ohm_streaming/tests/test_metadata_enrichment.py

"""
Tests for metadata enrichment with Kitsu API fallback.
"""
import pytest
from unittest.mock import AsyncMock, MagicMock, patch
from datetime import datetime, timedelta

from app.metadata_enrichment import MetadataEnricher
from app.models import AnimeMetadata


@pytest.fixture
async def enricher(temp_dir):
    """Create a MetadataEnricher instance with temp cache dir."""
    enricher = MetadataEnricher(cache_dir=temp_dir)
    yield enricher
    await enricher.close()


@pytest.fixture
def mock_kitsu_api():
    """Mock Kitsu API responses in raw Kitsu format."""
    mock_data = {
        'title': 'Naruto',
        'title_japanese': 'ナルト',
        'title_english': 'Naruto',
        'synopsis': 'A test synopsis from Kitsu',
        'genres': ['Action', 'Adventure'],
        'score': 8.5,
        'year': 2002,
        'episodes': 220,
        'status': 'Finished Airing',
        'images': {
            'jpg': {
                'large_image_url': 'https://kitsu.io/naruto-poster.jpg',
                'image_url': 'https://kitsu.io/naruto-poster-small.jpg'
            },
            'webp': {
                'large_image_url': 'https://kitsu.io/naruto-banner.jpg'
            }
        }
    }
    return mock_data


@pytest.fixture
def mock_kitsu_api_raw():
    """Mock raw Kitsu API response format."""
    return {
        'mal_id': 123,
        'title': 'Naruto',
        'title_japanese': 'ナルト',
        'title_english': 'Naruto',
        'episodes': 220,
        'status': 'Finished Airing',
        'score': 8.5,
        'synopsis': 'A test synopsis from Kitsu',
        'genres': ['Action', 'Adventure'],
        'images': {
            'jpg': {
                'image_url': 'https://kitsu.io/naruto-poster-small.jpg',
                'large_image_url': 'https://kitsu.io/naruto-poster.jpg'
            },
            'webp': {
                'image_url': 'https://kitsu.io/naruto-poster-small.webp',
                'large_image_url': 'https://kitsu.io/naruto-banner.jpg'
            }
        },
        'url': 'https://kitsu.io/anime/123',
        'subtype': 'TV',
        'year': 2002
    }


class TestMetadataEnricher:
    """Test MetadataEnricher functionality."""

    def test_init_creates_cache_dir(self, enricher, temp_dir):
        """Test that enricher creates cache directory."""
        assert enricher.cache_dir == temp_dir
        assert enricher.cache_file == temp_dir / "metadata_cache.json"

    def test_get_cache_key(self, enricher):
        """Test cache key generation."""
        key1 = enricher._get_cache_key("Naruto", "https://example.com/naruto")
        key2 = enricher._get_cache_key("Naruto", "https://example.com/naruto")
        key3 = enricher._get_cache_key("Naruto", "https://example.com/sasuke")

        # Same inputs should produce same key
        assert key1 == key2

        # Different URL should produce different key
        assert key1 != key3

    def test_get_missing_fields(self, enricher):
        """Test identification of missing fields."""
        # Complete metadata
        complete = {
            'synopsis': 'Test synopsis',
            'genres': ['Action'],
            'rating': '8.5/10',
            'release_year': 2020,
            'studio': 'Studio Pierrot',
            'poster_image': 'https://example.com/poster.jpg',
            'banner_image': 'https://example.com/banner.jpg',
            'total_episodes': 12,
            'status': 'Completed',
            'alternative_titles': ['Japanese Title']  # Now required for completeness
        }

        missing = enricher._get_missing_fields(complete)
        assert len(missing) == 0

        # Incomplete metadata
        incomplete = {
            'synopsis': 'Test synopsis',
            'genres': []  # Empty list counts as missing
        }

        missing = enricher._get_missing_fields(incomplete)
        assert 'rating' in missing
        assert 'release_year' in missing
        # Note: studio is not in KITSU_FIELDS, so it won't be detected as missing
        assert 'status' in missing
        assert 'genres' in missing  # Empty list is considered missing
        assert len(missing) >= 4

    def test_convert_kitsu_to_metadata(self, enricher, mock_kitsu_api):
        """Test conversion of Kitsu API response to metadata format."""
        metadata = enricher._convert_kitsu_to_metadata(mock_kitsu_api)

        assert metadata['synopsis'] == 'A test synopsis from Kitsu'
        assert metadata['genres'] == ['Action', 'Adventure']
        assert metadata['rating'] == '8.5/10'
        assert metadata['release_year'] == 2002
        assert metadata['poster_image'] == 'https://kitsu.io/naruto-poster.jpg'
        assert metadata['banner_image'] == 'https://kitsu.io/naruto-banner.jpg'
        assert metadata['total_episodes'] == 220
        assert metadata['status'] == 'Completed'
        assert 'ナルト' in metadata['alternative_titles']
        assert 'Naruto' in metadata['alternative_titles']

    def test_convert_kitsu_status_translation(self, enricher):
        """Test Kitsu status translation."""
        test_cases = [
            ('Airing', 'Ongoing'),
            ('Finished Airing', 'Completed'),
            ('To Be Aired', 'Upcoming'),
        ]

        for kitsu_status, expected_status in test_cases:
            metadata = enricher._convert_kitsu_to_metadata({
                'status': kitsu_status
            })
            assert metadata['status'] == expected_status

    def test_merge_metadata_prefer_provider(self, enricher, mock_kitsu_api):
        """Test that provider metadata takes priority over Kitsu."""
        provider_meta = {
            'synopsis': 'Provider synopsis (better)',
            'genres': ['Action'],
            'rating': '9.0/10',  # Different from Kitsu
            'release_year': 2002,
            'studio': 'Studio Pierrot',  # Not in Kitsu
        }

        kitsu_meta = enricher._convert_kitsu_to_metadata(mock_kitsu_api)

        merged = enricher._merge_metadata(provider_meta, kitsu_meta)

        # Provider data should be preserved
        assert merged['synopsis'] == 'Provider synopsis (better)'
        assert merged['rating'] == '9.0/10'
        assert merged['studio'] == 'Studio Pierrot'

        # Kitsu data should fill gaps
        assert merged['total_episodes'] == 220
        assert merged['status'] == 'Completed'

    def test_calculate_quality_score(self, enricher):
        """Test metadata quality score calculation."""
        # Complete metadata should have high score
        complete = {
            'synopsis': 'A detailed synopsis of the anime with lots of information',
            'genres': ['Action', 'Adventure', 'Fantasy'],
            'rating': '8.5/10',
            'release_year': 2020,
            'studio': 'Studio Pierrot',
            'poster_image': 'https://example.com/poster.jpg',
            'banner_image': 'https://example.com/banner.jpg',
            'total_episodes': 12,
            'status': 'Completed',
            'alternative_titles': ['Japanese Title']
        }

        score = enricher._calculate_quality_score(complete)
        assert score > 0.8  # Should be high quality

        # Minimal metadata should have low score
        minimal = {
            'synopsis': 'Short',
            'genres': ['Action']
        }

        score = enricher._calculate_quality_score(minimal)
        assert score < 0.5  # Should be low quality

    @pytest.mark.asyncio
    async def test_enrich_metadata_with_kitsu_fallback(self, enricher, mock_kitsu_api_raw):
        """Test enrichment with Kitsu API fallback."""
        provider_metadata = {
            'synopsis': 'Provider synopsis',
            'genres': ['Action'],
            # Missing: rating, release_year, poster_image, etc.
        }

        # Mock the Kitsu API search to return raw format
        with patch.object(enricher.kitsu_api, 'search_anime', return_value=[mock_kitsu_api_raw]):
            result = await enricher.enrich_metadata(
                provider_metadata=provider_metadata,
                title='Naruto',
                url='https://example.com/naruto',
                use_kitsu_fallback=True
            )

            # Should have Kitsu data
            assert result.rating == '8.5/10'
            assert result.release_year == 2002
            assert result.poster_image is not None
            assert result.total_episodes == 220
            assert result.status == 'Completed'

            # Should preserve provider data
            assert result.synopsis == 'Provider synopsis'

    @pytest.mark.asyncio
    async def test_enrich_metadata_without_kitsu_fallback(self, enricher):
        """Test enrichment without Kitsu fallback."""
        provider_metadata = {
            'synopsis': 'Provider synopsis',
            'genres': ['Action'],
        }

        result = await enricher.enrich_metadata(
            provider_metadata=provider_metadata,
            title='Naruto',
            url='https://example.com/naruto',
            use_kitsu_fallback=False
        )

        # Should only have provider data
        assert result.synopsis == 'Provider synopsis'
        assert result.genres == ['Action']
        assert result.rating is None  # No Kitsu fallback
        assert result.release_year is None

    @pytest.mark.asyncio
    async def test_enrich_metadata_caching(self, enricher, mock_kitsu_api_raw):
        """Test that enriched metadata is cached."""
        provider_metadata = {
            'synopsis': 'Provider synopsis',
            'genres': ['Action'],
        }

        with patch.object(enricher.kitsu_api, 'search_anime', return_value=[mock_kitsu_api_raw]) as mock_search:
            # First call should fetch from Kitsu
            result1 = await enricher.enrich_metadata(
                provider_metadata=provider_metadata,
                title='Naruto',
                url='https://example.com/naruto',
                use_kitsu_fallback=True
            )
            assert mock_search.call_count == 1

            # Second call should use cache
            result2 = await enricher.enrich_metadata(
                provider_metadata=provider_metadata,
                title='Naruto',
                url='https://example.com/naruto',
                use_kitsu_fallback=True
            )
            assert mock_search.call_count == 1  # No additional call

            # Results should be identical
            assert result1.model_dump() == result2.model_dump()

    @pytest.mark.asyncio
    async def test_enrich_search_results(self, enricher, mock_kitsu_api_raw):
        """Test enrichment of multiple search results."""
        search_results = [
            {
                'title': 'Naruto',
                'url': 'https://example.com/naruto',
                'metadata': {
                    'synopsis': 'Brief synopsis',
                    'genres': ['Action']
                }
            },
            {
                'title': 'One Piece',
                'url': 'https://example.com/onepiece',
                'metadata': {
                    'synopsis': 'Another synopsis',
                    'genres': ['Adventure']
                }
            },
            {
                'title': 'No Metadata',
                'url': 'https://example.com/nometa'
            }
        ]

        with patch.object(enricher.kitsu_api, 'search_anime', return_value=[mock_kitsu_api_raw]):
            enriched = await enricher.enrich_search_results(
                results=search_results,
                use_kitsu_fallback=True
            )

            # Should enrich results with metadata
            assert len(enriched) == 3

            # First result should be enriched
            assert enriched[0]['metadata']['rating'] == '8.5/10'
            assert enriched[0]['metadata']['release_year'] == 2002

            # Second result should also be enriched
            assert enriched[1]['metadata']['rating'] == '8.5/10'

            # Third result should have no metadata field
            assert 'metadata' not in enriched[2] or enriched[2].get('metadata') is None

    @pytest.mark.asyncio
    async def test_cache_expiry(self, enricher, mock_kitsu_api_raw):
        """Test that expired cache entries are removed."""
        provider_metadata = {'synopsis': 'Test'}

        # Add an expired entry to cache
        cache_key = enricher._get_cache_key('Test', 'https://example.com/test')
        enricher._cache[cache_key] = {
            'metadata': provider_metadata,
            'cached_at': (datetime.now() - timedelta(hours=25)).isoformat()  # Expired
        }
        enricher._cache_dirty = True

        with patch.object(enricher.kitsu_api, 'search_anime', return_value=[mock_kitsu_api_raw]) as mock_search:
            # Should fetch from Kitsu since cache is expired
            result = await enricher.enrich_metadata(
                provider_metadata=provider_metadata,
                title='Test',
                url='https://example.com/test',
                use_kitsu_fallback=True
            )

            assert mock_search.call_count == 1
            assert result.rating == '8.5/10'

    @pytest.mark.asyncio
    async def test_close_saves_cache(self, enricher):
        """Test that closing the enricher saves the cache."""
        # Add something to cache
        cache_key = 'test_key'
        enricher._cache[cache_key] = {
            'metadata': {'test': 'data'},
            'cached_at': datetime.now().isoformat()
        }
        enricher._cache_dirty = True

        await enricher.close()

        # Cache file should exist
        assert enricher.cache_file.exists()

    @pytest.mark.asyncio
    async def test_fetch_from_kitsu_error_handling(self, enricher):
        """Test error handling when Kitsu API fails."""
        provider_metadata = {'synopsis': 'Test'}

        with patch.object(enricher, '_fetch_from_kitsu', side_effect=Exception("API Error")):
            result = await enricher.enrich_metadata(
                provider_metadata=provider_metadata,
                title='NonExistent Anime',
                url='https://example.com/nonexistent',
                use_kitsu_fallback=True
            )

            # Should return provider metadata despite error
            assert result.synopsis == 'Test'
            assert result.rating is None


class TestMetadataEnrichmentIntegration:
    """Integration tests for metadata enrichment."""

    @pytest.mark.asyncio
    @pytest.mark.slow
    async def test_kitsu_api_integration(self):
        """Test actual Kitsu API integration (marked as slow)."""
        enricher = MetadataEnricher()

        try:
            # Search for a well-known anime
            results = await enricher.kitsu_api.search_anime('Naruto', limit=1)

            assert len(results) > 0
            assert 'title' in results[0]
            assert 'synopsis' in results[0] or 'genres' in results[0]

        finally:
            await enricher.close()

    @pytest.mark.asyncio
    @pytest.mark.slow
    async def test_full_enrichment_flow(self):
        """Test complete enrichment flow with real data (marked as slow)."""
        enricher = MetadataEnricher()

        try:
            # Simulate provider metadata with gaps
            provider_metadata = {
                'synopsis': 'Naruto Uzumaki wants to be the best ninja.',
                'genres': ['Action'],
                # Missing many fields
            }

            result = await enricher.enrich_metadata(
                provider_metadata=provider_metadata,
                title='Naruto',
                url='https://test.com/naruto',
                use_kitsu_fallback=True
            )

            # Should have enriched data
            assert result.synopsis is not None
            assert len(result.genres) > 0

            # Kitsu might have filled some gaps
            # (We can't assert specific fields as Kitsu responses may vary)
            quality_score = result.model_dump().get('_quality_score', 0)
            assert quality_score >= 0

        finally:
            await enricher.close()