Evret Test Suite

This directory covers the core behavior of Evret across metrics, evaluation, retrievers, framework adapters, judges, and Docker-backed integrations.

What Is Covered

Area	Path	Coverage
Metrics	`tests/metrics/`	Hit Rate, Recall, Precision, MRR, NDCG, and Average Precision scoring behavior.
Evaluation	`tests/evaluation/`	Dataset loading, evaluator orchestration, result summaries, exports, and default relevance matching.
Judges	`tests/judges/`	Token overlap judging and LLM provider wiring with mocked clients.
Retrievers	`tests/retrievers/`	Base retriever validation plus Qdrant, Chroma, Milvus, and Weaviate adapter behavior with mocked clients.
Framework integrations	`tests/integrations/`	LangChain and LlamaIndex adapter behavior when optional packages are installed.
Docker integrations	`tests/integration/`	End-to-end Qdrant and Chroma retrieval against real containers.
Shared utilities	`tests/test_utils.py`	Name, k-value, and dataset path validation helpers.

For the default test suite:

uv pip install -e ".[dev]"

For all optional integration coverage:

uv pip install -e ".[integration]"

Run the default suite:

pytest

Run a focused area:

pytest tests/metrics
pytest tests/evaluation
pytest tests/retrievers/test_qdrant.py

Run Docker-backed integration tests:

EVRET_RUN_INTEGRATION=1 pytest -m integration

The integration tests require Docker to be running. Without EVRET_RUN_INTEGRATION=1, they are skipped.

Run everything except Docker-backed tests:

pytest -m "not integration"