Test LeetCode Solutions Locally with pytest

Solving on leetcode.com means round-tripping between browser tabs. A generated practice environment moves that loop into your terminal: every problem ships with a parametrized pytest suite, so testing your solution is one command in your own editor, with your own debugger and no page reloads.

One-time setup

pip install leetcode-py-sdk
pip install pytest

The package pulls in loguru, which the suites use to log every case. pytest is not a package dependency, so install it in the same environment. Or use uv: uv tool install leetcode-py-sdk plus uv pip install pytest in your active venv.

Generate a problem, see red

lcpy gen -n 1
cd two_sum
python -m pytest test_solution.py

The suite runs immediately and mostly fails, because solution.py arrives as a TODO stub:

15 failed, 3 passed in 0.86s

Every failing case is named by its parametrized id, and the assertion output shows the inputs and the expected result, so you know exactly which contract you broke.

Implement, go green

Replace the TODO in solution.py with your approach:

class Solution:
    def two_sum(self, nums: list[int], target: int) -> list[int]:
        seen: dict[int, int] = {}
        for i, n in enumerate(nums):
            if target - n in seen:
                return [seen[target - n], i]
            seen[n] = i
        return []

Rerun the suite:

python -m pytest test_solution.py

18 passed in 0.45s

Never edit the tests to make them pass. The suite is the contract; the 10+ cases per problem include the LeetCode examples plus edge cases (empty results, negatives, duplicates, boundary sizes), so green here means green, not “green for the two examples I checked”.

Iterate on one case at a time

Mid-debug, you want one case, not eighteen. pytest flags that earn their keep on these suites:

Command	Effect
`python -m pytest -k nums0`	run only the case whose id matches
`python -m pytest -x`	stop at the first failure
`python -m pytest -s`	show the per-case loguru output

python -m pytest test_solution.py -k nums0

1 passed, 17 deselected in 0.26s

With -s, each case logs its inputs and verdict:

2026-08-24 14:41:49 | DEBUG | Running test_two_sum(nums=[2, 7, 11, 15], target=9, expected=[0, 1])
2026-08-24 14:41:49 | DEBUG | Test passed! ✨

A whole practice tree works the same

Generate a collection and plain pytest discovers every suite under the current directory:

lcpy gen -t blind-75
python -m pytest          # every problem, every case

Pick your set in Collections or browse the Catalog.

Where to go next

Anatomy of a test suite: how the parametrized cases, helpers, and logging fit together, and how one suite covers multiple solution classes
Problem anatomy: what each of the six generated files is for
lcpy CLI reference: every gen and list option

Welcome

Getting Started

Practice

CLI

Catalog

Contributing

Troubleshooting

Test LeetCode Solutions Locally with pytest

One-time setup

Generate a problem, see red

Implement, go green

Iterate on one case at a time

A whole practice tree works the same

Where to go next

​One-time setup

​Generate a problem, see red

​Implement, go green

​Iterate on one case at a time

​A whole practice tree works the same

​Where to go next

One-time setup

Generate a problem, see red

Implement, go green

Iterate on one case at a time

A whole practice tree works the same

Where to go next