Convert JavaScript number to string of 64bit double precision floating point representation (IEEE 754)

gistfile1.js

// Convert a JavaScript number to IEEE-754 Double Precision
// value represented as an array of 8 bytes (octets)
//
// http://cautionsingularityahead.blogspot.com/2010/04/javascript-and-ieee754-redux.html

function toIEEE754(v, ebits, fbits) {

    var bias = (1 << (ebits - 1)) - 1;

    // Compute sign, exponent, fraction
    var s, e, f;
    if (isNaN(v)) {
        e = (1 << bias) - 1; f = 1; s = 0;
    }
    else if (v === Infinity || v === -Infinity) {
        e = (1 << bias) - 1; f = 0; s = (v < 0) ? 1 : 0;
    }
    else if (v === 0) {
        e = 0; f = 0; s = (1 / v === -Infinity) ? 1 : 0;
    }
    else {
        s = v < 0;
        v = Math.abs(v);

        if (v >= Math.pow(2, 1 - bias)) {
            var ln = Math.min(Math.floor(Math.log(v) / Math.LN2), bias);
            e = ln + bias;
            f = v * Math.pow(2, fbits - ln) - Math.pow(2, fbits);
        }
        else {
            e = 0;
            f = v / Math.pow(2, 1 - bias - fbits);
        }
    }
     
    // Pack sign, exponent, fraction
    var i, bits = [];
    for (i = fbits; i; i -= 1) { bits.push(f % 2 ? 1 : 0); f = Math.floor(f / 2); }
    for (i = ebits; i; i -= 1) { bits.push(e % 2 ? 1 : 0); e = Math.floor(e / 2); }
    bits.push(s ? 1 : 0);
    bits.reverse();
    var str = bits.join('');
     
    // Bits to bytes
    var bytes = [];
    while (str.length) {
        bytes.push(parseInt(str.substring(0, 8), 2));
        str = str.substring(8);
    }
    return bytes;
}

function fromIEEE754(bytes, ebits, fbits) {

    // Bytes to bits
    var bits = [];
    for (var i = bytes.length; i; i -= 1) {
        var byte = bytes[i - 1];
        for (var j = 8; j; j -= 1) {
            bits.push(byte % 2 ? 1 : 0); byte = byte >> 1;
        }
    }
    bits.reverse();
    var str = bits.join('');
   
    // Unpack sign, exponent, fraction
    var bias = (1 << (ebits - 1)) - 1;
    var s = parseInt(str.substring(0, 1), 2) ? -1 : 1;
    var e = parseInt(str.substring(1, 1 + ebits), 2);
    var f = parseInt(str.substring(1 + ebits), 2);
     
    // Produce number
    if (e === (1 << ebits) - 1) {
        return f !== 0 ? NaN : s * Infinity;
    }
    else if (e > 0) {
        return s * Math.pow(2, e - bias) * (1 + f / Math.pow(2, fbits));
    }
    else if (f !== 0) {
        return s * Math.pow(2, -(bias-1)) * (f / Math.pow(2, fbits));
    }
    else {
        return s * 0;
    }
}

function fromIEEE754Double(b) { return fromIEEE754(b, 11, 52); }
function   toIEEE754Double(v) { return   toIEEE754(v, 11, 52); }
function fromIEEE754Single(b) { return fromIEEE754(b,  8, 23); }
function   toIEEE754Single(v) { return   toIEEE754(v,  8, 23); }


// Convert array of octets to string binary representation
// by bartaz

function toIEEE754DoubleString(v) {
    return toIEEE754Double(v)
            .map(function(n){ for(n = n.toString(2);n.length < 8;n="0"+n); return n })
            .join('')
            .replace(/(.)(.{11})(.{52})/, "$1 $2 $3")
}

this snippet is really useful, thanks :)

I had a question about NaN which with the above implementation is represented in octets as [ 127, 240, 0, 0, 0, 0, 0, 1 ] however ieee754-visualization show that NaN is represented as [ 127, 248, 0, 0, 0, 0, 0, 0 ] which is correct since it's using TypedArrays, I think that this:

if (isNaN(v)) {
  e = (1 << bias) - 1; f = 1; s = 0;
}

should be:

if (isNaN(v)) {
  e = (1 << bias) - 1; f = Math.pow(2, 51); s = 0;
}

could not make out for what it is worth.
nevertheless, the heritage-crime ieee 754 gave us is always really annoying when it comes to covert into decimal (god should have given us 16 or least 8 fingers), since there is no accurate math about it.
about the issue :
power function would not let you get away with an underflow when it comes to tiny numbers, so it should have been
f = ((v * Math.pow(2, - ln)-1) * Math.pow(2, fbits); in line 28 and
f = ((v * Math.pow(2, bias - 1)) * Math.pow(2, fbits); in line 32