Support legend label mapping

[jbednar]

HoloViews automatically constructs legends as color keys when plotting categorical data, now even for datashaded plots:

As mentioned in #4806 , in many cases the data itself isn't sufficient to generate such a legend. E.g. in https://examples.pyviz.org/census, the data has a categorical column with 'a' or 'w', but the legend should say 'Asian' or 'White'. Similarly in https://examples.pyviz.org/ship_traffic , the data has a numerical category code, but the legend should say a human-readable value like 'Fishing' or 'Cargo'.

For data like this, it would be useful to be able to supply a legend_labels dictionary option to HoloViews, mapping from a column or dimension value in the data to a human-readable label.

[korsbakken]

Does #5342 actually work for all backends? By using the legend_labels option, I get the expected behavior with bokeh, but it appears to have no effect whatsoever with matplotlib.

I test with a figure that has legend labels ('Total', 'Coal', 'Oil', 'Natural gas', 'Oil') to begin with (if I do not use the legend_labels option). The figure is stored in a variable called raw_fig. I test with the following code:

raw_fig.options(
    legend_labels={
        'Natural gas': 'a',
        'Coal': 'b',
        'Other': 'c',
        'Total': 'd',
        'Oil': 'e'
    }
)

With bokeh extension, the legend labels change to a, b, c, d, e as expected. With matplotlib extension and everything else identically the same, they do not. I can also switch back and forth between the two extensions. When bokeh is active, the changed labels are displayed. When matplotlib is active, the original labels are displayed.

I am using the Jupyter Notebook extension in Visual Studio code. with holoviews version 1.15.4 and matplotlib version 3.6.3, running on WSL Ubuntu 20.04 LTS under Windows 11.

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.