Using ints w/decimal place tracking for handling currency in PHP

Question

I have an issue, heh. I'm doing some calcs with bitcoins in PHP. When I call to the API, I get back JSON with bitcoin values. They are strings with a precision of 1millionth of a decimal.

I was tracking them as floats in my app, but that's leading to many precision issues and loss.

So, I decided to translate each string to an integer and store the decimal "power". I perform an operation, say subtraction, and then return the result to their original power.

My code (I know it looks like crap, been messing with this all night):

$val1 = 12.01;
$val2 = 11.01;

translateBTC($val1, $val1T, $val1Pow);
translateBTC($val2, $val2T, $val2Pow);
echo "Subtracing $val1 - $val2\n";
echo "Val1 = $val1T, power = $val1Pow\n";
echo "Val2 = $val2T, power = $val2Pow\n";

subtractBTC($val1T, $val1Pow, $val2T, $val2Pow, $res, $resP);
echo "Result = $res\n";

function translateBTC($btc, &$val, &$pow)
{
    $v1Str  = (string) $btc;
    $v1DPos = strpos($v1Str, '.');

    $v1Int = '';
    $v1Last = 0;
    for($i=0;$i<strlen($v1Str);$i++)
    {
        if ($v1Str[$i] != '.')
        {
            $v1Int .= $v1Str[$i];
        }

        if ($seen)
        {
            if ($v1Str[$i] != 0)
                $v1Last = $i;                
        }

        if ($v1Str[$i] == '.')
        {
            // Were now tracking past the ., so pow.
            $seen = TRUE;
        }
    }

    $val = (int) $v1Int;
    $pow = $v1Last - $v1DPos;
}

function subtractBTC($val1, $val1P, $val2, $val2P, &$result, &$rPow)
{
    $highP = ($val1P > $val2P)? $val1P : $val2P;
    $lowP  = ($val1P < $val2P)? $val1P : $val2P;
    $pStr  = str_pad('', ($highP - $lowP), '0');
    $val2 .= ($val1 > $val2)? $pStr : '';
    $val1 .= ($val2 > $val1)? $pStr : '';

    $val1 = (int)$val1;
    $val2 = (int)$val2;

    echo "val1($val1) - val2($val2)\n";

    $tmpResult = $val1 - $val2;
    $rPow   = $highP;

    $tmpResult = (string) $tmpResult;
    echo "tmpResult = ($tmpResult), rPow = $rPow\n";
    ...

At this point, I'm struggling with converting the result back to the proper base, for all the different values...I've got to be going about this all wrong - is there a currency type I can use for this sort of precision?

Here's the debug output:

Val1 = 1201, power = 2
Val2 = 1101, power = 2
val1(1201) - val2(1101)
tmpResult = (100), rPow = 2
001Result = 10.0

TIA SO!

Answer 1

Look into the BC Math extension which lets you operate on arbitrary precision numbers stored as strings.