Language-wise Regular Expressions.

zodLanguageValidation.js

import { z } from "zod"; // Ensure Zod is imported

/**
 * Regular expressions for various languages.
 * Each language key maps to its corresponding Unicode regex pattern.
 */
const languageRegexPatterns = {
    english: /^[A-Za-z\s]*$/, // English
    bangla: /^[\u0980-\u09FF\s]*$/, // Bangla (Bengali)
    hindi: /^[\u0900-\u097F\s]*$/, // Hindi (Devanagari)
    arabic: /^[\u0600-\u06FF\s]*$/, // Arabic
    urdu: /^[\u0600-\u06FF\s]*$/, // Urdu (Uses Arabic script)
    persian: /^[\u0600-\u06FF\s]*$/, // Persian (Farsi, Dari)
    pashto: /^[\u0600-\u06FF\s]*$/, // Pashto (Similar to Arabic)
    chinese: /^[\u4E00-\u9FFF\s]*$/, // Chinese (Simplified & Traditional)
    mandarin: /^[\u4E00-\u9FFF\s]*$/, // Mandarin (Uses Chinese characters)
    japanese: /^[\u3040-\u30FF\u4E00-\u9FFF\s]*$/, // Japanese (Hiragana, Katakana, Kanji)
    korean: /^[\uAC00-\uD7A3\s]*$/, // Korean (Hangul)
    tamil: /^[\u0B80-\u0BFF\s]*$/, // Tamil
    telugu: /^[\u0C00-\u0C7F\s]*$/, // Telugu
    marathi: /^[\u0900-\u097F\s]*$/, // Marathi (Devanagari)
    gujarati: /^[\u0A80-\u0AFF\s]*$/, // Gujarati
    punjabi: /^[\u0A00-\u0A7F\s]*$/, // Punjabi (Gurmukhi)
    sinhala: /^[\u0D80-\u0DFF\s]*$/, // Sinhala (Sri Lanka)
    burmese: /^[\u1000-\u109F\s]*$/, // Burmese (Myanmar)
    thai: /^[\u0E00-\u0E7F\s]*$/, // Thai
    lao: /^[\u0E80-\u0EFF\s]*$/, // Lao
    cambodian: /^[\u1780-\u17FF\s]*$/, // Khmer (Cambodian)
    hebrew: /^[\u0590-\u05FF\s]*$/, // Hebrew
    greek: /^[\u0370-\u03FF\s]*$/, // Greek
    russian: /^[\u0400-\u04FF\s]*$/, // Russian (Cyrillic)
    ukrainian: /^[\u0400-\u04FF\s]*$/, // Ukrainian (Cyrillic)
    bulgarian: /^[\u0400-\u04FF\s]*$/, // Bulgarian (Cyrillic)
    serbian: /^[\u0400-\u04FF\s]*$/, // Serbian (Cyrillic)
    georgian: /^[\u10A0-\u10FF\s]*$/, // Georgian
    armenian: /^[\u0530-\u058F\s]*$/, // Armenian
    vietnamese: /^[\p{Script=Latin}\s]*$/u, // Vietnamese (Uses Latin but with diacritics)
    french: /^[\p{Script=Latin}\s]*$/u, // French (Uses Latin script)
    spanish: /^[\p{Script=Latin}\s]*$/u, // Spanish (Uses Latin script)
    portuguese: /^[\p{Script=Latin}\s]*$/u, // Portuguese (Uses Latin script)
    german: /^[\p{Script=Latin}\s]*$/u, // German (Uses Latin script)
    italian: /^[\p{Script=Latin}\s]*$/u, // Italian (Uses Latin script)
    dutch: /^[\p{Script=Latin}\s]*$/u, // Dutch (Uses Latin script)
    polish: /^[\p{Script=Latin}\s]*$/u, // Polish (Uses Latin script)
    turkish: /^[\p{Script=Latin}\s]*$/u, // Turkish (Uses Latin script)
    malay: /^[\p{Script=Latin}\s]*$/u, // Malay (Uses Latin script)
    indonesian: /^[\p{Script=Latin}\s]*$/u, // Indonesian (Uses Latin script)
    filipino: /^[\p{Script=Latin}\s]*$/u, // Filipino (Tagalog, Uses Latin script)
};

/**
 * Creates a Zod schema for a non-empty string field.
 * @param {string} fieldName - The name of the field to validate.
 * @returns {z.ZodString} - A Zod schema that requires a non-empty string.
 */
const nonEmptyString = (fieldName) =>
    z.string().nonempty(`${fieldName} is required`);

/**
 * Creates a Zod schema for validating text in a specific language.
 *
 * @param {string} fieldName - The name of the field to validate.
 * @param {string} languageKey - The key representing the language (e.g., "english", "bangla", "arabic").
 * @returns {z.ZodString} - A Zod schema that enforces the language constraint.
 * @throws {Error} - Throws an error if the provided language key is invalid.
 *
 * @example
 * const banglaSchema = validLanguageString("Name", "bangla");
 * banglaSchema.safeParse("বাংলা নাম"); // ✅ Valid
 * banglaSchema.safeParse("English Name"); // ❌ Invalid
 */
const validLanguageString = (fieldName, languageKey) => {
    const languageRegex = languageRegexPatterns[languageKey];

    if (!languageRegex) {
        throw new Error(`Invalid language key: ${languageKey}`);
    }

    return nonEmptyString(fieldName).refine(
        (value) => languageRegex.test(value),
        {
            message: `${fieldName} must contain only ${languageKey} characters.`,
        }
    );
};

// ✅ Example Usage:
const banglaSchema = validLanguageString("Name", "bangla");
const englishSchema = validLanguageString("Name", "english");
const arabicSchema = validLanguageString("Name", "arabic");

// 🔹 Test Cases:
console.log(banglaSchema.safeParse("বাংলা নাম")); // ✅ Valid
console.log(englishSchema.safeParse("English Name")); // ✅ Valid
console.log(arabicSchema.safeParse("عربي النص")); // ✅ Valid
console.log(englishSchema.safeParse("বাংলা নাম")); // ❌ Invalid
console.log(banglaSchema.safeParse("English Name")); // ❌ Invalid