测试与 pytest

可靠软件离不开自动化测试。pytest 以简洁断言、fixture 和参数化著称，是 Python 项目事实标准。模块组织见 python-fundamentals；Agent 工具链也需测试，参考 ai-agents-essentials；/roadmaps/python。

学习目标

• 编写 pytest 测试函数与 assert 语句
• 使用 fixture 管理测试数据与临时目录
• 了解 mock、覆盖率与 CI 集成要点

基础测试

# tests/test_stats.py
from pathlib import Path
from lesson_tools.stats import count_chars

def test_count_chars(tmp_path: Path) -> None:
    (tmp_path / "a.md").write_text("hello", encoding="utf-8")
    (tmp_path / "b.md").write_text("world!", encoding="utf-8")
    result = count_chars(tmp_path)
    assert result == {"a.md": 5, "b.md": 6}

pytest -v
pytest tests/test_stats.py::test_count_chars

fixture 与参数化

import pytest

@pytest.fixture
def sample_draft(tmp_path: Path) -> Path:
    (tmp_path / "pya-testing.md").write_text("# Testing\n", encoding="utf-8")
    return tmp_path

@pytest.mark.parametrize("text,expected", [
    ("", 0),
    ("abc", 3),
    ("中文", 2),
])
def test_len_text(text: str, expected: int) -> None:
    assert len(text) == expected

mock 外部依赖

from unittest.mock import AsyncMock, patch
import asyncio

async def fetch_remote(slug: str) -> str:
    ...  # 实际会请求 HTTP

@pytest.mark.asyncio
async def test_fetch_remote() -> None:
    with patch("mymodule.fetch_remote", new_callable=AsyncMock) as mock:
        mock.return_value = "cached content"
        result = await fetch_remote("pya-testing")
        assert "content" in result

安装 pytest-asyncio 以测试 async 代码。

CI 清单

• PR 必跑 pytest、ruff check、mypy
• 覆盖率目标：核心模块 ≥ 80%（非硬性，视项目而定）
• 测试应快、独立、可重复

练习

• 为 packaging 课的 count_chars 补一个空目录边界测试
• 用 @pytest.mark.parametrize 测试 slugify 的 3 组输入
• 在 GitHub Actions 或本地脚本中串联 pytest + ruff