Implement plot span map

[0Hughman0]

Note this is a port of this PR to lumispy

Description of the change

A few sentences and/or a bulleted list to describe and motivate the change:

• implemented plot_span_map. This is a tool for automatically plotting a spatial map of intensity of a hyperspectral map over a number of ranges defined by a SpanROI.

Progress of the PR

• Change implemented (can be split into several points),
• update docstring (if appropriate),
• update user guide (if appropriate),
• add an changelog entry in the upcoming_changes folder (see upcoming_changes/README.rst),
• Check formatting changelog entry in the readthedocs doc build of this PR (link in github checks)
• add tests,
• ready for review.

Minimal example of the bug fix or the new feature

import numpy as np
import hyperspy.api as hs

s = hs.signals.Signal1D(data=np.random.random((64, 64, 1024)), 
                        axes=[{'name': 'x', 'size': 64}, 
                              {'name': 'y', 'size': 64}, 
                              {'name': 'sig', 'size': 1024}])

hs.plot.plot_roi_map(s, rois=2)

[jlaehne]

Great for porting this upstream. Will definetely be a very valuable feature!

[codecov]

Codecov Report

Attention: 2 lines in your changes are missing coverage. Please review.

Comparison is base (224dca7) 80.86% compared to head (39b7a2b) 80.97%.
Report is 10 commits behind head on RELEASE_next_major.

Files	Patch %	Lines
hyperspy/drawing/_widgets/range.py	50.00%	2 Missing ⚠️

Additional details and impacted files

@@                  Coverage Diff                   @@
##           RELEASE_next_major    #3224      +/-   ##
======================================================
+ Coverage               80.86%   80.97%   +0.10%     
======================================================
  Files                     140      140              
  Lines                   20399    20458      +59     
  Branches                 4825     4834       +9     
======================================================
+ Hits                    16495    16565      +70     
+ Misses                   2829     2825       -4     
+ Partials                 1075     1068       -7     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ericpre]

Thank you @0Hughman0, this looks in very good shape!

I think that it would be possible to make more general by supporting the following cases:

1. signal dimension of 2, use 2D roi instead of 1D
2. navigation dimension other than 2

Would you interested in trying to implement this in this PR? If not, we (not necessarily including yourself) can try to do this in a follow up PR?

For point 1, I think that we would need to change the name of the function, maybe to something along the line of plot_roi_map in which case, it would be better to do it before the 2.0 release.

[jlaehne]

Tests on windows are failing?

[0Hughman0]

No problem.

Hyperspy and Lumispy were huge helps over my PhD, happy to contribute.

I can imagine there are other scenarios where similar tools would be useful. As someone who's only really worked with hyperspectral maps, I can find it a bit mindblowing to consider more complex options!

My worry is having an all-in-one implementation for every combination of signals and navigation dimensions could get rather messy and very difficult to implement, document and test.

Is there a way I can get some toy datasets to play with? Don't mind having a go to see how things turn out.

At the risk of going a bit off course in the discussion, I don't know if you have seen or heard of plum - it allows multiple dispatch in Python. Would allow for separating out implementations of method and function calls, depending on the shape of the data. Potentially very useful I think for this library.

[ericpre]

I can imagine there are other scenarios where similar tools would be useful. As someone who's only really worked with hyperspectral maps, I can find it a bit mindblowing to consider more complex options!

My worry is having an all-in-one implementation for every combination of signals and navigation dimensions could get rather messy and very difficult to implement, document and test.

It is actually quite simple when using a suitable approach.

Is there a way I can get some toy datasets to play with? Don't mind having a go to see how things turn out.

There are some dataset available at https://pyxem.readthedocs.io/en/stable/open_datasets_workflows.html

At the risk of going a bit off course in the discussion, I don't know if you have seen or heard of plum - it allows multiple dispatch in Python. Would allow for separating out implementations of method and function calls, depending on the shape of the data. Potentially very useful I think for this library.

Thank you for sharing, I can't see straightaway where it would be useful but this is good that it is exist. With the structure of signals classes and the class registration mechanism (https://hyperspy.org/hyperspy-doc/dev/user_guide/signal/signal_basics.html#hyperspy-extensions) we already have a mechanism in place to automatically set the correct class (and functionality) depending on the signal type ("diffraction", "eels", etc.) and dimensionality.

[0Hughman0]

I'm having a great deal of difficulty getting the pytest-mpl tests to work during CI.

I have made sure I'm using matplotlib==3.5.1, but that doesn't seem to have helped.

Suggestions would be welcome!

Thanks!

[ericpre]

Can you check your version of freetype? If needs to be as follow:

hyperspy/conda_environment_dev.yml

Lines 6 to 7 in 70ca076

	- freetype=2.10
	- matplotlib-base=3.5.1

I thought that this was documented in developer guide but it is not!

If this doesn't help, we can update the baseline images for you.

[0Hughman0]

Thanks, things seem to be passing now.

Failures on windows seem to be something to do with running out of memory when building fonts?

I thought that this was documented in developer guide but it is not!

yes I think maybe the dev install instructions need to be updated for 2.0 perhaps?

I think version 3.5.1 of matplotlib is not compatible with Python 3.11, which I think caused me issues early on.

[0Hughman0]

Helloo, happy to put in some time to get this mergable, but not sure what to do next?

[jlaehne]

I propose to first get the present functionality merged and leave an implementation for 2D for a separate PR - even if you want to have a go at it, I think it is the approach that is manageable more easily.

As quite a bit has happened to the main branch in the meantime, I would propose to get the current version of RELEASE_next_major merged into this branch and then see what has to be done to get all the tests passing.

[ericpre]

Thank you @0Hughman0, this looks good to me. I did some tweaks and fixed an already existing bug that this PR highlights.

Sorry that it took some times to come back to this PR, but we were busy with other things! :)

[ericpre]

@jlaehne, do you want to do a last review?

[jlaehne]

Just some small improvements to the documentation.

[jlaehne]

Hyperspy and Lumispy were huge helps over my PhD, happy to contribute.

Thanks @0Hughman0 - it was a long way, but this has become a really neat feature! If you have other code from your PhD that you want to contribute, we're happy for suggestions :-)

[0Hughman0]

This is great! Thanks so much for getting this across the line.

Looking forward to installing 2.0 and having a go!

I'll have a dig through my codebase to see if there's any other handy snippets. If I get the time, I can maybe pick up an issue or two.

My biggest coding project during my PhD was Cassini which is a tool I used to organise my data and analysis. You guys should check it out if you have the time 😁.