Graphics for file flow

[lea-hagen]

This adds visualizations of what files are created when a BEAST run is split by either subgrids or source density. These are added to the docs and the code is in plotting. To help untangle the mess of lines, some of them are color-coded to separately follow the SED grid, noise model, and photometry.

plot_graphic_file_flow.py has options to change the numbers of subgrids, SD bins/sub-bins, etc. I settled on the current numbers because sometimes the graphs get rearranged in unhelpful ways.

Partially addresses #680

---

Subgrid splitting

---

Source Density splitting

[lea-hagen]

I'll check on the codestyle things. But in the meantime, @karllark do you happen know what's going on with docs error about not being able to read the image file?

[lgtm-com]

This pull request introduces 1 alert when merging a5629c3 into b1950c6 - view on LGTM.com

new alerts:

• 1 for Unused import

[karllark]

@karllark do you happen know what's going on with docs error about not being able to read the image file?

You need to add the png file to the [options.package_data] section in setup.cfg. Png files are not automatically included (unlike py files).

[lea-hagen]

Oh right! I'm pretty sure you've had to explain that to me a few times now 😆

[lea-hagen]

Hmm, that seems to not be the problem. I think that line in setup.cfg is for things in the beast folder, not in the docs. Looking back at #624, there weren't changes to any configuration files (except for adding the graphviz dependency), and as far as I can tell, those images worked fine.

[karllark]

Humm....do the docs build on your computer? Another thought, are the permissions on those files ok?

[karllark]

Ok, the issue is the filename is wrong in the rst file. In the docs it is images/beast-graphic-file-flow-sd.png and the file is actually called images/beast-graphic-file-flow-sourceden.png.

[lea-hagen]

Thanks for catching that!! Hopefully things are good to go now...

[lea-hagen]

It looks like the coverage test is failing because it's having download problems. I noticed that yesterday too, but with a different file.

[karllark]

I agree the coverage tests are an unrelated download issue.

Can you fix the minor codacy issue?

[lea-hagen]

I just checked on the Codacy issue. I'm a little confused - it says "No blank lines allowed after function docstring", but only finds one instance of it, when there are actually three functions with what I assume are blank lines after the function docstring. Is this something you're familiar with?

[karllark]

Is this something you're familiar with?

Nope. If black is ok with the file, then say the word and I'll ignore the codacy check.

[lea-hagen]

Yeah, I re-blackened and nothing was changed. I did some googling, and from discussion here, it seems like it might be varying interpretations of one of the PEPs:
https://github.com/PyCQA/pydocstyle/issues/361

(and apparently I inadvertently linked that issue and this PR... woops 😬)

[karllark]

One last minor change. Can you change the title of the rst page to Graphical Models as now there is more than one?

[lea-hagen]

Done!

[codecov-io]

Codecov Report

Merging #684 (93ce6db) into master (b1950c6) will decrease coverage by 0.67%.
The diff coverage is 4.43%.

@@            Coverage Diff             @@
##           master     #684      +/-   ##
==========================================
- Coverage   44.91%   44.24%   -0.68%     
==========================================
  Files          98       99       +1     
  Lines        9477     9638     +161     
==========================================
+ Hits         4257     4264       +7     
- Misses       5220     5374     +154     

Impacted Files	Coverage Δ
beast/main.py	16.12% <ø> (ø)
beast/plotting/plot_graphic_model.py	24.00% <ø> (ø)
beast/plotting/plot_graphic_file_flow.py	4.43% <4.43%> (ø)
beast/tools/density_map.py	21.21% <0.00%> (-0.67%)	⬇️
beast/tools/run/make_ast_inputs.py	14.75% <0.00%> (ø)
beast/tools/split_catalog_using_map.py	14.54% <0.00%> (ø)
beast/observationmodel/ast/make_ast_xy_list.py	5.19% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b1950c6...93ce6db. Read the comment docs.

[karllark]

Looks great.