add_shape, add_layout_image, add_annotation for multiple facets at once, add_hline and add_vline (+add_vrect/add_hrect)#2558
add_shape, add_layout_image, add_annotation for multiple facets at once, add_hline and add_vline (+add_vrect/add_hrect)#2558nicholas-esterer wants to merge 59 commits intomasterfrom
Conversation
Need more comprehensive tests for all the other add_* functions and for all the possible combinations of row and col specification.
Simply pass a tuple for row and/or col where the second arg == 'all'.
(maybe this is overkill though, as a user could easily pass list(range(...)) instead)
| if row is not None: | ||
| grid_ref = self._validate_get_grid_ref() | ||
| refs = grid_ref[row - 1][col - 1] | ||
| # TODO It is assumed that grid_ref has an equal number of columns in |
There was a problem hiding this comment.
Yes, this can be not-true, you can have really odd layouts: https://plotly.com/python/subplots/#multiple-custom-sized-subplots
There was a problem hiding this comment.
Hey so I checked this out and it turns out even in this case grid_ref is still a rectangle, just some of the entries are None. So we could still easily iterate through the grid_ref using something like
for r in range(1,len(grid_ref)+1):
for c in range(1,len(grid_ref[0])+1):
sub_plot=fig.get_subplot(row=r,col=c)
if sub_plot is not None:
....We just check if it is not None before doing anything to it 🔧
Not yet tested.
|
I greatly refactored the old approach and added basically this one function (it calls some helpers which you can check out in the code):
The idea is to change |
|
|
Recursion is only of depth 1.
…w and col Still need to add the product argument though.
|
We have decided to not support the cartesian product mode for |
for the add_trace and add_{shape,layout_image,annotation} Figure methods.
|
Hmm seeing now that the |
|
Ok so the PR doesn't affect |
This makes it possible to replace a shape in the layout with one where its one axis is fixed to the paper. This makes it possible to easily implement functions like add_vline, which add a horizontal line to a plot.
Still need to write tests though.
No tests written yet
|
For reviewers: |
| from contextlib import contextmanager | ||
| from copy import deepcopy, copy | ||
| import itertools | ||
| import pdb |
There was a problem hiding this comment.
this probably shouldn't be here?
There was a problem hiding this comment.
(the pdb part I mean)
There was a problem hiding this comment.
itertools yes, pdb no
|
I think reusing |
Making sure that position is for any orientation of the shape.
They can be added using the annotation keyword argument:
fig.add_hline(y=1,annotation=go.layout.Annotation(text="example 1"))
or you can specify their parameters using the annotation_ prefixed
keywords
fig.add_hline(y=1,annotation_text="example 2")
There's also a quick way to specify some commonly encountered positions,
through the annotation_position argument
fig.add_vrect(y=1,annotation_text="example 3",
annotation_position="inside bottom left")
Moved the tests for axis spanning shapes and their annotated versions to test_autoshapes.
|
hline, vline, hrect, vrect, can be annotated. To see an example of what this looks like, you can run from the root of this repo: VISUALIZE=1 python packages/python/plotly/plotly/tests/test_core/test_autoshapes/test_annotated_shapes.pyNote: probably commit 8a62ba7 has to get checked out before you can do this. |
Made common file to share functions for all tests Wrote spec of annotation shape tests.
See packages/python/plotly/plotly/tests/test_core/test_autoshapes/test_annotated_shapes.py for information on how to configure the test (it may need to be updated from time to time).
|
Annotations look great! Some comments:
|
…shapes do not try and set the most recently added layout object's xref and yref if we wanted to add it to a plot with one subplot/facet.
Updated regex in EnumeratedValidator.build_regex_replacement
Before Python3.5 it seems keys were not sorted by default (although json.dumps says they always aren't by default, but this might be related to some other serialization).
|
But if we say |
|
That sounds good |
Fails on bad description.
|
how close are the docs to being done here? Need any help or guidance? |
|
Just have to change them to reflect the behaviour where specifying no row or col puts the shapes on all subplots. And I have to document the |
|
Actually I was mistaken, it is sufficient to have the documentation appear when calling |
|
No, we need to have docs written in the MD files :) Also: the build output from |
|
OK, so once we release plotly.js 1.57.0, i'll update |
3 participants
Closes #2141 and closes #2140
Making it possible to specify multiple rows and columns when using
add_shape,add_layout_imageoradd_annotation.This pull request is to share the work, so a lot is missing still.
Code PR
plotly.graph_objects, my modifications concern thecodegenfiles and not generated files.modified existing tests.
new tutorial notebook (please see the doc checklist as well).