"""Universal thumbnail URL extractor for tube pages.

Direct scrapery (search-only) zwracają RawScene z thumbnail_url=None dla większości
tube'ów (xnxx, hdporn92, sxyland, sxyprn). Detail page zawiera URL miniatury w
jednym z patternów:

1. **OpenGraph** (najbardziej powszechne):
   `<meta property="og:image" content="https://cdn.../thumb.jpg">`
2. **Twitter Card** (fallback gdy og:image brak):
   `<meta name="twitter:image" content="...">`
3. **Schema.org VideoObject LD-JSON**:
   `"thumbnailUrl": "https://..."` lub `"thumbnailUrl": ["url1", "url2"]`
4. **html5player** (KVS-based — xnxx/xvideos):
   `html5player.setThumbUrl('https://thumb-cdn77.xnxx-cdn.com/.../t.jpg')`

Funkcja zwraca pierwszy znaleziony URL (string), lub None gdy żaden nie pasuje.
"""
from __future__ import annotations

import re

_OG_IMAGE_RE = re.compile(
    r'<meta\s+property="og:image"\s+content="([^"]+)"', re.IGNORECASE
)
_TWITTER_IMAGE_RE = re.compile(
    r'<meta\s+name="twitter:image(?::src)?"\s+content="([^"]+)"', re.IGNORECASE
)
_LD_THUMB_RE = re.compile(
    r'"thumbnailUrl"\s*:\s*(?:"([^"]+)"|\[\s*"([^"]+)")', re.IGNORECASE
)
_KVS_THUMB_RE = re.compile(
    r"setThumbUrl\(\s*['\"]([^'\"]+)['\"]", re.IGNORECASE
)
# hqporner: stare WordPress bez og:image / schema. Poster `_main.jpg` siedzi w
# atrybucie onclick (`changeImage('//cdn/_main.jpg', 'cover_<id>')`) lub w preload
# array (`preload_<id> = ['//.../_1.jpg', ...]`). Wyciągamy `_main.jpg` z CDN-a
# fastporndelivery.hqporner.com — to canonical poster, frames `_1..N.jpg` to
# hover animation.
_HQPORNER_THUMB_RE = re.compile(
    r"['\"](//[a-z0-9.\-]*hqporner[a-z0-9.\-]*/imgs/[^'\"]+_main\.jpg)['\"]",
    re.IGNORECASE,
)


def extract_thumbnail_url(html: str) -> str | None:
    """Zwraca pierwszą znalezioną miniaturkę URL lub None.

    Kolejność: OG → Twitter → LD-JSON → KVS html5player. og:image jest
    najpopularniejszy (większość WordPress + KVS-based tubes); pozostałe to
    fallback dla niszowych templatek.
    """
    if not html:
        return None

    if (m := _OG_IMAGE_RE.search(html)):
        url = m.group(1).strip()
        if url and not url.startswith("data:"):
            return url

    if (m := _TWITTER_IMAGE_RE.search(html)):
        url = m.group(1).strip()
        if url and not url.startswith("data:"):
            return url

    if (m := _LD_THUMB_RE.search(html)):
        url = (m.group(1) or m.group(2) or "").strip()
        if url:
            return url

    if (m := _KVS_THUMB_RE.search(html)):
        url = m.group(1).strip()
        if url:
            return url

    if (m := _HQPORNER_THUMB_RE.search(html)):
        url = m.group(1).strip()
        if url:
            # Protocol-relative `//cdn/path` → https.
            return "https:" + url if url.startswith("//") else url

    return None