@exodus/bytes 1.0.0-rc.8 → 1.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -0,0 +1,86 @@
1
+ import { asciiPrefix, decodeLatin1 } from './latin1.js'
2
+ import encodings from './single-byte.encodings.js'
3
+ import { decode2string } from './_utils.js'
4
+
5
+ export const E_STRICT = 'Input is not well-formed for this encoding'
6
+ const xUserDefined = 'x-user-defined'
7
+ const iso8i = 'iso-8859-8-i'
8
+
9
+ export const assertEncoding = (encoding) => {
10
+ if (Object.hasOwn(encodings, encoding) || encoding === xUserDefined || encoding === iso8i) return
11
+ throw new RangeError('Unsupported encoding')
12
+ }
13
+
14
+ const r = 0xff_fd
15
+
16
+ function getEncoding(encoding) {
17
+ assertEncoding(encoding)
18
+ if (encoding === xUserDefined) return Array.from({ length: 128 }, (_, i) => 0xf7_80 + i)
19
+ if (encoding === iso8i) encoding = 'iso-8859-8'
20
+ let prev = 127
21
+ return encodings[encoding].map((x) => (x === r ? x : (prev += x))) // eslint-disable-line no-return-assign
22
+ }
23
+
24
+ const mappers = new Map()
25
+ const decoders = new Map()
26
+
27
+ // Used only on Node.js, no reason to optimize for anything else
28
+ // E.g. avoiding .from and filling zero-initialized arr manually is faster on Hermes, but we avoid this codepath on Hermes completely
29
+ export function encodingMapper(encoding) {
30
+ const cached = mappers.get(encoding)
31
+ if (cached) return cached
32
+
33
+ const codes = getEncoding(encoding)
34
+ const incomplete = codes.includes(0xff_fd)
35
+ let map
36
+ const mapper = (arr, start = 0) => {
37
+ if (!map) {
38
+ map = new Uint16Array(256).map((_, i) => i) // Unicode subset
39
+ map.set(Uint16Array.from(codes), 128)
40
+ }
41
+
42
+ const o = Uint16Array.from(start === 0 ? arr : arr.subarray(start)) // copy to modify in-place, also those are 16-bit now
43
+ let i = 0
44
+ for (const end7 = o.length - 7; i < end7; i += 8) {
45
+ o[i] = map[o[i]]
46
+ o[i + 1] = map[o[i + 1]]
47
+ o[i + 2] = map[o[i + 2]]
48
+ o[i + 3] = map[o[i + 3]]
49
+ o[i + 4] = map[o[i + 4]]
50
+ o[i + 5] = map[o[i + 5]]
51
+ o[i + 6] = map[o[i + 6]]
52
+ o[i + 7] = map[o[i + 7]]
53
+ }
54
+
55
+ for (const end = o.length; i < end; i++) o[i] = map[o[i]]
56
+ return o
57
+ }
58
+
59
+ mappers.set(encoding, { mapper, incomplete })
60
+ return { mapper, incomplete }
61
+ }
62
+
63
+ export function encodingDecoder(encoding) {
64
+ const cached = decoders.get(encoding)
65
+ if (cached) return cached
66
+
67
+ let strings
68
+ const codes = getEncoding(encoding)
69
+ const incomplete = codes.includes(0xff_fd)
70
+ const decoder = (arr, loose = false) => {
71
+ if (!strings) {
72
+ const allCodes = Array.from({ length: 128 }, (_, i) => i).concat(codes)
73
+ while (allCodes.length < 256) allCodes.push(allCodes.length)
74
+ strings = allCodes.map((c) => String.fromCharCode(c))
75
+ }
76
+
77
+ const prefix = decodeLatin1(arr, 0, asciiPrefix(arr))
78
+ if (prefix.length === arr.length) return prefix
79
+ const suffix = decode2string(arr, prefix.length, arr.length, strings)
80
+ if (!loose && incomplete && suffix.includes('\uFFFD')) throw new TypeError(E_STRICT)
81
+ return prefix + suffix
82
+ }
83
+
84
+ decoders.set(encoding, decoder)
85
+ return decoder
86
+ }
@@ -0,0 +1,180 @@
1
+ import { decodeLatin1, encodeCharcodes } from './latin1.js'
2
+ import { isLE } from './_utils.js'
3
+
4
+ export const E_STRICT = 'Input is not well-formed utf16'
5
+ export const E_STRICT_UNICODE = 'Input is not well-formed Unicode'
6
+
7
+ const replacementCodepoint = 0xff_fd
8
+ const replacementCodepointSwapped = 0xfd_ff
9
+
10
+ const to16 = (a) => new Uint16Array(a.buffer, a.byteOffset, a.byteLength / 2) // Requires checked length and alignment!
11
+
12
+ export function to16input(u8, le) {
13
+ // Assume even number of bytes
14
+ if (le === isLE) return to16(u8.byteOffset % 2 === 0 ? u8 : Uint8Array.from(u8))
15
+
16
+ const res = new Uint8Array(u8.length)
17
+
18
+ let i = 0
19
+ for (const last3 = u8.length - 3; i < last3; i += 4) {
20
+ const x0 = u8[i]
21
+ const x1 = u8[i + 1]
22
+ const x2 = u8[i + 2]
23
+ const x3 = u8[i + 3]
24
+ res[i] = x1
25
+ res[i + 1] = x0
26
+ res[i + 2] = x3
27
+ res[i + 3] = x2
28
+ }
29
+
30
+ for (const last = u8.length - 1; i < last; i += 2) {
31
+ const x0 = u8[i]
32
+ const x1 = u8[i + 1]
33
+ res[i] = x1
34
+ res[i + 1] = x0
35
+ }
36
+
37
+ return to16(res)
38
+ }
39
+
40
+ export const decode = (u16, loose = false, checked = false) => {
41
+ if (checked || isWellFormed(u16)) return decodeLatin1(u16, 0, u16.length) // it's capable of decoding Uint16Array to UTF-16 as well
42
+ if (!loose) throw new TypeError(E_STRICT)
43
+ return decodeLatin1(toWellFormed(Uint16Array.from(u16)), 0, u16.length) // cloned for replacement
44
+ }
45
+
46
+ export function encode(str, loose = false, checked = false, swapped = false) {
47
+ const arr = new Uint16Array(str.length)
48
+ if (checked) return swapped ? encodeCheckedSwapped(str, arr) : encodeChecked(str, arr)
49
+ return swapped ? encodeUncheckedSwapped(str, arr, loose) : encodeUnchecked(str, arr, loose)
50
+ }
51
+
52
+ // Splitting paths into small functions helps (at least on SpiderMonkey)
53
+ /* eslint-disable @exodus/mutable/no-param-reassign-prop-only */
54
+
55
+ const encodeChecked = (str, arr) => encodeCharcodes(str, arr) // Same as encodeLatin1, but with Uint16Array
56
+
57
+ function encodeCheckedSwapped(str, arr) {
58
+ // TODO: faster path for Hermes? See encodeCharcodes
59
+ const length = str.length
60
+ for (let i = 0; i < length; i++) {
61
+ const x = str.charCodeAt(i)
62
+ arr[i] = ((x & 0xff) << 8) | (x >> 8)
63
+ }
64
+
65
+ return arr
66
+ }
67
+
68
+ // lead: d800 - dbff, trail: dc00 - dfff
69
+
70
+ function encodeUnchecked(str, arr, loose = false) {
71
+ // TODO: faster path for Hermes? See encodeCharcodes
72
+ const length = str.length
73
+ for (let i = 0; i < length; i++) {
74
+ const code = str.charCodeAt(i)
75
+ arr[i] = code
76
+ if (code >= 0xd8_00 && code < 0xe0_00) {
77
+ // An unexpected trail or a lead at the very end of input
78
+ if (code > 0xdb_ff || i + 1 >= length) {
79
+ if (!loose) throw new TypeError(E_STRICT_UNICODE)
80
+ arr[i] = replacementCodepoint
81
+ } else {
82
+ const next = str.charCodeAt(i + 1) // Process valid pairs immediately
83
+ if (next < 0xdc_00 || next >= 0xe0_00) {
84
+ if (!loose) throw new TypeError(E_STRICT_UNICODE)
85
+ arr[i] = replacementCodepoint
86
+ } else {
87
+ i++ // consume next
88
+ arr[i] = next
89
+ }
90
+ }
91
+ }
92
+ }
93
+
94
+ return arr
95
+ }
96
+
97
+ function encodeUncheckedSwapped(str, arr, loose = false) {
98
+ // TODO: faster path for Hermes? See encodeCharcodes
99
+ const length = str.length
100
+ for (let i = 0; i < length; i++) {
101
+ const code = str.charCodeAt(i)
102
+ arr[i] = ((code & 0xff) << 8) | (code >> 8)
103
+ if (code >= 0xd8_00 && code < 0xe0_00) {
104
+ // An unexpected trail or a lead at the very end of input
105
+ if (code > 0xdb_ff || i + 1 >= length) {
106
+ if (!loose) throw new TypeError(E_STRICT_UNICODE)
107
+ arr[i] = replacementCodepointSwapped
108
+ } else {
109
+ const next = str.charCodeAt(i + 1) // Process valid pairs immediately
110
+ if (next < 0xdc_00 || next >= 0xe0_00) {
111
+ if (!loose) throw new TypeError(E_STRICT_UNICODE)
112
+ arr[i] = replacementCodepointSwapped
113
+ } else {
114
+ i++ // consume next
115
+ arr[i] = ((next & 0xff) << 8) | (next >> 8)
116
+ }
117
+ }
118
+ }
119
+ }
120
+
121
+ return arr
122
+ }
123
+
124
+ export function toWellFormed(u16) {
125
+ const length = u16.length
126
+ for (let i = 0; i < length; i++) {
127
+ const code = u16[i]
128
+ if (code >= 0xd8_00 && code < 0xe0_00) {
129
+ // An unexpected trail or a lead at the very end of input
130
+ if (code > 0xdb_ff || i + 1 >= length) {
131
+ u16[i] = replacementCodepoint
132
+ } else {
133
+ const next = u16[i + 1] // Process valid pairs immediately
134
+ if (next < 0xdc_00 || next >= 0xe0_00) {
135
+ u16[i] = replacementCodepoint
136
+ } else {
137
+ i++ // consume next
138
+ }
139
+ }
140
+ }
141
+ }
142
+
143
+ return u16
144
+ }
145
+
146
+ export function isWellFormed(u16) {
147
+ const length = u16.length
148
+ let i = 0
149
+
150
+ // Speedup with u32, by skipping to the first surrogate
151
+ // Only implemented for aligned input for now, but almost all input is aligned (pooled Buffer or 0 offset)
152
+ if (length > 32 && u16.byteOffset % 4 === 0) {
153
+ const u32length = (u16.byteLength / 4) | 0
154
+ const u32 = new Uint32Array(u16.buffer, u16.byteOffset, u32length)
155
+ for (const last3 = u32length - 3; ; i += 4) {
156
+ if (i >= last3) break // loop is fast enough for moving this here to be _very_ useful, likely due to array access checks
157
+ const a = u32[i]
158
+ const b = u32[i + 1]
159
+ const c = u32[i + 2]
160
+ const d = u32[i + 3]
161
+ if (a & 0x80_00_80_00 || b & 0x80_00_80_00 || c & 0x80_00_80_00 || d & 0x80_00_80_00) break
162
+ }
163
+
164
+ for (; i < u32length; i++) if (u32[i] & 0x80_00_80_00) break
165
+ i *= 2
166
+ }
167
+
168
+ for (; i < length; i++) {
169
+ const code = u16[i]
170
+ if (code >= 0xd8_00 && code < 0xe0_00) {
171
+ // An unexpected trail or a lead at the very end of input
172
+ if (code > 0xdb_ff || i + 1 >= length) return false
173
+ i++ // consume next
174
+ const next = u16[i] // Process valid pairs immediately
175
+ if (next < 0xdc_00 || next >= 0xe0_00) return false
176
+ }
177
+ }
178
+
179
+ return true
180
+ }
package/hex.d.ts ADDED
@@ -0,0 +1,22 @@
1
+ /// <reference types="node" />
2
+
3
+ import type { OutputFormat, Uint8ArrayBuffer } from './array.js';
4
+
5
+ /**
6
+ * Encodes a Uint8Array to a lowercase hex string
7
+ * @param arr - The input bytes
8
+ * @returns The hex encoded string
9
+ */
10
+ export function toHex(arr: Uint8ArrayBuffer): string;
11
+
12
+ /**
13
+ * Decodes a hex string to bytes
14
+ * Unlike Buffer.from(), throws on invalid input
15
+ * @param str - The hex encoded string (case-insensitive)
16
+ * @param format - Output format (default: 'uint8')
17
+ * @returns The decoded bytes
18
+ */
19
+ export function fromHex(str: string, format?: 'uint8'): Uint8ArrayBuffer;
20
+ export function fromHex(str: string, format: 'buffer'): Buffer;
21
+ export function fromHex(str: string, format?: OutputFormat): Uint8ArrayBuffer | Buffer;
22
+
package/hex.node.js CHANGED
@@ -5,6 +5,7 @@ import { E_HEX } from './fallback/hex.js'
5
5
  if (Buffer.TYPED_ARRAY_SUPPORT) throw new Error('Unexpected Buffer polyfill')
6
6
 
7
7
  const { toHex: webHex } = Uint8Array.prototype // Modern engines have this
8
+ const denoBug = Buffer.from('ag', 'hex').length > 0
8
9
 
9
10
  export function toHex(arr) {
10
11
  assertUint8(arr)
@@ -20,6 +21,7 @@ export const fromHex = Uint8Array.fromHex
20
21
  : (str, format = 'uint8') => {
21
22
  if (typeof str !== 'string') throw new TypeError('Input is not a string')
22
23
  if (str.length % 2 !== 0) throw new SyntaxError(E_HEX)
24
+ if (denoBug && /[^\dA-Fa-f]/.test(str)) throw new SyntaxError(E_HEX)
23
25
  const buf = Buffer.from(str, 'hex') // will stop on first non-hex character, so we can just validate length
24
26
  if (buf.length * 2 !== str.length) throw new SyntaxError(E_HEX)
25
27
  return typedView(buf, format)
package/multi-byte.js ADDED
@@ -0,0 +1,13 @@
1
+ import { assertUint8 } from './assert.js'
2
+ import { multibyteDecoder } from './fallback/multi-byte.js'
3
+
4
+ export function createMultibyteDecoder(encoding, loose = false) {
5
+ const jsDecoder = multibyteDecoder(encoding, loose) // asserts
6
+ let streaming = false
7
+ return (arr, stream = false) => {
8
+ assertUint8(arr)
9
+ if (!streaming && arr.byteLength === 0) return ''
10
+ streaming = stream
11
+ return jsDecoder(arr, stream)
12
+ }
13
+ }
@@ -0,0 +1,25 @@
1
+ import { assertUint8 } from './assert.js'
2
+ import { isDeno } from './fallback/_utils.js'
3
+ import { isAsciiSuperset, multibyteDecoder } from './fallback/multi-byte.js'
4
+ import { isAscii } from 'node:buffer'
5
+
6
+ const toBuf = (x) => Buffer.from(x.buffer, x.byteOffset, x.byteLength)
7
+
8
+ export function createMultibyteDecoder(encoding, loose = false) {
9
+ const jsDecoder = multibyteDecoder(encoding, loose) // asserts
10
+ let streaming = false
11
+ const asciiSuperset = isAsciiSuperset(encoding)
12
+ return (arr, stream = false) => {
13
+ assertUint8(arr)
14
+ if (!streaming) {
15
+ if (arr.byteLength === 0) return ''
16
+ if (asciiSuperset && isAscii(arr)) {
17
+ if (isDeno) return toBuf(arr).toString()
18
+ return toBuf(arr).latin1Slice(0, arr.byteLength) // .latin1Slice is faster than .asciiSlice
19
+ }
20
+ }
21
+
22
+ streaming = stream
23
+ return jsDecoder(arr, stream)
24
+ }
25
+ }
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@exodus/bytes",
3
- "version": "1.0.0-rc.8",
3
+ "version": "1.0.0",
4
4
  "description": "Various operations on Uint8Array data",
5
5
  "scripts": {
6
6
  "lint": "eslint .",
@@ -10,6 +10,10 @@
10
10
  "test:spidermonkey": "exodus-test --engine=spidermonkey:bundle",
11
11
  "test:hermes": "exodus-test --engine=hermes:bundle",
12
12
  "test:quickjs": "exodus-test --engine=quickjs:bundle",
13
+ "test:xs": "exodus-test --engine=xs:bundle",
14
+ "test:engine262": "exodus-test --engine=engine262:bundle",
15
+ "test:deno": "exodus-test --engine=deno:pure",
16
+ "test:bun": "exodus-test --engine=bun:pure",
13
17
  "test:electron:bundle": "exodus-test --engine=electron:bundle",
14
18
  "test:electron:as-node": "exodus-test --engine=electron-as-node:test",
15
19
  "test:chrome:puppeteer": "exodus-test --engine=chrome:puppeteer",
@@ -17,6 +21,7 @@
17
21
  "test:webkit:playwright": "exodus-test --engine=webkit:playwright",
18
22
  "test:firefox:puppeteer": "exodus-test --engine=firefox:puppeteer",
19
23
  "test:firefox:playwright": "exodus-test --engine=firefox:playwright",
24
+ "test:servo:bundle": "exodus-test --engine=servo:bundle",
20
25
  "test": "exodus-test",
21
26
  "jsvu": "jsvu",
22
27
  "playwright": "exodus-test --playwright",
@@ -25,54 +30,96 @@
25
30
  },
26
31
  "repository": {
27
32
  "type": "git",
28
- "url": "git+https://github.com/ExodusMovement/bytes.git"
33
+ "url": "git+https://github.com/ExodusOSS/bytes.git"
29
34
  },
30
35
  "author": "Exodus Movement, Inc.",
31
36
  "license": "MIT",
32
37
  "bugs": {
33
- "url": "https://github.com/ExodusMovement/bytes/issues"
38
+ "url": "https://github.com/ExodusOSS/bytes/issues"
34
39
  },
35
- "homepage": "https://github.com/ExodusMovement/bytes#readme",
40
+ "homepage": "https://github.com/ExodusOSS/bytes#readme",
36
41
  "engines": {
37
- "node": "^20.19.0 || >=22.13.0"
42
+ "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
38
43
  },
39
44
  "type": "module",
40
45
  "files": [
41
46
  "/fallback/_utils.js",
42
- "/fallback/latin1.js",
43
47
  "/fallback/base32.js",
44
48
  "/fallback/base64.js",
49
+ "/fallback/encoding.js",
50
+ "/fallback/encoding.labels.js",
51
+ "/fallback/encoding.util.js",
45
52
  "/fallback/hex.js",
53
+ "/fallback/latin1.js",
54
+ "/fallback/multi-byte.encodings.cjs",
55
+ "/fallback/multi-byte.encodings.json",
56
+ "/fallback/multi-byte.js",
57
+ "/fallback/multi-byte.table.js",
58
+ "/fallback/single-byte.encodings.js",
59
+ "/fallback/single-byte.js",
60
+ "/fallback/utf16.js",
46
61
  "/fallback/utf8.js",
47
62
  "/array.js",
63
+ "/array.d.ts",
48
64
  "/assert.js",
49
65
  "/base32.js",
50
66
  "/base58.js",
51
67
  "/base58check.js",
52
68
  "/base64.js",
69
+ "/base64.d.ts",
53
70
  "/bech32.js",
54
- "/hex.node.js",
71
+ "/bigint.js",
72
+ "/encoding.js",
73
+ "/encoding-lite.js",
55
74
  "/hex.js",
56
- "/utf8.node.js",
75
+ "/hex.d.ts",
76
+ "/hex.node.js",
77
+ "/multi-byte.js",
78
+ "/multi-byte.node.js",
79
+ "/single-byte.js",
80
+ "/single-byte.node.js",
81
+ "/utf16.js",
82
+ "/utf16.node.js",
57
83
  "/utf8.js",
84
+ "/utf8.d.ts",
85
+ "/utf8.node.js",
58
86
  "/wif.js"
59
87
  ],
60
88
  "exports": {
61
- "./array.js": "./array.js",
89
+ "./array.js": {
90
+ "types": "./array.d.ts",
91
+ "default": "./array.js"
92
+ },
62
93
  "./base32.js": "./base32.js",
63
94
  "./base58.js": "./base58.js",
64
95
  "./base58check.js": "./base58check.js",
65
- "./base64.js": "./base64.js",
96
+ "./base64.js": {
97
+ "types": "./base64.d.ts",
98
+ "default": "./base64.js"
99
+ },
66
100
  "./bech32.js": "./bech32.js",
101
+ "./bigint.js": "./bigint.js",
67
102
  "./hex.js": {
103
+ "types": "./hex.d.ts",
68
104
  "node": "./hex.node.js",
69
105
  "default": "./hex.js"
70
106
  },
107
+ "./multi-byte.js": {
108
+ "node": "./multi-byte.node.js",
109
+ "default": "./multi-byte.js"
110
+ },
111
+ "./single-byte.js": {
112
+ "node": "./single-byte.node.js",
113
+ "default": "./single-byte.js"
114
+ },
115
+ "./encoding.js": "./encoding.js",
116
+ "./encoding-lite.js": "./encoding-lite.js",
71
117
  "./utf16.js": {
72
118
  "node": "./utf16.node.js",
73
119
  "default": "./utf16.js"
74
120
  },
75
121
  "./utf8.js": {
122
+ "types": "./utf8.d.ts",
76
123
  "node": "./utf8.node.js",
77
124
  "default": "./utf8.js"
78
125
  },
@@ -91,14 +138,15 @@
91
138
  "@exodus/crypto": "^1.0.0-rc.30",
92
139
  "@exodus/eslint-config": "^5.24.0",
93
140
  "@exodus/prettier": "^1.0.0",
94
- "@exodus/test": "^1.0.0-rc.108",
141
+ "@exodus/test": "^1.0.0-rc.109",
95
142
  "@noble/hashes": "^2.0.1",
96
143
  "@scure/base": "^1.2.6",
97
144
  "@stablelib/base64": "^2.0.1",
98
145
  "@stablelib/hex": "^2.0.1",
99
- "@types/node": "^22.13.0",
146
+ "@types/node": "^22.12.0",
100
147
  "base-x": "^5.0.1",
101
148
  "base32.js": "^0.1.0",
149
+ "base58-js": "^3.0.3",
102
150
  "base64-js": "^1.5.1",
103
151
  "bech32": "^2.0.0",
104
152
  "bs58": "^6.0.0",
@@ -114,11 +162,12 @@
114
162
  "hextreme": "^1.0.7",
115
163
  "hi-base32": "^0.5.1",
116
164
  "iconv-lite": "^0.7.0",
117
- "jsvu": "^3.0.0",
165
+ "jsvu": "^3.0.3",
118
166
  "text-encoding": "^0.7.0",
119
167
  "typescript": "^5.9.3",
120
168
  "uint8array-tools": "^0.0.9",
121
169
  "utf8": "^3.0.0",
170
+ "whatwg-encoding": "^3.1.1",
122
171
  "wif": "^5.0.0"
123
172
  },
124
173
  "prettier": "@exodus/prettier",
package/single-byte.js ADDED
@@ -0,0 +1,54 @@
1
+ import { assertUint8 } from './assert.js'
2
+ import { canDecoders } from './fallback/_utils.js'
3
+ import { assertEncoding, encodingDecoder } from './fallback/single-byte.js'
4
+
5
+ const { TextDecoder } = globalThis
6
+
7
+ let windows1252works
8
+
9
+ function shouldUseNative(enc) {
10
+ // https://issues.chromium.org/issues/468458388
11
+ // Also might be incorrectly imlemented on platforms as Latin1 (e.g. in Node.js) or regress
12
+ // This is the most significant single-byte encoding, 'ascii' and 'latin1' alias to this
13
+ // Even after Chrome bug is fixed, this should serve as a quick correctness check that it's actually windows-1252
14
+ if (enc === 'windows-1252') {
15
+ if (windows1252works === undefined) {
16
+ windows1252works = false
17
+ try {
18
+ const u = new Uint8Array(9) // using 9 bytes is significant to catch the bug
19
+ u[8] = 128
20
+ windows1252works = new TextDecoder(enc).decode(u).codePointAt(8) === 0x20_ac
21
+ } catch {}
22
+ }
23
+
24
+ return windows1252works
25
+ }
26
+
27
+ // iso-8859-16 is somehow broken in WebKit, at least on CI
28
+ return enc !== 'iso-8859-16'
29
+ }
30
+
31
+ export function createSinglebyteDecoder(encoding, loose = false) {
32
+ assertEncoding(encoding)
33
+
34
+ if (canDecoders && shouldUseNative(encoding)) {
35
+ // In try, as not all encodings might be implemented in all engines which have native TextDecoder
36
+ try {
37
+ const decoder = new TextDecoder(encoding, { fatal: !loose })
38
+ return (arr) => {
39
+ assertUint8(arr)
40
+ if (arr.byteLength === 0) return ''
41
+ return decoder.decode(arr)
42
+ }
43
+ } catch {}
44
+ }
45
+
46
+ const jsDecoder = encodingDecoder(encoding)
47
+ return (arr) => {
48
+ assertUint8(arr)
49
+ if (arr.byteLength === 0) return ''
50
+ return jsDecoder(arr, loose)
51
+ }
52
+ }
53
+
54
+ export const windows1252toString = createSinglebyteDecoder('windows-1252')
@@ -0,0 +1,62 @@
1
+ import { assertUint8 } from './assert.js'
2
+ import { isAscii } from 'node:buffer'
3
+ import { isDeno, isLE } from './fallback/_utils.js'
4
+ import { asciiPrefix } from './fallback/latin1.js'
5
+ import { encodingMapper, encodingDecoder, E_STRICT } from './fallback/single-byte.js'
6
+
7
+ const toBuf = (x) => Buffer.from(x.buffer, x.byteOffset, x.byteLength)
8
+
9
+ function latin1Prefix(arr, start) {
10
+ let p = start | 0
11
+ const length = arr.length
12
+ for (const len3 = length - 3; p < len3; p += 4) {
13
+ if ((arr[p] & 0xe0) === 0x80) return p
14
+ if ((arr[p + 1] & 0xe0) === 0x80) return p + 1
15
+ if ((arr[p + 2] & 0xe0) === 0x80) return p + 2
16
+ if ((arr[p + 3] & 0xe0) === 0x80) return p + 3
17
+ }
18
+
19
+ for (; p < length; p++) {
20
+ if ((arr[p] & 0xe0) === 0x80) return p
21
+ }
22
+
23
+ return length
24
+ }
25
+
26
+ export function createSinglebyteDecoder(encoding, loose = false) {
27
+ const latin1path = encoding === 'windows-1252'
28
+ if (isDeno) {
29
+ const jsDecoder = encodingDecoder(encoding) // asserts
30
+ return (arr) => {
31
+ assertUint8(arr)
32
+ if (arr.byteLength === 0) return ''
33
+ if (isAscii(arr)) return toBuf(arr).toString()
34
+ return jsDecoder(arr, loose) // somewhy faster on Deno anyway, TODO: optimize?
35
+ }
36
+ }
37
+
38
+ const { incomplete, mapper } = encodingMapper(encoding) // asserts
39
+ return (arr) => {
40
+ assertUint8(arr)
41
+ if (arr.byteLength === 0) return ''
42
+ if (isAscii(arr)) return toBuf(arr).latin1Slice(0, arr.byteLength) // .latin1Slice is faster than .asciiSlice
43
+
44
+ // Node.js TextDecoder is broken, so we can't use it. It's also slow anyway
45
+
46
+ let prefixBytes = asciiPrefix(arr)
47
+ let prefix = ''
48
+ if (latin1path) prefixBytes = latin1Prefix(arr, prefixBytes)
49
+ if (prefixBytes > 64 || prefixBytes === arr.length) {
50
+ prefix = toBuf(arr).latin1Slice(0, prefixBytes) // .latin1Slice is faster than .asciiSlice
51
+ if (prefixBytes === arr.length) return prefix
52
+ }
53
+
54
+ const b = toBuf(mapper(arr, prefix.length)) // prefix.length can mismatch prefixBytes
55
+ if (!isLE) b.swap16()
56
+ const suffix = b.ucs2Slice(0, b.byteLength)
57
+ if (!loose && incomplete && suffix.includes('\uFFFD')) throw new TypeError(E_STRICT)
58
+ return prefix + suffix
59
+ }
60
+ }
61
+
62
+ export const windows1252toString = createSinglebyteDecoder('windows-1252')