Use PrettyTables.pretty_table for StructArray display

General Usage
1

In my VSCode REPL I want the normal display of StructArray instances to be PrettyTables.pretty_table. What should I do?

2

Hi @jar1 !

Did you try to overload the show method? It should work.

1 Like
3

I can pirate show as you suggest.

julia> Base.show(io::IO, ::MIME"text/plain", x::StructArray) = pretty_table(io, x)

julia> StructArray(a=rand(5), b=rand(5))
┌──────────┬──────────┐
│        a │        b │
│  Float64 │  Float64 │
├──────────┼──────────┤
│ 0.557374 │ 0.415028 │
│ 0.809813 │ 0.628098 │
│ 0.248757 │ 0.320603 │
│ 0.762648 │ 0.207177 │
│ 0.478039 │ 0.693585 │
└──────────┴──────────┘

However that seems like a much bigger change than is necessary or safe. I don’t want to overload it for all output, because that could have adverse side effects, since I’m not aware of what other programs in my runtime are using show in one way or another. I only want to overload the display in my REPL.

4

If StructArray does not support changing the printing mechanism, I think overloading show is the only way.

2 Likes
5

This seems to work. It changes the repl display without changing other outputs. It’s still piracy of display and it relies on nonpublic REPL.outstream but still it seems better than pirating show.

import REPL

function Base.display(r::REPL.REPLDisplay, x::StructArray)
    io = REPL.outstream(r.repl)
    println(io, """StructArray{$(join(propertynames(x), ", "))}""")
    pretty_table(io, x)
end

julia> StructArray(a=rand(5), b=rand(5))
StructArray{a, b}
┌───────────┬───────────┐
│         a │         b │
│   Float64 │   Float64 │
├───────────┼───────────┤
│  0.165508 │   0.75791 │
│  0.440973 │  0.460919 │
│ 0.0858889 │  0.392867 │
│  0.471203 │ 0.0932289 │
│  0.410608 │  0.433334 │
└───────────┴───────────┘

julia> let io = IOBuffer()
           show(io, "text/plain", StructArray(a=rand(5), b=rand(5)))
           seekstart(io)
           println(read(io, String))
       end;
5-element StructArray(::Vector{Float64}, ::Vector{Float64}) with eltype @NamedTuple{a::Float64, b::Float64}:
 (a = 0.6178495422742597, b = 0.1238638495845299)
 (a = 0.30289555128884693, b = 0.27873172204566277)
 (a = 0.4013228628193214, b = 0.6438433505704091)
 (a = 0.8092062230983714, b = 0.3295303757547783)
 (a = 0.5775566148192897, b = 0.6535287178798409)