"""Industry taxonomy file + deterministic label → slug mapping."""
from __future__ import annotations

import pytest

from companyatlas.registry.industries import (
    industry_index,
    is_valid_slug,
    load_industries,
    map_industry,
    map_label,
    map_sic,
    top_level_of,
    top_level_slugs,
)

EXPECTED = {
    "software industry": ["software"],
    "Software": ["software"],
    "banking": ["banking"],
    "bank": ["banking"],
    "video game industry": ["gaming"],
    "automotive industry": ["automotive"],
    "insurance": ["insurance"],
    "pharmaceutical industry": ["pharmaceuticals"],
    "semiconductor industry": ["semiconductors"],
    "e-commerce": ["e-commerce"],
    "retail": ["retail"],
    "telecommunications industry": ["telecommunications"],
    "airline": ["airlines"],
    "film industry": ["entertainment"],
    "financial services": ["financial-services"],
    "oil and gas industry": ["oil-gas"],
    "renewable energy": ["renewables"],
    "biotechnology": ["biotechnology"],
    "hospitality industry": ["hospitality"],
    "aerospace manufacturer": ["aerospace-defense"],
    "arms industry": ["defense"],
    "food industry": ["food-beverage"],
    "fashion": ["apparel"],
    "mass media": ["media"],
    "real estate": ["real-estate"],
    "construction industry": ["construction"],
    "steel industry": ["mining"],
    "chemical industry": ["chemicals"],
    "cloud computing": ["cloud-infrastructure"],
    "artificial intelligence": ["artificial-intelligence"],
    "payment system": ["payments"],
    "asset management": ["asset-management"],
    "management consulting": ["consulting"],
    "agriculture": ["agriculture"],
    "robotics": ["robotics"],
    "machinery industry": ["industrial-machinery"],
    "public utility": ["utilities"],
    "glass industry": ["materials"],
    "health care industry": ["healthcare"],
    "medical device": ["medical-devices"],
    "toy industry": ["consumer-goods"],
    "logistics": ["logistics"],
    "shipping": ["shipping"],
    "education": ["education"],
    "tourism": ["travel"],
    "think tank": ["education"],          # longest keyword wins over "tank" (defense)
    "cybersecurity": ["cybersecurity"],
    "social media": ["internet"],
}


@pytest.mark.parametrize(("label", "slugs"), sorted(EXPECTED.items()))
def test_map_label(label: str, slugs: list[str]) -> None:
    assert map_industry(label) == slugs


def test_unknown_and_empty() -> None:
    assert map_industry("") == []
    assert map_industry(None) == []
    assert map_industry("zzz nothing here") == []
    assert map_label("post-production") == []      # hyphenated words are not split into keywords


def test_multi_label_order_and_dedupe() -> None:
    assert map_industry(["banking", "insurance", "banking", "software industry"]) == ["banking", "insurance", "software"]
    assert map_industry(["a", "b", "c", "d", "e"] + ["banking"] * 3, limit=1) == ["banking"]


def test_taxonomy_integrity() -> None:
    inds = load_industries()
    slugs = [i.slug for i in inds]
    assert len(slugs) == len(set(slugs)) >= 45
    for i in inds:
        assert i.name and i.description
        assert len(i.keywords) >= 3, i.slug
        assert i.parent is None or i.parent in slugs
        assert i.parent is None or industry_index()[i.parent].parent is None, "max two levels"
    required = {"technology", "software", "artificial-intelligence", "semiconductors", "cloud-infrastructure", "cybersecurity", "internet",
                "e-commerce", "fintech", "payments", "banking", "insurance", "asset-management", "real-estate", "construction", "retail",
                "consumer-goods", "food-beverage", "apparel", "energy", "oil-gas", "renewables", "utilities", "mining", "chemicals", "materials",
                "manufacturing", "industrial-machinery", "automotive", "aerospace-defense", "transportation", "logistics", "airlines", "shipping",
                "telecommunications", "media", "entertainment", "gaming", "healthcare", "biotechnology", "pharmaceuticals", "medical-devices",
                "hospitality", "travel", "education", "professional-services", "consulting", "agriculture", "robotics", "defense"}
    assert required <= set(slugs)
    assert "technology" in top_level_slugs() and "software" not in top_level_slugs()
    assert top_level_of("software") == "technology" and top_level_of("technology") == "technology"
    assert is_valid_slug("banking") and not is_valid_slug("banks")


def test_map_sic() -> None:
    assert map_sic(7372) == "software"
    assert map_sic("3674") == "semiconductors"
    assert map_sic(6021) == "banking"
    assert map_sic(2836) == "biotechnology"
    assert map_sic(4512) == "airlines"
    assert map_sic(1311) == "oil-gas"
    assert map_sic(9999) is None
    assert map_sic(None) is None and map_sic("") is None and map_sic("abc") is None