Match url ids but not image data blobs in svg salting

[jkrumbiegel]

Fixes #3440 which was a regression introduced by #3435 because the regex doing more efficient salting accidentally also matched image base64 blobs, making them invalid:

xlink:href="data:image/png;base64,iVBOR ... SuQmCC-f0f66279"/>

[MakieBot]

Compile Times benchmark

Note, that these numbers may fluctuate on the CI servers, so take them with a grain of salt. All benchmark results are based on the mean time and negative percent mean faster than the base branch. Note, that GLMakie + WGLMakie run on an emulated GPU, so the runtime benchmark is much slower. Results are from running:

using_time = @ctime using Backend
# Compile time
create_time = @ctime fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @ctime Makie.colorbuffer(display(fig))
# Runtime
create_time = @benchmark fig = scatter(1:4; color=1:4, colormap=:turbo, markersize=20, visible=true)
display_time = @benchmark Makie.colorbuffer(fig)

	using	create	display	create	display
GLMakie	3.50s (3.48, 3.53) 0.02+-	387.59ms (385.65, 390.25) 1.77+-	475.80ms (464.85, 516.81) 18.35+-	7.16ms (7.11, 7.21) 0.04+-	25.42ms (25.32, 25.52) 0.07+-
master	3.48s (3.46, 3.51) 0.02+-	398.36ms (388.20, 402.65) 5.55+-	469.76ms (467.02, 473.27) 2.08+-	7.13ms (7.03, 7.28) 0.08+-	25.38ms (25.30, 25.53) 0.09+-
evaluation	0.99x invariant, 0.02s (0.94d, 0.10p, 0.02std)	1.03x faster ✓, -10.77ms (-2.61d, 0.00p, 3.66std)	0.99x invariant, 6.04ms (0.46d, 0.42p, 10.21std)	1.00x invariant, 0.03ms (0.53d, 0.35p, 0.06std)	1.00x invariant, 0.04ms (0.51d, 0.36p, 0.08std)
CairoMakie	3.07s (3.01, 3.15) 0.05+-	324.16ms (314.05, 339.91) 9.81+-	142.80ms (139.87, 146.98) 2.43+-	7.15ms (7.06, 7.27) 0.08+-	604.05μs (598.29, 609.47) 3.91+-
master	3.03s (2.98, 3.13) 0.05+-	318.53ms (311.60, 327.43) 5.56+-	141.12ms (138.21, 145.29) 2.34+-	7.28ms (7.07, 7.44) 0.13+-	608.02μs (602.53, 612.57) 4.08+-
evaluation	0.99x invariant, 0.04s (0.67d, 0.23p, 0.05std)	0.98x invariant, 5.63ms (0.71d, 0.22p, 7.68std)	0.99x invariant, 1.68ms (0.70d, 0.21p, 2.39std)	1.02x faster ✓, -0.12ms (-1.20d, 0.05p, 0.10std)	1.01x invariant, -3.97μs (-0.99d, 0.09p, 4.00std)
WGLMakie	3.81s (3.70, 4.02) 0.11+-	321.91ms (313.82, 349.87) 12.63+-	8.81s (8.57, 9.14) 0.23+-	9.19ms (8.88, 9.76) 0.31+-	70.50ms (67.69, 77.03) 3.12+-
master	3.75s (3.69, 3.90) 0.07+-	322.73ms (314.24, 342.72) 9.28+-	8.74s (8.52, 9.14) 0.21+-	9.17ms (8.92, 9.53) 0.22+-	69.98ms (67.48, 74.74) 2.69+-
evaluation	0.99x invariant, 0.05s (0.55d, 0.33p, 0.09std)	1.00x invariant, -0.83ms (-0.07d, 0.89p, 10.95std)	0.99x invariant, 0.07s (0.33d, 0.55p, 0.22std)	1.00x invariant, 0.03ms (0.10d, 0.85p, 0.26std)	0.99x invariant, 0.52ms (0.18d, 0.74p, 2.91std)

[j-fu]

Still broken here:
[ee78f7c6] Makie v0.20.2 https://github.com/MakieOrg/Makie.jl#eb3e086

[jkrumbiegel]

Can you make an MWE that still fails? The broken triplot examples show up correctly in the docs build on firefox and chrome for me https://docs.makie.org/previews/PR3443/reference/plots/triplot/

[j-fu]

It looks ok with display, inkscape, but not in the browser, and not with gimp.

[j-fu]

Ah was confused by the monorepo structure. It was not sufficient to add Makie from the commit, but also had to take CairoMakie from this commit.

Now I can confirm it works - sorry for the confusion & thanks for the quick act!