encoding/ascii85.ts: encode() returns a wrong result with a 4-byte subarray

[dahlia]

Describe the bug

The encode() function from encoding/ascii.ts returns a wrong result with a subarray. More specifically, if a subarray is a 4-byte (or multiples of 4 bytes) slice of a buffer longer than it, and the slice refers to a non-zero offset in the buffer, the function encodes the first 4 bytes of the buffer instead of the subarray bytes.

Steps to Reproduce

import { encode } from "https://deno.land/std@0.178.0/encoding/ascii85.ts";
import { assertEquals } from "https://deno.land/std@0.178.0/testing/asserts.ts";

Deno.test("base85 encoding", () => {
  const data = new Uint8Array([0x73, 0x70, 0x61, 0x6d]);
  const encoded = encode(data);
  assertEquals(encoded, "F)YQ)");
});

Deno.test("base85 encoding with a 4-byte subarray", () => {
  const data = new Uint8Array(
    [0x01, 0x02, 0x03, 0x04, 0x73, 0x70, 0x61, 0x6d],
  );
  const encoded = encode(data.subarray(4));

  // The following assertion fails; it returns "!<N?+" instead, which encodes
  // the first 4 bytes of the array: [0x01, 0x02, 0x03, 0x04].
  assertEquals(encoded, "F)YQ)");
});

Expected behavior

The below test should pass, but it does not.

Environment

• OS: Ubuntu 22.04 & Windows 11 Pro
• Deno version: 1.30.3
• std version: 0.178.0

[kamilogorek]

Probably the same issue that I've fixed here - #3208

https://github.com/denoland/deno_std/blob/69fa1e956c2469f168756d757b1653323327ed53/encoding/ascii85.ts#L78-L79

Should be trivial to port the patch there with the appropriate test.