Bug summary

When adding legend to an axes withloc set to"best", the created legend does not try to avoid line collections added viaadd_collection.

It seems_auto_legend_data only returns the offsets of a collection, so only scatter-plot-like collections are avoided by the legend. Line collections that are intended to draw data lines always have zero orNone offset values and are thus not avoided when creating the legend.

Code for reproduction

importmatplotlib.pyplotaspltimportmatplotlib.collectionsfig,ax=plt.subplots(1,1,figsize=(7,4),constrained_layout=True)lc=matplotlib.collections.LineCollection([[(0,0), (1,1)]],label='Foo')ax.add_collection(lc)ax.legend()plt.show()

Actual outcome

Expected outcome

Additional information

I work this around by using a patch that essentially adds the following code after the branch here:

        elif (            isinstance(artist, Collection)            and (                artist.get_offsets() is None                or (np.array(artist.get_offsets()) == 0).all()            )        ):            transform, _, _, paths = artist._prepare_points()            lines += [transform.transform_path(path) for path in paths]

But I'm not familar enough with the code to say this won't create other problems, and it seems to be extremely slow with large line collections.

Operating system

No response

Matplotlib Version

3.5.0

Matplotlib Backend

No response

Python version

No response

Jupyter version

No response

Installation

No response