139 lines
11 KiB
TypeScript
139 lines
11 KiB
TypeScript
|
|
import { register, Wordlist } from './wordlist';
|
|
|
|
import { toUtf8String } from '../utils/utf8';
|
|
|
|
import * as errors from '../utils/errors';
|
|
|
|
const data = [
|
|
|
|
// 4-kana words
|
|
'AQRASRAGBAGUAIRAHBAghAURAdBAdcAnoAMEAFBAFCBKFBQRBSFBCXBCDBCHBGFBEQBpBBpQBIkBHNBeOBgFBVCBhBBhNBmOBmRBiHBiFBUFBZDBvFBsXBkFBlcBjYBwDBMBBTBBTRBWBBWXXaQXaRXQWXSRXCFXYBXpHXOQXHRXhRXuRXmXXbRXlXXwDXTRXrCXWQXWGaBWaKcaYgasFadQalmaMBacAKaRKKBKKXKKjKQRKDRKCYKCRKIDKeVKHcKlXKjHKrYNAHNBWNaRNKcNIBNIONmXNsXNdXNnBNMBNRBNrXNWDNWMNFOQABQAHQBrQXBQXFQaRQKXQKDQKOQKFQNBQNDQQgQCXQCDQGBQGDQGdQYXQpBQpQQpHQLXQHuQgBQhBQhCQuFQmXQiDQUFQZDQsFQdRQkHQbRQlOQlmQPDQjDQwXQMBQMDQcFQTBQTHQrDDXQDNFDGBDGQDGRDpFDhFDmXDZXDbRDMYDRdDTRDrXSAhSBCSBrSGQSEQSHBSVRShYShkSyQSuFSiBSdcSoESocSlmSMBSFBSFKSFNSFdSFcCByCaRCKcCSBCSRCCrCGbCEHCYXCpBCpQCIBCIHCeNCgBCgFCVECVcCmkCmwCZXCZFCdRClOClmClFCjDCjdCnXCwBCwXCcRCFQCFjGXhGNhGDEGDMGCDGCHGIFGgBGVXGVEGVRGmXGsXGdYGoSGbRGnXGwXGwDGWRGFNGFLGFOGFdGFkEABEBDEBFEXOEaBEKSENBENDEYXEIgEIkEgBEgQEgHEhFEudEuFEiBEiHEiFEZDEvBEsXEsFEdXEdREkFEbBEbRElFEPCEfkEFNYAEYAhYBNYQdYDXYSRYCEYYoYgQYgRYuRYmCYZTYdBYbEYlXYjQYRbYWRpKXpQopQnpSFpCXpIBpISphNpdBpdRpbRpcZpFBpFNpFDpFopFrLADLBuLXQLXcLaFLCXLEhLpBLpFLHXLeVLhILdHLdRLoDLbRLrXIABIBQIBCIBsIBoIBMIBRIXaIaRIKYIKRINBINuICDIGBIIDIIkIgRIxFIyQIiHIdRIbYIbRIlHIwRIMYIcRIRVITRIFBIFNIFQOABOAFOBQOaFONBONMOQFOSFOCDOGBOEQOpBOLXOIBOIFOgQOgFOyQOycOmXOsXOdIOkHOMEOMkOWWHBNHXNHXWHNXHDuHDRHSuHSRHHoHhkHmRHdRHkQHlcHlRHwBHWcgAEgAggAkgBNgBQgBEgXOgYcgLXgHjgyQgiBgsFgdagMYgWSgFQgFEVBTVXEVKBVKNVKDVKYVKRVNBVNYVDBVDxVSBVSRVCjVGNVLXVIFVhBVhcVsXVdRVbRVlRhBYhKYhDYhGShxWhmNhdahdkhbRhjohMXhTRxAXxXSxKBxNBxEQxeNxeQxhXxsFxdbxlHxjcxFBxFNxFQxFOxFoyNYyYoybcyMYuBQuBRuBruDMuCouHBudQukkuoBulVuMXuFEmCYmCRmpRmeDmiMmjdmTFmFQiADiBOiaRiKRiNBiNRiSFiGkiGFiERipRiLFiIFihYibHijBijEiMXiWBiFBiFCUBQUXFUaRUNDUNcUNRUNFUDBUSHUCDUGBUGFUEqULNULoUIRUeEUeYUgBUhFUuRUiFUsXUdFUkHUbBUjSUjYUwXUMDUcHURdUTBUrBUrXUrQZAFZXZZaRZKFZNBZQFZCXZGBZYdZpBZLDZIFZHXZHNZeQZVRZVFZmXZiBZvFZdFZkFZbHZbFZwXZcCZcRZRBvBQvBGvBLvBWvCovMYsAFsBDsaRsKFsNFsDrsSHsSFsCXsCRsEBsEHsEfspBsLBsLDsIgsIRseGsbRsFBsFQsFSdNBdSRdCVdGHdYDdHcdVbdySduDdsXdlRdwXdWYdWcdWRkBMkXOkaRkNIkNFkSFkCFkYBkpRkeNkgBkhVkmXksFklVkMBkWDkFNoBNoaQoaFoNBoNXoNaoNEoSRoEroYXoYCoYbopRopFomXojkowXorFbBEbEIbdBbjYlaRlDElMXlFDjKjjSRjGBjYBjYkjpRjLXjIBjOFjeVjbRjwBnXQnSHnpFnLXnINnMBnTRwXBwXNwXYwNFwQFwSBwGFwLXwLDweNwgBwuHwjDwnXMBXMpFMIBMeNMTHcaQcNBcDHcSFcCXcpBcLXcLDcgFcuFcnXcwXccDcTQcrFTQErXNrCHrpFrgFrbFrTHrFcWNYWNbWEHWMXWTR',
|
|
|
|
// 5-kana words
|
|
'ABGHABIJAEAVAYJQALZJAIaRAHNXAHdcAHbRAZJMAZJRAZTRAdVJAklmAbcNAjdRAMnRAMWYAWpRAWgRAFgBAFhBAFdcBNJBBNJDBQKBBQhcBQlmBDEJBYJkBYJTBpNBBpJFBIJBBIJDBIcABOKXBOEJBOVJBOiJBOZJBepBBeLXBeIFBegBBgGJBVJXBuocBiJRBUJQBlXVBlITBwNFBMYVBcqXBTlmBWNFBWiJBWnRBFGHBFwXXKGJXNJBXNZJXDTTXSHSXSVRXSlHXCJDXGQJXEhXXYQJXYbRXOfXXeNcXVJFXhQJXhEJXdTRXjdXXMhBXcQTXRGBXTEBXTnQXFCXXFOFXFgFaBaFaBNJaBCJaBpBaBwXaNJKaNJDaQIBaDpRaEPDaHMFamDJalEJaMZJaFaFaFNBaFQJaFLDaFVHKBCYKBEBKBHDKXaFKXGdKXEJKXpHKXIBKXZDKXwXKKwLKNacKNYJKNJoKNWcKDGdKDTRKChXKGaRKGhBKGbRKEBTKEaRKEPTKLMDKLWRKOHDKVJcKdBcKlIBKlOPKFSBKFEPKFpFNBNJNJBQNBGHNBEPNBHXNBgFNBVXNBZDNBsXNBwXNNaRNNJDNNJENNJkNDCJNDVDNGJRNJiDNZJNNsCJNJFNNFSBNFCXNFEPNFLXNFIFQJBFQCaRQJEQQLJDQLJFQIaRQOqXQHaFQHHQQVJXQVJDQhNJQmEIQZJFQsJXQJrFQWbRDJABDBYJDXNFDXCXDXLXDXZDDXsJDQqXDSJFDJCXDEPkDEqXDYmQDpSJDOCkDOGQDHEIDVJDDuDuDWEBDJFgSBNDSBSFSBGHSBIBSBTQSKVYSJQNSJQiSJCXSEqXSJYVSIiJSOMYSHAHSHaQSeCFSepQSegBSHdHSHrFShSJSJuHSJUFSkNRSrSrSWEBSFaHSJFQSFCXSFGDSFYXSFODSFgBSFVXSFhBSFxFSFkFSFbBSFMFCADdCJXBCXaFCXKFCXNFCXCXCXGBCXEJCXYBCXLDCXIBCXOPCXHXCXgBCXhBCXiBCXlDCXcHCJNBCJNFCDCJCDGBCDVXCDhBCDiDCDJdCCmNCpJFCIaRCOqXCHCHCHZJCViJCuCuCmddCJiFCdNBCdHhClEJCnUJCreSCWlgCWTRCFBFCFNBCFYBCFVFCFhFCFdSCFTBCFWDGBNBGBQFGJBCGBEqGBpBGBgQGNBEGNJYGNkOGNJRGDUFGJpQGHaBGJeNGJeEGVBlGVKjGiJDGvJHGsVJGkEBGMIJGWjNGFBFGFCXGFGBGFYXGFpBGFMFEASJEAWpEJNFECJVEIXSEIQJEOqXEOcFEeNcEHEJEHlFEJgFEhlmEmDJEmZJEiMBEUqXEoSREPBFEPXFEPKFEPSFEPEFEPpFEPLXEPIBEJPdEPcFEPTBEJnXEqlHEMpREFCXEFODEFcFYASJYJAFYBaBYBVXYXpFYDhBYCJBYJGFYYbRYeNcYJeVYiIJYZJcYvJgYvJRYJsXYsJFYMYMYreVpBNHpBEJpBwXpQxFpYEJpeNDpJeDpeSFpeCHpHUJpHbBpHcHpmUJpiiJpUJrpsJuplITpFaBpFQqpFGBpFEfpFYBpFpBpFLJpFIDpFgBpFVXpFyQpFuFpFlFpFjDpFnXpFwXpJFMpFTBLXCJLXEFLXhFLXUJLXbFLalmLNJBLSJQLCLCLGJBLLDJLHaFLeNFLeSHLeCXLepFLhaRLZsJLsJDLsJrLocaLlLlLMdbLFNBLFSBLFEHLFkFIBBFIBXFIBaQIBKXIBSFIBpHIBLXIBgBIBhBIBuHIBmXIBiFIBZXIBvFIBbFIBjQIBwXIBWFIKTRIQUJIDGFICjQIYSRIINXIJeCIVaRImEkIZJFIvJRIsJXIdCJIJoRIbBQIjYBIcqXITFVIreVIFKFIFSFIFCJIFGFIFLDIFIBIJFOIFgBIFVXIJFhIFxFIFmXIFdHIFbBIJFrIJFWOBGBOQfXOOKjOUqXOfXBOqXEOcqXORVJOFIBOFlDHBIOHXiFHNTRHCJXHIaRHHJDHHEJHVbRHZJYHbIBHRsJHRkDHWlmgBKFgBSBgBCDgBGHgBpBgBIBgBVJgBuBgBvFgKDTgQVXgDUJgGSJgOqXgmUMgZIJgTUJgWIEgFBFgFNBgFDJgFSFgFGBgFYXgJFOgFgQgFVXgFhBgFbHgJFWVJABVQKcVDgFVOfXVeDFVhaRVmGdViJYVMaRVFNHhBNDhBCXhBEqhBpFhBLXhNJBhSJRheVXhhKEhxlmhZIJhdBQhkIJhbMNhMUJhMZJxNJgxQUJxDEkxDdFxSJRxplmxeSBxeCXxeGFxeYXxepQxegBxWVcxFEQxFLXxFIBxFgBxFxDxFZtxFdcxFbBxFwXyDJXyDlcuASJuDJpuDIBuCpJuGSJuIJFueEFuZIJusJXudWEuoIBuWGJuFBcuFKEuFNFuFQFuFDJuFGJuFVJuFUtuFdHuFTBmBYJmNJYmQhkmLJDmLJomIdXmiJYmvJRmsJRmklmmMBymMuCmclmmcnQiJABiJBNiJBDiBSFiBCJiBEFiBYBiBpFiBLXiBTHiJNciDEfiCZJiECJiJEqiOkHiHKFieNDiHJQieQcieDHieSFieCXieGFieEFieIHiegFihUJixNoioNXiFaBiFKFiFNDiFEPiFYXitFOitFHiFgBiFVEiFmXiFitiFbBiFMFiFrFUCXQUIoQUIJcUHQJUeCEUHwXUUJDUUqXUdWcUcqXUrnQUFNDUFSHUFCFUFEfUFLXUtFOZBXOZXSBZXpFZXVXZEQJZEJkZpDJZOqXZeNHZeCDZUqXZFBQZFEHZFLXvBAFvBKFvBCXvBEPvBpHvBIDvBgFvBuHvQNJvFNFvFGBvFIBvJFcsXCDsXLXsXsXsXlFsXcHsQqXsJQFsEqXseIFsFEHsFjDdBxOdNpRdNJRdEJbdpJRdhZJdnSJdrjNdFNJdFQHdFhNkNJDkYaRkHNRkHSRkVbRkuMRkjSJkcqDoSJFoEiJoYZJoOfXohEBoMGQocqXbBAFbBXFbBaFbBNDbBGBbBLXbBTBbBWDbGJYbIJHbFQqbFpQlDgQlOrFlVJRjGEBjZJRnXvJnXbBnEfHnOPDngJRnxfXnUJWwXEJwNpJwDpBwEfXwrEBMDCJMDGHMDIJMLJDcQGDcQpHcqXccqNFcqCXcFCJRBSBRBGBRBEJRBpQTBNFTBQJTBpBTBVXTFABTFSBTFCFTFGBTFMDrXCJrXLDrDNJrEfHrFQJrFitWNjdWNTR',
|
|
|
|
// 6-kana words
|
|
'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',
|
|
|
|
// 7-kana words
|
|
'ABCDEFACNJTRAMBDJdAcNJVXBLNJEBXSIdWRXErNJkXYDJMBXZJCJaXMNJaYKKVJKcKDEJqXKDcNJhKVJrNYKbgJVXKFVJSBNBYBwDNJeQfXNJeEqXNhGJWENJFiJRQlIJbEQJfXxDQqXcfXQFNDEJQFwXUJDYcnUJDJIBgQDIUJTRDJFEqDSJQSJFSJQIJFSOPeZtSJFZJHCJXQfXCTDEqFGJBSJFGJBOfXGJBcqXGJHNJDGJRLiJEJfXEqEJFEJPEFpBEJYJBZJFYBwXUJYiJMEBYJZJyTYTONJXpQMFXFpeGIDdpJFstXpJFcPDLBVSJRLHQJqXLJFZJFIJBNJDIJBUqXIBkFDJIJEJPTIYJGWRIJeQPDIJeEfHIJFsJXOqGDSFHXEJqXgJCsJCgGQJqXgdQYJEgFMFNBgJFcqDVJwXUJVJFZJchIgJCCxOEJqXxOwXUJyDJBVRuscisciJBiJBieUtqXiJFDJkiFsJXQUGEZJcUJFsJXZtXIrXZDZJDrZJFNJDZJFstXvJFQqXvJFCJEsJXQJqkhkNGBbDJdTRbYJMEBlDwXUJMEFiJFcfXNJDRcNJWMTBLJXC',
|
|
|
|
// 8-kana words
|
|
'BraFUtHBFSJFdbNBLJXVJQoYJNEBSJBEJfHSJHwXUJCJdAZJMGjaFVJXEJPNJBlEJfFiJFpFbFEJqIJBVJCrIBdHiJhOPFChvJVJZJNJWxGFNIFLueIBQJqUHEJfUFstOZJDrlXEASJRlXVJXSFwVJNJWD',
|
|
|
|
// 9-kana words
|
|
'QJEJNNJDQJEJIBSFQJEJxegBQJEJfHEPSJBmXEJFSJCDEJqXLXNJFQqXIcQsFNJFIFEJqXUJgFsJXIJBUJEJfHNFvJxEqXNJnXUJFQqD',
|
|
|
|
// 10-kana words
|
|
'IJBEJqXZJ'
|
|
];
|
|
|
|
// Maps each character into its kana value (the index)
|
|
const mapping = "~~AzB~X~a~KN~Q~D~S~C~G~E~Y~p~L~I~O~eH~g~V~hxyumi~~U~~Z~~v~~s~~dkoblPjfnqwMcRTr~W~~~F~~~~~Jt"
|
|
|
|
let words: Array<string> = null;
|
|
|
|
function loadWords() {
|
|
if (words !== null) { return; }
|
|
words = [];
|
|
|
|
// Transforms for normalizing (sort is a not quite UTF-8)
|
|
var transform = {};
|
|
|
|
// Delete the diacritic marks
|
|
transform[toUtf8String([227, 130, 154])] = false;
|
|
transform[toUtf8String([227, 130, 153])] = false;
|
|
|
|
// Some simple transforms that sort out most of the order
|
|
transform[toUtf8String([227, 130, 133])] = toUtf8String([227, 130, 134]);
|
|
transform[toUtf8String([227, 129, 163])] = toUtf8String([227, 129, 164]);
|
|
transform[toUtf8String([227, 130, 131])] = toUtf8String([227, 130, 132]);
|
|
transform[toUtf8String([227, 130, 135])] = toUtf8String([227, 130, 136]);
|
|
|
|
|
|
// Normalize words using the transform
|
|
function normalize(word) {
|
|
var result = '';
|
|
for (var i = 0; i < word.length; i++) {
|
|
let kana = word[i];
|
|
var target = transform[kana];
|
|
if (target === false) { continue; }
|
|
if (target) { kana = target; }
|
|
result += kana;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
// Sort how the Japanese list is sorted
|
|
function sortJapanese(a, b) {
|
|
a = normalize(a);
|
|
b = normalize(b);
|
|
if (a < b) { return -1; }
|
|
if (a > b) { return 1; }
|
|
return 0;
|
|
}
|
|
|
|
// Load all the words
|
|
for (let length = 3; length <= 9; length++) {
|
|
let d = data[length - 3];
|
|
for (let offset = 0; offset < d.length; offset += length) {
|
|
let word = [];
|
|
for (let i = 0; i < length; i++) {
|
|
let k = mapping.indexOf(d[offset + i]);
|
|
word.push(227);
|
|
word.push((k & 0x40) ? 130: 129);
|
|
word.push((k & 0x3f) + 128);
|
|
}
|
|
words.push(toUtf8String(word));
|
|
}
|
|
}
|
|
words.sort(sortJapanese);
|
|
|
|
// For some reason kyoku and kiyoku are flipped; we'll just manually fix it
|
|
let kyoku = words[442];
|
|
words[442] = words[443];
|
|
words[443] = kyoku;
|
|
}
|
|
|
|
/*
|
|
var fs = require('fs');
|
|
fs.readFileSync('lang-ja.txt').toString().split('\x0a').forEach(function(d, i) {
|
|
if (d !== words[i]) { console.log(d, words[i], i, toUtf8Bytes(d)); }
|
|
});
|
|
*/
|
|
|
|
|
|
class LangJa extends Wordlist {
|
|
constructor() {
|
|
super('ja');
|
|
}
|
|
|
|
getWord(index) {
|
|
loadWords();
|
|
return words[index];
|
|
}
|
|
|
|
getWordIndex(word) {
|
|
loadWords();
|
|
return words.indexOf(word);
|
|
}
|
|
|
|
split(mnemonic: string): Array<string> {
|
|
if (!mnemonic.normalize) {
|
|
errors.throwError('Japanese is unsupported on this platform; missing String.prototype.normalize', errors.UNSUPPORTED_OPERATION, { operation: 'String.prototype.normalize' });
|
|
}
|
|
return mnemonic.split(/(?:\u3000| )+/g);
|
|
}
|
|
|
|
join(words: Array<string>): string {
|
|
return words.join('\u3000');
|
|
}
|
|
}
|
|
|
|
const langJa = new LangJa();
|
|
register(langJa);
|
|
|
|
export { langJa };
|
|
|