Add hatched fill for GR and PyPlot

[pearlzli]

I took a crack at adding a fillstyle argument which allows a few different hatching patterns, as suggested on Discourse and Gitter. I (kind of arbitrarily) followed the PyPlot convention in letting the possible values of fillstyle be :/, :\, :|, :-, :+, :x, and nothing (the default, which keeps the solid fill).

In PyPlot:

using Plots, PyPlot
pyplot()
y = rand(10)
plot(y .+ 1, fillrange = y, fillstyle = :/)

In GR, I unfortunately can't figure out how to change the color of the hatching from black:

using Plots
gr()
y = rand(10)
plot(y .+ 1, fillrange = y, fillstyle = :/)

Would love to get your feedback and suggestions!

Fix #1337.

[mkborregaard]

I love this!

[daschw]

That's amazing!

[daschw]

I tried with GR.setlinecolorind, GR.setfillcolorind, GR.setmarkercolorind and GR.settextcolorind, but nothing seems to change the hatch color. Is this not possible @jheinen?

[jheinen]

I tried with GR.setlinecolorind, GR.setfillcolorind, GR.setmarkercolorind and GR.settextcolorind, but nothing seems to change the hatch color. Is this not possible @jheinen?

Seems to be a bug in GR. Will check ...

[pearlzli]

I pushed some commits which fix setting the fill transparency level in PyPlot:

using Plots
pyplot()
y = rand(10)
plot(y .+ 1, fillrange = y, fillstyle = :/, fillalpha = 1)
plot(y .+ 1, fillrange = y, fillstyle = :/, fillalpha = 0.25)

This was already working in GR:

[pearlzli]

(Starting a new post for a separate thought.)

I also noticed (and haven't yet figured out how to address) a slightly bad thing in PyPlot. When using fillrange, you can set the line color/alpha independently from the fill color/alpha, but it doesn't show up in the legend, and it also doesn't work for plotting Shapes.

using Plots, PyPlot
pyplot()
y = rand(10)
Plots.plot(y .+ 1, fillrange = y, fillstyle = :/, linewidth = 2, linecolor = 2)
verts = [(1,0), (0,1), (-1,0), (0,-1), (1,0)]
Plots.plot(Shape(verts), fillstyle = :/, linewidth = 2, linecolor = 2)

It seems like this isn't a problem in GR, though it'll be easier to tell once we can change the hatch fill color:

[jheinen]

The GR problems should be fixed in GR.jl v0.53.0

[pearlzli]

Thanks, @jheinen! I've been a little swamped lately, but will try to make more progress on this next week.

[t-bltg]

@pearlzli, would you be able to find time to finish this ? (or is it maybe ready for review ?).

That would allow me to add support to filled pattern for the Gaston (gnuplot) backend ...

[pearlzli]

Hi @t-bltg and everyone - sorry for totally dropping the ball on this! I'll try to take a crack at this again over the weekend.

[pearlzli]

Sorry again for my delay on this! Using an updated version of GR.jl, changing the hatch color works perfectly:

using Plots
gr()
y = rand(10)
hatchstyles = [:/ :\ :| :- :+ :x nothing]
hatchnames = [":/" ":\\" ":|" ":-" ":+" ":x" "nothing"]
plot(repeat(y .+ 1, outer = (1,7)), fillrange = repeat(y, outer = (1,7)),
     fillstyle = hatchstyles, label = "fillstyle = " .* hatchnames,
     layout = (4,2))

I think I've also addressed the problem with setting the line color independently of the fill color in PyPlot that I mentioned in #3107 (comment):

using Plots
pyplot()
y = rand(10)
verts = [(1,0), (0,1), (-1,0), (0,-1), (1,0)]
p1 = plot(repeat(y .+ 1, outer = (1,2)), fillrange = repeat(y, outer = (1,2)),
          fillstyle = [:/ nothing], linecolor = 2, label = "fillrange", layout = (1,2))
p2 = plot([Shape(verts) Shape(verts)], fillstyle = [:/ nothing],
          linecolor = 2, label = "shape", layout = (1,2))
plot(p1, p2, layout = (2,1))

[codecov]

Codecov Report

Merging #3107 (fdf33d1) into master (0742d47) will increase coverage by 0.44%.
The diff coverage is 69.23%.

@@            Coverage Diff             @@
##           master    #3107      +/-   ##
==========================================
+ Coverage   62.97%   63.41%   +0.44%     
==========================================
  Files          28       28              
  Lines        7070     7372     +302     
==========================================
+ Hits         4452     4675     +223     
- Misses       2618     2697      +79     

Impacted Files	Coverage Δ
src/args.jl	74.02% <0.00%> (+0.91%)	⬆️
src/backends/gr.jl	89.39% <80.00%> (+0.18%)	⬆️
src/utils.jl	54.89% <100.00%> (+0.10%)	⬆️
src/plot.jl	66.20% <0.00%> (-1.22%)	⬇️
src/animation.jl	37.36% <0.00%> (-0.42%)	⬇️
src/themes.jl	13.95% <0.00%> (-0.34%)	⬇️
src/axes.jl	85.06% <0.00%> (-0.11%)	⬇️
src/Plots.jl	31.42% <0.00%> (ø)
src/plotattr.jl	0.00% <0.00%> (ø)
src/backends/gaston.jl	0.00% <0.00%> (ø)
... and 15 more

---

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 0742d47...fdf33d1. Read the comment docs.

[t-bltg]

LGTM, thanks for finalizing 👍

[jeanpauphilet]

Hi,
Thanks for adding this cool feature. I have noticed an issue when saving the figure as a pdf file. I am using Plots and GR. I am using the same code as @pearlzli.

using Plots
gr()
y = rand(10)
hatchstyles = [:/ :\ :| :- :+ :x nothing]
hatchnames = [":/" ":\\" ":|" ":-" ":+" ":x" "nothing"]
plot(repeat(y .+ 1, outer = (1,7)), fillrange = repeat(y, outer = (1,7)),
      fillstyle = hatchstyles, label = "fillstyle = " .* hatchnames,
      layout = (4,2))

The output looks fine on a Jupyter notebook. However when I save it as a PDF file, the hatch color is back to black. To save it, I do:

Plots.savefig("test.pdf")

Using GRUtils.savefig instead produced a corrupted pdf file.

Thanks for your help