"""sxyprn.com — direct HTML scrape search results.
Sxyprn search jest oparte na `?type=videos&query=` GET endpoint który zwraca
HTML strony z linkami. Scene URL format: `https://sxyprn.com/post/.html`.
Page'owanie sxyprn niespójne — często single-page results dla query (~24 wyników).
"""
from __future__ import annotations
import re
from app.connectors.direct_scrapers._search_base import BaseSearchScraper
class SxyPrnScraper(BaseSearchScraper):
sitetag = "sxyprncom"
_search_url_template = "https://sxyprn.com/?type=videos&query={query}&page={page}"
_scene_url_re = re.compile(
r'href="(?P/post/(?P[a-z0-9]+))\.html"',
)
def _title_from_slug(self, slug: str) -> str:
# sxyprn post ID to nieczytelny hash — placeholder, title backfill przy resolve.
return f"sxyprn:{slug}"