Organize test suite into classes

Currently tests are organized into modules and then separated into sections using comment rules e.g.

# pin_download ===================================================================

This can make it hard to navigate the test suite and also results in longer test names: the test names include both the category information and the specifics of the test.

Pytest allows you to organize tests into classes. Besides adding some clarity to the test organization, this also allows the user to have more fine-grained control about which tests run.

So I would propose changing the following:

# pin_write ===================================================================

def test_board_pin_write_default_title(board):

    df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})
    meta = board.pin_write(df, "df_csv", title=None, type="csv")
    assert meta.title == "df_csv: a pinned 3 x 2 DataFrame"


def test_board_pin_write_prepare_pin(board, tmp_dir2):

    df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})

    meta = board.prepare_pin_version(
        str(tmp_dir2), df, "df_csv", title=None, type="csv"
    )
    assert meta.file == "df_csv.csv"
    assert (tmp_dir2 / "data.txt").exists()
    assert (tmp_dir2 / "df_csv.csv").exists()
    assert not (tmp_dir2 / "df_csv.csv").is_dir()


def test_board_pin_write_roundtrip(board):

    df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})

    assert not board.pin_exists("df_csv")

    board.pin_write(df, "df_csv", type="csv")

    assert board.pin_exists("df_csv")

    loaded_df = board.pin_read("df_csv")
    assert loaded_df.equals(df)


def test_board_pin_write_type_not_specified_error(board):
    class C:
        pass

    with pytest.raises(NotImplementedError):
        board.pin_write(C(), "cool_pin")

To this:

class TestBoard:
    ... # All the other sets of tests on pins.board.BaseBoard methods...
    class TestPinWrite:
        def test_default_title(self, board):
        
            df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})
            meta = board.pin_write(df, "df_csv", title=None, type="csv")
            assert meta.title == "df_csv: a pinned 3 x 2 DataFrame"
        
        
        def test_prepare_pin(self, board, tmp_dir2):
        
            df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})
        
            meta = board.prepare_pin_version(
                str(tmp_dir2), df, "df_csv", title=None, type="csv"
            )
            assert meta.file == "df_csv.csv"
            assert (tmp_dir2 / "data.txt").exists()
            assert (tmp_dir2 / "df_csv.csv").exists()
            assert not (tmp_dir2 / "df_csv.csv").is_dir()
        
        
        def test_roundtrip(self, board):
        
            df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})
        
            assert not board.pin_exists("df_csv")
        
            board.pin_write(df, "df_csv", type="csv")
        
            assert board.pin_exists("df_csv")
        
            loaded_df = board.pin_read("df_csv")
            assert loaded_df.equals(df)
        
        
        def test_type_not_specified_error(self, board):
            class C:
                pass
        
            with pytest.raises(NotImplementedError):
                board.pin_write(C(), "cool_pin")
        
        ... # etc.

My plan for making this change would be one test file at a time.

Begin with the test file containing the pin_write section for pins.board.BaseBoard, then identify the other comment-separated test sections one file at a time. Convert those sections into nested test classes and shorten method names while preserving the existing assertions and fixtures. Run the affected pytest file after each conversion and confirm the full suite still passes.