"""URL-hash dedup (ADR-0015: dedup runs before classification, so we never
spend a Groq call on something we're about to discard as already-seen)."""

import hashlib


def url_hash(url: str) -> str:
    # Trailing slashes and query strings are common no-op variants of the same
    # article URL — strip them so the same piece of content doesn't get
    # rescraped as "new" on a second poll just because of a tracking param.
    normalized = url.strip().rstrip("/").split("?", 1)[0].split("#", 1)[0]
    return hashlib.sha256(normalized.encode("utf-8")).hexdigest()