feat(Base64): lenient decoding

Strip non-base64 characters before decoding instead of only stripping trailing padding.

Co-authored-by: Amp <amp@ampcode.com>
Amp-Thread-ID: https://ampcode.com/threads/T-019c079b-e704-766b-95e5-24dbb1dace30

Author: jxom

.changeset/gentle-bytes-decode.md

@@ -0,0 +1,5 @@
+---
+"ox": patch
+---
+
+Base64.toBytes now strips invalid characters before decoding for lenient parsing.

src/core/Base64.ts

@@ -238,7 +238,7 @@ export declare namespace fromString {
  * @returns The Base64 decoded {@link ox#Bytes.Bytes}.
  */
 export function toBytes(value: string): Bytes.Bytes {
-  const base64 = value.replace(/=+$/, '')
+  const base64 = value.replace(/[^A-Za-z0-9+/_-]/g, '')
 
   const size = base64.length
 

src/core/_test/Base64.test.ts

@@ -59,6 +59,16 @@ describe('toBytes', () => {
       Bytes.fromString('hello wo�d'),
     )
   })
+
+  test('strips invalid characters', () => {
+    expect(Base64.toBytes('aGVsbG8!')).toStrictEqual(Bytes.fromString('hello'))
+    expect(Base64.toBytes('  aGVsbG8  ')).toStrictEqual(
+      Bytes.fromString('hello'),
+    )
+    expect(Base64.toBytes('aGVs\nbG8=')).toStrictEqual(
+      Bytes.fromString('hello'),
+    )
+  })
 })
 
 describe('toHex', () => {