Extracting bit fields

mzslater

I need to decode a message that is coming in from a serial port (from an AIS receiver that indicates ship positions) in a packed binary format. It is 424 bits long, encoded as a series of 6-bit chunks, each represented by an ASCII character. Within these 424 bits are fields of various numbers of bits, from 1 to 42. There's no padding, so field boundaries typically don't line up on the 6-bit character boundaries.

To restate it generally, the problem is to extract arbitrarily positioned bit fields from a data stream several hundred bits long.

I'd appreciate any suggestions about an efficient way to tackle this in PHP.

Michael
www.BoatingSF.com

Weedpacket

[man]unpack[/man]might be useful.

Hmm: "...a series of 6-bit chunks, each represented by an ASCII character....". It's not base64 encoding is it? If it is, base64_decode will get rid of the padding and leave you with 53 8-bit bytes (as a string).

mzslater

Converting the 6-bit characters into binary is easy (it uses a variant of ASCII with a simple algorithm to convert). The part I'm struggling with is extracting all the odd-length bit fields from a long series of 6-bit values. For example, there's a 40-bit field that consists of 2 bits from one 6-bit "character", 6 bits each from the next 6 characters, and 2 bits from the last character. I have a clumsy approach (bitwise AND to get the needed bits from one character, shift and add) that works but it seems there must be something more elegant.

Ideally, I'd want some library function that lets me create a binary number several hundred bits long, and that then has extract functions where I can specify the start and end bit position for a field.

Michael

Weedpacket

Okay, here's something I whacked together. I'm not thrilled by it - the thought of using a whole byte to store one bit is mildly offensive - but it can be thought of as an intermediate stage.

function bitsequence_to_string($word)
{
	$array = unpack("H*hex", $word);
	$hex = $array['hex'];
	$nibbles = array(
		'0' => '0000', '1' => '0001', '2' => '0010', '3' => '0011',
		'4' => '0100', '5' => '0101', '6' => '0110', '7' => '0111',
		'8' => '1000', '9' => '1001', 'a' => '1010', 'b' => '1011',
		'c' => '1100', 'd' => '1101', 'e' => '1110', 'f' => '1111');
	$binary = strtr($hex, $nibbles);
	return $binary;
}

Now the string functions can be used to chop that up and extract chunks and compare bits and whatever else. If you want to convert a string of bit characters into a number, [man]bindec[/man] will make the conversion (but watch out for overflow!).