See Explorer output code within a tab

[ahuang11]

[maximlt]

Oops failing tests 🙃

Nice feature! Could you please add some tests and documentation?

In your screenshot is it normal there's no kind in the call to hvplot()?

Related work/thoughts

• there's an issue somewhere to turn plot_code into a Parameter (that'd maybe just code/plot_repr/code_repr/... to avoid breaking current users of plot_code)
• plot_code doesn't infer the variable name, it defaults to df as shown in your example. I found this cool trick that leverages f-strings to obtain the name of a variable, might be worth considering.

my_df = pd.DataFrame()
var_name = f'{my_df=}'.split('=')[0]

[maximlt]

Also @ahuang11 you could please add to the 0.9.0 milestone the PRs you need in that release?

[maximlt]

I just gave a quick review before @hoxbro's one.

I'm the one who suggested adding the code Parameter to the explorer. IIRC, this actually was a suggestion made by @MarcSkovMadsen a while back, I'd be happy to hear from you Marc whether this is what you expected at the time?

If it's alright, I believe the new code Parameter should be documented. I'm also wondering whether plot_code() should be deprecated, the only advantage it has is that you can provide a name for the data variable. So now we have .plot_code(var_name='df'), .code and _build_code_snippet(). We could have just code so not allow users to provide a variable name? The minimum we should do is merge plot_code and _build_code_snippet, if we want to expose such a method as public API.

[maximlt]

I would not include the import, it doesn't bring much in my opinion.

[ahuang11]

I disagree; it reduces the friction for the user to immediately get started--else, the user needs to manually type import hvplot.pandas

If they do not want the import, they can simply copy up to the line below.

[maximlt]

IIRC you have to run the import first in a notebook for the explorer to work. What use case do you have in mind?

[ahuang11]

This doesn't necessarily have to run in notebook, but if they do, they can copy it to a script or in the future, if they launch the explorer from CLI.

[ahuang11]

Initially I wanted to merge _build_code_snippet and plot_code, but I believe you suggested:

there's an issue somewhere to turn plot_code into a Parameter (that'd maybe just code/plot_repr/code_repr/... to avoid breaking current users of plot_code)

I agree we can simply deprecate plot_code.

[maximlt]

Sorry I wasn't clear enough, I meant that we couldn't just remove plot_code or turn it into a Parameter. I believe though it's fine changing its code, I doubt there's someone doing any kind of parsing on its output.

[ahuang11]

Okay I refactored this so that _build_code_snippet is now merged into plot_code and also cleaned up unnecessary redundancy with _code. Lastly, I renamed some internal variables for clarity.

[ahuang11]

Tests failing seem unrelated to this PR:

File /usr/share/miniconda3/envs/test-environment/lib/python3.8/site-packages/geoviews/data/geom_dict.py:327, in geom_from_dict(geom, xdim, ydim, single_type, multi_type)
    325 split_holes = geom.pop('holes', None)
    326 if split_holes is not None and len(split_holes) != len(split_geoms):
--> 327     raise DataError('Polygons with holes containing multi-geometries '
    328                     'must declare a list of holes for each geometry.')
    330 if single_type is Point:
    331     if len(splits) > 1 or any(len(g) > 1 for g in split_geoms):

DataError: Polygons with holes containing multi-geometries must declare a list of holes for each geometry.

A new version of maybe geopandas / shapely may have broke something; works locally.

rasm = xr.tutorial.open_dataset('rasm').load()



rasm.hvplot.contourf(
    'xc', 'yc', crs=ccrs.PlateCarree(), projection=ccrs.PlateCarree(),
    ylim=(0, 90), frame_width=800, cmap='viridis', levels=10,
    coastline=True
)

[maximlt]

• I merged main into this branch
• I decided against having import.<backend> displayed in the code
  • I believe it's not always going to be very useful
  • We've had issues (resolved now I think) with handling multiple imports
  • So, I wasn't sure about this feature, and as it's adding is easier than removing, I prefer postponing adding this feature in a next release, if we decide it's the right thing to do