/**
* Encode the given string using Huffman-encoding.
*
* @param string $input Input string containing only symbols that are mapped to codes.
* @param int $bits Passed by reference, will be populated with the number of bits used to encode the input string.
* @return string Huffman-encoded string.
*/
public function encode(string $input, int &$bits = NULL) : string
{
if ($this->symbols === NULL) {
$this->symbols = $this->code->getEncoderData();
}
$encoded = '';
$offset = 7;
$byte = 0;
for ($inputLen = strlen($input), $i = 0; $i < $inputLen; $i++) {
list($code, $len) = $this->symbols[$input[$i]];
for ($j = $len - 1; $j >= 0; $j--) {
// Get next code bit and shift it into the encoded byte at the next bit offset.
$byte |= ($code >> $j & 1) << $offset--;
$bits++;
if ($offset == -1) {
$offset = 7;
$encoded .= chr($byte);
$byte = 0;
}
}
}
// Pad encoded string with "1" bits.
if ($offset < 7) {
do {
$byte |= 1 << $offset--;
} while ($offset >= 0);
$encoded .= chr($byte);
}
return $encoded;
}
protected function decodeCanonical(string $input, int $bits) : string
{
if ($this->symbol === NULL) {
list($this->symbol, $this->first) = $this->code->getCanonicalDecoderData();
}
$decoded = '';
$buffer = NULL;
$len = strlen($input);
$byteOffset = 0;
$bitOffset = 7;
while (true) {
$code = 0;
$codeLen = 0;
do {
if ($bits-- == 0) {
break;
}
if ($buffer === NULL) {
if ($byteOffset == $len) {
if ($code === 0 || $this->isHuffmanPaddingCode($code)) {
return $decoded;
}
throw new \RuntimeException('Cannot read beyond end of Huffman-encoded string');
}
$buffer = ord($input[$byteOffset++]);
}
// Read next bit and and append it as LSB (least significant bit) to the code.
$code = $code << 1 | $buffer >> $bitOffset-- & 1;
if ($bitOffset == -1) {
$bitOffset = 7;
$buffer = NULL;
}
} while ($code > $this->first[++$codeLen]);
if ($bits == 0 && $codeLen == 0) {
return $decoded;
}
$char = $this->symbol[$codeLen][$this->first[$codeLen] - $code] ?? NULL;
if ($char !== NULL) {
$decoded .= $char;
if ($bits == 0) {
return $decoded;
}
continue;
}
if ($this->isHuffmanPaddingCode($code)) {
return $decoded;
}
break;
}
throw new \RuntimeException('Invalid Huffman code detected');
}
