hunspell-reader/dist/affToDicInfo.js

A library for reading Hunspell Dictionary Files

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.affToDicInfo = void 0;
const sync_1 = require("@cspell/cspell-pipe/sync");
const textUtils_1 = require("./textUtils");
function affToDicInfo(aff, locale) {
    const alphabetInfo = extractAlphabet(aff, locale);
    return {
        ...alphabetInfo,
        ...extractSuggestionEditCosts(aff, alphabetInfo),
        locale,
        alphabet: (0, textUtils_1.toRange)(alphabetInfo.alphabet, 5),
        accents: (0, textUtils_1.toRange)([...alphabetInfo.accents].sort().join('')),
    };
}
exports.affToDicInfo = affToDicInfo;
function extractAlphabet(aff, locale) {
    const sources = [
        aff.MAP,
        aff.TRY,
        aff.KEY,
        aff.REP?.flatMap((rep) => [rep.match, rep.replaceWith]),
        aff.ICONV?.flatMap((cov) => [cov.from, cov.to]),
        aff.OCONV?.flatMap((cov) => [cov.from, cov.to]),
        extractFxLetters(aff.PFX),
        extractFxLetters(aff.SFX),
    ];
    const setOfLetters = new Set(sources
        .filter(isDefined)
        .flatMap((a) => a)
        .map((a) => a.normalize())
        .flatMap((a) => [...a, ...a.toLocaleLowerCase(locale), ...a.toLocaleUpperCase(locale)])
        .map((a) => a.trim())
        .filter((a) => !!a));
    const alphabet = [...setOfLetters].sort().join('').replace(/\P{L}/gu, '');
    const accents = new Set(alphabet.normalize('NFD').replace(/\P{M}/gu, ''));
    return { locale, alphabet, accents };
}
function isDefined(a) {
    return a !== undefined;
}
function extractSuggestionEditCosts(aff, alphaInfo) {
    const suggestionEditCosts = [];
    suggestionEditCosts.push(...calcCapsAndAccentReplacements(alphaInfo), ...calcAffMapReplacements(aff), ...calcAffRepReplacements(aff));
    return {
        suggestionEditCosts,
    };
}
function calcAffMapReplacements(aff) {
    if (!aff.MAP)
        return [];
    const map = aff.MAP.sort().join('|');
    return [{ map, replace: 1, description: 'Hunspell Aff Map' }];
}
function calcAffRepReplacements(aff) {
    if (!aff.REP)
        return [];
    return createCostMaps(aff.REP.map((rep) => [rep.match, rep.replaceWith]), { map: '', replace: 75, description: 'Hunspell Replace Map' });
}
function calcCapsAndAccentReplacements(alphaInfo) {
    const { locale, alphabet } = alphaInfo;
    const letters = [...alphabet];
    const capForms = letters.map((letter) => calcCapitalizationForms(letter, locale));
    const accentForms = calcAccentForms(letters);
    const mapCrossAccent = calcCrossAccentCapsMap(accentForms, locale);
    return [
        ...createCostMaps(capForms, { map: '', replace: 1, description: 'Capitalization change.' }),
        ...createCostMaps(accentForms, { map: '', replace: 1, description: 'Replace Accents' }),
        ...createCostMaps(mapCrossAccent, { map: '', replace: 2, description: 'Capitalization and Accent change.' }),
    ];
}
function createCostMaps(formMaps, base) {
    const forms = formMaps.map((forms) => joinCharMap(forms));
    const mapValues = [...new Set(forms)].sort().filter((a) => !!a);
    return [...groupsOfN(mapValues, 6)].map((mapValues) => ({ ...base, map: mapValues.join('|') }));
}
function calcCapitalizationForms(letter, locale) {
    const forms = new Set();
    forms.add(letter);
    forms.add(letter.toUpperCase());
    forms.add(letter.toLowerCase());
    forms.add(letter.toLocaleUpperCase(locale));
    forms.add(letter.toLocaleLowerCase(locale));
    forms.add(letter.toLocaleUpperCase(locale).toLocaleLowerCase(locale));
    forms.add(letter.toLocaleLowerCase(locale).toLocaleUpperCase(locale));
    return forms;
}
function calcAccentForms(letters) {
    const forms = new Map();
    function getForm(letter) {
        const f = forms.get(letter);
        if (f)
            return f;
        const s = new Set();
        forms.set(letter, s);
        return s;
    }
    for (const letter of letters) {
        const base = (0, textUtils_1.removeAccents)(letter);
        const formCollection = getForm(base);
        formCollection.add(base);
        formCollection.add(letter);
        // addAccents(base, accents, formCollection);
    }
    return [...forms.values()].filter((s) => s.size > 1);
}
function joinCharMap(values) {
    return [...values]
        .sort()
        .map((a) => (a.length > 1 ? '(' + a + ')' : a))
        .join('');
}
function calcCrossAccentCapsMap(accentForms, locale) {
    function calc(form) {
        return new Set((0, sync_1.pipe)(form, (0, sync_1.opConcatMap)((letter) => calcCapitalizationForms(letter, locale))));
    }
    const values = (0, sync_1.pipe)(accentForms, (0, sync_1.opMap)(calc));
    return [...values];
}
// function addAccents(cleanLetter: string, accents: Iterable<string>, collection: Set<string>) {
//     for (const accent of accents) {
//         collection.add(applyAccent(cleanLetter, accent));
//     }
// }
// function applyAccent(letter: string, accent: string): string {
//     const withAccent = (letter + accent).normalize('NFC');
//     return removeLooseAccents(withAccent);
// }
function extractFxLetters(fxm) {
    if (!fxm)
        return undefined;
    const substations = (0, sync_1.pipe)(fxm.values(), (0, sync_1.opConcatMap)((f) => f.substitutionSets.values()), (0, sync_1.opConcatMap)((s) => s.substitutions));
    const partials = (0, sync_1.pipe)(substations, (0, sync_1.opConcatMap)((sub) => [sub.remove, sub.attach]));
    return [...partials];
}
function* groupsOfN(values, n) {
    let buffer = [];
    for (const item of values) {
        buffer.push(item);
        if (buffer.length >= n) {
            yield buffer;
            buffer = [];
        }
    }
    if (buffer.length) {
        yield buffer;
    }
}
//# sourceMappingURL=affToDicInfo.js.map