fixing hover_data and hover_name bugs in path API #2524
Conversation
| assert fig.layout.coloraxis.colorbar.title.text == "lifeExp" | ||
|
|
||
| df = px.data.tips() | ||
| fig = px.sunburst(df, path=["sex", "day", "time", "smoker"], hover_name="smoker") |
There was a problem hiding this comment.
this actually errors out on master (because smoker is in path but isn't accounted for in hover_name)
| fig = px.sunburst( | ||
| df, path=["continent", "country"], color="lifeExp", hover_data=df.columns | ||
| ) | ||
| assert fig.layout.coloraxis.colorbar.title.text == "lifeExp" |
There was a problem hiding this comment.
on master this appears as lifeExpadditional_col_for_hover because we rewrite args["color"]
| assert "%{hovertext}" in fig.data[0].hovertemplate # represented as '%{hovertext}' | ||
|
|
||
| df = px.data.tips() | ||
| fig = px.sunburst(df, path=["sex", "day", "time", "smoker"], custom_data=["smoker"]) |
There was a problem hiding this comment.
this also errors out on master (because smoker is in path but isn't accounted for in custom_data)
| for col_name in path: | ||
| series_to_copy = df[col_name] | ||
| new_col_name = col_name + "_path_copy" | ||
| path = [new_col_name if x == col_name else x for x in path] |
There was a problem hiding this comment.
the new approach is just to make a full copy of everything in path even if they're not in color or whatever and not touch any of the other args, instead of spot-checking color and hover_data... Maybe there's a case here I'm missing but it's not in the tests and this approach seems more future-friendly.
There was a problem hiding this comment.
probably we don't need the if x == col_name here? Since x is in path.
There was a problem hiding this comment.
it's an elegant solution. I had not thought that column names in path were disposable since they would not appear in the final hovertemplate.
There was a problem hiding this comment.
you're right that comprehension is weird... check out the new version?
|
Thanks for the improvement! I left a small comment since I think one |
nicolaskruchten
force-pushed
the
path_hover_fixes
branch
from
99664ad
to
2f401e0
Compare
This approach is a bit less brittle I think.