tests

Tests

Directory Structure

tests/
├── conftest.py              # Root fixtures + skipped test registry
├── test_*.py                # Unit tests (~170 files)
├── integration/             # Cross-component integration tests (~25 files)
├── e2e/                     # End-to-end CLI and library usage tests
│   ├── conftest.py          # E2E fixtures (myspell_cmd, e2e_test_db, run_myspell)
│   └── test_*.py            # CLI E2E, library usage, custom providers, robustness
├── stress/                  # Concurrency, batch processing, compound stress tests
├── fixtures/                # Shared test data
│   ├── myanmar_test_samples.py  # Ground-truth segmentations for Myanmar text
│   ├── config_templates.py      # YAML/JSON config templates for config tests
│   └── benchmarks/              # POS gold standard data
└── benchmarks/              # Benchmark suite (separate from tests/; lives at repo root)

Running Tests

# All tests
pytest tests/

# Single file
pytest tests/test_syllable_rules.py

# Single test method
pytest tests/test_syllable_rules.py::TestSyllableValidator::test_valid_syllable

# By marker
pytest tests/ -m unit
pytest tests/ -m integration
pytest tests/ -m "not slow"

# With coverage
pytest tests/ --cov=src/myspellchecker --cov-report=term-missing

# Pattern matching
pytest tests/ -k "context"

# Verbose
pytest tests/ -v

Markers

Defined in pyproject.toml under [tool.pytest.ini_options]:

Marker	Purpose	When to use
unit	Fast, isolated single-function tests	Default for most tests
integration	Cross-component interaction tests	Tests that wire multiple real components
slow	Tests that take >5s	Large corpus processing, full pipeline runs
e2e	End-to-end tests via CLI or public API	Tests that invoke myspellchecker as a user would
benchmark	Performance measurement tests	Requires pytest-benchmark installed

Usage: @pytest.mark.unit on test class or method.

Key Fixtures (conftest.py)

Session-scoped (run once per test session)

Fixture	Auto	Purpose
mock_resource_downloads_session	Yes	Prevents HuggingFace downloads; stubs WordTokenizer with RegexSegmenter fallback
patch_default_db	Yes	Patches SQLiteProvider.__init__ to inject a test DB when no path is given
test_database_path	No	Creates a temporary SQLite DB with sample syllables, words, and bigrams

Per-test (run before each test)

Fixture	Auto	Purpose
reset_grammar_config_singleton	Yes	Clears GrammarRuleConfig singleton to prevent cross-test pollution
reset_rich_console	Yes	Redirects Rich console output to StringIO
reset_singletons	Yes	Clears all singleton instances (clear_all_singletons)
clear_lru_caches	Yes	Clears known @lru_cache functions (segmenter, logger)
reset_logging	Yes	Restores myspellchecker logger handlers/propagation after each test
mock_console	No	Returns a MagicMock PipelineConsole (use explicitly when needed)

E2E fixtures (tests/e2e/conftest.py)

Fixture	Scope	Purpose
myspell_cmd	Session	Resolves the myspellchecker CLI command path
e2e_test_db	Session	Builds a small test DB from a tiny corpus via the pipeline
run_myspell	Function	Helper to invoke CLI with args and optional stdin input

Skipped Test Registry

The top of conftest.py documents every skipped and xfailed test, organized by category:

1. Cython extension not compiled -- Needs python setup.py build_ext --inplace
2. Optional dependency not installed -- transformers, torch, jsonschema, pytest-benchmark, pyyaml
3. Transformers mocked by another module -- Test ordering issue
4. SpellChecker not available -- Needs a built DB or active patch_default_db
5. Production DB not available -- Needs data/mySpellChecker_production.db
6. xfail: known limitations -- Pali dictionary entries not yet added
7. Data-dependent -- Conditional on runtime data (e.g., SKIPPED_CONTEXT_WORDS)
8. OS-dependent -- Environment-specific guards (e.g., symlink loops)

Writing New Tests

Naming

• File: test_<module_or_feature>.py
• Class: Test<ComponentName> (optional, group related tests)
• Method: test_<behavior_under_test>

Markers

Always mark tests appropriately:

import pytest

@pytest.mark.unit
class TestMyFeature:
    def test_basic_case(self):
        ...

    @pytest.mark.slow
    def test_large_corpus(self):
        ...

Using fixtures

The auto-use fixtures handle singleton/cache cleanup automatically. For tests needing a database:

def test_with_db(test_database_path):
    from myspellchecker.providers.sqlite import SQLiteProvider
    provider = SQLiteProvider(database_path=str(test_database_path))
    ...

Parametrize patterns

Use @pytest.mark.parametrize for Myanmar text validation:

@pytest.mark.parametrize("text,expected", [
    ("မြန်မာ", True),
    ("xyz", False),
])
def test_validation(text, expected):
    assert validate(text) == expected

Mocking heavy dependencies

For ONNX/semantic tests, mock the inference backend:

from unittest.mock import MagicMock, patch

def test_semantic_checker():
    mock_backend = MagicMock()
    mock_backend.predict.return_value = [0.95]
    with patch("myspellchecker.algorithms.semantic_checker.InferenceBackend", return_value=mock_backend):
        ...

Coverage requirement

Maintain >=65% code coverage. Check with:

pytest tests/ --cov=src/myspellchecker --cov-fail-under=65