"""
Subsector ids the news classifier is allowed to use for `business_types`.
Sourced from data/sectors.ts via scripts/export-subsectors.mjs — re-run that
script and this file picks up the change automatically, no code edit needed.
See ADR-0015 for why subsector-level (not sector or niche) was chosen.
"""

import json
from pathlib import Path

_PATH = Path(__file__).resolve().parent / "subsectors.json"
SUBSECTORS: list[dict] = json.loads(_PATH.read_text(encoding="utf-8"))
VALID_IDS: set[str] = {s["id"] for s in SUBSECTORS}


def labels_for_prompt() -> str:
    """'id — label' lines, for embedding the allowed list in the classifier prompt."""
    return "\n".join(f"- {s['id']} — {s['label']}" for s in SUBSECTORS)


def valid_subset(ids: list[str]) -> list[str]:
    """Drops any id the model hallucinated that isn't in our taxonomy."""
    return [i for i in ids if i in VALID_IDS]