goon/scripts/refresh_sxyprn_thumbs.py

"""Odśwież wygasłe miniaturki sxyprn z (żywych) stron post + oznacz martwe (bug 2026-06-10).

sxyprn/trafficdeposit thumbnaile są podpisane czasowo i ROTUJĄ (token wygasa po
~tygodniach → 404), ALE żywa strona /post/<id>.html zawiera świeży poster sceny w
`og:image` / `<video poster>` (post-id w ścieżce, current timestamp). Search/listingi
NIE re-surfaceują starych postów (0 overlap), więc jedyna droga to per-post page fetch.

Trzy wyniki per źródło:
  - 'thumb' → strona żyje + ma świeży poster → UPDATE thumbnail_url
  - 'dead'  → "Post Not Found" (post skasowany na sxyprn) → mark dead_at (martwy link,
              nie tylko brak miniatury, i tak by nie zagrał)
  - 'none'  → fetch fail / brak postera ale strona żyje → zostaw bez zmian

Wznawialne (--offset), idempotent. Thumbnaile rotują → odpalać periodycznie
(_job_refresh_sxyprn_thumbs co 12h).

Użycie (kontener worker):
  python scripts/refresh_sxyprn_thumbs.py [--limit N] [--offset M]
"""
from __future__ import annotations

import re
import sys
import time

from sqlalchemy import text

from app.db import session_scope
from app.extractors import browser_get

_OG_RE = re.compile(r"og:image[\"'][^>]*content=[\"']([^\"']+)", re.IGNORECASE)
_OG_RE2 = re.compile(r"content=[\"']([^\"']+)[\"'][^>]*property=[\"']og:image", re.IGNORECASE)
_POSTER_RE = re.compile(r"<video[^>]*poster=[\"']([^\"']+)", re.IGNORECASE)


def resolve_post(page_url: str) -> tuple[str, str | None]:
    """Zwraca ('thumb', url) | ('dead', None) | ('none', None)."""
    try:
        html = browser_get(page_url, timeout=25).text
    except Exception:
        return ("none", None)
    if "Post Not Found" in html:
        return ("dead", None)
    m = _OG_RE.search(html) or _OG_RE2.search(html) or _POSTER_RE.search(html)
    if not m:
        return ("none", None)
    u = m.group(1).strip()
    if u.startswith("//"):
        u = "https:" + u
    if "trafficdeposit.com" not in u and "sxyprn" not in u:
        return ("none", None)
    return ("thumb", u)


def refresh_batch(rows: list[tuple]) -> tuple[int, int, int]:
    """rows: [(pb_id, page_url), ...]. Returns (refreshed, marked_dead, untouched)."""
    refreshed = marked_dead = untouched = 0
    for pbid, page_url in rows:
        status, val = resolve_post(page_url)
        if status == "thumb":
            with session_scope() as s:
                s.execute(
                    text("UPDATE playback_sources SET thumbnail_url=:t WHERE id=:i").bindparams(
                        t=val, i=pbid
                    )
                )
                s.commit()
            refreshed += 1
        elif status == "dead":
            with session_scope() as s:
                s.execute(
                    text(
                        "UPDATE playback_sources SET dead_at=now(), "
                        "dead_reason='sxyprn Post Not Found (refresh sweep)' WHERE id=:i"
                    ).bindparams(i=pbid)
                )
                s.commit()
            marked_dead += 1
        else:
            untouched += 1
        time.sleep(0.25)
    return refreshed, marked_dead, untouched


def main() -> None:
    limit = 1_000_000
    offset = 0
    for i, a in enumerate(sys.argv):
        if a == "--limit" and i + 1 < len(sys.argv):
            limit = int(sys.argv[i + 1])
        if a == "--offset" and i + 1 < len(sys.argv):
            offset = int(sys.argv[i + 1])

    with session_scope() as s:
        rows = s.execute(
            text(
                "SELECT id, page_url FROM playback_sources "
                "WHERE origin='tube:sxyprncom' AND dead_at IS NULL "
                "ORDER BY id OFFSET :off LIMIT :lim"
            ).bindparams(off=offset, lim=limit)
        ).all()

    print(f"sxyprn sources: {len(rows)} (offset={offset})", flush=True)
    refreshed = marked_dead = untouched = 0
    CHUNK = 200
    for i in range(0, len(rows), CHUNK):
        r, d, u = refresh_batch(rows[i : i + CHUNK])
        refreshed += r
        marked_dead += d
        untouched += u
        print(
            f"  {min(i+CHUNK, len(rows))}/{len(rows)} refreshed={refreshed} dead={marked_dead} untouched={untouched}",
            flush=True,
        )
    print(f"DONE refreshed={refreshed} dead={marked_dead} untouched={untouched}", flush=True)


if __name__ == "__main__":
    main()