ethers.js/packages/wordlists/src.ts/lang-ja.ts

import { hexlify } from "@ethersproject/bytes";
import { id } from "@ethersproject/hash";
import { toUtf8Bytes, toUtf8String } from "@ethersproject/strings";

import { logger } from "./logger.js";
import { Wordlist } from "./wordlist.js";


const data = [

    // 4-kana words
    "AQRASRAGBAGUAIRAHBAghAURAdBAdcAnoAMEAFBAFCBKFBQRBSFBCXBCDBCHBGFBEQBpBBpQBIkBHNBeOBgFBVCBhBBhNBmOBmRBiHBiFBUFBZDBvFBsXBkFBlcBjYBwDBMBBTBBTRBWBBWXXaQXaRXQWXSRXCFXYBXpHXOQXHRXhRXuRXmXXbRXlXXwDXTRXrCXWQXWGaBWaKcaYgasFadQalmaMBacAKaRKKBKKXKKjKQRKDRKCYKCRKIDKeVKHcKlXKjHKrYNAHNBWNaRNKcNIBNIONmXNsXNdXNnBNMBNRBNrXNWDNWMNFOQABQAHQBrQXBQXFQaRQKXQKDQKOQKFQNBQNDQQgQCXQCDQGBQGDQGdQYXQpBQpQQpHQLXQHuQgBQhBQhCQuFQmXQiDQUFQZDQsFQdRQkHQbRQlOQlmQPDQjDQwXQMBQMDQcFQTBQTHQrDDXQDNFDGBDGQDGRDpFDhFDmXDZXDbRDMYDRdDTRDrXSAhSBCSBrSGQSEQSHBSVRShYShkSyQSuFSiBSdcSoESocSlmSMBSFBSFKSFNSFdSFcCByCaRCKcCSBCSRCCrCGbCEHCYXCpBCpQCIBCIHCeNCgBCgFCVECVcCmkCmwCZXCZFCdRClOClmClFCjDCjdCnXCwBCwXCcRCFQCFjGXhGNhGDEGDMGCDGCHGIFGgBGVXGVEGVRGmXGsXGdYGoSGbRGnXGwXGwDGWRGFNGFLGFOGFdGFkEABEBDEBFEXOEaBEKSENBENDEYXEIgEIkEgBEgQEgHEhFEudEuFEiBEiHEiFEZDEvBEsXEsFEdXEdREkFEbBEbRElFEPCEfkEFNYAEYAhYBNYQdYDXYSRYCEYYoYgQYgRYuRYmCYZTYdBYbEYlXYjQYRbYWRpKXpQopQnpSFpCXpIBpISphNpdBpdRpbRpcZpFBpFNpFDpFopFrLADLBuLXQLXcLaFLCXLEhLpBLpFLHXLeVLhILdHLdRLoDLbRLrXIABIBQIBCIBsIBoIBMIBRIXaIaRIKYIKRINBINuICDIGBIIDIIkIgRIxFIyQIiHIdRIbYIbRIlHIwRIMYIcRIRVITRIFBIFNIFQOABOAFOBQOaFONBONMOQFOSFOCDOGBOEQOpBOLXOIBOIFOgQOgFOyQOycOmXOsXOdIOkHOMEOMkOWWHBNHXNHXWHNXHDuHDRHSuHSRHHoHhkHmRHdRHkQHlcHlRHwBHWcgAEgAggAkgBNgBQgBEgXOgYcgLXgHjgyQgiBgsFgdagMYgWSgFQgFEVBTVXEVKBVKNVKDVKYVKRVNBVNYVDBVDxVSBVSRVCjVGNVLXVIFVhBVhcVsXVdRVbRVlRhBYhKYhDYhGShxWhmNhdahdkhbRhjohMXhTRxAXxXSxKBxNBxEQxeNxeQxhXxsFxdbxlHxjcxFBxFNxFQxFOxFoyNYyYoybcyMYuBQuBRuBruDMuCouHBudQukkuoBulVuMXuFEmCYmCRmpRmeDmiMmjdmTFmFQiADiBOiaRiKRiNBiNRiSFiGkiGFiERipRiLFiIFihYibHijBijEiMXiWBiFBiFCUBQUXFUaRUNDUNcUNRUNFUDBUSHUCDUGBUGFUEqULNULoUIRUeEUeYUgBUhFUuRUiFUsXUdFUkHUbBUjSUjYUwXUMDUcHURdUTBUrBUrXUrQZAFZXZZaRZKFZNBZQFZCXZGBZYdZpBZLDZIFZHXZHNZeQZVRZVFZmXZiBZvFZdFZkFZbHZbFZwXZcCZcRZRBvBQvBGvBLvBWvCovMYsAFsBDsaRsKFsNFsDrsSHsSFsCXsCRsEBsEHsEfspBsLBsLDsIgsIRseGsbRsFBsFQsFSdNBdSRdCVdGHdYDdHcdVbdySduDdsXdlRdwXdWYdWcdWRkBMkXOkaRkNIkNFkSFkCFkYBkpRkeNkgBkhVkmXksFklVkMBkWDkFNoBNoaQoaFoNBoNXoNaoNEoSRoEroYXoYCoYbopRopFomXojkowXorFbBEbEIbdBbjYlaRlDElMXlFDjKjjSRjGBjYBjYkjpRjLXjIBjOFjeVjbRjwBnXQnSHnpFnLXnINnMBnTRwXBwXNwXYwNFwQFwSBwGFwLXwLDweNwgBwuHwjDwnXMBXMpFMIBMeNMTHcaQcNBcDHcSFcCXcpBcLXcLDcgFcuFcnXcwXccDcTQcrFTQErXNrCHrpFrgFrbFrTHrFcWNYWNbWEHWMXWTR",

    // 5-kana words
    "ABGHABIJAEAVAYJQALZJAIaRAHNXAHdcAHbRAZJMAZJRAZTRAdVJAklmAbcNAjdRAMnRAMWYAWpRAWgRAFgBAFhBAFdcBNJBBNJDBQKBBQhcBQlmBDEJBYJkBYJTBpNBBpJFBIJBBIJDBIcABOKXBOEJBOVJBOiJBOZJBepBBeLXBeIFBegBBgGJBVJXBuocBiJRBUJQBlXVBlITBwNFBMYVBcqXBTlmBWNFBWiJBWnRBFGHBFwXXKGJXNJBXNZJXDTTXSHSXSVRXSlHXCJDXGQJXEhXXYQJXYbRXOfXXeNcXVJFXhQJXhEJXdTRXjdXXMhBXcQTXRGBXTEBXTnQXFCXXFOFXFgFaBaFaBNJaBCJaBpBaBwXaNJKaNJDaQIBaDpRaEPDaHMFamDJalEJaMZJaFaFaFNBaFQJaFLDaFVHKBCYKBEBKBHDKXaFKXGdKXEJKXpHKXIBKXZDKXwXKKwLKNacKNYJKNJoKNWcKDGdKDTRKChXKGaRKGhBKGbRKEBTKEaRKEPTKLMDKLWRKOHDKVJcKdBcKlIBKlOPKFSBKFEPKFpFNBNJNJBQNBGHNBEPNBHXNBgFNBVXNBZDNBsXNBwXNNaRNNJDNNJENNJkNDCJNDVDNGJRNJiDNZJNNsCJNJFNNFSBNFCXNFEPNFLXNFIFQJBFQCaRQJEQQLJDQLJFQIaRQOqXQHaFQHHQQVJXQVJDQhNJQmEIQZJFQsJXQJrFQWbRDJABDBYJDXNFDXCXDXLXDXZDDXsJDQqXDSJFDJCXDEPkDEqXDYmQDpSJDOCkDOGQDHEIDVJDDuDuDWEBDJFgSBNDSBSFSBGHSBIBSBTQSKVYSJQNSJQiSJCXSEqXSJYVSIiJSOMYSHAHSHaQSeCFSepQSegBSHdHSHrFShSJSJuHSJUFSkNRSrSrSWEBSFaHSJFQSFCXSFGDSFYXSFODSFgBSFVXSFhBSFxFSFkFSFbBSFMFCADdCJXBCXaFCXKFCXNFCXCXCXGBCXEJCXYBCXLDCXIBCXOPCXHXCXgBCXhBCXiBCXlDCXcHCJNBCJNFCDCJCDGBCDVXCDhBCDiDCDJdCCmNCpJFCIaRCOqXCHCHCHZJCViJCuCuCmddCJiFCdNBCdHhClEJCnUJCreSCWlgCWTRCFBFCFNBCFYBCFVFCFhFCFdSCFTBCFWDGBNBGBQFGJBCGBEqGBpBGBgQGNBEGNJYGNkOGNJRGDUFGJpQGHaBGJeNGJeEGVBlGVKjGiJDGvJHGsVJGkEBGMIJGWjNGFBFGFCXGFGBGFYXGFpBGFMFEASJEAWpEJNFECJVEIXSEIQJEOqXEOcFEeNcEHEJEHlFEJgFEhlmEmDJEmZJEiMBEUqXEoSREPBFEPXFEPKFEPSFEPEFEPpFEPLXEPIBEJPdEPcFEPTBEJnXEqlHEMpREFCXEFODEFcFYASJYJAFYBaBYBVXYXpFYDhBYCJBYJGFYYbRYeNcYJeVYiIJYZJcYvJgYvJRYJsXYsJFYMYMYreVpBNHpBEJpBwXpQxFpYEJpeNDpJeDpeSFpeCHpHUJpHbBpHcHpmUJpiiJpUJrpsJuplITpFaBpFQqpFGBpFEfpFYBpFpBpFLJpFIDpFgBpFVXpFyQpFuFpFlFpFjDpFnXpFwXpJFMpFTBLXCJLXEFLXhFLXUJLXbFLalmLNJBLSJQLCLCLGJBLLDJLHaFLeNFLeSHLeCXLepFLhaRLZsJLsJDLsJrLocaLlLlLMdbLFNBLFSBLFEHLFkFIBBFIBXFIBaQIBKXIBSFIBpHIBLXIBgBIBhBIBuHIBmXIBiFIBZXIBvFIBbFIBjQIBwXIBWFIKTRIQUJIDGFICjQIYSRIINXIJeCIVaRImEkIZJFIvJRIsJXIdCJIJoRIbBQIjYBIcqXITFVIreVIFKFIFSFIFCJIFGFIFLDIFIBIJFOIFgBIFVXIJFhIFxFIFmXIFdHIFbBIJFrIJFWOBGBOQfXOOKjOUqXOfXBOqXEOcqXORVJOFIBOFlDHBIOHXiFHNTRHCJXHIaRHHJDHHEJHVbRHZJYHbIBHRsJHRkDHWlmgBKFgBSBgBCDgBGHgBpBgBIBgBVJgBuBgBvFgKDTgQVXgDUJgGSJgOqXgmUMgZIJgTUJgWIEgFBFgFNBgFDJgFSFgFGBgFYXgJFOgFgQgFVXgFhBgFbHgJFWVJABVQKcVDgFVOfXVeDFVhaRVmGdViJYVMaRVFNHhBNDhBCXhBEqhBpFhBLXhNJBhSJRheVXhhKEhxlmhZIJhdBQhkIJhbMNhMUJhMZJxNJgxQUJxDEkxDdFxSJRxplmxeSBxeCXxeGFxeYXxepQxegBxWVcxFEQxFLXxFIBxFgBxFxDxFZtxFdcxFbBxFwXyDJXyDlcuASJuDJpuDIBuCpJuGSJuIJFueEFuZIJusJXudWEuoIBuWGJuFBcuFKEuFNFuFQFuFDJuFGJuFVJuFUtuFdHuFTBmBYJmNJYmQhkmLJDmLJomIdXmiJYmvJRmsJRmklmmMBymMuCmclmmcnQiJABiJBNiJBDiBSFiBCJiBEFiBYBiBpFiBLXiBTHiJNciDEfiCZJiECJiJEqiOkHiHKFieNDiHJQieQcieDHieSFieCXieGFieEFieIHiegFihUJixNoioNXiFaBiFKFiFNDiFEPiFYXitFOitFHiFgBiFVEiFmXiFitiFbBiFMFiFrFUCXQUIoQUIJcUHQJUeCEUHwXUUJDUUqXUdWcUcqXUrnQUFNDUFSHUFCFUFEfUFLXUtFOZBXOZXSBZXpFZXVXZEQJZEJkZpDJZOqXZeNHZeCDZUqXZFBQZFEHZFLXvBAFvBKFvBCXvBEPvBpHvBIDvBgFvBuHvQNJvFNFvFGBvFIBvJFcsXCDsXLXsXsXsXlFsXcHsQqXsJQFsEqXseIFsFEHsFjDdBxOdNpRdNJRdEJbdpJRdhZJdnSJdrjNdFNJdFQHdFhNkNJDkYaRkHNRkHSRkVbRkuMRkjSJkcqDoSJFoEiJoYZJoOfXohEBoMGQocqXbBAFbBXFbBaFbBNDbBGBbBLXbBTBbBWDbGJYbIJHbFQqbFpQlDgQlOrFlVJRjGEBjZJRnXvJnXbBnEfHnOPDngJRnxfXnUJWwXEJwNpJwDpBwEfXwrEBMDCJMDGHMDIJMLJDcQGDcQpHcqXccqNFcqCXcFCJRBSBRBGBRBEJRBpQTBNFTBQJTBpBTBVXTFABTFSBTFCFTFGBTFMDrXCJrXLDrDNJrEfHrFQJrFitWNjdWNTR",

    // 6-kana words
    "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",

    // 7-kana words
    "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",

    // 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 _wordlist: null | Array<string> = null;

function hex(word: string) {
    return hexlify(toUtf8Bytes(word));
}

const KiYoKu = "0xe3818de38284e3818f";
const KyoKu = "0xe3818de38283e3818f"

function toString(data: Array<number>): string {
    return toUtf8String(new Uint8Array(data));
}

function loadWords(): Array<string> {
    if (_wordlist !== null) { return _wordlist; }

    const wordlist = [];

    // Transforms for normalizing (sort is a not quite UTF-8)
    const transform: { [key: string]: string | boolean } = {};

    // Delete the diacritic marks
    transform[toString([227, 130, 154])] = false;
    transform[toString([227, 130, 153])] = false;

    // Some simple transforms that sort out most of the order
    transform[toString([227, 130, 133])] = toString([227, 130, 134]);
    transform[toString([227, 129, 163])] = toString([227, 129, 164]);
    transform[toString([227, 130, 131])] = toString([227, 130, 132]);
    transform[toString([227, 130, 135])] = toString([227, 130, 136]);


    // Normalize words using the transform
    function normalize(word: string) {
        let result = "";
        for (let i = 0; i < word.length; i++) {
            let kana = word[i];
            const target = transform[kana];
            if (target === false) { continue; }
            if (target) { kana = <string>target; }
            result += kana;
        }
        return result;
    }

    // Sort how the Japanese list is sorted
    function sortJapanese(a: string, b: string) {
        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++) {
        const d = data[length - 3];
        for (let offset = 0; offset < d.length; offset += length) {
            const word = [];
            for (let i = 0; i < length; i++) {
                 const k = mapping.indexOf(d[offset + i]);
                 word.push(227);
                 word.push((k & 0x40) ? 130: 129);
                 word.push((k & 0x3f) + 128);
            }
            wordlist.push(toString(word));
        }
    }
    wordlist.sort(sortJapanese);

    // For some reason kyoku and kiyoku are flipped in node (!!).
    // The order SHOULD be:
    //   - kyoku
    //   - kiyoku

    // This should ignore "if", but that doesn't work here??
    /* c8 ignore start */
    if (hex(wordlist[442]) === KiYoKu && hex(wordlist[443]) === KyoKu) {
        const tmp = wordlist[442];
        wordlist[442] = wordlist[443];
        wordlist[443] = tmp;
    }
    /* c8 ignore stop */

    // Verify the computed list matches the official list
    /* istanbul ignore if */
    const checksum = id(wordlist.join("\n") + "\n");
    /* c8 ignore start */
    if (checksum !== "0xcb36b09e6baa935787fd762ce65e80b0c6a8dabdfbc3a7f86ac0e2c4fd111600") {
        throw new Error("BIP39 Wordlist for ja (Japanese) FAILED");
    }
    /* c8 ignore stop */

    _wordlist = wordlist;

    return wordlist;
}

class LangJa extends Wordlist {
    constructor() { super("ja"); }

    getWord(index: number): string {
        const words = loadWords();
        if (index < 0 || index >= words.length) {
            logger.throwArgumentError(`invalid word index: ${ index }`, "index", index);
        }
        return words[index];
    }

    getWordIndex(word: string): number {
        return loadWords().indexOf(word);
    }

    split(mnemonic: string): Array<string> {
        //logger.assertNormalize();
        return mnemonic.split(/(?:\u3000| )+/g);
    }

    join(words: Array<string>): string {
        return words.join("\u3000");
    }
}

export const langJa = new LangJa();