.cursor/rules/python_tests.mdc - develop - vshmidt/streamlit

.cursor/rules/python_tests.mdc
46 строк2 KB
Laura Wilby
[chore] clarify anti-regression test assertion rules for AI agents (#13886)
10 фев 2026, 22:20
10 фев 2026, 22:20dea9a52
---
description:
globs: lib/tests/**/*.py
alwaysApply: false
---

# Python Unit Test Guide

We use the unit tests to cover internal behavior that can work without the web / backend counterpart.
We aim for high unit test coverage (90% or higher) of our Python code in `lib/streamlit`.

## Key Principles

- Prefer pytest or pytest plugins over unittest.
- For every new test function, please add a brief docstring comment (numpydoc style).
- New tests should be fully annotated with types.
- Imports should be at the top-level of the test file. Only use imports inside test functions when there is a specific reason (e.g., integration requirements, circular import issues, testing import behavior, or within `AppTest` functions).
- Skip tests (via `pytest.mark.skipif`) requiring CI secrets if the environment variables are not set.
- Parameterized Tests: Use `@parameterized.expand` whenever it is possible to combine overlapping tests with varying inputs.
- Anti-regression assertions: Where practical, go beyond testing the happy path by also covering a plausible failure mode or edge case. Good examples:
  - Test that invalid input raises the expected exception.
  - Test a boundary condition (empty list, zero, `None`, max length).
  - Assert that a side effect does **not** occur (e.g., a read-only operation must not mutate state).
  - Assert that a return value does not contain a plausible-but-wrong entry (e.g., a filter function must not include excluded items).
  - **Do NOT** add assertions that are logically implied by an earlier assertion in the same test. For example, if you already assert `x is True`, do not also assert `x is not False` — that is a tautology and adds no value. Similarly, do not assert both sides of a simple boolean or enum when one implies the other.
- Prefer targeted negatives over exhaustive matrices: Add one high-signal negative check per behavior; don't balloon test cases without a regression history.

## Running tests

- Run all with (execute from repo root):

```bash
make python-tests
```

- Run a specific test file with:

```bash
uv run pytest lib/tests/streamlit/my_example_test.py
```

- Run a specific test inside a test file with:

```bash
uv run pytest lib/tests/streamlit/my_example_test.py -k test_that_something_works
```