apps/whatsapp-worker/src/normalizeWolof.ts

/**
 * Wolof STT Normalizer Utility v2.0
 * Includes change tracking and WhatsApp message shortening.
 */

const NORMALIZATION_RULES: Record<string, string> = {
    // A. Corrections fréquentes verbes
    "damae": "damay",
    "dama": "damay",
    "dma": "damay",
    "jai": "jaay",
    "jaai": "jaay",
    "jaye": "jaay",
    "jendi": "jënd",
    "fei": "fey",
    "fay": "fey",

    // B. Produits / business courant
    "yere": "yére",
    "yare": "yére",
    "sandwiche": "sandwich",
    "pan": "mburu",
    "cafe": "café",
    "sabu": "sabu",
    "omo": "omo",
    "patat": "patas",
    "ognon": "sooble",
    "riz": "ceeb",

    // C. Lieux
    "yof": "Yoff",
    "dakar": "Dakar",
    "pikine": "Pikine",
    "guediawaye": "Guédiawaye",
    "keur": "kër",
    "ker": "kër",
    "sikarche": "ci kër",
    "sikarshe": "ci kër",
    "sikarce": "ci kër",
    "sikaarché": "ci kër",
    "quartier": "quartier",
    "banlieu": "banlieue",

    // D. Mots fonctionnels
    "si": "ci",
    "fane": "fan",
    "fana": "fan",
    "lana": "lan",
    "lanna": "lan",
    "nakka": "naka",
    "nakha": "naka",

    // E. Finance
    "niak": "ñàkk",
    "niakk": "ñàkk",
    "dencal": "denc", // requested dencal -> denc
    "limal": "lim",
    "ganee": "gañ",
    "gane": "gañ",
    "borom": "boroom",
    "xaalisou": "xaalis",
    "xaliss": "xaalis",
};

const CAPITALIZED_PLACES = ["Yoff", "Dakar", "Pikine", "Guédiawaye"];

export interface NormalizationResult {
    normalizedText: string;
    changes: string[]; // Format: ["damae -> damay", ...]
}

/**
 * Normalizes Wolof STT output to standard orthography with change tracking.
 */
export function normalizeWolof(rawText: string): NormalizationResult {
    if (!rawText) return { normalizedText: '', changes: [] };

    let text = rawText.trim().replace(/\s{2,}/g, " ");
    const changes: string[] = [];

    // Basic cleaning while keeping specific characters
    const words = text.split(" ");
    const processedWords = words.map(word => {
        const lowerWord = word.toLowerCase().replace(/[.,/#!$%^&*;:{}=\-_`~()]/g, "");
        if (NORMALIZATION_RULES[lowerWord]) {
            const replacement = NORMALIZATION_RULES[lowerWord];
            if (lowerWord !== replacement.toLowerCase()) {
                changes.push(`${lowerWord} -> ${replacement}`);
            }
            return replacement;
        }

        // Handle capitalized places
        const matchingPlace = CAPITALIZED_PLACES.find(p => p.toLowerCase() === lowerWord);
        if (matchingPlace) {
            if (matchingPlace !== word) {
                changes.push(`${word} -> ${matchingPlace}`);
            }
            return matchingPlace;
        }

        return word;
    });

    let normalizedText = processedWords.join(" ");

    // Preserve ñ and ë is already handled by not removing them in regex above if they are part of word

    // Capitalize first letter of phrase
    if (normalizedText.length > 0) {
        normalizedText = normalizedText.charAt(0).toUpperCase() + normalizedText.slice(1);
    }

    // Special Business Verbs Logic (Ensure common ones are standard)
    // jaay (vendre), jënd (acheter), fey (payer), denc (épargner), lim (compter)
    // These are already in NORMALIZATION_RULES.

    return { normalizedText, changes: Array.from(new Set(changes)) };
}

/**
 * WhatsApp-friendly message shortener.
 */
export function shortenForWhatsApp(text: string, _lang: string = 'FR'): string[] {
    const messages: string[] = [];

    // 1. Emoji limit (Max 2)
    const emojiRegex = /(\p{Emoji_Presentation}|\p{Emoji}\uFE0F)/gu;
    const emojis = text.match(emojiRegex) || [];
    let cleanText = text;
    if (emojis.length > 2) {
        let count = 0;
        cleanText = text.replace(emojiRegex, (match) => {
            count++;
            return count <= 2 ? match : '';
        });
    }

    // 2. Remove "marketing" repetitive phrases (heuristic)
    cleanText = cleanText.replace(/Bienvenue sur XAMLÉ, l'académie de l'entrepreneur informel/gi, "Xamle");

    // 3. Length check and split (Max 280)
    if (cleanText.length > 280) {
        // Try to split by sentence
        const sentences = cleanText.split(/(?<=[.!?])\s+/);
        let current = "";
        for (const s of sentences) {
            if ((current + s).length > 280) {
                if (current) messages.push(current.trim());
                current = s;
            } else {
                current += (current ? " " : "") + s;
            }
        }
        if (current) messages.push(current.trim());
    } else {
        messages.push(cleanText.trim());
    }

    // Limit to 2 messages maximum
    return messages.slice(0, 2);
}

/**
 * Formats a bilingual message (Wolof first, FR second on new line)
 */
export function formatBilingual(wo: string, fr: string): string {
    return `${wo}\n(FR) ${fr}`;
}