"""providers/anidb.py — anidb.app source (primary).

Sits behind Cloudflare: plain curl gets 403. All requests go through the
shared `_get` (curl_cffi Chrome impersonation). Parsers are regex against
live markup — they break here, in one place, when the layout changes.
"""
import re

from ..upstream import _cached, _get

BASE = "https://anidb.app"

_ANIME_LINK = re.compile(r'anime/([a-z0-9-]+)-(\d+)"[^>]*title="([^"]*)"')


def _search(q):
    """Browse page -> [{id, slug_id, name, poster}]"""
    html = _get(f"{BASE}/browse", params={"q": q}).text
    out, seen = [], set()
    for m in re.finditer(
        r'<a href="https://anidb\.app/anime/([a-z0-9-]+-\d+)"[^>]*title="([^"]*)"[^>]*>(.*?)</a>',
        html, re.S):
        slug_id, name, block = m.group(1), m.group(2), m.group(3)
        if slug_id in seen:
            continue
        seen.add(slug_id)
        img = re.search(r'<img src="([^"]+)"', block)
        out.append({"id": int(slug_id.rsplit("-", 1)[1]), "slug_id": slug_id,
                    "name": name, "poster": img.group(1) if img else None})
    return out


def _details(slug_id):
    """Details page -> {name, synopsis, mal_id, poster, seasons}"""
    p = _get(f"{BASE}/anime/{slug_id}").text
    mal = re.search(r'myanimelist\.net/anime/(\d+)', p)
    syn = re.search(r'>Synopsis</h2>.*?<p[^>]*>(.*?)</p>', p, re.S)
    h1 = re.search(r'<h1[^>]*>(.*?)</h1>', p, re.S)
    post = re.search(r'src="(https://cdn\.xlsbox\.com/poster/[^"\']+)"', p)
    seasons, seen = [], set()
    i = p.find(">Seasons<")
    if i > 0:
        cnt = re.search(r">(\d+) entries in this series<", p[i:i + 400])
        n = int(cnt.group(1)) if cnt else 10  # fall back to cap if count markup changes
        for m in _ANIME_LINK.finditer(p[i:i + 20000]):
            sid = f"{m.group(1)}-{m.group(2)}"
            if sid in seen:
                continue
            seen.add(sid)
            seasons.append({"slug_id": sid, "name": m.group(3), "id": int(m.group(2))})
            if len(seasons) == n:
                break
    return {"id": int(slug_id.rsplit("-", 1)[1]),
            "name": h1.group(1).strip() if h1 else None,
            "synopsis": re.sub(r"<[^>]+>", "", syn.group(1)).strip() if syn else None,
            "mal_id": int(mal.group(1)) if mal else None,
            "poster": post.group(1) if post else None,
            "seasons": seasons}


def _episodes(anime_id):
    """JSON API -> [{id, number, filler}]"""
    d = _get(f"{BASE}/api/frontend/anime/{anime_id}/episodes").json()
    return [{"id": e["id"], "number": e["number"], "filler": bool(e.get("filler"))}
            for e in d["episodes"]]


def _languages(ep_id):
    """JSON API -> [{code, name, embed_url}] (jpn=sub, eng=dub)"""
    return _get(f"{BASE}/api/frontend/episode/{ep_id}/languages").json()["languages"]


def _master_m3u8(ep_id, lang="eng"):
    """Embed page -> 'file: <master m3u8>' URL (None if unavailable)"""
    for l in languages(ep_id):
        if l["code"] == lang:
            m = re.search(r"file: '([^']*)'", _get(l["embed_url"]).text)
            if m:
                return m.group(1)
    return None


# Provider interface. Cache keys are namespaced with "anidb:" so a shared
# cache dict can't leak another provider's value into this one.
search = lambda q: _cached(f"anidb:search:{q.lower()}", 300, _search, q)
details = lambda s: _cached(f"anidb:details:{s}", 3600, _details, s)
episodes = lambda i: _cached(f"anidb:episodes:{i}", 300, _episodes, i)  # 5m: airing shows gain eps
languages = lambda e: _cached(f"anidb:languages:{e}", 3600, _languages, e)
master_m3u8 = lambda e, l="eng": _cached(f"anidb:master:{e}:{l}", 3600, _master_m3u8, e, l)