@exodus/bytes 1.6.0 → 1.8.0

package/README.md

@@ -1,5 +1,9 @@
 # `@exodus/bytes`
 
+[![](https://flat.badgen.net/npm/v/@exodus/bytes)](https://npmjs.org/package/@exodus/bytes)
+![](https://flat.badgen.net/npm/dm/@exodus/bytes)
+[![](https://flat.badgen.net/npm/license/@exodus/bytes)](https://github.com/ExodusOSS/bytes/blob/HEAD/LICENSE)
+
 `Uint8Array` conversion to and from `base64`, `base32`, `base58`, `hex`, `utf8`, `utf16`, `bech32` and `wif`
 
 And a [`TextEncoder` / `TextDecoder` polyfill](#textencoder--textdecoder-polyfill)
@@ -31,13 +35,14 @@ See [Performance](./Performance.md) for more info
 
 ```js
 import { TextDecoder, TextEncoder } from '@exodus/bytes/encoding.js'
+import { TextDecoderStream, TextEncoderStream } from '@exodus/bytes/encoding.js' // Requires Streams
 ```
 
 Less than half the bundle size of [text-encoding](https://npmjs.com/text-encoding), [whatwg-encoding](https://npmjs.com/whatwg-encoding) or [iconv-lite](https://npmjs.com/iconv-lite) (gzipped or not).\
 Also [much faster](#fast) than all of those.
 
 > [!TIP]
-> See also the [lite version](#lite-version) to get this down to 9 KiB gzipped.
+> See also the [lite version](#lite-version) to get this down to 10 KiB gzipped.
 
 Spec compliant, passing WPT and covered with extra tests.\
 Moreover, tests for this library uncovered [bugs in all major implementations](https://docs.google.com/spreadsheets/d/1pdEefRG6r9fZy61WHGz0TKSt8cO4ISWqlpBN5KntIvQ/edit).\
@@ -77,11 +82,12 @@ _These are only provided as a compatibility layer, prefer hardened APIs instead
 If you don't need support for legacy multi-byte encodings, you can use the lite import:
 ```js
 import { TextDecoder, TextEncoder } from '@exodus/bytes/encoding-lite.js'
+import { TextDecoderStream, TextEncoderStream } from '@exodus/bytes/encoding-lite.js' // Requires Streams
 ```
 
-This reduces the bundle size 10x:\
-from 90 KiB gzipped for `@exodus/bytes/encoding.js` to 9 KiB gzipped for `@exodus/bytes/encoding-lite.js`.\
-(For comparison, `text-encoding` module is 190 KiB gzipped, and `iconv-lite` is 194 KiB gzipped).
+This reduces the bundle size 9x:\
+from 90 KiB gzipped for `@exodus/bytes/encoding.js` to 10 KiB gzipped for `@exodus/bytes/encoding-lite.js`.\
+(For comparison, `text-encoding` module is 190 KiB gzipped, and `iconv-lite` is 194 KiB gzipped):
 
 It still supports `utf-8`, `utf-16le`, `utf-16be` and all single-byte encodings specified by the spec,
 the only difference is support for legacy multi-byte encodings.
@@ -121,32 +127,90 @@ import { utf16fromStringLoose, utf16toStringLoose } from '@exodus/bytes/utf16.js
 ### `@exodus/bytes/single-byte.js`
 
 ```js
-import { createSinglebyteDecoder } from '@exodus/bytes/single-byte.js'
-import { windows1252toString } from '@exodus/bytes/single-byte.js'
+import { createSinglebyteDecoder, createSinglebyteEncoder } from '@exodus/bytes/single-byte.js'
+import { windows1252toString, windows1252fromString } from '@exodus/bytes/single-byte.js'
 ```
 
-Decode the legacy single-byte encodings according to the [Encoding standard](https://encoding.spec.whatwg.org/)
-([§9](https://encoding.spec.whatwg.org/#legacy-single-byte-encodings) and
-[§14.5](https://encoding.spec.whatwg.org/#x-user-defined)).
+Decode / encode the legacy single-byte encodings according to the
+[Encoding standard](https://encoding.spec.whatwg.org/)
+([§9](https://encoding.spec.whatwg.org/#legacy-single-byte-encodings),
+[§14.5](https://encoding.spec.whatwg.org/#x-user-defined)),
+and [unicode.org](https://unicode.org/Public/MAPPINGS/ISO8859) `iso-8859-*` mappings.
 
-Supports all single-byte encodings listed in the standard:
+Supports all single-byte encodings listed in the WHATWG Encoding standard:
 `ibm866`, `iso-8859-2`, `iso-8859-3`, `iso-8859-4`, `iso-8859-5`, `iso-8859-6`, `iso-8859-7`, `iso-8859-8`,
 `iso-8859-8-i`, `iso-8859-10`, `iso-8859-13`, `iso-8859-14`, `iso-8859-15`, `iso-8859-16`, `koi8-r`, `koi8-u`,
 `macintosh`, `windows-874`, `windows-1250`, `windows-1251`, `windows-1252`, `windows-1253`, `windows-1254`,
 `windows-1255`, `windows-1256`, `windows-1257`, `windows-1258`, `x-mac-cyrillic` and `x-user-defined`.
 
+Also supports `iso-8859-1`, `iso-8859-9`, `iso-8859-11` as defined at
+[unicode.org](https://unicode.org/Public/MAPPINGS/ISO8859)
+(and all other `iso-8859-*` encodings there as they match WHATWG).
+
+> [!NOTE]
+> While all `iso-8859-*` encodings supported by the [WHATWG Encoding standard](https://encoding.spec.whatwg.org/) match
+> [unicode.org](https://unicode.org/Public/MAPPINGS/ISO8859), the WHATWG Encoding spec doesn't support
+> `iso-8859-1`, `iso-8859-9`, `iso-8859-11`, and instead maps them as labels to `windows-1252`, `windows-1254`, `windows-874`.\
+> `createSinglebyteDecoder()` (unlike `TextDecoder` or `legacyHookDecode()`) does not do such mapping,
+> so its results will differ from `TextDecoder` for those encoding names.
+
+```js
+> new TextDecoder('iso-8859-1').encoding
+'windows-1252'
+> new TextDecoder('iso-8859-9').encoding
+'windows-1254'
+> new TextDecoder('iso-8859-11').encoding
+'windows-874'
+> new TextDecoder('iso-8859-9').decode(Uint8Array.of(0x80, 0x81, 0xd0))
+'€\x81Ğ' // this is actually decoded according to windows-1254 per TextDecoder spec
+> createSinglebyteDecoder('iso-8859-9')(Uint8Array.of(0x80, 0x81, 0xd0))
+'\x80\x81Ğ' // this is iso-8859-9 as defined at https://unicode.org/Public/MAPPINGS/ISO8859/8859-9.txt
+```
+
 ##### `createSinglebyteDecoder(encoding, loose = false)`
 
-Create a decoder for a supported one-byte `encoding`, given it's lowercased name `encoding`.
+Create a decoder for a supported one-byte `encoding`, given its lowercased name `encoding`.
 
 Returns a function `decode(arr)` that decodes bytes to a string.
 
+##### `createSinglebyteEncoder(encoding, { mode = 'fatal' })`
+
+Create an encoder for a supported one-byte `encoding`, given its lowercased name `encoding`.
+
+Returns a function `encode(string)` that encodes a string to bytes.
+
+In `'fatal'` mode (default), will throw on non well-formed strings or any codepoints which could
+not be encoded in the target encoding.
+
+##### `latin1toString(arr)`
+
+Decode `iso-8859-1` bytes to a string.
+
+There is no loose variant for this encoding, all bytes can be decoded.
+
+Same as:
+```js
+const latin1toString = createSinglebyteDecoder('iso-8859-1')
+```
+
+Note: this is different from `new TextDecoder('iso-8859-1')` and `new TextDecoder('latin1')`, as
+those alias to `new TextDecoder('windows-1252')`.
+
+##### `latin1fromString(string)`
+
+Encode a string to `iso-8859-1` bytes.
+
+Will throw on non well-formed strings or any codepoints which could not be encoded in `iso-8859-1`.
+
+Same as:
+```js
+const latin1fromString = createSinglebyteEncoder('iso-8859-1', { mode: 'fatal' })
+```
+
 ##### `windows1252toString(arr)`
 
 Decode `windows-1252` bytes to a string.
 
-Also supports `ascii` and `latin-1` as those are strict subsets of `windows-1252`.
-
 There is no loose variant for this encoding, all bytes can be decoded.
 
 Same as:
@@ -154,6 +218,17 @@ Same as:
 const windows1252toString = createSinglebyteDecoder('windows-1252')
 ```
 
+##### `windows1252fromString(string)`
+
+Encode a string to `windows-1252` bytes.
+
+Will throw on non well-formed strings or any codepoints which could not be encoded in `windows-1252`.
+
+Same as:
+```js
+const windows1252fromString = createSinglebyteEncoder('windows-1252', { mode: 'fatal' })
+```
+
 ### `@exodus/bytes/multi-byte.js`
 
 ```js
@@ -171,7 +246,7 @@ Supports all legacy multi-byte encodings listed in the standard:
 
 ##### `createMultibyteDecoder(encoding, loose = false)`
 
-Create a decoder for a supported legacy multi-byte `encoding`, given it's lowercased name `encoding`.
+Create a decoder for a supported legacy multi-byte `encoding`, given its lowercased name `encoding`.
 
 Returns a function `decode(arr, stream = false)` that decodes bytes to a string.
 
@@ -188,6 +263,8 @@ import { fromBigInt, toBigInt } from '@exodus/bytes/bigint.js'
 
 ### `@exodus/bytes/hex.js`
 
+Implements Base16 from [RFC4648](https://datatracker.ietf.org/doc/html/rfc4648) (no differences from [RFC3548](https://datatracker.ietf.org/doc/html/rfc4648)).
+
 ```js
 import { fromHex, toHex } from '@exodus/bytes/hex.js'
 ```
@@ -197,6 +274,8 @@ import { fromHex, toHex } from '@exodus/bytes/hex.js'
 
 ### `@exodus/bytes/base64.js`
 
+Implements Base64 from [RFC4648](https://datatracker.ietf.org/doc/html/rfc4648) (no differences from [RFC3548](https://datatracker.ietf.org/doc/html/rfc4648)).
+
 ```js
 import { fromBase64, toBase64 } from '@exodus/bytes/base64.js'
 import { fromBase64url, toBase64url } from '@exodus/bytes/base64.js'
@@ -211,6 +290,8 @@ import { fromBase64any } from '@exodus/bytes/base64.js'
 
 ### `@exodus/bytes/base32.js`
 
+Implements Base32 from [RFC4648](https://datatracker.ietf.org/doc/html/rfc4648) (no differences from [RFC3548](https://datatracker.ietf.org/doc/html/rfc4648)).
+
 ```js
 import { fromBase32, toBase32 } from '@exodus/bytes/base32.js'
 import { fromBase32hex, toBase32hex } from '@exodus/bytes/base32.js'
@@ -223,6 +304,8 @@ import { fromBase32hex, toBase32hex } from '@exodus/bytes/base32.js'
 
 ### `@exodus/bytes/bech32.js`
 
+Implements [BIP-0173](https://github.com/bitcoin/bips/blob/master/bip-0173.mediawiki#specification) and [BIP-0350](https://github.com/bitcoin/bips/blob/master/bip-0350.mediawiki#specification).
+
 ```js
 import { fromBech32, toBech32 } from '@exodus/bytes/bech32.js'
 import { fromBech32m, toBech32m } from '@exodus/bytes/base32.js'
@@ -284,6 +367,7 @@ On non-Node.js, requires peer dependency [@exodus/crypto](https://www.npmjs.com/
 
 ```js
 import { TextDecoder, TextEncoder } from '@exodus/bytes/encoding.js'
+import { TextDecoderStream, TextEncoderStream } from '@exodus/bytes/encoding.js' // Requires Streams
 
 // Hooks for standards
 import { getBOMEncoding, legacyHookDecode, labelToName, normalizeEncoding } from '@exodus/bytes/encoding.js'
@@ -291,7 +375,9 @@ import { getBOMEncoding, legacyHookDecode, labelToName, normalizeEncoding } from
 
 Implements the [Encoding standard](https://encoding.spec.whatwg.org/):
 [TextDecoder](https://encoding.spec.whatwg.org/#interface-textdecoder),
-[TextEncoder](https://encoding.spec.whatwg.org/#interface-textdecoder),
+[TextEncoder](https://encoding.spec.whatwg.org/#interface-textencoder),
+[TextDecoderStream](https://encoding.spec.whatwg.org/#interface-textdecoderstream),
+[TextEncoderStream](https://encoding.spec.whatwg.org/#interface-textencoderstream),
 some [hooks](https://encoding.spec.whatwg.org/#specification-hooks) (see below).
 
 #### `new TextDecoder(label = 'utf-8', { fatal = false, ignoreBOM = false })`
@@ -300,7 +386,21 @@ some [hooks](https://encoding.spec.whatwg.org/#specification-hooks) (see below).
 
 #### `new TextEncoder()`
 
-[TextEncoder](https://encoding.spec.whatwg.org/#interface-textdecoder) implementation/polyfill.
+[TextEncoder](https://encoding.spec.whatwg.org/#interface-textencoder) implementation/polyfill.
+
+#### `new TextDecoderStream(label = 'utf-8', { fatal = false, ignoreBOM = false })`
+
+[TextDecoderStream](https://encoding.spec.whatwg.org/#interface-textdecoderstream) implementation/polyfill.
+
+Requires [Streams](https://streams.spec.whatwg.org/) to be either supported by the platform or
+[polyfilled](https://npmjs.com/package/web-streams-polyfill).
+
+#### `new TextEncoderStream()`
+
+[TextEncoderStream](https://encoding.spec.whatwg.org/#interface-textencoderstream) implementation/polyfill.
+
+Requires [Streams](https://streams.spec.whatwg.org/) to be either supported by the platform or
+[polyfilled](https://npmjs.com/package/web-streams-polyfill).
 
 #### `labelToName(label)`
 
@@ -370,6 +470,7 @@ new TextDecoder(getBOMEncoding(input) ?? fallbackEncoding).decode(input)
 
 ```js
 import { TextDecoder, TextEncoder } from '@exodus/bytes/encoding-lite.js'
+import { TextDecoderStream, TextEncoderStream } from '@exodus/bytes/encoding-lite.js' // Requires Streams
 
 // Hooks for standards
 import { getBOMEncoding, legacyHookDecode, labelToName, normalizeEncoding } from '@exodus/bytes/encoding-lite.js'
@@ -393,7 +494,9 @@ To avoid inconsistencies, the exported classes and methods are exactly the same
 > lite = require('@exodus/bytes/encoding-lite.js')
 [Module: null prototype] {
   TextDecoder: [class TextDecoder],
+  TextDecoderStream: [class TextDecoderStream],
   TextEncoder: [class TextEncoder],
+  TextEncoderStream: [class TextEncoderStream],
   getBOMEncoding: [Function: getBOMEncoding],
   labelToName: [Function: labelToName],
   legacyHookDecode: [Function: legacyHookDecode],
@@ -406,7 +509,9 @@ Error: Legacy multi-byte encodings are disabled in /encoding-lite.js, use /encod
 > full = require('@exodus/bytes/encoding.js')
 [Module: null prototype] {
   TextDecoder: [class TextDecoder],
+  TextDecoderStream: [class TextDecoderStream],
   TextEncoder: [class TextEncoder],
+  TextEncoderStream: [class TextEncoderStream],
   getBOMEncoding: [Function: getBOMEncoding],
   labelToName: [Function: labelToName],
   legacyHookDecode: [Function: legacyHookDecode],

package/base32.js

@@ -1,5 +1,6 @@
 import { assertEmptyRest } from './assert.js'
 import { typedView } from './array.js'
+import { E_STRING } from './fallback/_utils.js'
 import * as js from './fallback/base32.js'
 
 // See https://datatracker.ietf.org/doc/html/rfc4648
@@ -25,7 +26,7 @@ export function fromBase32hex(str, options) {
 }
 
 function fromBase32common(str, isBase32Hex, padding, format, rest) {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   if (rest !== null) assertEmptyRest(rest)
 
   if (padding === true) {

package/base58.js

@@ -1,6 +1,6 @@
 import { typedView } from './array.js'
 import { assertUint8 } from './assert.js'
-import { nativeDecoder, nativeEncoder, isHermes } from './fallback/_utils.js'
+import { nativeDecoder, nativeEncoder, isHermes, E_STRING } from './fallback/_utils.js'
 import { encodeAscii, decodeAscii } from './fallback/latin1.js'
 
 const alphabet58 = [...'123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz']
@@ -122,7 +122,7 @@ function toBase58core(arr, alphabet, codes) {
 }
 
 function fromBase58core(str, alphabet, codes, format = 'uint8') {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   const length = str.length
   if (length === 0) return typedView(new Uint8Array(), format)
 

package/base64.js

@@ -1,6 +1,6 @@
 import { assertUint8, assertEmptyRest } from './assert.js'
 import { typedView } from './array.js'
-import { isHermes, skipWeb } from './fallback/_utils.js'
+import { isHermes, skipWeb, E_STRING } from './fallback/_utils.js'
 import { decodeLatin1, encodeLatin1 } from './fallback/latin1.js'
 import * as js from './fallback/base64.js'
 
@@ -79,7 +79,7 @@ export function fromBase64any(str, { format = 'uint8', padding = 'both', ...rest
 }
 
 function fromBase64common(str, isBase64url, padding, format, rest) {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   if (rest !== null) assertEmptyRest(rest)
   const auto = padding === 'both' ? str.endsWith('=') : undefined
   // Older JSC supporting Uint8Array.fromBase64 lacks proper checks

package/bech32.js

@@ -1,5 +1,5 @@
 import { assertUint8 } from './assert.js'
-import { nativeEncoder } from './fallback/_utils.js'
+import { nativeEncoder, E_STRING } from './fallback/_utils.js'
 import { decodeAscii, encodeAscii, encodeLatin1 } from './fallback/latin1.js'
 
 const alphabet = [...'qpzry9x8gf2tvdw0s3jn54khce6mua7l']
@@ -12,7 +12,6 @@ const E_MIXED = 'Mixed-case string'
 const E_PADDING = 'Padding is invalid'
 const E_CHECKSUM = 'Invalid checksum'
 const E_CHARACTER = 'Non-bech32 character'
-const E_STRING = 'Input is not a string'
 
 // nativeEncoder path uses encodeAscii which asserts ascii, otherwise we have 0-255 bytes from encodeLatin1
 const c2x = new Int8Array(nativeEncoder ? 128 : 256).fill(-1)
@@ -179,6 +178,9 @@ function assertDecodeArgs(str, limit) {
   if (typeof limit !== 'number' || str.length < 8 || !(str.length <= limit)) throw new Error(E_SIZE)
 }
 
+// this is instant on 8-bit strings
+const NON_LATIN = /[^\x00-\xFF]/ // eslint-disable-line no-control-regex
+
 function fromBech32enc(str, limit, encoding) {
   assertDecodeArgs(str, limit)
   const lower = str.toLowerCase()
@@ -195,6 +197,7 @@ function fromBech32enc(str, limit, encoding) {
   if (wordsLength < 0) throw new Error(E_SIZE)
   const bytesLength = (wordsLength * 5) >> 3
   const slice = str.slice(split + 1)
+  if (!nativeEncoder && NON_LATIN.test(slice)) throw new SyntaxError(E_CHARACTER) // otherwise can't use encodeLatin1
   const c = nativeEncoder ? encodeAscii(slice, E_CHARACTER) : encodeLatin1(slice) // suboptimal, but only affects non-Hermes barebones
   const bytes = new Uint8Array(bytesLength)
 

package/encoding-lite.d.ts

@@ -0,0 +1 @@
+export * from './encoding.js'

package/encoding-lite.js

@@ -1,6 +1,8 @@
 export {
   TextDecoder,
   TextEncoder,
+  TextDecoderStream,
+  TextEncoderStream,
   normalizeEncoding,
   getBOMEncoding,
   labelToName,

package/encoding.d.ts

@@ -0,0 +1,58 @@
+/// <reference types="node" />
+
+/**
+ * Converts an encoding label to its name, as an ASCII-lowercased string
+ * @param label - The encoding label to normalize
+ * @returns The normalized encoding name, or null if invalid
+ */
+export function normalizeEncoding(label: string): string | null;
+
+/**
+ * Implements BOM sniff (https://encoding.spec.whatwg.org/#bom-sniff) legacy hook.
+ * @param input - The bytes to check for BOM
+ * @returns The encoding ('utf-8', 'utf-16le', 'utf-16be'), or null if no BOM found
+ */
+export function getBOMEncoding(
+  input: ArrayBufferLike | ArrayBufferView
+): 'utf-8' | 'utf-16le' | 'utf-16be' | null;
+
+/**
+ * Implements decode (https://encoding.spec.whatwg.org/#decode) legacy hook.
+ * @param input - The bytes to decode
+ * @param fallbackEncoding - The encoding to use if no BOM detected (default: 'utf-8')
+ * @returns The decoded string
+ */
+export function legacyHookDecode(
+  input: ArrayBufferLike | ArrayBufferView,
+  fallbackEncoding?: string
+): string;
+
+/**
+ * Converts an encoding label to its name, as a case-sensitive string.
+ * @param label - The encoding label
+ * @returns The proper case encoding name, or null if invalid
+ */
+export function labelToName(label: string): string | null;
+
+/**
+ * Text decoder for decoding bytes to strings in various encodings
+ * Supports strict and lossy modes
+ */
+export const TextDecoder: typeof globalThis.TextDecoder;
+
+/**
+ * Text encoder for encoding strings to UTF-8 bytes
+ */
+export const TextEncoder: typeof globalThis.TextEncoder;
+
+/**
+ * Transform stream wrapper for TextDecoder
+ * Decodes chunks of bytes to strings
+ */
+export const TextDecoderStream: typeof globalThis.TextDecoderStream;
+
+/**
+ * Transform stream wrapper for TextEncoder
+ * Encodes chunks of strings to UTF-8 bytes
+ */
+export const TextEncoderStream: typeof globalThis.TextEncoderStream;

package/encoding.js

@@ -6,6 +6,8 @@ setMultibyteDecoder(createMultibyteDecoder)
 export {
   TextDecoder,
   TextEncoder,
+  TextDecoderStream,
+  TextEncoderStream,
   normalizeEncoding,
   getBOMEncoding,
   labelToName,

package/fallback/_utils.js

@@ -5,14 +5,8 @@ export const isHermes = Boolean(globalThis.HermesInternal)
 export const isDeno = Boolean(globalThis.Deno)
 export const isLE = new Uint8Array(Uint16Array.of(258).buffer)[0] === 2
 
-let isNative = (x) => {
-  if (!x) return false
-  if (haveNativeBuffer) return true // we consider Node.js TextDecoder/TextEncoder native
-  const s = `${x}`
-  // See https://github.com/facebook/hermes/pull/1855#issuecomment-3659386410
-  return s.includes('[native code]') || s.includes(`[bytecode]`) // Static Hermes has [bytecode] for contrib, which includes TextEncoder/TextDecoder
-}
-
+// We consider Node.js TextDecoder/TextEncoder native
+let isNative = (x) => x && (haveNativeBuffer || `${x}`.includes('[native code]'))
 if (!haveNativeBuffer && isNative(() => {})) isNative = () => false // e.g. XS, we don't want false positives
 
 export const nativeEncoder = isNative(TextEncoder) ? new TextEncoder() : null
@@ -134,3 +128,5 @@ export const toBuf = (x) =>
   x.byteLength <= 64 && x.BYTES_PER_ELEMENT === 1
     ? Buffer.from(x)
     : Buffer.from(x.buffer, x.byteOffset, x.byteLength)
+
+export const E_STRING = 'Input is not a string'

package/fallback/encoding.js

@@ -68,7 +68,16 @@ function isAnyUint8Array(x) {
 const fromSource = (x) => {
   if (x instanceof Uint8Array) return x
   if (ArrayBuffer.isView(x)) return new Uint8Array(x.buffer, x.byteOffset, x.byteLength)
-  if (isAnyArrayBuffer(x)) return new Uint8Array(x)
+  if (isAnyArrayBuffer(x)) {
+    if ('detached' in x) return x.detached === true ? new Uint8Array() : new Uint8Array(x)
+    // Old engines without .detached, try-catch
+    try {
+      return new Uint8Array(x)
+    } catch {
+      return new Uint8Array()
+    }
+  }
+
   throw new TypeError('Argument must be a SharedArrayBuffer, ArrayBuffer or ArrayBufferView')
 }
 
@@ -259,6 +268,79 @@ export class TextEncoder {
   }
 }
 
+const E_NO_STREAMS = 'TransformStream global not present in the environment'
+
+// https://encoding.spec.whatwg.org/#interface-textdecoderstream
+export class TextDecoderStream {
+  constructor(encoding = 'utf-8', options = {}) {
+    if (!globalThis.TransformStream) throw new Error(E_NO_STREAMS)
+    const decoder = new TextDecoder(encoding, options)
+    const transform = new TransformStream({
+      transform: (chunk, controller) => {
+        const value = decoder.decode(fromSource(chunk), { stream: true })
+        if (value) controller.enqueue(value)
+      },
+      flush: (controller) => {
+        // https://streams.spec.whatwg.org/#dom-transformer-flush
+        const value = decoder.decode()
+        if (value) controller.enqueue(value)
+        // No need to call .terminate() (Node.js is wrong)
+      },
+    })
+
+    define(this, 'encoding', decoder.encoding)
+    define(this, 'fatal', decoder.fatal)
+    define(this, 'ignoreBOM', decoder.ignoreBOM)
+    define(this, 'readable', transform.readable)
+    define(this, 'writable', transform.writable)
+  }
+
+  get [Symbol.toStringTag]() {
+    return 'TextDecoderStream'
+  }
+}
+
+// https://encoding.spec.whatwg.org/#interface-textencoderstream
+// Only UTF-8 per spec
+export class TextEncoderStream {
+  constructor() {
+    if (!globalThis.TransformStream) throw new Error(E_NO_STREAMS)
+    let lead
+    const transform = new TransformStream({
+      // https://encoding.spec.whatwg.org/#encode-and-enqueue-a-chunk
+      // Not identical in code, but reuses loose mode to have identical behavior
+      transform: (chunk, controller) => {
+        let s = String(chunk) // DOMString, might contain unpaired surrogates
+        if (s.length === 0) return
+        if (lead) {
+          s = lead + s
+          lead = null
+        }
+
+        const last = s.charCodeAt(s.length - 1) // Can't come from previous lead due to length check
+        if ((last & 0xfc_00) === 0xd8_00) {
+          lead = s[s.length - 1]
+          s = s.slice(0, -1)
+        }
+
+        if (s) controller.enqueue(utf8fromStringLoose(s))
+      },
+      // https://encoding.spec.whatwg.org/#encode-and-flush
+      flush: (controller) => {
+        if (lead) controller.enqueue(Uint8Array.of(0xef, 0xbf, 0xbd))
+      },
+    })
+
+    define(this, 'encoding', 'utf-8')
+    define(this, 'readable', transform.readable)
+    define(this, 'writable', transform.writable)
+  }
+
+  get [Symbol.toStringTag]() {
+    return 'TextEncoderStream'
+  }
+}
+
 // Warning: unlike whatwg-encoding, returns lowercased labels
 // Those are case-insensitive and that's how TextDecoder encoding getter normalizes them
 export function getBOMEncoding(input) {

package/fallback/hex.js

@@ -1,5 +1,5 @@
 import { assertUint8 } from '../assert.js'
-import { nativeDecoder, nativeEncoder, decode2string } from './_utils.js'
+import { nativeDecoder, nativeEncoder, decode2string, E_STRING } from './_utils.js'
 import { encodeAscii, decodeAscii } from './latin1.js'
 
 let hexArray // array of 256 bytes converted to two-char hex strings
@@ -52,7 +52,7 @@ export function toHex(arr) {
 }
 
 export function fromHex(str) {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   if (str.length % 2 !== 0) throw new SyntaxError(E_HEX)
 
   const length = str.length / 2 // this helps Hermes in loops

package/fallback/latin1.js

@@ -6,12 +6,19 @@ import {
   isHermes,
   isDeno,
   isLE,
+  skipWeb,
 } from './_utils.js'
 
+const { atob } = globalThis
+const { toBase64: web64 } = Uint8Array.prototype
+
 // See http://stackoverflow.com/a/22747272/680742, which says that lowest limit is in Chrome, with 0xffff args
 // On Hermes, actual max is 0x20_000 minus current stack depth, 1/16 of that should be safe
 const maxFunctionArgs = 0x20_00
 
+// toBase64+atob path is faster on everything where fromBase64 is fast
+const useLatin1atob = web64 && atob && !skipWeb
+
 export function asciiPrefix(arr) {
   let p = 0 // verified ascii bytes
   const length = arr.length
@@ -46,6 +53,18 @@ export function decodeLatin1(arr, start = 0, stop = arr.length) {
   stop |= 0
   const total = stop - start
   if (total === 0) return ''
+
+  if (
+    useLatin1atob &&
+    total >= 256 &&
+    total < 1e8 &&
+    arr.toBase64 === web64 &&
+    arr.BYTES_PER_ELEMENT === 1
+  ) {
+    const sliced = start === 0 && stop === arr.length ? arr : arr.subarray(start, stop)
+    return atob(sliced.toBase64())
+  }
+
   if (total > maxFunctionArgs) {
     let prefix = ''
     for (let i = start; i < stop; ) {
@@ -107,8 +126,23 @@ export const encodeCharcodes = isHermes
       return arr
     }
 
+export function encodeAsciiPrefix(x, s) {
+  let i = 0
+  for (const len3 = s.length - 3; i < len3; i += 4) {
+    const x0 = s.charCodeAt(i), x1 = s.charCodeAt(i + 1), x2 = s.charCodeAt(i + 2), x3 = s.charCodeAt(i + 3) // prettier-ignore
+    if ((x0 | x1 | x2 | x3) >= 128) break
+    x[i] = x0
+    x[i + 1] = x1
+    x[i + 2] = x2
+    x[i + 3] = x3
+  }
+
+  return i
+}
+
 /* eslint-enable @exodus/mutable/no-param-reassign-prop-only */
 
+// Warning: can be used only on checked strings, converts strings to 8-bit
 export const encodeLatin1 = (str) => encodeCharcodes(str, new Uint8Array(str.length))
 
 // Expects nativeEncoder to be present

package/fallback/multi-byte.js

@@ -474,6 +474,7 @@ const mappers = {
 export const isAsciiSuperset = (enc) => enc !== 'iso-2022-jp' // all others are ASCII supersets and can use fast path
 
 export function multibyteDecoder(enc, loose = false) {
+  if (typeof loose !== 'boolean') throw new TypeError('loose option should be boolean')
   if (!Object.hasOwn(mappers, enc)) throw new RangeError('Unsupported encoding')
 
   // Input is assumed to be typechecked already

package/fallback/single-byte.encodings.js

@@ -8,54 +8,75 @@ const h = (x) => new Array(x).fill(r)
 
 // Common ranges
 
-// prettier-ignore
-const k8a = [9345,2,10,4,4,4,4,8,8,8,8,68,4,4,4,4,1,1,1,-627,640,-903,1,46,28,1,-8645,8833,-8817,2,5,64,9305,1,1,-8449]
-// prettier-ignore
-const k8b = [-30,1,21,-18,1,15,-17,18,-13,...e(7),16,-15,1,1,1,-13,-4,26,-1,-20,17,5,-4,-2,3]
-const p1 = [8237, -8235, 8089, -7816, 7820, 8, -6, 1]
-const p2 = [-99, 12, 20, -12, 17, 37, -29, 2]
-// prettier-ignore
-const p3 = [1,1,65,-63,158,-156,1,1,1,40,30,42,-46,6,-66,1,83,-6,-6,-67,176,...p2,-114,121,-119,1,1,155,-49,25,16,-142,159,2,-158,38,42,-46,6,-35,1,52,-6,-6,-36,145,...p2,-83,90,-88,1,1,124,-49,25,16,-111,128,2]
-const i0 = e(33)
 // prettier-ignore
 const i2 = [-40,-147,1,64,-62,117,-51,-63,69,-67,79,-77,79,-77,1,64,2,51,4,-116,1,124,-122,1,129,22,-148,150,-148,1,133,-131,118,-116,1,33,-31,86,-51,-32,38,-36,48,-46,48,-46,1,33,2,51,4,-85,1,93,-91,1,98,22,-117,119,-117,1,102,374]
-const i4a = [-75, -63, ...e(5), 104, -34, -67, 79, -77, 75, -73, 1]
-const i4b = [34, -32, ...e(5), 73, -34, -36, 48, -46, 44, -42, 1]
-const i7 = [721, 1, 1, -719, 721, -719, 721, ...e(19), r, 2, ...e(43), r]
-const i8 = [...e(26), r, r, 6692, 1, r]
+const i4a = [-75, -63, e(5), 104, -34, -67, 79, -77, 75, -73, 1]
+const i4b = [34, -32, e(5), 73, -34, -36, 48, -46, 44, -42, 1]
+const i7 = [721, 1, 1, -719, 721, -719, 721, e(19), r, 2, e(43), r]
+const i8 = [e(26), r, r, 6692, 1, r]
+const i9 = [79, -77, e(11), 84, 46, -127, e(16), 48, -46, e(11), 53, 46]
+const iB = [3425, e(57), h(4), 5, e(28), h(4)]
+const p2 = [-99, 12, 20, -12, 17, 37, -29, 2]
+const p1 = [8237, -8235, 8089, -7816, 7820, 8, -6, 1]
 const w0 = [8237, -8235, 8089, -8087, 8091, 8, -6, 1, -8089, 8104]
 const w8 = [8072, 1, 3, 1, 5, -15, 1]
-const w1 = [...w8, -7480, 7750, -8129, 7897, -7911, -182]
-const w3 = [...w8, -8060, 8330, -8328, 8096, -8094]
+const w1 = [w8, -7480, 7750, -8129, 7897, -7911, -182]
+const w3 = [w8, -8060, 8330, -8328, 8096, -8094]
 const m0 = [8558, -8328, 8374, -66, -8539, 16, 8043, -8070]
+// prettier-ignore
+const p3 = [1,1,65,-63,158,-156,1,1,1,40,30,42,-46,6,-66,1,83,-6,-6,-67,176,p2,-114,121,-119,1,1,155,-49,25,16,-142,159,2,-158,38,42,-46,6,-35,1,52,-6,-6,-36,145,p2,-83,90,-88,1,1,124,-49,25,16,-111,128,2]
+// prettier-ignore
+const k8a = [9345,2,10,4,4,4,4,8,8,8,8,68,4,4,4,4,1,1,1,-627,640,-903,1,46,28,1,-8645,8833,-8817,2,5,64,9305,1,1,-8449]
+// prettier-ignore
+const k8b = [-30,1,21,-18,1,15,-17,18,-13,e(7),16,-15,1,1,1,-13,-4,26,-1,-20,17,5,-4,-2,3]
 
 // prettier-ignore
-export default {
-  ibm866: [913,...e(47),8530,1,1,-145,34,61,1,-12,-1,14,-18,6,6,-1,-1,-75,4,32,-8,-16,-28,60,34,1,-5,-6,21,-3,-6,-16,28,-5,1,-4,1,-12,-1,-6,1,24,-1,-82,-12,124,-4,8,4,-16,-8512,...e(15),-78,80,-77,80,-77,80,-73,80,-942,8553,-8546,8547,-260,-8306,9468,-9472],
-  'iso-8859-10': [...i0,100,14,16,8,-2,14,-143,148,-43,80,6,23,-208,189,-32,-154,85,14,16,8,-2,14,-128,133,-43,80,6,23,7831,-7850,-32,...i4a,1,1,117,7,-121,1,1,1,146,-144,154,-152,...e(5),...i4b,1,1,86,7,-90,1,1,1,115,-113,123,-121,1,1,1,1,58],
-  'iso-8859-13': [...i0,8061,-8059,1,1,8058,-8056,1,49,-47,173,-171,1,1,1,24,-22,1,1,1,8041,-8039,...p3,7835],
-  'iso-8859-14': [...i0,7522,1,-7520,103,1,7423,-7523,7641,-7639,7641,-119,231,-7749,1,202,7334,1,-7423,1,7455,1,-7563,7584,43,-42,44,-35,147,-111,1,-36,-7585,...e(15),165,-163,...e(5),7572,-7570,...e(5),153,-151,...e(16),134,-132,...e(5),7541,-7539,...e(5),122],
-  'iso-8859-15': [...i0,1,1,1,8201,-8199,187,-185,186,-184,...e(10),202,-200,1,1,199,-197,1,1,151,1,37],
-  'iso-8859-16': [...i0,100,1,60,8043,-142,-7870,-185,186,-184,367,-365,206,-204,205,1,-203,1,91,54,59,7840,-8039,1,199,-113,268,-350,151,1,37,4,-188,1,1,64,-62,66,-64,...e(9),65,51,-113,1,1,124,-122,132,22,-151,1,1,1,60,258,-315,1,1,1,33,-31,35,-33,...e(9),34,51,-82,1,1,93,-91,101,22,-120,1,1,1,29,258],
-  'iso-8859-2': [...i0,100,468,-407,-157,153,29,-179,1,184,-2,6,21,-204,208,-2,-203,85,470,-409,-142,138,29,364,-527,169,-2,6,21,355,-351,-2,...i2],
-  'iso-8859-3': [...i0,134,434,-565,1,r,128,-125,1,136,46,-64,22,-135,r,206,-203,119,-117,1,1,1,112,-110,1,121,46,-64,22,-120,r,191,-188,1,1,r,2,70,-2,-65,...e(8),r,2,1,1,1,76,-74,1,69,-67,1,1,1,144,-16,-125,1,1,1,r,2,39,-2,-34,...e(8),r,2,1,1,1,45,-43,1,38,-36,1,1,1,113,-16,380],
-  'iso-8859-4': [...i0,100,52,30,-178,132,19,-148,1,184,-78,16,68,-185,208,-206,1,85,470,-388,-163,117,19,395,-527,169,-78,16,68,-29,52,-51,...i4a,92,-26,53,7,-22,-98,1,1,1,1,154,-152,1,1,140,2,-139,...i4b,61,-26,53,7,-22,-67,1,1,1,1,123,-121,1,1,109,2,366],
-  'iso-8859-5': [...i0,865,...e(11),-863,865,...e(65),7367,-7365,...e(11),-949,951,1],
-  'iso-8859-6': [...i0,r,r,r,4,...h(7),1384,-1375,...h(13),1390,r,r,r,4,r,2,...e(25),r,r,r,r,r,6,...e(18),...h(13)],
-  'iso-8859-7': [...i0,8056,1,-8054,8201,3,-8201,1,1,1,721,-719,1,1,r,8040,-8037,1,1,1,721,1,1,-719,...i7],
-  'iso-8859-8': [...i0,r,2,...e(7),46,-44,...e(14),62,-60,1,1,1,...h(32),8025,-6727,...i8],
-  'koi8-r': [...k8a,8450,...e(14),-8544,8545,...e(10),-9411,933,...k8b,-28,...k8b],
-  'koi8-u': [...k8a,3,8448,-8446,1,8448,1,1,1,1,-8394,-51,8448,1,1,1,-8544,3,8543,-8541,1,8543,1,1,1,1,-8410,-130,-869,933,...k8b,-28,...k8b],
-  macintosh: [69,1,2,2,8,5,6,5,-1,2,2,-1,2,2,2,-1,2,1,2,-1,2,1,2,2,-1,2,2,-1,5,-1,2,1,7972,-8048,-14,1,4,8059,-8044,41,-49,-5,8313,-8302,-12,8632,-8602,18,8518,-8557,8627,1,-8640,16,8525,15,-2,-7759,7787,-8577,16,751,-707,18,-57,-30,11,...m0,32,3,18,125,1,7872,1,8,1,-5,1,-7970,9427,-9419,121,7884,104,-115,1,56007,1,-56033,-8042,8035,4,18,-8046,8,-9,10,-3,5,1,1,-3,7,1,63531,-63533,8,1,-2,88,405,22,-557,553,1,1,-546,549,-2,-20],
-  'windows-1250': [...w0,-7888,7897,-7903,10,25,-4,-233,...w8,-8060,8330,-8129,7897,-7903,10,25,-4,-218,551,17,-407,-157,96,-94,1,1,1,181,-179,1,1,1,205,-203,1,554,-409,-142,1,1,1,1,77,90,-164,130,416,-415,62,...i2],
-  'windows-1251': [899,1,7191,-7111,7115,8,-6,1,139,-124,-7207,7216,-7215,2,-1,4,67,7110,1,3,1,5,-15,1,-8060,8330,-7369,7137,-7136,2,-1,4,-959,878,80,-86,-868,1004,-1002,1,858,-856,859,-857,1,1,1,857,-855,1,853,80,59,-988,1,1,922,7365,-7362,-921,925,-83,80,2,-71,...e(63)],
-  'windows-1252': [...p1,-7515,7530,-7888,7897,-7911,-197,240,-238,1,...w1,225,-6],
-  'windows-1253': [...p1,-8089,8104,-8102,8111,-8109,1,1,1,1,...w3,1,1,1,1,741,1,-739,1,1,1,1,1,1,r,2,1,1,1,8039,-8037,1,1,1,721,-719,1,1,...i7],
-  'windows-1254': [...p1,-7515,7530,-7888,7897,-7911,-197,1,1,1,...w1,1,218,-216,...e(47),79,-77,...e(11),84,46,-127,...e(16),48,-46,...e(11),53,46],
-  'windows-1255': [...p1,-7515,7530,-8102,8111,-8109,1,1,1,1,...w8,-7480,7750,-8328,8096,-8094,...e(7),8199,-8197,1,1,1,1,46,-44,...e(14),62,-60,1,1,1,1,1265,...e(19),45,1,1,1,1,...h(7),-36,...i8],
-  'windows-1256': [8237,-6702,6556,-7816,7820,8,-6,1,-7515,7530,-6583,6592,-7911,1332,18,-16,39,6505,1,3,1,5,-15,1,-6507,6777,-6801,6569,-7911,7865,1,-6483,-1562,1388,-1386,...e(7),1557,-1555,...e(14),1378,-1376,1,1,1,1377,162,-160,...e(21),-1375,1376,1,1,1,6,1,1,1,-1379,1380,-1378,1379,1,1,1,-1377,1,1,1,1,1374,1,-1372,1,1372,1,1,1,-1370,1371,1,-1369,1370,-1368,1369,-1367,1,7954,1,-6461],
-  'windows-1257': [...w0,-8102,8111,-8109,28,543,-527,-40,...w3,19,556,-572,1,r,2,1,1,r,2,1,49,-47,173,-171,1,1,1,24,-22,...e(5),...p3,347],
-  'windows-1258': [...p1,-7515,7530,-8102,8111,-7911,-197,1,1,1,...w8,-7480,7750,-8328,8096,-7911,-182,1,218,-216,...e(34),64,-62,...e(7),565,-563,1,1,65,-63,568,-566,1,204,-202,1,1,1,1,1,1,211,340,-548,1,1,1,33,-31,...e(7),534,-532,1,1,34,-32,562,-560,1,173,-171,1,1,1,1,1,1,180,7931],
-  'windows-874': [8237,-8235,1,1,1,8098,-8096,...e(10),...w8,-8060,...e(8),3425,...e(57),r,r,r,r,5,...e(28),r,r,r,r],
-  'x-mac-cyrillic': [913,...e(31),7153,-8048,992,-1005,4,8059,-8044,848,-856,-5,8313,-7456,80,7694,-7773,80,7627,-8557,8627,1,-7695,-929,988,-137,-4,80,-77,80,-78,80,-79,80,-2,-83,-857,...m0,875,80,-79,80,-7,7102,1,8,1,-5,1,-7970,7975,-7184,80,-79,80,7351,-7445,80,-2,-31,...e(30),7262]
+const maps = {
+  ibm866: [913,e(47),8530,1,1,-145,34,61,1,-12,-1,14,-18,6,6,-1,-1,-75,4,32,-8,-16,-28,60,34,1,-5,-6,21,-3,-6,-16,28,-5,1,-4,1,-12,-1,-6,1,24,-1,-82,-12,124,-4,8,4,-16,-8512,e(15),-78,80,-77,80,-77,80,-73,80,-942,8553,-8546,8547,-260,-8306,9468,-9472],
+  'koi8-r': [k8a,8450,e(14),-8544,8545,e(10),-9411,933,k8b,-28,k8b],
+  'koi8-u': [k8a,3,8448,-8446,1,8448,1,1,1,1,-8394,-51,8448,1,1,1,-8544,3,8543,-8541,1,8543,1,1,1,1,-8410,-130,-869,933,k8b,-28,k8b],
+  'x-mac-cyrillic': [913,e(31),7153,-8048,992,-1005,4,8059,-8044,848,-856,-5,8313,-7456,80,7694,-7773,80,7627,-8557,8627,1,-7695,-929,988,-137,-4,80,-77,80,-78,80,-79,80,-2,-83,-857,m0,875,80,-79,80,-7,7102,1,8,1,-5,1,-7970,7975,-7184,80,-79,80,7351,-7445,80,-2,-31,e(30),7262],
+  macintosh: [69,1,2,2,8,5,6,5,-1,2,2,-1,2,2,2,-1,2,1,2,-1,2,1,2,2,-1,2,2,-1,5,-1,2,1,7972,-8048,-14,1,4,8059,-8044,41,-49,-5,8313,-8302,-12,8632,-8602,18,8518,-8557,8627,1,-8640,16,8525,15,-2,-7759,7787,-8577,16,751,-707,18,-57,-30,11,m0,32,3,18,125,1,7872,1,8,1,-5,1,-7970,9427,-9419,121,7884,104,-115,1,56007,1,-56033,-8042,8035,4,18,-8046,8,-9,10,-3,5,1,1,-3,7,1,63531,-63533,8,1,-2,88,405,22,-557,553,1,1,-546,549,-2,-20],
+  'windows-874': [8237,-8235,1,1,1,8098,-8096,e(10),w8,-8060,e(8),iB],
 }
+
+// windows-1250 - windows-1258
+// prettier-ignore
+;[
+  [w0,-7888,7897,-7903,10,25,-4,-233,w8,-8060,8330,-8129,7897,-7903,10,25,-4,-218,551,17,-407,-157,96,-94,1,1,1,181,-179,1,1,1,205,-203,1,554,-409,-142,1,1,1,1,77,90,-164,130,416,-415,62,i2],
+  [899,1,7191,-7111,7115,8,-6,1,139,-124,-7207,7216,-7215,2,-1,4,67,7110,1,3,1,5,-15,1,-8060,8330,-7369,7137,-7136,2,-1,4,-959,878,80,-86,-868,1004,-1002,1,858,-856,859,-857,1,1,1,857,-855,1,853,80,59,-988,1,1,922,7365,-7362,-921,925,-83,80,2,-71,e(63)],
+  [p1,-7515,7530,-7888,7897,-7911,-197,240,-238,1,w1,225,-6],
+  [p1,-8089,8104,-8102,8111,-8109,1,1,1,1,w3,1,1,1,1,741,1,-739,e(6),r,2,1,1,1,8039,-8037,1,1,1,721,-719,1,1,i7],
+  [p1,-7515,7530,-7888,7897,-7911,-197,1,1,1,w1,1,218,-216,e(47),i9],
+  [p1,-7515,7530,-8102,8111,-8109,1,1,1,1,w8,-7480,7750,-8328,8096,-8094,e(7),8199,-8197,1,1,1,1,46,-44,e(14),62,-60,1,1,1,1,1265,e(19),45,1,1,1,1,h(7),-36,i8],
+  [8237,-6702,6556,-7816,7820,8,-6,1,-7515,7530,-6583,6592,-7911,1332,18,-16,39,6505,1,3,1,5,-15,1,-6507,6777,-6801,6569,-7911,7865,1,-6483,-1562,1388,-1386,e(7),1557,-1555,e(14),1378,-1376,1,1,1,1377,162,-160,e(21),-1375,1376,1,1,1,6,1,1,1,-1379,1380,-1378,1379,1,1,1,-1377,1,1,1,1,1374,1,-1372,1,1372,1,1,1,-1370,1371,1,-1369,1370,-1368,1369,-1367,1,7954,1,-6461],
+  [w0,-8102,8111,-8109,28,543,-527,-40,w3,19,556,-572,1,r,2,1,1,r,2,1,49,-47,173,-171,1,1,1,24,-22,e(5),p3,347],
+  [p1,-7515,7530,-8102,8111,-7911,-197,1,1,1,w8,-7480,7750,-8328,8096,-7911,-182,1,218,-216,e(34),64,-62,e(7),565,-563,1,1,65,-63,568,-566,1,204,-202,e(6),211,340,-548,1,1,1,33,-31,e(7),534,-532,1,1,34,-32,562,-560,1,173,-171,e(6),180,7931],
+].forEach((m, i) => {
+  maps[`windows-${i + 1250}`] = m
+});
+
+// iso-8859-1 - iso-8859-16
+// prettier-ignore
+;[
+  [], // Actual Latin1 / Unicode subset, non-WHATWG, which maps iso-8859-1 to windows-1252
+  [100,468,-407,-157,153,29,-179,1,184,-2,6,21,-204,208,-2,-203,85,470,-409,-142,138,29,364,-527,169,-2,6,21,355,-351,-2,i2],
+  [134,434,-565,1,r,128,-125,1,136,46,-64,22,-135,r,206,-203,119,-117,1,1,1,112,-110,1,121,46,-64,22,-120,r,191,-188,1,1,r,2,70,-2,-65,e(8),r,2,1,1,1,76,-74,1,69,-67,1,1,1,144,-16,-125,1,1,1,r,2,39,-2,-34,e(8),r,2,1,1,1,45,-43,1,38,-36,1,1,1,113,-16,380],
+  [100,52,30,-178,132,19,-148,1,184,-78,16,68,-185,208,-206,1,85,470,-388,-163,117,19,395,-527,169,-78,16,68,-29,52,-51,i4a,92,-26,53,7,-22,-98,1,1,1,1,154,-152,1,1,140,2,-139,i4b,61,-26,53,7,-22,-67,1,1,1,1,123,-121,1,1,109,2,366],
+  [865,e(11),-863,865,e(65),7367,-7365,e(11),-949,951,1],
+  [r,r,r,4,h(7),1384,-1375,h(13),1390,r,r,r,4,r,2,e(25),h(5),6,e(18),h(13)],
+  [8056,1,-8054,8201,3,-8201,1,1,1,721,-719,1,1,r,8040,-8037,1,1,1,721,1,1,-719,i7],
+  [r,2,e(7),46,-44,e(14),62,-60,1,1,1,h(32),8025,-6727,i8],
+  [e(47),i9], // non-WHATWG, which maps iso-8859-9 to windows-1254
+  [100,14,16,8,-2,14,-143,148,-43,80,6,23,-208,189,-32,-154,85,14,16,8,-2,14,-128,133,-43,80,6,23,7831,-7850,-32,i4a,1,1,117,7,-121,1,1,1,146,-144,154,-152,e(5),i4b,1,1,86,7,-90,1,1,1,115,-113,123,-121,1,1,1,1,58],
+  iB, // non-WHATWG, which maps iso-8859-11 to windows-874
+  null, // no 12
+  [8061,-8059,1,1,8058,-8056,1,49,-47,173,-171,1,1,1,24,-22,1,1,1,8041,-8039,p3,7835],
+  [7522,1,-7520,103,1,7423,-7523,7641,-7639,7641,-119,231,-7749,1,202,7334,1,-7423,1,7455,1,-7563,7584,43,-42,44,-35,147,-111,1,-36,-7585,e(15),165,-163,e(5),7572,-7570,e(5),153,-151,e(16),134,-132,e(5),7541,-7539,e(5),122],
+  [1,1,1,8201,-8199,187,-185,186,-184,e(10),202,-200,1,1,199,-197,1,1,151,1,37],
+  [100,1,60,8043,-142,-7870,-185,186,-184,367,-365,206,-204,205,1,-203,1,91,54,59,7840,-8039,1,199,-113,268,-350,151,1,37,4,-188,1,1,64,-62,66,-64,e(9),65,51,-113,1,1,124,-122,132,22,-151,1,1,1,60,258,-315,1,1,1,33,-31,35,-33,e(9),34,51,-82,1,1,93,-91,101,22,-120,1,1,1,29,258],
+].forEach((m, i) => {
+  if (m) maps[`iso-8859-${i + 1}`] = [e(33), m]
+})
+
+export default maps

package/fallback/single-byte.js

@@ -1,6 +1,6 @@
 import { asciiPrefix, decodeAscii, decodeLatin1 } from './latin1.js'
 import encodings from './single-byte.encodings.js'
-import { decode2string } from './_utils.js'
+import { decode2string, nativeDecoder } from './_utils.js'
 
 export const E_STRICT = 'Input is not well-formed for this encoding'
 const xUserDefined = 'x-user-defined'
@@ -18,11 +18,13 @@ function getEncoding(encoding) {
   if (encoding === xUserDefined) return Array.from({ length: 128 }, (_, i) => 0xf7_80 + i)
   if (encoding === iso8i) encoding = 'iso-8859-8'
   let prev = 127
-  return encodings[encoding].map((x) => (x === r ? x : (prev += x))) // eslint-disable-line no-return-assign
+  const enc = encodings[encoding].flat().flat().flat() // max depth is 3, rechecked by tests
+  return enc.map((x) => (x === r ? x : (prev += x))) // eslint-disable-line no-return-assign
 }
 
 const mappers = new Map()
 const decoders = new Map()
+const encmaps = new Map()
 
 // Used only on Node.js, no reason to optimize for anything else
 // E.g. avoiding .from and filling zero-initialized arr manually is faster on Hermes, but we avoid this codepath on Hermes completely
@@ -31,7 +33,7 @@ export function encodingMapper(encoding) {
   if (cached) return cached
 
   const codes = getEncoding(encoding)
-  const incomplete = codes.includes(0xff_fd)
+  const incomplete = codes.includes(r)
   let map
   const mapper = (arr, start = 0) => {
     if (!map) {
@@ -63,10 +65,12 @@ export function encodingMapper(encoding) {
 export function encodingDecoder(encoding) {
   const cached = decoders.get(encoding)
   if (cached) return cached
+  const isLatin1 = encoding === 'iso-8859-1'
+  if (isLatin1 && !nativeDecoder) return (arr, loose = false) => decodeLatin1(arr) // native decoder is faster for ascii below
 
   let strings
   const codes = getEncoding(encoding)
-  const incomplete = codes.includes(0xff_fd)
+  const incomplete = codes.includes(r)
   const decoder = (arr, loose = false) => {
     if (!strings) {
       const allCodes = Array.from({ length: 128 }, (_, i) => i).concat(codes)
@@ -76,6 +80,7 @@ export function encodingDecoder(encoding) {
 
     const prefixLen = asciiPrefix(arr)
     if (prefixLen === arr.length) return decodeAscii(arr)
+    if (isLatin1) return decodeLatin1(arr) // TODO: check if decodeAscii with subarray is faster for small prefixes too
     const prefix = decodeLatin1(arr, 0, prefixLen) // TODO: check if decodeAscii with subarray is faster for small prefixes too
     const suffix = decode2string(arr, prefix.length, arr.length, strings)
     if (!loose && incomplete && suffix.includes('\uFFFD')) throw new TypeError(E_STRICT)
@@ -85,3 +90,21 @@ export function encodingDecoder(encoding) {
   decoders.set(encoding, decoder)
   return decoder
 }
+
+export function encodeMap(encoding) {
+  const cached = encmaps.get(encoding)
+  if (cached) return cached
+
+  const codes = getEncoding(encoding)
+  let max = 128
+  while (codes.length < 128) codes.push(128 + codes.length)
+  for (const code of codes) if (code > max && code !== r) max = code
+  const map = new Uint8Array(max + 1) // < 10 KiB for all except macintosh, 63 KiB for macintosh
+  for (let i = 0; i < 128; i++) {
+    map[i] = i
+    if (codes[i] !== r) map[codes[i]] = 128 + i
+  }
+
+  encmaps.set(encoding, map)
+  return map
+}

package/fallback/utf8.js

@@ -1,3 +1,5 @@
+import { encodeAsciiPrefix } from './latin1.js'
+
 export const E_STRICT = 'Input is not well-formed utf8'
 export const E_STRICT_UNICODE = 'Input is not well-formed Unicode'
 
@@ -152,9 +154,10 @@ export function encode(string, loose) {
   const length = string.length
   let small = true
   let bytes = new Uint8Array(length) // assume ascii
-  let p = 0
 
-  for (let i = 0; i < length; i++) {
+  let i = encodeAsciiPrefix(bytes, string)
+  let p = i
+  for (; i < length; i++) {
     let code = string.charCodeAt(i)
     if (code < 0x80) {
       bytes[p++] = code

package/hex.node.js

@@ -1,5 +1,6 @@
 import { assertUint8 } from './assert.js'
 import { typedView } from './array.js'
+import { E_STRING } from './fallback/_utils.js'
 import { E_HEX } from './fallback/hex.js'
 
 if (Buffer.TYPED_ARRAY_SUPPORT) throw new Error('Unexpected Buffer polyfill')
@@ -19,7 +20,7 @@ export function toHex(arr) {
 export const fromHex = Uint8Array.fromHex
   ? (str, format = 'uint8') => typedView(Uint8Array.fromHex(str), format)
   : (str, format = 'uint8') => {
-      if (typeof str !== 'string') throw new TypeError('Input is not a string')
+      if (typeof str !== 'string') throw new TypeError(E_STRING)
       if (str.length % 2 !== 0) throw new SyntaxError(E_HEX)
       if (denoBug && /[^\dA-Fa-f]/.test(str)) throw new SyntaxError(E_HEX)
       const buf = Buffer.from(str, 'hex') // will stop on first non-hex character, so we can just validate length

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@exodus/bytes",
-  "version": "1.6.0",
+  "version": "1.8.0",
   "description": "Various operations on Uint8Array data",
   "scripts": {
     "lint": "eslint .",
@@ -72,7 +72,9 @@
     "/bech32.js",
     "/bigint.js",
     "/encoding.js",
+    "/encoding.d.ts",
     "/encoding-lite.js",
+    "/encoding-lite.d.ts",
     "/hex.js",
     "/hex.d.ts",
     "/hex.node.js",
@@ -117,8 +119,14 @@
       "node": "./single-byte.node.js",
       "default": "./single-byte.js"
     },
-    "./encoding.js": "./encoding.js",
-    "./encoding-lite.js": "./encoding-lite.js",
+    "./encoding.js": {
+      "types": "./encoding.d.ts",
+      "default": "./encoding.js"
+    },
+    "./encoding-lite.js": {
+      "types": "./encoding-lite.d.ts",
+      "default": "./encoding-lite.js"
+    },
     "./utf16.js": {
       "node": "./utf16.node.js",
       "default": "./utf16.js"
@@ -145,6 +153,7 @@
     "@exodus/prettier": "^1.0.0",
     "@exodus/test": "^1.0.0-rc.109",
     "@noble/hashes": "^2.0.1",
+    "@petamoriken/float16": "^3.9.3",
     "@scure/base": "^1.2.6",
     "@stablelib/base64": "^2.0.1",
     "@stablelib/hex": "^2.0.1",
@@ -172,7 +181,7 @@
     "typescript": "^5.9.3",
     "uint8array-tools": "^0.0.9",
     "utf8": "^3.0.0",
-    "whatwg-encoding": "^3.1.1",
+    "web-streams-polyfill": "^4.2.0",
     "wif": "^5.0.0"
   },
   "prettier": "@exodus/prettier",

package/single-byte.js

@@ -1,15 +1,17 @@
 import { assertUint8 } from './assert.js'
-import { canDecoders } from './fallback/_utils.js'
-import { assertEncoding, encodingDecoder } from './fallback/single-byte.js'
+import { canDecoders, nativeEncoder, isHermes, skipWeb, E_STRING } from './fallback/_utils.js'
+import { encodeAscii, encodeAsciiPrefix, encodeLatin1 } from './fallback/latin1.js'
+import { assertEncoding, encodingDecoder, encodeMap, E_STRICT } from './fallback/single-byte.js'
 
-const { TextDecoder } = globalThis
+const { TextDecoder, btoa } = globalThis
 
 let windows1252works
 
 // prettier-ignore
 const skipNative = new Set([
-  'iso-8859-16', // iso-8859-16 is somehow broken in WebKit, at least on CI
+  'iso-8859-1', 'iso-8859-9', 'iso-8859-11', // non-WHATWG
   'iso-8859-6', 'iso-8859-8', 'iso-8859-8-i', // slow in all 3 engines
+  'iso-8859-16', // iso-8859-16 is somehow broken in WebKit, at least on CI
 ])
 
 function shouldUseNative(enc) {
@@ -34,6 +36,7 @@ function shouldUseNative(enc) {
 }
 
 export function createSinglebyteDecoder(encoding, loose = false) {
+  if (typeof loose !== 'boolean') throw new TypeError('loose option should be boolean')
   assertEncoding(encoding)
 
   if (canDecoders && shouldUseNative(encoding)) {
@@ -56,4 +59,79 @@ export function createSinglebyteDecoder(encoding, loose = false) {
   }
 }
 
+const NON_LATIN = /[^\x00-\xFF]/ // eslint-disable-line no-control-regex
+
+function encode(s, m) {
+  const len = s.length
+  const x = new Uint8Array(len)
+  let i = nativeEncoder ? 0 : encodeAsciiPrefix(x, s)
+
+  if (!isHermes) {
+    for (const len3 = len - 3; i < len3; i += 4) {
+      const x0 = s.charCodeAt(i), x1 = s.charCodeAt(i + 1), x2 = s.charCodeAt(i + 2), x3 = s.charCodeAt(i + 3) // prettier-ignore
+      const c0 = m[x0], c1 = m[x1], c2 = m[x2], c3 = m[x3] // prettier-ignore
+      if ((!c0 && x0) || (!c1 && x1) || (!c2 && x2) || (!c3 && x3)) throw new TypeError(E_STRICT)
+
+      x[i] = c0
+      x[i + 1] = c1
+      x[i + 2] = c2
+      x[i + 3] = c3
+    }
+  }
+
+  for (; i < len; i++) {
+    const x0 = s.charCodeAt(i)
+    const c0 = m[x0]
+    if (!c0 && x0) return null
+    x[i] = c0
+  }
+
+  return x
+}
+
+// fromBase64+btoa path is faster on everything where fromBase64 is fast
+const useLatin1btoa = Uint8Array.fromBase64 && btoa && !skipWeb
+
+export function createSinglebyteEncoder(encoding, { mode = 'fatal' } = {}) {
+  // TODO: replacement, truncate (replacement will need varying length)
+  if (mode !== 'fatal') throw new Error('Unsupported mode')
+  const m = encodeMap(encoding) // asserts
+  const isLatin1 = encoding === 'iso-8859-1'
+
+  // No single-byte encoder produces surrogate pairs, so any surrogate is invalid
+  // This needs special treatment only to decide how many replacement chars to output, one or two
+  // Not much use in running isWellFormed, most likely cause of error is unmapped chars, not surrogate pairs
+  return (s) => {
+    if (typeof s !== 'string') throw new TypeError(E_STRING)
+    if (isLatin1) {
+      // max limit is to not produce base64 strings that are too long
+      if (useLatin1btoa && s.length >= 1024 && s.length < 1e8) {
+        try {
+          return Uint8Array.fromBase64(btoa(s)) // fails on non-latin1
+        } catch {
+          throw new TypeError(E_STRICT)
+        }
+      }
+
+      if (NON_LATIN.test(s)) throw new TypeError(E_STRICT)
+      return encodeLatin1(s)
+    }
+
+    // Instead of an ASCII regex check, encode optimistically - this is faster
+    // Check for 8-bit string with a regex though, this is instant on 8-bit strings so doesn't hurt the ASCII fast path
+    if (nativeEncoder && !NON_LATIN.test(s)) {
+      try {
+        return encodeAscii(s, E_STRICT)
+      } catch {}
+    }
+
+    const res = encode(s, m)
+    if (!res) throw new TypeError(E_STRICT)
+    return res
+  }
+}
+
+export const latin1toString = createSinglebyteDecoder('iso-8859-1')
+export const latin1fromString = createSinglebyteEncoder('iso-8859-1')
 export const windows1252toString = createSinglebyteDecoder('windows-1252')
+export const windows1252fromString = createSinglebyteEncoder('windows-1252')

package/single-byte.node.js

@@ -1,8 +1,8 @@
 import { assertUint8 } from './assert.js'
 import { isAscii } from 'node:buffer'
-import { isDeno, isLE, toBuf } from './fallback/_utils.js'
+import { isDeno, isLE, toBuf, E_STRING } from './fallback/_utils.js'
 import { asciiPrefix } from './fallback/latin1.js'
-import { encodingMapper, encodingDecoder, E_STRICT } from './fallback/single-byte.js'
+import { encodingMapper, encodingDecoder, encodeMap, E_STRICT } from './fallback/single-byte.js'
 
 function latin1Prefix(arr, start) {
   let p = start | 0
@@ -22,7 +22,7 @@ function latin1Prefix(arr, start) {
 }
 
 export function createSinglebyteDecoder(encoding, loose = false) {
-  const latin1path = encoding === 'windows-1252'
+  if (typeof loose !== 'boolean') throw new TypeError('loose option should be boolean')
   if (isDeno) {
     const jsDecoder = encodingDecoder(encoding) // asserts
     return (arr) => {
@@ -33,11 +33,13 @@ export function createSinglebyteDecoder(encoding, loose = false) {
     }
   }
 
+  const isLatin1 = encoding === 'iso-8859-1'
+  const latin1path = encoding === 'windows-1252'
   const { incomplete, mapper } = encodingMapper(encoding) // asserts
   return (arr) => {
     assertUint8(arr)
     if (arr.byteLength === 0) return ''
-    if (isAscii(arr)) return toBuf(arr).latin1Slice(0, arr.byteLength) // .latin1Slice is faster than .asciiSlice
+    if (isLatin1 || isAscii(arr)) return toBuf(arr).latin1Slice() // .latin1Slice is faster than .asciiSlice
 
     // Node.js TextDecoder is broken, so we can't use it. It's also slow anyway
 
@@ -57,4 +59,59 @@ export function createSinglebyteDecoder(encoding, loose = false) {
   }
 }
 
+const NON_LATIN = /[^\x00-\xFF]/ // eslint-disable-line no-control-regex
+
+export function createSinglebyteEncoder(encoding, { mode = 'fatal' } = {}) {
+  // TODO: replacement, truncate (replacement will need varying length)
+  if (mode !== 'fatal') throw new Error('Unsupported mode')
+  const m = encodeMap(encoding) // asserts
+  const isLatin1 = encoding === 'iso-8859-1'
+
+  return (s) => {
+    if (typeof s !== 'string') throw new TypeError(E_STRING)
+    if (isLatin1) {
+      if (NON_LATIN.test(s)) throw new TypeError(E_STRICT)
+      const b = Buffer.from(s, 'latin1')
+      return new Uint8Array(b.buffer, b.byteOffset, b.byteLength)
+    }
+
+    // Instead of an ASCII regex check, encode optimistically - this is faster
+    // Check for 8-bit string with a regex though, this is instant on 8-bit strings so doesn't hurt the ASCII fast path
+    if (!NON_LATIN.test(s)) {
+      const b = Buffer.from(s, 'utf8') // ascii/latin1 coerces, we need to check
+      if (b.length === s.length) return new Uint8Array(b.buffer, b.byteOffset, b.byteLength)
+    }
+
+    const len = s.length
+    let i = 0
+    const b = Buffer.from(s, 'utf-16le') // aligned
+    if (!isLE) b.swap16()
+    const x = new Uint16Array(b.buffer, b.byteOffset, b.byteLength / 2)
+    for (const len3 = len - 3; i < len3; i += 4) {
+      const x0 = x[i], x1 = x[i + 1], x2 = x[i + 2], x3 = x[i + 3] // prettier-ignore
+      const c0 = m[x0], c1 = m[x1], c2 = m[x2], c3 = m[x3] // prettier-ignore
+      if (!(c0 && c1 && c2 && c3) && ((!c0 && x0) || (!c1 && x1) || (!c2 && x2) || (!c3 && x3))) {
+        throw new TypeError(E_STRICT)
+      }
+
+      x[i] = c0
+      x[i + 1] = c1
+      x[i + 2] = c2
+      x[i + 3] = c3
+    }
+
+    for (; i < len; i++) {
+      const x0 = x[i]
+      const c0 = m[x0]
+      if (!c0 && x0) throw new TypeError(E_STRICT)
+      x[i] = c0
+    }
+
+    return new Uint8Array(x)
+  }
+}
+
+export const latin1toString = createSinglebyteDecoder('iso-8859-1')
+export const latin1fromString = createSinglebyteEncoder('iso-8859-1')
 export const windows1252toString = createSinglebyteDecoder('windows-1252')
+export const windows1252fromString = createSinglebyteEncoder('windows-1252')

package/utf16.js

@@ -1,5 +1,5 @@
 import * as js from './fallback/utf16.js'
-import { canDecoders, isLE } from './fallback/_utils.js'
+import { canDecoders, isLE, E_STRING } from './fallback/_utils.js'
 
 const { TextDecoder } = globalThis // Buffer is optional
 const ignoreBOM = true
@@ -18,7 +18,7 @@ const { E_STRICT, E_STRICT_UNICODE } = js
 const to8 = (a) => new Uint8Array(a.buffer, a.byteOffset, a.byteLength)
 
 function encode(str, loose = false, format = 'uint16') {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   if (format !== 'uint16' && format !== 'uint8-le' && format !== 'uint8-be') {
     throw new TypeError('Unknown format')
   }

package/utf16.node.js

@@ -1,4 +1,4 @@
-import { isDeno, isLE } from './fallback/_utils.js'
+import { isDeno, isLE, E_STRING } from './fallback/_utils.js'
 import { E_STRICT, E_STRICT_UNICODE } from './fallback/utf16.js'
 
 if (Buffer.TYPED_ARRAY_SUPPORT) throw new Error('Unexpected Buffer polyfill')
@@ -9,7 +9,7 @@ const to8 = (a) => new Uint8Array(a.buffer, a.byteOffset, a.byteLength)
 // Unlike utf8, operates on Uint16Arrays by default
 
 function encode(str, loose = false, format = 'uint16') {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   if (format !== 'uint16' && format !== 'uint8-le' && format !== 'uint8-be') {
     throw new TypeError('Unknown format')
   }

package/utf8.js

@@ -1,6 +1,6 @@
 import { assertUint8 } from './assert.js'
 import { typedView } from './array.js'
-import { isHermes, nativeDecoder, nativeEncoder } from './fallback/_utils.js'
+import { isHermes, nativeDecoder, nativeEncoder, E_STRING } from './fallback/_utils.js'
 import { asciiPrefix, decodeLatin1 } from './fallback/latin1.js'
 import * as js from './fallback/utf8.js'
 
@@ -44,7 +44,7 @@ function deLoose(str, loose, res) {
 }
 
 function encode(str, loose = false) {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   if (str.length === 0) return new Uint8Array() // faster than Uint8Array.of
   if (nativeEncoder) return deLoose(str, loose, nativeEncoder.encode(str))
   // No reason to use unescape + encodeURIComponent: it's slower than JS on normal engines, and modern Hermes already has TextEncoder

package/utf8.node.js

@@ -1,5 +1,6 @@
 import { assertUint8 } from './assert.js'
 import { typedView } from './array.js'
+import { E_STRING } from './fallback/_utils.js'
 import { E_STRICT, E_STRICT_UNICODE } from './fallback/utf8.js'
 import { isAscii } from 'node:buffer'
 
@@ -17,7 +18,7 @@ try {
 }
 
 function encode(str, loose = false) {
-  if (typeof str !== 'string') throw new TypeError('Input is not a string')
+  if (typeof str !== 'string') throw new TypeError(E_STRING)
   const strLength = str.length
   if (strLength === 0) return new Uint8Array() // faster than Uint8Array.of
   let res