#!/usr/bin/env python3
"""Parse curated URL lists, diff against wiki/catalog.json, report gaps."""
import json, re, unicodedata
from pathlib import Path

CATALOG = Path(__file__).resolve().parents[1] / "wiki" / "catalog.json"

# Games extracted from the 7 source lists (deduped, normalized keys).
LIST_ENTRIES = [
    # UfzGCYzA
    ("Abstract", "https://www.pokecommunity.com/threads/pokémon-abstract-version.527324/"),
    ("Alchemist", "https://www.pokecommunity.com/threads/pokémon-alchemist.378280/"),
    ("AfricanVs", "https://mega.nz/file/UgZj3BCQ#Xc6243i1iMfBTTJR7NfEsA7mFgfCFj37e7_u_B1EvMM"),
    ("Awakening", "https://www.pokecommunity.com/threads/pokémon-awakening.535910/"),
    ("Bizarre", "https://drive.google.com/file/d/1NXXr96vzeMD8sGplV1sWZFSph2vZijaK/view"),
    ("Bushido", "https://eeveeexpo.com/bushido/"),
    ("Empyrean", "https://www.pokecommunity.com/threads/pokémon-empyrean-v1-1-final-arc-released.411322/"),
    ("Eon Guardians", "https://www.pokecommunity.com/threads/pokémon-eon-guardians.527913/"),
    ("Floral Tempus", "https://www.pokecommunity.com/threads/pokémon-floral-tempus-ex-1-9-4-released-12-gym-badges-elite-4-champion.409175/"),
    ("HGSS Sevii", "https://www.pokecommunity.com/threads/pokémon-hgss-sevii-islands-hoenn-postgame.434636/"),
    ("Infinity", "https://eeveeexpo.com/infinity/"),
    ("Insurgence", "https://p-insurgence.com"),
    ("Itinerant", "https://www.pokecommunity.com/threads/pokemon-itinerant.528218/"),
    ("Infinite Fusion", "https://www.pokecommunity.com/threads/v6-4-pokémon-infinite-fusion-new-pokémon-character-customization-and-more.347883/"),
    ("Legends of the Arena", "https://www.pokecommunity.com/threads/pokémon-legends-of-the-arena.298738/"),
    ("Mega Adventures", "https://www.pokecommunity.com/threads/pokemon-mega-adventure.362058/"),
    ("Nightmare", "https://www.pokecommunity.com/threads/pokémon-nightmare.504856/"),
    ("Pokerogue", "https://pokerogue.net"),
    ("Poketale", "https://www.mediafire.com/file/n6yzxnclf2uys7c/Poketale.zip/file"),
    ("Realidea System", "https://www.pokecommunity.com/threads/pokémon-realidea-system-now-playable-in-english.481090/"),
    ("Reborn", "https://www.rebornevo.com/pr/index.html/"),
    ("Reloaded", "https://pokemon-reloaded.blogspot.com"),
    ("Reminiscencia", "https://www.pokecommunity.com/threads/pokémon-reminiscencia-full-release-spa-en-translation-complete.529202/"),
    ("Shattered Light", "https://eeveeexpo.com/shattered-light/"),
    ("Solar Light Lunar Dark", "https://pokemonsolarlightlunardark.fandom.com/wiki/Pokémon_Solar_Light_%26_Lunar_Dark_Wiki"),
    ("Tectonic", "https://www.tectonic-game.com"),
    ("Uranium", "https://pokemon-uranium.fandom.com/wiki/Main_Page"),
    ("Xenoverse", "https://mega.nz/file/FbBgiTTK#qJpzZeUb5XqO5rLt6meW5_tOXeQ3K6y3_mslGNUf41M"),
    ("Zeta", "https://zo.p-insurgence.com"),
    ("Atlas", None), ("Unbreakable Ties", None), ("Bioterror", None), ("Fuso Meteor", None),
    ("Comet", None), ("Ruckus", None), ("Island", None), ("Impetu", None),
    ("Slowpoke Shack", None), ("Empire", None), ("Flux", None),
    # S1CSW465
    ("Consonancia", "https://www.pokecommunity.com/threads/pokémon-consonancia-full-game.537329/"),
    ("Concealed", "https://www.pokecommunity.com/threads/concealed.527325/"),
    ("Chaos in Vesita", "https://www.pokecommunity.com/threads/pokémon-chaos-in-vesita-new-version-out-now-eng-ger.491767/"),
    ("Dark Horizon", "https://eeveeexpo.com/threads/9332/"),
    ("Decay", "https://eeveeexpo.com/decay/"),
    ("Fire Ash", "https://www.pokecommunity.com/threads/pokémon-fire-ash.399096/"),
    ("Gaidr Deluxe", "https://www.pokecommunity.com/threads/pokémon-gadir-deluxe-complete-v-1-3.539451/"),
    ("Hero Legacy", "https://www.pokecommunity.com/threads/pokémon-hero-legacy-eng-de-now-with-new-game-mode.538518/"),
    ("Hollow Woods", "https://www.pokecommunity.com/threads/pokémon-hollow-woods.525865/"),
    ("Hydro Bliss", "https://eeveeexpo.com/threads/9100/"),
    ("Jam Festival", "https://www.pokecommunity.com/threads/pkmn-jam-festival.536702/"),
    ("Keishou", "https://www.pokecommunity.com/threads/pokémon-keishou-v2-06-1.504806/"),
    ("Prisme", "https://pokemonprisme.com"),
    ("Re:Union", "https://projetreunion.com"),
    ("Relict", "https://phiongames.blogspot.com/2025/12/descarga-pokemon-relict.html"),
    # 6wPjUcrE
    ("Spades and Clubs", "https://senta-storage-system.neocities.org/main"),
    ("Giratina's Legend", "https://www.pokecommunity.com/threads/pokemon-giratinas-legend-07-01-24.289561"),
    ("Crown", "https://www.pokecommunity.com/threads/pokémon-crown.463821"),
    ("Odyssey II", None),  # Discord only
    ("Coral", "https://www.pokecommunity.com/threads/pokémon-coral-version-2022-demo-out-now.402361"),
    ("DarkFire", "https://www.pokecommunity.com/threads/new-2024-beta-2-1-out-now-pokémon-darkfire.421395/"),
    ("Sovereign of the Skies", "https://www.pokecommunity.com/threads/sovereign-of-the-skies-new-beta-2-0-0-version-released-06-10-2021.292651/"),
    ("Legacy Edition", None),  # Discord only
    ("Iridium", None),  # Discord only
    ("Light Platinum DS", None),  # Twitter only
    # ZwA8ng8E
    ("Deserted", "https://eeveeexpo.com/threads/5344/"),
    # 1S4acLQA
    ("Saiph", "https://www.pokecommunity.com/threads/pokémon-saiph-the-vytroverse-part-1-full-game-released.420990"),
    ("Sors", "https://www.pokecommunity.com/threads/pokémon-sors-the-vytroverse-part-2-full-game-v1-3-available-now.433238"),
    ("Saiph 2", "https://www.pokecommunity.com/threads/pokémon-saiph-2-the-vytroverse-part-3-full-game-v1-4-0-update-available.458509"),
    ("Sors 2", "https://www.pokecommunity.com/threads/pokémon-sors-2-the-vytroverse-part-4-tech-demo-available.526108"),
    ("Victory Fire", "https://www.pokecommunity.com/threads/pokemon-victory-fire-version-2-75-released.285031/"),
    ("Resolute", "https://www.pokecommunity.com/threads/pokemon-resolute-version-version-2-97-released.294565/"),
    ("Mega Power", "https://www.pokecommunity.com/threads/pokemon-mega-power-completed-beta-5-73-released.325422/"),
    ("Nameless", "https://www.pokecommunity.com/threads/pokemon-nameless-version-complete-beta-5-35-released-with-darker-future-episode.419717/"),
    ("Ruby Destiny", None),
    ("ROWE", "https://www.pokecommunity.com/threads/pokemon-r-o-w-e-2-0-an-open-world-version-of-pokémon-emerald-gen-9-16-badges-sevii-islands-following-pokémon-costumes-and-much-more.442592/"),
    ("Gaia", "https://www.pokecommunity.com/threads/pokémon-gaia-version.326118/"),
    ("Vanguard", None),
    # tARt0w7x
    ("Adventure Red", "https://www.pokecommunity.com/threads/pokémon-adventure-red-chapter-new-beta-expansion.298920/"),
    ("AshGray", "https://www.pokecommunity.com/threads/pokémon-ashgray-version-beta-4-5-released.180722/"),
    ("Azure Horizons", "https://hacksrepairman.blogspot.com/2020/03/pokemon-azure-horizons-fixed-beta2.html"),
    ("Crystal Clear", "https://shockslayer.com/crystal-clear/"),
    ("Dark Violet", "https://www.pokecommunity.com/threads/pokémon-darkviolet-full-version-released.291789/"),
    ("Dreams", "https://www.pokecommunity.com/threads/dreams-completed-version-1-5-1-now-available.443298/"),
    ("Elite Redux", "https://www.pokecommunity.com/threads/pokémon-elite-redux-v2-1-complete-—-unique-multi-ability-difficulty-hack.499227/"),
    ("Emerald Rogue", "https://www.pokecommunity.com/threads/pokemon-emerald-rogue.479406/"),
    ("Emerald Seaglass", "https://ko-fi.com/s/4a1535f351"),
    ("Elysium", "https://www.pokecommunity.com/threads/pokémon-elysium-gba.502953/"),
    ("Fire Red Extended", "https://www.pokecommunity.com/threads/pokémon-fire-red-extended-version.466535/"),
    ("Flora Sky", "https://www.pokemoncoders.com/pokemon-flora-sky/"),
    ("Glazed", "https://www.pokemoncoders.com/pokemon-glazed/"),
    ("Light Platinum", "https://www.pokemoncoders.com/pokemon-light-platinum-rom-hack/"),
    ("Liquid Crystal", "https://www.pokecommunity.com/threads/pokémon-liquid-crystal-3-3-xxxxx-live-beta.242023/"),
    ("Mariomon", "https://www.pokecommunity.com/threads/super-mariomon.535764/"),
    ("Odyssey", "https://www.pokecommunity.com/threads/pokémon-odyssey-complete-v4-0-1.488536/"),
    ("Pisces", "https://pebblerplatoon.miraheze.org/wiki/Pisces_Download"),
    ("Prism", "https://rainbowdevs.com/"),
    ("Polished Crystal", "https://www.pokecommunity.com/threads/pokémon-polished-crystal-update-3-1-1.373172/"),
    ("Radical Red", "https://www.pokecommunity.com/threads/pokémon-radical-red-version-4-1-released-gen-9-dlc-pokemon-character-customization-now-available.437688/"),
    ("Recharged Series", "https://jaizu.moe"),
    ("FireRed Rocket Edition", "https://www.pokecommunity.com/threads/pokémon-firered-rocket-edition-completed.360725/"),
    ("Scorched Silver", "https://www.pokecommunity.com/threads/pokémon-scorched-silver-v1-3-complete.529230/"),
    ("Sienna", "https://www.pokecommunity.com/threads/hack-of-the-year-2010-pokémon-sienna-complete-version-released.202372/"),
    ("Snakewood", "https://www.pokecommunity.com/threads/pokémon-snakewood-version.235371/"),
    ("Team Rocket Edition", "https://www.pokecommunity.com/threads/pokémon-team-rocket-edition-dragonsden-version-kanto-sevii-johto-next-release-dlc-season-4.527368/"),
    ("The Pit", "https://www.pokecommunity.com/threads/the-pit-v2-roguelite-style-hack.528423/"),
    ("Unbound", "https://www.pokecommunity.com/threads/pokémon-unbound-completed.382178/"),
    ("Vega", "https://www.pokecommunity.com/threads/pokémon-vega-and-altair-sirius-english-version.365959/"),
    ("Vega Fairy EX", "https://www.pokecommunity.com/threads/pokemon-vega-fairy-edition-ex-minus-v1-4-balance-updates-and-skarmory-evo.475538/"),
    # Google Doc
    ("Lazarus", "https://ko-fi.com/nemo622"),
    ("Renegade Platinum", "https://projectpokemon.org/home/forums/topic/52294-pokémon-renegade-platinum/"),
    ("Heart n Soul", "https://www.hackdex.app/hack/pokemon-heart-and-soul"),
    ("Dreamstone Mysteries", "https://github.com/dsmyst/dreamstone-mysteries"),
    ("Rejuvenation", None),
]

ALIASES = {
    "adventure red": "adventure red chapter",
    "ashgray": "ash gray",
    "ash gray": "ashgray",
    "firered rocket edition": "firered rocket edition",
    "fire red extended": "fire red extended",
    "darkfire": "dark fire",
    "gaidr deluxe": "gadir deluxe",
    "reunion": "re union",
    "re:union": "re union",
    "re union dx": "re union",
    "rowe": "r o w e",
    "r.o.w.e": "r o w e",
    "itinerant": "iternant",
    "keishou": "kieshou",
    "giratina's legend": "giratina",
    "vega fairy ex": "vega fairy",
    "team rocket edition": "team rocket edition",
    "mega adventures": "mega adventure",
    "reminiscencia": "reminiscensia",
    "spades and clubs": "spades",
    "odyssey ii": "odyssey 2",
    "heroes of lemuria": "odyssey 2",
    "light platinum ds": "light platinum ds",
    "heart n soul": "heart and soul",
    "recharged series": "recharged",
}


def norm(s: str) -> str:
    s = unicodedata.normalize("NFKD", s).encode("ascii", "ignore").decode().lower()
    s = re.sub(r"^pok[eé]mon\s+", "", s)
    s = re.sub(r"[^a-z0-9]+", " ", s).strip()
    return ALIASES.get(s, s)


def load_catalog():
    data = json.loads(CATALOG.read_text(encoding="utf-8"))
    by_norm = {}
    for h in data["hacks"]:
        title = h.get("title") or h.get("stem", "")
        key = norm(title)
        by_norm.setdefault(key, []).append(h)
        if h.get("stem"):
            by_norm.setdefault(norm(h["stem"]), []).append(h)
    return by_norm


def match_catalog(name, catalog):
    key = norm(name)
    hits = catalog.get(key, [])
    if hits:
        return hits[0]
    # fuzzy: substring
    for ck, hits in catalog.items():
        if key in ck or ck in key:
            return hits[0]
    return None


def main():
    catalog = load_catalog()
    seen = set()
    rows = []
    for name, url in LIST_ENTRIES:
        nk = norm(name)
        if nk in seen:
            continue
        seen.add(nk)
        hit = match_catalog(name, catalog)
        rows.append({
            "name": name,
            "url": url,
            "in_catalog": hit is not None,
            "catalog_title": hit["title"] if hit else None,
            "platform": hit.get("platform") if hit else None,
            "library_path": hit.get("library_path") if hit else None,
        })

    missing_cat = [r for r in rows if not r["in_catalog"]]
    in_cat = [r for r in rows if r["in_catalog"]]

    print(f"Total unique list entries: {len(rows)}")
    print(f"In catalog: {len(in_cat)}")
    print(f"NOT in catalog: {len(missing_cat)}")
    print("\n=== NOT IN CATALOG ===")
    for r in sorted(missing_cat, key=lambda x: x["name"].lower()):
        print(f"  {r['name']:30} {r['url'] or '(no URL)'}")

    print("\n=== IN CATALOG (for library check) ===")
    for r in sorted(in_cat, key=lambda x: x["name"].lower()):
        lp = r["library_path"] or "—"
        print(f"  {r['name']:30} | {r['platform'] or '?':5} | {lp}")


if __name__ == "__main__":
    main()