f6081669a0
Zgloszenie 6799253d: filmy Tushy Raw bez okladek, linki dzialaja. Okladki sieci pandanetwork (i*.wp.com/pandanetwork.club/.../<numer>h.jpg) to kopie okladek Adult Empire pod tym samym numerem produktu; zrodlo oddaje 404, a pamiec wp.com wygasa (40 z 40 losowych martwych). Normalizacja nowych filmow podmienia adres na imgs1cdn.adultempire.com/products/<NN>/<numer>h.jpg. Jednorazowo na produkcji przepisane 51 900 okladek (kopia starych adresow w movies_poster_backup_20261012), probka 80 z 80 zywych z VPS. Zgodnosc numeru z filmem sprawdzona po tytulach i wzrokowo na okladkach. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
86 lines
2.9 KiB
Python
86 lines
2.9 KiB
Python
"""RawMovie → NormalizedMovie. Reuse normalize_studio/performer/tag z scenes.py."""
|
|
from __future__ import annotations
|
|
|
|
import re
|
|
from dataclasses import dataclass, field
|
|
from datetime import date
|
|
|
|
from app.connectors.base import RawMovie, RawMovieChapter, RawPlaybackSource
|
|
from app.normalize.scenes import (
|
|
NormalizedPerformer,
|
|
NormalizedStudio,
|
|
NormalizedTag,
|
|
normalize_performer,
|
|
normalize_studio,
|
|
normalize_tag,
|
|
)
|
|
from app.normalize.text import clean_description, normalize, slugify, unescape_html
|
|
|
|
|
|
# Okladki z sieci pandanetwork (i*.wp.com/pandanetwork.club/.../<numer>h.jpg) to kopie okladek
|
|
# Adult Empire pod tym samym numerem produktu. Zrodlo oddaje juz 404, a pamiec podreczna wp.com
|
|
# wygasa: 2026-10-12 40 z 40 losowych okladek martwych, zgloszenie 6799253d (Tushy Raw bez
|
|
# okladek). Oryginal z Adult Empire jest pod numerem z nazwy pliku (sprawdzone wzrokowo i po
|
|
# tytulach na probce).
|
|
_PANDA_POSTER_RE = re.compile(r"pandanetwork\.club/.*/(\d{6,8})h(?:-\d)?\.jpe?g", re.IGNORECASE)
|
|
|
|
|
|
def stable_poster_url(url: str | None) -> str | None:
|
|
if not url:
|
|
return url
|
|
m = _PANDA_POSTER_RE.search(url)
|
|
if not m:
|
|
return url
|
|
pid = m.group(1)
|
|
return f"https://imgs1cdn.adultempire.com/products/{pid[-2:]}/{pid}h.jpg"
|
|
|
|
|
|
@dataclass
|
|
class NormalizedMovie:
|
|
external_id: str
|
|
title: str
|
|
title_normalized: str
|
|
slug: str
|
|
release_year: int | None
|
|
release_date: date | None
|
|
description: str | None
|
|
duration_sec: int | None
|
|
director: str | None
|
|
country: str | None
|
|
rating: float | None
|
|
poster_url: str | None
|
|
backdrop_url: str | None
|
|
url: str | None
|
|
|
|
studio: NormalizedStudio | None = None
|
|
performers: list[NormalizedPerformer] = field(default_factory=list)
|
|
tags: list[NormalizedTag] = field(default_factory=list)
|
|
chapters: list[RawMovieChapter] = field(default_factory=list)
|
|
playback_sources: list[RawPlaybackSource] = field(default_factory=list)
|
|
cross_source_refs: dict[str, str] = field(default_factory=dict)
|
|
|
|
|
|
def normalize_movie(raw: RawMovie) -> NormalizedMovie:
|
|
title = unescape_html(raw.title)
|
|
return NormalizedMovie(
|
|
external_id=raw.external_id,
|
|
title=title,
|
|
title_normalized=normalize(title),
|
|
slug=slugify(title),
|
|
release_year=raw.release_year,
|
|
release_date=raw.release_date,
|
|
description=clean_description(raw.description),
|
|
duration_sec=raw.duration_sec,
|
|
director=raw.director,
|
|
country=raw.country,
|
|
rating=raw.rating,
|
|
poster_url=stable_poster_url(raw.poster_url),
|
|
backdrop_url=raw.backdrop_url,
|
|
url=raw.url,
|
|
studio=normalize_studio(raw.studio) if raw.studio else None,
|
|
performers=[normalize_performer(p) for p in raw.performers],
|
|
tags=[normalize_tag(t) for t in raw.tags],
|
|
chapters=list(raw.chapters),
|
|
playback_sources=list(raw.playback_sources),
|
|
cross_source_refs=dict(raw.cross_source_refs),
|
|
)
|