Download tests/unit/test_chroma_provider.py from Amogh1221/serinity: direct link, hf CLI and curl.
- Browser
- Download file 5.59 kB
-
https://huggingface.co/spaces/Amogh1221/serinity/resolve/main/tests/unit/test_chroma_provider.py
- Command line
-
hf download hf://spaces/Amogh1221/serinity/tests/unit/test_chroma_provider.py
-
curl -L -o test_chroma_provider.py https://huggingface.co/spaces/Amogh1221/serinity/resolve/main/tests/unit/test_chroma_provider.py
5.59 kB
| import pytest | |
| from unittest.mock import MagicMock, patch | |
| from providers.chroma_provider import ChromaVectorStore | |
| class TestChromaVectorStore: | |
| """Tests for ChromaVectorStore, mocking ChromaDB, OllamaEmbeddings, and BM25.""" | |
| def test_init_empty_collection(self, MockChroma, MockEmbeddings): | |
| """If collection is empty, ensemble_retriever is set to None and no BM25 index is built.""" | |
| mock_chroma_instance = MockChroma.return_value | |
| # Mock count = 0 | |
| mock_chroma_instance._collection.count.return_value = 0 | |
| store = ChromaVectorStore( | |
| host="http://localhost:11434", | |
| embedding_model="nomic-embed-text", | |
| persist_dir="./fake_chroma_db", | |
| collection_name="test-knowledge" | |
| ) | |
| assert store.ensemble_retriever is None | |
| mock_chroma_instance.get.assert_not_called() | |
| def test_init_builds_bm25_from_scratch(self, mock_dump, mock_open, MockEnsemble, MockBM25, mock_exists, MockChroma, MockEmbeddings): | |
| """If collection has documents and no cache exists, it pulls docs from Chroma and builds BM25 cache.""" | |
| mock_chroma_instance = MockChroma.return_value | |
| mock_chroma_instance._collection.count.return_value = 10 | |
| # Mock batch fetching: first call returns 10 docs, second returns 0 | |
| mock_chroma_instance.get.side_effect = [ | |
| {"documents": ["Doc 1", "Doc 2"]}, | |
| {"documents": []} | |
| ] | |
| store = ChromaVectorStore( | |
| host="http://localhost:11434", | |
| embedding_model="nomic-embed-text", | |
| persist_dir="./fake_chroma_db", | |
| collection_name="test-knowledge" | |
| ) | |
| # Verified it pulled from Chroma | |
| assert mock_chroma_instance.get.call_count == 2 | |
| # Verified it built BM25 | |
| MockBM25.assert_called_once_with(["Doc 1", "Doc 2"]) | |
| # Verified it cached to pickle | |
| mock_open.assert_called_once() | |
| mock_dump.assert_called_once() | |
| # Verified ensemble was created | |
| MockEnsemble.assert_called_once() | |
| assert store.ensemble_retriever == MockEnsemble.return_value | |
| def test_retrieve_with_ensemble(self, MockChroma, MockEmbeddings): | |
| """Retrieve uses the ensemble retriever if it is available.""" | |
| mock_chroma_instance = MockChroma.return_value | |
| mock_chroma_instance._collection.count.return_value = 0 # Forces ensemble to None initially | |
| store = ChromaVectorStore("host", "emb", "dir", "col") | |
| # Inject fake ensemble | |
| mock_ensemble = MagicMock() | |
| fake_doc1 = MagicMock() | |
| fake_doc1.page_content = "retrieved text 1" | |
| fake_doc2 = MagicMock() | |
| fake_doc2.page_content = "retrieved text 2" | |
| mock_ensemble.invoke.return_value = [fake_doc1, fake_doc2] | |
| store.ensemble_retriever = mock_ensemble | |
| result = store.retrieve("Anxiety criteria", k=2) | |
| assert result == "retrieved text 1\n\nretrieved text 2" | |
| mock_ensemble.invoke.assert_called_once_with("Anxiety criteria") | |
| def test_retrieve_without_ensemble_falls_back_to_mmr(self, MockChroma, MockEmbeddings): | |
| """If ensemble is None, it uses ChromaDB MMR search.""" | |
| mock_chroma_instance = MockChroma.return_value | |
| mock_chroma_instance._collection.count.return_value = 0 | |
| store = ChromaVectorStore("host", "emb", "dir", "col") | |
| assert store.ensemble_retriever is None | |
| mock_chroma_instance._collection.count.return_value = 10 | |
| fake_doc1 = MagicMock() | |
| fake_doc1.page_content = "fallback text 1" | |
| mock_chroma_instance.max_marginal_relevance_search.return_value = [fake_doc1] | |
| result = store.retrieve("Anxiety", k=1) | |
| assert result == "fallback text 1" | |
| mock_chroma_instance.max_marginal_relevance_search.assert_called_once_with( | |
| "Anxiety", k=1, fetch_k=10, lambda_mult=0.7 | |
| ) | |
| def test_retrieve_handles_exception_and_falls_back(self, MockChroma, MockEmbeddings): | |
| """Retrieve falls back to similarity search on MMR exception.""" | |
| mock_chroma_instance = MockChroma.return_value | |
| mock_chroma_instance._collection.count.return_value = 0 | |
| store = ChromaVectorStore("host", "emb", "dir", "col") | |
| mock_chroma_instance._collection.count.return_value = 10 | |
| mock_chroma_instance.max_marginal_relevance_search.side_effect = Exception("MMR Failed") | |
| fake_doc = MagicMock() | |
| fake_doc.page_content = "similarity text" | |
| mock_chroma_instance.similarity_search.return_value = [fake_doc] | |
| result = store.retrieve("Anxiety", k=1) | |
| assert result == "similarity text" | |
| mock_chroma_instance.similarity_search.assert_called_once_with("Anxiety", k=1) | |