PronounsPage/test/locales/data.test.ts

import type { SyncExpectationResult } from '@vitest/expect';
import { describe, expect, test } from 'vitest';

import type { Config } from '~/locale/config.ts';
import type { NounTemplatesData, PronounGroupData, PronounExamplesData, PronounData } from '~/locale/data.ts';
import allLocales from '~/locale/locales.ts';
import { loadSuml, loadTsv } from '~/server/loader.ts';
import { normaliseKey } from '~/src/buildPronoun.ts';
import { Example } from '~/src/language/examples.ts';
import type { VariantsFromBaseConverter } from '~/src/language/grammarTables.ts';
import { availableGenders, gendersWithNumerus } from '~/src/nouns.ts';
import type { NounsData } from '~/src/nouns.ts';

function toHaveValidMorphemes(actual: string, morphemes: string[]): SyncExpectationResult {
    const containedMorphemes = Example.parse(actual).parts
        .filter((part) => typeof part !== 'string' && part.type === 'morpheme')
        .map((part) => part.morpheme);
    const unknownMorphemes = containedMorphemes.filter((morpheme) => !morphemes.includes(morpheme));
    if (unknownMorphemes.length > 0) {
        return {
            message: () => `expected example '${actual}' to have valid morphemes,` +
                ` but these are unknown:\n${unknownMorphemes.join(', ')}`,
            pass: false,
        };
    } else {
        return {
            message: () => 'expected example to have invalid morphemes',
            pass: true,
        };
    }
}

interface CustomMatchers<R> {
    toHaveValidMorphemes(morphemes: string[]): R;
}

declare module 'vitest' {
    interface Assertion<T> extends CustomMatchers<T> {}
}

expect.extend({ toHaveValidMorphemes });

describe.each(allLocales)('data files of $code', async ({ code }) => {
    const config = await loadSuml<Config>(`locale/${code}/config.suml`);

    const pronouns = await loadTsv<PronounData<string>>(`locale/${code}/pronouns/pronouns.tsv`);
    const pronounGroups = await loadTsv<PronounGroupData>(`locale/${code}/pronouns/pronounGroups.tsv`);
    const examples = await loadTsv<PronounExamplesData>(`locale/${code}/pronouns/examples.tsv`);

    test('pronouns/pronouns.tsv match schema', async () => {
        if (pronouns.length === 0) {
            return;
        }
        const required = [
            'key',
            'description',
            'normative',
            'plural',
            'pluralHonorific',
            'pronounceable',
            ...config.pronouns.morphemes ?? [],
        ];
        const optional = ['history', 'thirdForm', 'smallForm', 'sourcesInfo', 'hidden'];
        const actual = Object.keys(pronouns[0]);
        expect(actual).toEqual(expect.arrayContaining(required));
        expect([...required, ...optional]).toEqual(expect.arrayContaining(actual));
    });
    test('pronouns/pronouns.tsv have unique keys', () => {
        const keys = new Set();
        for (const pronoun of pronouns) {
            const pronounKeys = pronoun.key
                .replace(/،/g, ',')
                .split(',')
                .map((key) => normaliseKey(key));
            for (const key of pronounKeys) {
                expect(keys).not.toContain(key);
                keys.add(key);
            }
        }
    });
    test('pronouns/pronounGroups.tsv match schema', () => {
        if (pronounGroups.length === 0) {
            return;
        }
        const required = ['name', 'pronouns'];
        const optional = ['key', 'description', 'hidden'];
        const actual = Object.keys(pronounGroups[0]);
        expect(actual).toEqual(expect.arrayContaining(required));
        expect([...required, ...optional]).toEqual(expect.arrayContaining(actual));
    });
    test('pronouns/pronounGroups.tsv reference pronouns by canonical name', () => {
        const knownPronounNames = pronouns.map((pronoun) => {
            return pronoun.key.replace(/،/g, ',').split(',')[0];
        });
        for (const pronounGroup of pronounGroups) {
            const pronounNames = pronounGroup.pronouns?.replace(/،/g, ',').split(',') ?? [];
            for (const pronounName of pronounNames) {
                expect(knownPronounNames).toContain(pronounName);
            }
        }
    });
    test('pronouns/examples.tsv match schema', () => {
        if (examples.length === 0) {
            return;
        }
        const required = [
            'singular',
        ];
        if (config.pronouns.plurals) {
            required.push('plural');
        }
        if (config.pronouns.honorifics) {
            required.push('isHonorific');
        }
        const actual = Object.keys(examples[0]);
        expect(actual).toEqual(required);
    });
    test('pronouns/examples.tsv contain valid morphemes', async () => {
        for (const example of examples) {
            expect(example.singular).toHaveValidMorphemes(config.pronouns.morphemes ?? []);
            if (example.plural) {
                expect(example.plural).toHaveValidMorphemes(config.pronouns.morphemes ?? []);
            }
        }
    });
    test('pronouns/examples.tsv contains plural examples when language has plurals', () => {
        const hasExamplesWithPlurals = examples.some((example) => {
            return example.plural && example.plural !== example.singular;
        });
        expect(hasExamplesWithPlurals).toBe(!!config.pronouns.plurals);
    });
    test('pronouns/examples.tsv contains honorific examples when language has honorifics', () => {
        const hasExamplesWithHonorifics = examples.some((example) => example.isHonorific);
        expect(hasExamplesWithHonorifics).toBe(!!config.pronouns.honorifics);
    });

    if (config.nouns.templates?.enabled) {
        const nounTemplates = await loadTsv<NounTemplatesData>(`locale/${code}/nouns/nounTemplates.tsv`);

        describe('nouns/nounTemplates.tsv', () => {
            test('match schema', () => {
                if (nounTemplates.length === 0) {
                    return;
                }
                const required = [...availableGenders(config),
                    ...availableGenders(config).map((gender) => `${gender}Pl`)];
                const actual = Object.keys(nounTemplates[0]);
                expect(actual).toEqual(required);
            });
            test('contains templates when templates are enabled', () => {
                expect(nounTemplates.length > 0).toBe(config.nouns.templates?.enabled);
            });
            test('have exactly one hyphen as placeholder for root', () => {
                for (const template of nounTemplates) {
                    for (const genderWithNumerus of gendersWithNumerus) {
                        const actual = template[genderWithNumerus];
                        if (typeof actual === 'string') {
                            expect(actual).toMatch(/^(?:[^-]*-[^-]*(?:\/|$))+/);
                        }
                    }
                }
            });
        });
    }

    let nounsData: NounsData | undefined;
    try {
        nounsData = await loadSuml<NounsData>(`locale/${code}/nouns/nounsData.suml`);
    } catch (error) {
        nounsData = undefined;
    }

    if (config.nouns.enabled && config.nouns.conventions?.enabled) {
        test('config.nouns.conventions.enabled requires properties in nouns/nounsData.suml', () => {
            expect(nounsData?.morphemes).toBeDefined();
            expect(nounsData?.conventions).toBeDefined();
            expect(nounsData?.groups).toBeDefined();
            expect(nounsData?.classes).toBeDefined();
            expect(nounsData?.classExample).toBeDefined();
            expect(nounsData?.examples).toBeDefined();
        });
    }

    if (nounsData) {
        describe('nouns/nounsData.suml', () => {
            if (nounsData.grammarTables) {
                test('grammar tables have valid variant types', async () => {
                    const grammarTableVariantsConverter = (await import(`~/locale/${code}/language/grammarTableVariantsConverter.ts`)).default as VariantsFromBaseConverter;

                    for (const grammarTable of nounsData.grammarTables!) {
                        for (const section of grammarTable.sections) {
                            if (!Array.isArray(section.variants)) {
                                expect(Object.keys(grammarTableVariantsConverter)).toContain(section.variants.type);
                            }
                        }
                    }
                });
            }
            if (nounsData.conventions) {
                test('conventions have valid morphemes', () => {
                    for (const convention of Object.values(nounsData.conventions!)) {
                        expect(nounsData.morphemes)
                            .toEqual(expect.arrayContaining(Object.keys(convention.morphemes)));
                    }
                });
            }
            if (nounsData.conventions && nounsData.groups) {
                test('convention groups reference conventions by key', () => {
                    for (const group of Object.values(nounsData.groups!)) {
                        expect(Object.keys(nounsData.conventions!)).toEqual(expect.arrayContaining(group.conventions));
                    }
                });
            }
            if (nounsData.cases && nounsData.declensions) {
                test('declensions have valid cases', () => {
                    for (const declension of Object.values(nounsData.declensions!)) {
                        for (const numerus of ['singular', 'plural'] as const) {
                            expect(Object.keys(nounsData.cases!))
                                .toEqual(expect.arrayContaining(Object.keys(declension[numerus] ?? {})));
                        }
                    }
                });
            }
        });
    }
});