tests/test_rag.py

"""
Tests for RAG service module.
"""

import pytest
from pathlib import Path
from unittest.mock import patch, MagicMock

from app.core.rag import (
    RAGService,
    CORPUS_DIR,
    VECTOR_STORE_PATH,
)


class TestRAGServiceInit:
    """Tests for RAGService initialization."""

    @patch("app.core.rag.get_embeddings_model")
    def test_creates_embeddings(self, mock_embeddings):
        """RAGService should initialize embeddings."""
        mock_embeddings.return_value = MagicMock()

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()

        assert mock_embeddings.called
        assert service.embeddings is not None

    @patch("app.core.rag.get_embeddings_model")
    @patch("app.core.rag.os.getenv")
    def test_fallback_without_pinecone(self, mock_getenv, mock_embeddings):
        """Should use fallback when Pinecone not configured."""
        mock_embeddings.return_value = MagicMock()
        mock_getenv.return_value = None  # No Pinecone config

        with patch.object(RAGService, "_init_fallback") as mock_fallback:
            service = RAGService()

        # Should attempt fallback
        mock_fallback.assert_called_once()


class TestRAGServicePaths:
    """Tests for RAG service path constants."""

    def test_corpus_dir_defined(self):
        """CORPUS_DIR should be defined."""
        assert CORPUS_DIR is not None
        assert isinstance(CORPUS_DIR, Path)

    def test_vector_store_path_defined(self):
        """VECTOR_STORE_PATH should be defined."""
        assert VECTOR_STORE_PATH is not None
        assert isinstance(VECTOR_STORE_PATH, Path)


class TestRAGServiceRetrieve:
    """Tests for RAGService retrieve functionality."""

    @patch("app.core.rag.get_embeddings_model")
    def test_retrieve_no_store(self, mock_embeddings):
        """Retrieve should handle missing vector store gracefully."""
        mock_embeddings.return_value = MagicMock()

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()
            service.vector_store = None

        result = service.retrieve("test query")
        assert result == []

    @patch("app.core.rag.get_embeddings_model")
    def test_format_docs_empty(self, mock_embeddings):
        """format_docs should handle empty list."""
        mock_embeddings.return_value = MagicMock()

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()

        result = service.format_docs([])
        assert "No relevant context" in result


class TestRAGServiceFormatDocs:
    """Tests for document formatting."""

    @patch("app.core.rag.get_embeddings_model")
    def test_format_single_doc(self, mock_embeddings):
        """Should format single document correctly."""
        from langchain_core.documents import Document

        mock_embeddings.return_value = MagicMock()

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()

        doc = Document(
            page_content="Test content",
            metadata={"source": "test.md"},
        )
        result = service.format_docs([doc])

        assert "Test content" in result
        assert "test.md" in result

    @patch("app.core.rag.get_embeddings_model")
    def test_format_multiple_docs(self, mock_embeddings):
        """Should format multiple documents."""
        from langchain_core.documents import Document

        mock_embeddings.return_value = MagicMock()

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()

        docs = [
            Document(page_content="Content 1", metadata={"source": "doc1.md"}),
            Document(page_content="Content 2", metadata={"source": "doc2.md"}),
        ]
        result = service.format_docs(docs)

        assert "Content 1" in result
        assert "Content 2" in result
        assert "doc1.md" in result
        assert "doc2.md" in result


class TestRAGServiceRetriever:
    """Tests for retriever interface."""

    @patch("app.core.rag.get_embeddings_model")
    def test_get_retriever_raises_without_store(self, mock_embeddings):
        """get_retriever should raise if no vector store."""
        mock_embeddings.return_value = MagicMock()

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()
            service.vector_store = None

        with pytest.raises(RuntimeError, match="not initialized"):
            service.get_retriever()

    @patch("app.core.rag.get_embeddings_model")
    def test_get_retriever_with_store(self, mock_embeddings):
        """get_retriever should return retriever when store exists."""
        mock_embeddings.return_value = MagicMock()
        mock_store = MagicMock()
        mock_retriever = MagicMock()
        mock_store.as_retriever.return_value = mock_retriever

        with patch.object(RAGService, "_initialize_vector_store"):
            service = RAGService()
            service.vector_store = mock_store

        retriever = service.get_retriever(k=5)

        mock_store.as_retriever.assert_called_once()
        assert retriever == mock_retriever