tests/unit/test_location_core.py

"""Unit tests for cores.location — geography, language, plates."""

from __future__ import annotations

from cores.location import (
    coords_to_country, coordinate_to_country, coords_to_city_estimate,
    detect_script, script_to_regions, text_to_country_hints,
    plate_format_to_country,
    LocationEvidenceCollector, LocationEstimator,
)


class TestGeography:
    def test_usa_coordinates(self):
        # New York
        assert coords_to_country(40.7, -74.0) == "United States"

    def test_france_coordinates(self):
        # Paris
        assert coords_to_country(48.85, 2.35) == "France"

    def test_japan_coordinates(self):
        # Tokyo
        assert coords_to_country(35.68, 139.76) == "Japan"

    def test_ocean_coordinates_returns_none(self):
        # Middle of the Pacific
        assert coords_to_country(0.0, -160.0) is None

    def test_coordinate_to_country_confidence(self):
        # Center of France — high confidence
        result = coordinate_to_country(47.0, 2.0)
        assert result["country"] == "France"
        assert result["confidence"] > 0.5

    def test_city_estimate_near_major_city(self):
        result = coords_to_city_estimate(40.7, -74.0)  # near NYC
        assert result is not None
        assert result["city"] == "New York"

    def test_city_estimate_far_from_cities(self):
        # Middle of ocean
        result = coords_to_city_estimate(0.0, -160.0)
        assert result is None


class TestLanguageDetection:
    def test_latin_script(self):
        assert detect_script("Hello World") == "latin"

    def test_cyrillic_script(self):
        assert detect_script("Привет мир") == "cyrillic"

    def test_arabic_script(self):
        assert detect_script("مرحبا") == "arabic"

    def test_cjk_script(self):
        assert detect_script("你好世界") == "cjk"

    def test_hangul_script(self):
        assert detect_script("안녕하세요") == "hangul"

    def test_thai_script(self):
        assert detect_script("สวัสดี") == "thai"

    def test_empty_text(self):
        assert detect_script("") == "unknown"

    def test_script_to_regions_cyrillic(self):
        regions = script_to_regions("cyrillic")
        countries = [r["country"] for r in regions]
        assert "Russia" in countries

    def test_text_to_country_hints_keyword(self):
        hints = text_to_country_hints("I visited London last summer")
        countries = [h["country"] for h in hints]
        assert "United Kingdom" in countries


class TestPlateFormats:
    def test_french_plate(self):
        result = plate_format_to_country("AB123CD")
        assert result is not None
        assert result["country"] == "France"

    def test_uk_plate(self):
        result = plate_format_to_country("AB12CDE")
        assert result is not None
        assert result["country"] == "United Kingdom"

    def test_spanish_plate(self):
        result = plate_format_to_country("1234ABC")
        assert result is not None
        assert result["country"] == "Spain"

    def test_russian_plate(self):
        result = plate_format_to_country("A123BC77")
        assert result is not None
        assert result["country"] == "Russia"

    def test_empty_plate(self):
        assert plate_format_to_country("") is None


class TestLocationEstimator:
    def test_gps_only(self):
        collector = LocationEvidenceCollector()
        collector.add_gps(48.85, 2.35)  # Paris
        estimator = LocationEstimator()
        result = estimator.estimate(collector.evidence, gps={"lat": 48.85, "lon": 2.35})
        assert len(result.candidate_countries) > 0
        assert result.candidate_countries[0]["country"] == "France"
        assert result.candidate_countries[0]["confidence"] > 0.5

    def test_ocr_text_evidence(self):
        collector = LocationEvidenceCollector()
        collector.add_ocr_text("Привет мир")  # Russian
        estimator = LocationEstimator()
        result = estimator.estimate(collector.evidence)
        countries = [c["country"] for c in result.candidate_countries]
        assert "Russia" in countries

    def test_no_evidence(self):
        collector = LocationEvidenceCollector()
        estimator = LocationEstimator()
        result = estimator.estimate(collector.evidence)
        assert result.candidate_countries == []
        assert result.overall_confidence == 0.0

    def test_conflicting_evidence(self):
        collector = LocationEvidenceCollector()
        # GPS says France
        collector.add_gps(48.85, 2.35)
        # OCR says Russia (conflict)
        collector.add_ocr_text("Привет мир")
        estimator = LocationEstimator()
        result = estimator.estimate(collector.evidence, gps={"lat": 48.85, "lon": 2.35})
        # GPS should win but conflict is noted
        assert len(result.candidate_countries) > 0