"""Tests for B22 Cross-Domain-Legal-Doc-Detector."""

from compliance.services.cross_domain_doc_check import (
    _site_origin_sld,
    _sld,
    check_cross_domain_docs,
)


class TestSld:
    def test_simple(self):
        assert _sld("www.bmw.de") == "bmw"

    def test_compound_tld(self):
        assert _sld("docs.example.co.uk") == "example"

    def test_no_www(self):
        assert _sld("elli.eco") == "elli"


class TestPrimaryDetection:
    def test_majority_wins(self):
        state = {"doc_entries": [
            {"url": "https://elli.eco/de/impressum"},
            {"url": "https://elli.eco/de/datenschutz"},
            {"url": "https://docs.logpay.de/_docs/agb.pdf"},
        ]}
        assert _site_origin_sld(state) == "elli"

    def test_auto_discovered_excluded(self):
        # discovery results don't influence primary detection
        state = {"doc_entries": [
            {"url": "https://elli.eco/de/impressum", "auto_discovered": False},
            {"url": "https://discovered.tld/foo", "auto_discovered": True},
        ]}
        assert _site_origin_sld(state) == "elli"


class TestCheck:
    def test_elli_logpay_pattern(self):
        state = {"doc_entries": [
            {"doc_type": "dse", "url": "https://www.elli.eco/de/datenschutz"},
            {"doc_type": "impressum",
             "url": "https://www.elli.eco/de/impressum"},
            {"doc_type": "agb",
             "url": "https://docs.logpay.de/_docs/de/"
                     "allgemeine_geschaeftsbedingungen_de_EM.pdf"},
        ]}
        findings = check_cross_domain_docs(state)
        assert len(findings) == 1
        f = findings[0]
        assert f["check_id"] == "CROSS-DOMAIN-DOC-001"
        assert f["severity"] == "HIGH"  # AGB is HIGH
        assert f["doc_type"] == "agb"
        assert f["site_sld"] == "elli"
        assert f["host_sld"] == "logpay"

    def test_same_subdomain_no_finding(self):
        # docs.bmw.de is same SLD as www.bmw.de — no finding
        state = {"doc_entries": [
            {"doc_type": "dse",
             "url": "https://www.bmw.de/de/datenschutz.html"},
            {"doc_type": "agb",
             "url": "https://docs.bmw.de/agb.pdf"},
        ]}
        findings = check_cross_domain_docs(state)
        assert findings == []

    def test_no_primary_no_finding(self):
        # No URLs at all
        state = {"doc_entries": []}
        assert check_cross_domain_docs(state) == []

    def test_severity_per_doc_type(self):
        state = {"doc_entries": [
            {"doc_type": "agb", "url": "https://acme.de/x"},
            {"doc_type": "dse",
             "url": "https://docs.thirdparty.com/agb"},
            {"doc_type": "impressum",
             "url": "https://www.other.com/impressum"},
        ]}
        findings = check_cross_domain_docs(state)
        sev_by_doc = {f["doc_type"]: f["severity"] for f in findings}
        # agb is on primary (acme.de) — no finding
        # dse on thirdparty.com → MEDIUM
        # impressum on other.com → INFO
        assert sev_by_doc.get("dse") == "MEDIUM"
        assert sev_by_doc.get("impressum") == "INFO"