"""Domain-level paywall hints.

We never fetch article pages, so a paywall can only be inferred from the host.
This is a curated, conservative list of hard/soft paywalls — enough to label a
card "subscription may be required" and to prefer readable stories for the lead
and for replacements. It will never be perfect; it's an honest hint, not a gate.
"""

from __future__ import annotations

from urllib.parse import urlsplit

# Host suffixes considered paywalled. Subdomains match (news.nature.com → nature.com).
PAYWALL_DOMAINS = {
    "newscientist.com",
    "nature.com",
    "nytimes.com",
    "wsj.com",
    "ft.com",
    "economist.com",
    "wired.com",
    "theatlantic.com",
    "washingtonpost.com",
    "bloomberg.com",
    "technologyreview.com",
    "newyorker.com",
    "scientificamerican.com",
    "nationalgeographic.com",
    "thetimes.co.uk",
    "telegraph.co.uk",
    "foreignpolicy.com",
    "hbr.org",
    "harpers.org",
}


def is_paywalled(url: str | None) -> bool:
    """Low-level DOMAIN rule. Keep this distinct from the source-aware decision so
    callers can tell 'domain says paywalled' from 'this source is overridden'."""
    host = urlsplit(url or "").netloc.lower()
    if host.startswith("www."):
        host = host[4:]
    return any(host == d or host.endswith("." + d) for d in PAYWALL_DOMAINS)


def is_paywalled_for_source(url: str | None, override: str | None = None) -> bool:
    """The EFFECTIVE paywall decision used for ranking/lead/badges: a per-source
    override (set in admin after inspecting the articles) wins over the domain
    rule — 'free' clears a false positive (e.g. NY Times Learning), 'paywalled'
    flags a false negative. NULL falls back to the domain rule."""
    if override == "free":
        return False
    if override == "paywalled":
        return True
    return is_paywalled(url)