from __future__ import annotations

import asyncio

import pytest
from conftest import FakeFetcher, fixture_bytes, fixture_text

from companyatlas.services import discovery as D
from companyatlas.taxonomy import OnboardingStatus, Surface

BASE = "https://www.acme-cloud.example"
COMPANY = {"id": "co_test", "slug": "ztest-acme", "display_name": "Acme Cloud", "canonical_domain": "acme-cloud.example", "website": BASE + "/", "tier": 2, "importance": 0.5}


def _fake_site(ff: FakeFetcher) -> None:
    ff.add(BASE + "/", fixture_text("generic_html", "homepage.html"))
    ff.add(BASE + "/robots.txt", "User-agent: *\nAllow: /\nSitemap: https://www.acme-cloud.example/sitemap.xml\n", content_type="text/plain")
    ff.add(BASE + "/sitemap.xml", fixture_bytes("sitemap", "sitemap_index.xml"), content_type="application/xml")
    ff.add(BASE + "/sitemap-pages.xml", fixture_bytes("sitemap", "sitemap_pages.xml"), content_type="application/xml")
    ff.add(BASE + "/sitemap-blog.xml", "<urlset></urlset>", content_type="application/xml")
    ff.add(BASE + "/careers", fixture_text("generic_html", "careers.html"))
    ff.add(BASE + "/pricing", fixture_text("generic_html", "pricing.html"))
    ff.add("https://boards-api.greenhouse.io/v1/boards/acmecloud/jobs?content=false", fixture_bytes("greenhouse", "stripe_jobs.json"), content_type="application/json")


@pytest.fixture(autouse=True)
def _no_dns(monkeypatch: pytest.MonkeyPatch) -> None:
    async def never(host: str) -> bool:
        return False
    monkeypatch.setattr(D, "_resolves", never)


def test_discover_company_dry_run(fake_fetcher: FakeFetcher) -> None:
    _fake_site(fake_fetcher)
    res = asyncio.run(D.discover_company(COMPANY, fetcher=fake_fetcher, dry_run=True))  # type: ignore[arg-type]
    assert res.status == OnboardingStatus.ACTIVE, res.error
    surfaces = {s["surface"]: s for s in res.sensors}
    assert {"homepage", "pricing", "careers", "jobs_board", "newsroom", "leadership", "locations", "sitemap", "legal_terms", "legal_privacy", "feed"} <= set(surfaces)
    gh = surfaces["jobs_board"]
    assert gh["connector_id"] == "greenhouse-v1" and gh["config"]["token"] == "acmecloud" and gh["config"].get("verified_job_count") == 20
    assert surfaces["careers"]["url"] == BASE + "/careers" and surfaces["careers"]["connector_id"] == "generic-html-v1"
    assert surfaces["sitemap"]["connector_id"] == "sitemap-v1"
    assert surfaces["feed"]["connector_id"] == "feed-v1"
    # tier 2 → ×0.75 on the surface base interval, clamped
    assert surfaces["pricing"]["base_interval_s"] == int(12 * 3600 * 0.75)
    assert surfaces["homepage"]["tier"] in ("C", "D") and 0 < surfaces["pricing"]["quality_score"] <= 100
    assert all(s["status"] == "pending" for s in res.sensors)
    assert len(res.sensors) <= 40 and len({s["canonical_url"] for s in res.sensors}) == len(res.sensors)
    # bounded politeness: no more than ~25 requests for a whole company
    assert res.requests <= 25
    assert all(c.startswith(("https://www.acme-cloud.example", "https://boards-api.greenhouse.io")) for c in fake_fetcher.calls)


def test_discover_no_website(fake_fetcher: FakeFetcher) -> None:
    from companyatlas.fetch import FetchError
    from companyatlas.taxonomy import FailureClass

    for u in (BASE + "/", "https://www.acme-cloud.example/", "https://acme-cloud.example/", "http://www.acme-cloud.example/", "http://acme-cloud.example/"):
        fake_fetcher.add_error(u, FetchError("dns", url=u, failure=FailureClass.DNS))
    res = asyncio.run(D.discover_company(COMPANY, fetcher=fake_fetcher, dry_run=True))  # type: ignore[arg-type]
    assert res.status == OnboardingStatus.NO_WEBSITE and not res.sensors


def test_redirect_to_other_domain_is_recorded(fake_fetcher: FakeFetcher) -> None:
    fake_fetcher.add(BASE + "/", fixture_text("generic_html", "homepage.html"), final_url="https://www.acme-group.example/")
    fake_fetcher.add("https://www.acme-group.example/robots.txt", "", content_type="text/plain")
    res = asyncio.run(D.discover_company(COMPANY, fetcher=fake_fetcher, dry_run=True))  # type: ignore[arg-type]
    assert res.redirect_domain == "acme-group.example" and res.canonical_domain == "acme-group.example"
    assert any("redirects" in n for n in res.notes)


def test_select_sensors_caps_and_prefers_best_candidate() -> None:
    from datetime import UTC, datetime

    cands = [D.Candidate(url=BASE + "/pricing", surface=str(Surface.PRICING), confidence=0.7, method="sitemap"),
             D.Candidate(url=BASE + "/plans", surface=str(Surface.PRICING), confidence=0.95, method="nav", verified=True),
             D.Candidate(url=BASE + "/x", surface=str(Surface.OTHER), confidence=0.3, method="nav")]
    rows = D.select_sensors(cands, company=COMPANY, canonical_domain="acme-cloud.example", now=datetime.now(UTC))
    assert len(rows) == 1 and rows[0]["url"] == BASE + "/plans" and rows[0]["discovery_method"] == "nav"