Internally, this puts the number in a single-element Float64Array typed array and gets its buffer to read the value’s raw bits:
Then it converts that buffer to a string of "1"s and "0"s by viewing the buffer as a Uint8Array and mapping each viewed byte to a binary string with .toString(2). Finally, it creates properties like exponent and mantissa by using  array indexing, .substring, and .slice to extract the relevant bits.
Similar to what asm.js and emscripten do. We even support negative zero correctly. :)