codebox/base64/codec.c

#include <stdint.h>
#include <stddef.h>
#include <stdlib.h>

#include "libbase64.h"
#include "codecs.h"

BASE64_ENC_FUNCTION(plain)
{
	// Assume that *out is large enough to contain the output.
	// Theoretically it should be 4/3 the length of src.
	const uint8_t *c = (const uint8_t *)src;
	uint8_t *o = (uint8_t *)out;

	// Use local temporaries to avoid cache thrashing:
	size_t outl = 0;
	struct base64_state st;
	st.bytes = state->bytes;
	st.carry = state->carry;

	// Turn three bytes into four 6-bit numbers:
	// in[0] = 00111111
	// in[1] = 00112222
	// in[2] = 00222233
	// in[3] = 00333333

	// Duff's device, a for() loop inside a switch() statement. Legal!
	switch (st.bytes)
	{
		for (;;)
		{
		case 0:

			// If we have 64-bit ints, pick off 6 bytes at a time for as long as we can,
			// but ensure that there are at least 8 bytes available to avoid segfaulting:
			while (srclen >= 8)
			{
				// Load string:
				uint64_t str = *(uint64_t *)c;

				// Reorder to 64-bit big-endian, if not already in that format. The
				// workset must be in big-endian, otherwise the shifted bits do not
				// carry over properly among adjacent bytes:
				str = cpu_to_be64(str);

				// Shift input by 6 bytes each round and mask in only the lower 6 bits;
				// look up the character in the Base64 encoding table and write it to
				// the output location:
				*o++ = base64_table_enc[(str >> 58) & 0x3F];
				*o++ = base64_table_enc[(str >> 52) & 0x3F];
				*o++ = base64_table_enc[(str >> 46) & 0x3F];
				*o++ = base64_table_enc[(str >> 40) & 0x3F];
				*o++ = base64_table_enc[(str >> 34) & 0x3F];
				*o++ = base64_table_enc[(str >> 28) & 0x3F];
				*o++ = base64_table_enc[(str >> 22) & 0x3F];
				*o++ = base64_table_enc[(str >> 16) & 0x3F];

				c += 6;		// 6 bytes of input
				outl += 8;	// 8 bytes of output
				srclen -= 6;
			}

			if (srclen-- == 0) {
				break;
			}
			*o++ = base64_table_enc[*c >> 2];
			st.carry = (*c++ << 4) & 0x30;
			st.bytes++;
			outl += 1;

			// Deliberate fallthrough:
			BASE64_FALLTHROUGH

		case 1:	if (srclen-- == 0) {
				break;
			}
			*o++ = base64_table_enc[st.carry | (*c >> 4)];
			st.carry = (*c++ << 2) & 0x3C;
			st.bytes++;
			outl += 1;

			// Deliberate fallthrough:
			BASE64_FALLTHROUGH

		case 2:	if (srclen-- == 0) {
				break;
			}
			*o++ = base64_table_enc[st.carry | (*c >> 6)];
			*o++ = base64_table_enc[*c++ & 0x3F];
			st.bytes = 0;
			outl += 2;
		}
	}
	state->bytes = st.bytes;
	state->carry = st.carry;
	*outlen = outl;
}

BASE64_DEC_FUNCTION(plain)
{
	int ret = 0;
	const uint8_t *c = (const uint8_t *)src;
	uint8_t *o = (uint8_t *)out;
	uint8_t q;

	// Use local temporaries to avoid cache thrashing:
	size_t outl = 0;
	struct base64_state st;
	st.eof = state->eof;
	st.bytes = state->bytes;
	st.carry = state->carry;

	// If we previously saw an EOF or an invalid character, bail out:
	if (st.eof) {
		*outlen = 0;
		ret = 0;
		// If there was a trailing '=' to check, check it:
		if (srclen && (st.eof == BASE64_AEOF)) {
			state->bytes = 0;
			state->eof = BASE64_EOF;
			ret = ((base64_table_dec[*c++] == 254) && (srclen == 1)) ? 1 : 0;
		}
		return ret;
	}

	// Turn four 6-bit numbers into three bytes:
	// out[0] = 11111122
	// out[1] = 22223333
	// out[2] = 33444444

	// Duff's device again:
	switch (st.bytes)
	{
		for (;;)
		{
		case 0:
			if (srclen-- == 0) {
				ret = 1;
				break;
			}
			if ((q = base64_table_dec[*c++]) >= 254) {
				st.eof = BASE64_EOF;
				// Treat character '=' as invalid for byte 0:
				break;
			}
			st.carry = q << 2;
			st.bytes++;

			// Deliberate fallthrough:
			BASE64_FALLTHROUGH

		case 1:	if (srclen-- == 0) {
				ret = 1;
				break;
			}
			if ((q = base64_table_dec[*c++]) >= 254) {
				st.eof = BASE64_EOF;
				// Treat character '=' as invalid for byte 1:
				break;
			}
			*o++ = st.carry | (q >> 4);
			st.carry = q << 4;
			st.bytes++;
			outl++;

			// Deliberate fallthrough:
			BASE64_FALLTHROUGH

		case 2:	if (srclen-- == 0) {
				ret = 1;
				break;
			}
			if ((q = base64_table_dec[*c++]) >= 254) {
				st.bytes++;
				// When q == 254, the input char is '='.
				// Check if next byte is also '=':
				if (q == 254) {
					if (srclen-- != 0) {
						st.bytes = 0;
						// EOF:
						st.eof = BASE64_EOF;
						q = base64_table_dec[*c++];
						ret = ((q == 254) && (srclen == 0)) ? 1 : 0;
						break;
					}
					else {
						// Almost EOF
						st.eof = BASE64_AEOF;
						ret = 1;
						break;
					}
				}
				// If we get here, there was an error:
				break;
			}
			*o++ = st.carry | (q >> 2);
			st.carry = q << 6;
			st.bytes++;
			outl++;

			// Deliberate fallthrough:
			BASE64_FALLTHROUGH

		case 3:	if (srclen-- == 0) {
				ret = 1;
				break;
			}
			if ((q = base64_table_dec[*c++]) >= 254) {
				st.bytes = 0;
				st.eof = BASE64_EOF;
				// When q == 254, the input char is '='. Return 1 and EOF.
				// When q == 255, the input char is invalid. Return 0 and EOF.
				ret = ((q == 254) && (srclen == 0)) ? 1 : 0;
				break;
			}
			*o++ = st.carry | q;
			st.carry = 0;
			st.bytes = 0;
			outl++;
		}
	}
	state->eof = st.eof;
	state->bytes = st.bytes;
	state->carry = st.carry;
	*outlen = outl;
	return ret;
}