tests/test_cache.py

"""
Testes para sistema de cache de embeddings.
"""

import pytest
import tempfile
import shutil
from pathlib import Path
from src.cache import EmbeddingCache


class TestEmbeddingCache:
    """Testes para classe EmbeddingCache."""

    @pytest.fixture
    def cache_dir(self):
        """Cria diretorio temporario para cache."""
        temp_dir = tempfile.mkdtemp()
        yield temp_dir
        shutil.rmtree(temp_dir)

    @pytest.fixture
    def cache(self, cache_dir):
        """Instancia de EmbeddingCache."""
        return EmbeddingCache(cache_dir=cache_dir)

    def test_cache_initialization(self, cache_dir):
        """Testa inicializacao do cache."""
        cache = EmbeddingCache(cache_dir=cache_dir)
        assert cache.cache_dir == Path(cache_dir)
        assert cache.cache_dir.exists()

    def test_get_cache_key(self, cache):
        """Testa geracao de chave de cache."""
        text = "Texto de teste"
        model_id = "model-123"

        key1 = cache._get_cache_key(text, model_id)
        key2 = cache._get_cache_key(text, model_id)

        assert key1 == key2
        assert len(key1) == 64  # SHA256 hex digest

    def test_get_cache_key_different_inputs(self, cache):
        """Testa que inputs diferentes geram keys diferentes."""
        key1 = cache._get_cache_key("texto1", "model1")
        key2 = cache._get_cache_key("texto2", "model1")
        key3 = cache._get_cache_key("texto1", "model2")

        assert key1 != key2
        assert key1 != key3
        assert key2 != key3

    def test_save_and_load_embedding(self, cache):
        """Testa salvar e carregar embedding."""
        text = "Texto de teste"
        model_id = "model-123"
        embedding = [0.1, 0.2, 0.3, 0.4, 0.5]

        # Salvar
        cache.save(text, model_id, embedding)

        # Carregar
        loaded = cache.load(text, model_id)

        assert loaded is not None
        assert len(loaded) == len(embedding)
        assert all(abs(a - b) < 1e-6 for a, b in zip(embedding, loaded))

    def test_load_nonexistent(self, cache):
        """Testa carregar embedding que nao existe."""
        result = cache.load("texto inexistente", "model-123")
        assert result is None

    def test_has_cache(self, cache):
        """Testa verificacao de existencia no cache."""
        text = "Texto de teste"
        model_id = "model-123"
        embedding = [0.1, 0.2, 0.3]

        assert not cache.has(text, model_id)

        cache.save(text, model_id, embedding)

        assert cache.has(text, model_id)

    def test_clear_cache(self, cache):
        """Testa limpeza do cache."""
        # Adicionar varios embeddings
        for i in range(5):
            cache.save(f"texto{i}", "model", [float(i)])

        # Verificar que existem
        assert cache.has("texto0", "model")
        assert cache.has("texto4", "model")

        # Limpar
        cache.clear()

        # Verificar que foram removidos
        assert not cache.has("texto0", "model")
        assert not cache.has("texto4", "model")

    def test_get_cache_stats(self, cache):
        """Testa estatisticas do cache."""
        # Cache vazio
        stats = cache.get_stats()
        assert stats['num_files'] == 0
        assert stats['total_size_bytes'] == 0

        # Adicionar embeddings
        for i in range(3):
            cache.save(f"texto{i}", "model", [float(i)] * 100)

        # Verificar estatisticas
        stats = cache.get_stats()
        assert stats['num_files'] == 3
        assert stats['total_size_bytes'] > 0

    def test_cache_with_large_embedding(self, cache):
        """Testa cache com embedding grande."""
        text = "Texto longo"
        model_id = "model-large"
        embedding = [float(i) for i in range(1000)]

        cache.save(text, model_id, embedding)
        loaded = cache.load(text, model_id)

        assert loaded is not None
        assert len(loaded) == 1000

    def test_cache_persistence(self, cache_dir):
        """Testa persistencia do cache entre instancias."""
        text = "Texto persistente"
        model_id = "model-123"
        embedding = [0.5, 0.6, 0.7]

        # Primeira instancia
        cache1 = EmbeddingCache(cache_dir=cache_dir)
        cache1.save(text, model_id, embedding)

        # Segunda instancia
        cache2 = EmbeddingCache(cache_dir=cache_dir)
        loaded = cache2.load(text, model_id)

        assert loaded is not None
        assert len(loaded) == len(embedding)


if __name__ == "__main__":
    pytest.main([__file__, "-v"])