Skip to content

pytest_discovery.md

Latest commit

 

History

History
170 lines (127 loc) · 4.33 KB

pytest_discovery.md

File metadata and controls

170 lines (127 loc) · 4.33 KB

Pytest-Style Test Discovery

The framework now uses pytest-style automatic test discovery - no pattern configuration needed!

How It Works

Just like pytest, the framework automatically discovers test files matching these patterns:

  • test_*.py - Files starting with test_

    • test_example.py
    • test_data_validation.py
    • test_my_feature.py

  • *_test.py - Files ending with _test

    • example_test.py
    • data_validation_test.py
    • my_feature_test.py

Automatic & Recursive

  • ✅ No configuration needed - Just name your files correctly
  • ✅ Recursive search - Finds tests in all subdirectories
  • ✅ Both patterns - Matches both test_* and *_test
  • ✅ Workspace support - Works for both local and workspace tests

Examples

Directory Structure

my_project/
├── tests/
│   ├── test_auth.py           ← Found!
│   ├── test_database.py        ← Found!
│   ├── integration/
│   │   ├── test_api.py         ← Found!
│   │   └── test_webhooks.py    ← Found!
│   ├── unit/
│   │   ├── validation_test.py  ← Found!
│   │   └── parser_test.py      ← Found!
│   └── helper.py               ← Skipped (doesn't match pattern)

Running Tests

# Discovers all test_* and *_test files automatically
dbx_test run --local

# Same automatic discovery for remote tests
dbx_test run --remote --profile dev

# Workspace tests also use automatic discovery
dbx_test run --remote --workspace-tests \
  --tests-dir "/Workspace/Repos/my-repo/tests"

Output

Discovering tests matching: test_* or *_test

Found 6 test notebook(s):
  • tests/test_auth.py
  • tests/test_database.py
  • tests/integration/test_api.py
  • tests/integration/test_webhooks.py
  • tests/unit/validation_test.py
  • tests/unit/parser_test.py

Running tests...

Migration from Pattern-Based Discovery

Before (Old Way)

# Had to specify pattern
dbx_test run --remote --pattern "test_*"
dbx_test run --remote --pattern "*integration*"

After (New Way)

# No pattern needed! Automatic discovery
dbx_test run --remote

# Just use descriptive names
# test_integration.py or integration_test.py

Best Practices

✅ DO: Use Clear Names

test_user_authentication.py    # Clear what it tests
test_data_pipeline.py          # Clear what it tests
api_integration_test.py        # Clear what it tests

❌ DON'T: Use Non-Standard Names

check_users.py          # Won't be discovered
validate_data.py        # Won't be discovered
tests_for_api.py        # Won't be discovered (wrong pattern)

Databricks Notebooks

In Databricks workspace, notebooks don't show .py extension in the UI, but the discovery still works:

Workspace Structure

/Workspace/Users/me/tests/
├── test_feature_a       ← Found!
├── test_feature_b       ← Found!
├── validation_test      ← Found!
└── helpers              ← Skipped

Running

dbx_test run --remote --workspace-tests \
  --profile adb \
  --tests-dir "/Workspace/Users/me/tests"

Configuration

The test_pattern setting in config/test_config.yml is now deprecated but kept for backwards compatibility:

paths:
  workspace_root: "/Workspace/tests"
  # test_pattern is no longer needed!
  # Discovery is automatic: test_* and *_test

FAQ

Q: Can I still use custom patterns?

A: No, the framework now uses pytest-style discovery only. This makes it more predictable and follows Python testing conventions.

Q: What if I have test files not matching the pattern?

A: Rename them to follow pytest conventions:

  • my_tests.py → test_my_feature.py or my_feature_test.py

Q: Does this work for workspace tests?

A: Yes! Automatic discovery works everywhere:

  • Local tests (--local)
  • Remote tests (--remote)
  • Workspace tests (--remote --workspace-tests)

Q: Can I have both patterns in the same directory?

A: Yes! You can mix:

tests/
├── test_feature.py      ← Found!
├── feature_test.py      ← Found!
└── another_test.py      ← Found!

Summary

✨ Just name your test files with test_* or *_test and the framework finds them automatically!

No configuration, no patterns, no hassle - just like pytest! 🎉