Skip to content

TESTING.md

Latest commit

 

History

History
132 lines (102 loc) · 3.66 KB

TESTING.md

File metadata and controls

132 lines (102 loc) · 3.66 KB

Testing Guide for semantic_corpus

This document explains how to run the tests for the semantic_corpus project, which uses live APIs and real downloads.

Test Categories

1. Unit Tests (No Network Required)

These tests don't require network access and run quickly:

pytest tests/test_corpus_core.py tests/test_metadata_processing.py -m "not live_api"

2. Live API Tests (Network Required)

These tests interact with real repositories and download actual papers:

# Run all live API tests
pytest -m "live_api"

# Run only Europe PMC tests
pytest tests/test_repository_interface.py::TestEuropePMCRepository -m "live_api"

# Run only arXiv tests
pytest tests/test_repository_interface.py::TestArxivRepository -m "live_api"

3. Integration Tests (Full Workflow)

These tests demonstrate complete workflows with real data:

pytest tests/test_integration_live.py

4. CLI Tests (Live APIs)

These tests verify the command-line interface with real repositories:

pytest tests/test_cli.py -m "live_api"

Running All Tests

With Live APIs (Recommended)

# Run all tests including live API tests
pytest

# Run with verbose output
pytest -v

# Run with coverage
pytest --cov=semantic_corpus

Without Live APIs (Fast, but limited)

# Skip live API tests for faster execution
pytest -m "not live_api"

Test Configuration

Environment Variables

  • SEMANTIC_CORPUS_TEST_TIMEOUT: Set timeout for API calls (default: 30 seconds)
  • SEMANTIC_CORPUS_TEST_LIMIT: Set maximum number of papers to download (default: 3)

Test Markers

  • @pytest.mark.live_api: Tests that use live APIs
  • @pytest.mark.network: Tests that require network access
  • @pytest.mark.integration: Integration tests
  • @pytest.mark.slow: Tests that take longer to run

What the Live API Tests Do

Repository Tests

  1. Search Tests: Query real repositories (Europe PMC, arXiv) for papers
  2. Metadata Tests: Retrieve actual paper metadata
  3. Download Tests: Download real papers (XML, PDF) to temporary directories

Integration Tests

  1. Full Workflow: Search → Download → Add to Corpus → Search in Corpus
  2. Statistics: Test corpus statistics with real downloaded papers
  3. Error Handling: Test error scenarios with live APIs

CLI Tests

  1. Search Command: Test semantic_corpus search with real repositories
  2. Download Command: Test semantic_corpus download with real papers
  3. Config File: Test configuration file support

Expected Test Results

Live API Tests Should:

  • ✅ Successfully search for papers using real queries
  • ✅ Download actual papers (XML/PDF files)
  • ✅ Verify downloaded files exist and have content
  • ✅ Add real papers to corpus and retrieve them
  • ✅ Perform searches within the corpus

Test Data

  • Query: "climate change" (reliable, returns results)
  • Limit: 2-3 papers per test (fast, sufficient for testing)
  • Formats: XML for Europe PMC, PDF for arXiv
  • Categories: cs.AI for arXiv (reliable results)

Troubleshooting

Network Issues

If tests fail due to network issues:

# Run without live API tests
pytest -m "not live_api"

API Rate Limits

If you hit rate limits:

# Run tests with delays
pytest --tb=short -x

Missing Dependencies

Install required packages:

pip install -e ".[dev]"

Test Output

Successful live API tests will show:

  • Real paper titles and abstracts
  • Downloaded file paths and sizes
  • Corpus statistics with actual data
  • Search results from real repositories

This demonstrates that the semantic_corpus system works with real scientific data and can be used for actual research workflows.