Legend rank per pie slice

[my-tien]

This PR makes it possible to pass arrays to the pie property 'legendrank' so that legend rank assignment can be configured per slice.

Changes:

Previously, traces.pie.legendrank only accepted one value that applied for all slices. With this PR, slice legend entries can be individually ranked in the legend. E.g. the below JSON would produce the displayed chart:

{
"data": [
{
"type": "pie",
"values": [1, 2, 3, 4, 5],
"labels": ["E", "D", "C", "B", "A"],
"legendrank": [5, 4, 3, 2, 1]
}
],
"layout": {
"width": 500,
"height": 400,
"legend": {
"title": {
"text": "A legend"
}
}
}
}

This fixes#7778

Note:

This feature is not expected to work when pie values are not set.

Disclaimer

I am required to add that…

the software is provided "as is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. in no event shall the authors or copyright holders be liable for any claim, damages or other liability, whether in an action of contract, tort or otherwise, arising from, out of or in connection with the software or the use or other dealings in the software.

[my-tien]

Hi guys, I'm not sure if the failed baselines-b64 test is related to my changes. It seems that the legendrank array doesn't take effect there?

Expected output (and actual output for the baseline test):

output for baselines-b64 test (note incorrect legend entry order):

[my-tien]

I fixed the baseline-b64 test mentioned in my previous comment.
It was a mistake in my code. In get_legend_data.js I only checked Array.isArray() on legendrank which failed for typed arrays.

This time the failed baseline test seems to contain only minimal pixel differences. Any advice here? @emilykl

This is diff-zz-pie-slice-legendrank.png

[my-tien]

Sorry for the spam, I saw that kaleido was bumped for the CI pipeline which explains the pixel differences. I updated the baseline image. All good from my side! Should I look into the failed flaky test? Or do you think this is fine?

[emilykl]

Hi @my-tien ! Don't worry about the flaky test for now, you're right that it's unrelated to this PR.

Could you link an issue to this PR? You can open a new one if needed.

I'm OK with this change, but could you explain why it's better for you to use legendrank rather than passing the data array in the same order you want the items to appear in the legend?

[my-tien]

Thanks Emily, I forgot to create an issue for this, sorry. I now added an issue and linked it in the Pull Request description. I tried to explain the issue there. Let me know if I should add anything to it!

[emilykl]

Thanks @my-tien! The justification in the linked issue makes sense, I had forgotten that pie legend entries sort automatically by value.

Could you rename the image mock and baseline image to remove the initial zz-? (I apologize for the workflow change, I realize it's different than what we've done in the past.) Once that's done I think this is ready to go.

[emilykl]

Thanks @my-tien !