Docstring formatting does not always respect line length in dynamic mode #9126
Description
First off, thanks so much for the great work on the docstring formatter! I'm very much looking forward to using it for the Polars code base.
When running the new docstring formatter on our code base, I found a number of examples where dynamic line length mode did not produce the expected result. See the three examples below:
def example1():
"""
Docstring example containing a class.
Examples
--------
>>> @pl.api.register_dataframe_namespace("split")
... class SplitFrame:
... def __init__(self, df: pl.DataFrame):
... self._df = df
...
... def by_first_letter_of_column_values(self, col: str) -> list[pl.DataFrame]:
... return [
... self._df.filter(pl.col(col).str.starts_with(c))
... for c in sorted(
... set(df.select(pl.col(col).str.slice(0, 1)).to_series())
... )
... ]
"""
class Nested:
def example2():
"""
Regular docstring of class method.
Examples
--------
>>> df = pl.DataFrame(
... {"foo": [1, 2, 3], "bar": [6, 7, 8], "ham": ["a", "b", "c"]}
... )
"""
def example3():
"""
Pragma comment.
Examples
--------
>>> af1, af2, af3 = pl.align_frames(
... df1, df2, df3, on="dt"
... ) # doctest: +IGNORE_RESULT
"""In ruff 0.1.8, running the ruff docstring formatter on these with line-length = 88 will produce line lengths over 88.
I think examples 1 and 2 are clear violations. Example 3 is debatable whether the result is desired or not.