windows-server-2003/ds/security/cryptoapi/common/pkifmt/base64.cpp

//+--------------------------------------------------------------------------
//
// Microsoft Windows
// Copyright (C) Microsoft Corporation, 1996 - 1999
//
// File:        base64.cpp
//
// Contents:    base64 encode/decode implementation
//
// History:     25-Jul-96       vich created
//
//---------------------------------------------------------------------------

#include <windows.h>
#include <assert.h>
#include <dbgdef.h>
#include "pkifmt.h"

//#define BASE64_STRICT		// enforce syntax check on input data
#undef BASE64_STRICT		// enforce syntax check on input data


// The following table translates an ascii subset to 6 bit values as follows
// (see RFC 1421 and/or RFC 1521):
//
//  input    hex (decimal)
//  'A' --> 0x00 (0)
//  'B' --> 0x01 (1)
//  ...
//  'Z' --> 0x19 (25)
//  'a' --> 0x1a (26)
//  'b' --> 0x1b (27)
//  ...
//  'z' --> 0x33 (51)
//  '0' --> 0x34 (52)
//  ...
//  '9' --> 0x3d (61)
//  '+' --> 0x3e (62)
//  '/' --> 0x3f (63)
//
// Encoded lines must be no longer than 76 characters.
// The final "quantum" is handled as follows:  The translation output shall
// always consist of 4 characters.  'x', below, means a translated character,
// and '=' means an equal sign.  0, 1 or 2 equal signs padding out a four byte
// translation quantum means decoding the four bytes would result in 3, 2 or 1
// unencoded bytes, respectively.
//
//  unencoded size    encoded data
//  --------------    ------------
//     1 byte		"xx=="
//     2 bytes		"xxx="
//     3 bytes		"xxxx"

#define CB_BASE64LINEMAX	64	// others use 64 -- could be up to 76

// Any other (invalid) input character value translates to 0x40 (64)

const BYTE abDecode[256] =
{
    /* 00: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* 10: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* 20: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 62, 64, 64, 64, 63,
    /* 30: */ 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 64, 64, 64, 64, 64, 64,
    /* 40: */ 64,  0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14,
    /* 50: */ 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 64, 64, 64, 64, 64,
    /* 60: */ 64, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
    /* 70: */ 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 64, 64, 64, 64, 64,
    /* 80: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* 90: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* a0: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* b0: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* c0: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* d0: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* e0: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
    /* f0: */ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
};


const UCHAR abEncode[] =
    /*  0 thru 25: */ "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
    /* 26 thru 51: */ "abcdefghijklmnopqrstuvwxyz"
    /* 52 thru 61: */ "0123456789"
    /* 62 and 63: */  "+/";

#define MOD4(x) ((x) & 3)

__inline BOOL
_IsBase64WhiteSpace(
    IN TCHAR const ch)
{
    return(
	ch == TEXT(' ') ||
	ch == TEXT('\t') ||
	ch == TEXT('\r') ||
	ch == TEXT('\n'));
}


DWORD
Base64DecodeA(
    IN TCHAR const     *pchIn,
    IN DWORD           cchIn,
    OPTIONAL OUT BYTE *pbOut,
    IN OUT DWORD      *pcbOut)
{
    DWORD dwErr;
    DWORD cchInDecode, cbOutDecode;
    TCHAR const *pchInEnd;
    TCHAR const *pchInT;
    BYTE *pbOutT;

    // Count the translatable characters, skipping whitespace & CR-LF chars.

    cchInDecode = 0;
    pchInEnd = &pchIn[cchIn];
    dwErr = ERROR_INVALID_DATA;
    for (pchInT = pchIn; pchInT < pchInEnd; pchInT++)
    {
	if (sizeof(abDecode) < (unsigned) *pchInT || abDecode[*pchInT] > 63)
	{
	    // Found a non-base64 character.  Decide what to do.

	    DWORD cch;

	    if (_IsBase64WhiteSpace(*pchInT))
	    {
		continue;		// skip all whitespace
	    }

	    // The length calculation may stop in the middle of the last
	    // translation quantum, because the equal sign padding characters
	    // are treated as invalid input.  If the last translation quantum
	    // is not 4 bytes long, there must be 3, 2 or 1 equal sign(s).

	    if (0 != cchInDecode)
	    {
		cch = MOD4(cchInDecode);
		if (0 != cch)
		{
		    cch = 4 - cch;
		    while (0 != cch && pchInT < pchInEnd && '=' == *pchInT)
		    {
			pchInT++;
			cch--;
		    }
		}
//#ifdef BASE64_STRICT
		if (0 == cch)
//#endif
		{
		    break;
		}
	    }
#if DBG
            DbgPrintf(
                DBG_SS_TRACE,
                "Bad base64 data: \"%.*" szFMTTSTR "...\"\n",
                min(16, SAFE_SUBTRACT_POINTERS(pchInEnd, pchInT)),
                pchInT);
#endif //DBG
	    goto BadBase64Data;
	}
	cchInDecode++;			// only count valid base64 chars
    }
    assert(pchInT <= pchInEnd);
#ifdef BASE64_STRICT
    if (pchInT < pchInEnd)
    {
	TCHAR const *pch;
	DWORD cchEqual = 0;

	for (pch = pchInT; pch < pchInEnd; pch++)
	{
	    if (!_IsBase64WhiteSpace(*pch))
	    {
		// Allow up to 3 extra trailing equal signs.
		if (TEXT('=') == *pch && 3 > cchEqual)
		{
		    cchEqual++;
		    continue;
		}
#if DBG
		DbgPrintf(DBG_SS_TRACE,
		    "Bad trailing base64 data: \"%.*" szFMTTSTR "...\"\n",
		    min(16, SAFE_SUBTRACT_POINTERS(pchInEnd, pch)),
		    pch);
#endif //DBG
		goto BadTrailingBase64Data;
	    }
	}
#if DBG
	if (0 != cchEqual)
	{
	    DbgPrintf(DBG_SS_TRACE,
		"Ignored trailing base64 data: \"%.*" szFMTTSTR "\"\n",
		cchEqual,
		TEXT("==="));
	}
#endif //DBG
    }
#endif
    pchInEnd = pchInT;		// don't process any trailing stuff again

    // We know how many translatable characters are in the input buffer, so now
    // set the output buffer size to three bytes for every four (or fraction of
    // four) input bytes.  Compensate for a fractional translation quantum.

    cbOutDecode = ((cchInDecode + 3) >> 2) * 3;
    switch (cchInDecode % 4)
    {
	case 1:
	case 2:
	    cbOutDecode -= 2;
	    break;

	case 3:
	    cbOutDecode--;
	    break;
    }

    pbOutT = pbOut;

    if (NULL == pbOut)
    {
	pbOutT += cbOutDecode;
    }
    else
    {
	// Decode one quantum at a time: 4 bytes ==> 3 bytes

	//assert(cbOutDecode <= *pcbOut);
        if (cbOutDecode > *pcbOut)
        {
            *pcbOut = cbOutDecode;
            dwErr = ERROR_MORE_DATA;
            goto MoreDataError;
        }

	pchInT = pchIn;
	while (cchInDecode > 0)
	{
	    DWORD i;
	    BYTE ab4[4];

	    ZeroMemory(ab4, sizeof(ab4));
	    for (i = 0; i < min(sizeof(ab4)/sizeof(ab4[0]), cchInDecode); i++)
	    {
		while (
		    sizeof(abDecode) > (unsigned) *pchInT &&
		    63 < abDecode[*pchInT])
		{
		    pchInT++;
		}
		assert(pchInT < pchInEnd);
		ab4[i] = (BYTE) *pchInT++;
	    }

	    // Translate 4 input characters into 6 bits each, and deposit the
	    // resulting 24 bits into 3 output bytes by shifting as appropriate.

	    // out[0] = in[0]:in[1] 6:2
	    // out[1] = in[1]:in[2] 4:4
	    // out[2] = in[2]:in[3] 2:6

	    *pbOutT++ =
		(BYTE) ((abDecode[ab4[0]] << 2) | (abDecode[ab4[1]] >> 4));

	    if (i > 2)
	    {
		*pbOutT++ =
		  (BYTE) ((abDecode[ab4[1]] << 4) | (abDecode[ab4[2]] >> 2));
	    }
	    if (i > 3)
	    {
		*pbOutT++ = (BYTE) ((abDecode[ab4[2]] << 6) | abDecode[ab4[3]]);
	    }
	    cchInDecode -= i;
	}
	assert((DWORD) (pbOutT - pbOut) <= cbOutDecode);
    }
    *pcbOut = SAFE_SUBTRACT_POINTERS(pbOutT, pbOut);

    dwErr = ERROR_SUCCESS;
ErrorReturn:
    return dwErr;

SET_ERROR(MoreDataError, dwErr)
SET_ERROR(BadBase64Data, dwErr)
#ifdef BASE64_STRICT
SET_ERROR(BadTrailingBase64Data, dwErr)
#endif
}


// Encode a BYTE array into a Base64 text string.
// Use CR-LF pairs for line breaks, unless CRYPT_STRING_NOCR is set.
// Do not '\0' terminate the text string -- that's handled by the caller.
// Do not add -----BEGIN/END headers -- that's also handled by the caller.

DWORD
Base64EncodeA(
    IN BYTE const *pbIn,
    IN DWORD cbIn,
    IN DWORD Flags,
    OPTIONAL OUT TCHAR *pchOut,
    IN OUT DWORD *pcchOut)
{
    DWORD dwErr;
    TCHAR *pchOutT;
    DWORD cchOutEncode;
    BOOL fNoCR = 0 != (CRYPT_STRING_NOCR & Flags);

    // Allocate enough memory for full final translation quantum.

    cchOutEncode = ((cbIn + 2) / 3) * 4;

    // and enough for CR-LF pairs for every CB_BASE64LINEMAX character line.

    cchOutEncode +=
	(fNoCR? 1 : 2) *
	((cchOutEncode + CB_BASE64LINEMAX - 1) / CB_BASE64LINEMAX);

    pchOutT = pchOut;
    if (NULL == pchOut)
    {
	pchOutT += cchOutEncode;
	//printf("cchOut:       =%x Computed=%x\n", (DWORD) (pchOutT - pchOut), cchOutEncode);
    }
    else
    {
	DWORD cCol;

	if (cchOutEncode > *pcchOut)
	{
            *pcchOut = cchOutEncode;
	    dwErr = ERROR_MORE_DATA;
	    goto MoreDataError;
	}

	cCol = 0;
	while ((long) cbIn > 0)	// signed comparison -- cbIn can wrap
	{
	    BYTE ab3[3];

	    if (cCol == CB_BASE64LINEMAX/4)
	    {
		cCol = 0;
		if (!fNoCR)
		{
		    *pchOutT++ = '\r';
		}
		*pchOutT++ = '\n';
	    }
	    cCol++;
	    ZeroMemory(ab3, sizeof(ab3));

	    ab3[0] = *pbIn++;
	    if (cbIn > 1)
	    {
		ab3[1] = *pbIn++;
		if (cbIn > 2)
		{
		    ab3[2] = *pbIn++;
		}
	    }

	    *pchOutT++ = abEncode[ab3[0] >> 2];
	    *pchOutT++ = abEncode[((ab3[0] << 4) | (ab3[1] >> 4)) & 0x3f];
	    *pchOutT++ = (cbIn > 1)?
			abEncode[((ab3[1] << 2) | (ab3[2] >> 6)) & 0x3f] : '=';
	    *pchOutT++ = (cbIn > 2)? abEncode[ab3[2] & 0x3f] : '=';

	    cbIn -= 3;
	}

	// Append CR-LF only if there was input data

	if (pchOutT != pchOut)
	{
	    if (!fNoCR)
	    {
		*pchOutT++ = '\r';
	    }
	    *pchOutT++ = '\n';
	}
	//printf("cchOut: Actual=%x Computed=%x Buffer=%x\n", (DWORD) (pchOutT - pchOut), cchOutEncode, *pcchOut);
	assert((DWORD) (pchOutT - pchOut) == cchOutEncode);
    }
    *pcchOut = SAFE_SUBTRACT_POINTERS(pchOutT, pchOut);

    dwErr = ERROR_SUCCESS;
ErrorReturn:
    return dwErr;

SET_ERROR(MoreDataError, dwErr)
}


DWORD
Base64EncodeW(
    IN BYTE const *pbIn,
    IN DWORD cbIn,
    IN DWORD Flags,
    OUT WCHAR *wszOut,
    OUT DWORD *pcchOut)
{

    DWORD   cchOut;
    CHAR   *pch = NULL;
    DWORD   cch;
    DWORD   err;

    assert(pcchOut != NULL);

    // only want to know how much to allocate
    // we know all base64 char map 1-1 with unicode
    if (wszOut == NULL)
    {

        // get the number of characters
        *pcchOut = 0;
        err = Base64EncodeA(pbIn, cbIn, Flags, NULL, pcchOut);
    }

    // otherwise we have an output buffer
    else {

        // char count is the same be it ascii or unicode,
        cchOut = *pcchOut;
        cch = 0;
        err = ERROR_OUTOFMEMORY;
        pch = (CHAR *) malloc(cchOut);
        if (NULL != pch)
	{
            err = Base64EncodeA(pbIn, cbIn, Flags, pch, &cchOut);
	    if (ERROR_SUCCESS == err)
	    {
		// should not fail!
		cch = MultiByteToWideChar(0, 0, pch, cchOut, wszOut, *pcchOut);

		// check to make sure we did not fail
		assert(*pcchOut == 0 || cch != 0);
	    }
	}
    }

    if(pch != NULL)
        free(pch);

    return(err);
}


DWORD
Base64DecodeW(
    IN const WCHAR * wszIn,
    IN DWORD cch,
    OUT BYTE *pbOut,
    OUT DWORD *pcbOut)
{
    CHAR *pch;
    DWORD err = ERROR_SUCCESS;

    // in all cases we need to convert to an ascii string
    // we know the ascii string is less

    if ((pch = (CHAR *) malloc(cch)) == NULL)
    {
        err = ERROR_OUTOFMEMORY;
    }

    // we know no base64 wide char map to more than 1 ascii char
    else if (WideCharToMultiByte(0, 0, wszIn, cch, pch, cch, NULL, NULL) == 0)
    {
        err = ERROR_NO_DATA;
    }

    // get the length of the buffer
    else if (pbOut == NULL)
    {
        *pcbOut = 0;
        err = Base64Decode(pch, cch, NULL, pcbOut);
    }

    // otherwise fill in the buffer
    else {
        err = Base64Decode(pch, cch, pbOut, pcbOut);
    }

    if(pch != NULL)
        free(pch);

    return(err);
}