How can source Int16Array values be retained though conversion to Uint8Array

Question

I pass data through functions in this format

Int16Array > Uint8Array > Int16Array

How to make z be the same as the original x

var x = new Int16Array([17, -45]);
var y = new Uint8Array(x);
var z = new Int16Array(y);

Want to make z as [ 17, -45 ]

EDIT: FULL EXAMPLE

const Readable = require('stream').Readable;

const rs = new Readable({
    objectMode: false,
    read(){}
});

var data = new Int16Array([17, -45]);
rs.push(data);

rs.on('data', function(chunk) {
    var result = new Int16Array(chunk);
});

I want "result" to have value [17, -45]

Answer 1

Number	16 bit (BE)	In binary (BE)	Note
17	0x00 0x11	00000000 00010001
-45	0xff 0xd3	11111111 11010011	255,211 in decimal
45	0x00 0xd3	00000000 00101101	The bits are flipped

A 16 bit TypedArray with two values in it would need 2 * Int16Array.BYTES_PER_ELEMENT = 4 bytes to store data. You can check that yourself:

console.log(
  arr16 = new Int16Array([17, -45]),
  arr16.buffer
);

[17, -45] // two elements (decimal)
...<11 00 d3 ff> // of 4 bytes total

Note: 11 00 & d3 ff as opposed to 00 11 & ff d3. The order is different because the representation is Little Endian. Don't worry about that.

If you pass the buffer to a Int8Array, you'd get:

console.log(
  arr8 = new Uint8Array( arr16.buffer ),
  arr8.buffer
);

[17, 0, 211, 255] // four elements (decimal)
...<11 00 d3 ff> // but same 4 bytes total

Note: In fact arr8 and arr16 both refer to the exact same bytes, so if you do arr16[0]=3, arr8[0] will change, too!

The inverse is not that hard to imagine now:

new Int16Array( arr8.buffer );
[17, -45] // two elements (decimal)

---

Thus, your sample would be:

var x = new Int16Array([17, -45]);
var y = new Uint8Array(x.buffer); // the .buffer is important!
var z = new Int16Array(y.buffer);

console.log(x[1] === z[1]); // true;

---

If, rather than to a UintArray you want to convert to a Node.js Buffer:

// Important to use .buffer here, otherwise it'll use the decimal
// representation of your input, not the actual bytes!
const buffer = Buffer.from(arr16.buffer);

Then you'll need to add:

new Int16Array( buffer.buffer )

Because if you do this:

new Int16Array( buffer );

Then you're effectively doing this:

const decimalRepresentationOfBytesInBuffer = [...buffer.values()];
new Int16Array( decimalRepresentationOfBytesInBuffer );
[17, 0, 11, 255] // four elements (decimal)
...<11 00 00 00 d3 00 ff 00> // of 8 bytes total