base32.c

/**
 * base32 (de)coder implementation as specified by RFC4648.
 *
 * Copyright (c) 2010 Adrien Kunysz
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 **/

#include <assert.h>  // assert()
#include <limits.h>  // CHAR_BIT

#include "base32.h"

/**
 * Let this be a sequence of plain data before encoding:
 *
 *  01234567 01234567 01234567 01234567 01234567
 * +--------+--------+--------+--------+--------+
 * |< 0 >< 1| >< 2 ><|.3 >< 4.|>< 5 ><.|6 >< 7 >|
 * +--------+--------+--------+--------+--------+
 *
 * There are 5 octets of 8 bits each in each sequence.
 * There are 8 blocks of 5 bits each in each sequence.
 *
 * You probably want to refer to that graph when reading the algorithms in this
 * file. We use "octet" instead of "byte" intentionnaly as we really work with
 * 8 bits quantities. This implementation will probably not work properly on
 * systems that don't have exactly 8 bits per (unsigned) char.
 **/

static size_t min(size_t x, size_t y)
{
	return x < y ? x : y;
}

static const unsigned char PADDING_CHAR = '=';

/**
 * Pad the given buffer with len padding characters.
 */
static void pad(unsigned char *buf, int len)
{
	for (int i = 0; i < len; i++)
		buf[i] = PADDING_CHAR;
}

/**
 * This convert a 5 bits value into a base32 character.
 * Only the 5 least significant bits are used.
 */
static unsigned char encode_char(unsigned char c)
{
	static unsigned char base32[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567";
	return base32[c & 0x1F];  // 0001 1111
}

/**
 * Decode given character into a 5 bits value. 
 * Returns -1 iff the argument given was an invalid base32 character
 * or a padding character.
 */
static int decode_char(unsigned char c)
{
	char retval = -1;

	if (c >= 'A' && c <= 'Z')
		retval = c - 'A';
	if (c >= '2' && c <= '7')
		retval = c - '2' + 26;

	assert(retval == -1 || ((retval & 0x1F) == retval));

	return  retval;
}

/**
 * Given a block id between 0 and 7 inclusive, this will return the index of
 * the octet in which this block starts. For example, given 3 it will return 1
 * because block 3 starts in octet 1:
 *
 * +--------+--------+
 * | ......<|.3 >....|
 * +--------+--------+
 *  octet 1 | octet 2
 */
static int get_octet(int block)
{
	assert(block >= 0 && block < 8);
	return (block*5) / 8;
}

/**
 * Given a block id between 0 and 7 inclusive, this will return how many bits
 * we can drop at the end of the octet in which this block starts. 
 * For example, given block 0 it will return 3 because there are 3 bits
 * we don't care about at the end:
 *
 *  +--------+-
 *  |< 0 >...|
 *  +--------+-
 *
 * Given block 1, it will return -2 because there
 * are actually two bits missing to have a complete block:
 *
 *  +--------+-
 *  |.....< 1|..
 *  +--------+-
 **/
static int get_offset(int block)
{
	assert(block >= 0 && block < 8);
	return (8 - 5 - (5*block) % 8);
}

/**
 * Like "b >> offset" but it will do the right thing with negative offset.
 * We need this as bitwise shifting by a negative offset is undefined
 * behavior.
 */
static unsigned char shift_right(unsigned char byte, signed char offset)
{
	if (offset > 0)
		return byte >>  offset;
	else
		return byte << -offset;
}

static unsigned char shift_left(unsigned char byte, signed char offset)
{
	return shift_right(byte, - offset);
}

/**
 * Encode a sequence. A sequence is no longer than 5 octets by definition.
 * Thus passing a length greater than 5 to this function is an error. Encoding
 * sequences shorter than 5 octets is supported and padding will be added to the
 * output as per the specification.
 */
static void encode_sequence(const unsigned char *plain, int len, unsigned char *coded)
{
	assert(CHAR_BIT == 8);  // not sure this would work otherwise
	assert(len >= 0 && len <= 5);

	for (int block = 0; block < 8; block++) {
		int octet = get_octet(block);  // figure out which octet this block starts in
		int junk = get_offset(block);  // how many bits do we drop from this octet?

		if (octet >= len) { // we hit the end of the buffer
			pad(&coded[block], 8 - block);
			return;
		}

		unsigned char c = shift_right(plain[octet], junk);  // first part

		if (junk < 0  // is there a second part?
		&&  octet < len - 1)  // is there still something to read?
		{
			c |= shift_right(plain[octet+1], 8 + junk);
		}
		coded[block] = encode_char(c);
	}
}

void base32_encode(const unsigned char *plain, size_t len, unsigned char *coded)
{
	// All the hard work is done in encode_sequence(),
	// here we just need to feed it the data sequence by sequence.
	for (size_t i = 0, j = 0; i < len; i += 5, j += 8) {
		encode_sequence(&plain[i], min(len - i, 5), &coded[j]);
	}
}

static int decode_sequence(const unsigned char *coded, unsigned char *plain)
{
	assert(CHAR_BIT == 8);
	assert(coded && plain);

	plain[0] = 0;
	for (int block = 0; block < 8; block++) {
		int offset = get_offset(block);
		int octet = get_octet(block);

		int c = decode_char(coded[block]);
		if (c < 0)  // invalid char, stop here
			return octet;

		plain[octet] |= shift_left(c, offset);
		if (offset < 0) {  // does this block overflows to next octet?
			assert(octet < 4);
			plain[octet+1] = shift_left(c, 8 + offset);
		}
	}
	return 5;
}

size_t base32_decode(const unsigned char *coded, unsigned char *plain)
{
	size_t written = 0;
	for (size_t i = 0, j = 0; ; i += 8, j += 5) {
		int n = decode_sequence(&coded[i], &plain[j]);
		written += n;
		if (n < 5)
			return written;
	}
}