client/src/components/LeaderboardSection/hooks/useLanguageStats.js

import { useMemo, useRef } from "react";

const LANGUAGE_FAMILIES = {
  "Indo-European": {
    Germanic: ["english", "german", "dutch", "danish", "swedish", "icelandic"],
    Romance: [
      "french",
      "spanish",
      "italian",
      "portuguese",
      "romanian",
      "catalan",
      "galician",
    ],
    Slavic: [
      "russian",
      "polish",
      "czech",
      "slovak",
      "ukrainian",
      "bulgarian",
      "slovenian",
      "serbian",
      "croatian",
    ],
    Baltic: ["lithuanian", "latvian"],
    "Indo-Iranian": [
      "persian",
      "hindi",
      "bengali",
      "gujarati",
      "nepali",
      "marathi",
      "indic",
    ],
    Greek: ["greek"],
    Armenian: ["armenian"],
  },
  "Sino-Tibetan": ["chinese", "mandarin", "taiwanese"],
  Afroasiatic: [
    "arabic",
    "hebrew",
    "darija", // Variante de l'arabe
  ],
  Austronesian: [
    "indonesian",
    "malay",
    "filipino",
    "singlish", // Créole basé sur l'anglais mais avec influence malaise
  ],
  "Niger-Congo": ["swahili", "yoruba"],
  Dravidian: ["tamil", "telugu", "kannada", "malayalam"],
  Austroasiatic: ["vietnamese"],
  "Kra-Dai": ["thai"],
  Japonic: [
    "japanese",
    "日本語", // Japonais en caractères japonais
  ],
  Koreanic: ["korean"],
  Uralic: ["hungarian", "finnish", "estonian"],
  Turkic: ["turkish", "azerbaijani", "uzbek", "kazakh"],
  "Language Isolate": [
    "basque", // Langue isolée, pas de famille connue
  ],
  "Other Languages": [], // Will catch any language not in other families
};

const FAMILY_ORDER = [
  "Indo-European", // Plus grande famille en Europe et Asie du Sud
  "Sino-Tibetan", // Deuxième plus grande famille en Asie
  "Afroasiatic", // Principale famille en Afrique du Nord et Moyen-Orient
  "Austronesian", // Principale famille en Asie du Sud-Est insulaire
  "Niger-Congo", // Plus grande famille en Afrique subsaharienne
  "Dravidian", // Principale famille en Inde du Sud
  "Austroasiatic", // Importante en Asie du Sud-Est continentale
  "Kra-Dai", // Famille du thaï
  "Japonic", // Japonais
  "Koreanic", // Coréen
  "Uralic", // Famille finno-ougrienne
  "Turkic", // Famille des langues turciques
  "Language Isolate", // Langues isolées
  "Other Languages", // Autres langues non classifiées
];

const findLanguageFamily = (language) => {
  for (const [family, content] of Object.entries(LANGUAGE_FAMILIES)) {
    if (family === "Indo-European") {
      // Cas spécial pour l'Indo-Européen qui a des sous-familles
      for (const [subFamily, languages] of Object.entries(content)) {
        if (languages.includes(language.toLowerCase())) {
          return { family, subFamily };
        }
      }
    } else if (content.includes(language.toLowerCase())) {
      return { family, subFamily: language };
    }
  }
  return { family: "Other Languages", subFamily: language };
};

export const useLanguageStats = (leaderboards, filteredLeaderboards) => {
  const statsRef = useRef(null);
  const languagesRef = useRef(null);

  // Reset stats when leaderboards or filteredLeaderboards change
  if (leaderboards && filteredLeaderboards) {
    // Calculate unique languages from all leaderboards
    const langMap = new Map();
    const langFamilyMap = new Map();

    leaderboards.forEach((board) => {
      board.tags?.forEach((tag) => {
        if (tag.toLowerCase().startsWith("language:")) {
          const lang = tag.split(":")[1].toLowerCase();
          langMap.set(lang, (langMap.get(lang) || 0) + 1);
          const { family } = findLanguageFamily(lang);
          langFamilyMap.set(family, (langFamilyMap.get(family) || 0) + 1);
        }
      });
    });

    const langArray = Array.from(langMap.entries()).map(([lang, count]) => {
      const { family } = findLanguageFamily(lang);
      return [lang, count, family];
    });

    languagesRef.current = langArray
      .sort((a, b) => {
        const familyA = a[2];
        const familyB = b[2];

        const orderA = FAMILY_ORDER.indexOf(familyA);
        const orderB = FAMILY_ORDER.indexOf(familyB);

        if (orderA !== orderB) {
          return orderA - orderB;
        }

        // Sort alphabetically by language name instead of by count
        return a[0].localeCompare(b[0]);
      })
      .map(([lang]) => lang);

    // Calculate stats based on filtered leaderboards
    statsRef.current = new Map();
    languagesRef.current.forEach((lang) => {
      const count = filteredLeaderboards.filter((board) =>
        board.tags?.some(
          (tag) => tag.toLowerCase() === `language:${lang.toLowerCase()}`
        )
      ).length;
      statsRef.current.set(lang, count);
    });
  }

  return {
    languages: languagesRef.current,
    languageStats: statsRef.current,
    LANGUAGE_FAMILIES,
    findLanguageFamily,
  };
};