_ARG;

    /* Reject lengths that would wrap the encoded-size calculation below. */

    if (inLen >= (WOLFSSL_MAX_32BIT / 4))

        return BAD_FUNC_ARG;

    outSz = (inLen + 3 - 1) / 3 * 4;

    addSz = (outSz + BASE64_LINE_SZ - 1) / BASE64_LINE_SZ;  /* new lines */

    if (escaped == WC_ESC_NL_ENC)

        addSz *= 3;   /* instead of just \n, we're doing %0A triplet */

    else if (escaped == WC_NO_NL_ENC)

        addSz = 0;    /* encode without \n */

    outSz += addSz;

    /* if escaped we can't predetermine size for one pass encoding, but

     * make sure we have enough if no escapes are in input

     * Also need to ensure outLen valid before dereference */

    if (!outLen || (outSz > *outLen && !getSzOnly)) return BAD_FUNC_ARG;

    while (inLen > 2) {

        byte b1 = in[j++];

        byte b2 = in[j++];

        byte b3 = in[j++];

        /* encoded idx */

        byte e1 = b1 >> 2;

        byte e2 = (byte)(((b1 & 0x3) << 4) | (b2 >> 4));

        byte e3 = (byte)(((b2 & 0xF) << 2) | (b3 >> 6));

        byte e4 = b3 & 0x3F;

        /* store */

        ret = CEscape(escaped, e1, out, &i, *outLen, 0, getSzOnly);

        if (ret != 0) break;

        ret = CEscape(escaped, e2, out, &i, *outLen, 0, getSzOnly);

        if (ret != 0) break;

        ret = CEscape(escaped, e3, out, &i, *outLen, 0, getSzOnly);

        if (ret != 0) break;

        ret = CEscape(escaped, e4, out, &i, *outLen, 0, getSzOnly);

        if (ret != 0) break;

        inLen -= 3;

        /* Insert newline after BASE64_LINE_SZ, unless no \n requested */

        if (escaped != WC_NO_NL_ENC && (++n % (BASE64_LINE_SZ/4)) == 0 && inLen) {

            ret = CEscape(escaped, '\n', out, &i, *outLen, 1, getSzOnly);

            if (ret != 0) break;

        }

    }

    /* last integral */

    if (inLen && ret == 0) {

        int twoBytes = (inLen == 2);

        byte b1 = in[j++];

        byte b2 = (twoBytes) ? in[j++] : 0;

        byte e1 = b1 >> 2;

        byte e2 = (byte)(((b1 & 0x3) << 4) | (b2 >> 4));

        byte e3 = (byte)((b2 & 0xF) << 2);

        ret = CEscape(escaped, e1, out, &i, *outLen, 0, getSzOnly);

        if (ret == 0)

            ret = CEscape(escaped, e2, out, &i, *outLen, 0, getSzOnly);

        if (ret == 0) {

            /* third */

            if (twoBytes)

                ret = CEscape(escaped, e3, out, &i, *outLen, 0, getSzOnly);

            else

                ret = CEscape(escaped, '=', out, &i, *outLen, 1, getSzOnly);

        }

        /* fourth always pad */

        if (ret == 0)

            ret = CEscape(escaped, '=', out, &i, *outLen, 1, getSzOnly);

    }

    if (ret == 0 && escaped != WC_NO_NL_ENC)

        ret = CEscape(escaped, '\n', out, &i, *outLen, 1, getSzOnly);

    if (i != outSz && escaped != 1 && ret == 0)

        return ASN_INPUT_E;

/* If the output buffer has a room for an extra byte, add a null terminator */

    if (out && *outLen > i)

        out[i]= '\0';

    *outLen = i;

    if (ret == 0)

        return getSzOnly ? WC_NO_ERR_TRACE(LENGTH_ONLY_E) : 0;

    return ret;

}

/* Base64 Encode, PEM style, with \n line endings */

int Base64_Encode(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    return DoBase64_Encode(in, inLen, out, outLen, WC_STD_ENC);

}

/* Base64 Encode, with %0A escaped line endings instead of \n */

int Base64_EncodeEsc(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    return DoBase64_Encode(in, inLen, out, outLen, WC_ESC_NL_ENC);

}

int Base64_Encode_NoNl(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    return DoBase64_Encode(in, inLen, out, outLen, WC_NO_NL_ENC);

}

#endif /* WOLFSSL_BASE64_ENCODE */

#ifdef WOLFSSL_BASE16

static

const ALIGN64 byte hexDecode[] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,

                           BAD, BAD, BAD, BAD, BAD, BAD, BAD,

                           10, 11, 12, 13, 14, 15,  /* upper case A-F */

                           BAD, BAD, BAD, BAD, BAD, BAD, BAD, BAD,

                           BAD, BAD, BAD, BAD, BAD, BAD, BAD, BAD,

                           BAD, BAD, BAD, BAD, BAD, BAD, BAD, BAD,

                           BAD, BAD,  /* G - ` */

                           10, 11, 12, 13, 14, 15   /* lower case a-f */

                         };  /* A starts at 0x41 not 0x3A */

int Base16_Decode(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    word32 inIdx  = 0;

    word32 outIdx = 0;

    if (in == NULL || out == NULL || outLen == NULL)

        return BAD_FUNC_ARG;

    if (inLen == 1 && *outLen) {

        byte b = (byte)(in[inIdx++] - BASE16_MIN);  /* 0 starts at 0x30 */

        /* sanity check */

        if (b >=  sizeof(hexDecode)/sizeof(hexDecode[0]))

            return ASN_INPUT_E;

        b  = hexDecode[b];

        if (b == BAD)

            return ASN_INPUT_E;

        out[outIdx++] = b;

        *outLen = outIdx;

        return 0;

    }

    if (inLen % 2)

        return BAD_FUNC_ARG;

    if (*outLen < (inLen / 2))

        return BUFFER_E;

    while (inLen) {

        byte b  = (byte)(in[inIdx++] - BASE16_MIN);  /* 0 starts at 0x30 */

        byte b2 = (byte)(in[inIdx++] - BASE16_MIN);

        /* sanity checks */

        if (b >=  sizeof(hexDecode)/sizeof(hexDecode[0]))

            return ASN_INPUT_E;

        if (b2 >= sizeof(hexDecode)/sizeof(hexDecode[0]))

            return ASN_INPUT_E;

        b  = hexDecode[b];

        b2 = hexDecode[b2];

        if (b == BAD || b2 == BAD)

            return ASN_INPUT_E;

        out[outIdx++] = (byte)((b << 4) | b2);

        inLen -= 2;

    }

    *outLen = outIdx;

    return 0;

}

static

const ALIGN64 byte hexEncode[] = { '0', '1', '2', '3', '4', '5', '6', '7',

                                   '8', '9', 'A', 'B', 'C', 'D', 'E', 'F'

};

int Base16_Encode(const byte* in, word32 inLen, byte* out, word32* outLen)

{

    word32 outIdx = 0;

    word32 i;

    if (in == NULL || out == NULL || outLen == NULL)

        return BAD_FUNC_ARG;

    if (inLen > (WOLFSSL_MAX_32BIT / 2))

        return BAD_FUNC_ARG;

    if (*outLen < (2 * inLen))

        return BAD_FUNC_ARG;

    for (i = 0; i < inLen; i++) {

        byte hb = in[i] >> 4;

        byte lb = in[i] & 0x0f;

        hb = hexEncode[hb];

        lb = hexEncode[lb];

        out[outIdx++] = hb;

        out[outIdx++] = lb;

    }

    /* If the output buffer has a room for an extra byte, add a null terminator */

    if (*outLen > outIdx)

        out[outIdx++]= '\0';

    *outLen = outIdx;

    return 0;

}

#endif /* WOLFSSL_BASE16 */

#endif /* !NO_CODING */

#ifdef WOLFSSL_UTF8_DECODE

/**

 * Decode the UTF-8 encoding of one code point.

 *

 * The encodings that RFC 3629 Sec. 3 requires a decoder to reject are

 * rejected: overlong forms, the code points reserved for UTF-16 surrogates,

 * code points past the end of the Unicode range, the five and six octet forms

 * that RFC 3629 removed, and sequences whose continuation octets are missing

 * or malformed. Any of these decoding to a character would let one octet

 * sequence impersonate another.

 *

 * *inOutIdx only moves when a code point is decoded, so a caller that wants to

 * keep going after a bad sequence is free to choose how far to skip.

 *

 * @param [in]      in        Buffer holding UTF-8 encoded text.

 * @param [in]      inLen     Length of buffer in octets.

 * @param [in, out] inOutIdx  On in, index of the first octet to decode.

 *                            On out, index of the first octet after the code

 *                            point decoded.

 * @param [out]     cp        Code point decoded.

 * @return  0 on success.

 * @return  BAD_FUNC_ARG when in, inOutIdx or cp is NULL.

 * @return  BUFFER_E when no octets remain, or when the sequence needs more

 *          continuation octets than the buffer holds.

 * @return  ASN_INPUT_E when the octets are not a valid encoding of a code

 *          point.

 */

int wc_Utf8_DecodeChar(const byte* in, word32 inLen, word32* inOutIdx,

                       word32* cp)

{

    word32 idx;

    word32 c;

    word32 minCp;

    word32 need;

    word32 i;

    if ((in == NULL) || (inOutIdx == NULL) || (cp == NULL)) {

        return BAD_FUNC_ARG;

    }

    idx = *inOutIdx;

    if (idx >= inLen) {

        return BUFFER_E;

    }

    c = in[idx];

    if (c < 0x80U) {

        /* 0xxxxxxx: the octet is the code point. */

        *inOutIdx = idx + 1U;

        *cp = c;

        return 0;

    }

    /* Take the code point bits out of the lead octet and note how many

     * continuation octets follow and the smallest code point that the form is

     * allowed to carry. */

    if ((c & 0xE0U) == 0xC0U) {

        need = 1U;

        minCp = 0x80U;

        c &= 0x1FU;

    }

    else if ((c & 0xF0U) == 0xE0U) {

        need = 2U;

        minCp = 0x800U;

        c &= 0x0FU;

    }

    else if ((c & 0xF8U) == 0xF0U) {

        need = 3U;

        minCp = 0x10000U;

        c &= 0x07U;

    }

    else {

        /* A continuation octet with no lead, or a lead octet of one of the

         * longer forms that are no longer part of UTF-8. */

        return ASN_INPUT_E;

    }

    /* Every continuation octet must be in the buffer. */

    if ((inLen - idx) <= need) {

        return BUFFER_E;

    }

    for (i = 1; i <= need; i++) {

        /* 10xxxxxx. */

        if ((in[idx + i] & 0xC0U) != 0x80U) {

            return ASN_INPUT_E;

        }

        c = (c << 6U) | (word32)(in[idx + i] & 0x3FU);

    }

    if ((c < minCp) || (c > WC_UNICODE_MAX_CODEPOINT) ||

            ((c >= WC_UTF16_HI_SURROGATE_MIN) &&

             (c <= WC_UTF16_LO_SURROGATE_MAX))) {

        return ASN_INPUT_E;

    }

    *inOutIdx = idx + need + 1U;

    *cp = c;

    return 0;

}

#endif /* WOLFSSL_UTF8_DECODE */

/* An empty translation unit is a constraint violation in C89, so emit a

 * harmless typedef to keep it well-formed in case everything above is

 * compiled out. */

typedef int wolfssl_coding_dummy_decl;