mirror of
https://github.com/zebrajr/node.git
synced 2026-01-15 12:15:26 +00:00
lib: implement all 1-byte encodings in js
PR-URL: https://github.com/nodejs/node/pull/61093 Reviewed-By: Matteo Collina <matteo.collina@gmail.com> Reviewed-By: René <contact.9a5d6388@renegade334.me.uk>
This commit is contained in:
Nikita Skovoroda
@@ -13,24 +13,27 @@ const {
|
||||
StringPrototypeSlice,
|
||||
Symbol,
|
||||
SymbolToStringTag,
|
||||
Uint8Array,
|
||||
} = primordials;
|
||||
|
||||
const { FastBuffer } = require('internal/buffer');
|
||||
|
||||
const {
|
||||
ERR_ENCODING_NOT_SUPPORTED,
|
||||
ERR_INVALID_ARG_TYPE,
|
||||
ERR_INVALID_THIS,
|
||||
ERR_NO_ICU,
|
||||
} = require('internal/errors').codes;
|
||||
const kMethod = Symbol('method');
|
||||
const kHandle = Symbol('handle');
|
||||
const kFlags = Symbol('flags');
|
||||
const kEncoding = Symbol('encoding');
|
||||
const kDecoder = Symbol('decoder');
|
||||
const kFatal = Symbol('kFatal');
|
||||
const kUTF8FastPath = Symbol('kUTF8FastPath');
|
||||
const kWindows1252FastPath = Symbol('kWindows1252FastPath');
|
||||
const kIgnoreBOM = Symbol('kIgnoreBOM');
|
||||
|
||||
const { isSinglebyteEncoding, createSinglebyteDecoder } = require('internal/encoding/single-byte');
|
||||
|
||||
const {
|
||||
getConstructorOf,
|
||||
customInspectSymbol: inspect,
|
||||
@@ -55,11 +58,8 @@ const {
|
||||
encodeIntoResults,
|
||||
encodeUtf8String,
|
||||
decodeUTF8,
|
||||
decodeWindows1252,
|
||||
} = binding;
|
||||
|
||||
const { Buffer } = require('buffer');
|
||||
|
||||
function validateDecoder(obj) {
|
||||
if (obj == null || obj[kDecoder] !== true)
|
||||
throw new ERR_INVALID_THIS('TextDecoder');
|
||||
@@ -69,7 +69,7 @@ const CONVERTER_FLAGS_FLUSH = 0x1;
|
||||
const CONVERTER_FLAGS_FATAL = 0x2;
|
||||
const CONVERTER_FLAGS_IGNORE_BOM = 0x4;
|
||||
|
||||
const empty = new Uint8Array(0);
|
||||
const empty = new FastBuffer();
|
||||
|
||||
const encodings = new SafeMap([
|
||||
['unicode-1-1-utf-8', 'utf-8'],
|
||||
@@ -387,6 +387,24 @@ ObjectDefineProperties(
|
||||
[SymbolToStringTag]: { __proto__: null, configurable: true, value: 'TextEncoder' },
|
||||
});
|
||||
|
||||
function parseInput(input) {
|
||||
if (isAnyArrayBuffer(input)) {
|
||||
try {
|
||||
return new FastBuffer(input);
|
||||
} catch {
|
||||
return empty;
|
||||
}
|
||||
} else if (isArrayBufferView(input)) {
|
||||
try {
|
||||
return new FastBuffer(input.buffer, input.byteOffset, input.byteLength);
|
||||
} catch {
|
||||
return empty;
|
||||
}
|
||||
} else {
|
||||
throw new ERR_INVALID_ARG_TYPE('input', ['ArrayBuffer', 'ArrayBufferView'], input);
|
||||
}
|
||||
}
|
||||
|
||||
const TextDecoder =
|
||||
internalBinding('config').hasIntl ?
|
||||
makeTextDecoderICU() :
|
||||
@@ -420,10 +438,12 @@ function makeTextDecoderICU() {
|
||||
this[kFatal] = Boolean(options?.fatal);
|
||||
// Only support fast path for UTF-8.
|
||||
this[kUTF8FastPath] = enc === 'utf-8';
|
||||
this[kWindows1252FastPath] = enc === 'windows-1252';
|
||||
this[kHandle] = undefined;
|
||||
this[kMethod] = undefined;
|
||||
|
||||
if (!this[kUTF8FastPath] && !this[kWindows1252FastPath]) {
|
||||
if (isSinglebyteEncoding(this.encoding)) {
|
||||
this[kMethod] = createSinglebyteDecoder(this.encoding, this[kFatal]);
|
||||
} else if (!this[kUTF8FastPath]) {
|
||||
this.#prepareConverter();
|
||||
}
|
||||
}
|
||||
@@ -440,22 +460,18 @@ function makeTextDecoderICU() {
|
||||
|
||||
decode(input = empty, options = kEmptyObject) {
|
||||
validateDecoder(this);
|
||||
validateObject(options, 'options', kValidateObjectAllowObjectsAndNull);
|
||||
|
||||
if (this[kMethod]) return this[kMethod](parseInput(input));
|
||||
|
||||
this[kUTF8FastPath] &&= !(options?.stream);
|
||||
this[kWindows1252FastPath] &&= !(options?.stream);
|
||||
|
||||
if (this[kUTF8FastPath]) {
|
||||
return decodeUTF8(input, this[kIgnoreBOM], this[kFatal]);
|
||||
}
|
||||
|
||||
if (this[kWindows1252FastPath]) {
|
||||
return decodeWindows1252(input, this[kIgnoreBOM], this[kFatal]);
|
||||
}
|
||||
|
||||
this.#prepareConverter();
|
||||
|
||||
validateObject(options, 'options', kValidateObjectAllowObjectsAndNull);
|
||||
|
||||
let flags = 0;
|
||||
if (options !== null)
|
||||
flags |= options.stream ? 0 : CONVERTER_FLAGS_FLUSH;
|
||||
@@ -487,47 +503,40 @@ function makeTextDecoderJS() {
|
||||
validateObject(options, 'options', kValidateObjectAllowObjectsAndNull);
|
||||
|
||||
const enc = getEncodingFromLabel(encoding);
|
||||
if (enc === undefined || !hasConverter(enc))
|
||||
if (enc === undefined)
|
||||
throw new ERR_ENCODING_NOT_SUPPORTED(encoding);
|
||||
|
||||
let flags = 0;
|
||||
if (options !== null) {
|
||||
if (options.fatal) {
|
||||
throw new ERR_NO_ICU('"fatal" option');
|
||||
}
|
||||
flags |= options.fatal ? CONVERTER_FLAGS_FATAL : 0;
|
||||
flags |= options.ignoreBOM ? CONVERTER_FLAGS_IGNORE_BOM : 0;
|
||||
}
|
||||
|
||||
this[kDecoder] = true;
|
||||
// StringDecoder will normalize WHATWG encoding to Node.js encoding.
|
||||
this[kHandle] = new (lazyStringDecoder())(enc);
|
||||
this[kFlags] = flags;
|
||||
this[kEncoding] = enc;
|
||||
this[kIgnoreBOM] = Boolean(options?.ignoreBOM);
|
||||
this[kFatal] = Boolean(options?.fatal);
|
||||
this[kBOMSeen] = false;
|
||||
this[kMethod] = undefined;
|
||||
|
||||
if (isSinglebyteEncoding(enc)) {
|
||||
this[kMethod] = createSinglebyteDecoder(enc, this[kFatal]);
|
||||
} else {
|
||||
if (!hasConverter(enc)) throw new ERR_ENCODING_NOT_SUPPORTED(encoding);
|
||||
if (this[kFatal]) throw new ERR_NO_ICU('"fatal" option');
|
||||
// StringDecoder will normalize WHATWG encoding to Node.js encoding.
|
||||
this[kHandle] = new (lazyStringDecoder())(enc);
|
||||
}
|
||||
}
|
||||
|
||||
decode(input = empty, options = kEmptyObject) {
|
||||
validateDecoder(this);
|
||||
if (isAnyArrayBuffer(input)) {
|
||||
try {
|
||||
input = Buffer.from(input);
|
||||
} catch {
|
||||
input = empty;
|
||||
}
|
||||
} else if (isArrayBufferView(input)) {
|
||||
try {
|
||||
input = Buffer.from(input.buffer, input.byteOffset,
|
||||
input.byteLength);
|
||||
} catch {
|
||||
input = empty;
|
||||
}
|
||||
} else {
|
||||
throw new ERR_INVALID_ARG_TYPE('input',
|
||||
['ArrayBuffer', 'ArrayBufferView'],
|
||||
input);
|
||||
}
|
||||
input = parseInput(input);
|
||||
validateObject(options, 'options', kValidateObjectAllowObjectsAndNull);
|
||||
|
||||
if (this[kMethod]) return this[kMethod](input);
|
||||
|
||||
if (this[kFlags] & CONVERTER_FLAGS_FLUSH) {
|
||||
this[kBOMSeen] = false;
|
||||
}
|
||||
@@ -542,9 +551,7 @@ function makeTextDecoderJS() {
|
||||
this[kHandle].end(input) :
|
||||
this[kHandle].write(input);
|
||||
|
||||
if (result.length > 0 &&
|
||||
!this[kBOMSeen] &&
|
||||
!(this[kFlags] & CONVERTER_FLAGS_IGNORE_BOM)) {
|
||||
if (result.length > 0 && !this[kBOMSeen] && !this[kIgnoreBOM]) {
|
||||
// If the very first result in the stream is a BOM, and we are not
|
||||
// explicitly told to ignore it, then we discard it.
|
||||
if (result[0] === '\ufeff') {
|
||||
|
||||
155
lib/internal/encoding/single-byte.js
Normal file
155
lib/internal/encoding/single-byte.js
Normal file
@@ -0,0 +1,155 @@
|
||||
// Simplified version extracted from https://npmjs.com/package/@exodus/bytes codepath for 1-byte encodings
|
||||
// Copyright Exodus Movement. Licensed under MIT License.
|
||||
|
||||
'use strict';
|
||||
|
||||
const {
|
||||
Array,
|
||||
ArrayPrototypeFill,
|
||||
ObjectKeys,
|
||||
ObjectPrototypeHasOwnProperty,
|
||||
SafeArrayIterator,
|
||||
SafeMap,
|
||||
SafeSet,
|
||||
StringPrototypeIncludes,
|
||||
TypedArrayFrom,
|
||||
TypedArrayOf,
|
||||
TypedArrayPrototypeIncludes,
|
||||
TypedArrayPrototypeSet,
|
||||
Uint16Array,
|
||||
} = primordials;
|
||||
|
||||
const { isAscii } = require('buffer');
|
||||
|
||||
const { FastBuffer } = require('internal/buffer');
|
||||
|
||||
const {
|
||||
ERR_ENCODING_NOT_SUPPORTED,
|
||||
ERR_ENCODING_INVALID_ENCODED_DATA,
|
||||
} = require('internal/errors').codes;
|
||||
|
||||
const isBigEndian = new FastBuffer(TypedArrayOf(Uint16Array, 258).buffer)[1] === 2;
|
||||
|
||||
const it = (x) => new SafeArrayIterator(x);
|
||||
|
||||
/* fallback/single-byte.encodings.js */
|
||||
|
||||
const r = 0xfffd;
|
||||
const e = (x) => it(ArrayPrototypeFill(new Array(x), 1));
|
||||
const h = (x) => it(ArrayPrototypeFill(new Array(x), r));
|
||||
|
||||
/* eslint-disable @stylistic/js/max-len */
|
||||
|
||||
// Index tables from https://encoding.spec.whatwg.org/#legacy-single-byte-encodings
|
||||
// Each table in the spec lists only mapping from byte 0x80 onwards, as below that they are all ASCII and mapped as idenity
|
||||
// Here, 0xfffd (replacement charcode) designates a hole (unmapped offset), as not all encodings map all offsets
|
||||
// All other numbers are deltas from the last seen mapped value, starting with 0x7f (127, highest ASCII)
|
||||
// Thus, [0x80, 0x81, , 0x83] is stored as [1, 1, r, 2]
|
||||
// Truncation (length < 128) means that all remaining ones are mapped as identity (offset i => codepoint i), not unmapped
|
||||
const encodings = {
|
||||
'__proto__': null,
|
||||
'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': [...e(33), 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, -75, -63, ...e(5), 104, -34, -67, 79, -77, 75, -73, 1, 1, 1, 117, 7, -121, 1, 1, 1, 146, -144, 154, -152, ...e(5), 34, -32, ...e(5), 73, -34, -36, 48, -46, 44, -42, 1, 1, 1, 86, 7, -90, 1, 1, 1, 115, -113, 123, -121, 1, 1, 1, 1, 58],
|
||||
'iso-8859-13': [...e(33), 8061, -8059, 1, 1, 8058, -8056, 1, 49, -47, 173, -171, 1, 1, 1, 24, -22, 1, 1, 1, 8041, -8039, 1, 1, 65, -63, 158, -156, 1, 1, 1, 40, 30, 42, -46, 6, -66, 1, 83, -6, -6, -67, 176, -99, 12, 20, -12, 17, 37, -29, 2, -114, 121, -119, 1, 1, 155, -49, 25, 16, -142, 159, 2, -158, 38, 42, -46, 6, -35, 1, 52, -6, -6, -36, 145, -99, 12, 20, -12, 17, 37, -29, 2, -83, 90, -88, 1, 1, 124, -49, 25, 16, -111, 128, 2, 7835],
|
||||
'iso-8859-14': [...e(33), 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': [...e(33), 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': [...e(33), 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': [...e(33), 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, -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],
|
||||
'iso-8859-3': [...e(33), 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': [...e(33), 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, -75, -63, ...e(5), 104, -34, -67, 79, -77, 75, -73, 1, 92, -26, 53, 7, -22, -98, 1, 1, 1, 1, 154, -152, 1, 1, 140, 2, -139, 34, -32, ...e(5), 73, -34, -36, 48, -46, 44, -42, 1, 61, -26, 53, 7, -22, -67, 1, 1, 1, 1, 123, -121, 1, 1, 109, 2, 366],
|
||||
'iso-8859-5': [...e(33), 865, ...e(11), -863, 865, ...e(65), 7367, -7365, ...e(11), -949, 951, 1],
|
||||
'iso-8859-6': [...e(33), 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': [...e(33), 8056, 1, -8054, 8201, 3, -8201, 1, 1, 1, 721, -719, 1, 1, r, 8040, -8037, 1, 1, 1, 721, 1, 1, -719, 721, 1, 1, -719, 721, -719, 721, ...e(19), r, 2, ...e(43), r],
|
||||
'iso-8859-8': [...e(33), r, 2, ...e(7), 46, -44, ...e(14), 62, -60, 1, 1, 1, ...h(32), 8025, -6727, ...e(26), r, r, 6692, 1, r],
|
||||
'koi8-r': [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, 8450, ...e(14), -8544, 8545, ...e(10), -9411, 933, -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, -28, -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],
|
||||
'koi8-u': [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, 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, -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, -28, -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],
|
||||
'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, 8558, -8328, 8374, -66, -8539, 16, 8043, -8070, 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': [8237, -8235, 8089, -8087, 8091, 8, -6, 1, -8089, 8104, -7888, 7897, -7903, 10, 25, -4, -233, 8072, 1, 3, 1, 5, -15, 1, -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, -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],
|
||||
'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': [8237, -8235, 8089, -7816, 7820, 8, -6, 1, -7515, 7530, -7888, 7897, -7911, -197, 240, -238, 1, 8072, 1, 3, 1, 5, -15, 1, -7480, 7750, -8129, 7897, -7911, -182, 225, -6],
|
||||
'windows-1253': [8237, -8235, 8089, -7816, 7820, 8, -6, 1, -8089, 8104, -8102, 8111, -8109, 1, 1, 1, 1, 8072, 1, 3, 1, 5, -15, 1, -8060, 8330, -8328, 8096, -8094, 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, 721, 1, 1, -719, 721, -719, 721, ...e(19), r, 2, ...e(43), r],
|
||||
'windows-1254': [8237, -8235, 8089, -7816, 7820, 8, -6, 1, -7515, 7530, -7888, 7897, -7911, -197, 1, 1, 1, 8072, 1, 3, 1, 5, -15, 1, -7480, 7750, -8129, 7897, -7911, -182, 1, 218, -216, ...e(47), 79, -77, ...e(11), 84, 46, -127, ...e(16), 48, -46, ...e(11), 53, 46],
|
||||
'windows-1255': [8237, -8235, 8089, -7816, 7820, 8, -6, 1, -7515, 7530, -8102, 8111, -8109, 1, 1, 1, 1, 8072, 1, 3, 1, 5, -15, 1, -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, ...e(26), r, r, 6692, 1, r],
|
||||
'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': [8237, -8235, 8089, -8087, 8091, 8, -6, 1, -8089, 8104, -8102, 8111, -8109, 28, 543, -527, -40, 8072, 1, 3, 1, 5, -15, 1, -8060, 8330, -8328, 8096, -8094, 19, 556, -572, 1, r, 2, 1, 1, r, 2, 1, 49, -47, 173, -171, 1, 1, 1, 24, -22, ...e(5), 1, 1, 65, -63, 158, -156, 1, 1, 1, 40, 30, 42, -46, 6, -66, 1, 83, -6, -6, -67, 176, -99, 12, 20, -12, 17, 37, -29, 2, -114, 121, -119, 1, 1, 155, -49, 25, 16, -142, 159, 2, -158, 38, 42, -46, 6, -35, 1, 52, -6, -6, -36, 145, -99, 12, 20, -12, 17, 37, -29, 2, -83, 90, -88, 1, 1, 124, -49, 25, 16, -111, 128, 2, 347],
|
||||
'windows-1258': [8237, -8235, 8089, -7816, 7820, 8, -6, 1, -7515, 7530, -8102, 8111, -7911, -197, 1, 1, 1, 8072, 1, 3, 1, 5, -15, 1, -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), 8072, 1, 3, 1, 5, -15, 1, -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, 8558, -8328, 8374, -66, -8539, 16, 8043, -8070, 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],
|
||||
};
|
||||
|
||||
/* eslint-enable @stylistic/js/max-len */
|
||||
|
||||
/* fallback/single-byte.js + single-byte.node.js, simplified */
|
||||
|
||||
const l256 = { __proto__: null, length: 256 };
|
||||
|
||||
function getEncoding(encoding) {
|
||||
if (encoding === 'x-user-defined') {
|
||||
// https://encoding.spec.whatwg.org/#x-user-defined-decoder, 14.5.1. x-user-defined decoder
|
||||
return TypedArrayFrom(Uint16Array, l256, (_, i) => (i >= 0x80 ? 0xf700 + i : i));
|
||||
}
|
||||
|
||||
if (!ObjectPrototypeHasOwnProperty(encodings, encoding)) {
|
||||
throw new ERR_ENCODING_NOT_SUPPORTED(encoding);
|
||||
}
|
||||
|
||||
const map = TypedArrayFrom(Uint16Array, l256, (_, i) => i); // Unicode subset
|
||||
let prev = 127;
|
||||
map.set(TypedArrayFrom(Uint16Array, it(encodings[encoding]), (x) => (x === r ? x : (prev += x))), 128);
|
||||
return map;
|
||||
}
|
||||
|
||||
const supported = new SafeSet(it(ObjectKeys(encodings))).add('iso-8859-8-i').add('x-user-defined');
|
||||
const isSinglebyteEncoding = (enc) => supported.has(enc);
|
||||
|
||||
const decodersLoose = new SafeMap();
|
||||
const decodersFatal = new SafeMap();
|
||||
|
||||
function createSinglebyteDecoder(encoding, fatal) {
|
||||
const id = encoding === 'iso-8859-8-i' ? 'iso-8859-8' : encoding;
|
||||
const decoders = fatal ? decodersFatal : decodersLoose;
|
||||
const cached = decoders.get(id);
|
||||
if (cached) return cached;
|
||||
|
||||
const map = getEncoding(id);
|
||||
const incomplete = TypedArrayPrototypeIncludes(map, r);
|
||||
|
||||
// Expects type-checked Buffer input
|
||||
const decoder = (buf) => {
|
||||
if (buf.byteLength === 0) return '';
|
||||
if (isAscii(buf)) return buf.latin1Slice(); // .latin1Slice is faster than .asciiSlice
|
||||
const o = new Uint16Array(buf.length);
|
||||
TypedArrayPrototypeSet(o, buf); // Copy to modify in-place, also those are 16-bit now
|
||||
|
||||
let i = 0;
|
||||
for (const end7 = o.length - 7; i < end7; i += 8) {
|
||||
o[i] = map[o[i]];
|
||||
o[i + 1] = map[o[i + 1]];
|
||||
o[i + 2] = map[o[i + 2]];
|
||||
o[i + 3] = map[o[i + 3]];
|
||||
o[i + 4] = map[o[i + 4]];
|
||||
o[i + 5] = map[o[i + 5]];
|
||||
o[i + 6] = map[o[i + 6]];
|
||||
o[i + 7] = map[o[i + 7]];
|
||||
}
|
||||
|
||||
for (const end = o.length; i < end; i++) o[i] = map[o[i]];
|
||||
|
||||
const b = new FastBuffer(o.buffer, o.byteOffset, o.byteLength);
|
||||
if (isBigEndian) b.swap16();
|
||||
const string = b.ucs2Slice();
|
||||
if (fatal && incomplete && StringPrototypeIncludes(string, '\uFFFD')) {
|
||||
throw new ERR_ENCODING_INVALID_ENCODED_DATA(encoding, undefined);
|
||||
}
|
||||
return string;
|
||||
};
|
||||
|
||||
decoders.set(id, decoder);
|
||||
return decoder;
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
isSinglebyteEncoding,
|
||||
createSinglebyteDecoder,
|
||||
getEncoding, // for tests
|
||||
};
|
||||
@@ -1,7 +1,6 @@
|
||||
#include "encoding_binding.h"
|
||||
#include "ada.h"
|
||||
#include "env-inl.h"
|
||||
#include "node_buffer.h"
|
||||
#include "node_errors.h"
|
||||
#include "node_external_reference.h"
|
||||
#include "simdutf.h"
|
||||
@@ -424,8 +423,6 @@ void BindingData::CreatePerIsolateProperties(IsolateData* isolate_data,
|
||||
SetMethodNoSideEffect(isolate, target, "decodeUTF8", DecodeUTF8);
|
||||
SetMethodNoSideEffect(isolate, target, "toASCII", ToASCII);
|
||||
SetMethodNoSideEffect(isolate, target, "toUnicode", ToUnicode);
|
||||
SetMethodNoSideEffect(
|
||||
isolate, target, "decodeWindows1252", DecodeWindows1252);
|
||||
}
|
||||
|
||||
void BindingData::CreatePerContextProperties(Local<Object> target,
|
||||
@@ -443,77 +440,6 @@ void BindingData::RegisterTimerExternalReferences(
|
||||
registry->Register(DecodeUTF8);
|
||||
registry->Register(ToASCII);
|
||||
registry->Register(ToUnicode);
|
||||
registry->Register(DecodeWindows1252);
|
||||
}
|
||||
|
||||
void BindingData::DecodeWindows1252(const FunctionCallbackInfo<Value>& args) {
|
||||
Environment* env = Environment::GetCurrent(args);
|
||||
|
||||
CHECK_GE(args.Length(), 1);
|
||||
if (!(args[0]->IsArrayBuffer() || args[0]->IsSharedArrayBuffer() ||
|
||||
args[0]->IsArrayBufferView())) {
|
||||
return node::THROW_ERR_INVALID_ARG_TYPE(
|
||||
env->isolate(),
|
||||
"The \"input\" argument must be an instance of ArrayBuffer, "
|
||||
"SharedArrayBuffer, or ArrayBufferView.");
|
||||
}
|
||||
|
||||
bool ignore_bom = args[1]->IsTrue();
|
||||
|
||||
ArrayBufferViewContents<uint8_t> buffer(args[0]);
|
||||
const uint8_t* data = buffer.data();
|
||||
size_t length = buffer.length();
|
||||
|
||||
if (ignore_bom && length > 0 && data[0] == 0xFF) {
|
||||
data++;
|
||||
length--;
|
||||
}
|
||||
|
||||
if (length == 0) {
|
||||
return args.GetReturnValue().SetEmptyString();
|
||||
}
|
||||
|
||||
// Windows-1252 specific mapping for bytes 128-159
|
||||
// These differ from Latin-1/ISO-8859-1
|
||||
static const uint16_t windows1252_mapping[32] = {
|
||||
0x20AC, 0x0081, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021, // 80-87
|
||||
0x02C6, 0x2030, 0x0160, 0x2039, 0x0152, 0x008D, 0x017D, 0x008F, // 88-8F
|
||||
0x0090, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014, // 90-97
|
||||
0x02DC, 0x2122, 0x0161, 0x203A, 0x0153, 0x009D, 0x017E, 0x0178 // 98-9F
|
||||
};
|
||||
|
||||
std::string result;
|
||||
result.reserve(length * 3); // Reserve space for UTF-8 output
|
||||
|
||||
for (size_t i = 0; i < length; i++) {
|
||||
uint8_t byte = data[i];
|
||||
uint32_t codepoint;
|
||||
|
||||
// Check if byte is in the special Windows-1252 range (128-159)
|
||||
if (byte >= 0x80 && byte <= 0x9F) {
|
||||
codepoint = windows1252_mapping[byte - 0x80];
|
||||
} else {
|
||||
// For all other bytes, Windows-1252 is identical to Latin-1
|
||||
codepoint = byte;
|
||||
}
|
||||
|
||||
// Convert codepoint to UTF-8
|
||||
if (codepoint < 0x80) {
|
||||
result.push_back(static_cast<char>(codepoint));
|
||||
} else if (codepoint < 0x800) {
|
||||
result.push_back(static_cast<char>(0xC0 | (codepoint >> 6)));
|
||||
result.push_back(static_cast<char>(0x80 | (codepoint & 0x3F)));
|
||||
} else {
|
||||
result.push_back(static_cast<char>(0xE0 | (codepoint >> 12)));
|
||||
result.push_back(static_cast<char>(0x80 | ((codepoint >> 6) & 0x3F)));
|
||||
result.push_back(static_cast<char>(0x80 | (codepoint & 0x3F)));
|
||||
}
|
||||
}
|
||||
|
||||
Local<Value> ret;
|
||||
if (ToV8Value(env->context(), result, env->isolate()).ToLocal(&ret)) {
|
||||
args.GetReturnValue().Set(ret);
|
||||
}
|
||||
}
|
||||
|
||||
} // namespace encoding_binding
|
||||
|
||||
@@ -31,8 +31,6 @@ class BindingData : public SnapshotableObject {
|
||||
static void EncodeInto(const v8::FunctionCallbackInfo<v8::Value>& args);
|
||||
static void EncodeUtf8String(const v8::FunctionCallbackInfo<v8::Value>& args);
|
||||
static void DecodeUTF8(const v8::FunctionCallbackInfo<v8::Value>& args);
|
||||
static void DecodeWindows1252(
|
||||
const v8::FunctionCallbackInfo<v8::Value>& args);
|
||||
|
||||
static void ToASCII(const v8::FunctionCallbackInfo<v8::Value>& args);
|
||||
static void ToUnicode(const v8::FunctionCallbackInfo<v8::Value>& args);
|
||||
|
||||
4
test/fixtures/encoding/README.md
vendored
Normal file
4
test/fixtures/encoding/README.md
vendored
Normal file
@@ -0,0 +1,4 @@
|
||||
1. <https://encoding.spec.whatwg.org/>
|
||||
2. <https://encoding.spec.whatwg.org/encodings.json>
|
||||
3. <https://encoding.spec.whatwg.org/#indexes>
|
||||
4. <https://encoding.spec.whatwg.org/#legacy-single-byte-encodings>
|
||||
465
test/fixtures/encoding/encodings.json
vendored
Normal file
465
test/fixtures/encoding/encodings.json
vendored
Normal file
@@ -0,0 +1,465 @@
|
||||
[
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"unicode-1-1-utf-8",
|
||||
"unicode11utf8",
|
||||
"unicode20utf8",
|
||||
"utf-8",
|
||||
"utf8",
|
||||
"x-unicode20utf8"
|
||||
],
|
||||
"name": "UTF-8"
|
||||
}
|
||||
],
|
||||
"heading": "The Encoding"
|
||||
},
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"866",
|
||||
"cp866",
|
||||
"csibm866",
|
||||
"ibm866"
|
||||
],
|
||||
"name": "IBM866"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatin2",
|
||||
"iso-8859-2",
|
||||
"iso-ir-101",
|
||||
"iso8859-2",
|
||||
"iso88592",
|
||||
"iso_8859-2",
|
||||
"iso_8859-2:1987",
|
||||
"l2",
|
||||
"latin2"
|
||||
],
|
||||
"name": "ISO-8859-2"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatin3",
|
||||
"iso-8859-3",
|
||||
"iso-ir-109",
|
||||
"iso8859-3",
|
||||
"iso88593",
|
||||
"iso_8859-3",
|
||||
"iso_8859-3:1988",
|
||||
"l3",
|
||||
"latin3"
|
||||
],
|
||||
"name": "ISO-8859-3"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatin4",
|
||||
"iso-8859-4",
|
||||
"iso-ir-110",
|
||||
"iso8859-4",
|
||||
"iso88594",
|
||||
"iso_8859-4",
|
||||
"iso_8859-4:1988",
|
||||
"l4",
|
||||
"latin4"
|
||||
],
|
||||
"name": "ISO-8859-4"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatincyrillic",
|
||||
"cyrillic",
|
||||
"iso-8859-5",
|
||||
"iso-ir-144",
|
||||
"iso8859-5",
|
||||
"iso88595",
|
||||
"iso_8859-5",
|
||||
"iso_8859-5:1988"
|
||||
],
|
||||
"name": "ISO-8859-5"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"arabic",
|
||||
"asmo-708",
|
||||
"csiso88596e",
|
||||
"csiso88596i",
|
||||
"csisolatinarabic",
|
||||
"ecma-114",
|
||||
"iso-8859-6",
|
||||
"iso-8859-6-e",
|
||||
"iso-8859-6-i",
|
||||
"iso-ir-127",
|
||||
"iso8859-6",
|
||||
"iso88596",
|
||||
"iso_8859-6",
|
||||
"iso_8859-6:1987"
|
||||
],
|
||||
"name": "ISO-8859-6"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatingreek",
|
||||
"ecma-118",
|
||||
"elot_928",
|
||||
"greek",
|
||||
"greek8",
|
||||
"iso-8859-7",
|
||||
"iso-ir-126",
|
||||
"iso8859-7",
|
||||
"iso88597",
|
||||
"iso_8859-7",
|
||||
"iso_8859-7:1987",
|
||||
"sun_eu_greek"
|
||||
],
|
||||
"name": "ISO-8859-7"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csiso88598e",
|
||||
"csisolatinhebrew",
|
||||
"hebrew",
|
||||
"iso-8859-8",
|
||||
"iso-8859-8-e",
|
||||
"iso-ir-138",
|
||||
"iso8859-8",
|
||||
"iso88598",
|
||||
"iso_8859-8",
|
||||
"iso_8859-8:1988",
|
||||
"visual"
|
||||
],
|
||||
"name": "ISO-8859-8"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csiso88598i",
|
||||
"iso-8859-8-i",
|
||||
"logical"
|
||||
],
|
||||
"name": "ISO-8859-8-I"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatin6",
|
||||
"iso-8859-10",
|
||||
"iso-ir-157",
|
||||
"iso8859-10",
|
||||
"iso885910",
|
||||
"l6",
|
||||
"latin6"
|
||||
],
|
||||
"name": "ISO-8859-10"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"iso-8859-13",
|
||||
"iso8859-13",
|
||||
"iso885913"
|
||||
],
|
||||
"name": "ISO-8859-13"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"iso-8859-14",
|
||||
"iso8859-14",
|
||||
"iso885914"
|
||||
],
|
||||
"name": "ISO-8859-14"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csisolatin9",
|
||||
"iso-8859-15",
|
||||
"iso8859-15",
|
||||
"iso885915",
|
||||
"iso_8859-15",
|
||||
"l9"
|
||||
],
|
||||
"name": "ISO-8859-15"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"iso-8859-16"
|
||||
],
|
||||
"name": "ISO-8859-16"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cskoi8r",
|
||||
"koi",
|
||||
"koi8",
|
||||
"koi8-r",
|
||||
"koi8_r"
|
||||
],
|
||||
"name": "KOI8-R"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"koi8-ru",
|
||||
"koi8-u"
|
||||
],
|
||||
"name": "KOI8-U"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csmacintosh",
|
||||
"mac",
|
||||
"macintosh",
|
||||
"x-mac-roman"
|
||||
],
|
||||
"name": "macintosh"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"dos-874",
|
||||
"iso-8859-11",
|
||||
"iso8859-11",
|
||||
"iso885911",
|
||||
"tis-620",
|
||||
"windows-874"
|
||||
],
|
||||
"name": "windows-874"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1250",
|
||||
"windows-1250",
|
||||
"x-cp1250"
|
||||
],
|
||||
"name": "windows-1250"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1251",
|
||||
"windows-1251",
|
||||
"x-cp1251"
|
||||
],
|
||||
"name": "windows-1251"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"ansi_x3.4-1968",
|
||||
"ascii",
|
||||
"cp1252",
|
||||
"cp819",
|
||||
"csisolatin1",
|
||||
"ibm819",
|
||||
"iso-8859-1",
|
||||
"iso-ir-100",
|
||||
"iso8859-1",
|
||||
"iso88591",
|
||||
"iso_8859-1",
|
||||
"iso_8859-1:1987",
|
||||
"l1",
|
||||
"latin1",
|
||||
"us-ascii",
|
||||
"windows-1252",
|
||||
"x-cp1252"
|
||||
],
|
||||
"name": "windows-1252"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1253",
|
||||
"windows-1253",
|
||||
"x-cp1253"
|
||||
],
|
||||
"name": "windows-1253"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1254",
|
||||
"csisolatin5",
|
||||
"iso-8859-9",
|
||||
"iso-ir-148",
|
||||
"iso8859-9",
|
||||
"iso88599",
|
||||
"iso_8859-9",
|
||||
"iso_8859-9:1989",
|
||||
"l5",
|
||||
"latin5",
|
||||
"windows-1254",
|
||||
"x-cp1254"
|
||||
],
|
||||
"name": "windows-1254"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1255",
|
||||
"windows-1255",
|
||||
"x-cp1255"
|
||||
],
|
||||
"name": "windows-1255"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1256",
|
||||
"windows-1256",
|
||||
"x-cp1256"
|
||||
],
|
||||
"name": "windows-1256"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1257",
|
||||
"windows-1257",
|
||||
"x-cp1257"
|
||||
],
|
||||
"name": "windows-1257"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"cp1258",
|
||||
"windows-1258",
|
||||
"x-cp1258"
|
||||
],
|
||||
"name": "windows-1258"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"x-mac-cyrillic",
|
||||
"x-mac-ukrainian"
|
||||
],
|
||||
"name": "x-mac-cyrillic"
|
||||
}
|
||||
],
|
||||
"heading": "Legacy single-byte encodings"
|
||||
},
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"chinese",
|
||||
"csgb2312",
|
||||
"csiso58gb231280",
|
||||
"gb2312",
|
||||
"gb_2312",
|
||||
"gb_2312-80",
|
||||
"gbk",
|
||||
"iso-ir-58",
|
||||
"x-gbk"
|
||||
],
|
||||
"name": "GBK"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"gb18030"
|
||||
],
|
||||
"name": "gb18030"
|
||||
}
|
||||
],
|
||||
"heading": "Legacy multi-byte Chinese (simplified) encodings"
|
||||
},
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"big5",
|
||||
"big5-hkscs",
|
||||
"cn-big5",
|
||||
"csbig5",
|
||||
"x-x-big5"
|
||||
],
|
||||
"name": "Big5"
|
||||
}
|
||||
],
|
||||
"heading": "Legacy multi-byte Chinese (traditional) encodings"
|
||||
},
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"cseucpkdfmtjapanese",
|
||||
"euc-jp",
|
||||
"x-euc-jp"
|
||||
],
|
||||
"name": "EUC-JP"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csiso2022jp",
|
||||
"iso-2022-jp"
|
||||
],
|
||||
"name": "ISO-2022-JP"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csshiftjis",
|
||||
"ms932",
|
||||
"ms_kanji",
|
||||
"shift-jis",
|
||||
"shift_jis",
|
||||
"sjis",
|
||||
"windows-31j",
|
||||
"x-sjis"
|
||||
],
|
||||
"name": "Shift_JIS"
|
||||
}
|
||||
],
|
||||
"heading": "Legacy multi-byte Japanese encodings"
|
||||
},
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"cseuckr",
|
||||
"csksc56011987",
|
||||
"euc-kr",
|
||||
"iso-ir-149",
|
||||
"korean",
|
||||
"ks_c_5601-1987",
|
||||
"ks_c_5601-1989",
|
||||
"ksc5601",
|
||||
"ksc_5601",
|
||||
"windows-949"
|
||||
],
|
||||
"name": "EUC-KR"
|
||||
}
|
||||
],
|
||||
"heading": "Legacy multi-byte Korean encodings"
|
||||
},
|
||||
{
|
||||
"encodings": [
|
||||
{
|
||||
"labels": [
|
||||
"csiso2022kr",
|
||||
"hz-gb-2312",
|
||||
"iso-2022-cn",
|
||||
"iso-2022-cn-ext",
|
||||
"iso-2022-kr",
|
||||
"replacement"
|
||||
],
|
||||
"name": "replacement"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"unicodefffe",
|
||||
"utf-16be"
|
||||
],
|
||||
"name": "UTF-16BE"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"csunicode",
|
||||
"iso-10646-ucs-2",
|
||||
"ucs-2",
|
||||
"unicode",
|
||||
"unicodefeff",
|
||||
"utf-16",
|
||||
"utf-16le"
|
||||
],
|
||||
"name": "UTF-16LE"
|
||||
},
|
||||
{
|
||||
"labels": [
|
||||
"x-user-defined"
|
||||
],
|
||||
"name": "x-user-defined"
|
||||
}
|
||||
],
|
||||
"heading": "Legacy miscellaneous encodings"
|
||||
}
|
||||
]
|
||||
134
test/fixtures/encoding/single-byte/index-ibm866.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-ibm866.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-ibm866.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: db6fe14a559d1601a7667338d83704773d5708dbc641e1ad3c5e21405770f05e
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0410 А (CYRILLIC CAPITAL LETTER A)
|
||||
1 0x0411 Б (CYRILLIC CAPITAL LETTER BE)
|
||||
2 0x0412 В (CYRILLIC CAPITAL LETTER VE)
|
||||
3 0x0413 Г (CYRILLIC CAPITAL LETTER GHE)
|
||||
4 0x0414 Д (CYRILLIC CAPITAL LETTER DE)
|
||||
5 0x0415 Е (CYRILLIC CAPITAL LETTER IE)
|
||||
6 0x0416 Ж (CYRILLIC CAPITAL LETTER ZHE)
|
||||
7 0x0417 З (CYRILLIC CAPITAL LETTER ZE)
|
||||
8 0x0418 И (CYRILLIC CAPITAL LETTER I)
|
||||
9 0x0419 Й (CYRILLIC CAPITAL LETTER SHORT I)
|
||||
10 0x041A К (CYRILLIC CAPITAL LETTER KA)
|
||||
11 0x041B Л (CYRILLIC CAPITAL LETTER EL)
|
||||
12 0x041C М (CYRILLIC CAPITAL LETTER EM)
|
||||
13 0x041D Н (CYRILLIC CAPITAL LETTER EN)
|
||||
14 0x041E О (CYRILLIC CAPITAL LETTER O)
|
||||
15 0x041F П (CYRILLIC CAPITAL LETTER PE)
|
||||
16 0x0420 Р (CYRILLIC CAPITAL LETTER ER)
|
||||
17 0x0421 С (CYRILLIC CAPITAL LETTER ES)
|
||||
18 0x0422 Т (CYRILLIC CAPITAL LETTER TE)
|
||||
19 0x0423 У (CYRILLIC CAPITAL LETTER U)
|
||||
20 0x0424 Ф (CYRILLIC CAPITAL LETTER EF)
|
||||
21 0x0425 Х (CYRILLIC CAPITAL LETTER HA)
|
||||
22 0x0426 Ц (CYRILLIC CAPITAL LETTER TSE)
|
||||
23 0x0427 Ч (CYRILLIC CAPITAL LETTER CHE)
|
||||
24 0x0428 Ш (CYRILLIC CAPITAL LETTER SHA)
|
||||
25 0x0429 Щ (CYRILLIC CAPITAL LETTER SHCHA)
|
||||
26 0x042A Ъ (CYRILLIC CAPITAL LETTER HARD SIGN)
|
||||
27 0x042B Ы (CYRILLIC CAPITAL LETTER YERU)
|
||||
28 0x042C Ь (CYRILLIC CAPITAL LETTER SOFT SIGN)
|
||||
29 0x042D Э (CYRILLIC CAPITAL LETTER E)
|
||||
30 0x042E Ю (CYRILLIC CAPITAL LETTER YU)
|
||||
31 0x042F Я (CYRILLIC CAPITAL LETTER YA)
|
||||
32 0x0430 а (CYRILLIC SMALL LETTER A)
|
||||
33 0x0431 б (CYRILLIC SMALL LETTER BE)
|
||||
34 0x0432 в (CYRILLIC SMALL LETTER VE)
|
||||
35 0x0433 г (CYRILLIC SMALL LETTER GHE)
|
||||
36 0x0434 д (CYRILLIC SMALL LETTER DE)
|
||||
37 0x0435 е (CYRILLIC SMALL LETTER IE)
|
||||
38 0x0436 ж (CYRILLIC SMALL LETTER ZHE)
|
||||
39 0x0437 з (CYRILLIC SMALL LETTER ZE)
|
||||
40 0x0438 и (CYRILLIC SMALL LETTER I)
|
||||
41 0x0439 й (CYRILLIC SMALL LETTER SHORT I)
|
||||
42 0x043A к (CYRILLIC SMALL LETTER KA)
|
||||
43 0x043B л (CYRILLIC SMALL LETTER EL)
|
||||
44 0x043C м (CYRILLIC SMALL LETTER EM)
|
||||
45 0x043D н (CYRILLIC SMALL LETTER EN)
|
||||
46 0x043E о (CYRILLIC SMALL LETTER O)
|
||||
47 0x043F п (CYRILLIC SMALL LETTER PE)
|
||||
48 0x2591 ░ (LIGHT SHADE)
|
||||
49 0x2592 ▒ (MEDIUM SHADE)
|
||||
50 0x2593 ▓ (DARK SHADE)
|
||||
51 0x2502 │ (BOX DRAWINGS LIGHT VERTICAL)
|
||||
52 0x2524 ┤ (BOX DRAWINGS LIGHT VERTICAL AND LEFT)
|
||||
53 0x2561 ╡ (BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE)
|
||||
54 0x2562 ╢ (BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE)
|
||||
55 0x2556 ╖ (BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE)
|
||||
56 0x2555 ╕ (BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE)
|
||||
57 0x2563 ╣ (BOX DRAWINGS DOUBLE VERTICAL AND LEFT)
|
||||
58 0x2551 ║ (BOX DRAWINGS DOUBLE VERTICAL)
|
||||
59 0x2557 ╗ (BOX DRAWINGS DOUBLE DOWN AND LEFT)
|
||||
60 0x255D ╝ (BOX DRAWINGS DOUBLE UP AND LEFT)
|
||||
61 0x255C ╜ (BOX DRAWINGS UP DOUBLE AND LEFT SINGLE)
|
||||
62 0x255B ╛ (BOX DRAWINGS UP SINGLE AND LEFT DOUBLE)
|
||||
63 0x2510 ┐ (BOX DRAWINGS LIGHT DOWN AND LEFT)
|
||||
64 0x2514 └ (BOX DRAWINGS LIGHT UP AND RIGHT)
|
||||
65 0x2534 ┴ (BOX DRAWINGS LIGHT UP AND HORIZONTAL)
|
||||
66 0x252C ┬ (BOX DRAWINGS LIGHT DOWN AND HORIZONTAL)
|
||||
67 0x251C ├ (BOX DRAWINGS LIGHT VERTICAL AND RIGHT)
|
||||
68 0x2500 ─ (BOX DRAWINGS LIGHT HORIZONTAL)
|
||||
69 0x253C ┼ (BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL)
|
||||
70 0x255E ╞ (BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE)
|
||||
71 0x255F ╟ (BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE)
|
||||
72 0x255A ╚ (BOX DRAWINGS DOUBLE UP AND RIGHT)
|
||||
73 0x2554 ╔ (BOX DRAWINGS DOUBLE DOWN AND RIGHT)
|
||||
74 0x2569 ╩ (BOX DRAWINGS DOUBLE UP AND HORIZONTAL)
|
||||
75 0x2566 ╦ (BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL)
|
||||
76 0x2560 ╠ (BOX DRAWINGS DOUBLE VERTICAL AND RIGHT)
|
||||
77 0x2550 ═ (BOX DRAWINGS DOUBLE HORIZONTAL)
|
||||
78 0x256C ╬ (BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL)
|
||||
79 0x2567 ╧ (BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE)
|
||||
80 0x2568 ╨ (BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE)
|
||||
81 0x2564 ╤ (BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE)
|
||||
82 0x2565 ╥ (BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE)
|
||||
83 0x2559 ╙ (BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE)
|
||||
84 0x2558 ╘ (BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE)
|
||||
85 0x2552 ╒ (BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE)
|
||||
86 0x2553 ╓ (BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE)
|
||||
87 0x256B ╫ (BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE)
|
||||
88 0x256A ╪ (BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE)
|
||||
89 0x2518 ┘ (BOX DRAWINGS LIGHT UP AND LEFT)
|
||||
90 0x250C ┌ (BOX DRAWINGS LIGHT DOWN AND RIGHT)
|
||||
91 0x2588 █ (FULL BLOCK)
|
||||
92 0x2584 ▄ (LOWER HALF BLOCK)
|
||||
93 0x258C ▌ (LEFT HALF BLOCK)
|
||||
94 0x2590 ▐ (RIGHT HALF BLOCK)
|
||||
95 0x2580 ▀ (UPPER HALF BLOCK)
|
||||
96 0x0440 р (CYRILLIC SMALL LETTER ER)
|
||||
97 0x0441 с (CYRILLIC SMALL LETTER ES)
|
||||
98 0x0442 т (CYRILLIC SMALL LETTER TE)
|
||||
99 0x0443 у (CYRILLIC SMALL LETTER U)
|
||||
100 0x0444 ф (CYRILLIC SMALL LETTER EF)
|
||||
101 0x0445 х (CYRILLIC SMALL LETTER HA)
|
||||
102 0x0446 ц (CYRILLIC SMALL LETTER TSE)
|
||||
103 0x0447 ч (CYRILLIC SMALL LETTER CHE)
|
||||
104 0x0448 ш (CYRILLIC SMALL LETTER SHA)
|
||||
105 0x0449 щ (CYRILLIC SMALL LETTER SHCHA)
|
||||
106 0x044A ъ (CYRILLIC SMALL LETTER HARD SIGN)
|
||||
107 0x044B ы (CYRILLIC SMALL LETTER YERU)
|
||||
108 0x044C ь (CYRILLIC SMALL LETTER SOFT SIGN)
|
||||
109 0x044D э (CYRILLIC SMALL LETTER E)
|
||||
110 0x044E ю (CYRILLIC SMALL LETTER YU)
|
||||
111 0x044F я (CYRILLIC SMALL LETTER YA)
|
||||
112 0x0401 Ё (CYRILLIC CAPITAL LETTER IO)
|
||||
113 0x0451 ё (CYRILLIC SMALL LETTER IO)
|
||||
114 0x0404 Є (CYRILLIC CAPITAL LETTER UKRAINIAN IE)
|
||||
115 0x0454 є (CYRILLIC SMALL LETTER UKRAINIAN IE)
|
||||
116 0x0407 Ї (CYRILLIC CAPITAL LETTER YI)
|
||||
117 0x0457 ї (CYRILLIC SMALL LETTER YI)
|
||||
118 0x040E Ў (CYRILLIC CAPITAL LETTER SHORT U)
|
||||
119 0x045E ў (CYRILLIC SMALL LETTER SHORT U)
|
||||
120 0x00B0 ° (DEGREE SIGN)
|
||||
121 0x2219 ∙ (BULLET OPERATOR)
|
||||
122 0x00B7 · (MIDDLE DOT)
|
||||
123 0x221A √ (SQUARE ROOT)
|
||||
124 0x2116 № (NUMERO SIGN)
|
||||
125 0x00A4 ¤ (CURRENCY SIGN)
|
||||
126 0x25A0 ■ (BLACK SQUARE)
|
||||
127 0x00A0 (NO-BREAK SPACE)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-10.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-10.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-10.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 02c2b5590d8ccda9931008c471f6ee2c590b2c8fe5e6ccb3b08638115d778507
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
34 0x0112 Ē (LATIN CAPITAL LETTER E WITH MACRON)
|
||||
35 0x0122 Ģ (LATIN CAPITAL LETTER G WITH CEDILLA)
|
||||
36 0x012A Ī (LATIN CAPITAL LETTER I WITH MACRON)
|
||||
37 0x0128 Ĩ (LATIN CAPITAL LETTER I WITH TILDE)
|
||||
38 0x0136 Ķ (LATIN CAPITAL LETTER K WITH CEDILLA)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x013B Ļ (LATIN CAPITAL LETTER L WITH CEDILLA)
|
||||
41 0x0110 Đ (LATIN CAPITAL LETTER D WITH STROKE)
|
||||
42 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
43 0x0166 Ŧ (LATIN CAPITAL LETTER T WITH STROKE)
|
||||
44 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x016A Ū (LATIN CAPITAL LETTER U WITH MACRON)
|
||||
47 0x014A Ŋ (LATIN CAPITAL LETTER ENG)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
50 0x0113 ē (LATIN SMALL LETTER E WITH MACRON)
|
||||
51 0x0123 ģ (LATIN SMALL LETTER G WITH CEDILLA)
|
||||
52 0x012B ī (LATIN SMALL LETTER I WITH MACRON)
|
||||
53 0x0129 ĩ (LATIN SMALL LETTER I WITH TILDE)
|
||||
54 0x0137 ķ (LATIN SMALL LETTER K WITH CEDILLA)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x013C ļ (LATIN SMALL LETTER L WITH CEDILLA)
|
||||
57 0x0111 đ (LATIN SMALL LETTER D WITH STROKE)
|
||||
58 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
59 0x0167 ŧ (LATIN SMALL LETTER T WITH STROKE)
|
||||
60 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
61 0x2015 ― (HORIZONTAL BAR)
|
||||
62 0x016B ū (LATIN SMALL LETTER U WITH MACRON)
|
||||
63 0x014B ŋ (LATIN SMALL LETTER ENG)
|
||||
64 0x0100 Ā (LATIN CAPITAL LETTER A WITH MACRON)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x012E Į (LATIN CAPITAL LETTER I WITH OGONEK)
|
||||
72 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x0116 Ė (LATIN CAPITAL LETTER E WITH DOT ABOVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x00D0 Ð (LATIN CAPITAL LETTER ETH)
|
||||
81 0x0145 Ņ (LATIN CAPITAL LETTER N WITH CEDILLA)
|
||||
82 0x014C Ō (LATIN CAPITAL LETTER O WITH MACRON)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x0168 Ũ (LATIN CAPITAL LETTER U WITH TILDE)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x0172 Ų (LATIN CAPITAL LETTER U WITH OGONEK)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x00DD Ý (LATIN CAPITAL LETTER Y WITH ACUTE)
|
||||
94 0x00DE Þ (LATIN CAPITAL LETTER THORN)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x0101 ā (LATIN SMALL LETTER A WITH MACRON)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x012F į (LATIN SMALL LETTER I WITH OGONEK)
|
||||
104 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x0117 ė (LATIN SMALL LETTER E WITH DOT ABOVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x00F0 ð (LATIN SMALL LETTER ETH)
|
||||
113 0x0146 ņ (LATIN SMALL LETTER N WITH CEDILLA)
|
||||
114 0x014D ō (LATIN SMALL LETTER O WITH MACRON)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x0169 ũ (LATIN SMALL LETTER U WITH TILDE)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x0173 ų (LATIN SMALL LETTER U WITH OGONEK)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x00FD ý (LATIN SMALL LETTER Y WITH ACUTE)
|
||||
126 0x00FE þ (LATIN SMALL LETTER THORN)
|
||||
127 0x0138 ĸ (LATIN SMALL LETTER KRA)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-13.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-13.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-13.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 40736338e964ab520407cebcb01329f8d450abf6ce12bf88b74b655b60e43300
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x0156 Ŗ (LATIN CAPITAL LETTER R WITH CEDILLA)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x0157 ŗ (LATIN SMALL LETTER R WITH CEDILLA)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
64 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
65 0x012E Į (LATIN CAPITAL LETTER I WITH OGONEK)
|
||||
66 0x0100 Ā (LATIN CAPITAL LETTER A WITH MACRON)
|
||||
67 0x0106 Ć (LATIN CAPITAL LETTER C WITH ACUTE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
71 0x0112 Ē (LATIN CAPITAL LETTER E WITH MACRON)
|
||||
72 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x0179 Ź (LATIN CAPITAL LETTER Z WITH ACUTE)
|
||||
75 0x0116 Ė (LATIN CAPITAL LETTER E WITH DOT ABOVE)
|
||||
76 0x0122 Ģ (LATIN CAPITAL LETTER G WITH CEDILLA)
|
||||
77 0x0136 Ķ (LATIN CAPITAL LETTER K WITH CEDILLA)
|
||||
78 0x012A Ī (LATIN CAPITAL LETTER I WITH MACRON)
|
||||
79 0x013B Ļ (LATIN CAPITAL LETTER L WITH CEDILLA)
|
||||
80 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
81 0x0143 Ń (LATIN CAPITAL LETTER N WITH ACUTE)
|
||||
82 0x0145 Ņ (LATIN CAPITAL LETTER N WITH CEDILLA)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x014C Ō (LATIN CAPITAL LETTER O WITH MACRON)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x0172 Ų (LATIN CAPITAL LETTER U WITH OGONEK)
|
||||
89 0x0141 Ł (LATIN CAPITAL LETTER L WITH STROKE)
|
||||
90 0x015A Ś (LATIN CAPITAL LETTER S WITH ACUTE)
|
||||
91 0x016A Ū (LATIN CAPITAL LETTER U WITH MACRON)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x017B Ż (LATIN CAPITAL LETTER Z WITH DOT ABOVE)
|
||||
94 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
97 0x012F į (LATIN SMALL LETTER I WITH OGONEK)
|
||||
98 0x0101 ā (LATIN SMALL LETTER A WITH MACRON)
|
||||
99 0x0107 ć (LATIN SMALL LETTER C WITH ACUTE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
103 0x0113 ē (LATIN SMALL LETTER E WITH MACRON)
|
||||
104 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x017A ź (LATIN SMALL LETTER Z WITH ACUTE)
|
||||
107 0x0117 ė (LATIN SMALL LETTER E WITH DOT ABOVE)
|
||||
108 0x0123 ģ (LATIN SMALL LETTER G WITH CEDILLA)
|
||||
109 0x0137 ķ (LATIN SMALL LETTER K WITH CEDILLA)
|
||||
110 0x012B ī (LATIN SMALL LETTER I WITH MACRON)
|
||||
111 0x013C ļ (LATIN SMALL LETTER L WITH CEDILLA)
|
||||
112 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
113 0x0144 ń (LATIN SMALL LETTER N WITH ACUTE)
|
||||
114 0x0146 ņ (LATIN SMALL LETTER N WITH CEDILLA)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x014D ō (LATIN SMALL LETTER O WITH MACRON)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x0173 ų (LATIN SMALL LETTER U WITH OGONEK)
|
||||
121 0x0142 ł (LATIN SMALL LETTER L WITH STROKE)
|
||||
122 0x015B ś (LATIN SMALL LETTER S WITH ACUTE)
|
||||
123 0x016B ū (LATIN SMALL LETTER U WITH MACRON)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x017C ż (LATIN SMALL LETTER Z WITH DOT ABOVE)
|
||||
126 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
127 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-14.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-14.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-14.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 2c8651cfc08b1f35b17919ee5379f2fa006af3ec809f11b3b7f470785580542b
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x1E02 Ḃ (LATIN CAPITAL LETTER B WITH DOT ABOVE)
|
||||
34 0x1E03 ḃ (LATIN SMALL LETTER B WITH DOT ABOVE)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x010A Ċ (LATIN CAPITAL LETTER C WITH DOT ABOVE)
|
||||
37 0x010B ċ (LATIN SMALL LETTER C WITH DOT ABOVE)
|
||||
38 0x1E0A Ḋ (LATIN CAPITAL LETTER D WITH DOT ABOVE)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x1E80 Ẁ (LATIN CAPITAL LETTER W WITH GRAVE)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x1E82 Ẃ (LATIN CAPITAL LETTER W WITH ACUTE)
|
||||
43 0x1E0B ḋ (LATIN SMALL LETTER D WITH DOT ABOVE)
|
||||
44 0x1EF2 Ỳ (LATIN CAPITAL LETTER Y WITH GRAVE)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
48 0x1E1E Ḟ (LATIN CAPITAL LETTER F WITH DOT ABOVE)
|
||||
49 0x1E1F ḟ (LATIN SMALL LETTER F WITH DOT ABOVE)
|
||||
50 0x0120 Ġ (LATIN CAPITAL LETTER G WITH DOT ABOVE)
|
||||
51 0x0121 ġ (LATIN SMALL LETTER G WITH DOT ABOVE)
|
||||
52 0x1E40 Ṁ (LATIN CAPITAL LETTER M WITH DOT ABOVE)
|
||||
53 0x1E41 ṁ (LATIN SMALL LETTER M WITH DOT ABOVE)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x1E56 Ṗ (LATIN CAPITAL LETTER P WITH DOT ABOVE)
|
||||
56 0x1E81 ẁ (LATIN SMALL LETTER W WITH GRAVE)
|
||||
57 0x1E57 ṗ (LATIN SMALL LETTER P WITH DOT ABOVE)
|
||||
58 0x1E83 ẃ (LATIN SMALL LETTER W WITH ACUTE)
|
||||
59 0x1E60 Ṡ (LATIN CAPITAL LETTER S WITH DOT ABOVE)
|
||||
60 0x1EF3 ỳ (LATIN SMALL LETTER Y WITH GRAVE)
|
||||
61 0x1E84 Ẅ (LATIN CAPITAL LETTER W WITH DIAERESIS)
|
||||
62 0x1E85 ẅ (LATIN SMALL LETTER W WITH DIAERESIS)
|
||||
63 0x1E61 ṡ (LATIN SMALL LETTER S WITH DOT ABOVE)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x0174 Ŵ (LATIN CAPITAL LETTER W WITH CIRCUMFLEX)
|
||||
81 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
82 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x1E6A Ṫ (LATIN CAPITAL LETTER T WITH DOT ABOVE)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x00DD Ý (LATIN CAPITAL LETTER Y WITH ACUTE)
|
||||
94 0x0176 Ŷ (LATIN CAPITAL LETTER Y WITH CIRCUMFLEX)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x0175 ŵ (LATIN SMALL LETTER W WITH CIRCUMFLEX)
|
||||
113 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
114 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x1E6B ṫ (LATIN SMALL LETTER T WITH DOT ABOVE)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x00FD ý (LATIN SMALL LETTER Y WITH ACUTE)
|
||||
126 0x0177 ŷ (LATIN SMALL LETTER Y WITH CIRCUMFLEX)
|
||||
127 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-15.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-15.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-15.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: a560aba47bccd7510a6ac77f671fe75dca3800f05cf6d676910c311a8f8ff079
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x00A1 ¡ (INVERTED EXCLAMATION MARK)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x20AC € (EURO SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x00AA ª (FEMININE ORDINAL INDICATOR)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x00BA º (MASCULINE ORDINAL INDICATOR)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
61 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
62 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
63 0x00BF ¿ (INVERTED QUESTION MARK)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x00D0 Ð (LATIN CAPITAL LETTER ETH)
|
||||
81 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
82 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x00DD Ý (LATIN CAPITAL LETTER Y WITH ACUTE)
|
||||
94 0x00DE Þ (LATIN CAPITAL LETTER THORN)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x00F0 ð (LATIN SMALL LETTER ETH)
|
||||
113 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
114 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x00FD ý (LATIN SMALL LETTER Y WITH ACUTE)
|
||||
126 0x00FE þ (LATIN SMALL LETTER THORN)
|
||||
127 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-16.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-16.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-16.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 55676320d2d1b6e6909f5b3d741a7cf0cefc84e920aa4474afc091459111c2e3
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
34 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
35 0x0141 Ł (LATIN CAPITAL LETTER L WITH STROKE)
|
||||
36 0x20AC € (EURO SIGN)
|
||||
37 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
38 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x0218 Ș (LATIN CAPITAL LETTER S WITH COMMA BELOW)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x0179 Ź (LATIN CAPITAL LETTER Z WITH ACUTE)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x017A ź (LATIN SMALL LETTER Z WITH ACUTE)
|
||||
47 0x017B Ż (LATIN CAPITAL LETTER Z WITH DOT ABOVE)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
51 0x0142 ł (LATIN SMALL LETTER L WITH STROKE)
|
||||
52 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
53 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
57 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
58 0x0219 ș (LATIN SMALL LETTER S WITH COMMA BELOW)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
61 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
62 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
63 0x017C ż (LATIN SMALL LETTER Z WITH DOT ABOVE)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x0102 Ă (LATIN CAPITAL LETTER A WITH BREVE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x0106 Ć (LATIN CAPITAL LETTER C WITH ACUTE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x0110 Đ (LATIN CAPITAL LETTER D WITH STROKE)
|
||||
81 0x0143 Ń (LATIN CAPITAL LETTER N WITH ACUTE)
|
||||
82 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x0150 Ő (LATIN CAPITAL LETTER O WITH DOUBLE ACUTE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x015A Ś (LATIN CAPITAL LETTER S WITH ACUTE)
|
||||
88 0x0170 Ű (LATIN CAPITAL LETTER U WITH DOUBLE ACUTE)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
94 0x021A Ț (LATIN CAPITAL LETTER T WITH COMMA BELOW)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x0103 ă (LATIN SMALL LETTER A WITH BREVE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x0107 ć (LATIN SMALL LETTER C WITH ACUTE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x0111 đ (LATIN SMALL LETTER D WITH STROKE)
|
||||
113 0x0144 ń (LATIN SMALL LETTER N WITH ACUTE)
|
||||
114 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x0151 ő (LATIN SMALL LETTER O WITH DOUBLE ACUTE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x015B ś (LATIN SMALL LETTER S WITH ACUTE)
|
||||
120 0x0171 ű (LATIN SMALL LETTER U WITH DOUBLE ACUTE)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
126 0x021B ț (LATIN SMALL LETTER T WITH COMMA BELOW)
|
||||
127 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-2.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-2.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-2.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 9569c67f22d0b57790e1c407c6eecf227e4562322dc296de43cdab7a0152ec73
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
34 0x02D8 ˘ (BREVE)
|
||||
35 0x0141 Ł (LATIN CAPITAL LETTER L WITH STROKE)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x013D Ľ (LATIN CAPITAL LETTER L WITH CARON)
|
||||
38 0x015A Ś (LATIN CAPITAL LETTER S WITH ACUTE)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
42 0x015E Ş (LATIN CAPITAL LETTER S WITH CEDILLA)
|
||||
43 0x0164 Ť (LATIN CAPITAL LETTER T WITH CARON)
|
||||
44 0x0179 Ź (LATIN CAPITAL LETTER Z WITH ACUTE)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
47 0x017B Ż (LATIN CAPITAL LETTER Z WITH DOT ABOVE)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
50 0x02DB ˛ (OGONEK)
|
||||
51 0x0142 ł (LATIN SMALL LETTER L WITH STROKE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x013E ľ (LATIN SMALL LETTER L WITH CARON)
|
||||
54 0x015B ś (LATIN SMALL LETTER S WITH ACUTE)
|
||||
55 0x02C7 ˇ (CARON)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
58 0x015F ş (LATIN SMALL LETTER S WITH CEDILLA)
|
||||
59 0x0165 ť (LATIN SMALL LETTER T WITH CARON)
|
||||
60 0x017A ź (LATIN SMALL LETTER Z WITH ACUTE)
|
||||
61 0x02DD ˝ (DOUBLE ACUTE ACCENT)
|
||||
62 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
63 0x017C ż (LATIN SMALL LETTER Z WITH DOT ABOVE)
|
||||
64 0x0154 Ŕ (LATIN CAPITAL LETTER R WITH ACUTE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x0102 Ă (LATIN CAPITAL LETTER A WITH BREVE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x0139 Ĺ (LATIN CAPITAL LETTER L WITH ACUTE)
|
||||
70 0x0106 Ć (LATIN CAPITAL LETTER C WITH ACUTE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x011A Ě (LATIN CAPITAL LETTER E WITH CARON)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x010E Ď (LATIN CAPITAL LETTER D WITH CARON)
|
||||
80 0x0110 Đ (LATIN CAPITAL LETTER D WITH STROKE)
|
||||
81 0x0143 Ń (LATIN CAPITAL LETTER N WITH ACUTE)
|
||||
82 0x0147 Ň (LATIN CAPITAL LETTER N WITH CARON)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x0150 Ő (LATIN CAPITAL LETTER O WITH DOUBLE ACUTE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x0158 Ř (LATIN CAPITAL LETTER R WITH CARON)
|
||||
89 0x016E Ů (LATIN CAPITAL LETTER U WITH RING ABOVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x0170 Ű (LATIN CAPITAL LETTER U WITH DOUBLE ACUTE)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x00DD Ý (LATIN CAPITAL LETTER Y WITH ACUTE)
|
||||
94 0x0162 Ţ (LATIN CAPITAL LETTER T WITH CEDILLA)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x0155 ŕ (LATIN SMALL LETTER R WITH ACUTE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x0103 ă (LATIN SMALL LETTER A WITH BREVE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x013A ĺ (LATIN SMALL LETTER L WITH ACUTE)
|
||||
102 0x0107 ć (LATIN SMALL LETTER C WITH ACUTE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x011B ě (LATIN SMALL LETTER E WITH CARON)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x010F ď (LATIN SMALL LETTER D WITH CARON)
|
||||
112 0x0111 đ (LATIN SMALL LETTER D WITH STROKE)
|
||||
113 0x0144 ń (LATIN SMALL LETTER N WITH ACUTE)
|
||||
114 0x0148 ň (LATIN SMALL LETTER N WITH CARON)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x0151 ő (LATIN SMALL LETTER O WITH DOUBLE ACUTE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x0159 ř (LATIN SMALL LETTER R WITH CARON)
|
||||
121 0x016F ů (LATIN SMALL LETTER U WITH RING ABOVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x0171 ű (LATIN SMALL LETTER U WITH DOUBLE ACUTE)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x00FD ý (LATIN SMALL LETTER Y WITH ACUTE)
|
||||
126 0x0163 ţ (LATIN SMALL LETTER T WITH CEDILLA)
|
||||
127 0x02D9 ˙ (DOT ABOVE)
|
||||
127
test/fixtures/encoding/single-byte/index-iso-8859-3.txt
vendored
Normal file
127
test/fixtures/encoding/single-byte/index-iso-8859-3.txt
vendored
Normal file
@@ -0,0 +1,127 @@
|
||||
# For details on index index-iso-8859-3.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: af8f1e12df79b768322b5e83613698cdc619438270a2fc359554331c805054a3
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0126 Ħ (LATIN CAPITAL LETTER H WITH STROKE)
|
||||
34 0x02D8 ˘ (BREVE)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
38 0x0124 Ĥ (LATIN CAPITAL LETTER H WITH CIRCUMFLEX)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x0130 İ (LATIN CAPITAL LETTER I WITH DOT ABOVE)
|
||||
42 0x015E Ş (LATIN CAPITAL LETTER S WITH CEDILLA)
|
||||
43 0x011E Ğ (LATIN CAPITAL LETTER G WITH BREVE)
|
||||
44 0x0134 Ĵ (LATIN CAPITAL LETTER J WITH CIRCUMFLEX)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
47 0x017B Ż (LATIN CAPITAL LETTER Z WITH DOT ABOVE)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x0127 ħ (LATIN SMALL LETTER H WITH STROKE)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x0125 ĥ (LATIN SMALL LETTER H WITH CIRCUMFLEX)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x0131 ı (LATIN SMALL LETTER DOTLESS I)
|
||||
58 0x015F ş (LATIN SMALL LETTER S WITH CEDILLA)
|
||||
59 0x011F ğ (LATIN SMALL LETTER G WITH BREVE)
|
||||
60 0x0135 ĵ (LATIN SMALL LETTER J WITH CIRCUMFLEX)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
63 0x017C ż (LATIN SMALL LETTER Z WITH DOT ABOVE)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x010A Ċ (LATIN CAPITAL LETTER C WITH DOT ABOVE)
|
||||
70 0x0108 Ĉ (LATIN CAPITAL LETTER C WITH CIRCUMFLEX)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
81 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
82 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x0120 Ġ (LATIN CAPITAL LETTER G WITH DOT ABOVE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x011C Ĝ (LATIN CAPITAL LETTER G WITH CIRCUMFLEX)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x016C Ŭ (LATIN CAPITAL LETTER U WITH BREVE)
|
||||
94 0x015C Ŝ (LATIN CAPITAL LETTER S WITH CIRCUMFLEX)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x010B ċ (LATIN SMALL LETTER C WITH DOT ABOVE)
|
||||
102 0x0109 ĉ (LATIN SMALL LETTER C WITH CIRCUMFLEX)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
113 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
114 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x0121 ġ (LATIN SMALL LETTER G WITH DOT ABOVE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x011D ĝ (LATIN SMALL LETTER G WITH CIRCUMFLEX)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x016D ŭ (LATIN SMALL LETTER U WITH BREVE)
|
||||
126 0x015D ŝ (LATIN SMALL LETTER S WITH CIRCUMFLEX)
|
||||
127 0x02D9 ˙ (DOT ABOVE)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-4.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-4.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-4.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 72f29c92344d351fe9e74a946e7e0468d76d542c6894ff82982cb652ebe0feb7
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
34 0x0138 ĸ (LATIN SMALL LETTER KRA)
|
||||
35 0x0156 Ŗ (LATIN CAPITAL LETTER R WITH CEDILLA)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x0128 Ĩ (LATIN CAPITAL LETTER I WITH TILDE)
|
||||
38 0x013B Ļ (LATIN CAPITAL LETTER L WITH CEDILLA)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
42 0x0112 Ē (LATIN CAPITAL LETTER E WITH MACRON)
|
||||
43 0x0122 Ģ (LATIN CAPITAL LETTER G WITH CEDILLA)
|
||||
44 0x0166 Ŧ (LATIN CAPITAL LETTER T WITH STROKE)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
50 0x02DB ˛ (OGONEK)
|
||||
51 0x0157 ŗ (LATIN SMALL LETTER R WITH CEDILLA)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x0129 ĩ (LATIN SMALL LETTER I WITH TILDE)
|
||||
54 0x013C ļ (LATIN SMALL LETTER L WITH CEDILLA)
|
||||
55 0x02C7 ˇ (CARON)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
58 0x0113 ē (LATIN SMALL LETTER E WITH MACRON)
|
||||
59 0x0123 ģ (LATIN SMALL LETTER G WITH CEDILLA)
|
||||
60 0x0167 ŧ (LATIN SMALL LETTER T WITH STROKE)
|
||||
61 0x014A Ŋ (LATIN CAPITAL LETTER ENG)
|
||||
62 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
63 0x014B ŋ (LATIN SMALL LETTER ENG)
|
||||
64 0x0100 Ā (LATIN CAPITAL LETTER A WITH MACRON)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x012E Į (LATIN CAPITAL LETTER I WITH OGONEK)
|
||||
72 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x0116 Ė (LATIN CAPITAL LETTER E WITH DOT ABOVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x012A Ī (LATIN CAPITAL LETTER I WITH MACRON)
|
||||
80 0x0110 Đ (LATIN CAPITAL LETTER D WITH STROKE)
|
||||
81 0x0145 Ņ (LATIN CAPITAL LETTER N WITH CEDILLA)
|
||||
82 0x014C Ō (LATIN CAPITAL LETTER O WITH MACRON)
|
||||
83 0x0136 Ķ (LATIN CAPITAL LETTER K WITH CEDILLA)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x0172 Ų (LATIN CAPITAL LETTER U WITH OGONEK)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x0168 Ũ (LATIN CAPITAL LETTER U WITH TILDE)
|
||||
94 0x016A Ū (LATIN CAPITAL LETTER U WITH MACRON)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x0101 ā (LATIN SMALL LETTER A WITH MACRON)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x012F į (LATIN SMALL LETTER I WITH OGONEK)
|
||||
104 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x0117 ė (LATIN SMALL LETTER E WITH DOT ABOVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x012B ī (LATIN SMALL LETTER I WITH MACRON)
|
||||
112 0x0111 đ (LATIN SMALL LETTER D WITH STROKE)
|
||||
113 0x0146 ņ (LATIN SMALL LETTER N WITH CEDILLA)
|
||||
114 0x014D ō (LATIN SMALL LETTER O WITH MACRON)
|
||||
115 0x0137 ķ (LATIN SMALL LETTER K WITH CEDILLA)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x0173 ų (LATIN SMALL LETTER U WITH OGONEK)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x0169 ũ (LATIN SMALL LETTER U WITH TILDE)
|
||||
126 0x016B ū (LATIN SMALL LETTER U WITH MACRON)
|
||||
127 0x02D9 ˙ (DOT ABOVE)
|
||||
134
test/fixtures/encoding/single-byte/index-iso-8859-5.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-iso-8859-5.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-iso-8859-5.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: fa9b1f3f5242df43e2e7bca80e9b6997c67944f20a4af91ee06bacc4e132d9c9
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0401 Ё (CYRILLIC CAPITAL LETTER IO)
|
||||
34 0x0402 Ђ (CYRILLIC CAPITAL LETTER DJE)
|
||||
35 0x0403 Ѓ (CYRILLIC CAPITAL LETTER GJE)
|
||||
36 0x0404 Є (CYRILLIC CAPITAL LETTER UKRAINIAN IE)
|
||||
37 0x0405 Ѕ (CYRILLIC CAPITAL LETTER DZE)
|
||||
38 0x0406 І (CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
39 0x0407 Ї (CYRILLIC CAPITAL LETTER YI)
|
||||
40 0x0408 Ј (CYRILLIC CAPITAL LETTER JE)
|
||||
41 0x0409 Љ (CYRILLIC CAPITAL LETTER LJE)
|
||||
42 0x040A Њ (CYRILLIC CAPITAL LETTER NJE)
|
||||
43 0x040B Ћ (CYRILLIC CAPITAL LETTER TSHE)
|
||||
44 0x040C Ќ (CYRILLIC CAPITAL LETTER KJE)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x040E Ў (CYRILLIC CAPITAL LETTER SHORT U)
|
||||
47 0x040F Џ (CYRILLIC CAPITAL LETTER DZHE)
|
||||
48 0x0410 А (CYRILLIC CAPITAL LETTER A)
|
||||
49 0x0411 Б (CYRILLIC CAPITAL LETTER BE)
|
||||
50 0x0412 В (CYRILLIC CAPITAL LETTER VE)
|
||||
51 0x0413 Г (CYRILLIC CAPITAL LETTER GHE)
|
||||
52 0x0414 Д (CYRILLIC CAPITAL LETTER DE)
|
||||
53 0x0415 Е (CYRILLIC CAPITAL LETTER IE)
|
||||
54 0x0416 Ж (CYRILLIC CAPITAL LETTER ZHE)
|
||||
55 0x0417 З (CYRILLIC CAPITAL LETTER ZE)
|
||||
56 0x0418 И (CYRILLIC CAPITAL LETTER I)
|
||||
57 0x0419 Й (CYRILLIC CAPITAL LETTER SHORT I)
|
||||
58 0x041A К (CYRILLIC CAPITAL LETTER KA)
|
||||
59 0x041B Л (CYRILLIC CAPITAL LETTER EL)
|
||||
60 0x041C М (CYRILLIC CAPITAL LETTER EM)
|
||||
61 0x041D Н (CYRILLIC CAPITAL LETTER EN)
|
||||
62 0x041E О (CYRILLIC CAPITAL LETTER O)
|
||||
63 0x041F П (CYRILLIC CAPITAL LETTER PE)
|
||||
64 0x0420 Р (CYRILLIC CAPITAL LETTER ER)
|
||||
65 0x0421 С (CYRILLIC CAPITAL LETTER ES)
|
||||
66 0x0422 Т (CYRILLIC CAPITAL LETTER TE)
|
||||
67 0x0423 У (CYRILLIC CAPITAL LETTER U)
|
||||
68 0x0424 Ф (CYRILLIC CAPITAL LETTER EF)
|
||||
69 0x0425 Х (CYRILLIC CAPITAL LETTER HA)
|
||||
70 0x0426 Ц (CYRILLIC CAPITAL LETTER TSE)
|
||||
71 0x0427 Ч (CYRILLIC CAPITAL LETTER CHE)
|
||||
72 0x0428 Ш (CYRILLIC CAPITAL LETTER SHA)
|
||||
73 0x0429 Щ (CYRILLIC CAPITAL LETTER SHCHA)
|
||||
74 0x042A Ъ (CYRILLIC CAPITAL LETTER HARD SIGN)
|
||||
75 0x042B Ы (CYRILLIC CAPITAL LETTER YERU)
|
||||
76 0x042C Ь (CYRILLIC CAPITAL LETTER SOFT SIGN)
|
||||
77 0x042D Э (CYRILLIC CAPITAL LETTER E)
|
||||
78 0x042E Ю (CYRILLIC CAPITAL LETTER YU)
|
||||
79 0x042F Я (CYRILLIC CAPITAL LETTER YA)
|
||||
80 0x0430 а (CYRILLIC SMALL LETTER A)
|
||||
81 0x0431 б (CYRILLIC SMALL LETTER BE)
|
||||
82 0x0432 в (CYRILLIC SMALL LETTER VE)
|
||||
83 0x0433 г (CYRILLIC SMALL LETTER GHE)
|
||||
84 0x0434 д (CYRILLIC SMALL LETTER DE)
|
||||
85 0x0435 е (CYRILLIC SMALL LETTER IE)
|
||||
86 0x0436 ж (CYRILLIC SMALL LETTER ZHE)
|
||||
87 0x0437 з (CYRILLIC SMALL LETTER ZE)
|
||||
88 0x0438 и (CYRILLIC SMALL LETTER I)
|
||||
89 0x0439 й (CYRILLIC SMALL LETTER SHORT I)
|
||||
90 0x043A к (CYRILLIC SMALL LETTER KA)
|
||||
91 0x043B л (CYRILLIC SMALL LETTER EL)
|
||||
92 0x043C м (CYRILLIC SMALL LETTER EM)
|
||||
93 0x043D н (CYRILLIC SMALL LETTER EN)
|
||||
94 0x043E о (CYRILLIC SMALL LETTER O)
|
||||
95 0x043F п (CYRILLIC SMALL LETTER PE)
|
||||
96 0x0440 р (CYRILLIC SMALL LETTER ER)
|
||||
97 0x0441 с (CYRILLIC SMALL LETTER ES)
|
||||
98 0x0442 т (CYRILLIC SMALL LETTER TE)
|
||||
99 0x0443 у (CYRILLIC SMALL LETTER U)
|
||||
100 0x0444 ф (CYRILLIC SMALL LETTER EF)
|
||||
101 0x0445 х (CYRILLIC SMALL LETTER HA)
|
||||
102 0x0446 ц (CYRILLIC SMALL LETTER TSE)
|
||||
103 0x0447 ч (CYRILLIC SMALL LETTER CHE)
|
||||
104 0x0448 ш (CYRILLIC SMALL LETTER SHA)
|
||||
105 0x0449 щ (CYRILLIC SMALL LETTER SHCHA)
|
||||
106 0x044A ъ (CYRILLIC SMALL LETTER HARD SIGN)
|
||||
107 0x044B ы (CYRILLIC SMALL LETTER YERU)
|
||||
108 0x044C ь (CYRILLIC SMALL LETTER SOFT SIGN)
|
||||
109 0x044D э (CYRILLIC SMALL LETTER E)
|
||||
110 0x044E ю (CYRILLIC SMALL LETTER YU)
|
||||
111 0x044F я (CYRILLIC SMALL LETTER YA)
|
||||
112 0x2116 № (NUMERO SIGN)
|
||||
113 0x0451 ё (CYRILLIC SMALL LETTER IO)
|
||||
114 0x0452 ђ (CYRILLIC SMALL LETTER DJE)
|
||||
115 0x0453 ѓ (CYRILLIC SMALL LETTER GJE)
|
||||
116 0x0454 є (CYRILLIC SMALL LETTER UKRAINIAN IE)
|
||||
117 0x0455 ѕ (CYRILLIC SMALL LETTER DZE)
|
||||
118 0x0456 і (CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
119 0x0457 ї (CYRILLIC SMALL LETTER YI)
|
||||
120 0x0458 ј (CYRILLIC SMALL LETTER JE)
|
||||
121 0x0459 љ (CYRILLIC SMALL LETTER LJE)
|
||||
122 0x045A њ (CYRILLIC SMALL LETTER NJE)
|
||||
123 0x045B ћ (CYRILLIC SMALL LETTER TSHE)
|
||||
124 0x045C ќ (CYRILLIC SMALL LETTER KJE)
|
||||
125 0x00A7 § (SECTION SIGN)
|
||||
126 0x045E ў (CYRILLIC SMALL LETTER SHORT U)
|
||||
127 0x045F џ (CYRILLIC SMALL LETTER DZHE)
|
||||
89
test/fixtures/encoding/single-byte/index-iso-8859-6.txt
vendored
Normal file
89
test/fixtures/encoding/single-byte/index-iso-8859-6.txt
vendored
Normal file
@@ -0,0 +1,89 @@
|
||||
# For details on index index-iso-8859-6.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 85bb7b5c2dc75975afebe5743935ba4ed5a09c1e9e34e9bfb2ff80293f5d8bbc
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
44 0x060C ، (ARABIC COMMA)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
59 0x061B ؛ (ARABIC SEMICOLON)
|
||||
63 0x061F ؟ (ARABIC QUESTION MARK)
|
||||
65 0x0621 ء (ARABIC LETTER HAMZA)
|
||||
66 0x0622 آ (ARABIC LETTER ALEF WITH MADDA ABOVE)
|
||||
67 0x0623 أ (ARABIC LETTER ALEF WITH HAMZA ABOVE)
|
||||
68 0x0624 ؤ (ARABIC LETTER WAW WITH HAMZA ABOVE)
|
||||
69 0x0625 إ (ARABIC LETTER ALEF WITH HAMZA BELOW)
|
||||
70 0x0626 ئ (ARABIC LETTER YEH WITH HAMZA ABOVE)
|
||||
71 0x0627 ا (ARABIC LETTER ALEF)
|
||||
72 0x0628 ب (ARABIC LETTER BEH)
|
||||
73 0x0629 ة (ARABIC LETTER TEH MARBUTA)
|
||||
74 0x062A ت (ARABIC LETTER TEH)
|
||||
75 0x062B ث (ARABIC LETTER THEH)
|
||||
76 0x062C ج (ARABIC LETTER JEEM)
|
||||
77 0x062D ح (ARABIC LETTER HAH)
|
||||
78 0x062E خ (ARABIC LETTER KHAH)
|
||||
79 0x062F د (ARABIC LETTER DAL)
|
||||
80 0x0630 ذ (ARABIC LETTER THAL)
|
||||
81 0x0631 ر (ARABIC LETTER REH)
|
||||
82 0x0632 ز (ARABIC LETTER ZAIN)
|
||||
83 0x0633 س (ARABIC LETTER SEEN)
|
||||
84 0x0634 ش (ARABIC LETTER SHEEN)
|
||||
85 0x0635 ص (ARABIC LETTER SAD)
|
||||
86 0x0636 ض (ARABIC LETTER DAD)
|
||||
87 0x0637 ط (ARABIC LETTER TAH)
|
||||
88 0x0638 ظ (ARABIC LETTER ZAH)
|
||||
89 0x0639 ع (ARABIC LETTER AIN)
|
||||
90 0x063A غ (ARABIC LETTER GHAIN)
|
||||
96 0x0640 ـ (ARABIC TATWEEL)
|
||||
97 0x0641 ف (ARABIC LETTER FEH)
|
||||
98 0x0642 ق (ARABIC LETTER QAF)
|
||||
99 0x0643 ك (ARABIC LETTER KAF)
|
||||
100 0x0644 ل (ARABIC LETTER LAM)
|
||||
101 0x0645 م (ARABIC LETTER MEEM)
|
||||
102 0x0646 ن (ARABIC LETTER NOON)
|
||||
103 0x0647 ه (ARABIC LETTER HEH)
|
||||
104 0x0648 و (ARABIC LETTER WAW)
|
||||
105 0x0649 ى (ARABIC LETTER ALEF MAKSURA)
|
||||
106 0x064A ي (ARABIC LETTER YEH)
|
||||
107 0x064B ً (ARABIC FATHATAN)
|
||||
108 0x064C ٌ (ARABIC DAMMATAN)
|
||||
109 0x064D ٍ (ARABIC KASRATAN)
|
||||
110 0x064E َ (ARABIC FATHA)
|
||||
111 0x064F ُ (ARABIC DAMMA)
|
||||
112 0x0650 ِ (ARABIC KASRA)
|
||||
113 0x0651 ّ (ARABIC SHADDA)
|
||||
114 0x0652 ْ (ARABIC SUKUN)
|
||||
131
test/fixtures/encoding/single-byte/index-iso-8859-7.txt
vendored
Normal file
131
test/fixtures/encoding/single-byte/index-iso-8859-7.txt
vendored
Normal file
@@ -0,0 +1,131 @@
|
||||
# For details on index index-iso-8859-7.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: f53d8aeba36314ef950eef02ffcf11dff540638ce27dfe7a86b6ccc6875afb24
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
34 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x20AC € (EURO SIGN)
|
||||
37 0x20AF ₯ (DRACHMA SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x037A ͺ (GREEK YPOGEGRAMMENI)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
47 0x2015 ― (HORIZONTAL BAR)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x0384 ΄ (GREEK TONOS)
|
||||
53 0x0385 ΅ (GREEK DIALYTIKA TONOS)
|
||||
54 0x0386 Ά (GREEK CAPITAL LETTER ALPHA WITH TONOS)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x0388 Έ (GREEK CAPITAL LETTER EPSILON WITH TONOS)
|
||||
57 0x0389 Ή (GREEK CAPITAL LETTER ETA WITH TONOS)
|
||||
58 0x038A Ί (GREEK CAPITAL LETTER IOTA WITH TONOS)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x038C Ό (GREEK CAPITAL LETTER OMICRON WITH TONOS)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x038E Ύ (GREEK CAPITAL LETTER UPSILON WITH TONOS)
|
||||
63 0x038F Ώ (GREEK CAPITAL LETTER OMEGA WITH TONOS)
|
||||
64 0x0390 ΐ (GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS)
|
||||
65 0x0391 Α (GREEK CAPITAL LETTER ALPHA)
|
||||
66 0x0392 Β (GREEK CAPITAL LETTER BETA)
|
||||
67 0x0393 Γ (GREEK CAPITAL LETTER GAMMA)
|
||||
68 0x0394 Δ (GREEK CAPITAL LETTER DELTA)
|
||||
69 0x0395 Ε (GREEK CAPITAL LETTER EPSILON)
|
||||
70 0x0396 Ζ (GREEK CAPITAL LETTER ZETA)
|
||||
71 0x0397 Η (GREEK CAPITAL LETTER ETA)
|
||||
72 0x0398 Θ (GREEK CAPITAL LETTER THETA)
|
||||
73 0x0399 Ι (GREEK CAPITAL LETTER IOTA)
|
||||
74 0x039A Κ (GREEK CAPITAL LETTER KAPPA)
|
||||
75 0x039B Λ (GREEK CAPITAL LETTER LAMDA)
|
||||
76 0x039C Μ (GREEK CAPITAL LETTER MU)
|
||||
77 0x039D Ν (GREEK CAPITAL LETTER NU)
|
||||
78 0x039E Ξ (GREEK CAPITAL LETTER XI)
|
||||
79 0x039F Ο (GREEK CAPITAL LETTER OMICRON)
|
||||
80 0x03A0 Π (GREEK CAPITAL LETTER PI)
|
||||
81 0x03A1 Ρ (GREEK CAPITAL LETTER RHO)
|
||||
83 0x03A3 Σ (GREEK CAPITAL LETTER SIGMA)
|
||||
84 0x03A4 Τ (GREEK CAPITAL LETTER TAU)
|
||||
85 0x03A5 Υ (GREEK CAPITAL LETTER UPSILON)
|
||||
86 0x03A6 Φ (GREEK CAPITAL LETTER PHI)
|
||||
87 0x03A7 Χ (GREEK CAPITAL LETTER CHI)
|
||||
88 0x03A8 Ψ (GREEK CAPITAL LETTER PSI)
|
||||
89 0x03A9 Ω (GREEK CAPITAL LETTER OMEGA)
|
||||
90 0x03AA Ϊ (GREEK CAPITAL LETTER IOTA WITH DIALYTIKA)
|
||||
91 0x03AB Ϋ (GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA)
|
||||
92 0x03AC ά (GREEK SMALL LETTER ALPHA WITH TONOS)
|
||||
93 0x03AD έ (GREEK SMALL LETTER EPSILON WITH TONOS)
|
||||
94 0x03AE ή (GREEK SMALL LETTER ETA WITH TONOS)
|
||||
95 0x03AF ί (GREEK SMALL LETTER IOTA WITH TONOS)
|
||||
96 0x03B0 ΰ (GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS)
|
||||
97 0x03B1 α (GREEK SMALL LETTER ALPHA)
|
||||
98 0x03B2 β (GREEK SMALL LETTER BETA)
|
||||
99 0x03B3 γ (GREEK SMALL LETTER GAMMA)
|
||||
100 0x03B4 δ (GREEK SMALL LETTER DELTA)
|
||||
101 0x03B5 ε (GREEK SMALL LETTER EPSILON)
|
||||
102 0x03B6 ζ (GREEK SMALL LETTER ZETA)
|
||||
103 0x03B7 η (GREEK SMALL LETTER ETA)
|
||||
104 0x03B8 θ (GREEK SMALL LETTER THETA)
|
||||
105 0x03B9 ι (GREEK SMALL LETTER IOTA)
|
||||
106 0x03BA κ (GREEK SMALL LETTER KAPPA)
|
||||
107 0x03BB λ (GREEK SMALL LETTER LAMDA)
|
||||
108 0x03BC μ (GREEK SMALL LETTER MU)
|
||||
109 0x03BD ν (GREEK SMALL LETTER NU)
|
||||
110 0x03BE ξ (GREEK SMALL LETTER XI)
|
||||
111 0x03BF ο (GREEK SMALL LETTER OMICRON)
|
||||
112 0x03C0 π (GREEK SMALL LETTER PI)
|
||||
113 0x03C1 ρ (GREEK SMALL LETTER RHO)
|
||||
114 0x03C2 ς (GREEK SMALL LETTER FINAL SIGMA)
|
||||
115 0x03C3 σ (GREEK SMALL LETTER SIGMA)
|
||||
116 0x03C4 τ (GREEK SMALL LETTER TAU)
|
||||
117 0x03C5 υ (GREEK SMALL LETTER UPSILON)
|
||||
118 0x03C6 φ (GREEK SMALL LETTER PHI)
|
||||
119 0x03C7 χ (GREEK SMALL LETTER CHI)
|
||||
120 0x03C8 ψ (GREEK SMALL LETTER PSI)
|
||||
121 0x03C9 ω (GREEK SMALL LETTER OMEGA)
|
||||
122 0x03CA ϊ (GREEK SMALL LETTER IOTA WITH DIALYTIKA)
|
||||
123 0x03CB ϋ (GREEK SMALL LETTER UPSILON WITH DIALYTIKA)
|
||||
124 0x03CC ό (GREEK SMALL LETTER OMICRON WITH TONOS)
|
||||
125 0x03CD ύ (GREEK SMALL LETTER UPSILON WITH TONOS)
|
||||
126 0x03CE ώ (GREEK SMALL LETTER OMEGA WITH TONOS)
|
||||
98
test/fixtures/encoding/single-byte/index-iso-8859-8.txt
vendored
Normal file
98
test/fixtures/encoding/single-byte/index-iso-8859-8.txt
vendored
Normal file
@@ -0,0 +1,98 @@
|
||||
# For details on index index-iso-8859-8.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 7657a9ca3fa875990da960d3f812eea28dcd0ae6ed55a18d5394303c86f5484b
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0080 (<control>)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x0085
(<control>)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x0091 (<control>)
|
||||
18 0x0092 (<control>)
|
||||
19 0x0093 (<control>)
|
||||
20 0x0094 (<control>)
|
||||
21 0x0095 (<control>)
|
||||
22 0x0096 (<control>)
|
||||
23 0x0097 (<control>)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x00D7 × (MULTIPLICATION SIGN)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x00F7 ÷ (DIVISION SIGN)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
95 0x2017 ‗ (DOUBLE LOW LINE)
|
||||
96 0x05D0 א (HEBREW LETTER ALEF)
|
||||
97 0x05D1 ב (HEBREW LETTER BET)
|
||||
98 0x05D2 ג (HEBREW LETTER GIMEL)
|
||||
99 0x05D3 ד (HEBREW LETTER DALET)
|
||||
100 0x05D4 ה (HEBREW LETTER HE)
|
||||
101 0x05D5 ו (HEBREW LETTER VAV)
|
||||
102 0x05D6 ז (HEBREW LETTER ZAYIN)
|
||||
103 0x05D7 ח (HEBREW LETTER HET)
|
||||
104 0x05D8 ט (HEBREW LETTER TET)
|
||||
105 0x05D9 י (HEBREW LETTER YOD)
|
||||
106 0x05DA ך (HEBREW LETTER FINAL KAF)
|
||||
107 0x05DB כ (HEBREW LETTER KAF)
|
||||
108 0x05DC ל (HEBREW LETTER LAMED)
|
||||
109 0x05DD ם (HEBREW LETTER FINAL MEM)
|
||||
110 0x05DE מ (HEBREW LETTER MEM)
|
||||
111 0x05DF ן (HEBREW LETTER FINAL NUN)
|
||||
112 0x05E0 נ (HEBREW LETTER NUN)
|
||||
113 0x05E1 ס (HEBREW LETTER SAMEKH)
|
||||
114 0x05E2 ע (HEBREW LETTER AYIN)
|
||||
115 0x05E3 ף (HEBREW LETTER FINAL PE)
|
||||
116 0x05E4 פ (HEBREW LETTER PE)
|
||||
117 0x05E5 ץ (HEBREW LETTER FINAL TSADI)
|
||||
118 0x05E6 צ (HEBREW LETTER TSADI)
|
||||
119 0x05E7 ק (HEBREW LETTER QOF)
|
||||
120 0x05E8 ר (HEBREW LETTER RESH)
|
||||
121 0x05E9 ש (HEBREW LETTER SHIN)
|
||||
122 0x05EA ת (HEBREW LETTER TAV)
|
||||
125 0x200E (LEFT-TO-RIGHT MARK)
|
||||
126 0x200F (RIGHT-TO-LEFT MARK)
|
||||
134
test/fixtures/encoding/single-byte/index-koi8-r.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-koi8-r.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-koi8-r.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: c5497cd9071cb352c0e56b219154e539badf63de40b71578f09e2e11fe7d50ae
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x2500 ─ (BOX DRAWINGS LIGHT HORIZONTAL)
|
||||
1 0x2502 │ (BOX DRAWINGS LIGHT VERTICAL)
|
||||
2 0x250C ┌ (BOX DRAWINGS LIGHT DOWN AND RIGHT)
|
||||
3 0x2510 ┐ (BOX DRAWINGS LIGHT DOWN AND LEFT)
|
||||
4 0x2514 └ (BOX DRAWINGS LIGHT UP AND RIGHT)
|
||||
5 0x2518 ┘ (BOX DRAWINGS LIGHT UP AND LEFT)
|
||||
6 0x251C ├ (BOX DRAWINGS LIGHT VERTICAL AND RIGHT)
|
||||
7 0x2524 ┤ (BOX DRAWINGS LIGHT VERTICAL AND LEFT)
|
||||
8 0x252C ┬ (BOX DRAWINGS LIGHT DOWN AND HORIZONTAL)
|
||||
9 0x2534 ┴ (BOX DRAWINGS LIGHT UP AND HORIZONTAL)
|
||||
10 0x253C ┼ (BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL)
|
||||
11 0x2580 ▀ (UPPER HALF BLOCK)
|
||||
12 0x2584 ▄ (LOWER HALF BLOCK)
|
||||
13 0x2588 █ (FULL BLOCK)
|
||||
14 0x258C ▌ (LEFT HALF BLOCK)
|
||||
15 0x2590 ▐ (RIGHT HALF BLOCK)
|
||||
16 0x2591 ░ (LIGHT SHADE)
|
||||
17 0x2592 ▒ (MEDIUM SHADE)
|
||||
18 0x2593 ▓ (DARK SHADE)
|
||||
19 0x2320 ⌠ (TOP HALF INTEGRAL)
|
||||
20 0x25A0 ■ (BLACK SQUARE)
|
||||
21 0x2219 ∙ (BULLET OPERATOR)
|
||||
22 0x221A √ (SQUARE ROOT)
|
||||
23 0x2248 ≈ (ALMOST EQUAL TO)
|
||||
24 0x2264 ≤ (LESS-THAN OR EQUAL TO)
|
||||
25 0x2265 ≥ (GREATER-THAN OR EQUAL TO)
|
||||
26 0x00A0 (NO-BREAK SPACE)
|
||||
27 0x2321 ⌡ (BOTTOM HALF INTEGRAL)
|
||||
28 0x00B0 ° (DEGREE SIGN)
|
||||
29 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
30 0x00B7 · (MIDDLE DOT)
|
||||
31 0x00F7 ÷ (DIVISION SIGN)
|
||||
32 0x2550 ═ (BOX DRAWINGS DOUBLE HORIZONTAL)
|
||||
33 0x2551 ║ (BOX DRAWINGS DOUBLE VERTICAL)
|
||||
34 0x2552 ╒ (BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE)
|
||||
35 0x0451 ё (CYRILLIC SMALL LETTER IO)
|
||||
36 0x2553 ╓ (BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE)
|
||||
37 0x2554 ╔ (BOX DRAWINGS DOUBLE DOWN AND RIGHT)
|
||||
38 0x2555 ╕ (BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE)
|
||||
39 0x2556 ╖ (BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE)
|
||||
40 0x2557 ╗ (BOX DRAWINGS DOUBLE DOWN AND LEFT)
|
||||
41 0x2558 ╘ (BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE)
|
||||
42 0x2559 ╙ (BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE)
|
||||
43 0x255A ╚ (BOX DRAWINGS DOUBLE UP AND RIGHT)
|
||||
44 0x255B ╛ (BOX DRAWINGS UP SINGLE AND LEFT DOUBLE)
|
||||
45 0x255C ╜ (BOX DRAWINGS UP DOUBLE AND LEFT SINGLE)
|
||||
46 0x255D ╝ (BOX DRAWINGS DOUBLE UP AND LEFT)
|
||||
47 0x255E ╞ (BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE)
|
||||
48 0x255F ╟ (BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE)
|
||||
49 0x2560 ╠ (BOX DRAWINGS DOUBLE VERTICAL AND RIGHT)
|
||||
50 0x2561 ╡ (BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE)
|
||||
51 0x0401 Ё (CYRILLIC CAPITAL LETTER IO)
|
||||
52 0x2562 ╢ (BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE)
|
||||
53 0x2563 ╣ (BOX DRAWINGS DOUBLE VERTICAL AND LEFT)
|
||||
54 0x2564 ╤ (BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE)
|
||||
55 0x2565 ╥ (BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE)
|
||||
56 0x2566 ╦ (BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL)
|
||||
57 0x2567 ╧ (BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE)
|
||||
58 0x2568 ╨ (BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE)
|
||||
59 0x2569 ╩ (BOX DRAWINGS DOUBLE UP AND HORIZONTAL)
|
||||
60 0x256A ╪ (BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE)
|
||||
61 0x256B ╫ (BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE)
|
||||
62 0x256C ╬ (BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL)
|
||||
63 0x00A9 © (COPYRIGHT SIGN)
|
||||
64 0x044E ю (CYRILLIC SMALL LETTER YU)
|
||||
65 0x0430 а (CYRILLIC SMALL LETTER A)
|
||||
66 0x0431 б (CYRILLIC SMALL LETTER BE)
|
||||
67 0x0446 ц (CYRILLIC SMALL LETTER TSE)
|
||||
68 0x0434 д (CYRILLIC SMALL LETTER DE)
|
||||
69 0x0435 е (CYRILLIC SMALL LETTER IE)
|
||||
70 0x0444 ф (CYRILLIC SMALL LETTER EF)
|
||||
71 0x0433 г (CYRILLIC SMALL LETTER GHE)
|
||||
72 0x0445 х (CYRILLIC SMALL LETTER HA)
|
||||
73 0x0438 и (CYRILLIC SMALL LETTER I)
|
||||
74 0x0439 й (CYRILLIC SMALL LETTER SHORT I)
|
||||
75 0x043A к (CYRILLIC SMALL LETTER KA)
|
||||
76 0x043B л (CYRILLIC SMALL LETTER EL)
|
||||
77 0x043C м (CYRILLIC SMALL LETTER EM)
|
||||
78 0x043D н (CYRILLIC SMALL LETTER EN)
|
||||
79 0x043E о (CYRILLIC SMALL LETTER O)
|
||||
80 0x043F п (CYRILLIC SMALL LETTER PE)
|
||||
81 0x044F я (CYRILLIC SMALL LETTER YA)
|
||||
82 0x0440 р (CYRILLIC SMALL LETTER ER)
|
||||
83 0x0441 с (CYRILLIC SMALL LETTER ES)
|
||||
84 0x0442 т (CYRILLIC SMALL LETTER TE)
|
||||
85 0x0443 у (CYRILLIC SMALL LETTER U)
|
||||
86 0x0436 ж (CYRILLIC SMALL LETTER ZHE)
|
||||
87 0x0432 в (CYRILLIC SMALL LETTER VE)
|
||||
88 0x044C ь (CYRILLIC SMALL LETTER SOFT SIGN)
|
||||
89 0x044B ы (CYRILLIC SMALL LETTER YERU)
|
||||
90 0x0437 з (CYRILLIC SMALL LETTER ZE)
|
||||
91 0x0448 ш (CYRILLIC SMALL LETTER SHA)
|
||||
92 0x044D э (CYRILLIC SMALL LETTER E)
|
||||
93 0x0449 щ (CYRILLIC SMALL LETTER SHCHA)
|
||||
94 0x0447 ч (CYRILLIC SMALL LETTER CHE)
|
||||
95 0x044A ъ (CYRILLIC SMALL LETTER HARD SIGN)
|
||||
96 0x042E Ю (CYRILLIC CAPITAL LETTER YU)
|
||||
97 0x0410 А (CYRILLIC CAPITAL LETTER A)
|
||||
98 0x0411 Б (CYRILLIC CAPITAL LETTER BE)
|
||||
99 0x0426 Ц (CYRILLIC CAPITAL LETTER TSE)
|
||||
100 0x0414 Д (CYRILLIC CAPITAL LETTER DE)
|
||||
101 0x0415 Е (CYRILLIC CAPITAL LETTER IE)
|
||||
102 0x0424 Ф (CYRILLIC CAPITAL LETTER EF)
|
||||
103 0x0413 Г (CYRILLIC CAPITAL LETTER GHE)
|
||||
104 0x0425 Х (CYRILLIC CAPITAL LETTER HA)
|
||||
105 0x0418 И (CYRILLIC CAPITAL LETTER I)
|
||||
106 0x0419 Й (CYRILLIC CAPITAL LETTER SHORT I)
|
||||
107 0x041A К (CYRILLIC CAPITAL LETTER KA)
|
||||
108 0x041B Л (CYRILLIC CAPITAL LETTER EL)
|
||||
109 0x041C М (CYRILLIC CAPITAL LETTER EM)
|
||||
110 0x041D Н (CYRILLIC CAPITAL LETTER EN)
|
||||
111 0x041E О (CYRILLIC CAPITAL LETTER O)
|
||||
112 0x041F П (CYRILLIC CAPITAL LETTER PE)
|
||||
113 0x042F Я (CYRILLIC CAPITAL LETTER YA)
|
||||
114 0x0420 Р (CYRILLIC CAPITAL LETTER ER)
|
||||
115 0x0421 С (CYRILLIC CAPITAL LETTER ES)
|
||||
116 0x0422 Т (CYRILLIC CAPITAL LETTER TE)
|
||||
117 0x0423 У (CYRILLIC CAPITAL LETTER U)
|
||||
118 0x0416 Ж (CYRILLIC CAPITAL LETTER ZHE)
|
||||
119 0x0412 В (CYRILLIC CAPITAL LETTER VE)
|
||||
120 0x042C Ь (CYRILLIC CAPITAL LETTER SOFT SIGN)
|
||||
121 0x042B Ы (CYRILLIC CAPITAL LETTER YERU)
|
||||
122 0x0417 З (CYRILLIC CAPITAL LETTER ZE)
|
||||
123 0x0428 Ш (CYRILLIC CAPITAL LETTER SHA)
|
||||
124 0x042D Э (CYRILLIC CAPITAL LETTER E)
|
||||
125 0x0429 Щ (CYRILLIC CAPITAL LETTER SHCHA)
|
||||
126 0x0427 Ч (CYRILLIC CAPITAL LETTER CHE)
|
||||
127 0x042A Ъ (CYRILLIC CAPITAL LETTER HARD SIGN)
|
||||
134
test/fixtures/encoding/single-byte/index-koi8-u.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-koi8-u.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-koi8-u.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 19a4da2c3f245118bbc8019326f45a07832949938ff903f03d62ac4da1f61f40
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x2500 ─ (BOX DRAWINGS LIGHT HORIZONTAL)
|
||||
1 0x2502 │ (BOX DRAWINGS LIGHT VERTICAL)
|
||||
2 0x250C ┌ (BOX DRAWINGS LIGHT DOWN AND RIGHT)
|
||||
3 0x2510 ┐ (BOX DRAWINGS LIGHT DOWN AND LEFT)
|
||||
4 0x2514 └ (BOX DRAWINGS LIGHT UP AND RIGHT)
|
||||
5 0x2518 ┘ (BOX DRAWINGS LIGHT UP AND LEFT)
|
||||
6 0x251C ├ (BOX DRAWINGS LIGHT VERTICAL AND RIGHT)
|
||||
7 0x2524 ┤ (BOX DRAWINGS LIGHT VERTICAL AND LEFT)
|
||||
8 0x252C ┬ (BOX DRAWINGS LIGHT DOWN AND HORIZONTAL)
|
||||
9 0x2534 ┴ (BOX DRAWINGS LIGHT UP AND HORIZONTAL)
|
||||
10 0x253C ┼ (BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL)
|
||||
11 0x2580 ▀ (UPPER HALF BLOCK)
|
||||
12 0x2584 ▄ (LOWER HALF BLOCK)
|
||||
13 0x2588 █ (FULL BLOCK)
|
||||
14 0x258C ▌ (LEFT HALF BLOCK)
|
||||
15 0x2590 ▐ (RIGHT HALF BLOCK)
|
||||
16 0x2591 ░ (LIGHT SHADE)
|
||||
17 0x2592 ▒ (MEDIUM SHADE)
|
||||
18 0x2593 ▓ (DARK SHADE)
|
||||
19 0x2320 ⌠ (TOP HALF INTEGRAL)
|
||||
20 0x25A0 ■ (BLACK SQUARE)
|
||||
21 0x2219 ∙ (BULLET OPERATOR)
|
||||
22 0x221A √ (SQUARE ROOT)
|
||||
23 0x2248 ≈ (ALMOST EQUAL TO)
|
||||
24 0x2264 ≤ (LESS-THAN OR EQUAL TO)
|
||||
25 0x2265 ≥ (GREATER-THAN OR EQUAL TO)
|
||||
26 0x00A0 (NO-BREAK SPACE)
|
||||
27 0x2321 ⌡ (BOTTOM HALF INTEGRAL)
|
||||
28 0x00B0 ° (DEGREE SIGN)
|
||||
29 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
30 0x00B7 · (MIDDLE DOT)
|
||||
31 0x00F7 ÷ (DIVISION SIGN)
|
||||
32 0x2550 ═ (BOX DRAWINGS DOUBLE HORIZONTAL)
|
||||
33 0x2551 ║ (BOX DRAWINGS DOUBLE VERTICAL)
|
||||
34 0x2552 ╒ (BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE)
|
||||
35 0x0451 ё (CYRILLIC SMALL LETTER IO)
|
||||
36 0x0454 є (CYRILLIC SMALL LETTER UKRAINIAN IE)
|
||||
37 0x2554 ╔ (BOX DRAWINGS DOUBLE DOWN AND RIGHT)
|
||||
38 0x0456 і (CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
39 0x0457 ї (CYRILLIC SMALL LETTER YI)
|
||||
40 0x2557 ╗ (BOX DRAWINGS DOUBLE DOWN AND LEFT)
|
||||
41 0x2558 ╘ (BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE)
|
||||
42 0x2559 ╙ (BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE)
|
||||
43 0x255A ╚ (BOX DRAWINGS DOUBLE UP AND RIGHT)
|
||||
44 0x255B ╛ (BOX DRAWINGS UP SINGLE AND LEFT DOUBLE)
|
||||
45 0x0491 ґ (CYRILLIC SMALL LETTER GHE WITH UPTURN)
|
||||
46 0x045E ў (CYRILLIC SMALL LETTER SHORT U)
|
||||
47 0x255E ╞ (BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE)
|
||||
48 0x255F ╟ (BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE)
|
||||
49 0x2560 ╠ (BOX DRAWINGS DOUBLE VERTICAL AND RIGHT)
|
||||
50 0x2561 ╡ (BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE)
|
||||
51 0x0401 Ё (CYRILLIC CAPITAL LETTER IO)
|
||||
52 0x0404 Є (CYRILLIC CAPITAL LETTER UKRAINIAN IE)
|
||||
53 0x2563 ╣ (BOX DRAWINGS DOUBLE VERTICAL AND LEFT)
|
||||
54 0x0406 І (CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
55 0x0407 Ї (CYRILLIC CAPITAL LETTER YI)
|
||||
56 0x2566 ╦ (BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL)
|
||||
57 0x2567 ╧ (BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE)
|
||||
58 0x2568 ╨ (BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE)
|
||||
59 0x2569 ╩ (BOX DRAWINGS DOUBLE UP AND HORIZONTAL)
|
||||
60 0x256A ╪ (BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE)
|
||||
61 0x0490 Ґ (CYRILLIC CAPITAL LETTER GHE WITH UPTURN)
|
||||
62 0x040E Ў (CYRILLIC CAPITAL LETTER SHORT U)
|
||||
63 0x00A9 © (COPYRIGHT SIGN)
|
||||
64 0x044E ю (CYRILLIC SMALL LETTER YU)
|
||||
65 0x0430 а (CYRILLIC SMALL LETTER A)
|
||||
66 0x0431 б (CYRILLIC SMALL LETTER BE)
|
||||
67 0x0446 ц (CYRILLIC SMALL LETTER TSE)
|
||||
68 0x0434 д (CYRILLIC SMALL LETTER DE)
|
||||
69 0x0435 е (CYRILLIC SMALL LETTER IE)
|
||||
70 0x0444 ф (CYRILLIC SMALL LETTER EF)
|
||||
71 0x0433 г (CYRILLIC SMALL LETTER GHE)
|
||||
72 0x0445 х (CYRILLIC SMALL LETTER HA)
|
||||
73 0x0438 и (CYRILLIC SMALL LETTER I)
|
||||
74 0x0439 й (CYRILLIC SMALL LETTER SHORT I)
|
||||
75 0x043A к (CYRILLIC SMALL LETTER KA)
|
||||
76 0x043B л (CYRILLIC SMALL LETTER EL)
|
||||
77 0x043C м (CYRILLIC SMALL LETTER EM)
|
||||
78 0x043D н (CYRILLIC SMALL LETTER EN)
|
||||
79 0x043E о (CYRILLIC SMALL LETTER O)
|
||||
80 0x043F п (CYRILLIC SMALL LETTER PE)
|
||||
81 0x044F я (CYRILLIC SMALL LETTER YA)
|
||||
82 0x0440 р (CYRILLIC SMALL LETTER ER)
|
||||
83 0x0441 с (CYRILLIC SMALL LETTER ES)
|
||||
84 0x0442 т (CYRILLIC SMALL LETTER TE)
|
||||
85 0x0443 у (CYRILLIC SMALL LETTER U)
|
||||
86 0x0436 ж (CYRILLIC SMALL LETTER ZHE)
|
||||
87 0x0432 в (CYRILLIC SMALL LETTER VE)
|
||||
88 0x044C ь (CYRILLIC SMALL LETTER SOFT SIGN)
|
||||
89 0x044B ы (CYRILLIC SMALL LETTER YERU)
|
||||
90 0x0437 з (CYRILLIC SMALL LETTER ZE)
|
||||
91 0x0448 ш (CYRILLIC SMALL LETTER SHA)
|
||||
92 0x044D э (CYRILLIC SMALL LETTER E)
|
||||
93 0x0449 щ (CYRILLIC SMALL LETTER SHCHA)
|
||||
94 0x0447 ч (CYRILLIC SMALL LETTER CHE)
|
||||
95 0x044A ъ (CYRILLIC SMALL LETTER HARD SIGN)
|
||||
96 0x042E Ю (CYRILLIC CAPITAL LETTER YU)
|
||||
97 0x0410 А (CYRILLIC CAPITAL LETTER A)
|
||||
98 0x0411 Б (CYRILLIC CAPITAL LETTER BE)
|
||||
99 0x0426 Ц (CYRILLIC CAPITAL LETTER TSE)
|
||||
100 0x0414 Д (CYRILLIC CAPITAL LETTER DE)
|
||||
101 0x0415 Е (CYRILLIC CAPITAL LETTER IE)
|
||||
102 0x0424 Ф (CYRILLIC CAPITAL LETTER EF)
|
||||
103 0x0413 Г (CYRILLIC CAPITAL LETTER GHE)
|
||||
104 0x0425 Х (CYRILLIC CAPITAL LETTER HA)
|
||||
105 0x0418 И (CYRILLIC CAPITAL LETTER I)
|
||||
106 0x0419 Й (CYRILLIC CAPITAL LETTER SHORT I)
|
||||
107 0x041A К (CYRILLIC CAPITAL LETTER KA)
|
||||
108 0x041B Л (CYRILLIC CAPITAL LETTER EL)
|
||||
109 0x041C М (CYRILLIC CAPITAL LETTER EM)
|
||||
110 0x041D Н (CYRILLIC CAPITAL LETTER EN)
|
||||
111 0x041E О (CYRILLIC CAPITAL LETTER O)
|
||||
112 0x041F П (CYRILLIC CAPITAL LETTER PE)
|
||||
113 0x042F Я (CYRILLIC CAPITAL LETTER YA)
|
||||
114 0x0420 Р (CYRILLIC CAPITAL LETTER ER)
|
||||
115 0x0421 С (CYRILLIC CAPITAL LETTER ES)
|
||||
116 0x0422 Т (CYRILLIC CAPITAL LETTER TE)
|
||||
117 0x0423 У (CYRILLIC CAPITAL LETTER U)
|
||||
118 0x0416 Ж (CYRILLIC CAPITAL LETTER ZHE)
|
||||
119 0x0412 В (CYRILLIC CAPITAL LETTER VE)
|
||||
120 0x042C Ь (CYRILLIC CAPITAL LETTER SOFT SIGN)
|
||||
121 0x042B Ы (CYRILLIC CAPITAL LETTER YERU)
|
||||
122 0x0417 З (CYRILLIC CAPITAL LETTER ZE)
|
||||
123 0x0428 Ш (CYRILLIC CAPITAL LETTER SHA)
|
||||
124 0x042D Э (CYRILLIC CAPITAL LETTER E)
|
||||
125 0x0429 Щ (CYRILLIC CAPITAL LETTER SHCHA)
|
||||
126 0x0427 Ч (CYRILLIC CAPITAL LETTER CHE)
|
||||
127 0x042A Ъ (CYRILLIC CAPITAL LETTER HARD SIGN)
|
||||
134
test/fixtures/encoding/single-byte/index-macintosh.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-macintosh.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-macintosh.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: f2c6a4f6406b3e86a50a5dba4d2b7dd48e2e33c0d82aefe764535c934ec11764
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
1 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
2 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
3 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
4 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
5 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
6 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
7 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
8 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
9 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
10 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
11 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
12 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
13 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
14 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
15 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
16 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
17 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
18 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
19 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
20 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
21 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
22 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
23 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
24 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
25 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
26 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
27 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
28 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
29 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
30 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
31 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
32 0x2020 † (DAGGER)
|
||||
33 0x00B0 ° (DEGREE SIGN)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A7 § (SECTION SIGN)
|
||||
37 0x2022 • (BULLET)
|
||||
38 0x00B6 ¶ (PILCROW SIGN)
|
||||
39 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
40 0x00AE ® (REGISTERED SIGN)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x2122 ™ (TRADE MARK SIGN)
|
||||
43 0x00B4 ´ (ACUTE ACCENT)
|
||||
44 0x00A8 ¨ (DIAERESIS)
|
||||
45 0x2260 ≠ (NOT EQUAL TO)
|
||||
46 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
47 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
48 0x221E ∞ (INFINITY)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x2264 ≤ (LESS-THAN OR EQUAL TO)
|
||||
51 0x2265 ≥ (GREATER-THAN OR EQUAL TO)
|
||||
52 0x00A5 ¥ (YEN SIGN)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x2202 ∂ (PARTIAL DIFFERENTIAL)
|
||||
55 0x2211 ∑ (N-ARY SUMMATION)
|
||||
56 0x220F ∏ (N-ARY PRODUCT)
|
||||
57 0x03C0 π (GREEK SMALL LETTER PI)
|
||||
58 0x222B ∫ (INTEGRAL)
|
||||
59 0x00AA ª (FEMININE ORDINAL INDICATOR)
|
||||
60 0x00BA º (MASCULINE ORDINAL INDICATOR)
|
||||
61 0x03A9 Ω (GREEK CAPITAL LETTER OMEGA)
|
||||
62 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
63 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
64 0x00BF ¿ (INVERTED QUESTION MARK)
|
||||
65 0x00A1 ¡ (INVERTED EXCLAMATION MARK)
|
||||
66 0x00AC ¬ (NOT SIGN)
|
||||
67 0x221A √ (SQUARE ROOT)
|
||||
68 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
69 0x2248 ≈ (ALMOST EQUAL TO)
|
||||
70 0x2206 ∆ (INCREMENT)
|
||||
71 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
72 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
73 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
74 0x00A0 (NO-BREAK SPACE)
|
||||
75 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
76 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
77 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
78 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
79 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
80 0x2013 – (EN DASH)
|
||||
81 0x2014 — (EM DASH)
|
||||
82 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
83 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
84 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
85 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
86 0x00F7 ÷ (DIVISION SIGN)
|
||||
87 0x25CA ◊ (LOZENGE)
|
||||
88 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
89 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
90 0x2044 ⁄ (FRACTION SLASH)
|
||||
91 0x20AC € (EURO SIGN)
|
||||
92 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
93 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
94 0xFB01 fi (LATIN SMALL LIGATURE FI)
|
||||
95 0xFB02 fl (LATIN SMALL LIGATURE FL)
|
||||
96 0x2021 ‡ (DOUBLE DAGGER)
|
||||
97 0x00B7 · (MIDDLE DOT)
|
||||
98 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
99 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
100 0x2030 ‰ (PER MILLE SIGN)
|
||||
101 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
102 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
103 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
104 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
105 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
106 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
107 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
108 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
109 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
110 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
111 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
112 0xF8FF (<Private Use>)
|
||||
113 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
114 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
115 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
116 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
117 0x0131 ı (LATIN SMALL LETTER DOTLESS I)
|
||||
118 0x02C6 ˆ (MODIFIER LETTER CIRCUMFLEX ACCENT)
|
||||
119 0x02DC ˜ (SMALL TILDE)
|
||||
120 0x00AF ¯ (MACRON)
|
||||
121 0x02D8 ˘ (BREVE)
|
||||
122 0x02D9 ˙ (DOT ABOVE)
|
||||
123 0x02DA ˚ (RING ABOVE)
|
||||
124 0x00B8 ¸ (CEDILLA)
|
||||
125 0x02DD ˝ (DOUBLE ACUTE ACCENT)
|
||||
126 0x02DB ˛ (OGONEK)
|
||||
127 0x02C7 ˇ (CARON)
|
||||
134
test/fixtures/encoding/single-byte/index-windows-1250.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-windows-1250.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-windows-1250.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 0669455a7a1c70ba6003ea737991e8ee9adc455125c13cfe6705a361358de5fa
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0083 (<control>)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x0088 (<control>)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x015A Ś (LATIN CAPITAL LETTER S WITH ACUTE)
|
||||
13 0x0164 Ť (LATIN CAPITAL LETTER T WITH CARON)
|
||||
14 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
15 0x0179 Ź (LATIN CAPITAL LETTER Z WITH ACUTE)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x0098 (<control>)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x015B ś (LATIN SMALL LETTER S WITH ACUTE)
|
||||
29 0x0165 ť (LATIN SMALL LETTER T WITH CARON)
|
||||
30 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
31 0x017A ź (LATIN SMALL LETTER Z WITH ACUTE)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x02C7 ˇ (CARON)
|
||||
34 0x02D8 ˘ (BREVE)
|
||||
35 0x0141 Ł (LATIN CAPITAL LETTER L WITH STROKE)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x015E Ş (LATIN CAPITAL LETTER S WITH CEDILLA)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x017B Ż (LATIN CAPITAL LETTER Z WITH DOT ABOVE)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x02DB ˛ (OGONEK)
|
||||
51 0x0142 ł (LATIN SMALL LETTER L WITH STROKE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
58 0x015F ş (LATIN SMALL LETTER S WITH CEDILLA)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x013D Ľ (LATIN CAPITAL LETTER L WITH CARON)
|
||||
61 0x02DD ˝ (DOUBLE ACUTE ACCENT)
|
||||
62 0x013E ľ (LATIN SMALL LETTER L WITH CARON)
|
||||
63 0x017C ż (LATIN SMALL LETTER Z WITH DOT ABOVE)
|
||||
64 0x0154 Ŕ (LATIN CAPITAL LETTER R WITH ACUTE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x0102 Ă (LATIN CAPITAL LETTER A WITH BREVE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x0139 Ĺ (LATIN CAPITAL LETTER L WITH ACUTE)
|
||||
70 0x0106 Ć (LATIN CAPITAL LETTER C WITH ACUTE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x011A Ě (LATIN CAPITAL LETTER E WITH CARON)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x010E Ď (LATIN CAPITAL LETTER D WITH CARON)
|
||||
80 0x0110 Đ (LATIN CAPITAL LETTER D WITH STROKE)
|
||||
81 0x0143 Ń (LATIN CAPITAL LETTER N WITH ACUTE)
|
||||
82 0x0147 Ň (LATIN CAPITAL LETTER N WITH CARON)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x0150 Ő (LATIN CAPITAL LETTER O WITH DOUBLE ACUTE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x0158 Ř (LATIN CAPITAL LETTER R WITH CARON)
|
||||
89 0x016E Ů (LATIN CAPITAL LETTER U WITH RING ABOVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x0170 Ű (LATIN CAPITAL LETTER U WITH DOUBLE ACUTE)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x00DD Ý (LATIN CAPITAL LETTER Y WITH ACUTE)
|
||||
94 0x0162 Ţ (LATIN CAPITAL LETTER T WITH CEDILLA)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x0155 ŕ (LATIN SMALL LETTER R WITH ACUTE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x0103 ă (LATIN SMALL LETTER A WITH BREVE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x013A ĺ (LATIN SMALL LETTER L WITH ACUTE)
|
||||
102 0x0107 ć (LATIN SMALL LETTER C WITH ACUTE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x011B ě (LATIN SMALL LETTER E WITH CARON)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x010F ď (LATIN SMALL LETTER D WITH CARON)
|
||||
112 0x0111 đ (LATIN SMALL LETTER D WITH STROKE)
|
||||
113 0x0144 ń (LATIN SMALL LETTER N WITH ACUTE)
|
||||
114 0x0148 ň (LATIN SMALL LETTER N WITH CARON)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x0151 ő (LATIN SMALL LETTER O WITH DOUBLE ACUTE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x0159 ř (LATIN SMALL LETTER R WITH CARON)
|
||||
121 0x016F ů (LATIN SMALL LETTER U WITH RING ABOVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x0171 ű (LATIN SMALL LETTER U WITH DOUBLE ACUTE)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x00FD ý (LATIN SMALL LETTER Y WITH ACUTE)
|
||||
126 0x0163 ţ (LATIN SMALL LETTER T WITH CEDILLA)
|
||||
127 0x02D9 ˙ (DOT ABOVE)
|
||||
134
test/fixtures/encoding/single-byte/index-windows-1251.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-windows-1251.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-windows-1251.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 7592ef921679ba168b00a9e9afa3b4eebd67bf13dc7e84c4b6e120de856826e0
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0402 Ђ (CYRILLIC CAPITAL LETTER DJE)
|
||||
1 0x0403 Ѓ (CYRILLIC CAPITAL LETTER GJE)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0453 ѓ (CYRILLIC SMALL LETTER GJE)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x20AC € (EURO SIGN)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x0409 Љ (CYRILLIC CAPITAL LETTER LJE)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x040A Њ (CYRILLIC CAPITAL LETTER NJE)
|
||||
13 0x040C Ќ (CYRILLIC CAPITAL LETTER KJE)
|
||||
14 0x040B Ћ (CYRILLIC CAPITAL LETTER TSHE)
|
||||
15 0x040F Џ (CYRILLIC CAPITAL LETTER DZHE)
|
||||
16 0x0452 ђ (CYRILLIC SMALL LETTER DJE)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x0098 (<control>)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x0459 љ (CYRILLIC SMALL LETTER LJE)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x045A њ (CYRILLIC SMALL LETTER NJE)
|
||||
29 0x045C ќ (CYRILLIC SMALL LETTER KJE)
|
||||
30 0x045B ћ (CYRILLIC SMALL LETTER TSHE)
|
||||
31 0x045F џ (CYRILLIC SMALL LETTER DZHE)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x040E Ў (CYRILLIC CAPITAL LETTER SHORT U)
|
||||
34 0x045E ў (CYRILLIC SMALL LETTER SHORT U)
|
||||
35 0x0408 Ј (CYRILLIC CAPITAL LETTER JE)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x0490 Ґ (CYRILLIC CAPITAL LETTER GHE WITH UPTURN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x0401 Ё (CYRILLIC CAPITAL LETTER IO)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x0404 Є (CYRILLIC CAPITAL LETTER UKRAINIAN IE)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x0407 Ї (CYRILLIC CAPITAL LETTER YI)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x0406 І (CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
51 0x0456 і (CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
52 0x0491 ґ (CYRILLIC SMALL LETTER GHE WITH UPTURN)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x0451 ё (CYRILLIC SMALL LETTER IO)
|
||||
57 0x2116 № (NUMERO SIGN)
|
||||
58 0x0454 є (CYRILLIC SMALL LETTER UKRAINIAN IE)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x0458 ј (CYRILLIC SMALL LETTER JE)
|
||||
61 0x0405 Ѕ (CYRILLIC CAPITAL LETTER DZE)
|
||||
62 0x0455 ѕ (CYRILLIC SMALL LETTER DZE)
|
||||
63 0x0457 ї (CYRILLIC SMALL LETTER YI)
|
||||
64 0x0410 А (CYRILLIC CAPITAL LETTER A)
|
||||
65 0x0411 Б (CYRILLIC CAPITAL LETTER BE)
|
||||
66 0x0412 В (CYRILLIC CAPITAL LETTER VE)
|
||||
67 0x0413 Г (CYRILLIC CAPITAL LETTER GHE)
|
||||
68 0x0414 Д (CYRILLIC CAPITAL LETTER DE)
|
||||
69 0x0415 Е (CYRILLIC CAPITAL LETTER IE)
|
||||
70 0x0416 Ж (CYRILLIC CAPITAL LETTER ZHE)
|
||||
71 0x0417 З (CYRILLIC CAPITAL LETTER ZE)
|
||||
72 0x0418 И (CYRILLIC CAPITAL LETTER I)
|
||||
73 0x0419 Й (CYRILLIC CAPITAL LETTER SHORT I)
|
||||
74 0x041A К (CYRILLIC CAPITAL LETTER KA)
|
||||
75 0x041B Л (CYRILLIC CAPITAL LETTER EL)
|
||||
76 0x041C М (CYRILLIC CAPITAL LETTER EM)
|
||||
77 0x041D Н (CYRILLIC CAPITAL LETTER EN)
|
||||
78 0x041E О (CYRILLIC CAPITAL LETTER O)
|
||||
79 0x041F П (CYRILLIC CAPITAL LETTER PE)
|
||||
80 0x0420 Р (CYRILLIC CAPITAL LETTER ER)
|
||||
81 0x0421 С (CYRILLIC CAPITAL LETTER ES)
|
||||
82 0x0422 Т (CYRILLIC CAPITAL LETTER TE)
|
||||
83 0x0423 У (CYRILLIC CAPITAL LETTER U)
|
||||
84 0x0424 Ф (CYRILLIC CAPITAL LETTER EF)
|
||||
85 0x0425 Х (CYRILLIC CAPITAL LETTER HA)
|
||||
86 0x0426 Ц (CYRILLIC CAPITAL LETTER TSE)
|
||||
87 0x0427 Ч (CYRILLIC CAPITAL LETTER CHE)
|
||||
88 0x0428 Ш (CYRILLIC CAPITAL LETTER SHA)
|
||||
89 0x0429 Щ (CYRILLIC CAPITAL LETTER SHCHA)
|
||||
90 0x042A Ъ (CYRILLIC CAPITAL LETTER HARD SIGN)
|
||||
91 0x042B Ы (CYRILLIC CAPITAL LETTER YERU)
|
||||
92 0x042C Ь (CYRILLIC CAPITAL LETTER SOFT SIGN)
|
||||
93 0x042D Э (CYRILLIC CAPITAL LETTER E)
|
||||
94 0x042E Ю (CYRILLIC CAPITAL LETTER YU)
|
||||
95 0x042F Я (CYRILLIC CAPITAL LETTER YA)
|
||||
96 0x0430 а (CYRILLIC SMALL LETTER A)
|
||||
97 0x0431 б (CYRILLIC SMALL LETTER BE)
|
||||
98 0x0432 в (CYRILLIC SMALL LETTER VE)
|
||||
99 0x0433 г (CYRILLIC SMALL LETTER GHE)
|
||||
100 0x0434 д (CYRILLIC SMALL LETTER DE)
|
||||
101 0x0435 е (CYRILLIC SMALL LETTER IE)
|
||||
102 0x0436 ж (CYRILLIC SMALL LETTER ZHE)
|
||||
103 0x0437 з (CYRILLIC SMALL LETTER ZE)
|
||||
104 0x0438 и (CYRILLIC SMALL LETTER I)
|
||||
105 0x0439 й (CYRILLIC SMALL LETTER SHORT I)
|
||||
106 0x043A к (CYRILLIC SMALL LETTER KA)
|
||||
107 0x043B л (CYRILLIC SMALL LETTER EL)
|
||||
108 0x043C м (CYRILLIC SMALL LETTER EM)
|
||||
109 0x043D н (CYRILLIC SMALL LETTER EN)
|
||||
110 0x043E о (CYRILLIC SMALL LETTER O)
|
||||
111 0x043F п (CYRILLIC SMALL LETTER PE)
|
||||
112 0x0440 р (CYRILLIC SMALL LETTER ER)
|
||||
113 0x0441 с (CYRILLIC SMALL LETTER ES)
|
||||
114 0x0442 т (CYRILLIC SMALL LETTER TE)
|
||||
115 0x0443 у (CYRILLIC SMALL LETTER U)
|
||||
116 0x0444 ф (CYRILLIC SMALL LETTER EF)
|
||||
117 0x0445 х (CYRILLIC SMALL LETTER HA)
|
||||
118 0x0446 ц (CYRILLIC SMALL LETTER TSE)
|
||||
119 0x0447 ч (CYRILLIC SMALL LETTER CHE)
|
||||
120 0x0448 ш (CYRILLIC SMALL LETTER SHA)
|
||||
121 0x0449 щ (CYRILLIC SMALL LETTER SHCHA)
|
||||
122 0x044A ъ (CYRILLIC SMALL LETTER HARD SIGN)
|
||||
123 0x044B ы (CYRILLIC SMALL LETTER YERU)
|
||||
124 0x044C ь (CYRILLIC SMALL LETTER SOFT SIGN)
|
||||
125 0x044D э (CYRILLIC SMALL LETTER E)
|
||||
126 0x044E ю (CYRILLIC SMALL LETTER YU)
|
||||
127 0x044F я (CYRILLIC SMALL LETTER YA)
|
||||
134
test/fixtures/encoding/single-byte/index-windows-1252.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-windows-1252.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-windows-1252.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: e56d49d9176e9a412283cf29ac9bd613f5620462f2a080a84eceaf974cfa18b7
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x02C6 ˆ (MODIFIER LETTER CIRCUMFLEX ACCENT)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
13 0x008D (<control>)
|
||||
14 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x02DC ˜ (SMALL TILDE)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
29 0x009D (<control>)
|
||||
30 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
31 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x00A1 ¡ (INVERTED EXCLAMATION MARK)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x00AA ª (FEMININE ORDINAL INDICATOR)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x00BA º (MASCULINE ORDINAL INDICATOR)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x00BF ¿ (INVERTED QUESTION MARK)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x00D0 Ð (LATIN CAPITAL LETTER ETH)
|
||||
81 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
82 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x00DD Ý (LATIN CAPITAL LETTER Y WITH ACUTE)
|
||||
94 0x00DE Þ (LATIN CAPITAL LETTER THORN)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x00F0 ð (LATIN SMALL LETTER ETH)
|
||||
113 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
114 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x00FD ý (LATIN SMALL LETTER Y WITH ACUTE)
|
||||
126 0x00FE þ (LATIN SMALL LETTER THORN)
|
||||
127 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
131
test/fixtures/encoding/single-byte/index-windows-1253.txt
vendored
Normal file
131
test/fixtures/encoding/single-byte/index-windows-1253.txt
vendored
Normal file
@@ -0,0 +1,131 @@
|
||||
# For details on index index-windows-1253.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 49fdc881a3488904dd1e8dfba9aef3258454249958b611bcded1d4c981ab5561
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x0088 (<control>)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x008A (<control>)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x0098 (<control>)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x009A (<control>)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0385 ΅ (GREEK DIALYTIKA TONOS)
|
||||
34 0x0386 Ά (GREEK CAPITAL LETTER ALPHA WITH TONOS)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x2015 ― (HORIZONTAL BAR)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x0384 ΄ (GREEK TONOS)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x0388 Έ (GREEK CAPITAL LETTER EPSILON WITH TONOS)
|
||||
57 0x0389 Ή (GREEK CAPITAL LETTER ETA WITH TONOS)
|
||||
58 0x038A Ί (GREEK CAPITAL LETTER IOTA WITH TONOS)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x038C Ό (GREEK CAPITAL LETTER OMICRON WITH TONOS)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x038E Ύ (GREEK CAPITAL LETTER UPSILON WITH TONOS)
|
||||
63 0x038F Ώ (GREEK CAPITAL LETTER OMEGA WITH TONOS)
|
||||
64 0x0390 ΐ (GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS)
|
||||
65 0x0391 Α (GREEK CAPITAL LETTER ALPHA)
|
||||
66 0x0392 Β (GREEK CAPITAL LETTER BETA)
|
||||
67 0x0393 Γ (GREEK CAPITAL LETTER GAMMA)
|
||||
68 0x0394 Δ (GREEK CAPITAL LETTER DELTA)
|
||||
69 0x0395 Ε (GREEK CAPITAL LETTER EPSILON)
|
||||
70 0x0396 Ζ (GREEK CAPITAL LETTER ZETA)
|
||||
71 0x0397 Η (GREEK CAPITAL LETTER ETA)
|
||||
72 0x0398 Θ (GREEK CAPITAL LETTER THETA)
|
||||
73 0x0399 Ι (GREEK CAPITAL LETTER IOTA)
|
||||
74 0x039A Κ (GREEK CAPITAL LETTER KAPPA)
|
||||
75 0x039B Λ (GREEK CAPITAL LETTER LAMDA)
|
||||
76 0x039C Μ (GREEK CAPITAL LETTER MU)
|
||||
77 0x039D Ν (GREEK CAPITAL LETTER NU)
|
||||
78 0x039E Ξ (GREEK CAPITAL LETTER XI)
|
||||
79 0x039F Ο (GREEK CAPITAL LETTER OMICRON)
|
||||
80 0x03A0 Π (GREEK CAPITAL LETTER PI)
|
||||
81 0x03A1 Ρ (GREEK CAPITAL LETTER RHO)
|
||||
83 0x03A3 Σ (GREEK CAPITAL LETTER SIGMA)
|
||||
84 0x03A4 Τ (GREEK CAPITAL LETTER TAU)
|
||||
85 0x03A5 Υ (GREEK CAPITAL LETTER UPSILON)
|
||||
86 0x03A6 Φ (GREEK CAPITAL LETTER PHI)
|
||||
87 0x03A7 Χ (GREEK CAPITAL LETTER CHI)
|
||||
88 0x03A8 Ψ (GREEK CAPITAL LETTER PSI)
|
||||
89 0x03A9 Ω (GREEK CAPITAL LETTER OMEGA)
|
||||
90 0x03AA Ϊ (GREEK CAPITAL LETTER IOTA WITH DIALYTIKA)
|
||||
91 0x03AB Ϋ (GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA)
|
||||
92 0x03AC ά (GREEK SMALL LETTER ALPHA WITH TONOS)
|
||||
93 0x03AD έ (GREEK SMALL LETTER EPSILON WITH TONOS)
|
||||
94 0x03AE ή (GREEK SMALL LETTER ETA WITH TONOS)
|
||||
95 0x03AF ί (GREEK SMALL LETTER IOTA WITH TONOS)
|
||||
96 0x03B0 ΰ (GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS)
|
||||
97 0x03B1 α (GREEK SMALL LETTER ALPHA)
|
||||
98 0x03B2 β (GREEK SMALL LETTER BETA)
|
||||
99 0x03B3 γ (GREEK SMALL LETTER GAMMA)
|
||||
100 0x03B4 δ (GREEK SMALL LETTER DELTA)
|
||||
101 0x03B5 ε (GREEK SMALL LETTER EPSILON)
|
||||
102 0x03B6 ζ (GREEK SMALL LETTER ZETA)
|
||||
103 0x03B7 η (GREEK SMALL LETTER ETA)
|
||||
104 0x03B8 θ (GREEK SMALL LETTER THETA)
|
||||
105 0x03B9 ι (GREEK SMALL LETTER IOTA)
|
||||
106 0x03BA κ (GREEK SMALL LETTER KAPPA)
|
||||
107 0x03BB λ (GREEK SMALL LETTER LAMDA)
|
||||
108 0x03BC μ (GREEK SMALL LETTER MU)
|
||||
109 0x03BD ν (GREEK SMALL LETTER NU)
|
||||
110 0x03BE ξ (GREEK SMALL LETTER XI)
|
||||
111 0x03BF ο (GREEK SMALL LETTER OMICRON)
|
||||
112 0x03C0 π (GREEK SMALL LETTER PI)
|
||||
113 0x03C1 ρ (GREEK SMALL LETTER RHO)
|
||||
114 0x03C2 ς (GREEK SMALL LETTER FINAL SIGMA)
|
||||
115 0x03C3 σ (GREEK SMALL LETTER SIGMA)
|
||||
116 0x03C4 τ (GREEK SMALL LETTER TAU)
|
||||
117 0x03C5 υ (GREEK SMALL LETTER UPSILON)
|
||||
118 0x03C6 φ (GREEK SMALL LETTER PHI)
|
||||
119 0x03C7 χ (GREEK SMALL LETTER CHI)
|
||||
120 0x03C8 ψ (GREEK SMALL LETTER PSI)
|
||||
121 0x03C9 ω (GREEK SMALL LETTER OMEGA)
|
||||
122 0x03CA ϊ (GREEK SMALL LETTER IOTA WITH DIALYTIKA)
|
||||
123 0x03CB ϋ (GREEK SMALL LETTER UPSILON WITH DIALYTIKA)
|
||||
124 0x03CC ό (GREEK SMALL LETTER OMICRON WITH TONOS)
|
||||
125 0x03CD ύ (GREEK SMALL LETTER UPSILON WITH TONOS)
|
||||
126 0x03CE ώ (GREEK SMALL LETTER OMEGA WITH TONOS)
|
||||
134
test/fixtures/encoding/single-byte/index-windows-1254.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-windows-1254.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-windows-1254.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: e80a27adf377438be8ba5bd223875ea56d6a4d47f958cce1c957a2c446825caa
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x02C6 ˆ (MODIFIER LETTER CIRCUMFLEX ACCENT)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x02DC ˜ (SMALL TILDE)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x00A1 ¡ (INVERTED EXCLAMATION MARK)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x00AA ª (FEMININE ORDINAL INDICATOR)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x00BA º (MASCULINE ORDINAL INDICATOR)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x00BF ¿ (INVERTED QUESTION MARK)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x00C3 Ã (LATIN CAPITAL LETTER A WITH TILDE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x00CC Ì (LATIN CAPITAL LETTER I WITH GRAVE)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x011E Ğ (LATIN CAPITAL LETTER G WITH BREVE)
|
||||
81 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
82 0x00D2 Ò (LATIN CAPITAL LETTER O WITH GRAVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x0130 İ (LATIN CAPITAL LETTER I WITH DOT ABOVE)
|
||||
94 0x015E Ş (LATIN CAPITAL LETTER S WITH CEDILLA)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x00E3 ã (LATIN SMALL LETTER A WITH TILDE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x00EC ì (LATIN SMALL LETTER I WITH GRAVE)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x011F ğ (LATIN SMALL LETTER G WITH BREVE)
|
||||
113 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
114 0x00F2 ò (LATIN SMALL LETTER O WITH GRAVE)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x0131 ı (LATIN SMALL LETTER DOTLESS I)
|
||||
126 0x015F ş (LATIN SMALL LETTER S WITH CEDILLA)
|
||||
127 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
124
test/fixtures/encoding/single-byte/index-windows-1255.txt
vendored
Normal file
124
test/fixtures/encoding/single-byte/index-windows-1255.txt
vendored
Normal file
@@ -0,0 +1,124 @@
|
||||
# For details on index index-windows-1255.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: cd7fb43c97eefa1651084d92d02af53ad668bd848528c18c3b1af5c06b499651
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x02C6 ˆ (MODIFIER LETTER CIRCUMFLEX ACCENT)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x008A (<control>)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x02DC ˜ (SMALL TILDE)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x009A (<control>)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x00A1 ¡ (INVERTED EXCLAMATION MARK)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x20AA ₪ (NEW SHEQEL SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x00D7 × (MULTIPLICATION SIGN)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x00F7 ÷ (DIVISION SIGN)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x00BF ¿ (INVERTED QUESTION MARK)
|
||||
64 0x05B0 ְ (HEBREW POINT SHEVA)
|
||||
65 0x05B1 ֱ (HEBREW POINT HATAF SEGOL)
|
||||
66 0x05B2 ֲ (HEBREW POINT HATAF PATAH)
|
||||
67 0x05B3 ֳ (HEBREW POINT HATAF QAMATS)
|
||||
68 0x05B4 ִ (HEBREW POINT HIRIQ)
|
||||
69 0x05B5 ֵ (HEBREW POINT TSERE)
|
||||
70 0x05B6 ֶ (HEBREW POINT SEGOL)
|
||||
71 0x05B7 ַ (HEBREW POINT PATAH)
|
||||
72 0x05B8 ָ (HEBREW POINT QAMATS)
|
||||
73 0x05B9 ֹ (HEBREW POINT HOLAM)
|
||||
74 0x05BA ֺ (HEBREW POINT HOLAM HASER FOR VAV)
|
||||
75 0x05BB ֻ (HEBREW POINT QUBUTS)
|
||||
76 0x05BC ּ (HEBREW POINT DAGESH OR MAPIQ)
|
||||
77 0x05BD ֽ (HEBREW POINT METEG)
|
||||
78 0x05BE ־ (HEBREW PUNCTUATION MAQAF)
|
||||
79 0x05BF ֿ (HEBREW POINT RAFE)
|
||||
80 0x05C0 ׀ (HEBREW PUNCTUATION PASEQ)
|
||||
81 0x05C1 ׁ (HEBREW POINT SHIN DOT)
|
||||
82 0x05C2 ׂ (HEBREW POINT SIN DOT)
|
||||
83 0x05C3 ׃ (HEBREW PUNCTUATION SOF PASUQ)
|
||||
84 0x05F0 װ (HEBREW LIGATURE YIDDISH DOUBLE VAV)
|
||||
85 0x05F1 ױ (HEBREW LIGATURE YIDDISH VAV YOD)
|
||||
86 0x05F2 ײ (HEBREW LIGATURE YIDDISH DOUBLE YOD)
|
||||
87 0x05F3 ׳ (HEBREW PUNCTUATION GERESH)
|
||||
88 0x05F4 ״ (HEBREW PUNCTUATION GERSHAYIM)
|
||||
96 0x05D0 א (HEBREW LETTER ALEF)
|
||||
97 0x05D1 ב (HEBREW LETTER BET)
|
||||
98 0x05D2 ג (HEBREW LETTER GIMEL)
|
||||
99 0x05D3 ד (HEBREW LETTER DALET)
|
||||
100 0x05D4 ה (HEBREW LETTER HE)
|
||||
101 0x05D5 ו (HEBREW LETTER VAV)
|
||||
102 0x05D6 ז (HEBREW LETTER ZAYIN)
|
||||
103 0x05D7 ח (HEBREW LETTER HET)
|
||||
104 0x05D8 ט (HEBREW LETTER TET)
|
||||
105 0x05D9 י (HEBREW LETTER YOD)
|
||||
106 0x05DA ך (HEBREW LETTER FINAL KAF)
|
||||
107 0x05DB כ (HEBREW LETTER KAF)
|
||||
108 0x05DC ל (HEBREW LETTER LAMED)
|
||||
109 0x05DD ם (HEBREW LETTER FINAL MEM)
|
||||
110 0x05DE מ (HEBREW LETTER MEM)
|
||||
111 0x05DF ן (HEBREW LETTER FINAL NUN)
|
||||
112 0x05E0 נ (HEBREW LETTER NUN)
|
||||
113 0x05E1 ס (HEBREW LETTER SAMEKH)
|
||||
114 0x05E2 ע (HEBREW LETTER AYIN)
|
||||
115 0x05E3 ף (HEBREW LETTER FINAL PE)
|
||||
116 0x05E4 פ (HEBREW LETTER PE)
|
||||
117 0x05E5 ץ (HEBREW LETTER FINAL TSADI)
|
||||
118 0x05E6 צ (HEBREW LETTER TSADI)
|
||||
119 0x05E7 ק (HEBREW LETTER QOF)
|
||||
120 0x05E8 ר (HEBREW LETTER RESH)
|
||||
121 0x05E9 ש (HEBREW LETTER SHIN)
|
||||
122 0x05EA ת (HEBREW LETTER TAV)
|
||||
125 0x200E (LEFT-TO-RIGHT MARK)
|
||||
126 0x200F (RIGHT-TO-LEFT MARK)
|
||||
134
test/fixtures/encoding/single-byte/index-windows-1256.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-windows-1256.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-windows-1256.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 161bdb381f16408e8bebcc8f5310c4190af0e359de8d9bbaa3628ce2f0875509
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x067E پ (ARABIC LETTER PEH)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x02C6 ˆ (MODIFIER LETTER CIRCUMFLEX ACCENT)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x0679 ٹ (ARABIC LETTER TTEH)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
13 0x0686 چ (ARABIC LETTER TCHEH)
|
||||
14 0x0698 ژ (ARABIC LETTER JEH)
|
||||
15 0x0688 ڈ (ARABIC LETTER DDAL)
|
||||
16 0x06AF گ (ARABIC LETTER GAF)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x06A9 ک (ARABIC LETTER KEHEH)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x0691 ڑ (ARABIC LETTER RREH)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
29 0x200C (ZERO WIDTH NON-JOINER)
|
||||
30 0x200D (ZERO WIDTH JOINER)
|
||||
31 0x06BA ں (ARABIC LETTER NOON GHUNNA)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x060C ، (ARABIC COMMA)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x06BE ھ (ARABIC LETTER HEH DOACHASHMEE)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x061B ؛ (ARABIC SEMICOLON)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x061F ؟ (ARABIC QUESTION MARK)
|
||||
64 0x06C1 ہ (ARABIC LETTER HEH GOAL)
|
||||
65 0x0621 ء (ARABIC LETTER HAMZA)
|
||||
66 0x0622 آ (ARABIC LETTER ALEF WITH MADDA ABOVE)
|
||||
67 0x0623 أ (ARABIC LETTER ALEF WITH HAMZA ABOVE)
|
||||
68 0x0624 ؤ (ARABIC LETTER WAW WITH HAMZA ABOVE)
|
||||
69 0x0625 إ (ARABIC LETTER ALEF WITH HAMZA BELOW)
|
||||
70 0x0626 ئ (ARABIC LETTER YEH WITH HAMZA ABOVE)
|
||||
71 0x0627 ا (ARABIC LETTER ALEF)
|
||||
72 0x0628 ب (ARABIC LETTER BEH)
|
||||
73 0x0629 ة (ARABIC LETTER TEH MARBUTA)
|
||||
74 0x062A ت (ARABIC LETTER TEH)
|
||||
75 0x062B ث (ARABIC LETTER THEH)
|
||||
76 0x062C ج (ARABIC LETTER JEEM)
|
||||
77 0x062D ح (ARABIC LETTER HAH)
|
||||
78 0x062E خ (ARABIC LETTER KHAH)
|
||||
79 0x062F د (ARABIC LETTER DAL)
|
||||
80 0x0630 ذ (ARABIC LETTER THAL)
|
||||
81 0x0631 ر (ARABIC LETTER REH)
|
||||
82 0x0632 ز (ARABIC LETTER ZAIN)
|
||||
83 0x0633 س (ARABIC LETTER SEEN)
|
||||
84 0x0634 ش (ARABIC LETTER SHEEN)
|
||||
85 0x0635 ص (ARABIC LETTER SAD)
|
||||
86 0x0636 ض (ARABIC LETTER DAD)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x0637 ط (ARABIC LETTER TAH)
|
||||
89 0x0638 ظ (ARABIC LETTER ZAH)
|
||||
90 0x0639 ع (ARABIC LETTER AIN)
|
||||
91 0x063A غ (ARABIC LETTER GHAIN)
|
||||
92 0x0640 ـ (ARABIC TATWEEL)
|
||||
93 0x0641 ف (ARABIC LETTER FEH)
|
||||
94 0x0642 ق (ARABIC LETTER QAF)
|
||||
95 0x0643 ك (ARABIC LETTER KAF)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x0644 ل (ARABIC LETTER LAM)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x0645 م (ARABIC LETTER MEEM)
|
||||
100 0x0646 ن (ARABIC LETTER NOON)
|
||||
101 0x0647 ه (ARABIC LETTER HEH)
|
||||
102 0x0648 و (ARABIC LETTER WAW)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x0649 ى (ARABIC LETTER ALEF MAKSURA)
|
||||
109 0x064A ي (ARABIC LETTER YEH)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x064B ً (ARABIC FATHATAN)
|
||||
113 0x064C ٌ (ARABIC DAMMATAN)
|
||||
114 0x064D ٍ (ARABIC KASRATAN)
|
||||
115 0x064E َ (ARABIC FATHA)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x064F ُ (ARABIC DAMMA)
|
||||
118 0x0650 ِ (ARABIC KASRA)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x0651 ّ (ARABIC SHADDA)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x0652 ْ (ARABIC SUKUN)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x200E (LEFT-TO-RIGHT MARK)
|
||||
126 0x200F (RIGHT-TO-LEFT MARK)
|
||||
127 0x06D2 ے (ARABIC LETTER YEH BARREE)
|
||||
132
test/fixtures/encoding/single-byte/index-windows-1257.txt
vendored
Normal file
132
test/fixtures/encoding/single-byte/index-windows-1257.txt
vendored
Normal file
@@ -0,0 +1,132 @@
|
||||
# For details on index index-windows-1257.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: cc7256bdd10a5b8dc7fb6f994659f307dfcae60def9aa6c29d811f85e2842c47
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0083 (<control>)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x0088 (<control>)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x008A (<control>)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x008C (<control>)
|
||||
13 0x00A8 ¨ (DIAERESIS)
|
||||
14 0x02C7 ˇ (CARON)
|
||||
15 0x00B8 ¸ (CEDILLA)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x0098 (<control>)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x009A (<control>)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x009C (<control>)
|
||||
29 0x00AF ¯ (MACRON)
|
||||
30 0x02DB ˛ (OGONEK)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x0156 Ŗ (LATIN CAPITAL LETTER R WITH CEDILLA)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x0157 ŗ (LATIN SMALL LETTER R WITH CEDILLA)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
64 0x0104 Ą (LATIN CAPITAL LETTER A WITH OGONEK)
|
||||
65 0x012E Į (LATIN CAPITAL LETTER I WITH OGONEK)
|
||||
66 0x0100 Ā (LATIN CAPITAL LETTER A WITH MACRON)
|
||||
67 0x0106 Ć (LATIN CAPITAL LETTER C WITH ACUTE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x0118 Ę (LATIN CAPITAL LETTER E WITH OGONEK)
|
||||
71 0x0112 Ē (LATIN CAPITAL LETTER E WITH MACRON)
|
||||
72 0x010C Č (LATIN CAPITAL LETTER C WITH CARON)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x0179 Ź (LATIN CAPITAL LETTER Z WITH ACUTE)
|
||||
75 0x0116 Ė (LATIN CAPITAL LETTER E WITH DOT ABOVE)
|
||||
76 0x0122 Ģ (LATIN CAPITAL LETTER G WITH CEDILLA)
|
||||
77 0x0136 Ķ (LATIN CAPITAL LETTER K WITH CEDILLA)
|
||||
78 0x012A Ī (LATIN CAPITAL LETTER I WITH MACRON)
|
||||
79 0x013B Ļ (LATIN CAPITAL LETTER L WITH CEDILLA)
|
||||
80 0x0160 Š (LATIN CAPITAL LETTER S WITH CARON)
|
||||
81 0x0143 Ń (LATIN CAPITAL LETTER N WITH ACUTE)
|
||||
82 0x0145 Ņ (LATIN CAPITAL LETTER N WITH CEDILLA)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x014C Ō (LATIN CAPITAL LETTER O WITH MACRON)
|
||||
85 0x00D5 Õ (LATIN CAPITAL LETTER O WITH TILDE)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x0172 Ų (LATIN CAPITAL LETTER U WITH OGONEK)
|
||||
89 0x0141 Ł (LATIN CAPITAL LETTER L WITH STROKE)
|
||||
90 0x015A Ś (LATIN CAPITAL LETTER S WITH ACUTE)
|
||||
91 0x016A Ū (LATIN CAPITAL LETTER U WITH MACRON)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x017B Ż (LATIN CAPITAL LETTER Z WITH DOT ABOVE)
|
||||
94 0x017D Ž (LATIN CAPITAL LETTER Z WITH CARON)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x0105 ą (LATIN SMALL LETTER A WITH OGONEK)
|
||||
97 0x012F į (LATIN SMALL LETTER I WITH OGONEK)
|
||||
98 0x0101 ā (LATIN SMALL LETTER A WITH MACRON)
|
||||
99 0x0107 ć (LATIN SMALL LETTER C WITH ACUTE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x0119 ę (LATIN SMALL LETTER E WITH OGONEK)
|
||||
103 0x0113 ē (LATIN SMALL LETTER E WITH MACRON)
|
||||
104 0x010D č (LATIN SMALL LETTER C WITH CARON)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x017A ź (LATIN SMALL LETTER Z WITH ACUTE)
|
||||
107 0x0117 ė (LATIN SMALL LETTER E WITH DOT ABOVE)
|
||||
108 0x0123 ģ (LATIN SMALL LETTER G WITH CEDILLA)
|
||||
109 0x0137 ķ (LATIN SMALL LETTER K WITH CEDILLA)
|
||||
110 0x012B ī (LATIN SMALL LETTER I WITH MACRON)
|
||||
111 0x013C ļ (LATIN SMALL LETTER L WITH CEDILLA)
|
||||
112 0x0161 š (LATIN SMALL LETTER S WITH CARON)
|
||||
113 0x0144 ń (LATIN SMALL LETTER N WITH ACUTE)
|
||||
114 0x0146 ņ (LATIN SMALL LETTER N WITH CEDILLA)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x014D ō (LATIN SMALL LETTER O WITH MACRON)
|
||||
117 0x00F5 õ (LATIN SMALL LETTER O WITH TILDE)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x0173 ų (LATIN SMALL LETTER U WITH OGONEK)
|
||||
121 0x0142 ł (LATIN SMALL LETTER L WITH STROKE)
|
||||
122 0x015B ś (LATIN SMALL LETTER S WITH ACUTE)
|
||||
123 0x016B ū (LATIN SMALL LETTER U WITH MACRON)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x017C ż (LATIN SMALL LETTER Z WITH DOT ABOVE)
|
||||
126 0x017E ž (LATIN SMALL LETTER Z WITH CARON)
|
||||
127 0x02D9 ˙ (DOT ABOVE)
|
||||
134
test/fixtures/encoding/single-byte/index-windows-1258.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-windows-1258.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-windows-1258.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 198bacedfcf24390e219240a7b776b6cec34cff070330b08a601a69c67f7eb24
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x201A ‚ (SINGLE LOW-9 QUOTATION MARK)
|
||||
3 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
4 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x2020 † (DAGGER)
|
||||
7 0x2021 ‡ (DOUBLE DAGGER)
|
||||
8 0x02C6 ˆ (MODIFIER LETTER CIRCUMFLEX ACCENT)
|
||||
9 0x2030 ‰ (PER MILLE SIGN)
|
||||
10 0x008A (<control>)
|
||||
11 0x2039 ‹ (SINGLE LEFT-POINTING ANGLE QUOTATION MARK)
|
||||
12 0x0152 Π(LATIN CAPITAL LIGATURE OE)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x02DC ˜ (SMALL TILDE)
|
||||
25 0x2122 ™ (TRADE MARK SIGN)
|
||||
26 0x009A (<control>)
|
||||
27 0x203A › (SINGLE RIGHT-POINTING ANGLE QUOTATION MARK)
|
||||
28 0x0153 œ (LATIN SMALL LIGATURE OE)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x0178 Ÿ (LATIN CAPITAL LETTER Y WITH DIAERESIS)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x00A1 ¡ (INVERTED EXCLAMATION MARK)
|
||||
34 0x00A2 ¢ (CENT SIGN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A4 ¤ (CURRENCY SIGN)
|
||||
37 0x00A5 ¥ (YEN SIGN)
|
||||
38 0x00A6 ¦ (BROKEN BAR)
|
||||
39 0x00A7 § (SECTION SIGN)
|
||||
40 0x00A8 ¨ (DIAERESIS)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x00AA ª (FEMININE ORDINAL INDICATOR)
|
||||
43 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
44 0x00AC ¬ (NOT SIGN)
|
||||
45 0x00AD (SOFT HYPHEN)
|
||||
46 0x00AE ® (REGISTERED SIGN)
|
||||
47 0x00AF ¯ (MACRON)
|
||||
48 0x00B0 ° (DEGREE SIGN)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x00B2 ² (SUPERSCRIPT TWO)
|
||||
51 0x00B3 ³ (SUPERSCRIPT THREE)
|
||||
52 0x00B4 ´ (ACUTE ACCENT)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x00B6 ¶ (PILCROW SIGN)
|
||||
55 0x00B7 · (MIDDLE DOT)
|
||||
56 0x00B8 ¸ (CEDILLA)
|
||||
57 0x00B9 ¹ (SUPERSCRIPT ONE)
|
||||
58 0x00BA º (MASCULINE ORDINAL INDICATOR)
|
||||
59 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
60 0x00BC ¼ (VULGAR FRACTION ONE QUARTER)
|
||||
61 0x00BD ½ (VULGAR FRACTION ONE HALF)
|
||||
62 0x00BE ¾ (VULGAR FRACTION THREE QUARTERS)
|
||||
63 0x00BF ¿ (INVERTED QUESTION MARK)
|
||||
64 0x00C0 À (LATIN CAPITAL LETTER A WITH GRAVE)
|
||||
65 0x00C1 Á (LATIN CAPITAL LETTER A WITH ACUTE)
|
||||
66 0x00C2 Â (LATIN CAPITAL LETTER A WITH CIRCUMFLEX)
|
||||
67 0x0102 Ă (LATIN CAPITAL LETTER A WITH BREVE)
|
||||
68 0x00C4 Ä (LATIN CAPITAL LETTER A WITH DIAERESIS)
|
||||
69 0x00C5 Å (LATIN CAPITAL LETTER A WITH RING ABOVE)
|
||||
70 0x00C6 Æ (LATIN CAPITAL LETTER AE)
|
||||
71 0x00C7 Ç (LATIN CAPITAL LETTER C WITH CEDILLA)
|
||||
72 0x00C8 È (LATIN CAPITAL LETTER E WITH GRAVE)
|
||||
73 0x00C9 É (LATIN CAPITAL LETTER E WITH ACUTE)
|
||||
74 0x00CA Ê (LATIN CAPITAL LETTER E WITH CIRCUMFLEX)
|
||||
75 0x00CB Ë (LATIN CAPITAL LETTER E WITH DIAERESIS)
|
||||
76 0x0300 ̀ (COMBINING GRAVE ACCENT)
|
||||
77 0x00CD Í (LATIN CAPITAL LETTER I WITH ACUTE)
|
||||
78 0x00CE Î (LATIN CAPITAL LETTER I WITH CIRCUMFLEX)
|
||||
79 0x00CF Ï (LATIN CAPITAL LETTER I WITH DIAERESIS)
|
||||
80 0x0110 Đ (LATIN CAPITAL LETTER D WITH STROKE)
|
||||
81 0x00D1 Ñ (LATIN CAPITAL LETTER N WITH TILDE)
|
||||
82 0x0309 ̉ (COMBINING HOOK ABOVE)
|
||||
83 0x00D3 Ó (LATIN CAPITAL LETTER O WITH ACUTE)
|
||||
84 0x00D4 Ô (LATIN CAPITAL LETTER O WITH CIRCUMFLEX)
|
||||
85 0x01A0 Ơ (LATIN CAPITAL LETTER O WITH HORN)
|
||||
86 0x00D6 Ö (LATIN CAPITAL LETTER O WITH DIAERESIS)
|
||||
87 0x00D7 × (MULTIPLICATION SIGN)
|
||||
88 0x00D8 Ø (LATIN CAPITAL LETTER O WITH STROKE)
|
||||
89 0x00D9 Ù (LATIN CAPITAL LETTER U WITH GRAVE)
|
||||
90 0x00DA Ú (LATIN CAPITAL LETTER U WITH ACUTE)
|
||||
91 0x00DB Û (LATIN CAPITAL LETTER U WITH CIRCUMFLEX)
|
||||
92 0x00DC Ü (LATIN CAPITAL LETTER U WITH DIAERESIS)
|
||||
93 0x01AF Ư (LATIN CAPITAL LETTER U WITH HORN)
|
||||
94 0x0303 ̃ (COMBINING TILDE)
|
||||
95 0x00DF ß (LATIN SMALL LETTER SHARP S)
|
||||
96 0x00E0 à (LATIN SMALL LETTER A WITH GRAVE)
|
||||
97 0x00E1 á (LATIN SMALL LETTER A WITH ACUTE)
|
||||
98 0x00E2 â (LATIN SMALL LETTER A WITH CIRCUMFLEX)
|
||||
99 0x0103 ă (LATIN SMALL LETTER A WITH BREVE)
|
||||
100 0x00E4 ä (LATIN SMALL LETTER A WITH DIAERESIS)
|
||||
101 0x00E5 å (LATIN SMALL LETTER A WITH RING ABOVE)
|
||||
102 0x00E6 æ (LATIN SMALL LETTER AE)
|
||||
103 0x00E7 ç (LATIN SMALL LETTER C WITH CEDILLA)
|
||||
104 0x00E8 è (LATIN SMALL LETTER E WITH GRAVE)
|
||||
105 0x00E9 é (LATIN SMALL LETTER E WITH ACUTE)
|
||||
106 0x00EA ê (LATIN SMALL LETTER E WITH CIRCUMFLEX)
|
||||
107 0x00EB ë (LATIN SMALL LETTER E WITH DIAERESIS)
|
||||
108 0x0301 ́ (COMBINING ACUTE ACCENT)
|
||||
109 0x00ED í (LATIN SMALL LETTER I WITH ACUTE)
|
||||
110 0x00EE î (LATIN SMALL LETTER I WITH CIRCUMFLEX)
|
||||
111 0x00EF ï (LATIN SMALL LETTER I WITH DIAERESIS)
|
||||
112 0x0111 đ (LATIN SMALL LETTER D WITH STROKE)
|
||||
113 0x00F1 ñ (LATIN SMALL LETTER N WITH TILDE)
|
||||
114 0x0323 ̣ (COMBINING DOT BELOW)
|
||||
115 0x00F3 ó (LATIN SMALL LETTER O WITH ACUTE)
|
||||
116 0x00F4 ô (LATIN SMALL LETTER O WITH CIRCUMFLEX)
|
||||
117 0x01A1 ơ (LATIN SMALL LETTER O WITH HORN)
|
||||
118 0x00F6 ö (LATIN SMALL LETTER O WITH DIAERESIS)
|
||||
119 0x00F7 ÷ (DIVISION SIGN)
|
||||
120 0x00F8 ø (LATIN SMALL LETTER O WITH STROKE)
|
||||
121 0x00F9 ù (LATIN SMALL LETTER U WITH GRAVE)
|
||||
122 0x00FA ú (LATIN SMALL LETTER U WITH ACUTE)
|
||||
123 0x00FB û (LATIN SMALL LETTER U WITH CIRCUMFLEX)
|
||||
124 0x00FC ü (LATIN SMALL LETTER U WITH DIAERESIS)
|
||||
125 0x01B0 ư (LATIN SMALL LETTER U WITH HORN)
|
||||
126 0x20AB ₫ (DONG SIGN)
|
||||
127 0x00FF ÿ (LATIN SMALL LETTER Y WITH DIAERESIS)
|
||||
126
test/fixtures/encoding/single-byte/index-windows-874.txt
vendored
Normal file
126
test/fixtures/encoding/single-byte/index-windows-874.txt
vendored
Normal file
@@ -0,0 +1,126 @@
|
||||
# For details on index index-windows-874.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: b416583ce125e38474381b31b401a98b19ecf2e57e0998e78a1e18b14894905d
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x20AC € (EURO SIGN)
|
||||
1 0x0081 (<control>)
|
||||
2 0x0082 (<control>)
|
||||
3 0x0083 (<control>)
|
||||
4 0x0084 (<control>)
|
||||
5 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
6 0x0086 (<control>)
|
||||
7 0x0087 (<control>)
|
||||
8 0x0088 (<control>)
|
||||
9 0x0089 (<control>)
|
||||
10 0x008A (<control>)
|
||||
11 0x008B (<control>)
|
||||
12 0x008C (<control>)
|
||||
13 0x008D (<control>)
|
||||
14 0x008E (<control>)
|
||||
15 0x008F (<control>)
|
||||
16 0x0090 (<control>)
|
||||
17 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
18 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
19 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
20 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
21 0x2022 • (BULLET)
|
||||
22 0x2013 – (EN DASH)
|
||||
23 0x2014 — (EM DASH)
|
||||
24 0x0098 (<control>)
|
||||
25 0x0099 (<control>)
|
||||
26 0x009A (<control>)
|
||||
27 0x009B (<control>)
|
||||
28 0x009C (<control>)
|
||||
29 0x009D (<control>)
|
||||
30 0x009E (<control>)
|
||||
31 0x009F (<control>)
|
||||
32 0x00A0 (NO-BREAK SPACE)
|
||||
33 0x0E01 ก (THAI CHARACTER KO KAI)
|
||||
34 0x0E02 ข (THAI CHARACTER KHO KHAI)
|
||||
35 0x0E03 ฃ (THAI CHARACTER KHO KHUAT)
|
||||
36 0x0E04 ค (THAI CHARACTER KHO KHWAI)
|
||||
37 0x0E05 ฅ (THAI CHARACTER KHO KHON)
|
||||
38 0x0E06 ฆ (THAI CHARACTER KHO RAKHANG)
|
||||
39 0x0E07 ง (THAI CHARACTER NGO NGU)
|
||||
40 0x0E08 จ (THAI CHARACTER CHO CHAN)
|
||||
41 0x0E09 ฉ (THAI CHARACTER CHO CHING)
|
||||
42 0x0E0A ช (THAI CHARACTER CHO CHANG)
|
||||
43 0x0E0B ซ (THAI CHARACTER SO SO)
|
||||
44 0x0E0C ฌ (THAI CHARACTER CHO CHOE)
|
||||
45 0x0E0D ญ (THAI CHARACTER YO YING)
|
||||
46 0x0E0E ฎ (THAI CHARACTER DO CHADA)
|
||||
47 0x0E0F ฏ (THAI CHARACTER TO PATAK)
|
||||
48 0x0E10 ฐ (THAI CHARACTER THO THAN)
|
||||
49 0x0E11 ฑ (THAI CHARACTER THO NANGMONTHO)
|
||||
50 0x0E12 ฒ (THAI CHARACTER THO PHUTHAO)
|
||||
51 0x0E13 ณ (THAI CHARACTER NO NEN)
|
||||
52 0x0E14 ด (THAI CHARACTER DO DEK)
|
||||
53 0x0E15 ต (THAI CHARACTER TO TAO)
|
||||
54 0x0E16 ถ (THAI CHARACTER THO THUNG)
|
||||
55 0x0E17 ท (THAI CHARACTER THO THAHAN)
|
||||
56 0x0E18 ธ (THAI CHARACTER THO THONG)
|
||||
57 0x0E19 น (THAI CHARACTER NO NU)
|
||||
58 0x0E1A บ (THAI CHARACTER BO BAIMAI)
|
||||
59 0x0E1B ป (THAI CHARACTER PO PLA)
|
||||
60 0x0E1C ผ (THAI CHARACTER PHO PHUNG)
|
||||
61 0x0E1D ฝ (THAI CHARACTER FO FA)
|
||||
62 0x0E1E พ (THAI CHARACTER PHO PHAN)
|
||||
63 0x0E1F ฟ (THAI CHARACTER FO FAN)
|
||||
64 0x0E20 ภ (THAI CHARACTER PHO SAMPHAO)
|
||||
65 0x0E21 ม (THAI CHARACTER MO MA)
|
||||
66 0x0E22 ย (THAI CHARACTER YO YAK)
|
||||
67 0x0E23 ร (THAI CHARACTER RO RUA)
|
||||
68 0x0E24 ฤ (THAI CHARACTER RU)
|
||||
69 0x0E25 ล (THAI CHARACTER LO LING)
|
||||
70 0x0E26 ฦ (THAI CHARACTER LU)
|
||||
71 0x0E27 ว (THAI CHARACTER WO WAEN)
|
||||
72 0x0E28 ศ (THAI CHARACTER SO SALA)
|
||||
73 0x0E29 ษ (THAI CHARACTER SO RUSI)
|
||||
74 0x0E2A ส (THAI CHARACTER SO SUA)
|
||||
75 0x0E2B ห (THAI CHARACTER HO HIP)
|
||||
76 0x0E2C ฬ (THAI CHARACTER LO CHULA)
|
||||
77 0x0E2D อ (THAI CHARACTER O ANG)
|
||||
78 0x0E2E ฮ (THAI CHARACTER HO NOKHUK)
|
||||
79 0x0E2F ฯ (THAI CHARACTER PAIYANNOI)
|
||||
80 0x0E30 ะ (THAI CHARACTER SARA A)
|
||||
81 0x0E31 ั (THAI CHARACTER MAI HAN-AKAT)
|
||||
82 0x0E32 า (THAI CHARACTER SARA AA)
|
||||
83 0x0E33 ำ (THAI CHARACTER SARA AM)
|
||||
84 0x0E34 ิ (THAI CHARACTER SARA I)
|
||||
85 0x0E35 ี (THAI CHARACTER SARA II)
|
||||
86 0x0E36 ึ (THAI CHARACTER SARA UE)
|
||||
87 0x0E37 ื (THAI CHARACTER SARA UEE)
|
||||
88 0x0E38 ุ (THAI CHARACTER SARA U)
|
||||
89 0x0E39 ู (THAI CHARACTER SARA UU)
|
||||
90 0x0E3A ฺ (THAI CHARACTER PHINTHU)
|
||||
95 0x0E3F ฿ (THAI CURRENCY SYMBOL BAHT)
|
||||
96 0x0E40 เ (THAI CHARACTER SARA E)
|
||||
97 0x0E41 แ (THAI CHARACTER SARA AE)
|
||||
98 0x0E42 โ (THAI CHARACTER SARA O)
|
||||
99 0x0E43 ใ (THAI CHARACTER SARA AI MAIMUAN)
|
||||
100 0x0E44 ไ (THAI CHARACTER SARA AI MAIMALAI)
|
||||
101 0x0E45 ๅ (THAI CHARACTER LAKKHANGYAO)
|
||||
102 0x0E46 ๆ (THAI CHARACTER MAIYAMOK)
|
||||
103 0x0E47 ็ (THAI CHARACTER MAITAIKHU)
|
||||
104 0x0E48 ่ (THAI CHARACTER MAI EK)
|
||||
105 0x0E49 ้ (THAI CHARACTER MAI THO)
|
||||
106 0x0E4A ๊ (THAI CHARACTER MAI TRI)
|
||||
107 0x0E4B ๋ (THAI CHARACTER MAI CHATTAWA)
|
||||
108 0x0E4C ์ (THAI CHARACTER THANTHAKHAT)
|
||||
109 0x0E4D ํ (THAI CHARACTER NIKHAHIT)
|
||||
110 0x0E4E ๎ (THAI CHARACTER YAMAKKAN)
|
||||
111 0x0E4F ๏ (THAI CHARACTER FONGMAN)
|
||||
112 0x0E50 ๐ (THAI DIGIT ZERO)
|
||||
113 0x0E51 ๑ (THAI DIGIT ONE)
|
||||
114 0x0E52 ๒ (THAI DIGIT TWO)
|
||||
115 0x0E53 ๓ (THAI DIGIT THREE)
|
||||
116 0x0E54 ๔ (THAI DIGIT FOUR)
|
||||
117 0x0E55 ๕ (THAI DIGIT FIVE)
|
||||
118 0x0E56 ๖ (THAI DIGIT SIX)
|
||||
119 0x0E57 ๗ (THAI DIGIT SEVEN)
|
||||
120 0x0E58 ๘ (THAI DIGIT EIGHT)
|
||||
121 0x0E59 ๙ (THAI DIGIT NINE)
|
||||
122 0x0E5A ๚ (THAI CHARACTER ANGKHANKHU)
|
||||
123 0x0E5B ๛ (THAI CHARACTER KHOMUT)
|
||||
134
test/fixtures/encoding/single-byte/index-x-mac-cyrillic.txt
vendored
Normal file
134
test/fixtures/encoding/single-byte/index-x-mac-cyrillic.txt
vendored
Normal file
@@ -0,0 +1,134 @@
|
||||
# For details on index index-x-mac-cyrillic.txt see the Encoding Standard
|
||||
# https://encoding.spec.whatwg.org/
|
||||
#
|
||||
# Identifier: 73e8e7642c6fa9de29d42819b47fba55b58666fb1e339faeb4a89a0bd7c24d43
|
||||
# Date: 2024-09-18
|
||||
|
||||
0 0x0410 А (CYRILLIC CAPITAL LETTER A)
|
||||
1 0x0411 Б (CYRILLIC CAPITAL LETTER BE)
|
||||
2 0x0412 В (CYRILLIC CAPITAL LETTER VE)
|
||||
3 0x0413 Г (CYRILLIC CAPITAL LETTER GHE)
|
||||
4 0x0414 Д (CYRILLIC CAPITAL LETTER DE)
|
||||
5 0x0415 Е (CYRILLIC CAPITAL LETTER IE)
|
||||
6 0x0416 Ж (CYRILLIC CAPITAL LETTER ZHE)
|
||||
7 0x0417 З (CYRILLIC CAPITAL LETTER ZE)
|
||||
8 0x0418 И (CYRILLIC CAPITAL LETTER I)
|
||||
9 0x0419 Й (CYRILLIC CAPITAL LETTER SHORT I)
|
||||
10 0x041A К (CYRILLIC CAPITAL LETTER KA)
|
||||
11 0x041B Л (CYRILLIC CAPITAL LETTER EL)
|
||||
12 0x041C М (CYRILLIC CAPITAL LETTER EM)
|
||||
13 0x041D Н (CYRILLIC CAPITAL LETTER EN)
|
||||
14 0x041E О (CYRILLIC CAPITAL LETTER O)
|
||||
15 0x041F П (CYRILLIC CAPITAL LETTER PE)
|
||||
16 0x0420 Р (CYRILLIC CAPITAL LETTER ER)
|
||||
17 0x0421 С (CYRILLIC CAPITAL LETTER ES)
|
||||
18 0x0422 Т (CYRILLIC CAPITAL LETTER TE)
|
||||
19 0x0423 У (CYRILLIC CAPITAL LETTER U)
|
||||
20 0x0424 Ф (CYRILLIC CAPITAL LETTER EF)
|
||||
21 0x0425 Х (CYRILLIC CAPITAL LETTER HA)
|
||||
22 0x0426 Ц (CYRILLIC CAPITAL LETTER TSE)
|
||||
23 0x0427 Ч (CYRILLIC CAPITAL LETTER CHE)
|
||||
24 0x0428 Ш (CYRILLIC CAPITAL LETTER SHA)
|
||||
25 0x0429 Щ (CYRILLIC CAPITAL LETTER SHCHA)
|
||||
26 0x042A Ъ (CYRILLIC CAPITAL LETTER HARD SIGN)
|
||||
27 0x042B Ы (CYRILLIC CAPITAL LETTER YERU)
|
||||
28 0x042C Ь (CYRILLIC CAPITAL LETTER SOFT SIGN)
|
||||
29 0x042D Э (CYRILLIC CAPITAL LETTER E)
|
||||
30 0x042E Ю (CYRILLIC CAPITAL LETTER YU)
|
||||
31 0x042F Я (CYRILLIC CAPITAL LETTER YA)
|
||||
32 0x2020 † (DAGGER)
|
||||
33 0x00B0 ° (DEGREE SIGN)
|
||||
34 0x0490 Ґ (CYRILLIC CAPITAL LETTER GHE WITH UPTURN)
|
||||
35 0x00A3 £ (POUND SIGN)
|
||||
36 0x00A7 § (SECTION SIGN)
|
||||
37 0x2022 • (BULLET)
|
||||
38 0x00B6 ¶ (PILCROW SIGN)
|
||||
39 0x0406 І (CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
40 0x00AE ® (REGISTERED SIGN)
|
||||
41 0x00A9 © (COPYRIGHT SIGN)
|
||||
42 0x2122 ™ (TRADE MARK SIGN)
|
||||
43 0x0402 Ђ (CYRILLIC CAPITAL LETTER DJE)
|
||||
44 0x0452 ђ (CYRILLIC SMALL LETTER DJE)
|
||||
45 0x2260 ≠ (NOT EQUAL TO)
|
||||
46 0x0403 Ѓ (CYRILLIC CAPITAL LETTER GJE)
|
||||
47 0x0453 ѓ (CYRILLIC SMALL LETTER GJE)
|
||||
48 0x221E ∞ (INFINITY)
|
||||
49 0x00B1 ± (PLUS-MINUS SIGN)
|
||||
50 0x2264 ≤ (LESS-THAN OR EQUAL TO)
|
||||
51 0x2265 ≥ (GREATER-THAN OR EQUAL TO)
|
||||
52 0x0456 і (CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I)
|
||||
53 0x00B5 µ (MICRO SIGN)
|
||||
54 0x0491 ґ (CYRILLIC SMALL LETTER GHE WITH UPTURN)
|
||||
55 0x0408 Ј (CYRILLIC CAPITAL LETTER JE)
|
||||
56 0x0404 Є (CYRILLIC CAPITAL LETTER UKRAINIAN IE)
|
||||
57 0x0454 є (CYRILLIC SMALL LETTER UKRAINIAN IE)
|
||||
58 0x0407 Ї (CYRILLIC CAPITAL LETTER YI)
|
||||
59 0x0457 ї (CYRILLIC SMALL LETTER YI)
|
||||
60 0x0409 Љ (CYRILLIC CAPITAL LETTER LJE)
|
||||
61 0x0459 љ (CYRILLIC SMALL LETTER LJE)
|
||||
62 0x040A Њ (CYRILLIC CAPITAL LETTER NJE)
|
||||
63 0x045A њ (CYRILLIC SMALL LETTER NJE)
|
||||
64 0x0458 ј (CYRILLIC SMALL LETTER JE)
|
||||
65 0x0405 Ѕ (CYRILLIC CAPITAL LETTER DZE)
|
||||
66 0x00AC ¬ (NOT SIGN)
|
||||
67 0x221A √ (SQUARE ROOT)
|
||||
68 0x0192 ƒ (LATIN SMALL LETTER F WITH HOOK)
|
||||
69 0x2248 ≈ (ALMOST EQUAL TO)
|
||||
70 0x2206 ∆ (INCREMENT)
|
||||
71 0x00AB « (LEFT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
72 0x00BB » (RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK)
|
||||
73 0x2026 … (HORIZONTAL ELLIPSIS)
|
||||
74 0x00A0 (NO-BREAK SPACE)
|
||||
75 0x040B Ћ (CYRILLIC CAPITAL LETTER TSHE)
|
||||
76 0x045B ћ (CYRILLIC SMALL LETTER TSHE)
|
||||
77 0x040C Ќ (CYRILLIC CAPITAL LETTER KJE)
|
||||
78 0x045C ќ (CYRILLIC SMALL LETTER KJE)
|
||||
79 0x0455 ѕ (CYRILLIC SMALL LETTER DZE)
|
||||
80 0x2013 – (EN DASH)
|
||||
81 0x2014 — (EM DASH)
|
||||
82 0x201C “ (LEFT DOUBLE QUOTATION MARK)
|
||||
83 0x201D ” (RIGHT DOUBLE QUOTATION MARK)
|
||||
84 0x2018 ‘ (LEFT SINGLE QUOTATION MARK)
|
||||
85 0x2019 ’ (RIGHT SINGLE QUOTATION MARK)
|
||||
86 0x00F7 ÷ (DIVISION SIGN)
|
||||
87 0x201E „ (DOUBLE LOW-9 QUOTATION MARK)
|
||||
88 0x040E Ў (CYRILLIC CAPITAL LETTER SHORT U)
|
||||
89 0x045E ў (CYRILLIC SMALL LETTER SHORT U)
|
||||
90 0x040F Џ (CYRILLIC CAPITAL LETTER DZHE)
|
||||
91 0x045F џ (CYRILLIC SMALL LETTER DZHE)
|
||||
92 0x2116 № (NUMERO SIGN)
|
||||
93 0x0401 Ё (CYRILLIC CAPITAL LETTER IO)
|
||||
94 0x0451 ё (CYRILLIC SMALL LETTER IO)
|
||||
95 0x044F я (CYRILLIC SMALL LETTER YA)
|
||||
96 0x0430 а (CYRILLIC SMALL LETTER A)
|
||||
97 0x0431 б (CYRILLIC SMALL LETTER BE)
|
||||
98 0x0432 в (CYRILLIC SMALL LETTER VE)
|
||||
99 0x0433 г (CYRILLIC SMALL LETTER GHE)
|
||||
100 0x0434 д (CYRILLIC SMALL LETTER DE)
|
||||
101 0x0435 е (CYRILLIC SMALL LETTER IE)
|
||||
102 0x0436 ж (CYRILLIC SMALL LETTER ZHE)
|
||||
103 0x0437 з (CYRILLIC SMALL LETTER ZE)
|
||||
104 0x0438 и (CYRILLIC SMALL LETTER I)
|
||||
105 0x0439 й (CYRILLIC SMALL LETTER SHORT I)
|
||||
106 0x043A к (CYRILLIC SMALL LETTER KA)
|
||||
107 0x043B л (CYRILLIC SMALL LETTER EL)
|
||||
108 0x043C м (CYRILLIC SMALL LETTER EM)
|
||||
109 0x043D н (CYRILLIC SMALL LETTER EN)
|
||||
110 0x043E о (CYRILLIC SMALL LETTER O)
|
||||
111 0x043F п (CYRILLIC SMALL LETTER PE)
|
||||
112 0x0440 р (CYRILLIC SMALL LETTER ER)
|
||||
113 0x0441 с (CYRILLIC SMALL LETTER ES)
|
||||
114 0x0442 т (CYRILLIC SMALL LETTER TE)
|
||||
115 0x0443 у (CYRILLIC SMALL LETTER U)
|
||||
116 0x0444 ф (CYRILLIC SMALL LETTER EF)
|
||||
117 0x0445 х (CYRILLIC SMALL LETTER HA)
|
||||
118 0x0446 ц (CYRILLIC SMALL LETTER TSE)
|
||||
119 0x0447 ч (CYRILLIC SMALL LETTER CHE)
|
||||
120 0x0448 ш (CYRILLIC SMALL LETTER SHA)
|
||||
121 0x0449 щ (CYRILLIC SMALL LETTER SHCHA)
|
||||
122 0x044A ъ (CYRILLIC SMALL LETTER HARD SIGN)
|
||||
123 0x044B ы (CYRILLIC SMALL LETTER YERU)
|
||||
124 0x044C ь (CYRILLIC SMALL LETTER SOFT SIGN)
|
||||
125 0x044D э (CYRILLIC SMALL LETTER E)
|
||||
126 0x044E ю (CYRILLIC SMALL LETTER YU)
|
||||
127 0x20AC € (EURO SIGN)
|
||||
@@ -88,6 +88,7 @@ expected.beforePreExec = new Set([
|
||||
'NativeModule internal/process/signal',
|
||||
'Internal Binding fs',
|
||||
'NativeModule internal/encoding',
|
||||
'NativeModule internal/encoding/single-byte',
|
||||
'NativeModule internal/blob',
|
||||
'NativeModule internal/fs/utils',
|
||||
'NativeModule fs',
|
||||
|
||||
@@ -1,53 +0,0 @@
|
||||
// Flags: --expose-internals
|
||||
|
||||
'use strict';
|
||||
|
||||
require('../common');
|
||||
|
||||
const assert = require('node:assert');
|
||||
const { internalBinding } = require('internal/test/binding');
|
||||
const binding = internalBinding('encoding_binding');
|
||||
|
||||
// Windows-1252 specific tests
|
||||
{
|
||||
// Test Windows-1252 special characters in 128-159 range
|
||||
// These differ from Latin-1
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x80), false, false), '€');
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x82), false, false), '‚');
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x83), false, false), 'ƒ');
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x9F), false, false), 'Ÿ');
|
||||
}
|
||||
|
||||
{
|
||||
// Test Windows-1252 characters outside 128-159 range (same as Latin-1)
|
||||
const buf = Uint8Array.from([0xC1, 0xE9, 0xF3]);
|
||||
assert.strictEqual(binding.decodeWindows1252(buf, false, false), 'Áéó');
|
||||
}
|
||||
|
||||
{
|
||||
// Empty input
|
||||
const buf = Uint8Array.from([]);
|
||||
assert.strictEqual(binding.decodeWindows1252(buf, false, false), '');
|
||||
}
|
||||
|
||||
// Windows-1252 specific tests
|
||||
{
|
||||
// Test Windows-1252 special characters in 128-159 range
|
||||
// These differ from Latin-1
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x80), false, false), '€');
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x82), false, false), '‚');
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x83), false, false), 'ƒ');
|
||||
assert.strictEqual(binding.decodeWindows1252(Uint8Array.of(0x9F), false, false), 'Ÿ');
|
||||
}
|
||||
|
||||
{
|
||||
// Test Windows-1252 characters outside 128-159 range (same as Latin-1)
|
||||
const buf = Uint8Array.from([0xC1, 0xE9, 0xF3]);
|
||||
assert.strictEqual(binding.decodeWindows1252(buf, false, false), 'Áéó');
|
||||
}
|
||||
|
||||
{
|
||||
// Empty input
|
||||
const buf = Uint8Array.from([]);
|
||||
assert.strictEqual(binding.decodeWindows1252(buf, false, false), '');
|
||||
}
|
||||
84
test/parallel/test-whatwg-encoding-singlebyte.mjs
Normal file
84
test/parallel/test-whatwg-encoding-singlebyte.mjs
Normal file
@@ -0,0 +1,84 @@
|
||||
// Test that single-byte encodings mappings are correct and match spec
|
||||
// Works without Intl
|
||||
|
||||
// From: https://github.com/ExodusOSS/bytes/blob/master/tests/single-byte.test.js
|
||||
// Copyright Exodus Movement. Licensed under MIT License.
|
||||
|
||||
import '../common/index.mjs';
|
||||
import { readFileSync } from 'node:fs';
|
||||
import { join } from 'node:path';
|
||||
import { test, describe } from 'node:test';
|
||||
import raw from '../fixtures/encoding/encodings.json' with { type: 'json' };
|
||||
|
||||
const groups = new Map(raw.map((x) => [x.heading, x.encodings.map((x) => x.name.toLowerCase())]));
|
||||
const encodings = groups.get('Legacy single-byte encodings');
|
||||
|
||||
describe('single-byte encodings are supersets of ascii', () => {
|
||||
for (const encoding of encodings) {
|
||||
test(encoding, (t) => {
|
||||
const loose = new TextDecoder(encoding);
|
||||
const fatal = new TextDecoder(encoding, { fatal: true });
|
||||
for (let i = 0; i < 128; i++) {
|
||||
const str = String.fromCodePoint(i);
|
||||
t.assert.strictEqual(loose.decode(Uint8Array.of(i)), str, i);
|
||||
t.assert.strictEqual(fatal.decode(Uint8Array.of(i)), str, i);
|
||||
}
|
||||
});
|
||||
}
|
||||
});
|
||||
|
||||
describe('single-byte encodings index', () => {
|
||||
for (const encoding of encodings) {
|
||||
test(encoding, (t) => {
|
||||
const loose = new TextDecoder(encoding);
|
||||
const fatal = new TextDecoder(encoding, { fatal: true });
|
||||
const file = encoding === 'iso-8859-8-i' ? `index-iso-8859-8.txt` : `index-${encoding}.txt`;
|
||||
const text = readFileSync(join(import.meta.dirname, '../fixtures/encoding/single-byte', file), 'utf8');
|
||||
const rows = text
|
||||
.split('\n')
|
||||
.map((x) => x.trim())
|
||||
.filter((x) => x && x[0] !== '#')
|
||||
.map((x) => x.split('\t'))
|
||||
.map(([istr, codeHex, description]) => {
|
||||
const i = Number(istr);
|
||||
t.assert.ok(i < 128);
|
||||
const code = parseInt(codeHex.slice(2), 16);
|
||||
t.assert.strictEqual(`${i}`, istr);
|
||||
t.assert.strictEqual('0x' + code.toString(16).padStart(4, '0').toUpperCase(), codeHex);
|
||||
t.assert.ok(code && code !== 0xff_fd && code <= 0xff_ff); // Can't be a replacement char, has to be <= 16-bit
|
||||
t.assert.ok(code < 0xd8_00 || code >= 0xe0_00); // not a surrogate
|
||||
return [i, { i, code, description }];
|
||||
});
|
||||
|
||||
t.assert.ok(rows.length <= 128);
|
||||
const known = new Map(rows);
|
||||
t.assert.strictEqual(rows.length, known.size); // all unique
|
||||
|
||||
for (let i = 0; i < 128; i++) {
|
||||
const row = known.get(i);
|
||||
const byte = i + 128;
|
||||
if (row) {
|
||||
t.assert.strictEqual(i, row.i);
|
||||
const str = String.fromCodePoint(row.code);
|
||||
t.assert.strictEqual(fatal.decode(Uint8Array.of(byte)), str, row.description);
|
||||
t.assert.strictEqual(loose.decode(Uint8Array.of(byte)), str, row.description);
|
||||
} else {
|
||||
t.assert.throws(() => fatal.decode(Uint8Array.of(byte)), TypeError);
|
||||
t.assert.strictEqual(loose.decode(Uint8Array.of(byte)), '\uFFFD');
|
||||
}
|
||||
}
|
||||
});
|
||||
}
|
||||
});
|
||||
|
||||
// https://encoding.spec.whatwg.org/#x-user-defined-decoder
|
||||
test('x-user-defined', (t) => {
|
||||
const encoding = 'x-user-defined';
|
||||
const loose = new TextDecoder(encoding);
|
||||
const fatal = new TextDecoder(encoding, { fatal: true });
|
||||
for (let byte = 0; byte < 256; byte++) {
|
||||
const str = String.fromCodePoint(byte >= 0x80 ? 0xF780 + byte - 0x80 : byte);
|
||||
t.assert.strictEqual(fatal.decode(Uint8Array.of(byte)), str, byte);
|
||||
t.assert.strictEqual(loose.decode(Uint8Array.of(byte)), str, byte);
|
||||
}
|
||||
});
|
||||
@@ -4,5 +4,4 @@ export interface EncodingBinding {
|
||||
decodeUTF8(buffer: ArrayBufferView | ArrayBuffer | SharedArrayBuffer, ignoreBOM?: boolean, hasFatal?: boolean): string;
|
||||
toASCII(input: string): string;
|
||||
toUnicode(input: string): string;
|
||||
decodeWindows1252(buffer: ArrayBufferView | ArrayBuffer | SharedArrayBuffer, ignoreBOM?: boolean, hasFatal?: boolean): string;
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user