statatest 0.1.0

Pytest-inspired testing and code coverage framework for Stata

statatest

A pytest-inspired testing and code coverage framework for Stata.

Documentation | GitHub | PyPI

Features

• Test Discovery: Automatically find and run test_*.do files
• Rich Assertions: Built on Stata's native assert with detailed failure messages
• Fixture System: Reusable setup/teardown with pytest-like scoping
• Code Coverage: Line-level coverage via SMCL comment instrumentation
• CI Integration: JUnit XML output for GitHub Actions, LCOV for Codecov
• Backward Compatible: Works with existing test patterns

Installation

Using uv (recommended)

# Install globally
uv tool install statatest

# Or run directly without installing
uvx statatest tests/

Using pip

pip install statatest

Quick Start

# Run all tests in tests/ directory
statatest tests/

# Run with coverage
statatest tests/ --coverage

# Generate JUnit XML for CI
statatest tests/ --junit-xml=junit.xml

Writing Tests

// tests/test_myfunction.do

// @marker: unit
program define test_basic_functionality
    // Setup
    clear
    set obs 10
    gen x = _n

    // Test
    myfunction x, gen(y)

    // Assert
    assert_var_exists y
    assert_equal _N, expected(10)
end

Assertions

Function	Purpose	Example
assert_equal	Value equality	assert_equal "\r(mean)'"", expected("5")`
assert_true	Boolean true	assert_true _N > 0
assert_false	Boolean false	assert_false missing(x)
assert_error	Command should fail	assert_error "invalid_command"
assert_var_exists	Variable exists	assert_var_exists myvar
assert_approx_equal	Float comparison	assert_approx_equal r(mean), expected(0.5) tol(0.01)

Fixtures

Create reusable setup/teardown functions with conftest.do:

// tests/conftest.do - Shared fixtures

program define fixture_sample_panel
    clear
    set obs 100
    gen int firm_id = ceil(_n / 10)
    gen int year = 2010 + mod(_n, 10)
    gen double revenue = exp(rnormal(15, 2))
end

program define fixture_sample_panel_teardown
    clear
end

Use fixtures in your tests:

// tests/test_analysis.do
// @uses_fixture: sample_panel

program define test_panel_analysis
    use_fixture sample_panel

    assert_obs_count 100
    assert_var_exists revenue
end

Built-in fixtures:

Fixture	Purpose
fixture_tempfile	Temporary file path ($fixture_tempfile_path)
fixture_empty_dataset	Empty dataset with optional obs count
fixture_seed	Reproducible random seed

Configuration

Create statatest.toml in your project root:

[tool.statatest]
testpaths = ["tests"]
test_files = ["test_*.do"]
stata_executable = "stata-mp"

[tool.statatest.coverage]
source = ["code/functions"]
omit = ["tests/*"]

[tool.statatest.reporting]
junit_xml = "junit.xml"
lcov = "coverage.lcov"

GitHub Actions Integration

name: Stata Tests

on: [push, pull_request]

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: astral-sh/setup-uv@v5
      - run:
          uvx statatest tests/ --junit-xml=junit.xml --coverage
          --cov-report=lcov
      - uses: codecov/codecov-action@v5
        with:
          files: coverage.lcov

Requirements

• Python: 3.11+
• Stata: 16+

License

MIT License - see LICENSE for details.

Acknowledgments

• Inspired by pytest
• SMCL parsing patterns adapted from mcp-stata (AGPL-3.0)